Yes Downloadpdf145050886.pdf 2

MORPHOLOGICAL DEVELOPMENT IN
THE INTERLANGUAGE OF
ENGLISH LEARNERS OF XHOSA
A thesis submitted in fulfilment of the

requirements for the degree of
DOCTOR OF PHILOSOPHY
of
RHODES UNIVERSITY
by
CAROL BONNIN HOBSON
December 1999
The financial assistance of the Joint Research Committee of Rhodes University is gratefully
acknowledged.
ABSTRACT
This study investigates the development of morphology in the interlanguage of English learners
of Xhosa. A quasi-longitudinal research design is used to trace development in the oral
interlanguage of six learners of Xhosa for a period of eight months. The elicitation tasks employed
range from fairly unstructured conversation tasks to highly structured sentence-manipulation
tasks. The learners have varying levels of competence at the beginning of the study and they are
exposed to input mainly in formal contexts of learning.
One of the aims of the study is to investigate whether the features of interlanguage identified in
other studies appear in the learner language in this study. Most other studies discussed in the
literature have investigated the features of the interlanguage produced by learners of analytic and
inflectional languages. However, this study analyses the interlanguage of learners of an
agglutinative language.
Studies of other languages have concluded that learners do not use inflectional or agreement
morphology at early stages of development and this conclusion is tested for learners of an
agglutinative language in this study. Since agreement and inflectional morphology play a central
role in conveying meaning in Xhosa, it is found that learners use morphology from the beginning
of the learning process. Although forms may be used incorrectly and the functions of forms may
be restricted, morphemes appear in the interlanguage of learners of this study earlier than other
studies predict.
One of the characteristics of early interlanguage and an early form of learner language called the
Basic Variety (Klein & Perdue 1997) is the lack of morphology, but this feature proves to be
inadequate as a measure of early development in the interlanguage of learners of a language such
as Xhosa. This study concludes, therefore, that the presence of morphology in the interlanguage
of learners of Xhosa cannot be an indicator of advanced language development.

TABLE OF CONTENTS
Page
LIST OF TABLES i
LIST OF APPENDICES iii
LIST OF ABBREVIATIONS iv
ACKNOWLEDGEMENTS v
CHAPTER 1: INTRODUCTION 1
1.1 Background to the study 1
1.2 Aims 3
1.3 Outline of the thesis 3
CHAPTER 2: LITERATURE REVIEW 5

2.0 Introduction 5
2.1 Contrastive Analysis 6
2.1.1 Main tenets of contrastive analysis 6
2.1.2 Problems with contrastive analysis 6
2.1.3 The relevance of contrastive analysis 8
2.1.4 Conclusion 9
2.2 Error Analysis 9
2.2.1 Characteristics of error analysis studies 9
2.2.2 Criticisms of error analysis 10
2.2.3 Errors vs. Mistakes 12
2.2.4 Conclusion 13
2.3 Morpheme Studies 13
2.3.1 Background 13
2.3.2 Criticisms of morpheme studies 14
2.3.3 Conclusion 15
2.4 Natural Routes 16
2.4.1 Background 16
2.4.2 Role of L1 18
2.4.3 Influence of elicitation tasks 18
2.4.4 Influence of input 19
2.4.5 Conclusion 20
2.5 Interlanguage 21
2.5.1 Introduction 21
2.5.2 Key concepts in interlanguage 21
2.5.3 Composition of the IL system 24
2.5.3.1 Universal elements 25
2.5.3.2 Formulaic elements 25
2.5.3.3 Conclusion 26
2.5.4 Features of interlanguage 26
2.5.5 Interlanguage development 28
2.5.5.1 General development 28
2.5.5.2 Simplification 29
2.5.5.3 Development of the interlanguage system 29
2.5.5.3.1 Strategies for development 31
2.5.5.3.2 Morphological development 34
2.5.5.3.3 Utterance development 37
2.5.6 Conclusion 38
2.6 Basic Variety 38
2.6.1 Theoretical background 39
2.6.2 Characteristics of the Basic Variety 40
2.6.3 Elements of the Basic Variety 42
2.6.3.1 Lexical items 42
2.6.3.2 Morphology 43
2.6.3.3 Functions expressed in the Basic Variety 43
2.6.3.4 Utterance structure 43
2.6.3.4.1 Phrasal constraints 44
2.6.3.4.2 Semantic constraints 44
2.6.3.4.3 Pragmatic constraints 45
2.6.4 Development of learner language 45
2.6.4.1 Pronoun development 46
2.6.4.2 Reasons for development 46
2.6.5 Conclusion 47
2.7 Formal vs. naturalistic acquisition 47
2.7.1 Limitations of the studies 50
2.7.2 Reasons for better performance 52
2.8 Conclusion 53
CHAPTER 3: METHODOLOGY 55
3.0 Introduction 55
3.1 Methodological approach 56
3.1.1 Longitudinal and cross-sectional studies 56
3.1.2 Case study approach 58
3.1.3 Qualitative methods 59
3.2 Data collection procedures 60
3.2.1 Selection criteria for the sample 60
3.2.1.1 Background of participants 61
3.2.2 Learning context 66
3.2.3 Data collection 67
3.2.3.1 Interviewer 68
3.2.3.2 Data capture methods 69
3.2.3.3 Data types 69
3.2.3.3.1 Interview content 70
3.2.3.3.2 Elicitation tasks 71
3.3 Analysis procedures 76
3.3.1 General principles of IL analysis 77
3.3.2 Analysis of IL as a system 78
3.3.3 Units of analysis 78
3.3.4 Characteristics of formulaic speech 79
3.3.5 Form-focussed analysis 79
3.3.5.1 Error analysis procedures 80
3.3.5.2 Problems with error analysis procedures 81
3.3.6 Functional analysis 83
3.3.6.1 Function-to-form analysis 83
3.3.6.2 Form-to-function analysis 84
3.3.6.3 Functions investigated in my study 85
3.4 Limitations of the study 85
3.4.1 System qua system 86
3.4.2 Competence/performance 86
3.4.3 Determining the norm 87
3.4.4 Classifying an error 88
3.4.5 Different levels of competence 89
3.4.6 Generalisability based on subjects and tasks 90
3.4.7 Elicitation procedures 90
3.4.8 Appropriate linguistic frameworks for SLA studies 91
3.4.9 Interviewer interference 92
3.4.10 Time frames for data collection 92
3.4.11 Amount of data 92
3.5 Conclusion 93
CHAPTER 4: FORMS AND FUNCTIONS IN THE DATA 94

4.0 Introduction 94
4.1 Size of the data set 94
4.2 Observations on the interlanguage of each learner 96
4.2.1 Strategies for communicating 99
4.2.2 Relationships between elements of the utterance 105
4.3 Functions evident in the data set 107
4.3.1 Referential function 108
4.3.2 Spatial reference 109
4.3.3 Temporal reference 111
4.4 Morphemes evident in the data set 114
4.4.1 Noun prefixes 114
4.4.1.1 Patterns of noun prefix use 116
4.4.1.1.1 Unmarked and marked classes 117
4.4.1.1.2 Class 2 for class 2a 117
4.4.1.1.3 Animacy cues 117
4.4.1.1.4 Opposite of animacy 117
4.4.1.1.5 Singular and plural substitution 118
4.4.1.1.6 Class 2 and 6 interchangeability 118
4.4.1.1.7 E-/i- confusion 118
4.4.1.1.8 Xhosa concords on English words 118
4.4.1.2 Patterns of noun prefix omission 119
4.4.1.2.1 No Xhosa prefixes on English words 119
4.4.1.2.2 English in utterances 120
4.4.1.2.3 Other forms added 120
4.4.1.2.4 Initial vowels 120
4.4.1.2.5 Beginnings of utterances 120
4.4.1.2.6 Correct omission on vocatives 120
4.4.2 Subject concords 121
4.4.2.1 Patterns of use of the subject concord 122
4.4.2.1.1 Pronoun confusion 122
4.4.2.1.2 Animacy 123
4.4.2.1.3 Opposite of animacy 123
4.4.2.1.4 Class 2 and 6 confusion 124
4.4.2.1.5 Hidden forms 124
4.4.2.1.6 Singular and plural confusion 124
4.4.2.1.7 Plural forms interchanged 125
4.4.2.1.8 Two referents with a singular subject
concord 125
4.4.2.1.9 Subject concords on vowel-commencing
verbs 125
4.4.2.1.10 Existential forms 125
4.4.2.2 Patterns of subject concord omission in the data 126
4.4.2.2.1 Vocabulary retrieval problems 126
4.4.2.2.2 Intervening morphemes 127
4.4.2.2.3 After xa 127
4.4.2.2.4 After pronouns 127
4.4.2.2.5 No ndi- form 127
4.4.2.2.6 Repetition 127
4.4.2.2.7 Thembi supplies the verb 128
4.4.2.3 Infinitive forms 128
4.4.3 The verbal formative -ya- 129
4.4.4 Negative subject concords 131
4.4.4.1 Patterns of use of the negative subject concord 133
4.4.4.1.1 Singular/plural substitution 133
4.4.4.1.4 Class 6 problems 134
4.4.4.1.5 Past negatives 134
4.4.4.2 Negatives of nouns 134
4.4.4.3 Use of the negative terminative i- 135
4.4.5 Verbal extensions 136
4.4.6 Auxiliaries 136
4.4.7 Participials 137
4.4.8 Copulatives 137
4.4.8.1 Usage patterns for the copulative 138
4.4.8.1.1 Non-copulative forms in copulative position 139
4.4.8.1.3 Opposite of animacy cues 139
4.4.8.2 Copulative omission patterns in the data 139
4.4.8.2.1 Omissions with demonstratives 140
4.4.8.2.2 Leave to the context 140
4.4.8.2.3 Verbal functions 140
4.4.9 Locatives 140
4.4.9.1 Omission patterns in the locative 143
4.4.9.1.1 Omissions on English words 144
4.4.9.1.2 No “is” form 144
4.4.9.1.3 Omit prepositional forms 144
4.4.10 Demonstratives 145
4.4.10.1 Patterns of use of the demonstrative 146
4.4.10.1.2 Overgeneralised forms 146
4.4.10.1.4 Plural forms are used interchangeably 147
4.4.10.1.5 Proximity distinctions not made 147
4.4.10.1.6 Correct usage 147
4.4.11 Possessive 148
4.4.11.1 Patterns of possessive concord use 150
4.4.11.1.3 Singular/plural interchangeability 151
4.4.11.1.4 Marked classes 151
4.4.11.1.6 Class 1 overgeneralisation 152
4.4.11.1.7 Overuse of ka- 152
4.4.11.1.8 Generalised e- form 152
4.4.11.2 Patterns of omission 153
4.4.12 Qualificatives 153
4.4.12.1 Patterns of use of the qualificative 155
4.4.12.2 Omission patterns of the qualificative in the data 156
4.4.12.2.1 On numerals 156
4.4.12.2.2 Stems with lu- 156
4.4.12.2.3 Lack of coalescence 156
4.4.13 Qualificative a- morpheme 157
4.4.13.1 Patterns of use of the qualificative a- morpheme 158
4.4.13.2 Omission patterns of the qualificative a- morpheme 158
4.4.13.2.1 Processing constraints 158
4.4.13.2.2 No subordinate clause marking 158
4.4.13.2.3 Stems beginning with lu- 158
4.4.13.3 Verbal relatives 159
4.4.14 Na-/nga-: “with” 159
4.4.15 Na-: Associative copulative 161
4.4.15.1 Patterns of use of the associative concord 161
4.4.15.1.1 Lack of correct negative forms 161
4.4.15.1.2 Formal form 161
4.4.15.1.3 Lack of coalescence 162
4.4.16 Na-: Additive conjunctions 162
4.4.16.1 Patterns of additive conjunction use 163
4.4.16.1.1 Not on lists 163
4.4.16.1.2 Discourse markers 163
4.4.16.1.3 On infinitives 163
4.4.16.1.4 On verbs 163
4.4.17 Object concords 164
4.4.18 Njenga- 166
4.5 Conclusion 166
CHAPTER 5: ACQUISITION PATTERNS IN THE INTERLANGUAGE

OF LEARNERS OF XHOSA 168
5.0 Introduction 168
5.1 Principles and patterns in morpheme use and omission 168
5.1.1 Singular forms for plural forms and vice versa 168
5.1.2 Mix forms 170
5.1.3 Animacy cues 170
5.1.4 Opposite of animacy 171
5.1.5 Marked and unmarked 171
5.1.6 Beginnings of words 172
5.1.7 Overgeneralised forms 173
5.1.8 Redundancy 174
5.1.9 Restricted functions of morphemes 174
5.1.10 Distance from the noun 174
5.1.11 Processing problems 175
5.1.12 Discreteness of forms 175
5.1.13 Influence of English 176
5.1.14 Structure of words 176
5.1.15 Phonological constraints 176
5.1.16 Multiples functions for one form 177
5.1.17 Rule application 177
5.1.18 Conclusion 177
5.2 Levels of development 178
5.2.1 Givón’s pragmatic-syntactic continuum 178
5.2.1.1 Conjunction use 179
5.2.1.2 Noun:verb ratio 180
5.2.1.3 Verb complexity 180
5.2.1.4 Use of grammatical morphology 181
5.2.1.4.1 Morpheme suppliance 181
5.2.1.4.2 Target-like use 182
5.2.1.4.3 Ninety-percent suppliance 183
5.2.1.4.4 Number of different morphemes in the data 183
5.2.1.4.5 Range of morphemes used to realise the
referential, spatial and temporal functions 184
5.2.1.5 Reliability of rankings 185
5.2.2 Corder’s characteristics of simplified systems 185
5.2.2.1 Fixed word order 186
5.2.2.2 Simple personal pronoun system 186
5.2.2.3 Few grammatical function words or grammatical
categories 188
5.2.2.4 Absence of article system 188
5.2.2.5 Some deictic words 188
5.2.2.6 Syntactic relations shown by word order 188
5.3 Sequences of morphological development 189
5.3.1 Order of teaching 191
5.4 General sequences of development for morphemes 193
5.4.1 Order of appearance 194
5.4.2 Morpheme suppliance 196
5.4.3 Comparison with English sequences of development 197
5.5 Explanation of the sequence of development in my study 198
5.5.1 Morphemes as foundations in utterances 198
5.5.2 Links to thematic roles 199
5.5.3 Agreement functions vs. Semantic functions 199
5.5.4 Invariant forms 200
5.5.5 Intraphrasal forms 200
5.5.6 Extension of the basic utterance 200
5.5.7 Lexical equivalence to English 200
5.5.8 Embedded forms 201
5.5.9 Free vs. Bound forms 201
5.6 Avoidance patterns in the data 201
5.7 Performance on different tasks 202
5.8 Conclusion 206
CHAPTER 6: COMPARISONS WITH THE BASIC VARIETY AND

CONCLUSION 207
6.0 Introduction 207
6.1 Comparisons with the ESF study 207
6.1.1 Why might there be points of similarity? 207
6.1.2 Why might there be differences between the BV and the learner
language in my study? 208
6.1.3 Comparison of Basic Variety forms and forms in the IL of learners
in my study 209
6.1.3.1 Linguistic elements in the BV and my study 209
6.1.3.2 Phrasal constraints 211
6.1.3.3 Levels of development 212
6.1.4 Conclusion 213
6.2 Suggestions for further research 214
6.3 Conclusion 215
REFERENCES 217
APPENDICES 243
i
LIST OF TABLES
Page
TABLE 2.1 FEATURES OF THE PRAGMATIC AND SYNTACTIC MODE 27
TABLE 3.1 INTERVIEW SCHEDULE 70
TABLE 4.1 NUMBER OF UTTERANCES AND MORPHEMES IN THE DATA 95

TABLE 4.2 CONJUNCTIONS USED BY THE LEARNERS 107
TABLE 4.3 FORMS USED TO REALISE THE REFERENTIAL FUNCTION 109
TABLE 4.4 FORMS USED TO REALISE SPATIAL REFERENCE 110
TABLE 4.5 ADVERBS USED BY THE LEARNERS 111
TABLE 4.6 FORMS USED TO CONVEY THE TEMPORAL FUNCTION 112
TABLE 4.7 SUMMARY OF NOUN PREFIX DATA 115
TABLE 4.8 PATTERNS OF NOUN PREFIX USE 116
TABLE 4.9 PATTERNS OF NOUN PREFIX OMISSION 119
TABLE 4.10 SUMMARY OF SUBJECT CONCORD DATA 122
TABLE 4.11 PATTERNS OF SUBJECT CONCORD USE 122
TABLE 4.12 SUBJECT CONCORD OMISSIONS IN THE DATA 126
TABLE 4.13 SUMMARY OF THE USE OF -YA- 130
TABLE 4.14 SUMMARY OF NEGATIVE SUBJECT CONCORD USE 132
TABLE 4.15 PATTERNS OF USE OF THE NEGATIVE SUBJECT CONCORD 133
TABLE 4.16 SUMMARY OF LACK OF NEGATIVE TERMINATIVE -I FOR
NEGATIVE VERBS 135
TABLE 4.17 SUMMARY OF COPULATIVE DATA 138
TABLE 4.18 COPULATIVE OMISSION PATTERNS IN THE DATA 139
TABLE 4.19 SUMMARY OF LOCATIVE MORPHEMES 141
TABLE 4.20 OMISSION PATTERNS IN THE LOCATIVE 144
TABLE 4.21 SUMMARY OF DEMONSTRATIVE DATA 145
TABLE 4.22 PATTERNS OF USE OF THE DEMONSTRATIVE 146
TABLE 4.23 SUMMARY OF POSSESSIVE CONCORD DATA 149
TABLE 4.24 PATTERNS OF POSSESSIVE CONCORD USE 150
TABLE 4.25 SUMMARY OF QUALIFICATIVE CONCORD DATA 154
TABLE 4.26 OMISSION PATTERNS OF THE QUALIFICATIVE IN THE DATA 156
TABLE 4.27 SUMMARY OF QUALIFICATIVE A- CONCORD 157
TABLE 4.28 SUMMARY OF NA-/NGA- 160
TABLE 4.29 SUMMARY OF THE ASSOCIATIVE COPULATIVE 161
TABLE 4.30 SUMMARY OF USE OF THE ADDITIVE CONJUNCTION 162
TABLE 4.31 PATTERNS OF ADDITIVE CONJUNCTION USE 163
TABLE 4.32 SUMMARY OF THE USE OF OBJECT CONCORDS 164
ii
TABLE 5.1 CONJUNCTION USE IN THE DATA 179

TABLE 5.2 RANGE OF FORMS ADDED TO THE VERB 181
TABLE 5.3 NUMBER OF MORPHEMES USED TO REALISE EACH
FUNCTION 184
TABLE 5.4 COMPARISON OF ORDER OF TEACHING, APPEARANCE
AND ACCURACY FOR ANN 192
TABLE 5.5 SUPPLIANCE OF TOP FIFTEEN MORPHEMES 197
TABLE 5.6 COMPARISON OF ENGLISH AND XHOSA SEQUENCES OF
DEVELOPMENT 198
TABLE 6.1 COMPARISON OF LINGUISTIC ELEMENTS IN THE BV AND

MY STUDY 210
iii
LIST OF APPENDICES
Page
APPENDIX A BIOGRAPHICAL/ATTITUDINAL QUESTIONNAIRE 243
APPENDIX B A SHORT SUMMARY OF THE BACKGROUND OF 244
THE PARTICIPANTS
APPENDIX C INTERVIEW MATERIAL 246
APPENDIX D NUMBER OF UTTERANCES IN THE DATA 255
APPENDIX E COUNTING MORPHEMES 256
APPENDIX F XHOSA STRUCTURE 257
APPENDIX G1 INCORRECT NOUN PREFIXES 261
APPENDIX G2 NOUN PREFIX OMISSIONS 262
APPENDIX G3 INCORRECT SUBJECT CONCORDS 263
APPENDIX G4 SUBJECT CONCORD OMISSIONS 264
APPENDIX G5 SUPERFLUOUS USE OF -YA- 265
APPENDIX G6 INCORRECT NEGATIVE SUBJECT CONCORDS 266
APPENDIX G7 LACK OF I-ENDING ON NEGATIVES 267
APPENDIX G8 INCORRECT COPULATIVE 268
APPENDIX G9 COPULATIVE OMISSIONS 269
APPENDIX G10 LOCATIVE OMISSIONS 270
APPENDIX G11 INCORRECT LOCATIVES 271
APPENDIX G12 ONLY INITIAL PART OF LOCATIVE 272
APPENDIX G13 ONLY FINAL PART OF LOCATIVE 273
APPENDIX G14 INCORRECT DEMONSTRATIVES 274
APPENDIX G15 INCORRECT POSSESSIVE CONCORDS 275
APPENDIX G16 POSSESSIVE CONCORD OMISSION 276
APPENDIX G17 INCORRECT QUALIFICATIVE CONCORDS 277
APPENDIX G18 QUALIFICATIVE CONCORD OMISSIONS 278
APPENDIX G19 INCORRECT QUALIFICATIVE A- CONCORDS 279
APPENDIX G20 QUALIFICATIVE A- CONCORD OMISSIONS 280
iv
LIST OF ABBREVIATIONS
ADJ Adjective
BV Basic Variety
C Conversation
CA Contrastive Analysis
CCH Creative Construction Hypothesis
Cop Copulative
Det Determiner
EA Error Analysis
ESF European Science Foundation
FL Foreign Language
II Interlingual Identification
IL Interlanguage
L1 First Language
L2 Second Language
MT Mother Tongue
NC Noun Class
NIVES Non-native Institutionalised Varieties of English
NL Native Language
NP Noun Phrase
NS Native Speaker
O Object
PD Picture Description
PP Prepositional Phrase
S Subject
SB Sentence building
SL Source Language
SLA Second- Language Acquisition
SVO Subject Verb Object
T Translation
TL Target Language
UG Universal Grammar
V Verb
VP Verb Phrase
v
Acknowledgements
I would like to thank the following people for their contributions and support:
C The six participants who committed themselves to a longitudinal study and without whom
this study could not have been conducted;
C Thembi, whose unfailing sense of humour and excellent interviewing skills ensured that
the participants completed the interviews;
C Dr John Claughton, Ms Ntosh Mazwi and Ms Bulelwa Nosilela of the African Languages
Department at Rhodes University, who were willing to give me access to their classes and
learners;
C Ms Bulelwa Nosilela who proofread the Xhosa parts of this study;
C My supervisor, Professor Vivian de Klerk, who is an excellent role model in all spheres
of academic life. Her comments on drafts and encouragement during the study were
invaluable;
C My supervisor, Professor Gary Barkhuizen, who taught me much of what I know about
research. His careful reading of my work improved it substantially, and his early words of
encouragement kept me motivated;
C My colleagues, who carried extra burdens while I completed the study;
C My family, whose contributions to my education and my life empowered me to
contemplate doing this study;
C Kevin Chadd, who lived through the highs and lows of this project with me, and had to
learn about interlanguage to survive.
This thesis is dedicated to my grandfather, Donald Bonnin Hobson.

1
CHAPTER 1
INTRODUCTION
L2 [second-language] acquisition research is concerned with investigating the processes

and stages that learners undergo in acquiring L2s and subsequent languages. (Nunan
1996:349)
1.1 BACKGROUND TO THE STUDY

The term interlanguage, which refers to the language produced by learners acquiring an additional
language, was coined by Selinker (1972), although studies of the interlanguage phenomenon
predate the term by several years (Corder 1967). A variety of studies has supported the
hypotheses that interlanguage is systematic, permeable to influence from the first language and
follows a natural sequence of development (Adjémian 1976; Dulay & Burt 1974a; Lalleman
1996).
However, the majority of studies investigating the development of interlanguage (IL) has focussed
on speakers of a variety of different languages learning Indo-European (and usually Germanic and
Italic) languages. Most studies are of learners of English (see Larsen-Freeman & Long, 1991, for
an overview), although there are also studies of learners of other languages such as Dutch,
French, Spanish and German (e.g. Carroll 1999; Extra & van Hout 1996; Lalleman 1983;
Lightbown 1980; Jordens 1988; Meisel et al. 1981). These studies provide insights into
acquisition patterns in a variety of areas including word order, the development of questions, the
use of inflectional morphemes and utterance structure. The patterns of acquisition found in these
studies seem to be fairly universal amongst learners of these languages, but it should be noted that
the languages are all fairly similar in terms of language typology.
The Germanic and Italic languages are mainly analytical, inflectional or a mixture of the two
(Crystal 1997), and inflectional morphemes therefore have varying degrees of salience in these
languages. Although many studies have focussed on the acquisition of inflectional morphemes in
English, there are relatively few of these morphemes in English, compared with many other
languages. As a result, a different perspective on second-language acquisition (SLA) can be
obtained by looking at languages with typologies which differ from those of Indo-European
languages.
2
Slobin (1992:9) found in studies of L1 acquisition that typological factors were important: “It
seems that we attend to typological factors when encountering languages that have unfamiliar
characteristics, such as Eskimo polysynthesis, Bantu prefix concord, and Semitic
morphophonemic patterns”. Unfortunately, studies of languages with these typologies are rare
and, in particular, the process of learning African languages is rarely documented (a study by
Musau, 1995, is a notable exception).
The theoretical foundation of this study falls into the scope of interlanguage (IL) studies which
have developed from contrastive analysis studies, error analysis studies and the morpheme studies.
Also discussed is the more recent writing on the Basic Variety, an early learner variety which
emerged from a cross-linguistic study of the IL of early learners of a range of different languages
who were learning in naturalistic settings (Perdue 1993a, b).
One of the main findings of many SLA studies is that the interlanguages of early learners have
many common characteristics (Larsen-Freeman & Long 1991). Typically, grammatical
morphology is omitted and there is a reliance on the context, rather than morphology, to perform
many of the functions needed by the learners in their communication.
However, inflectional morphemes play a much bigger role in agglutinative languages than they
do in Germanic and Italic languages because much of the meaning and many of the grammatical
relationships are conveyed by the use of morphemes. Demuth (1992:269) concludes then that it
is possible “that children learning morphologically rich languages produce more morphology
earlier than their more analytic language counterparts”. There are therefore two possibilities for
the acquisition of grammatical morphology:
C first-language learners of morphologically rich languages learn morphemes early, as
Demuth (1992) proposes; or
C grammatical morphology is universally lacking at early stages of L2 learning (as many
studies have shown).
A study of L2 learners of an agglutinative language can shed light on which of the two views is
correct for L2 acquisition of an agglutinative language. My study therefore investigates the
language development of English learners of Xhosa, a Niger-Congo agglutinative language.
My study traces morphological development and some functional development in the IL of six
3
English-speaking learners of Xhosa, who have varying levels of Xhosa competence. The learners
were interviewed seven times over a period of eight months using a range of elicitation tasks,
ranging from fairly unstructured, spontaneous conversations to highly structured utterance-
manipulation tasks.
The study takes both a formal and a functional approach to the data analysis. This dual approach
mitigates some of the negative evaluation of errors which is associated with error analysis and the
formal approach, and it provides an interesting perspective on the range of morphemes used to
realise particular functions in the data. In addition, the functional approach sheds light on what
the learners actually need to do with their IL at early stages of development.
1.2 AIMS
The central aims of the study are:
a) To provide a qualitative description and analysis of the morphological development in
the interlanguage of selected English mother-tongue speakers learning Xhosa;
b) To determine the sequence of acquisition of morphemes across learners and to compare
this sequence to common sequences of acquisition found in other studies;
c) To explain why learners tend to display preference patterns in terms of which morphemes
they omit and which morphemes they use;
d) To determine whether the lack of grammatical morphology is a feature of the IL of
learners of an agglutinative language;
e) To determine the extent to which the acquisition patterns in this study conform to the
patterns of the Basic Variety.
1.3 OUTLINE OF THE THESIS

Chapter 2 discusses the theoretical foundation of my study. It includes an historical perspective
on the development of IL studies, including a discussion of contrastive analysis, error analysis and
the morpheme studies. In addition, there is a focus on the findings of a variety of IL studies with
regard to the features of IL. This is followed by a discussion of the Basic Variety, a more recent
development in IL studies. The chapter concludes with a discussion of the influence of formal
instruction on IL.
4
Chapter 3 outlines the methodology used in the study and provides motivations for the use of
formal and functional approaches to the analysis of the data. The limitations of the methodology
are discussed in this chapter.
Chapter 4 presents some of the results of the study. The chapter is divided to reflect the functional
and formal analysis approaches to the data. In the section on how functions are realised in the
data, the range of morphemes (and other means) used to convey each function is presented. In the
section on the formal characteristics of the data, the range of morphemes found in the IL of the
learners is presented. For each of the morphemes, a table of results indicating correct usage,
incorrect choices and omissions of morphemes is included. In addition, there are examples of
different usage and omission patterns where common patterns are found across learners.
Chapter 5 discusses common patterns of use and omission across different morphemes, and links
these patterns to those found in other studies. The level of development of learners in my study
is analysed, and problems with applying available methods of measuring the level of development
to learners of an agglutinative language are mentioned. Finally, a classification of early- and later-
acquired morphemes is made and a tentative sequence of development of morphemes is
suggested.
In the final chapter, the findings of my study are compared with those of the European Science
Foundation study (Klein & Perdue 1997; Perdue 1993a, b) in order to find similarities and
differences in selected areas between the interlanguage produced in my study and the Basic
Variety. Finally, suggestions for further research are made.
5
CHAPTER 2
LITERATURE REVIEW
2.0 INTRODUCTION
My study investigates the characteristics of the interlanguage (IL) of English learners of Xhosa
and compares these characteristics to those found in studies of other ILs. The preliminary
questions which need to be considered in a survey of the relevant literature are therefore:
C what are the characteristics of other ILs;
C which theoretical and methodological perspectives influenced these studies; and
C why do the ILs in these studies look the way they do?
My study is influenced primarily by the theoretical traditions of IL and the more recent findings
pertaining to the Basic Variety (BV) (Perdue 1993a, b). However, even the most recent
developments in SLA theory and data analysis cannot ignore the roots of the discipline entirely,
and since this study uses several of the basic ideas and data analysis techniques of earlier
paradigms, a range of relevant areas will be reviewed. This chapter therefore traces the history
of studies of IL, beginning with contrastive analysis (CA), moving on to error analysis (EA) and
the morpheme studies from which the notion of IL developed, to the findings of BV studies.
Selinker (1992:4) states that IL has its roots in CA and EA and that the three areas of CA/EA/IL
are inextricably linked. This view is also expressed by Corder (1981) in his discussion of EA as
a basic learner strategy and a starting point for the study of transfer in ILs. Each of these
approaches developed as a result of changing views of the status of the language produced by
learners, changing perceptions of the forces driving development and the need for methodological
innovation in the study of this form of language. In the absence of a theoretical framework which
adequately describes and explains all second-language acquisition (SLA) data, it is also probably
best to take an eclectic approach (Tarone 1994). Furthermore, any study also needs to consider
a variety of SLA phenomena including variability, systematicity, transfer, staged development and
incompleteness (Towell & Hawkins 1994). Each of the four approaches (CA, EA, IL, BV)
discussed below accounts for each of these areas with varying degrees of success.
6
2.1 CONTRASTIVE ANALYSIS

2.1.1 Main tenets of contrastive analysis
The study of SLA as we know it today is rooted in early CA, which became the dominant
approach during the 1950s and 1960s. According to Lado (1957), the purpose of CA is to
carefully describe the source language (SL) and the target language (TL) in order to develop
effective pedagogical materials. The basic assumption of CA is that learning a second language
(L2) entails transferring the linguistic forms and meanings of the first language (L1) to the L2 by
learning a set of habits. Contrastive analysts predicted that some languages would be easier to
learn than others, because where languages differed greatly in structure, the learner would be
required to automatise a more complex set of habits. This automatisation would take longer than
if languages were structurally more similar.
Three main theoretical claims result from CA:

C the learner expects to find equivalence between the SL and TL;
C learning the TL requires the learner to learn something which is different from the SL;
and
C the learner finds it difficult to learn the new language when structures and patterns in the
two languages differ (Selinker 1992).
Studies taking a CA perspective therefore focussed primarily on transfer phenomena and

especially negative transfer, which occurred when languages differed in structure. As a result of
these structural differences, learning a language was not a simple matter of transferring a form
directly from L1 to L2. The primary focus of CA studies was therefore on difference and types
of difference (Long & Sato 1984).
Detailed analyses of similarities and differences were carried out by comparing languages in terms
of mainly phonology and syntax and, to a lesser extent, semantics (Fries 1945; James 1980; Lado
1957; Weinreich cited in Selinker 1989). Many later studies of transfer take a contrastive
approach, although a detailed CA is not always carried out.
2.1.2 Problems with contrastive analysis

Unfortunately, CA in its original formulation proved to be seriously flawed when applied to data
from learners across the world in different language-learning situations. Most seriously, the
strong predictions of difficulty and ease of learning which are intuitively appealing were not
7
always borne out by studies of learner language (see Lalleman 1996 for an overview; Wardhaugh
1970). In particular, some researchers found that when there was a great degree of difference
between languages, learners seemed to be able to produce the form correctly (i.e. there was no
negative transfer), whereas if there was a small degree of difference learners seemed to find it
more difficult to produce the correct form (Kellerman 1979; Odlin 1989; Towell & Hawkins
1994; Whitman & Jackson 1972; Wode 1978). As a result, some areas of error were not
predicted by CA (Hyltenstam 1977). Furthermore, students tended to avoid difficult areas to
reduce the possibility of making errors, and thus the full range of possible errors was not available
for study in this approach (Schachter 1974).
From an explanatory point of view, another limitation of CA studies lies in the extreme role of
transfer posited by early theorists like Lado (1957). Early theorists believed that language transfer
was the main process in SLA. However, later studies show that many errors are not simply
traceable to the L1 (Dulay & Burt 1973; Felix 1980; George 1972; Richards 1985). In addition,
some errors may be a result of performance problems (Zobl 1984) and errors are subject to
variability (Zobl 1982, 1984). Several theorists concluded that although there is some role for
transfer, learners choose in an active and principled way whether or not to transfer and what to
transfer (Gass 1979, 1984a; Kellerman 1979; Selinker 1992). Contrastive analysis does not
account for this active role of the learner, because it is primarily interested in the languages as
linguistic systems and products rather than in learners using complex psycholinguistic processes
(Long & Sato 1984; van Els et al. 1984). As a result of the failure of the “strong” version of CA,
Wardhaugh (1970) suggested a “weak” version of CA which proposed that the findings of CA
could be used to explain transfer after the fact. This version had limited explanatory value,
although it was later incorporated as part of EA (James 1998).
From a theoretical perspective, there are profound problems with comparing languages at a
superficial or surface level if one views them as independent systems (see section 3.4.1 for a more
detailed discussion). For example, it is highly controversial whether one can meaningfully
compare English and German, because one is comparing the elements of one system with the
elements of another system and trying to find similarities and differences across independent
systems. The value of the elements may not be the same in each system and these elements may
therefore be incomparable. Furthermore, Lattey (cited in Selinker 1992) points out that even
8
though structures across languages have the same name, they may not in fact be the same (e.g.
clitics in French and clitics in Xhosa) and the size of the structure to be compared may also be
problematic. For example, should one study questions, as Langacker did (cited in Selinker 1992),
or should one look only at “primary questions”, as Armagost did (cited in Selinker 1992)? A
further problem encountered when comparing learner language cross-linguistically is that a
learner and a language may use a variety of strategies and structures to express a particular
concept or function (Comrie 1984), making direct comparison difficult.
2.1.3 The relevance of contrastive analysis

Despite the problems mentioned above, CA can be a useful approach to SLA. For example, there
are patterns across languages and one would be missing a great deal of regularity if one failed
to compare languages and elements of those languages at all. One of the solutions to the problem
of comparing surface features is to posit regularities at the deep level or abstract level (Jake
1998). These regularities may take the form of common grammatical units which occur
universally in learner languages. These units have been termed interlingual identifications (IIs),
but they have not yet been identified since “... no unit of linguistic theory, as these units are
currently conceived, could fit [the criterion] of a unit identified interlingually across three
linguistic systems (NL [native language], TL and IL) ...” (Selinker 1972:225). Selinker
(1972:227-228) describes these IIs as
... a new type of psycholinguistic unit, available to an individual whenever he [sic]
attempts to produce sentences in a second language. This interlingual unit stretches, we
hypothesize, across three linguistic systems: NL, IL, and TL, and becomes available to
the idealized second-language learner who will not achieve native-speaker competence
in the TL, whenever he attempts to produce a TL norm. These units become available to
the learner only after he has switched his psychic set or state from the native-speaker
domain to the new domain of interlingual identifications. I would like to postulate further
that these relevant units of interlingual identifications do not come from anywhere; they
are latent in the brain in a latent psychological structure, available to the individual
whenever he wishes to attempt to produce the norm of any TL.
Selinker (1992) and James (1994) feel that these IIs are still relevant, because although a
comprehensive linguistic theory of IIs has not yet been developed and these units are as yet
undiscovered and undescribed, numerous studies have upheld this view of a learner contrastively
“setting up interlingual identifications” (Selinker 1992:210).
9
The problem lies in trying to overcome the mind-set of analysts who have been trained in the use
of current linguistic terminology and units and requiring them to study the phenomenon in an
entirely different way with a set of as yet unavailable analytical tools. This area of research
promises to be fruitful if analysts can commit themselves to studying learner languages cross-
linguistically as independent systems, so that regularities can be noted. It is only once regularities
across learner languages are observed that the role of transfer and cross-linguistic influence could
be determined more accurately.
2.1.4 Conclusion
In conclusion, transfer may be one aspect of SLA, but it does not explain it fully. An adequate
explanation of the process of development in SLA is therefore not provided by CA (Towell &
Hawkins 1994), although Zobl (1984:79) says CA “was not an acquisition theory; or,
alternatively, it lacked one”. Contrastive analysis was largely abandoned during the 1970s, but
it is perpetuated in a modified form in transfer analysis (James 1990). Transfer analysis is
concerned mainly with processes such as cross-linguistic influence (Giacobbe 1992; Kellerman
1986; Kellerman & Sharwood Smith 1986) and language transfer (Gass & Selinker 1992; Odlin
1989). This newer version of CA is not the same as the original because “you are comparing IL
with MT [mother tongue] and not MT with TL” (James 1998:5). My study will take a transfer
analysis perspective at times, although English and Xhosa are structurally very different and one
would expect less transfer than in more closely related languages such as Romance languages.
2.2 ERROR ANALYSIS

2.2.1 Characteristics of error analysis studies
During the 1970s, error analysis (EA) emerged as the next major development in the study of
SLA, and although the studies sometimes attempted to explain how second languages are learnt,
EA remained primarily a methodological approach rather than a theory of SLA. The central focus
in EA is on the L2 learner rather than on the system, as in CA (van Els et al. 1984), although the
“systematicity in development and the common processes posited to explain development” (Long
& Sato 1984:256) became central features of EA. As the pendulum swung away from CA,
theorists claimed that errors could be explained in terms of the target language (TL) only, with
no reference whatsoever to the source language (SL), i.e. there was no transfer. In reality, this
proved to be too extreme a view once data were carefully analysed.
10
The EA studies focussed mainly on the performance of learners of a few languages who had
learned in formal contexts and were studied in experimental conditions. These studies had two
primary aims. The first aim was to provide explanations for errors, which were attributed to
learner-internal cognitive processes and learner-external causes. Learner-internal errors include:
C overgeneralisations (Schumann 1978);
C developmental patterns which remain the same regardless of the L1 of the learner (Dulay
& Burt 1973, 1974a, 1974b; Huang & Hatch 1978; Richards 1985);
C ignorance of rule restrictions (Richards 1985);
C incomplete application of rules (Richards 1985); and
C learning strategies (Selinker 1972).
Learner-external errors related to problems with the input received by the learner, especially in
the case of formal instruction (Faerch et al. 1984) and included errors such as transfer of training
(Felix 1981; Selinker 1972; Stenson 1974). As a result of finding learner-external causes of
errors, the second, later aim of EA was to relate the social context of learning to the errors
produced (Faerch et al. 1984; Seliger & Long 1983). These studies looked closely at the input
available to the learner and studied the errors in relation to the input received.
2.2.2 Criticisms of error analysis

Error analysis has been strongly criticised, both from a theoretical and a methodological point of
view. Firstly, in an EA the norm is the TL and any deviation from the target is viewed as an error.
However, determining a norm is problematic because it depends on a variety of factors including
the linguistic context, “the medium (spoken or written language), the social context (formal or
informal), and the relation between speaker and hearer (symmetrical or asymmetrical)” (van Els
et al. 1984:47). Deviation from the norm is viewed negatively, which means that these studies do
not acknowledge the creative processes learners use in building the new language. They therefore
ignore a large part of the developmental process.
From a methodological point of view, many limitations are discussed in the literature. Firstly, EA
measures production (which may be fairly restricted), rather than perception (which may be less
restricted) (Alexander 1979). Secondly, EA studies focus on only a small part of the production
data (i.e. the errors) rather than all the learner language produced (Alexander 1979; Corder 1975;
Schachter & Celce-Murcia 1977). This means that some “errors” would not appear to be errors
because they seem to be well-formed, although they may be misformed from a pragmatic point
11
of view (Zydatiss cited in Alexander 1979). Furthermore, learners may avoid some of the TL
constructions because they do not know how to produce them or because certain structures are
perceived as difficult and more likely to induce error (Alexander 1979; Kleinmann 1977;
Schachter 1974). The group that does produce these constructions, albeit with errors, is not
directly comparable to the group which avoids the constructions and therefore makes fewer
errors overall.
Another methodological problem is that the task used to elicit data may have an effect on the
errors produced so that different types and numbers of errors may be produced in different tasks.
Schachter & Celce-Murcia (1977) claim that errors were also often classified very subjectively
and that analysts did not always know enough about the languages they were studying to notice
subtle but important differences. Analysts did not always correctly identify L1 influence on the
learner language since different L1s may influence the source of the error. For example, what is
probably a transfer error for a speaker of one language may be a simplification error for a
speaker of another language. Related to this point is the way in which errors are classifed and
quantified. Some studies ascribe errors to one source when there could have been more than one
source and other studies ascribe errors to several sources when there was only one source. As
Long & Sato (1984:257) note:
Explanations were often impressionistic and vague. Two or more sources of error were
often plausible, yet analysts sometimes opted for just one. This is a criticism taken up by
Burt, Dulay and Krashen (1982), who see the root of the problem as researchers’
attempts to describe and classify errors simultaneously. Burt et al. argue for a two-stage
analysis. First, errors should be described, e.g. by reference to linguistic domain (word
order, morphology, lexis, etc.) or ‘surface strategy’ (omission, addition, misinformation
or misordering). Only then should causes, such as overgeneralization or interference, be
attributed.
Another quantification problem occurs when an error is found over a larger linguistic domain than
a word (Schachter & Celce-Murcia 1977). In some cases, one error may create additional errors
in a text and it may be difficult to decide how to quantify these error forms. Quantification is also
problematic since some studies count error types (the occurrence of an error is noted once) and
some count tokens (every example of the error is counted; e.g. Lennon 1991). Making
comparisons across studies is therefore unreliable and comparing error frequencies or generalising
the results is not a simple matter (Nickel 1989; Schachter & Celce-Murcia 1977).
12
2.2.3 Errors vs. Mistakes

At the level of analysis, deciding whether a deviation is an error or a mistake is another problem.
Corder (1967, 1971, 1981) contends that mistakes should not be included in the quantification
or analysis of errors and this is the approach taken by most analysts. Johnson (1988) believes that
mistakes can be corrected by the learner, but in practice determining whether a learner cannot
correct his/her own deviant utterances is very problematic (James 1998). Errors occur when the
learner does not know the rule and needs to be taught it or when the learner needs to be shown
that the wrong knowledge or partial knowledge has been applied to the particular situation
(Shaughnessy 1977). In a different view, Edge (1989) rejects this error-mistake classification and
calls all deviations from the norm mistakes. These mistakes include:
C slips, which are a result of “processing problems or carelessness” (Edge 1989:11);
C errors, which are comprehensible but which the learner is unable to correct, although the
form has been taught; and
C attempts, which are fairly incomprehensible and uncorrectable by the learner.
Snow (cited in James 1998) argues for two steps in error development. The first step is the
presence of errors which the learner does not recognise as errors, and the second step is the
presence of errors that the learner recognises as errors but which he/she cannot correct. The
mistake, where the learner is able to correct a wrong form, may be a third step. In other words,
mistakes are a performance problem rather than a competence problem (Corder 1967), rather like
the lapses made by L1 speakers (Johnson 1988). This performance-competence distinction is
maintained by most theorists in distinguishing errors from mistakes.
Another way of determining whether a deviant form should be classified as an error or a mistake
is to decide on the gravity of the error. In order to do this, James (1994:191) believes that
criteria for error gravity need to be established (e.g. “are lexical errors more serious than
grammatical?”), as well as who will judge the gravity (e.g. L1 teachers/L2 teachers/non-teachers).
An additional criterion is that errors have a lack of speaker intention, otherwise they may be
classified as deviances (James 1998). The classification of an utterance as deviant is further
confused by the distinction between unacceptability and ungrammaticality; e.g. a grammatical
utterance may be unacceptable because of non-linguistic factors (de Beaugrande & Dressler
1981). Acceptability is judged by use in a particular context, while grammaticality can be judged
by a native speaker of the language and a grammatical utterance is necessarily acceptable as well
13
(Lyons 1968).
2.2.4 Conclusion
Despite the above criticisms and methodological difficulties, there is evidence of a more positive
approach to EA in some recent writing. According to James (1998), there are two reasons for
the continued use of EA when investigating SLA data. Firstly, the empirical design is simple, with
a clear indication of an error if a particular norm is chosen. Secondly, teachers play this normative
role and encourage their students to achieve these target norms. This negative view of error may
be held by many teachers, but many SLA theorists tend to regard errors in a much more positive
way because they regard them as signs of creative hypothesis construction and testing.
In conclusion, James (1998) feels that those doing IL studies and those engaged in EA analyses
have different goals, the former concerned with developing a theory of acquisition and the latter
with pedagogic goals. However, like Cook (1993), he feels that EA is an effective way of dealing
with data in the absence of a suitable analytical framework in IL studies and it is for this reason
that an EA is carried out on the data in my study. An EA alone does not provide a sufficient
description or explanation of learner language, but it has a significant contribution to make as part
of an analysis of this type of language because it can offer insights into the sequence of
acquisition, the patterns of acquisition and the types of structures which learners find difficult.
2.3 MORPHEME STUDIES

2.3.1 Background
At the same time as studies in EA and L1 acquisition were flourishing, a series of studies was
conducted to see whether L2 development patterns conformed to those of L1 acquisition. The
morpheme studies, as they became known, were different from contrastive analysis (CA) and
early error analysis (EA), since researchers viewed learner language and “errors” positively. They
sought to discover the developmental patterns in learner language and to account for the learner’s
active participation in the construction of his/her communicative system. This focus on the learner
was congruent with Selinker’s (1972) views of IL which had been published at about the same
time.
More recently, morpheme studies have been linked to performance analysis (Long & Sato 1984)
14
because of their focus on performance rather than competence during data collection. They have
also been linked to creative construction, which Sharwood Smith (1994:49) says takes a “target-
oriented and incremental” approach. Second-language input provides “the database for L2
acquisition” and studies focus on the similarities between L1 and L2 development. Examples of
morpheme studies include Bailey et al. (1974), Dulay & Burt (1973, 1974a, 1974b), Fathman
(1978), Krashen et al. (1976), Larsen-Freeman (1975), Lightbown (1983) and Pica (1983).
The main findings of these studies include:

C the discovery of natural sequences of morphological development;
C a significant congruence between L1 and L2 development; and
C the description of developmental errors.
The focus of the studies is on developmental errors and, as a result, there is a minimal role
ascribed to input, the L1 and transfer (except as a performance phenomenon when the learner is
under stress to perform what is not known according to Sharwood Smith, 1994). Again, this is
an extreme position which needs modification in the light of later studies.
2.3.2 Criticisms of morpheme studies

There have been many criticisms of the morpheme studies, mostly based on methodological
weaknesses, but the findings have been reinforced repeatedly in other studies and it cannot be
disputed that they stimulated vast amounts of research on the learning of different languages.
With regard to criticisms of the studies, Lightbown (1984) feels that the results have been
overinterpreted because later studies have shown that input and the L1 also have a significant
influence on IL (e.g. White 1977). She also mentions methodological problems such as the
difficulty of replication because of the wide range of variables in any study or population group,
the failure to investigate the input that learners receive and its influence on IL development, the
small range of tasks making up the data, and the need for piloting and testing data-collection
instruments to ensure baseline data from speakers of the L1 and the L2. Early criticisms related
to the elicitation instrument (Bilingual Syntax Measure), because it was thought that the common
developmental sequences found were an artifact of the data-collection methods. However, later
studies refuted this criticism when researchers found the same orders using different instruments
(Long & Sato 1984).
15
Long & Sato (1984) discuss further limitations of the morpheme studies:
C morpheme use in obligatory contexts is the criterion used to show acquisition, but this
is not the same as mastery of the morpheme, because it ignores contexts where the
morpheme is used inappropriately;
C accurate use of a morpheme does not imply that the learner knows the function of the
morpheme;
C early development of morphemes in terms of sequence of appearance is not measured;
rather the sequence of accurate use is measured;
C a functional perspective on morpheme use is not taken, and thus no variation in
development is noted if the morpheme has a variety of functions (see also Sato 1990);
C avoidance of morphemes is not considered (see also Schachter 1974);
C only a small part of the language is studied, i.e. a few morphemes;
C the order is based on English so that it is not possible to generalise or even compare it to
other languages.
The studies are also usually cross-sectional rather than longitudinal (Andersen 1991), and thus
ordering of morphemes cannot be said to be absolute in the way that the morpheme studies said
they were1. Finally, individual variation in performance is not adequately reported because data
are grouped (Andersen 1977, 1978; Rosansky 1976).
Fathman (1978) reported that the structures which were analysed were based on studies of L1
development. As a result, structures which were unique to SLA may have been missed. Errors
with ambiguous sources may not have been accounted for in the data which may provide an
explanation for the low rate of L1 errors which was reported (White 1977). In short:
The morpheme order approach is useful for describing the overall sequence of target-like
use of distinct structural items as an accuracy measure (Larsen-Freeman 1978), but I
think it is not suited to capture various developmental regularities. Notice that the
criticism reviewed above did not question the morpheme order approach as a useful
notion to describe language acquisition. In Wode et al. 1978 we argue that the notion
itself is inadequate as a tool to provide insights into how language is learned, because by
focusing on target-like performance morpheme order studies necessarily miss all those
developments leading up to, and preceding, the final states of achievements. This includes
missing several types of data well known to occur in many different types of language
acquisition, including some quite prevalent in many different structural areas of L2
acquisition. (Wode 1981:64)
2.3.3 Conclusion
Andersen (1991) says that morpheme studies lost their relevance and were abandoned partly
1
Nevertheless, Kessler & Idar (1979) found that a longitudinal study of naturalistic acquisition of
English by a Vietnamese adult and child followed the sequences of the morpheme studies.
16
because of their inability to explain the process of how learners learnt an L2. They could be part
of a theoretical explanation of SLA, but they would never account for SLA adequately (Hatch
1979). On the other hand, they proved invaluable in starting the trend towards studying learner
language as systematic and showing active participation on the part of the learner in system-
building. In addition, notions such as natural sequences of development and developmental errors
are still widely accepted today and these two concepts will be addressed in my study.
2.4 NATURAL ROUTES

One of the most important findings of the morpheme studies was the existence of natural routes
of development in SLA. These routes usually refer to particular languages, but there are also
cross-linguistic generalisations which can be made, especially for the acquisition of Indo-
European languages; e.g. external negation is the first step in the acquisition of negation.
However, further research is needed to identify natural routes of development in languages which
are not part of the Indo-European family. It might be possible, for example, to find functional
commonalities rather than formal ones across languages.
2.4.1 Background
A fairly limited number of languages and morphemes have been investigated in search of
sequences of development. English as the L2 has been the main focus of most studies and there
has been little comparative work on other ILs, especially on ILs of non-Indo European languages.
The acquisition orders of negation (e.g. Hyltenstam 1977; Wode 1981), word order and
interrogatives (e.g. Carroll 1990; Huang cited in Sato 1990; Ravem cited in Towell & Hawkins
1994; Wagner-Gough 1975) have been studied extensively in English, German, and to a lesser
extent, Dutch. The native languages of the learners have tended to be Spanish, Italian, Chinese,
Turkish and Arabic (see Lalleman 1996 for an overview). So, although one can now investigate
and probably assume that there are natural sequences of acquisition in all languages, it is not clear
which morphemes (or functions) are likely to appear earlier and which later, if one were to
compare the development of different learner languages.
The earliest mention of natural routes of development is found in Corder (1967) who proposed
a built-in syllabus for learners. The morpheme studies explored and developed this point so that
Krashen (1981, 1982) included the natural order hypothesis in his model of SLA. Lalleman
17
(1996:20) concludes that:

(i)t is generally agreed upon that there are set orders of acquisition whereby certain
structures can be learned only after others have been acquired. The acquisition order of
a number of (unrelated) morphological features and (related) syntactic features has been
shown to be universal to a large extent (Krashen 1981, Dulay et al. 1982). Most
researchers, therefore, conclude that the available material has been sufficiently analyzed
to consider the ‘universal route of acquisition’ proven.
The sense in which the word ‘universal’ is used here is ambiguous. Studies which have shown
a universal route of acquisition which is independent of L1 background or learning situation for
learners of a particular language, support the claim of a universal acquisition for a language.
However, if the word ‘universal’ is used in the sense of a universal order of acquisition for all
languages, regardless of L1 background or the L2 to be learned, then this claim has definitely not
yet been proven. It is likely that there could be a universal sequence of development in a
functional sense, but the specific morphemes and syntactic patterns of different languages are
unlikely to be acquired in exactly the same order.
Most studies have focussed on the acquisition of forms and not of functions since they have
looked at particular languages rather than compared sequences cross-linguistically. In addition,
the languages which have been studied tend to belong to the Indo-European family and these
results cannot necessarily be generalised to languages with different typologies, such as the
agglutinative African languages (unless, perhaps, one considers a functional rather than a formal
view). For example, one could investigate whether the English acquisition sequence -ing, plural
-s, copula, auxiliary, article, irregular past, regular past, third-person singular agreement,
possessive (Krashen cited in Zobl 1995) holds true from a functional view, i.e. do learners
typically need to do the same kind of things with their new language and therefore need forms
which express the same kinds of functions? I would argue that this might be a more logical and
fruitful approach and it might provide explanations for typical early forms, especially if one could
assume that human beings are relatively similar in terms of how they process and use language
across the world (Felix 1984).
Several studies support a common developmental route for the learning of English by speakers
of different L1s. These include Ravem’s (cited in Towell & Hawkins 1994) study of a Norwegian
child, Cazden et al.’s (1975) study of Spanish speakers, Hakuta’s (1974, 1976) study of a
18
Japanese child, Wode’s (1981) study of German children, and Lightbown et al.’s (cited in
Allwright 1984) study of French children and adolescents. All of these studies found similar
orders which were also similar to other early morpheme studies.
2.4.2 Role of L1
In support of common routes of development, several studies conclude that the influence of the
L1 is seen mainly in the rate of development rather than the sequence of development. Firstly, the
typological proximity of the L1 to the L2 may affect the rate of acquisition. For example, Zobl
(1984) points out that a Swedish learner of English can make use of parsing routines similar to
those of Swedish, whereas a Turkish learner of English cannot do this as easily because of the
agglutinative structure of Turkish. Zobl (1984:95) concludes that the “degree of initial success
with which familiar parsing routines can be applied to the L2 may have a determining influence
on how far down on the scale of language differentiation the complexification process begins”.
Kellerman (1984) feels, however, that the influence of the L1 can work both ways because if the
developmental pattern is found in the L1 of the learner, the learner may either move more quickly
through that stage or may in fact fossilise at that stage (see also studies by Ravem cited in Towell
& Hawkins 1994; Schumann 1978; Wode 1978; Zobl cited in Odlin 1989).
When comparing the results of numerous studies of natural routes, it becomes apparent that there
is some controversy about whether there is a set order of acquisition since not all results show
exactly the same route of development. For example, a number of studies show the effect of the
L1 on acquisition orders (Hakuta 1976; Larsen-Freeman 1975; Mace-Matluck 1979; Schachter
1974). Schumann (1979) and Zobl (1980) argue strongly for substantial L1 influences on
development. Odlin (1989) shows that this view is supported by comparative studies of the
development of negation, where Spanish learners used no as the negative particle while Japanese
speakers used no and not. The choice of particles can be related to the negative formation rules
in the L1s of these learners (Stauble 1984). From a different perspective, Kellerman (cited in
Kellerman 1984) concludes that L1 influence can be seen clearly if one takes avoidance strategies
into account.
2.4.3 Influence of elicitation tasks

It also appears that different types of data yield slightly different results. Although natural
19
sequences have been observed in spontaneous speech, the order has been shown to be disrupted
if more formal language production is studied; e.g. translation, grammar exercises, written forms
(Ellis 1987b; Fuller cited in Allwright 1984; Krashen et al. 1976). Ghrib’s (1987) study of a
Tunisian learner of English attributes differences in the sequence of development to the mixture
of spontaneous and elicited data on which her study was based. For example, she finds the regular
past tense appears before the irregular in elicited speech but vice versa in spontaneous speech.
2.4.4 Influence of input

The type of input which the learners receive also seems to make a difference to the sequence of
development. Larsen-Freeman (1991:320) concludes that “a recurring finding was the correlation
between the frequency of certain forms in the input and their appearance in learners’ ILs”. In
addition, Sajavaara (cited in Allwright 1984) found differences in sequences of acquisition for
Finnish students who only had exposure to formal input. Lightbown (1984:246) feels that
disturbed orders in instruction-only contexts are a result of exposure to a “limited and distorted
version of [TL] - so distorted that it may be said to constitute a different ‘target-language’”.
Eubank (cited in Zobl 1995) and Weinert (1987) also found disturbed orders with distorted input
and explicit instruction as the input. Wode (1981:305) argues that learners in natural situations
follow a strict sequence of development, but that learners in instructed environments have
developmental orders “characterised by rather loose chronological ordering” (although error
types are the same).
Lightbown (1983) shows that accuracy on the plural was worse than on the auxiliary for French
learners in Canada. This finding led her to conclude that although the natural order was not
confirmed it was also not disconfirmed, and that it might be better not to posit too strict an order.
Allwright (1984:213) agrees with this conclusion and feels that although ordering may not be
invariant, it is feasible to posit a “reliable and stable ... sequence”. Even more tentatively,
Andersen (1978) prefers to use implicational hierarchies and says that there is no absolute route
but that one can discuss tendencies, i.e. some morphemes appear earlier and some appear later.
This is not a particularly helpful observation, as it is rather general.
Nevertheless, several writers show that formal instruction has no effect on natural orders,
regardless of whether learners are in naturalistic or formal situations and regardless of the L1 of
20
the learners (Bailey et al. 1974; Ellis 1984, 1989; Felix 1981; Krashen et al. 1977; Lalleman 1996;
Meisel et al. 1981; Perkins & Larsen-Freeman 1975; Pienemann 1984; Towell & Hawkins 1994).
To explain this, Felix (1981:110) argues that instructed learners follow the same route as
naturalistic learners because when the formal environment demands that they produce the
structure, they have to rely on the same natural processes of development and “natural acquisition
principles” that naturalistic learners use. Krashen (1985) accounts for similarities between the
language of learners in formal and naturalistic situations with his non-interface position. In this
view, formal instruction has no influence on the acquisition of a language because formal learning
and unconscious acquisition are entirely unrelated and the two knowledge representations never
interact. What the learner produces is therefore a result of acquisition rather than formal
instruction. Providing a further two possible explanations, Allwright (1984) suggests that the
natural syllabus may override what is available in the classroom, or possibly classroom discourse
is not particularly different from the input received in the natural environment.
2.4.5 Conclusion
Providing an explanation for the natural order which has been found has proved problematic.
Doughty (1991:436) studied the acquisition of relative clause types after formal instruction and
concluded that natural sequences are a result of a “universal ordering of difficulty, although
exactly what constitutes difficulty has not yet been clearly established”. She relates difficulty to
psycholinguistic processing and concludes that markedness also affects acquisition once the
learner is ready to acquire a particular structure. Zobl (1995) concludes that a mixture of
functional categories and Universal Grammar (UG) provides an explanation for natural orders.
Despite the limitations discussed above, the discovery of natural routes of acquisition was
important for a number of reasons. Firstly, it substantiated the idea that acquisition was regular
and reinforced the notion of systematicity in learner language. Secondly, its value lay in the
contribution it could make to teaching, since language-learning materials could be graded and
presented according to the sequence of development. Thirdly, it showed that formal instruction
probably had less effect on acquisition than previously believed, which meant that studies of
learners in formal and informal learning conditions could be compared.
21
2.5 INTERLANGUAGE
2.5.1 Introduction
The term interlanguage (IL), which refers to the language produced by learners acquiring an
additional language, was coined by Selinker (1972), although studies of the IL phenomenon and
the notion that learner language should be studied as an autonomous system predate the term by
several years (e.g. Corder 1967). Some writers (e.g. Lalleman 1996) classify IL as part of the
Creative Construction Hypothesis (CCH) paradigm which has the central tenet that the learner
is an active participant in the learning process and produces a system which is not too different
from other natural languages (Long & Sato 1984). On the other hand, Sharwood Smith (1994)
argues against IL falling under the CCH because he says this branch of research was not
concerned with interim grammars but rather with finding common patterns of development.
2.5.2 Key concepts in interlanguage

The original formulations in Selinker’s (1972) seminal paper include:
C that the learner moves through a series of intermediate stages from the L1 to the L2;
C that the learner’s aim is to move from the linguistic system of the IL to the L2 system;
C that the output of the learner is not describable in terms of the linguistic units of the L1
and/or the L2; and
C that 95% of learners never actually achieve the L2 system.
An IL is an independent, structured linguistic system which includes both errors and non-errors
(cf. EA which looked only at errors) produced by a learner trying to produce a target language
(TL). An IL is idiosyncratic to a learner, but as discussed above in section 2.4, there is a fairly
common range of IL structures and patterns used by learners with the same L1 and different L1s.
Corder’s (1971) use of an alternative term for IL, idiosyncratic dialect, emphasizes that
individual learners do not share the same ILs, although research showing commonalities across
learners may make it feasible to talk of “mini-dialects” for a group of learners in the same class
(James 1998; Sharwood Smith 1994). Input, which is fairly similar across formal contexts
because of common classroom discourse patterns, may also lead to similarities in IL (Allwright
1984), and so common development may result from similar environmental factors (Snow &
Ferguson cited in Felix 1984).
Nemser (1971) uses the term approximative system for IL to show that the learner moves closer
and closer towards the TL as he/she processes more and more of the TL system and Corder’s
22
(1967) use of transitional competence has a similar focus on movement from L1 to L2. The
notion of movement from L1 to L2 is controversial and Corder (cited in van Els et al. 1984:69)
says that movement should not be seen as movement from one language to another but “as a
movement through a series of increasingly complex stages”. Both the terms approximative system
and transitional competence have been rejected by some because of their connotations of
comparison between IL and TL, although Nemser’s discussion includes the point that IL is
distinct from the L1 and the L2 (as Selinker and others argue).
A further point of difference is that Nemser (1971:116) defined the learner-language system as
a “deviant” form of the TL, which is not a view which Selinker holds. Selinker (1972) strongly
rejects the notion that IL should be compared to the TL and insists that IL is a system in its own
right. Nevertheless, he maintains the view that there is movement towards a target and James
(1998:17) also suggests that “as long as FL [foreign language] learners are prepared to see and
call themselves learners, the assumption that they wish to conform is surely a reasonable one to
make”. He believes that learners themselves are comparing their language to the TL so that some
form of comparison is necessary (see also Jake 1998 for a similar view). This is a view not
strongly supported by IL theorists, but one practised widely in the analysis of data, even when
functional approaches are used.
Regarding IL as a system proved to be a good theoretical concept, but actually analysing the data
in this fashion has proved highly problematic. James (1994) says that the first step in an analysis
is to describe the IL. However, later it is also necessary to compare systems (L1, L2, IL) in order
to reach the desired level of explanation. Perdue and his colleagues (1993a) took a step in the
right direction when they analysed the data from their cross-linguistic study of learner language
as a system in its own right (see section 2.6.1). However, they too admit to using descriptive
categories based on other established languages and they were also unable to avoid completely
the “trap” of comparisons between the IL and the TL (Klein & Perdue 1997). In defence, they
say that if one is to regard learner language as a natural language (conforming to UG rules with
internal consistency), which many do (e.g. Adjémian 1976; Pinker 1984), then one cannot expect
to find a whole new set of categories which apply only to learner languages (see also Pienemann
1992). Granted, one may find that learners do not use the categories in a predictable way (e.g.
they may have different functional purposes when they use the categories and they may have split
23
the categories in a non-target-like and non-source-like way), but the basic categories are still
available when they learn the language. It is the unique function-form match which the researcher
should be interested in rather than attempting to discover entirely new categories for these
languages. James (1998:94) makes pertinent points when he asks:
So is it ever justifiable to describe learners’ errors in terms appropriate to, or even derived
from, the TL? It would be if the two codes were closely related, or ‘cognate’, if they were
two dialects of the same language for example. Now Corder (1971) called the learner’s
version of the TL their ‘idiosyncratic dialect’. This must imply that the learner’s and the
NSs’ [native speakers] codes are dialects of the same language, the first ‘idiosyncratic’
and incomplete, the NSs’, by contrast, a complete and a social rather than an
idiosyncratic, private one. Being co-dialects of the same language, they should be
describable in terms of the same grammar.
A strong TL comparison approach seems to suppose that all learners can and will eventually
become indistinguishable from L1 speakers, a notion which Selinker & Lamendella (1978) and
Scovel (1988) reject. Nevertheless, Selinker (1992) notes that some learners can manage to
appear to be native-like in some conditions and situations and often control their performance by
avoiding certain constructions and types of talk. The lack of complete development of the TL is
described as fossilisation. Parts of the system appear to be impermeable to outside influence and
new hypotheses. A number of reasons have been cited for this phenomenon and they include:
C the notion that the learner learns as much as is necessary to communicate (Klein &
Perdue 1997);
C phonological fossilisation interferes with morphological development (Sato 1990);
C non-target-like forms are automatised (Hulstijn 1989);
C there are problems with parameter-setting in a UG view (Hale cited in Nakuma 1998);
C the L1 interferes in the form of negative transfer (Selinker 1992; Selinker & Lakshamanan
1992); and/or
C motivation levels are not high enough (Nakuma 1998).
Adjémian (1976) contributed to the concept of IL by emphasizing its permeability to influence

from the L1. Adjémian is comparing L1 and L2 development here and shows that L2
development may show signs of L1 features which the development of L1 logically cannot show
because the L1 learner does not have another language to draw on. Permeability to influence
from the L1 may be problematic with regard to the notion of systematicity, as it seems to be
acceptable to compare the L1 and the IL, but not the IL and the L2. One point in favour here may
be that the learner knows the L1 and therefore incorporates it into the system, whereas he/she
does not know the L2 so that it cannot be incorporated properly into the system. However, it
24
might also be possible to say that the IL is influenced by the L2. This leaves two possibilities with
regard to the structure of the IL. Either, there are two sources of influence (L1 and L2) and a
separate system called the IL; or there are two sources of influence (L1 and L2) which combine
to make up the new system. Selinker (1992) seems to argue for the former approach although
he does not place much emphasis on the influence of the L1, regarding it as only one of several
learner strategies for learning the new language. A further option which he explores is to search
for a system of interlingual identifications (IIs) which form a system of comparison at a
subconscious level (see section 2.1.3). This would give the L1 and the L2 similar low levels of
influence in the formation of the IL. If there was a universal system of IIs, one could then say that
the L1 and the L2 have very little influence. However, if the system of IIs is between the L1 and
the TL then one returns to the contrastive problem of comparing two languages, albeit at the
deep level, and both languages have a major role to play in the formation of the IL. Unfortunately
(as discussed in section 2.1.3), an adequate theory of IIs does not exist and the preceding
questions cannot yet be answered.
2.5.3 Composition of the IL system

Bialystok & Sharwood Smith (1985) point out that there are two ways of viewing IL. On the one
hand, it can be seen as the product of a set of highly structured hypotheses making up the
underlying competence. The product is revealed in performance. This product can be studied as
a diachronic series of systems which develop as second-language acquisition (SLA) progresses,
or as a language system at a particular point in time (Cooreman & Kilborn 1991). On the other
hand, IL can be seen as the system of underlying competence which needs to be investigated in
terms of the psychological processes at work. It seems then that both the product and the system
need to be investigated.
There are different views of the “composition of the IL system” (Bialystok & Sharwood Smith
1985:102-103). Some, like Selinker, see it as a “single system composed of rules which have been
developed via different processes”, while others, like Adjémian (1976), see the system as
composed of “a combination of separate knowledge sources” consisting of the L1 and an L2-
based system. Robert (1989:219-220) views a language as being composed of two aspects: a
“computational system”, which is the set of rules for structuring syntax, phonology and
semantics, and a “conceptual system” which relates to thematic relations. Whereas fully-fledged
25
languages make use of both systems to a great extent, it is claimed that IL users rely on the
conceptual system more than on the computational system. The conceptual system is more
universal whereas the computational system would be language-specific.
Bialystok & Sharwood Smith (1985) argue instead for a knowledge component (underlying
linguistic system) and a control component (retrieval procedures) in the description of IL (see
also Cook cited in Towell 1987). The control component is the set of retrieval procedures used
to access the underlying linguistic system. Either or both of these components need to be
considered in the analysis of IL data and may be different from or similar to those of the native
speaker (see also Towell 1987). Their definition of IL is therefore:
... the systematic language performance (in production and recognition of utterances) by
second-language learners who have not achieved sufficient levels of analysis of linguistic
knowledge or control of processing to be identified completely with native speakers.
(Bialystok & Sharwood Smith 1985:116)
Second-language acquisition development in this view requires increased knowledge and/or

better control.
2.5.3.1 Universal elements

The IL may also consist of a universal component. Despite extensive writing in the area (e.g.
Cook 1996; DeKydtspotter et al. 1998; Felix 1995; Flynn 1991; Gass 1984b, 1995; Gregg 1989;
Kanno 1998; Kaplan 1998; Lardiere 1998; Schachter 1992; Schwartz & Sprouse 1994;
Sharwood Smith 1994; White 1992, 1998), theorists cannot agree if IL is a natural language
which is UG-based. It is possible, if interlingual identifications (IIs) are not based on UG (which
might be the case if UG is unavailable to older learners), that the new system is entirely separate
from the native language (NL) and the TL. On the other hand, another argument is that the
system is partially separate, has to be based on the NL and the TL to some extent because of
transfer, is linked via IIs and should be investigable in a UG framework (e.g. White cited in
Selinker 1992). A further possibility is that IL develops from UG with a “learning mechanism
which incorporates a theory of markedness” (Zobl 1992:176).
2.5.3.2 Formulaic elements

Formulaic utterances are a major component of the IL system. They serve a vitally important
26
communicative function and may be unanalysed chunks of language at early stages (Altman 1997;
Klein 1986). For example, Altman (1997) points out that inflected verbs may be produced as
unanalyzed wholes but this does not mean that the learner can use the rules productively to form
other inflected verbs. Apart from their communicative function, Fillmore (cited in Myles et al.
1998) suggests that formulaic expressions may also be useful data on which language learning
is based because they can be used as frames for further development. Similarly, Myles et al.
(1998, 1999) who studied the development of pronominal systems, conclude that formulaic
expressions aid communication in the early stages and that they are used as the basis of
hypothesis testing. Hakuta’s (1974) discussion of formulaic expressions shows that learners
perceive formulaic expressions as units, both syntactically and semantically.
2.5.3.3 Conclusion
Towell & Hawkins (1994) discuss the different origins of the elements of IL and conclude that
some of them have UG origins, some have transfer origins, some are formulaic, and some are a
result of learned linguistic knowledge. It appears then that there is a knowledge system which
may consist of hypotheses about the TL, universal elements such as a conceptual system relating
to thematic relations, and formulaic utterances. In addition, there is a control system which
consists of procedures to access the knowledge system. The product of the knowledge and
control systems will be discussed in the next section.
2.5.4 Features of interlanguage

There have been a variety of attempts at describing IL features and comparisons have been made
between IL, child language and pidgins, because some features appear to be similar (Andersen
1983). Kachru (cited in Williams 1989) found that many features of non-native institutionalized
varieties of English (NIVES) can be found in learner language and suggests that they have their
roots in individual SLA. Theorists comparing IL to “reduced” forms of language like child
language and pidgins see it as “a variety of language which is both formally and communicatively
reduced when compared to languages used as native languages by adults” (Faerch et al.
1984:271). However, equating IL with other non-native forms of language has met with criticism
because some (e.g. Bickerton 1983; Myhill 1991) believe that the social contexts in which each
occur are incomparable, although Adamson (1989) suggests that a common bioprogram guides
development in these different contexts. Mitchell & Myles (1998) also conclude that pidgins and
27
ILs do not share all their syntactic features and that pidgin systems are more stable than most ILs.
Nevertheless, Givón (1979) provides a framework which describes the features of different forms
of language, in which he distinguishes a pragmatic mode of communication which occurs in
“reduced” languages and a syntactic mode of communication which occurs in fully-developed
languages. Table 2.1 presents some of the features of the two modes.
Table 2.1 Features of the pragmatic and syntactic mode

Pragmatic mode Syntactic mode
1. Topic-comment structure 1. Subject-predicate structure

2. Loose coordination 2. Tight coordination
3. Slow rate of delivery 3. Fast rate of delivery
4. Small chunks under one intonation contour 4. Large chunks under one intonation
contour
5. Lower noun/verb ratio in discourse 5. Higher noun/verb ratio
6. More simple verbs 6. More complex verbs
7. No use of grammatical morphology 7. Extensive use of grammatical
morphology
(Adapted from Givón 1979:98)
Sato (1988) reports that only the lack of grammatical IL morphology has been studied extensively
and that the evidence of a low noun/verb ratio is not conclusive, although Perdue’s (1993a,b)
study supports both lack of morphology and low noun/verb rations. Pienemann (1992) believes
that ILs have a common lexicon, part of the rule system of the TL and a set of idiosyncratic rules.
However, these aspects have not been fully explored and a full set of lexical items and rules is not
available. A more detailed account of the features of IL (or “simple codes”) is supplied by Corder
(1977:2):
... a simple or virtually non-existent morphological system, a more-or-less fixed word
order, a simple personal pronoun system, a small number of grammatical function words
and grammatical categories, little or no use of the copula, absence of an article system
(less often the absence of deictic words). The semantic functions of these and other
systematic systems such as tense and aspect are typically performed, when at all, by
lexical means, e.g. adverbs, or some “imperial form”. The basic syntactic relations are
expressed by word order.
The IL does not therefore consist of the same phonological, morphological and syntactic
categories of the target language (TL) or the source language (SL), but it is still unclear to what
degree the system is a simplified version of fully-fledged languages and to what degree it is based
28
on universal categories.
2.5.5 Interlanguage development

2.5.5.1 General development
The initial hypothesis (starting point) of learner language is controversial in SLA theory.
Contrastive analysts believed that the L1 is the starting point with successive restructuring until
the learner’s language approximates the TL (Fries 1945, Lado 1957). Development occurs when
the basic system is elaborated by incorporating the rules of the TL. Corder (cited in Gass 1984a)
believes that this is the correct position for phonology, but Zobl (1984:85) criticises this
perspective because he feels that “(i)n restructuring distinctions are lost and subsequently
elaborated along different lines ... unmarking or despecification of the L1 grammar would have
to take place for any movement to occur in the direction of the target”. He also refutes the idea
that the L1 in its full form is the starting point of IL because of empirical data showing that
certain structures that one would expect to be transferred because of their equivalent positions
in the L2 are in fact not transferred.
Zobl (1984) joins others (e.g. Sharwood Smith & Rutherford cited in Selinker 1992) in believing
that the initial hypothesis is an independent system (core grammar, IL base) which resides in the
learner, especially for syntax (Corder cited in Gass 1984a). This core grammar might be made
up of universal rules of human language (Selinker 1984, 1992) and, according to Corder (cited
in Gass 1984a), it is based on the system used when the L1 developed. This may be akin to the
notion of Universal Grammar (UG), although Corder does not commit himself on this point. His
1977 paper suggests that semantic aspects such as agency, animacy and spatial location may be
sufficient to analyse IL data. He says that the common features found in the ILs of learners with
diverse L1 backgrounds may be due to a return to the basic system, rather than a need to use the
L1 to any great extent. The pull towards the basic code is therefore stronger than any L1
influences. However, Selinker (1992:33-34) feels that a combination of approaches is the correct
one:
My hypothesis is that the NL is part of where the L2 learner has to be on day one of
exposure to input from the TL, because of the pervasive reality of language transfer and,
therefore, interlingual identifications. But it cannot be the entire starting point because
of the reality of early fossilization of non-L1-like structures.
29
2.5.5.2 Simplification
Simple systems and simplification processes are controversial issues in IL writing. The core
system is a simple system (but not a simplified version of the NL or the TL because the learner
does not know the TL, according to Valdman, 1977). This simple system is elaborated as the
learner learns more about the TL, with the possibility that development is arrested at some point,
resulting in fossilization. Explaining the simplification which has taken place in simple codes,
Corder (1977) discusses two views of simplification. The first is that IL appears simple because
it is compared to a native language (NL), which means that ILs are described in terms of the
complexities of these fully-fledged languages. This is the view taken by Silva-Corvalán
(1991:330) when she says linguistic simplification includes “reduction of the inventory of
linguistic forms, semantic range, or language functions, and the elimination of alternative
structures at certain levels”. She also sees it as generalisation or overgeneralisation (see also
Preston cited in Silva-Corvalán 1991), which involves expanding the use of one form at the
expense of another form, resulting in a loss of variety in the forms used. In SLA, forms are not
being lost, but the overgeneralisation of a form is the starting point with progressive acquisition
of further forms in a process which is the mirror-image of language loss. Simplification in ILs may
therefore be the result of universal simplification processes, which removes the link to a specific
native language.
The second way of looking at simplification, which Corder (1977) also examines, is that NLs are
complex forms of simple codes and that there are language-specific complexification strategies
or rules (see also Klein & Perdue 1997 for a similar conclusion). According to Baker (1979),
from a learnability perspective, the idea of a simple code which is elaborated is a better one.
2.5.5.3 Development of the interlanguage system

When discussing further development of the IL system, it is essential to remember that internal
systematicity is a central principle of IL. Ellis (1985a) points out that IL is systematic from both
horizontal (synchronic) and vertical (diachronic) perspectives. Although it does not necessarily
detract from the systematic nature of IL, it is also necessary to posit a degree of variability
(Dickerson 1975; Huebner 1979). However, Sharwood Smith (1994) sees systematicity and
variability as incompatible, and he argues that absolute systematicity and independence of the
system has to be an idealisation because of the dynamic quality of the learner’s IL.
30
Development may be viewed in two ways: either as leaps from stage to stage, or as the gradual
spread of a rule to different areas and contexts of the IL. Both views lead to a role for variability
which, although it may be random at times, is more often contextually derived, with influence
from the L1 (permeability) and internal development having an effect on the choices made. The
task may also have an effect on variability (Tarone 1982, 1983, 1989). Any analysis of IL
therefore requires an account of situational, contextual and linguistic factors as well as their
interaction, as these factors may influence the product of the system (Sato 1988, 1990).
The focus on L1 and developmental errors is a combination of a contrastive analysis (CA)

approach, cognitive processes (Gass 1979; Kellerman 1977, 1979; Sharwood Smith 1979), and
developmental processes (Andersen 1983). Learning happens gradually as the process involves
learning forms, meanings and functions (Larsen-Freeman 1991) while rules change over time
(Ellis 1987a, b, c; Selinker & Douglas 1985). Rules need to cover all aspects of language from
the phonological to the discoursal levels (Klein 1991), and these rules must allow for interaction,
since each level of language is not learnt in an isolated fashion. Form-meaning relationships need
to be developed since
... there are also forms whose relationship with meaning is difficult to access in the L2.
These forms carry little semantic weight or have little perceptual salience, or the form-
meaning relationship may be difficult to grasp. ... Learners also need data as they
construct or set their interlanguage. They need to know how their interlanguage differs
from the L2. It might be said that they need to know what is ungrammatical, but since
interlanguage is systematic and, therefore, grammatical in its own way, one might simply
say that learners need to know what in their interlanguage is inconsistent with the L2.
Finally, learners need to have data on the potential of their interlanguage for expressing
relationships of form and meaning as well as the extent to which they can modify and
restructure their interlanguage toward L2 morphosyntax. (Pica 1998:11)
As new knowledge is gained and integrated into the underlying system, the functions of other
items in the system are narrowed or broadened and elements of the system are restructured or
rejected. Interlanguage does not develop in a linear fashion but is recursive and continually
restructured (Corder 1992). Carroll (1984) finds in her study of English learners of German that
IL data shows periods of focus where there is a disproportionate number of one type of structure
which has been newly acquired at the expense of other forms which may have been used earlier.
The notion of staged development from IL1 to ILn might not be a feasible one, as restructuring
may not necessarily move the learner forward but may in fact move him/her further away from
31
the TL if an incorrect hypothesis is made. Alternatively, it may create confusion, so that the
learner temporarily reverts to an earlier stage (backsliding). In addition, the degree of variation
apparent in an IL at any one time may preclude the notion of discretely staged development.
Corder (1992) prefers to use this model for phonological development, but he does not see this
as necessarily feasible in the case of other subsystems of language, partly because of the lack of
clarity about the starting point of IL. As a result, it is not always possible to predict that learners
will progress in exactly the same way and improve their competence in a particular element of
the system. For example, Clahsen (1995), discussing the development of German plurals, found
that although some learners increased their accuracy over time, some learners stayed at about the
same rate of accuracy even though other elements of their grammar did develop. Over the two-
year period, none of the learners he investigated reached 100% accuracy.
Fillmore (cited in Johnson 1992) sees L2 development as comprising three stages:

C gesturing without verbal communication;
C communication which does not focus on correctness of form; and
C a concern for form.
Cazden (1968) offers a different classification into four periods in L1 acquisition:

C no inflection is apparent;
C formulaic utterances are produced (there are no errors but little communication beyond
the set of known utterances);
C much communication with many errors and overgeneralizations;
C 90% correct use of a form.
These L1 acquisition stages seem to be true of L2 development as well (Klein 1986). With regard
to the grammaticalisation of utterances, Skiba & Dittmar (1992) conclude that the first stage is
where learners put words next to each other but they are not explicitly related. The second stage
occurs when syntax starts to develop and relations between words are shown more explicitly. The
third stage is where the IL and TL start to converge, and syntactic and morphological relations
are fairly target-like. Bardovi-Harlig & Bofman (1989) conclude that advanced learners are better
at syntax than at morphology because they focus on what is communicatively important rather
than on what is less important (and often redundant) to communication.
2.5.5.3.1 Strategies for development

The learner makes use of a number of strategies for accessing and analysing input data and
32
producing output. Wolf Quintero (cited in Towell & Hawkins 1994:49) provides a number of
strategies for analysing and producing language:
* conservatism: ‘the initial hypothesis will be the most conservative possible ... even if
a learner notices complexity in the input data’ ...
* continuity: ‘a preference for items that combine to be adjacent’ (p. 44)
* uniqueness: an ‘initial preference for one-to-one correspondences between forms and
their meaning’ (p. 45)
* cumulative development: ‘development must proceed in stages and ... each stage will
contain the previous stage plus something more’ (p. 45)
* generalisation: ‘avoid exceptions’ (p. 45)
* pre-emption: ‘when a structure is generalised to related lexical items without direct
evidence from the input, the hypothesis will be noted as tentative ... If there is never
confirming evidence, the hypothesis will be lost’ (p. 46)
Apart from these strategies, Hatch (1978) feels that development is a result of interaction so that
forms and functions come to be matched over time and the basis for development is discourse.
Pfaff (1987:100) notes that “(d)iscourse functions develop before grammatical functions ...” (see
also Huebner 1983; Jordens cited in Cooreman & Kilborn 1991).
Fillmore (cited in Myles et al. 1999) and Nattinger & DeCarrico (1992) take a different view,
suggesting rather that formulaic utterances form the basis of grammatical development. Formulaic
utterances contain complex grammatical structures which are available to the learner for analysis.
Ellis (1996) agrees with this view and, in addition, he concludes that memorizing formulaic or
non-formulaic sequences of language forms a major part of language learning. These sequences
move from short-term memory to long-term memory, where they are analysed implicitly for
regularities (Ellis & Schmidt 1997). In contrast, Krashen & Scarcella (1978) maintain a non-
interface position between learned knowledge and acquired knowledge. Zobl (1979) rejects this
view based on evidence from creative construction strategies. Wode (1981) also rejects
Fillmore’s (cited in Wode 1981) position because it does not explain how elements which do not
occur linearly are acquired (e.g. phonological and semantic features, discourse structure rules).
Furthermore, there are developmental errors in the speech of a learner which cannot be attributed
to any formulaic expressions, the role of the L1 is not explained and formulaic utterances may
be altered at later stages to conform to the developmental level at that time.
Schachter (1992) points out that many theorists have accepted the view that language learning
involves a process of hypothesis-making and hypothesis-testing, although it is not clear in what
33
way learners use negative and positive evidence or seek disconfirmation for their hypotheses.
Fodor & Crain (cited in Zobl 1992) feel that the conservativeness of the learning process is a
result of hypothesis-testing which relies on the input and unmarked versions of a grammar. Less
conservative development only occurs once marked versions of a structure are noticed in the
input. Beck et al. (1995) and Schwartz (1993) argue that positive evidence is the main factor in
L2 grammar development. Although they do not rule out a role for negative evidence in SLA,
they do not believe that it plays a role as evidence in constructing L2 grammar systems.
Schachter (1992) says that hypotheses are formulated on the basis of experience with the
language and input and that the formulation-testing-acceptance/rejection process is cyclical and
continuous, with the learner focussing at different times on different hypotheses.
Common to both versions of the cognitive approach is the idea that L2 learners initially
decode, analyse, store and produce - i.e. process - material from the new language in
ways which are determined by general cognitive factors like the ‘perceptual saliency’ of
the material, the ‘continuity’ of elements in that material, the basic ‘conservatism’ of
learners in not extending hypotheses to domains not warranted by the input. This
approach considers that people perceive events in terms of ‘actors’, ‘actions’ and
‘persons or things acted upon’, and that these are more ‘salient’ than the place where the
event took place, or the time it took place, or the manner in which it took place. By
extension it is considered that L2 learners will attend to and acquire new ways of
expressing ‘actors’, ‘actions’ and ‘people or things acted upon’ before they will attend
to and acquire adverbials dealing with the place, time and manner of the event. (Towell
& Hawkins 1994:46)
Tarone (1988) and Ellis (1992) believe that noticing is the crucial aspect for inclusion of a new
structure in the grammar. Ellis (1992) sees development spreading according to tasks or contexts.
However, Towell & Hawkins (1994) feel that this view implies a random entry point by the
learner rather than an entry point which can be related to the L1. They cite as evidence studies
of learners of English and French which show that these learners differ systematically in where
they place the object pronoun, with French learners easily learning to place the pronoun
postverbally in English, but English learners going through a stage of placing the French pronoun
postverbally rather than preverbally (the same sequence is found for English learners of Spanish
according to Andersen, 1991).
This section has discussed the general features of IL development which seem to result from a
series of cycles of hypothesis-making and hypothesis-testing, possibly limited by a universal set
of constraints on what kind of hypotheses can be made. The performance displayed by the learner
34
is subject to variability according to task and level of development. As this section has focussed
mainly on the general features of development, the next section moves on to a discussion of
morphological development in IL, an area which is particularly relevant to my study.
2.5.5.3.2 Morphological development

Lack of morphology seems to be one of the main characteristics of IL and Schumann (1982:338)
explains the logical problem of explaining lack of morphology in the following way:
Absence of morphology is the aspect of simplification that is most problematic because,
as many researchers have pointed out, simplification implies having something and then
eliminating it. If learners do not supply morphology it is probably because they do not
know it, not because they choose not to use it. But I regard lack of morphology as
simplification because learners do not process it in the input, and therefore, it is not
available in their output. Thus absence of morphology results from a processing
constraint which leads to product level utterances which are morphologically simple in
comparison to well formed TL speech.
Bybee (1985, 1991) discusses the acquisition of morphology based on a variety of different
languages in terms of order and process. Although many studies focus on L1 acquisition,
VanPatten (1984) found that L1 and L2 morphological acquisition is almost identical. Suzman’s
(1982, 1999) studies of the acquisition of Zulu also confirm similarities between L1 and L2
acquisition. Bybee (1991:80) presents a hierarchy of morphemes which are ordered from most
semantically relevant to least semantically relevant in the verb (aspect, tense, mood, number,
person), a hierarchy upheld by several studies (e.g. Antinucci & Miller 1976 for L1 acquisition;
Kumpf 1984; Nixon cited in Andersen 1991; Paradis et al. 1998 for L2 acquisition).
The following acquisition patterns are also found:

C case-marking is learnt before less semantically motivated morphemes such as gender
(Pfaff 1987);
C basic forms are selected on the basis of markedness or semantic simplicity and/or
frequency of use (Bybee 1991);
C more relevant functors are placed closer to the lexical item so that semantically important
morphemes are more likely to be added to the base than morphemes which do not have
such important semantic functions (Slobin 1985);
C “free forms are acquired before bound forms” (Wode 1981:306); and
C verbal morphology develops before substantive morphology (Connors 1988).
Wode (1981) generalises from data on the acquisition of the plural and says that the rule for an
inflection proceeds in the following way: it will first be used with stems that take the inflection
35
in the target language, while other stems will not be inflected. The rule will then spread within
the appropriate class of stems. Newly acquired items may continue to be left uninflected, even
if they belong to the set in question. Next there will be generalizations to include some of the
remaining stem classes in non-target-like ways. At first, only one inflected form is used in all
contexts. The second stage occurs when the learner learns more than one form of the inflection,
but words are produced as lexical items with no awareness that there are two parts to the lexical
item. Only later is a productive rule used which is when the use of correctly formed irregular
items falls away as all forms are regularised. The final stage is correct use of irregular and regular
forms, but this stage may not be reached by many learners.
With regard to pronominal development, Felix (1981) records the confusion of pronouns in the
early stages of learning (see also Fillmore and Wode, both cited in Felix 1981). He finds a
preference for the use of the proper noun rather than the pronoun during spontaneous
conversation (see also Broeder 1991). Myles et al.’s (1999) study of French pronoun
development shows development of the third-person pronoun based on formulaic chunks which
included the first-person pronoun. Initially, the chunk is used inappropriately and the use of je
(I) is overextended. This is followed by overextension of the chunk which includes je (I) with the
addition of a lexical NP. The third stage is where the pronoun is omitted and replaced by a NP,
and the final stage is where the third-person pronoun is used correctly.
Agreement is a morpheme which may have little semantic impact on utterances and these forms
are learnt late. Learners of Brazilian Portuguese initially use the third-person agreement
morpheme (Simoes & Stoel-Gammon cited in Bybee 1991), although this is a later form for
learners of English. A further stage occurs when first-person agreement is added to an inflected
form, so that double morphological marking occurs for learners of Brazilian Portuguese and
Polish (Simoes & Stoel-Gammon and Smoczynska, both cited in Bybee 1991). In a discussion
of the use of the third-person agreement marker in English L2, Makoni (1996) notes that
redundancy does not necessarily stop the learner from using the marker as some have argued
before. Rather, the presence of a pronoun requiring the agreement marker triggers the use of the
marker more often than zero anaphor does (see also Young 1988, 1991 for similar findings).
Carroll’s (1999) study of absolute beginner learners of the French gender system found that
36
learners learnt natural gender most easily and that a variety of other factors influenced the
learning of non-natural gender marking. Phonological cues are not particularly useful for learning
gender but semantic patterning which enables the learner to map conceptual categories to
morphosyntactic categories aids learning. These learners all made generalisations about gender
patterning which were not related to their L1 or to the TL at times.
Krashen (1981) and Seliger (1979) argue that where there is simple one-to-one mapping of form
to function, the morpheme will be easy to learn, but where there are a variety of forms for a
function or a variety of functions for a form, learners will find these morphemes more difficult
to acquire. African languages of the Bantu family make extensive use of morphology, since they
have an agglutinative structure and the range of forms and functions is usually wider than the
range found in the morphology of Indo-European languages. In fact, Slobin (1982a, b) concludes
that L1 learners of agglutinative languages may learn inflections earlier than learners of non-
fusional languages because of the important role played by inflections in these languages.
In order for L2 learners to cope with the multiplicity of morphological forms and functions in
Swahili (an African language of the Bantu family), Musau (1995) identifies the use of three types
of strategies (see also Faerch & Kasper 1983). These include:
C functional reduction which is shown by “topic avoidance”, “message abandonment” or
“meaning replacement”;
C formal reduction where a simple system is utilised to avoid errors; and
C compensatory strategies involving “code-switching, interlingual transfer, or
overgeneralization (Musau 1995:298).
He found that learners of Swahili overgeneralised a few concords (classes 9 and 10) to all classes
of nouns, possibly because they may be the most productive (Bokamba 1977) and neutral (Besha
cited in Musau 1995). This pattern has also been seen in learners whose L1 has a similar class
system, learners who do not have a similar class system and in L1 learners of other Bantu
languages (Tsonope cited in Musau 1995). Musau (1995) explains the pattern by relating it to
Andersen’s (1984) One-to-One Principle. In the IL of learners of Swahili, one form (e.g. class
9 or 10) is used to achieve one function (e.g. agreement) with singular/plural distinctions being
made by the use of classes 9 or 10 respectively. Musau (1995) notes a similar pattern in the
acquisition of adjectival concords.
37
Morphological development is usually found to be a fairly late development in the acquisition of

Indo-European languages, but a question remains about the acquisition of morphology in an
agglutinative African language. Although there may be similar sequences of development from
a functional point of view, formal aspects of development may differ because of the wider range
of morpheme forms which tend to be needed to perform particular semantic functions.
2.5.5.3.3 Utterance development

The Kiel project (Felix 1978; Wode 1976, 1979, 1981) studied the acquisition of German and
English longitudinally. The study focussed on the development of negation, interrogation,
inflections, phonology, and intonation. The results of the study show that the “variable(s) that
determine(s) the linguistic structure of the learner’s utterances are the formal properties of
linguistic devices, i.e. word order, free vs. bound form, segmental vs. suprasegmental marking,
embedding, etc.” (Wode 1981:45).
In a pilot study of Perdue’s (1993a, b) study, Klein (1986:82) lists a similar set of utterance
organisation principles which he found in learners of German (cf. Givón 1979; Robert 1989):
(A) Put ‘given information’ before ‘new information’...
(B) Put what is spoken about before what is to be said about it ...
(C) Keep elements linked in terms of meaning close together ...
(D) Place elements of predominantly functional value consistently before (or
consistently behind) the corresponding elements of predominantly lexical value
...
(E) Place orientational elements (place, time, modality, etc.) at the beginning of an
utterance ...
(F) Mention events in their factual temporal order ...
(G) Indicate sentence modality (interrogative, declarative, requesting) by intonation...
(H) Mark rhematic information by intonation ...
Dittmar (1981) found that early IL relies on the use of adverbials to encode temporality rather
than morphemes indicating tense. Meisel (cited in Sato 1990:85) found extensive reliance on
discourse strategies like “interlocutor scaffolding, implicit reference, order of mention, and
contrast of two or more events” (see also Schumann 1987).
Klein (1986) argues that the pragmatic principles are not linked to a particular language and that
the early learner variety is not influenced by the L1, because the basic categories and inflections
have not yet been established. Word classes cannot always be determined by semantic means or
38
by transfer from the L1. The word class system develops as the learner becomes familiar with the
“distributional regularities and the morphological criteria which characterize the given word
class” (Klein 1986:90).
It seems, then, that utterance development in ILs begins at a pragmatic level and utterances only
grammaticalize at a later stage (if at all) for some learners, who fossilize at the pragmatic
organisation level.
2.5.6 Conclusion
Selinker’s notion of IL is a valuable one which has had an enormous impact on the way in which
SLA theorists conceive of learner language. For example, interlanguage is no longer widely
viewed amongst most theorists as a poor approximation of the TL, but a system in its own right
which is regular and rule-governed. It is therefore equal to other languages, although there are
notable differences between ILs and fully-fledged languages. Firstly, it is a system which can
change quite rapidly and it may have a higher degree of variability at any one time or over time.
Secondly, it is idiosyncratic to the learner to some extent, although there seem to be common
developmental routes across learners from the same and different language backgrounds. Thirdly,
some learners fossilise before they have a fully functional system which could be used in as wide
a range of contexts as their L1s.
2.6 BASIC VARIETY

The Basic Variety (BV) is the next major development which is partially based on the approaches
discussed thus far. The notion of a BV emerged from a study conducted under the auspices of
the European Science Foundation (ESF) study under the guidance of Perdue (1993a, b). A pilot
study was conducted in the early 1980s and the study itself took place from 1984-1987. The BV
is an hypothesized learner-language variety which is independent of the L1 and L2. It is
characterised by a range of syntactic and organisational principles. Klein & Perdue (1997) argue
that internal organisation and transitions are systematic and that BV systems are error-free and
not poor imitations of other languages. Other languages are merely stable systems of language
acquisition where learners stop learning because the learner’s variety and the input are identical.
39
2.6.1 Theoretical background

The theoretical background is similar to IL, although more rigorous methodological measures
have been implemented. Perdue (1993a) distinguishes the BV from IL on two fronts. Firstly, he
claims that even though they set out to do so, IL studies did not look at learner language as a
truly independent system with no reference to the target language (TL), mainly because of
methodological problems. Klein (1998) contends that IL studies are different from BV studies
because IL theorists view learner languages as in-between systems which are imperfect versions
of the TL. In contrast, he claims that the BV perspective truly looks at the system as independent
and studies the system of each learner in its own right. Of course, the system relies on the lexical
items of the TL (and the source language [SL] in some instances), but Klein & Perdue (1993)
hesitate to ascribe too much transfer effect to the SL, as they focus on the independence of the
learner system. There is also a move away from using only a syntactic analysis based on the
grammatical categories of fully-fledged languages towards an analysis which relies on
organisational principles which are more likely to be universal.
The study is remarkable in a number of respects (Perdue 1993b). Firstly, it is extensive in scope,
comparing the language development of five different TLs (Dutch, German, English, French and
Swedish) by forty adult learners who came from a variety of language backgrounds (Turkish,
Punjabi, Italian, Arabic, Spanish and Finnish). Learners with different SLs who were learning the
same TLs were compared so that L1 influence could be assessed. The project ran for thirty
months and longitudinal studies of this length are very rare, mainly because of their expense in
terms of both time and money, and the attrition of subjects. In addition, the study was of learners
in natural environments and finding adult immigrants willing to participate in such a study is
difficult. The findings of this study are valuable because of its cross-linguistic perspective and its
attempt to establish a theoretical framework which describes the early ILs of learners. A stronger
focus on the independence of the learner’s language system, the analytical framework and a
willingness to generalise it to some other naturalistic learning contexts distinguishes this study
from many other IL studies.
Perdue (1993b) hesitates to generalise his findings because the learners were learning in natural
environments. Earlier research on acquisition in both formal and informal environments, however,
seems to indicate that the learner language of most learners is fairly similar (see section 2.7).
40
Perdue’s (1993a) subjects did not all rely completely on naturalistic acquisition as a few had had
some form of formal tuition, although this seemed to have had little effect on the building of their
grammars. In particular, tuition appeared to have negligible long-term effects, which makes the
decision to generalise the findings only to naturalistic learners problematic to some extent.
Nevertheless, Perdue’s hesitation may be justified since Myles et al. (1999) found that although
early learners in instructed contexts show BV characteristics at early stages, they also focus on
memorising formulaic utterances which have been explicitly taught and these chunks form the
basis upon which development proceeds. They find that development seems to result from
“tension between complex but communicatively rich chunks on the one hand and simple but
communicatively inadequate structures on the other hand” (Myles et al. 1999:77). Perdue
(1993b) makes no allowance for these formulaic chunks and, in fact, they are specifically ignored
in the analysis of the data in the ESF study.
From an analytical point of view, the ESF study is more advanced than many other studies. The
analysts tried as far as possible to regard each learner’s language as a system without reference
to the categories of the SL and TL, although this was not always possible. At times, these
categories were useful, but there was also a lack of syntactic categories which were directly
comparable to the SL or TL. This finding forced them to choose an analysis based on
organisational principles in order to find patterns. Linked to this is the use of a more functional
than formal perspective in the analysis. This functional analysis is more difficult for the analyst
(but usually more informative) than a formal one, because it has to take a wide range of factors
into account; e.g. pragmatic factors and extralinguistic factors rather than just linguistic factors.
2.6.2 Characteristics of the Basic Variety2

Klein & Perdue (1993, 1997) conclude that the BV has three broad characteristics:
C it is largely independent of the L1 and the L2 because it does not share the structural
characteristics of either language but is structured by a small set of universal
organizational principles;
C about a third of learners fossilized at this point although they increased their lexicon and
became more fluent in the BV;
C it is very similar to pidgins (Holm 1990), foreigner talk (Roche cited in Klein & Perdue
2
Klein & Perdue (1997) believe that the BV is an I-language (an instantiation of UG), because it
satisfies the constraints on an I-language according to Minimalist Theory and because it is communicatively
efficient. Criticisms of these views are presented by Bierwisch (1997), Comrie (1997), and Meisel (1997).
41
1997) and agrammatical speech (Kolk & Heeschen 1992).
The notion of a universal core language has been posited by a number of researchers including
Corder (cited in Gass 1984a) with his idea of simple codes, Schumann (1978), Klein & Dittmar
(1979), Givón (1979) and Bickerton’s (1990) protolanguage.
Klein & Perdue (1997) believe that there is a universal process at work in BV construction
because of observation of similar developmental patterns across learners of a variety of languages
who themselves speak different languages. They believe that linguistic and organizing principles
are universal and therefore belong to the human language faculty.
Schwartz (1997:393) rejects the notion that the BV is not at all related to the L1 and instead
proposes that the L1 grammar is transferred and “it is the first ‘way station’ for TL input data,
imposing analyses on them and thus potentially deriving analyses quite distinct from those of the
TL native speaker”. She argues for a role for the L1 based on evidence that Punjabi and Turkish
learners of English and German used an SOV word order (which is congruent with their L1
structure) in their early BV, whereas Moroccan Arabic learners (whose L1 has an SVO order)
used SVO order. Vainikka & Young-Scholten (1996) report that Turkish and Korean learners
of German produced utterance-final verbs, which serves as evidence of the possible influence of
transfer. Furthermore, speakers of Italian and Spanish produce verb-initial utterances based on
their own languages, and change these to verb-final ones at an early stage of syntactic
development (Vainikka & Young-Scholten 1996). Broeder et al. (1993a, b) and Schenning & van
Hout (1994) show that the SL has a role to play when alternatives are available in the TL, in
which case the alternative most like the SL form tends to be used. Schwartz (1997) concludes
then that the interaction between L1 structure, TL input and UG needs to be analysed to find the
internal coherence in the system. She argues, in fact, that the BV as a universal form does not
exist and that the L1 plays a far more significant role than Klein & Perdue (1997) allow.
The TL also has some influence on the BV, since the system is related to the TL system to some
extent, because it uses TL vocabulary and the learner aims to learn the TL system. However, the
BV cannot be described fully in terms of the TL because of the idiosyncratic ways in which
learners build up their new system. Rather than the learner using the grammatical categories of
42
the TL, Perdue (1991:420) explains that the TL is used at a discourse level in the following way:
... the principles governing what information needs to be expressed by words, and what
information can be left implicit for the listener to infer, emerge particularly clearly in these
systems. Put another way, adult learners assume that the inferencing capacity they have
to bring to bear when analyzing TL input ... is available to the TL listener who attempts
to understand their speech.
2.6.3 Elements of the Basic Variety

The findings of the ESF study relate to the lexicon, morphological structure and utterance
organisation.
2.6.3.1 Lexical items

Klein & Perdue (1993, 1997) and Dietrich (1989a, b) catalogue the structures available to an
early learner of a language and say that the learner language will contain proper names, noun-like
words and verb-like words, although it is often difficult to decide whether something is a noun
or a verb in early speech (Coupier cited in Perdue 1991). An additional motivation for using the
term “verb-like” is that these words lack the argument structure which they would have in the
TL (Jordens 1997). Additional lexical items include adverbial-like words (usually temporal and
spatial denotations), a few numerals, a few personal pronouns (but no anaphoric pronoun forms
for inanimate objects), a word for negation, a few determiners which are demonstratives rather
than definite and indefinite articles (Carroll & Dietrich 1985) and a few more complex rote or
formulaic constructions. There is no copula, nor are there prepositions (except a few with lexical
meaning, usually denoting spatial and temporal relations).
Lexical items are based on the L2, although a few L1 items may be used, and there are more
open-class items than closed-class items (Klein & Perdue 1997). Lexical items tend to occur in
an invariant form which is usually the stem, infinitive or nominative form or an inflected form
borrowed directly from the TL. Word formation tends to be via noun-noun compounding rather
than derivational processes (Broeder et al. 1993b), with the SL playing a greater role here when
there are multiple or ambiguous TL word-formation processes. Single lexical items are preferred
to compound ones (Klein & Perdue 1993). Irregular forms are acquired before regular forms.
This is congruent with other findings which indicate a period of correct use of irregular forms
before the productive application of regular forms (e.g. Bybee 1991; Cazden 1968; Fillmore cited
in Seliger & Shohamy 1989; Klein 1986).
43
2.6.3.2 Morphology
With regard to morphology, the BV is characterised by a lack of morphological inflections with
“purely grammatical function” (Klein & Perdue 1997:332). The lack of inflection means that
there is no finite morphology marking tense and aspect, no agreement in number, case or gender
and no agreement by morphology (Carroll & Dietrich 1985). In contrast, however, in another
large study (Pavia Project) of learners of Italian in mainly naturalistic contexts, Giacalone Ramat
(1992:302) found that morphological sensitivity (“the ability to analyze words in the input into
their morphological components and to distinguish on a principled basis among variant forms of
a grammatical category ...”) developed quite early. Bardovi-Harlig & Bofman (1989) agree with
Giacalone Ramat (1992), suggesting that lack of morphological development may be related to
particular TLs. They suggest that learners of Turkish, which has a very regular morphological
structure, may show more accurate and less variable acquisition patterns, although this has not
been investigated.
2.6.3.3 Functions expressed in the Basic Variety

Form-function relations between words are shown by means of pragmatic, semantic and phrase-
structure constraints and BV users are able to perform a variety of functions relating to
temporality (duration, habituality, iterativity), spatial relations (location, change of location) and
types of situation (states, dynamic events) (Jordens 1997:291). In his discussion of the
development of temporality, Klein (1993:102) points out that “there is no way to mark
temporality by grammatical means” in the BV because of the lack of verbs and copulas. Time
reference is marked by the use of temporal adverbials, use of the time reference used by the
interviewer or by the context of the utterance (Klein & Perdue 1997). Using the time of the
utterance as time reference is particularly prevalent because of the use of film-retelling discourse
as data (Perdue 1990). Klein et al. (1993) found that contrary to other studies (e.g. Bickerton
1982; Weist 1986), tense marking precedes aspect marking, with aspectual forms such as early
use of -ing not having clear aspectual functions.
2.6.3.4 Utterance structure

At the level of utterance structure, there is an interaction of phrasal, semantic and pragmatic
constraints (Klein & Perdue 1993).
44
2.6.3.4.1 Phrasal constraints

Phrasal constraints relate to the form and order of the elements and Klein & Perdue (1997:314)
say that phrasal constraints in the BV are mainly of the following kind:
A. NP1-V-(NP2(NP2))
B. NP1-Cop-{NP2
ADJ
PP}
C. V-NP2
NP1 may be a proper name/optional determiner plus noun/pronoun/zero. NP2 may be a proper
name or optional determiner plus noun. The verbs are usually in the base form, although there
may be a V-ing form in English, infinitives (German, French) or an inflected form (Swedish), with
the SL having an effect on the form of the verb which is more prevalent (Klein & Perdue 1997).
Klein & Perdue (1993:27) state that “(t)he placement of adverbials and negation mainly depends
on topic-focus structure and semantic scope”. They note that adverbial items are usually directly
in front of or behind the item over which they have scope, or at the point between topic and focus
elements (especially in the case of negative particles) (Giacomi et al. and Dimroth & Klein, both
cited in Klein & Perdue 1997). This observation fits the placing of early negation particles found
in other studies with external negation of no and placement of the negative particle directly before
the item to be negated (e.g. Wode 1981).
The basic word order is Agent-Verb-Patient and Klein & Perdue (1997) claim that this is a
universal pattern for all learners in untutored learning situations. Comrie (1997) and Schwartz
(1997), however, question the validity of this claim as the TLs all had this basic word order and
it may therefore be a function of the TL input, especially since Punjabi and Turkish learners
initially use verb-final forms as their own languages do.
2.6.3.4.2 Semantic constraints

Semantic constraints relate to argument structure and case roles. At a semantic level, the
controller of the action is in the first position when verbs appear, although this may be modified
by the use of prepositions. Deciding on the “subject/object” function requires an investigation of
the semantic controller in the utterance, with the general rule being that “NP-referent with highest
control comes first” (Klein & Perdue 1997:316). If there are controllers of source and target
45
states, as in the arguments of the word give, the “controller of the source state outweighs
controller of target state” (Klein & Perdue 1997:315).
2.6.3.4.3 Pragmatic constraints

Pragmatic constraints relate to textual structure (Jordens 1997). They include a strong topic-
focus structure (cf. Givón 1979, 1984, 1985) and patterns of reference introduction and
maintenance (given-new patterns) (Klein & Perdue 1997). Perdue (1991) notes, however, that
early speech is characterised by scaffolding by the questioner, so that the focus-last sequence
could be an artifact of the elicitation task.
2.6.4 Development of learner language

Unlike Corder (1977), who saw the simple code as the initial state, Perdue (1996) discusses a
pre-basic variety which indicates that the BV is not the initial state of L2 acquisition. Perdue
(1996) says that the pre-basic variety is noun-based and based on semantic-pragmatic relations.
Non-finite verbs and NP-internal structure only emerge later in the BV. Klein et al. (1993) discuss
a pre-basic variety characterised by its lexical nature without clear distinctions between word
classes (although there is a predominance of noun-like words), no functional inflections, and few
complex constructions which are characterised by linking via pragmatic principles and context
dependency. Jordens (1997) says that verbs do not structure the utterances in terms of arguments
and case role assignment, although words cluster around this verb-like element. Organisation
using lexical principles and grammatical principles only appears later (Perdue & Klein 1993).
Development proceeds from the “nominal via infinite to finite utterance organisation” (Klein &
Perdue 1993:25). The nominal stage of development is the pre-basic variety discussed above. The
next stage involves the appearance of non-finite verbs which can assign arguments such as agent
and benefactor but which still lack verbal morphology. Verbal morphology needs to be acquired
to move to the third stage of target-like production of finite verbs. This stage is beyond the BV
and the IL looks very much like the L2. A further stage of use of subordinators can also be seen
in some learners (Perdue 1990). Development from stage to stage is “slow and gradual” (Klein
& Perdue 1993:25) with variable performance and backsliding at each stage. They conclude then
that:
... the acquisition of finiteness is not merely a question of adding some morphological
46
features, in this case verb inflexion, to the learner’s repertoire. It marks the transition
from a type of utterance organisation which is dominated by semantic and pragmatic
constraints, to a type of utterance organisation in which phrasal constraints (i.e.
syntactic constraints in the narrower sense of the word) gain equal weight. (Klein &
Perdue 1993:29)
2.6.4.1 Pronoun development

Pronoun development is variable but generally follows the pattern of “(i) singular ... before plural;
(ii) nominative ... before oblique; (iii) pronouns referring to animates ... before those referring to
inanimates and (iv) definitely referring NPs ... before overt pronouns” (Perdue 1991:417; see also
Broeder 1991 and Perdue 1990:994). However, Perdue (1990) cautions that the task design (use
of film-retelling) precludes display of development of first- and second-person pronouns.
2.6.4.2 Reasons for development

The main explanation of development is that it occurs when constraints come into conflict
(Perdue 1990). As new components are added to the system, changes occur in that system
because of changing interactions of the principles. For example, if a syntactic element is added,
then the weight of some of the other organisational principles may decrease. This shifting of the
weight of principles may result in patterns more characteristic of the TL. For example, the
number of noun-like elements decreases over time as verb-like elements are added. An increase
in utterance complexity is also noticed, with an increase in articles, prepositions, conjunctions and
pronouns only after the BV stage (Broeder et al. 1993a; Klein & Perdue 1993, 1997).
Development also occurs to improve communicative efficiency (Perdue & Klein 1992). This
development, however, may only occur in the lexicon rather than the utterance organisation after
an initial period of development. Development beyond the BV is more strongly influenced by the
TL because more specific aspects of the TL need to be acquired at this point, since the system
is complexified away from the simple basic system (Perdue & Klein 1992). Carroll & Becker
(1993:146) point out that development is limited by processing capacity, especially in a
conversational situation and because the learner is “building up a workable system of
communication by using it actively”. Perdue & Klein (1993) conclude that the L1 can influence
the rate of acquisition and ultimate achievement, but the sequence of development is not affected
by the L1 (as found by other researchers, see section 2.7).
Other reasons for development have been discussed in the literature and Klein et al. (1993)
47
conclude that length of stay in the TL community is not a causal variable for development.
Rather, it is the intensity of input that determines development. Development towards the finite
verb form is characterised by:
C acquisition of morphology as the distinction between non-finite and finite is acquired (cf.
Jordens 1988);
C more complex ways of expressing temporality;
C the development of case oppositions; and
C subject-predicate marking (Perdue & Klein 1993).
Parameter setting and triggering after exposure to input are also possible explanations for
development, although Jordens (1997) does not find this a convincing argument because of the
stability of the BV system. He offers the case of structural ambiguity as an additional driving
force behind development and feels that the acquisition of categories of fully-fledged languages
such as subject, object, noun, etc. are necessary to resolve structural ambiguity and develop
structural dependency and recursivity relationships (which are features of fully-fledged
languages).
2.6.5 Conclusion
The BV clearly fits the findings of many IL studies, but it adds considerably to our knowledge
of the organisational principles at work in learner language. The cross-linguistic perspective of
the study is illuminating and the common patterns found across learner languages supports many
previous studies. However, Perdue (1993a) repeatedly refuses to generalise the findings of the
ESF project to learners in formal situations, and I would argue that there are similarities which
should be considered. For this reason, an overview of the findings regarding the influence of
formal learning on ILs is considered in the next section.
2.7 FORMAL VS. NATURALISTIC ACQUISITION

Studies of development in classroom and naturalistic situations seem to point to the conclusion
that the route of development stays the same regardless of the learning situation but that the rate
of acquisition is faster and the ultimate attainment higher in classroom situations (Ellis 1985b;
Larsen-Freeman & Long 1991; Long 1983). Some structures such as third-person agreement -s
and plural -s in English also seem to be produced more accurately by learners with instruction.
Ellis & Roberts (1987) say this may point to earlier acquisition. On the other hand, there seem
to be difficult structures, many of which seem impervious to formal instruction (Larsen-Freeman
48
1998). These include items such as the French gender system (Clark 1985; Harley cited in
Schachter 1998; Karmiloff-Smith 1979), adverbial placement (White 1991; Trahey & White
1993) and English object pronouns (Yuan 1997).
The structures mentioned above may be complex and Robinson (1997b) found that learners not
explicitly taught complex rules performed better than those who had been given explicit
instruction. Schmidt (1990) and DeKeyser (1995) report similar results. Learners explicitly taught
simple rules perform better than those who are not given explicit instruction (Robinson 1997a,
b).
Sharwood Smith (1994:117) distinguishes studies which ask whether formal instruction can
disturb the natural order (e.g. Pienemann 1984) from those which ask “whether classroom
instruction displays the characteristics encountered in natural situations” (e.g. Felix & Hahn
1985). The latter studies can help to explain why similar routes are found if instruction and
natural discourse display the same characteristics, while the former type have less of a
comparative focus.
There have been studies of learning in four different types of learning settings:
C naturalistic settings (e.g. Clahsen 1987; Felix 1978; Huang cited in Felix 1981; Meisel
1987; Perdue 1993b; Trèvise 1987; Veronique 1987; von Stutterheim & Klein 1987);
C mixed settings (e.g. Butterworth cited in Felix 1981; Pienemann 1987; Schumann 1975);
C formal instruction only in the classroom (e.g. Ellis 1989; Felix 1981; Myles et al. 1999;
Weinert 1987); and
C formal instruction only in experimental conditions with artificial languages (e.g. de Graaff
1997; DeKeyser 1995; Ellis & Schmidt 1997; Yang & Givon 1997).
Felix (1981) notes that there is a crucial difference in the input in the different situations because
naturalistic learners tend to have exposure to a wide range of structures from the beginning,
whereas classroom learners tend to have a restricted number of structures in their input. They are
also more likely to be presented with isolated models containing the structures they should be
acquiring, so they may be more aware of the actual model that they should be attempting to
produce, even though the requirements for accurate production may be too numerous at early
stages. Some structures seem to be more amenable to teacher intervention and Felix (1981)
mentions that lexical negations like no and don’t could be more easily corrected than structures
49
which extend over longer chunks of language or may be more abstract, e.g. word order. He
concludes that there is nothing in the productions of instructed learners which would not also
be found in the language of learners in naturalistic situations. However, Doughty (1991) cautions
that Felix’s (1981) conclusions are based on noting similarities between the ILs of instructed and
naturalistic learners, rather than focussing on any advantages that instruction might have.
The question which many studies strive to answer is whether there is transfer from learnt
knowledge to the underlying system. Krashen (1981, 1982) argues a non-interface position
(learnt knowledge does not transfer), while others like McLaughlin (1990) and O’Malley &
Chamot (1990) argue a strong interface position (learnt knowledge transfers). Others take a
middle road and argue that instruction and input enhancement can assist in noticing, so that
implicit acquisition is facilitated (Ellis 1993; Long 1991; Tomlin & Villa 1994). De Graaff
(1997:251) concludes that
(a)lthough some evidence for the facilitative effect of explicit instruction on L2
acquisition has been found, little is known yet concerning the question of under which
specific learning circumstances and for exactly which aspects of grammar explicit
knowledge can be most facilitative for L2 acquisition. In Hulstijn and de Graaff (1994),
we argued that the following variables could possibly influence the effect of explicit
instruction: the target structure’s linguistic domain (in the core or in the periphery of
Universal Grammar), its complexity ..., its degree of semantic redundancy, its reliability
(the ratio of regular to irregular cases), its scope (number of phenomena covered), the
frequency with which it is manifested in the input, the competition between rule-based
learning and item-based learning ..., task modality (reception vs. production), type of
instruction or input enhancement, and individual learner characteristics.
Ellis & Roberts (1987) discuss a number of reasons for the more rapid development of classroom
learners. Firstly, if the input is tailored so that learners are developmentally ready to learn a
structure, it may ensure that the learners acquire the structure more rapidly than if they were left
to discover the rule for themselves. A second reason may be the consciousness-raising function
of instruction, so that learners may become aware of patterns in grammatical structure which
would have gone unnoticed otherwise. A third reason may be that the formality of the language
used in the classroom may ensure that the learner is exposed to structures which are rare in
everyday speech but more frequent in formal speech and in writing. The learner has more
exposure to these constructions and may therefore acquire them more easily than if there was
only occasional exposure to them in natural situations.
50
With regard to the long-term effects of formal instruction, Pienemann’s (1984) study seems to
suggest that there are no long-term benefits to teaching structures that learners are not ready to
process. Pienemann (1984) concludes that there are some structures which depend on the
developmental readiness of the learner and there are some structures which can be learnt at any
time. In addition, there can be backsliding after instruction to a previous state in which the earlier
form is produced rather than the taught form. Towell & Hawkins (1994) show that any
immediate effects of classroom instruction may wear off and leave few long-term effects.
Generalisability of learnt skills may be problematic (DeKeyser 1997; DeKeyser & Sokalski 1996)
and discrete exercises where the rules are applied to sentences and translations may well be
correctly completed. However, generalizations outside the classroom to situations which do not
require exact repetition of forms learnt in the class are not nearly as successful.
2.7.1 Limitations of the studies

Doughty (1991) argues for a cautious interpretation of the results of many studies on the effect
of formal instruction. She feels that many studies do not take into account (or make clear)
different amounts of natural and classroom exposure. They are therefore inappropriate predictors
of the effect of classroom instruction on the IL of learners. She further discusses the need to
specify the nature of instruction received by the learners, although comparisons of different forms
of instruction have not shown conclusive differences (Long 1983, 1988). Another criticism is that
it has been difficult to measure proficiency, which has tended to be measured in a global way and,
as a result, it is impossible to conclude whether the influence has been from formal instruction
or not. Long (1988) concludes that, although instruction seems to improve rates of learning and
levels of ultimate attainment, future studies need to be careful about subject selection and the
selection of structures so that studies are comparable.
Furthermore, Long (cited in Doughty 1991:460) notes that there is a difference between a focus
on form and a focus on forms, “where the former refers to instruction that, in some meaningful
way, draws learners’ attention to TL structure in context and the latter refers only to the
presentation of isolated linguistic structures”. Faster rates of acquisition have been noted where
there is a focus on form as opposed to no focus on form.
Jordens (1996) believes that proving the influence of formal instruction on IL development is not
51
an easy matter and that even within the classroom, different tasks may have different effects on
IL development. He distinguishes three different types of classroom activity: “(a) activities which
focus on the processing of input, (b) activities which focus on what is called ‘consciousness-
raising’ and (c) activities which focus on the productive use of taught formal knowledge”
(Jordens 1996:426). The first activity does not require the learner to produce any particular
structures in the input, but the input has been carefully selected to expose learners to input which
is ahead of their current competence and they are expected to find meaning in the input. The idea
is based on Krashen (1982) who believes that focus on meaning will accelerate development.
Other studies employing this activity include Ellis (1989), Pienemann (1984, 1989) and Macky
(cited in Pica 1998). Using input which has been structured to present structures according to an
accessibility hierarchy has also proved effective (Doughty 1991; Eckman et al. 1988). Doughty’s
(1991) study focussed on the acquisition of relativization, for which markedness hierarchies have
been clearly worked out. She found that “[i]nstruction incorporating unmarked data generalizes
only to unmarked contexts, whereas instruction incorporating marked data potentially generalizes
not only to that marked context but to other contexts as well” (Doughty 1991:464), although she
cannot provide an explanation for this finding.
The second activity, consciousness-raising (also called input enhancement), involves making the
learner aware of particular structures in the input. This is particularly useful for structures which
are not crucial to meaning and might therefore not be noticed otherwise. Examples of studies
employing input enhancement include Lightbown (cited in Pica 1998), Lightbown & Spada (cited
in Pica 1998), Rutherford (1987), Rutherford & Sharwood Smith (1985) and Sharwood Smith
(1991, 1993). Although the structures will not be acquired until the learner is developmentally
ready to acquire them, the learner has been “primed” (Ellis 1990:169).
The third activity involves explicit teaching of a grammatical structure or the provision of
information about that structure. Development occurs when learners “apply ... generalizations
first as formal operations and subsequently in more communicative settings” (Jordens 1996:426).
This last activity is probably the most often practised activity in formal settings, but there is a
great deal of controversy about its usefulness to students. Robinson (1996a, b) reports that
studies have had mixed results with regard to the use of formal instruction and they usually
conclude that the learner needs to be developmentally ready to acquire the item before instruction
52
shows any effect (see also Ellis 1989; Larsen-Freeman & Long 1991; Pienemann 1984, 1989,
1992).
A further method of instruction discussed by Pica (1998) is what she calls garden path
interaction where learners are presented with the general rule and led to believe that the general
rule can be applied to all instances. They then regularise forms which should be irregular. The
benefit is that when learners are then taught irregular forms they remember them more correctly
than those taught regular rules and irregularities at the same time (Tomasello & Herron 1988,
1989).
2.7.2 Reasons for better performance

Formal instruction may hasten development because of the nature of classroom discourse.
Classroom discourse contains a lot of feedback which provides learners with explicit feedback
about hypotheses they have made (Allwright 1984). Learners do not therefore have to wait for
negative evidence in the input when testing hypotheses. However, formal instruction does have
an impact on the development of spontaneous speech because teachers take up about two-thirds
of classroom talk time (Allwright 1984). This means that the learners have few practice
opportunities which are further depleted by the number of learners in a classroom all competing
for opportunities to talk.
Ellis (1987b, c) uses Tarone’s (1982) model of careful and casual styles to explain the better
performance of instructed learners by saying that classroom learners are better able to perform
tasks involving the careful style, as this is the type of learning which is required from them most
often. Tarone (1982) argues that development moves from careful to casual styles and Ellis
(1987b, c) concludes that instructed learners are at an advantage even in the casual style if they
are able to exploit their knowledge of other styles to produce this style. They are particularly
advantaged when they are able to practise this transfer in unplanned discourse in the classroom.
Pedagogical and social norms also affect development more strongly in the formal context. These
norms influence the ILs of classroom learners, making their ILs more permeable to change and
positively influencing development.
Schauble et al. (cited in McGroarty 1998) say that naturalistic situations allow learners to choose
53
their goals in a setting, whereas formal settings limit the choices of learners (cf. Spolsky 1989).
A good learner seeks out exposure to the language and interaction at the level that he/she is
comfortable with (Naiman et al. 1978; Stevick 1989) and formal teaching may be helpful to some
but have no effect for others (McGroarty 1998). Other studies (e.g. Lightbown & Spada 1990;
White 1991; White et al. 1991) have shown that formal instruction in immersion situations may
improve comprehension rather than production. However, there may still be a high level of
linguistic inaccuracy in complex clause structures, tense, aspect and sociolinguistic rules, even
after long periods of immersion.
Klein & Perdue (1997) claim that the Basic Variety (BV) is only generalisable to naturalistic
situations of acquisition and that the BV system has not been found for learners in classrooms.
They argue that in the classroom natural principles of acquisition as well as the effect of particular
teaching methods are at work. The learner’s system is thus influenced by an outside force which
has considerable influence of the structure of this system. In particular, learners are expected to
learn structures such as inflectional morphology, so it would be expected that they would also
produce these forms. However, they also note that learners performing outside the classroom do
not always transfer what they have learnt to their production in naturalistic situations. If transfer
occurs, then the BV would not be highly relevant to this study, but if it does not usually happen,
then the BV may be found in learners with formal learning backgrounds performing outside of
formal contexts.
2.8 CONCLUSION
This chapter has outlined the history of the second-language acquisition (SLA) field as it relates
to IL. As noted throughout the chapter, there are aspects of earlier research which need to be
taken into account and since there is no coherent SLA theory, it seems that an eclectic approach
provides a richer picture of IL development. A wide range of aspects needs to be taken into
account and no single theoretical or methodological perspective seems to provide an entirely
adequate explanation of all areas.
The choice of the BV as a background theory to my study needs to be justified. Firstly, numerous
studies show that formal instruction has less effect on IL than teachers would hope and that
developmental sequences remain the same regardless of the formal input. As a result, I would
54
argue that the BV must have some significance for my study, even if it shows where the BV and
the IL represented in my study are not related. Secondly, although some learners in this study
may have moved beyond the BV, some of the learners are at early stages of learning and their ILs
are therefore more likely to display characteristics of the BV. In addition, there is never 100%
correct suppliance of any morpheme in the data, which may indicate that learners backslide to
previous levels (in this case the BV level). Finally, the European Science Foundation (ESF) study
is an extension of IL studies, so it is necessary to discuss it so that a richer picture of early
development can be formed.
55
CHAPTER 3
METHODOLOGY
3.0 INTRODUCTION
The broad, primary aim of my study is to investigate the development of morphemes in the Xhosa IL
of English speakers. Related to this broad aim are the more specific aims of exploring:
C which morphemes are used;
C which functions are realised;
C how the functions are realised;
C what sequence of development of morphemes is evident;
C why this sequence is found in the data; and
C how the findings of my study relate to current theory and other studies.
This chapter explores the methodological approach of the study. Firstly, it outlines the rationale for
the choice of a longitudinal, qualitative approach. Then it proceeds with a detailed exposition of the
sample, and data collection and analysis procedures. Finally, the limitations of the methodological
procedures are highlighted.
My study borrows aspects of methodology from a range of studies, since it takes both a formal and
a functional approach to the data analysis. In this way, a more comprehensive picture of the language
data and development in the IL can be formed. The formal approach provides an indication of overall
development and trends in the data, while a functional approach assesses the development of a limited
number of functions in greater depth. Furthermore, functional approaches provide for a multilevel
analysis of data so that syntactic and pragmatic variables can also be shown to contribute to
interlanguage (IL) features. This multilevel analysis therefore allows a wider range of explanations
of the developmental patterns which are evident in the data.
Long & Sato (1984:279) point out four areas which are generally problematic in SLA methodology:
“focus on (1) product rather than process, (2) form rather than function, (3) single rather than
multiple levels of linguistic analysis, and (4) IL in isolation rather than in its linguistic and
conversational context”. My study tries to overcome some of these problems in a variety of ways.
56
With regard to the first area (focus on product rather than on process), both the product and the
process can be investigated because of the longitudinal nature of the study. The second area (focus
on form rather than function), is addressed since both form and function are considered in my study.
Following Pfaff (1987), I do not believe that the two can be separated in many cases. With regard to
the third area (single rather than multiple levels of analysis), the morphological, syntactic and
discoursal levels are considered to some extent in the explanation of data, although the focus of the
study is on the morphological level. The fourth area (taking context into account), is addressed in my
study, since the context of the IL production is taken into account during analysis and the use of a
qualitative paradigm allows for a focus on contextual factors.
3.1 METHODOLOGICAL APPROACH

My study takes a quasi-longitudinal, case study approach to data collection. The analysis of the data
is primarily qualitative. Reasons for the choice of this methodology are discussed in detail below.
3.1.1 Longitudinal and cross-sectional studies

Studies of SLA have usually taken one of three approaches to data collection. One approach is cross-
sectional, where data are collected from subjects on a single occasion. This approach is economical
and efficient since a large number of subjects can be studied at one time and development of a
particular form or function can be measured fairly easily. If subjects have different levels of
competence, inferences can be drawn about developmental sequences by taking samples of speech
or writing from learners of different levels and sequencing the different developmental patterns which
are found. The most well-known studies taking a cross-sectional approach are the morpheme studies
(e.g. Dulay & Burt 1973, 1974a, 1974b; Fathman 1978; Krashen et al. 1976; Pica 1983), although
other studies also use this approach (e.g. de Graaff 1997; Extra cited in van Els et al. 1984; Klein &
Dittmar 1979; Yang & Givon 1997).
A second approach is a longitudinal study where one or more subjects are followed over time and
data are collected by sampling at fairly regular intervals (e.g. Carroll 1984; Giacalone Ramat 1992;
Myles et al. 1999; Perdue 1993a; Skiba & Dittmar 1992). The longitudinal study can last for anything
from a few months to a few years (e.g. Ghrib’s 1987 study of a Turkish speaker for about 6 months;
57
Huebner’s 1983 study of a Hmong speaker for 12 months; Schumann’s 1978 study of a Spanish
speaker for 9 months).
The longitudinal study has a number of advantages over the cross-sectional study. Regular appraisal
of the learner’s development allows for closer documentation of the individual’s progress and a richer
picture of developmental sequences and/or lack of development can be painted. This approach also
allows the analyst to focus on a wide or narrow range of forms and functions, depending on the
purpose of the study and the stage of development of the learner. Regular monitoring of a learner’s
grammar can then also be linked to the input that the learner receives in order to assess whether it has
any impact on the grammar (Corder 1971, 1981; Selinker 1992).
Despite the benefit of being more methodologically sound than cross-sectional studies (Odlin 1989),
longitudinal studies are difficult to carry out for a number of reasons. These reasons include:
C high costs;
C constraints on the number of subjects it is possible to study;
C time constraints; and
C attrition of subjects over the period of the study.
Analysis is also more difficult because there is a range of variables which need to be accounted for.
Furthermore, it is not always possible to develop an effective analysis framework to account for all
of the data, which may include formulaic utterances, idiosyncratic rules and constraints, as well as TL
forms.
A third approach to data collection is to combine a cross-sectional and longitudinal approach if more
than one subject is studied and the learners are at different levels of development. The
Zweitsprachenerwerb Italienischer und Spanischer Arbeiter (ZISA) project (Meisel et al. 1981), the
Pavia Project (Giacalone Ramat 1992) and the European Science Foundation (ESF) study (Perdue
1993a) all used this pseudo-longitudinal or quasi-longitudinal approach. In these studies, “data from
the least advanced learners is hypothesized to represent data collected first, and data from the most
advanced learners [represents] the data collected last in a longitudinal study” (Faerch et al. 1984:297).
This approach combines the advantages of both longitudinal and cross-sectional approaches.
58
However, Young (1991) criticises pseudo-longitudinal studies, because he says that they assume that
IL changes only over the single dimension of time.
My study uses a pseudo-longitudinal approach since six subjects with different competence levels are
interviewed at regular intervals over a period of eight months. This combined approach enables the
investigation of both individual development and developmental sequences, since subjects have
different levels of competence. The use of this approach was a result of the availability of subjects,
as well as the expectation that significant staged development might not occur over the relatively
short period of the study.
3.1.2 Case study approach

My study has the goals and design of a case study since
the purpose of a case study is to describe the case in its context. Guided by a research
question, a researcher studies the case and those aspects of the environment that pertain to
that case and that shed light on the research question. (Johnson 1992:76)
A more in-depth approach is possible with a case study than was possible, for example, with the
morpheme studies. There is also a high degree of flexibility with regard to data collection methods
and a variety of data may be collected (Johnson 1992; Seliger & Shohamy 1989). Selinker (1992) and
Eckman (1994) state a preference for case studies where individual learners are studied so that the
variation and progress of the individual is not lost in data from a large group, which is usually the case
in large-scale quantitative studies. In addition, it is not always possible to test a particular hypothesis
when there is a complex task like language learning under investigation.
Despite these advantages, there may be validity problems with case studies. For example, the
researcher may struggle to maintain objectivity (van Els et al. 1984), although better validity can be
achieved by triangulation. A further limitation of this kind of study is that the learner’s language
development may be influenced by frequent interviewing or contact with the interviewer (van Els et
al. 1984).
Furthermore, case studies with a single learner or a small group of learners may not be good
59
predictors of the linguistic behaviour of a larger group of learners (Selinker 1992). In order to
counteract this prediction problem, Johnson (1992) says that it is possible to look at a variety of
studies and extract general principles which can then be compared with the case study findings.
Huebner (1991) notes, however, that generalisability across studies is difficult because of the range
of elicitation methods employed, all of which have an effect on the type of data produced. Like
Johnson (1992), Huebner (1991:13) prefers to see case studies as a useful “means of generating
hypotheses which can then be tested through quantitative studies”. Studies using a case study
approach include Hakuta’s (1976) study of a Japanese child learning English, Schumann’s (1978)
study of a Spanish speaker learning English, and Wong-Fillmore’s (cited in Johnson 1992) study of
five children learning English.
3.1.3 Qualitative methods

My study takes a qualitative rather than a quantitative approach to the analysis of the data, although
descriptive statistics are used where quantification is illuminating. One of the main advantages of
qualitative methods is that they enable the researcher to look at areas in syntax, semantics and
pragmatics where binary choices are not always available for quantitative analysis (Selinker 1992;
Tarone et al. 1976). Qualitative studies may also prove better when studying the process of second-
language acquisition (SLA) rather than the product (Rutherford 1984). Quantitative studies usually
require a fairly sophisticated level of statistical analysis which requires a very careful manipulation of
the variables involved in language production. Variable manipulation is not always possible (or
desirable), especially in a longitudinal study (Young 1991). Nevertheless, quantitative studies are not
superfluous, and Selinker (1992) argues that quantification might form part of an analysis.
Nevertheless, there are a number of limitations associated with qualitative analysis. Tarone (1987)
says the most significant limitation is the lack of generalizability which is possible when results are
based on qualitative studies. She warns that data can easily be overinterpreted and to avoid this,
contextual factors need to be taken into account, especially when data sets are limited. Similarly,
Roberts & Simonot (1987:135) point out that a number of contexts need to be taken into account
during analysis, including the “context created as the interaction unfolds ... contexts of previous
similar interactions ... and the wider social context ...”.
60
3.2 DATA COLLECTION PROCEDURES

3.2.1 Selection criteria for the sample
A number of factors need to be taken into account when selecting a sample. For example, Extra &
van Hout (1996) note that age, instruction and previous formal education seem to have an effect on
SLA and that these factors should be acknowledged when selecting a sample and analysing the data.
For this reason, a shortlist of students was compiled from a list of students registered for Non-Mother
Tongue (NMT)1 Xhosa I, II and III at Rhodes University in 1998. Selection criteria for shortlisting
were that:
C students were first-language English speakers;
C they had not studied any languages besides Afrikaans and Xhosa at school or university; and
C they were between eighteen and twenty-one years old.
I wished to restrict the study to English speakers so that it might be possible to generalise about
transfer or lack of transfer by keeping the L1 common. As all South African schoolchildren are
required to study an L2, it was impossible to select a sample which had not learnt any other language
besides Xhosa. This would have been preferable so that interference from an additional language
could be ruled out. Nevertheless, all of the participants had learnt Afrikaans as an L2 at school and
this allows for some measure of generalisability in the results.
The restriction on the age of the participants was included to promote homogeneity with regard to
the possible effects of a range of socio-cultural and cognitive factors. Learners who are much older
than the class average may have different motivations, life experiences and attitudes from their
younger classmates, and these factors may have an impact on their learning of the L2. Although age
itself does not seem to be a major factor in developmental sequences of language learning, it is still
preferable to keep the sample as homogenous as possible.
1
The second-language Xhosa course at Rhodes is known as Xhosa Non-Mother Tongue (NMT) and is
restricted to students whose first language is not Xhosa, although speakers of other African languages may do
the course. From 1998, students who had done Xhosa as a grade 12 subject were not allowed to enter the first-
year course since the course content was similar to the content of the school syllabus. However, this restriction
was new and two of the participants in my study had done the first-year course despite a grade 12 pass in Xhosa.
61
These students were then given a letter outlining my study and they were asked to indicate whether
they were willing to participate for the full data-collection period. They were not offered any
remuneration for their participation, which was entirely voluntary, although encouraged by lecturers
in the Department of African Languages. Six students (three at 1st year level, two at 2nd year level,
and one at 3rd year level) agreed to participate in my study.
3.2.1.1 Background of Participants

Using a semi-structured questionnaire, I interviewed each of the participants in English about the
following aspects, at the end of the year in which data were collected using a semi-structured
questionnaire (see Appendix A for the questionnaire):
C motivation for learning Xhosa;
C previous experiences of learning Xhosa;
C practice opportunities utilised; and
C attitudes towards the context of learning.
The participants are introduced below2.
Ann
Ann, aged 19 at the beginning of the study, started Xhosa I(NMT) as a complete beginner. She hoped
to be able to speak some Xhosa by the end of the year and she believed that she would be competent
enough to hold a conversation at the end of the course. She decided to take Xhosa because she
intends to do community development work in the Eastern Cape Province of South Africa where the
majority of people speak Xhosa. She felt that although she had had high expectations of being fluent
by the end of the year when she started the course, she had not practised enough and was not
confident enough to hold a conversation. The only interaction she had with Xhosa speakers outside
the classroom was when she greeted staff in her residence.
She spent about two hours a week working on Xhosa homework outside the class. She said that her
speaking ability was “not good” but that she could read and understand “a bit” by the end of the
2
Appendix B provides a short summary of the background of the participants for later reference.
62
course. When she was trying to understand a message, she focussed on the roots of words so that she
could understand the basic message, rather than the grammatical categories, which she felt were
mainly superfluous to meaning. She had enjoyed the classes and would encourage others to take the
course.
During the interviews, Ann was very hesitant about speaking Xhosa and resorted to using English
very often. She cited her lack of vocabulary as a major stumbling block to her ability to speak and
understand. She obtained 61% for the course at the end of the year.
Ben
Ben, aged 19 at the beginning of the study, was registered for the Xhosa I(NMT) course. He had first
learnt Xhosa as a young child, but he had lost most of his speaking ability after moving away from
a Xhosa area at the age of five. He then learnt “basic” Xhosa in grades six and seven and more
grammar-focussed Xhosa from grades eight to ten. He claimed that his comprehension skills were
still good as a result of his early contact with the language, and this is borne out in his interviews
where he rarely required questions to be repeated.
He was studying Xhosa because he thought it was an interesting language, it was an advantage to
know Xhosa and it was advantageous to know three languages. His goal was to learn to speak
Xhosa, but he very seldom spoke to people outside the classroom and he spent about two hours on
Xhosa homework outside the class every week. He felt that his comprehension skills were “very
good”, and his speaking, reading and writing was “fine”. He did not think that it was very important
to produce error-free speech in case other speakers thought he was trying to sound “posh”. His
attitude towards Xhosa was positive because he felt that classes were fun and he obtained good marks
for the subject (final mark of 85%). He would encourage others to study Xhosa and he was thinking
about doing the second-year course (although he did not do this).
Ben was fairly fluent during the interviews and code-switched when he did not know a Xhosa word.
He restricted himself to answering the questions in most cases and did not elaborate on utterances
unless specifically asked to do so.
63
Claire
Claire grew up on a farm in the Eastern Cape and had been exposed to Xhosa from an early age,
although she said she had lost much of her speaking ability when she went to school. Nevertheless,
she is a fairly fluent and very uninhibited Xhosa speaker in terms of speed and lack of hesitation, but
has a fairly limited range of topics and vocabulary at her disposal. Her speech also included slang,
which was not evident in the speech of other participants. She had never read or written Xhosa before
she started the Xhosa I(NMT) course and her motivation for taking the course was to learn to read
and write the language, and to regain some of the ability which she had lost. She also thought she
might like to be a teacher and she felt that Xhosa would be very advantageous in a teaching situation.
She said that she spoke Xhosa “all the time” to staff on campus and people in her residence and
enjoyed speaking the language. She found conversation classes extremely easy, but had many
difficulties with learning to write Xhosa and using “correct” grammatical constructions. She felt that
when she was speaking she was totally unaware of grammatical categories and did not know which
forms she was using when she constructed sentences. She did not mind if other speakers corrected
her spoken language and she felt that the amount of practice she had had during the year she had
studied Xhosa had improved her language structure and vocabulary. She was the only participant who
could hear and produce Xhosa tonology and could apply tones to new vocabulary if she heard the
vocabulary from Xhosa speakers. She classified her written work as “poor” because she based it on
her spoken language. She spent about half an hour a week outside the classroom practising grammar
and writing skills. Her final mark for the 1st year course was 60%.
Claire was very fluent in the interviews and always tried to say as much as possible. She frequently
commented on her lack of grammatical knowledge, but usually did not code-switch when she was
searching for an appropriate Xhosa word.
Karen
Karen, aged 19 at the beginning of the study, had studied Xhosa formally at school for seven years.
This formal instruction included two years of “basic” Xhosa with little focus on grammar in primary
school, and five years of Xhosa with a grammatical focus at secondary school. She decided to take
64
the university Xhosa I(NMT) and II(NMT) courses because she thought that it would be useful to
learn more Xhosa for communication and employment purposes, because she enjoyed the challenge
of learning the grammar of a new language, and because she had some background in it and felt that
this would make it an easy option as a university subject.
She enjoyed grammar and comprehension tests the most, but did not enjoy conversation classes. She
never engaged in conversation with Xhosa speakers outside the classroom and said that although she
knew this would be a good way to practise the language, she felt that she did not know enough
relevant vocabulary to hold general conversations and felt self-conscious about her lack of fluency.
She wanted to learn to speak better Xhosa, but her lack of self-confidence and the fact that she was
not usually forced to speak Xhosa meant that she did not make use of opportunities to interact with
Xhosa speakers. She preferred to write down an utterance to check her grammar rather than produce
it spontaneously. She felt that her comprehension skills were “fair” and that she could work out the
meaning of written texts fairly easily. She said her written errors were usually a result of direct
translation from English and she specifically mentioned that she found it difficult to produce correct
complex sentences and object concords.
She said that she spent very little time working on her Xhosa outside the classroom and would usually
do only the required homework for the week - “it gets pushed to one side”. She felt that she was not
particularly motivated, especially since her skills in reading and writing were not very useful outside
the classroom and she should rather be practising speaking skills. Nevertheless, she intended to
proceed to third-year Xhosa. Her final mark for the course was 84%.
Karen was a very hesitant and slow speaker in the interviews, although she tried to produced far more
complex utterances than any of the other participants. Her range of morphemes is also the widest and
she tended to use morphemes more correctly when compared to other participants.
Pat
Pat, aged 19 at the beginning of the study, was registered for the Xhosa II(NMT) course for the
second time, having failed it the year before. She had not studied Xhosa formally at school, but had
65
learnt to speak it before school on the farm on which she lived. However, when she went to boarding
school at the age of 6 she did not speak Xhosa except during the holidays, and she said she had lost
much of the speaking ability she had developed before school. She was studying Xhosa because her
cousins said they had enjoyed the course at Rhodes and she felt it would be good to “have another
language”. She spent about an hour a week outside class time doing assigned work and spoke Xhosa
to Xhosa speakers in shops and to Xhosa staff at a bar where she worked. She said that she could
usually work out what her interlocutors were saying, but that many people spoke very fast.
She felt that her speaking ability allowed her to “help out” when there were communication problems,
but that she would not be able to interpret a conversation. She also commented on the colloquial
nature of the language used in these situations, where she could code-switch when she did not know
a word, whereas the classroom required formal language. She said that outside the classroom she was
not inhibited about speaking Xhosa, but that in the classroom she was continually checking her
grammar, which made her a less fluent speaker in this context. She enjoyed conversation classes
where students were required to talk spontaneously. She said that she had good metalinguistic
knowledge and could always explain grammar rules in tests, but that she could not apply this
knowledge to the language which she produced, and thus her utterances were not usually correct.
Because there were usually many errors in her written work, she did not try to correct all of them but
might try to focus on a single error type to improve for the next time she wrote. She felt that
vocabulary was important for conversational skills and that grammar only became important when
writing. Her final mark for the course was 54%.
Pat was reasonably fluent in the interviews and seemed fairly self-confident when speaking. She code-
switched occasionally when conveying her message.
Sarah
Sarah, aged 20 at the beginning of the study, had studied Xhosa at school for ten years, and had
decided to register for Xhosa 3 because she felt that the workload was light. In addition, she enjoyed
studying Xhosa, she felt that Xhosa was relevant, and it would be useful for her future primary-school
teaching career. However, she felt that she had not really progressed much in her competence or
66
knowledge since grade 12, because there was very little pressure on students to improve and much
of the work was a repetition of school work - “I feel that all the Xhosa I know I learnt up to Matric
and that I’ve been sort of revising but less and less as the years go on and I’ve actually regressed I’d
say”.
She did only the assigned homework outside the class and never initiated conversations with Xhosa
speakers outside the classroom. She enjoyed literature classes because she felt that they helped her
to improve her reading and translation skills and she found them challenging. She did not feel that she
could speak Xhosa fluently, partly as a result of a lack of confidence to practise and partly because
there was too little pressure to improve speaking skills. She was worried about making errors and felt
that there was too much emphasis on correct grammar in the course, which impeded progress in
speaking ability. She felt that her understanding was “fine” and that she was better at writing because
there was time to think about grammatical constructions. She could sometimes hear different tones
but could not produce them - “the last thing I’m thinking of is tone [when I am speaking]”. She
obtained 64% for Xhosa 3 at the end of 1998.
Sarah was a fairly hesitant speaker, although she generally knew the vocabulary reasonably well. She
did not attempt to use a very wide range of morphemes when compared to Karen, although she had
had more years of formal exposure to Xhosa than the other participants.
3.2.2. Learning Context

The bi-weekly first-year grammar classes were observed throughout the data collection period, so
that appropriate vocabulary and grammatical constructions could be selected for the interviews. These
classes also gave me insight into the teaching methods used in these classes and the problems
experienced by the students.
First-year students attend two grammar classes and one class which focusses on cultural and phonetic
aspects of Xhosa per week. In addition, they are assigned to small groups which meet once a week
for a conversation class with a Xhosa speaker who teaches at a local school. These conversation
classes tend to be fairly structured and vocabulary, sentence structures and sentences are written on
67
the board. Dialogues may be written out and practised in these classes and there is usually little room
for spontaneous conversation, especially if the students are beginners.
At 2nd and 3rd year level, students have 5 and 6 classes per week respectively. These classes focus on
grammar, conversation skills, literature and phonology. These classes are much smaller than the first-
year class (four 2 nd and two 3 rd year students in 1998). More spontaneous conversation is encouraged
than in 1st year, although students are usually provided with written vocabulary lists for the topic
under discussion. It should be noted that these courses are primarily academic courses which focus
on the structures of Xhosa, and that communicative ability is of secondary importance.
3.2.3 Data collection

One of the most problematic features of most SLA research is that it seeks to investigate the
competence of the learner when researchers have access only to the performance of the learner.
Although some data collection procedures, such as grammaticality judgements, are claimed to access
competence reliably (Felix 1984; Sorace 1996), the claims are far from proven. As a result, Lalleman
(1996:13) concludes that
(a)nalyses based on intuitions do not seem to be more reliable than analyses based on
spontaneous speech data, and the best way to arrive at the right conclusions is to take
different types of language material of many types of learners into account ... .
My study follows Lalleman’s approach by including data from a range of different sources and
activities (see section 3.2.3.3.2 for further discussion).
The data in my study does not come from naturalistic interactional contexts, but at the same time it
is not as highly controlled as in an experimental setting. A typical feature of the experimental setting
is that only a single or a few aspects of the IL are studied (Bialystok 1990), whereas my study aims
to describe a range of elements in the data. In addition, the more controlled the experiment, the more
difficult it is to know whether the reflection of competence is genuine or whether it is an artifact of
the research situation. On the other hand, the more naturalistic the situation, the harder it is to control
for a myriad of influential factors in the SLA process. A further consideration in rejecting data
68
collection in entirely naturalistic interactional contexts in my study was that the participants rarely
(and never in some cases) conversed in the target language (TL) in natural situations which meant
that obtaining such data would have been virtually impossible. For this reason, each participant in my
study attended seven structured interviews which consisted of a variety of elicitation tasks over a
period of eight months.
3.2.3.1 Interviewer
The interviewer (Thembi) was chosen because she is a L1 Xhosa student and her English competence
is good. Five of the six interviewees were female, and since I felt that it would be desirable to match
the sex of the interviewer and interviewee as far as possible, a female interviewer was chosen. She
was also close in age to all the participants, and although she did not know any of the interviewees
at the beginning of the year, she established a relatively relaxed relationship during interviews with
all of them over the course of the year. I was not present at the interviews as I wanted the data to be
as spontaneous as possible, and I felt that my presence would be intrusive for both the interviewer
and the interviewee. However, I was entirely responsible for the formulation of elicitation exercises.
Thembi was approached to conduct the interviews at the beginning of the year in which the data were
collected and a pilot interview was conducted so that guidelines for interview conduct could be
clarified. We met to discuss the interview material before the beginning of the week in which
interviews were conducted. She was instructed to attempt to present the material in as uniform a way
as possible across all the interviews. Generally, she succeeded in this aim and conversations were
usually started in the same way with the same questions used to elicit data from interviewees.
Nevertheless, Thembi was given the freedom to adjust the vocabulary and structures she used
according to the competence levels of the participants, which helped to further interaction and ensure
more natural conversation. She was encouraged to prompt the interviewees to try and discover
meanings by repeating utterances and/or using alternative vocabulary and only to translate into
English for them as a last resort. As the year progressed, she increasingly participated in the
conversations as a conversational partner rather than as an interviewer, so that the artificial situation
of a series of questions directed at the interviewee could be avoided to some extent.
69
3.2.3.2 Data capture methods

The interviews for my study were tape-recorded. Phonetic transcriptions of spoken data were not
made since the phonetic level of IL was not the focus of the study. Instead, there was a fairly simple
transposition of speech into writing, with repetition and false starts marked. Thembi checked all
transcriptions after I had transcribed the interviews, to ensure greater reliability.
Tape-recording may be problematic for a number of reasons including hesitance on the part of the
subject, lack of clarity of the recording and the lack of extralinguistic cues available to the analyst
(Johnson 1992; Seliger & Shohamy 1989). However, recordings usually allow for a fuller set of data
than note-taking and since the interviews were conducted by someone other than the researcher, tape-
recording was regarded as the most feasible way of capturing data3.
3.2.3.3 Data types

The primary data comes from the interviews conducted over an eight-month period. Each of the
participants attended seven interviews from March to October and interviews lasted between 35 and
50 minutes each. Five different sets of interviews were used so that two of the interviews were
repeated. In this way, morphological development demonstrated in the same tasks over time could
be assessed more accurately. Table 3.1 sets out the interview schedule.
Video-recording was considered, but the presence of a video camera was deemed too
3
intrusive in a context where subjects were participating in what was, for them, a fairly infrequent
conversational encounter in the TL.
70
Table 3.1 Interview schedule

Interview number Time elapsed since last interview Additional comments
1 0 First interview held 3 weeks after start of
course
2 5 weeks Interview held after one-week vacation
3 4 weeks
4 10 weeks Mid-year examinations and vacation
between interviews 3 and 4
5 4 weeks Repeat of the 2nd interview
6 4 weeks One-week vacation included between
interviews 5 and 6
7 4 weeks Repeat of 3rd interview
3.2.3.3.1 Interview Content

The interviews consisted of eight different tasks, with task types repeated in the same order in each
interview. Each interview consisted of two conversations, one joint-storybuilding or role-playing task,
one picture-description task, one translation from Xhosa to English and vice-versa, one dictation task,
sentence-building tasks, sentence-manipulation tasks (e.g. change the sentence from positive to
negative), and a grammaticality-judgement task (interview material is included in Appendix C).
Interview material was the same for all subjects in order to exploit the benefits of the quasi-
longitudinal approach.
Seliger & Shohamy (1989:183) say that elicitation tasks should focus on getting data “where attention
is on the meaning of the utterances rather than on grammatical forms, and where the task constrains
speakers to produce the language which the researcher is interested in obtaining”. For some of the
tasks, this was achieved, although some tasks seemed to elicit greater focus on form than meaning
(e.g. translation).
Interview content was selected to match the goals of the elicitation tasks, which were:
C to generate spontaneous language data (conversation, story-building, role-playing, sentence-
building tasks);
71
C to test underlying competence (grammaticality judgement, dictation);

C to elicit examples of particular morphemes which would be obligatory in the context (sentence
manipulation, translation);
C to investigate development of morphology over time; and
C to investigate variation across tasks.
As different tasks have been shown to influence the IL produced (Selinker 1992; Tarone 1982, 1983,
1987), the tasks were repeated and the interview structure was kept as uniform as possible so that
assessments of morphological development could be more reliable for each task. Different tasks may
affect both fluency and accuracy (Ellis 1987b, c), since carefully controlled production may be more
accurate and less fluent, whereas more spontaneous production may be more fluent but less accurate.
A range of tasks can also limit the problems associated with the kind of data which learners present.
Corder (1981) and Extra & van Hout (1996) note that learners may only demonstrate a small part
of their repertoire which relates to what they feel most confident about in spontaneous conversation.
By artificially “forcing” learners to use certain morphemes in certain contexts (e.g. in the sentence-
manipulation task), it may be possible to investigate the underlying system more thoroughly.
3.2.3.3.2 Elicitation tasks

My study uses elicitation task-types commonly used in a variety of other studies. It relies to some
extent on the principles (rather than the elicitation tasks) used by the European Science Foundation
(ESF) study, which used “free conversation, pre-structured play acting, commenting on or retelling
video fragments of films (in particular silent movies), and small-scale experiments” (Extra & van Hout
1996:95). Experiments in the ESF study included translation, picture-naming and verb completion
and a few, less structured activities which had highly structured topics (e.g. giving directions). Extra
& van Hout (1996:97) note that the variety of exercises and the repetition of interviews “offered ...
both variation and structure”. It was felt that if results were to be compared with the ESF study, there
would need to be some points of similarity, although a range of other tasks was also included to
facilitate comparison with other studies.
72
Spontaneous conversation
Spontaneous conversation in natural conversation contexts is said to be the ideal IL to study, as it is
the “purest” manifestation of competence, primarily because there is minimal monitoring (Faerch et
al. 1984; Towell & Hawkins 1994). Studies using spontaneous speech as data include Extra & van
Hout (1996), Giacalone Ramat (1992), Tarone (1983), and Wode (1981). Sharwood Smith (1994)
argues that spontaneous speech better represents the underlying system than writing because there
is less monitoring, and it therefore allows more access to “automatized and implicit IL knowledge”
(Faerch et al. 1984:300). However, the availability of spontaneous conversation data is usually limited
by ethical problems. With my group of participants, it is also usually limited by their lack of
engagement in spontaneous Xhosa conversation. The interview format is therefore employed to
stimulate spontaneous speech in my study.
Bell (1984) claims that topic and topic familiarity are important variables to take into account when
studying IL utterances, and for this reason, two conversations with different topics were initiated in
each interview (see also Selinker & Douglas 1985; Veronique 1987). The primary criterion for topic
selection was the availability of vocabulary items which might be known to the participants.
Vocabulary and topics were selected by noting the vocabulary items which were introduced in the
first-year class4.
Story-building and role-playing

The conversation tasks tended to be rather one-sided and unnatural, especially in the early interviews,
because the interviewer usually asked questions and the interviewee answered these questions without
much elaboration. Faerch et al. (1984) mention role-playing as an effective elicitation method because
it requires more equal input from both participants. In the story-building task, the interviewer and the
interviewee together created a story on a given topic, each providing two to three items of
information per turn. In these tasks, the interviewee would be forced to abandon some of the
formulaic utterances which were prevalent in some of the conversations and jointly construct a
4
It was assumed that 2nd and 3rd year students would be familiar with vocabulary introduced in the first
year.
73
dialogue which was perhaps less predictable because of the greater contributions of the interviewer.
Picture Description
Picture-description tasks have been used widely in IL elicitation, mainly because it is believed that the
learner is inclined to focus more strongly on the pictorial elements and less strongly on the form of
the language that is produced (e.g. Bialystok & Frohlich 1980; Ghrib 1987; Giacalone Ramat 1992;
Sánchez & José Giménez 1998; van Els et al.1984; Váradi 1980; Yang & Givon 1997). As far as
possible, care was taken to use pictures relating to vocabulary which would be familiar to participants
in my study.
Translations
Studies using translations include Hölscher & Möhle (1987), Swain et al. (1974), and Yang & Givon
(1997), but the use of translation as an elicitation task is controversial in SLA theory. On the one
hand, some believe that translations adequately measure comprehension, lexicon, grammar and
transfer from L1 to L2 (Seliger & Shohamy 1989). On the other hand, others feel that translation
does not actually measure L2 proficiency because it is a highly specialised skill which can improve
with practice. Students become used to focussing on accuracy without much recourse to the L1
(transfer) when doing translation exercises (Sharwood Smith 1994). Nickel (1989) also found that
there were not necessarily more L1 transfer errors in translation than in other activities, despite a
belief that learners would focus on the surface form when translating. Dittmar (1981), who used an
oral translation task, confirmed the findings of other studies that learners did not produce IL
morphology. This finding confirms the view that translation tasks may not elicit a different form of
competence from other tasks.
The English-to-Xhosa translations in my study required oral translations of written stimuli. The
sentences for translation deliberately included grammatical constructions which students had been
taught in the formal situation and vocabulary which should have been familiar to them. Translation
tasks have the dual function of investigating the learner’s ability to recognise the functions of
morphemes in the L2, and the ability to use the morphemes in the grammatical context. In these tasks,
the difference between knowledge of function and form is particularly evident.
74
The translation task requiring translation from Xhosa to English in my study used a read-aloud
technique. The Xhosa passage was read aloud by participants and then translated into English. The
use of a read-aloud technique can illuminate competence as learners may read what they believe
should be there, rather than what is in fact there (Faerch et al. 1984; Schmidt cited in Young 1991;
Seliger & Shohamy 1989). This task did not work particularly well in my study because interviewees
read too slowly and often with too little comprehension to enable an adequate assessment of their
competence. For this reason, the reading data is not included in the analysis. The translations into
English served to illuminate comprehension of the meaning and function of morphemes, and for
learners with low levels of competence it showed where morphemes were still unknown. At times,
“wild” guessing in an attempt to match expected utterances with the context interfered with the
reliability of the data.
Dictation
Language-testing research (Oller 1979) has shown that dictation tests may allow the analyst access
to competence (see also Long & Sato 1984). When writing down a dictated text, the learner is
inclined to write down what he/she thinks should be there, rather than what has actually been said.
Dictation-test results in my study were used as comparisons with the results of oral tasks. One of the
problems with this task in my study was that participants had varying degrees of familiarity with the
vocabulary used. More familiar vocabulary was usually accompanied by more accurate use of
morphemes, whereas unfamiliar items were accompanied by poor use of morphology. Nevertheless,
the results of the dictation tests usually support the findings from the oral data from subjects.
Sentence building
Sentence-building tasks require students to construct sentences for each word from a list of words
given to them. This task requires the production of full sentences with the appropriate morphemes.
Students can choose vocabulary with which they are familiar, but the danger is always that they will
choose to reproduce formulaic utterances which do not shed light on their morphological
development.
75
Sentence manipulation
Sentence-manipulation (or stimulus modification, James 1998) tasks focus the attention of the
interviewees on particular grammatical morphemes (Long & Sato 1984). Studies by Ghrib (1987) and
van Els et al. (1984) use this elicitation method. The task requires them to produce these morphemes
in given sentences or to make lists of words into sentences using a particular morpheme (e.g.
negatives).
In my study, the sentences which needed to be manipulated were either available orally from the
interviewer or in written form. Oral forms required the student to remember the given sentence while
performing the task, while the written forms were usually easier as they required only the
manipulation of one or two components of the sentence.
Grammaticality judgement
The purpose of the grammaticality-judgement test is to test metalinguistic awareness, which is said
to test competence (Seliger & Shohamy 1989; Sorace 1996), although this claim is controversial.
Examples of studies using grammaticality judgement include Duffield et al. (1998); Tarone (cited in
Towell & Hawkins 1994), and Trahey & White (1993). Grammaticality-judgement tests have been
conducted in a variety of ways and different types of test include:
C oral production of test sentences or elicited imitation (Munnich et al. 1994);
C a selection of sentences which can be graded for acceptability (Chaudron 1983; Sorace 1996);
C a selection of sentences of which the grammatically incorrect or correct sentence has to be
indicated (Sorace cited in Lalleman 1996); and
C a selection of sentences containing errors which must be corrected (Seliger & Shohamy
1989).
Some of these tests require the correction of the sentences, while some require reasons to be given,
verbally or in written form, for why the sentence is unacceptable. The grammaticality test used in my
study required the student to correct errors in written sentences. A few distractor sentences which
contain no errors were used to test hypercorrection (Sajjadi & Tahriran 1992) and therefore ensure
greater reliability in the findings.
A grammaticality judgement which merely asks the learner to decide whether the sentence is correct
76
or incorrect is subject to problems because the more proficient the learner is, the easier it is to decide
on a grammatical sentence. However, an increase in proficiency does not necessarily correlate with
an ability to decide on grammaticality of a sentence, since the learner may conclude that it is
ungrammatical because he/she lacks experience of the sentence structure (Lalleman 1996). Requiring
interviewees to correct ungrammatical forms in my study avoided some of these problems.
Grammaticality-judgement tests as indicators of competence have numerous detractors. Munnich et

al. (1994) feel that the grammaticality-judgement test does not elicit actual linguistic behaviour in the
L2, but rather metalinguistic judgements about beliefs about the system (see also Gass 1994). In this
case, there may be a mismatch between the underlying linguistic system and the system of knowledge
about the underlying system. Birdsong (1989) concludes that grammaticality judgements are too
metalinguistic, explicit and unreliable. Furthermore, the link between competence and performance
may not be direct (Ellis 1994), since data may be affected by performance constraints. In this case,
processing rather than grammaticality may be the problem (Schachter 1989). Similarly, Paradis et al.
(1998:238) argue that “metalinguistic and explicit knowledge might interfere with the on-line
processing desired in the procedure”.
Written Work
The mode of discourse (oral or written) may have an influence on the language produced (Tarone
1987). For this reason, a sample of the exercises, essays and dictation tests written during the course
of the year by each student was collected so that spoken and written data could be compared. A
problem with the use of these exercises and essays for comparative purposes is that students have
access to charts of concords, dictionaries and other references when completing these assignments,
and they are probably not good indicators of actual competence with regard to function, and
particularly to form. Nevertheless, written work has value in terms of showing whether students know
and can manipulate linguistic rules and morphemes (possibly with the support of materials).
3.3 ANALYSIS PROCEDURES

Data was analysed from two different perspectives in my study. Firstly, a form-focussed approach
was employed using error analysis (EA), so that general trends with regard to morpheme use in the
77
data could be assessed. Secondly, a function-focussed approach was used to look at some of the
functions which the learners were able to perform in their IL. This latter approach enables a multilevel
analysis and it can be used successfully where the range of forms is limited. Morphemes used in the
TL may not be available to the learner, but the function may be carried out with the aid of other
strategies and so a functional approach provides a more comprehensive account of development. It
was hoped that the use of these two analysis procedures might throw light on patterns in the
development of morphemes, idiosyncratic usage of morphemes, as well as the ways in which
communication was achieved.
3.3.1 General principles of IL analysis

Analysis of IL needs to be carried out very carefully and in a cyclical fashion so that patterns can be
identified and genres (or tasks) can be compared (Johnson 1992). Difficulties may be encountered
because of the transitional nature of the system, the presence of idiosyncratic rules and the importance
of taking the L1 into account (Jagtman & Bongaerts 1994).
Corder (1981) advises that every utterance should be seen as idiosyncratic until it can be shown that
it is not. In fact, Selinker (1992) cautions that an analyst should not accept the first plausible
interpretation, but should continually look for alternative explanations. Furthermore, Corder (1981)
suggests that the analysis should be based on inferring the intended meaning of the learner before
conclusions can be reached about the presence or absence of error. In support of this view, Douglas
& Selinker (1994:121) say that “primary interlanguage data are always ambiguous; at a minimum, one
needs to gain access to interlanguage intention” and that the learner’s perspective needs to remain
at the forefront of any analysis. Perdue (1993a:13) cautions that
the nearer a learner’s production appears to be to the TL, the more tempting it becomes to
imagine a ‘corresponding’ TL version and use the analytic categories relevant to the latter
version to analyse the former. The ‘closeness fallacy’ ... is insidious and ubiquitous, and leads
inevitably to false dichotomies such as ‘error/non-error’.
Taking the above cautions into account, it is evident that a consideration of the context is a vitally
important aspect of the analysis (see also Selinker 1984). The discourse level is particularly important
here, and Long & Sato (1984) provide an example of an analysis of the use of the past tense where
78
looking at morphological indications of past tense is inadequate because the past time reference may
be established early in the discourse by non-morphological means. Isolated sentences may not
therefore provide the full picture.
Furthermore, personal characteristics of the learners need to be taken into account during analyses.
For example, a greater willingness to take risks may result in a greater amount of speech containing
more errors, whereas another learner may not make many errors but speak slowly, hesitantly and
produce only small amounts of speech (Klein 1986). This factor was clearly evident in the data of my
study.
3.3.2 Analysis of IL as a system

The systematicity of IL has been widely documented (see section 2.5.2), but in a quasi-longitudinal
study, systematicity may be difficult to determine. Adjémian (1976) and Selinker & Douglas (1985)
argue against the notion of a single system, and they prefer to posit a variety of different distinct and
coexistent systems which change over time and according to context (see also Faerch et al. 1984).
Transition is also systematic, so there is both horizontal (or synchronic) and vertical (or diachronic)
systematicity (Klein 1986; Perdue 1993a). A different view is that there are degrees of systematicity
(Labov cited in Young 1988) which need to be discovered for different areas of the data. In addition,
it should not be forgotten that it is the system apparent in performance, rather than competence,
which is actually analysed (Young 1988).
3.3.3 Units of analysis

Choosing a unit of analysis is problematic in IL studies. Crookes (1990:185) discusses a variety of
possibilities. The first option is the turn, defined as “one or more streams of speech bounded by
speech of another, usually an interlocutor”. A second possibility is the utterance, which is defined by
Crookes & Rulon (cited in Crookes 1990:187) as
... a stream of speech with at least one of the following characteristics:
(1) under one intonation contour ...
(2) bounded by pauses, and
(3) constituting a single semantic unit.
79
For the purposes of my study, the turn was often too large a unit of analysis, especially in some tasks
such as translation and picture description where the learner might provide information using a range
of unrelated sentences. Rather, the utterance was the preferred unit and the second and third criteria
mentioned by Crookes & Rulon above were particularly useful in determining an utterance in my
study. Furthermore, utterance length may be a better indicator of development than turn length which
depends much more heavily on the task and topic, since very short turns may be all that are required
in a conversation, for example.
3.3.4 Characteristics of formulaic speech

The IL of the participants contained varying amounts of formulaic speech, but identifying formulaic
utterances was not always an easy task. Examples of forms which are likely to be formulaic in the
early interlanguage of learners of English include: “How are you?”, “What is your name?” Bahns et
al. (1986) argue that identification is mainly intuitive, but there are other characteristics which can
be used to identify these speech forms.
Myles et al. (1999:50) believe that “overextension of these utterances in use (whether syntactically,
semantically, or pragmatically)” is one way to identify a formulaic utterance. The utterance is usually
used invariantly and fluently in particular contexts and it may be more complex or longer than other
forms which appear in the learner’s language (see also Weinert 1995). The learner may not be able
to substitute other forms into formulaic utterances and they are therefore used inappropriately. Myles
et al. (1999) find instances in their data where the pronoun reference needs to be changed from first
to third person if the utterance is to be correct, but the learner can only produce the formulaic
utterance which contains the first person. A further indication of a formulaic utterance is that the
language is usually “well formed” and “grammatically advanced compared to the rest of the learner’s
language” when a formulaic utterance is used (Myles et al. 1999:52). The primary function of
formulaic utterances seems to be to enable communication, but they may also be used to deduce the
patterns of the language (Pawley & Syder 1983; Wong-Fillmore cited in Myles et al. 1999).
3.3.5 Form-focussed analysis

The form-focussed analysis employed in this study takes the form of an EA. The approach has several
80
benefits, which include:

C being useful for discovering regularities at the morphological level;
C usefully showing which level of development has been reached; e.g. pragmatic, syntactic,
finite, non-finite, etc.;
C allowing one to take a comparative approach by comparing which morphemes are used by
learners and L1 speakers; and
C being able to determine sequences of development.
An overall picture of development can be gained by calculating the ratio of errors to the number of
words in the text (Faerch et al. 1984). However, Ellis (1989:306) strongly criticises a form-only
approach to determining acquisition orders because functions of morphemes are ignored, and
“accuracy and acquisition orders cannot be equated”. He goes on to argue that “there seems to be
no theoretical basis for comparing the acquisition of one feature (e.g. articles) with another (e.g.
auxiliary, be) which comes from a totally separate sub-system of the grammar of the language” (Ellis
1989:306).
3.3.5.1 Error analysis procedures

The first stage of the EA consisted of what James (1998:19) calls the “broad trawl”. This initial
analysis enables the researcher “to gain a first impression of the learner’s capacities and limitations,
to identify the areas of TL competence where they are most susceptible to error” (James 1998:19).
Part of the study consists of a description of the errors and James (1998) lists three reasons for
describing data:
C the need for labels so that intuitions can be compared with those of other people;
C the need to count errors; and
C the need to create categories of errors which can be compared intralingually and cross-
linguistically.
The focus of the EA is on morphological development, and Young (1991) feels that the use of
inflectional morphemes can be quantified fairly easily by categorising them in any of the following
ways:
C free or bound;
C easy or hard;
C one-to-one or one-to-many form-to-function relationships;
C likely to transfer or not;
81
C NP or VP morphemes; and
C single morpheme versus several allomorphs.
Gregg (1989) argues that one needs a well-defined set of categories before valid categorisation can
take place. In my study, data were categorised according to TL morpheme categories. Further
analysis occurred at the utterance level so that regularities of morpheme pairing could be noted. It
was also necessary to include multiple utterances in the analysis, since some morphemes were used
or omitted over a larger discourse unit. Correct use of the TL morpheme was noted as well as deviant
usage. Following van Els et al. (1984), usage is divided into three categories:
C no morpheme supplied;
C misformed morpheme supplied; and
C correct morpheme supplied.
There were also a few examples of inappropriate suppliance of morphemes, but this category did not
occur often enough to warrant setting up a separate category in most cases. The norm used when
deciding the status of a form was a written TL norm, as this was the variety most frequently
encountered by most of the participants.
The second stage of an error analysis is where one counts the number of obligatory contexts and then
calculates suppliance of the morpheme according to van Els et al.’s (1984) categories noted above.
Problems with this approach include instances where the learner avoids the structure. This is
especially problematic when comparing learners where one learner uses the form and another learner
avoids the form (Tarone 1987).
Error analysis as an analysis procedure has a number of shortcomings (see section 2.2.2), but it is
nevertheless a useful starting point for a description of the data (James 1998). In my study, results
of the EA are tabulated according to task and interview sequence so that developmental trends can
be investigated according to the two dimensions of task and time.
3.3.5.2 Problems with error analysis procedures

The use of TL labels on IL data is problematic, as Pienemann (1992) points out, especially if it is
82
assumed that IL is a system in its own right. However, no studies have completely reinvented
terminology which will effectively account for IL data, because IL categories have not been
determined. Broeder et al. (1993a, b), who worked with the corpus from the European Science
Foundation (ESF) study, count the number of words in the data and assign a word class category and
place of occurrence to each word. However, this approach is problematic in the sense that
identification of a word may be problematic (especially in spoken data), and placing it in a category
may be even more controversial (Broeder et al. 1993a). Nevertheless, for the purposes of analysis,
some level of categorisation needs to take place and I have assumed that there is some regularity in
distributional categories and that learners have some notion of a Xhosa word, although it may not be
the same as in the TL.
A further problem in an EA is deciding on what counts as an ‘obligatory context’ of use for a

morpheme. Tarone (1987:42) suggests that “an ‘obligatory context’ for any language form is a
linguistic context in which NSs of the TL would be obliged to supply that language form in order to
produce a grammatically correct utterance”. Superficially, this seems to be a fairly easy task, but it
is also necessary to look more broadly at the context so that alternative forms can be assessed. In my
study, the EA was performed twice: once by myself and once by a Xhosa mother-tongue speaker. The
analysis took place independently and results were discussed so that reasons for any discrepancies
between the analyses could be noted.
Another difficulty lies in trying to decide when a learner has acquired a morpheme. The 90% correct
use in obligatory context criterion has been widely used, but it is not entirely satisfactory. Pienemann
(1992:89) advises setting up additional acquisition criteria to measure developmental stages:
Consider the following example: a quantitative analysis reveals that a plural marker (e.g.,
English ‘-s’) is used 67% of times in plural contexts in a given sample. The researcher might
... decide that any frequency in the occurrence of a morpheme alone does not tell whether it
is being used productively. S/he may therefore decide to study lexical and morphological
variation in the use of this morpheme. In other words, s/he may wish to investigate how many
different lexical items the plural marker occurs with, to what extent the same lexical items
vary morphologically, and how many lexical items do not occur with a plural marker in a
plural context.
83
A final aspect to bear in mind is that generalisation of results across tasks can be problematic. In my
study, tables which mix results from different tasks and tables which separate results from different
tasks are provided.
3.3.6 Functional analysis

As a focus on errors in a classic EA paradigm is based mainly on form, a functional approach is also
useful. The functional approach allows one to investigate influences at levels beyond the
morphological level, which can give insight into the choices made at the morphological level. The use
of a restricted repertoire of forms to perform a wide range of functions is also more noticeable in a
functional analysis of the data. Many studies have focussed on order of acquisition, but this may not
be a true picture of the underlying processes of acquisition. A more useful and explanatorily adequate
analysis would take functional uses into account as well so that systematicity could be observed
(Rutherford 1984).
In order to overcome some of the shortcomings of a form-only analysis where many interesting
results and patterns of development are hidden, there are two possible functional analysis types: form-
function and function-form. Form-function analysis is defined by Long & Sato (1984:265) as “a
comprehensive analysis of the functional distribution of a particular form in a learner’s IL”. On the
other hand, function-form analysis “begins with a functional domain, such as the expression of
temporality, and documents the evolution of grammatical encoding of that functional domain” (Long
& Sato 1984:265). Bailey (1989:295) believes that both form and function need to be studied since
“nothing in language learning is purely form related or purely meaning related” and at different times
each may assume greater priority in the learning process. A form-function approach requires multiple-
level analyses, a focus on process rather than product, and a focus on context rather than individual
utterances (Extra & van Hout 1996).
3.3.6.1 Function-to-form analysis

One kind of function-to-form analysis starts at the genre (or task) level and then seeks to discover
the forms used in the genre (or task). This is motivated by a belief that the discourse functions
determine the linguistic forms which are used (Kumpf 1984). The analyst can then start to look at
84
how, for example, tense, aspect, modality and reference are realised in the discourse across utterances
at first. A further stage of development is when the functions are gradually found within utterances
as the language grammaticises (see Sato cited in Long and Sato, 1984, for an example of this
approach).
Sato (1990) investigates the development from parataxis (or the pragmatic level) to syntacticisation
in a function-to-form analysis of past tense development in two Vietnamese children learning English.
She found extensive use of parataxis but little development in syntacticisation, although more lexical
past verbs were encoded as the children’s IL developed. Although she expected linear development,
this did not happen (as Meisel et al. 1981 predicted). Development is slowed down by the availability
of discourse-pragmatic means of indicating past time reference, and phonological constraints play a
role in slowing down morphological development for the two learners in Sato’s study.
3.3.6.2 Form-to-function analysis

In my study, one would expect Andersen’s (1984) One-to-One Principle to be relevant, especially
since Musau (1995) demonstrates that this principle operates for learners of Swahili who reduce
morphological irregularities to single forms (see section 2.5.5.3.2). At early stages of development,
the learner prefers to use a single form for each function rather than multiple forms for a single
function. A form-to-function approach is thus relevant in my study.
Examples of function-to-form and form-to-function studies include Wagner-Gough (cited in Sato

1990) who studied the acquisition of -ing and found non-target-like and variable functional usage of
the morpheme. Huebner’s (1983) study of a Hmong speaker’s English L2 development takes a
functional approach. Other studies include Schumann’s (1978) study of a Spanish speaker learning
English, and Dittmar’s (1984) reanalysis of Spanish adults learning German. Karmiloff-Smith (1979)
looks at the acquisition of articles by French speakers and Pfaff (1987) investigates the determiner
and reference systems of Turkish learners of German using this approach. The ESF study (Perdue
1993a) also takes a functional approach.
Mitchell & Myles (1998) conclude that functionalist approaches have promoted systematic analysis
85
of a range of functions which have helped to describe IL and that their focus on the discourse and
pragmatic levels has been particularly illuminating. However, they tend to focus on early language
development and more studies are needed which investigate development of later and more complex
areas.
3.3.6.3 Functions investigated in my study

Three functional areas were investigated in my study. These were temporal reference, spatial
reference and pronominal reference.
Other studies (e.g. Bhardwaj et al. 1988; Dietrich et al. 1995; Meisel 1987; Trévise 1987) have found
that temporality is encoded:
C pragmatically;
C at the word level by the use of adverbial elements;
C by tense morphemes, to some extent in later development.
Spatial relationships are encoded non-verbally with the use of gestures as well as by the use of
adverbial elements (Carroll & Becker 1993; Klein & Perdue 1993; Perdue 1990). Reference can
range from repetition of the NP to pragmatic means to pronominal forms (Broeder 1991; Kumpf cited
in Long & Sato 1984; Perdue & Klein 1993; Pfaff cited in Rutherford 1984).
These functions were chosen because they had been studied before and the findings of my study could
therefore be compared to the findings of other studies. In addition, they seem to be primary functions
which need to be performed in the elicitation tasks. Since there are a variety of means of expressing
the functions, it is also possible to trace the development of these functions across tasks and over
time.
3.4 LIMITATIONS OF THE STUDY

The limitations of my study include several which are theoretically problematic in many other studies
in the IL field.
86
3.4.1 System qua system

One of the problems with IL analysis most frequently raised in the literature is the notion that IL
needs to be studied as an independent system, rather than a partially correct form of the TL. As
Selinker (1992:43) says:
Units of linguistic structure have linguistic ‘value’ only in terms of the place of these units
within a linguistic system and the constraints of such units within that system, while at the
same time individuals in a language contact situation clearly violate, on a regular basis, such
constraints.
Bley-Vroman (1983:6) describes the tendency to describe IL in terms of the TL system as the
“comparative fallacy”. Even IL studies have not managed to steer clear of the comparative fallacy,
although initially this was a seminal difference distinguishing them from other studies (Rutherford
1984). Analysing IL without reference to the TL is very difficult because it requires one to ignore
completely the response that one has when one sees an “error” in terms of the TL. Particularly in the
case of my study, where the learners seem to be exposed to a formal classroom norm and seem to
experiment very little outside the confines of the classroom, it is tempting to say that much of their
output is based on memorization and that very little in the way of independent hypothesis-making
and hypothesis-testing is taking place. Yet, there is evidence to suggest that this is occurring and it
is necessary to describe the system in its own terms.
3.4.2 Competence/performance
Discovering the competence of a learner using performance as the data is a problem with studies in
the IL paradigm. The reliance on performance data is a result of the impossibility of identifying and
studying competence. Performance may be variable while competence may not necessarily be
variable, and performance may not reflect what is in the underlying system (Swan 1987). However,
there are different views on the homogeneity of competence, and Tarone (1987) feels that there are
a variety of competences or capabilities rather than the single homogenous one which Chomsky
posits. Furthermore, although the competence/performance distinction is widely accepted, Seliger &
Shohamy (1989) question whether the distinction needs to be made for L2 acquisition or whether it
holds only for L1s.
87
3.4.3 Determining the norm

As mentioned above, during analysis it would be ideal to view the IL produced by the learner as a
system which could not be compared with forms in the TL. Furthermore, one would seek an “IL
grammar” made up of its own units rather than using units from the source language (SL) or target
language (TL) during the analysis process. Unfortunately, no adequate grammar of learner language
exists and work on interlingual identifications is still insufficiently developed. Consequently, most of
the analysis of IL relies on using grammatical categories which are more closely associated with fully-
fledged languages and the IL is implicitly or explicitly compared to the SL and the TL to some extent
(Faerch et al. 1984).
The problem with this approach is that a TL norm has to be provided. Languages always have a range
of styles and registers which may affect lexical and syntactic choices (Marshall 1989; Valdman 1989).
The learner needs to be able to control a range of language used appropriately in context to achieve
the goal of native-like competence (Bialystok 1998; Brown 1998), since what is acceptable in one
context of language use may not be acceptable in another context.
The reference form is usually an idealised form of the L2 or the L1 unless baseline data from L1
speakers is used (Selinker 1992). Norms generally used in IL studies tend to be assumed rather than
actual (Gass & Selinker 1992), and they tend to be based on written corpora (Alexander 1979).
However, using a TL norm may not be as problematic as some theorists claim since the learner may
well be comparing his/her forms with those of the TL. In fact, Jake (1998:341) posits the “target-
language principle” which says that “to the extent possible, construct the IL from TL lexical
structure”. A further relevant point is that these reference points for the learner may not be actual TL
norms but may be perceived TL norms against which the learner compares his/her current system
(Selinker 1992).
Judging norms is further complicated by the fact that native speakers may make errors although they
can usually correct themselves (Faerch et al. 1984:382). In addition, there may be some dispute with
regard to what the norm is, especially when using grammaticality judgements (Singh et al. 1982).
However, James (1994) says that native speakers are reliable informants.
88
One argument (Cook 1991) is that learners should not be judged by TL standards as they do not
speak this variety of the language, but that they should rather be judged according to their own
standards. James (1998) argues strongly against this view because he feels that there needs to be an
external rather than an internal norm with which to judge, otherwise no judgement can take place.
In order to resolve this conflict, Klein (1998) and Larsen-Freeman (1998) agree that learner data
should be analysed both from the perspective of a TL norm so that it can be understood why learners
fail to reach the norm and from a systematic perspective, where the data is analysed in its own right.
The use of a form-focussed approach and a functional approach in my study allows this dual analysis.
In my study, the choice of norm is further complicated by the relatively large differences between the
form of language used in the classroom and the forms of language produced in the wider community
of speakers (Gough 1995). The language of the classroom is formal, conservative and relatively
“unsullied” by the language of the wider community from the point of view of the purist. It is the
language of writing reflected in textbooks and older texts, but does not reflect everyday spoken
Xhosa in many ways. The urban spoken language tends to be influenced extensively by other
languages, although the influence is mainly in the realm of vocabulary. Nevertheless, the spoken
variety is much less formal and there are certain constructions which are necessary in the written form
which are not used in the spoken form.
For four of the learners who participated in the study (Ann, Ben, Karen, Sarah), I believe that it is
acceptable to use the formal language of the classroom as the norm since they have very little input
from speakers of Xhosa. Pat has received input from both Xhosa speakers and the classroom, and
deciding on a norm for her language is more difficult, although her exposure is primarily to the
written norms. Claire’s primary input has been from mother-tongue speakers of the language in a
rural area and her exposure to the formal norms of the classroom is limited. However, rural varieties
tend to be closer to the written norms than urban varieties so, although not ideal, it is possible to
apply the written norm to her speech.
3.4.4 Classifying an error

Deciding on what is an error and what is a mistake can be difficult, despite the various definitions of
89
the two given in the literature (see section 2.2.3). Also, there is the “knock-on” effect of an error in
the remainder of the utterance; e.g. if the noun class of the Xhosa noun is supplied incorrectly in the
subject, it will influence the form of the subject concord of the verb which must agree with it. The
problem is then to decide whether this constitutes two errors or one. In my study, I take the view that
this constitutes a single error since, although the noun class is incorrect, the form of the subject
concord is correct for the verb because it agrees with the noun class of the noun with which it is co-
referenced.
Using the categories of the TL is also problematic because one cannot be sure that the learner
understands the functions and forms of the categories. Avoidance may have occurred, or the learner
may overuse a single form rather than a range of forms which might be more like TL usage. In
addition, the use of a correct form does not always indicate understanding of the system since the
form may have been produced correctly as a result of an automated, frequently heard form (or even
a lucky guess in some cases).
3.4.5 Different levels of competence

The different levels of competence of the learners in my study is both a disadvantage and an
advantage. The main disadvantage is that it does not allow precise comparisons to be made because
learners control the elements of the language to very different degrees. For example, 15 different
morphemes appear in the spoken IL of Ann, while Karen uses 21 different morphemes. Karen’s much
more automated performance of some of the “basic” morphemes makes it difficult to compare her
IL with that of Ann.
On the other hand, different levels of competence can be viewed as an advantage, because it allows
one to develop a sense of the developmental routes of learners over a period of time. At this stage,
when the development of morphology in learners of Xhosa has not been studied extensively, it is
probably more useful to gain a broader view of the variety of morphemes they are able to manipulate,
rather than considering too closely the details of the stages they go through to acquire these
morphemes.
90
3.4.6 Generalisability based on subjects and tasks

A widely documented limitation of case studies relates to the generalisability of the findings. The
sample in my study is small and more heterogeneous than is desirable for conclusive findings in the
research area. However, it should be remembered that generalisability in any SLA study is
problematic because of the vast variety of personal, social, psychological and contextual factors which
play a role in the input, processing and output of the learner. Although no two learners are ever
exactly alike, certain generalisations seem to be feasible because they appear so often in the IL of
learners across the world who learn and speak different languages. Further research would be
necessary for more conclusive statements regarding common features of the language produced by
learners of Xhosa, but much of what is presented here is probably valid for the broader population
of English learners of Xhosa, and it may be valid for the broader population of speakers of non-
agglutinative languages who are learning an agglutinative language.
3.4.7 Elicitation procedures

Elicitation procedures are always problematic in SLA studies, because ideally researchers would like
to obtain data which are produced spontaneously in settings where the speaker is not aware of being
recorded and is genuinely trying to communicate for a real purpose. In my study (and many others),
it is not possible to obtain spontaneous data because the learners hardly ever (or even never) have
“real” conversations with “real” people for “real” communicative purposes. Class, race and power
differences in South Africa play a further role in reducing the amount of “real” contact between
English and Xhosa speakers. Indeed, even if there is contact, the English speaker can usually rely on
the Xhosa speaker to have at least some rudimentary command of English so that if their Xhosa fails,
they are usually able to resort to English instead.
Therefore, many learners of Xhosa have limited conversational practice in simulated environments
in the classroom. Teachers and peers who know their L1 provide extraordinary support for these
conversations by supplying vocabulary before-hand or even writing up sentence structures and
vocabulary on the blackboard. There is little real communicative purpose and little urgency to
produce a message which is fully coherent.
91
The second problem with access to spontaneous data relates to ethical considerations. As all parties
involved in the conversation would need to be aware of the recording of data, spontaneity may be
reduced. It would also involve recruiting “real” people in actual situations or contexts with the hope
that the learner would approach these people in order to interact with them. The learners in my study
are not in the same situation as those of the ESF study (Perdue 1993a) who were immigrants who
had to use the language of the host country in order to survive and go about their daily business.
Therefore, a variety of elicitation tasks had to be used with varying degrees of spontaneity in terms
of spoken data produced. These tasks had to be structured to accommodate learners of different
levels (Hulstijn 1989), which means that some learners would have been able to produce fairly fluent
data because they had practised the forms more often. On the other hand, some learners would have
struggled to produce the data because of their lack of familiarity with the forms and vocabulary
required by the tasks. The elicitation tasks may well have focussed on particularly easy or particularly
difficult areas and one always has to be aware that they may not adequately reflect the learner’s ability
in the language. However, the nature of the learning situation and the types of learners studied
required the use of elicitation tasks.
3.4.8 Appropriate linguistic frameworks for SLA studies

Adequate theoretical frameworks for the study of SLA have not been sufficiently developed to
account for all types of learner language. The problem lies in the lack of constructs to analyse the
data, as the grammatical categories of the TL are not always appropriate for the analysis of IL. It is
controversial whether analyses using the current approaches (i.e. comparing the IL and the TL;
studying IL as a system in its own right but using the categories of other languages; using a form-
function approach) are adequate. Analysts may well be missing important features of the IL and
grammatical development by imposing these fairly rigid categories on this type of language. On the
other hand, if IL is viewed as a species of language which is completely different from the TL and
which cannot be analysed with the available categories, then one risks not finding many similarities
with other fully-fledged languages. If general human language processing is taking place, then one
might expect that the IL is similar to the TL and finding a completely new linguistic framework for
studying it is unnecessary.
92
3.4.9 Interviewer interference

Ideally, the interviewer should not have used English at all during the interviews and should even
perhaps not have been able to speak English. However, the confidence and competence of the
students was such that I believe conversation with no resort to English might have caused students
to abandon the study.
Directions were therefore usually given in English and there are examples of clarification using
English in the data, although the majority of the interviews were in Xhosa. Occasionally there was
translation of a vocabulary item and the interviewee might ask for clarification in English. Some
resorted to English when they did not know the word in Xhosa. Ironically, this is probably a better
representation of the spontaneous conversation of these learners than if they had been allowed to hear
and speak only Xhosa, because so many of their interlocutors in the “real” world would be able to
give them this kind of help during the conversation. It would also have created an even more artificial
interview situation if everyone had pretended not to know any English.
3.4.10 Time frames for data collection

It would have been better to have exactly the same intervals between interviews, but the structure of
the university year prevented this and there is a ten-week break between interviews three and four.
A total of eight months of data is perhaps too short to find meaningful development, but the quasi-
longitudinal nature of the study plays a role in alleviating this problem to some extent. If it were
possible to predict which learners would continue learning for a period of a few years, it would be
better to obtain further samples of speech from these learners. This would prevent some of the
disparities between the subjects found in my study and enhance the predictive value of the study.
However, only a very small number of students at Rhodes (typically 4-6 out of about 30) go on to
the second year of study and only half of these students usually go on to the third year.
3.4.11 Amount of data

Although there are enough utterances for a reasonable study in a qualitative paradigm, there are not
enough examples of individual morphemes to test for statistical relevance in a quantitative analysis.
Trends can be analysed, but it is not possible to increase reliability with statistical calculations.
93
3.5 CONCLUSION
This chapter has shown that there is a wide range of possible data collection and analytical
approaches to the study of SLA. The primary aim of IL studies is to gather data about the
competence system underlying the performance by learners, but elicitation methods are not always
successful in this goal. Different tasks elicit different forms and show the variability in performance
of an IL system.
The use of multiple analysis procedures in my study aims to mitigate some of the problems associated
with form-only or function-only approaches. In this way, a more comprehensive picture of Xhosa IL
development has been made possible. Although the results may not be widely generalisable, the
benefits of a detailed case-study are apparent.
In the next chapter, the forms and functions represented in the data are presented and discussed.
94
CHAPTER 4
FORMS AND FUNCTIONS IN THE DATA
4.0 INTRODUCTION
This chapter presents the data from the study, focussing both on the functions learners are able
to represent in their utterances and the morphemes which they can and cannot produce. The
primary focus is on data from conversations, picture descriptions, translation and sentence-
building exercises, since these were the best sources of spontaneous data. Data from
grammaticality-judgement exercises, dictation tests, written class work and tests are presented to
support the analysis in chapter 5, but limitations of data from these elicitation tasks make direct
comparison with the oral tasks problematic (see section 3.2.3.3.2). Summaries of the suppliance,
omission or incorrect forms of morphemes are included in this chapter. In addition, detailed results
of performance on each of the four elicitation tasks for each learner for some morphemes are
included in Appendices G1-20, so that task variability can be shown.
The first section of this chapter presents information about the size of the data set. The second
section discusses how the functions are expressed in the data and the third section looks at the
formal characteristics of the morphemes represented in the data. Where common patterns are
evident, results from individuals are discussed together, but individual patterns are also presented.
This enables the retention of the case study approach and ensures that the significance of data
relating to one individual is not lost by aggregating results (Eckman 1994; Selinker 1992).
4.1 SIZE OF THE DATA SET

Table 4.1 sets out the total number of utterances and morphemes produced in each task by each
participant 1. Utterances were counted according to the criteria listed in section 3.3.3, but I found
that there were difficulties with this approach, since the boundaries of an utterance in spoken data
are often ambiguous. For this reason, I felt that it would be informative to count the number of
morphemes in the data. Besides indicating the scope of the data, this total also shows the
complexity of the learning task faced by these learners.
1
Appendix D provides the number of utterances produced in each task for each interview.
95
When counting morphemes, I took the approach of counting each prefix and stem (e.g. incwadi
= i + ncwadi) or each prefix, root and suffix where the suffix is changeable (e.g. bahamba = ba
+ hamb- + a) (but see Posthumus, 1994, for a different approach). Morphological divisions
followed the patterns set out by Gough et al. (1989). An example of the counting method is
included in Appendix E.
Table 4.1 Number of utterances and morphemes in the data

Name Number of Number of
utterances morphemes
Ann
Conversation 111 641
Picture description 90 458
Translation 44 317
Sentence building 102 749
Ben
Translation 41 317
Claire
Translation 50 393
Pat
Translation 47 405
Karen
Translation 48 389
Sarah
Translation 48 431
Total 2122 26 656
96
4.2 OBSERVATIONS ON THE INTERLANGUAGE OF EACH LEARNER

I have included general observations on the interlanguage of each participant to serve as a more
detailed picture of each learner than that provided in section 3.2.1.1., where attitudes and learning
experiences were considered.
Ann
Ann started as a complete beginner in the first-year Xhosa class. Her early interlanguage in the
first interview is characterised by formulaic speech and single verbs and nouns. She progresses
fairly slowly during the year and has difficulty remembering vocabulary. This means that she tends
to produce very little during the conversations, but is able to produce utterances when the words
are available to her. Progress is particularly noticeable in her dictation tests, where the texts begin
to resemble Xhosa texts towards the end, although she has not mastered any morpheme fully and
she employs a fairly limited range of morphemes. As Ghrib (1987:46) found in her study, many
morphemes do not appear in free conversation and need to be elicited.
The most obvious feature of Ann’s speech is that she uses concordial forms when she can retrieve
the necessary vocabulary. When she has retrieval problems, concords are omitted or incorrect
concords are selected. This is probably due to the high processing loads when vocabulary is not
available. Although many of the concordial forms are correct, she also uses more formulaic
utterances than other participants, so that her system may appear more correct and complex than
it is. Nevertheless, formulaic utterances do not account for the only correct uses of several
morphemes, especially at later stages, and it is evident that Ann has formulated a concordial
system by the end of the data-collection period.
Ann’s performance is fairly variable since the production of correct forms seems to depend largely
on her ability to retrieve the appropriate vocabulary. Task variation can be seen when comparing
conversation, translation and sentence building. Forms tend to be most variable and incorrect in
spontaneous conversation, where she cannot rely on formulaic utterances, and in translation,
where she is forced to retrieve the words and grammatical forms with which she is presented. She
is most fluent and correct in the sentence-building tasks, where she has the most control over the
content which she produces. This may mean, however, that she usually produces forms which are
automatised, formulaic or practised on a previous occasion, rather than new forms where she
97
applies rules to produce the utterances.
Ben
Ben is a 1st year student who has had some previous exposure to Xhosa as a young child and as
a scholar. Ben’s speech is fairly fluent and he employs a range of morphemes, although they are
not always correct. Most of his speech does not appear to be formulaic, although quite a lot of
it seems to be automatised, because he produces forms relatively quickly and repeats forms across
the interviews. He uses English when he does not know the relevant vocabulary, but he quite
often adds a Xhosa morpheme to the English word.
There is some variability in his use of morphemes, often based on the task. Picture descriptions
seem to cause the most problems while sentence manipulation from written data and building his
own sentences are the most fluent and accurate. He performs variably on translation tasks.
Throughout the interviews, he prefers to use a basic utterance structure: SVO (Locative). This
is particularly noticeable in the picture-description task, probably because he is concentrating on
the vocabulary retrieval in these tasks. He does not seem to develop much over the course of the
interviews, although his repertoire of morphemes expands to some extent. The number of
omissions and correct forms remains fairly stable across the data-collection period.
Claire
Claire had had extensive exposure to natural input but was learning Xhosa formally for the first
time in the year in which data were collected. Claire seems to know the functions of a wide range
of morphemes, although the forms she uses are not always correct. She produces long utterances
and a range of morphemes are used from the first interview. She uses a number of lexical items
fairly idiosyncratically, which may be a result of mishearing the oral input which she has received.
Shortened or abbreviated forms may also be a consequence of mishearing during extensive
interaction with L1 speakers. She rarely resorts to English and uses Xhosa discourse markers,
interjections and ideophones in her speech. Repetitions and clarifications from Thembi are usually
requested in Xhosa. She usually understands what is said and can communicate fairly effectively,
although mid-utterance changes in tense, for example, may be confusing to the hearer.
To some extent it seems that Claire’s Xhosa has fossilised and she does not show much
98
development over the course of the interviews. Perhaps communicative fluency is achieved in her
view and her system is impervious to change, as Selinker (1992) suggests. She has a fairly uniform
performance over different tasks (unlike some of the others who seem to have tasks where they
find it easier to produce correct forms) and she seems to make the same kinds of errors across
tasks. This may be a result of fossilisation or because she is not using formulaic forms as the other
participants may be doing. Nevertheless, picture descriptions contain the most errors and the
shortest utterances, and translation sometimes has more errors, perhaps because the necessary
forms have not been automatised. When automatised forms are relied on in general conversation,
there may be problems with producing new meanings. Claire may experience problems with the
task of translation (Sharwood Smith 1994 suggests that it is a practised skill), or a lack of
conscious knowledge of translation forms or how some areas are translated. There seems to be
less evidence in translation tasks of direct translation from English and she may “think in Xhosa”.
Pat
Pat is a 2nd year student with some exposure to natural input and two years of input in the formal
learning context at the beginning of the study. She uses a fairly wide range of morphemes and
does not resort to English very often. Her speech is characterised by a few idiosyncratic forms and
overuse of a few morphemes. Otherwise, she has variable degrees of correctness in her use of
concords, although she usually uses the concords rather than leaving them out. There is
development in the use of some concords but generally forms are fairly regular and fossilised. She
generally uses an SVO (Locative) utterance structure, but also expands on these elements with
modifiers. Picture descriptions cause the most problems. She uses fewer and simpler utterances,
mainly with verbs and nouns, rather than expanded by the use of modifiers. Translation is often
direct and translation exercises usually have many errors. Some English words are used when
Xhosa is not known, but Pat can hold a conversation and answer questions accurately. English
discourse markers such as “but”, “as well as” and “and” are used to link elements of the
utterances. Pat self-corrects some forms, which may indicate an underlying systematicity with
performance difficulties.
Karen
Karen is a 2nd year student with eight years of exposure to Xhosa in a formal learning context at
the beginning of the study. She gives the impression of a fairly sophisticated level of development
99
because of the wide range of morphemes which she uses. However, she makes many errors,
mainly as a result of using direct translation as her primary strategy for communication. At the
idiomatic and morphemic level, her utterances may thus sound fairly strange to the L1 listener.
Her performance on translation tasks is excellent, with most errors occurring in the conversation
and picture-description tasks. Where she has control over the content in the sentence-building
tasks there are few errors and she performs well on sentence-manipulation tasks. A lack of
vocabulary or unknown vocabulary seems to cause the most errors in concord usage. Complex
stories in the conversation contain many errors and words containing click sounds also seem to
introduce problems. She self-corrects many of her own errors.
Sarah
Sarah, a 3rd year student, has had 12 years of exposure to Xhosa in a formal learning context at
the beginning of the study. Her use of morphemes is usually correct in terms of the function of
the morphemes, but incorrect forms are sometimes chosen, even in fairly simple utterances.
Nevertheless, the range of morphemes which Sarah uses is fairly wide. She goes beyond using
only basic utterances and she uses some advanced structures such as auxiliary forms. Much of her
speech shows direct translation from English in terms of word order, and the morphemes which
she uses and omits. Her translation skills are markedly better than those of Ann, Ben, Claire and
Pat and this may be a result of practice in the formal context. Picture descriptions cause problems
for the selection of correct morphemes, but she does not usually try to convey particularly
complex messages, so there are not too many difficulties in conveying her meanings. She uses
English when she is unsure of the vocabulary and many of her errors are self-corrected. Words
which contain click sounds seem to cause problems and the links between words are problematic
when she is unsure of the vocabulary.
4.2.1 Strategies for communicating

This section on communication strategies is concerned with establishing how the learners manage
to initiate and sustain communication, and how they use the cues available in the context to
converse. Communication can be achieved by means of non-linguistic elements such as gestures
and drawings, transfer from the L1, code-switching and the use of formulaic utterances (Bialystok
1990; Wong-Fillmore cited in Wode 1981; Ghrib 1987; Rampton cited in Bialystok 1998). For
several learners, code-switching and the overgeneralisation of known vocabulary items are
100
important resources. For Claire, there is the option of using Xhosa discourse forms which she has
acquired from extensive interaction with native speakers.
Ann’s use of formulae help (or enable) her to communicate and may form the basis of further
development (as Wong-Fillmore cited in Myles et al., 1998, suggests). For example, when a
subject concord has been established in a formula it may be used in a new form, whereas it may
have been omitted otherwise. Ann’s lack of understanding of the elements of the formulae is
evident from the way she mixes formulae, mixes parts of formulae in answers and the lack of
pronominal reference changes during a conversation.
The following example illustrates how Ann retains the question word (phi “where”) in an answer:
T: Usisi wakho uhlala phi?
‘Where does your sister live?’
A: Usisi uhlala *phi2 eThekwini3 (A1C)4
Sister lives where in Durban
Usisi uhlala eThekwini.
‘Sister lives in Durban.’
The next example shows how she tries to use formulae in contexts where they are inappropriate:
T: Ibinjani iholide yakho? (A2C)
‘How was your holiday?’
A: Could I say *ndiphilile?
2
The asterisk is used on each word which contains an error. Where it is possible, the word which
contains the error under discussion has been put in bold, since an utterance may have several asterisked
words. English and Afrikaans words used in Xhosa utterances are underlined.
3
The examples follow the following format:
T: Interviewer’s utterance
A/B/C/P/K/S: Initial of the learner replying to Thembi’s utterance
Line 1: Xhosa utterance as produced by the subject
Line 2: Direct English translation
Line 3: Correct Xhosa form
Line 4: Correct English form
4
This code has three parts:
1: who produced the utterance: A = Ann, B = Ben, C = Claire, P = Pat, K = Karen, S = Sarah, T
= Thembi
2: the number of the interview
3: the task, C = conversation, T = translation, P = picture description, S = sentence building and
sentence manipulation
101
Could I say I was well?
In the first interview, Ann’s strategy for answering questions seems to be to answer by taking the
first element of a question and adding ndi- (I) to it. For example, she does not appear to
understand the first element of the following utterance:
T: Ngeziphi izifundo ozenzayo?
‘Which subjects do you do?’
A: *Ndingeziphi ... isiXhosa *Politics ... (A1C)
‘I which Xhosa Politics’
Ndenza ... isiXhosa nePolitics ...
‘I do ... Xhosa and Politics ...’
Ann can form questions by adding question words to an utterance, although her knowledge of
question words is restricted and she forgets the necessary vocabulary over the course of the
interviews.
At the discourse level, Ann focusses mainly on producing (sometimes uninflected) words and
leaves the interpretation of the utterances to the hearer. She uses code-switching and relies heavily
on the context to create meaning in her messages. There are also some examples of relying on
Thembi’s utterances. She seems to be able to recognise the stems of words which appear in
Thembi’s utterances and uses these to build her own utterances. For example, in interview 2, she
tries to include an adjective based on Thembi’s input:
A: What is “big”? (A2C)
T: If you want to say “big ears” you say ... like umlomo omkhulu big mouth
omkhulu
A: so I’d say like amehlo *umkhulu
So I’d say like eyes are big
Amehlo amakhulu
‘Eyes that are big’
Her communication is hampered by a lack of vocabulary so that utterance and discourse levels
suffer because she has to concentrate on vocabulary retrieval. Ann seems to be most comfortable
with forming sentences containing human referents or referents from class 1, so she usually
chooses one of these forms to enhance success in communication. In addition, she prefers the “I”
(ndi-) pronoun when building sentences or making general conversation.
102
Ben often communicates by building on Thembi’s utterances or copying the forms of her
utterances. He code-switches, although this is usually at the word level rather than at the phrase
or utterance level like Ann. He relies heavily on a basic subject-verb-object-locative pattern in
most of his utterances and he has sufficient vocabulary to communicate most of his messages. Ben
uses occasional word order transfer and some direct translation from English. An example is the
direct translation of “have no hair” as ananwele (negative formative+have+stem for hair).
Ben usually seems to know what is going on in the interviews and he can usually answer
questions. He sometimes tries to clarify understanding by translating Thembi’s Xhosa forms into
English before answering, but this may be a think-aloud process. The following is a typical turn
by Ben:
T: Uhambe nabani? (B2C)
You walked with whom?
B: *Ndihamba *nemoto *wam yam ja nomninawa wam siye eMonti for *iholide
I walk with car my my ja with younger brother my we went to East London for
holiday
Ndiye ngemoto yam nomninawa wam. Siye eMonti ngeholide.
‘I went with my car and/with my brother. We went to East London during the
holidays.’
Claire communicates mainly in Xhosa and, although there are English words in her utterances,
they normally have Xhosa morphemes attached to them. She uses Xhosa discourse markers in
many places and includes forms such as interjectives (e.g. yhu “whew”) to convey emotion. She
also seems to rely on direct translation to some extent, although she also has some idiomatic
forms. An example of direct translation from Claire is into ukutya (thing to eat) instead of adding
the possessive form, into yokutya (thing of to eat). Context sometimes plays a major role in
supporting meanings because some of her speech is quite confused and changes direction in a
single utterance. The following exchange exemplifies Claire’s utterances:
T: Ndixelele ngeklasi yakho nabafundi (C1C)
‘Tell me about your class and the students.’
C: Abafundi OK *andyazi yima *ayafunda nam kodwa mna ndiya *nxa ndiyathetha
kakuhle
Students OK I don’t know stop they study with me but me I go when I talk well
abanye abathethi *odwa ngoku *ndilibe *iXhosa yam yile nto ndiyafunda
isiXhosa *odwa
some don’t talk but now I forgot Xhosa my it is this thing I study Xhosa
siyafunda into *zam ndiyazi *ndiyafunda *ndiyathanda kodwa umfundi eklasini
u *andiyazi
103
we study a thing my I know I study I like but student in class (hesitancy) I don’t
know
Abanye abathethi kodwa ngoku ndilibele isiXhosa sam yile nto ndisifunda
isiXhosa. Abafundi, andibazi. Bayafunda nam kodwa mna ndithetha kakuhle.
Sifunda izinto endizaziyo. Ndiyafunda. Ndiyasithanda kodwa abafundi eklasini
andibazi.
‘Some don’t talk. But now I have forgotten my Xhosa and that is why I am
studying Xhosa. The students, I don’t know them. They study with me but I speak
well. We study things that I know. I study. I like (it) but the students in the class,
I don’t know them.’
Pat overgeneralises a few forms and uses a few basic utterance patterns wherever she can, e.g.
pronoun+thanda uku+stem (pronoun+like to+stem). She uses some English, but generally
manages to communicate fairly successfully using the context to convey meaning when she lacks
suitable forms. Overgeneralisation of forms also allows her to convey functions when forms are
not available to her. A typical turn of Pat’s would be:
Abantu *abavuya ukuphuma *ibedroom *kaloku *iyabanda kakhulu abantu *basela ikofu
no *bayaphunga ikofu *zishushu (P3C)
People are not happy to leave bedroom now cold very people drink coffee no drink coffee
is hot
Abantu abavuyi ukuphuma kwigumbi lokulala kuba kuyabanda kakhulu. Abantu
baphunga ikofu eshushu.
‘People are not happy to leave the bedroom because it is very cold. They drink coffee
which is hot.’
When producing questions, Pat substitutes nini (when) for ngubani (who), with the word ixesha
(time) suggesting overgeneralisation of the prototypical temporal question word, which is nini
(when). There is also an example of yintoni ixesha (what is the time?) which is a literal translation.
Karen communicates with some Xhosa discourse forms, a bit of code-switching and lots of direct
translation. Some of her code-switching has Xhosa concords added, e.g. ngechangi “with
change”. She overextends known words and she tends to tail off when she is unsure, rather than
completing the message. Karen’s direct translation causes clumsy forms, sociolinguistically
problematic forms and incorrect idiomatic forms in all the interviews. Word order problems result
from direct translation. A typical example of a turn in her conversations is:
... ngenye imini *kwakubo inja *ecinci ihlala efameni igama *enja *uSpotty uthanda uku
... one day there was dog that is small it lives on a farm name of dog Spotty he likes to
ukuya ne *nefama *ukuya *ukuze *ingela *ngena e e es es *esendle (K5C)
to go with with farmer to go so that ? enter (hesitancy) the veld
... ngenye imini kwakukho inja encinci. Ihlala efameni. Igama lenja nguSpotty. Uthanda
104
ukuya nofama endle.

‘... one day there was a small dog. It lives on a farm. The name of the dog is Spotty. He
likes to go with the farmer into the veld.’
Like Karen, Sarah communicates with some Xhosa discourse forms, code-switching and lots of
direct translation. Direct translation in Sarah’s utterances seems to be very common in all the
interviews, and this causes problems for tense selection, morphology which is added or omitted,
word order and vocabulary selection. Some code-switching has Xhosa concords added. She also
seems to rely on identifying the roots of oral forms with which she is presented. An interesting
example of her reliance on root identification is the following:
T: Zingaphi iimoto emfanekisweni? (S7P)
‘How many cars in the picture?’
S: How many?
T: (inaudible)
S: But phi means where
T: Zingaphi
‘How many’
S: OK sorry *zineemoto ezintandathu
They have cars that are six
Uneemoto ezinthandathu
‘It has six cars.’
Examination of overgeneralised vocabulary forms shows that some of the overextended forms are
displayed by several learners, while some are unique to individuals. Overextended forms are either
in the same semantic field or display phonetic similarities. Examples of overextended forms
include:
hamba (go) for ya (go to or towards) (Ann, Ben, Claire, Pat)
bona (see) for bukela (watch) (Ann)
jonga (look) for bukela (watch) (Ben, Claire)
jonga (look) for bona (see) (Claire)
jonga (look) and bona (see) for bonakala (appear or seem) (Sarah)
bukela (watch) for bona (see) (Pat, vice versa for Karen)
phunga (drink something hot) for sela (drink something cold) (Ben, vice versa for Pat)
sika (cut) for cheba (cut/shave off hair) (Claire, Pat, Sarah)
thetha (talk) for ncokola (chat) (Pat)
xelela (tell) for ncokola (chat) (Sarah)
vela (come from) for hlala (live) (Sarah)
ukutya for “food” and “eat” (Ann)
Overextension of these forms enables communication of similar meanings when vocabulary is

limited.
105
Although they rely on the context to a large extent, these learners are able to communicate their
messages at most times. They use a number of strategies to enable communication and these
include selecting vocabulary and forms with which they are familiar.
4.2.2 Relationship between elements of the utterance5

The relationship between words can be indicated by the word order or by agreement morphemes
(Felix 1981; Jordens cited in Cooreman & Kilborn 1991; Klein 1986; Meisel et al. 1981). The
simplest way to show the relationship between the elements of an utterance is to place them next
to each other and hope that the hearer can infer the intended meaning from the context. A more
advanced and target-like way is to use Xhosa morphemes since these show the relationship
between co-referenced elements of an utterance. However, the ordering of elements is variable.
In some contexts, the unmarked form is SVO, while in narratives, Gough (1992) finds that VS
is the unmarked form, e.g.
(concord)VS
bayabhala abantu
they write people
The subject concord acts as a co-reference with the subject which appears after the verb. Another
way of signalling relationships between elements of an utterance is to use conjunctions.
Ann relies on the context and juxtaposition of elements which should be next to each other to
convey her meaning when she is unable to produce the required morphemes. The context is an
important clue to the meaning which she attempts to convey in many cases, especially since she
does not always link her utterances carefully to the previous utterances by means of morphemes.
The SVO word order which Ann uses is English-based, but this is not necessarily problematic,
since ordering of basic utterances is usually the same in Xhosa and English. There is a fledgling
concordial system which attempts to link elements, although the forms of the co-referencing
concords are not always correct.
Ben’s use of a greater variety of forms than Ann means that he can supply the links between forms
rather than relying as heavily as Ann does on juxtaposition of elements. However, where he does
5
A brief introduction to the structure of Xhosa is provided in Appendix F.
106
not know the form of a morpheme, he also uses juxtaposition and English words, e.g.:
*Funa *ya eDulcies what’s tonight? (B2C)
Want go to Dulcies what’s tonight?
Ufuna ukuya eDulcies ngokuhlwa?
‘Do you want to go to Dulcies in the evening?’
Claire usually uses morphemes to link elements in the sentence and makes extensive use of
conjunctions in her utterances. Some of these conjunctions act as discourse markers to show
hesitation and give her time to think of vocabulary, but some show the relationship between
elements of the utterance. An extract from a conversation in interview 7 shows how she creates
links between parts of utterances by using morphemes, an English discourse marker and Xhosa
conjunctions:
T: Zinjani iimini zasehlotyeni (C7C)
‘What are the days of summer like?’
C: Zimnandi zinde zishushu and *ndiyathanda kakhulu ngoba ndiyathanda *qubha
They are nice they are long they are hot and I like very much because I like swim
ndiyathanda *ilwandle *ishushu kamnandi ndiyaqubha kaninzi *nxa *ishushu
...
I like sea it is hot nice I swim lots when it hot
Zimnandi zinde zishushu. Ndiyazithanda kakhulu ngoba ndiyathanda ukuqubha.
Ndiyaluthanda ulwandle. Kushushu kamnandi (ndiyaluthanda ulwandle xa
kushushu kamnandi). Ndiyaqubha kaninzi xa kushushu.
‘They are nice, they are long, they are hot. I like them very much because I like
to swim. I like the sea. It (indefinite) is nice and hot (or I like the sea when it is
very hot). I swim a lot when it is hot.’
Pat, Karen and Sarah rely on direct translation and morphemes to indicate relationships between
parts of their utterances. In the following example, direct translation means that the passive is not
included in uvuya ngexabiso (he is happy with the price), so the passive followed by the
copulative does not appear, e.g.:
Utata ubuza unovenkile ixabiso *yenja utata *uvuya *ngexabiso ngoku ke *uyathenga
(K2C)
Father asks shopkeeper price of dog father happy with price now he buys
Utata ubuza unovenkile ixabiso lenja. Utata uvuyiswa (passive) lixabiso ngoku ke
uyayithenga.
‘Father asks the shopkeeper the price of the dog. Father is happy with the price and now
he buys it (the dog).’
The conjunctions which appear in the data from all learners are included in Table 4.2 to show the
types of relationships which can be encoded by these learners.
107
Table 4.2 Conjunctions used by the learners

Ann Ben Claire Pat Karen Sarah
kodwa (but) % % % % % %
xa (when, if) x % % % % %
kanjalo (also/likewise) x x x x % x
okanye (or) x % % x % %
ukuze (that/in order that) x x % x % x
kuba (because) x % % x % %
ke (but/well) x x x x % x
ukuba (if) x x % x % %
ngokuba (because) x x % x % %
mhlawumbi (perhaps) x x x x % x
ngoba (because) x % % x x x
ngaphandle kwa- (except for) x x x x x %
noko (although) x x % x x x
The extensive use of conjunctions and morphemes shows that learners have moved beyond the
stage of mere juxtaposition of elements in most of the interlanguage. This earlier stage reappears
when they are struggling to retrieve vocabulary, but usually the more complex stage of using
conjunctions and morphemes is evident. Although morphemes and conjunctions are not always
used correctly, their functions are still usually conveyed and they enhance the meanings intended
in the messages.
4.3 FUNCTIONS EVIDENT IN THE DATA SET

I have chosen to start the discussion of the data by looking at what the learners in this study can
accomplish with the language they produce, since this approach is more congruent with the aims
of interlanguage (IL) studies. If one starts from an error analysis perspective, one is continually
tempted to focus on the deficiencies in the IL rather than the proficiency which the learner
exhibits. The first focus is therefore on the functions evident in the data as well as on how
108
particular functions are achieved in the IL. For these learners, many functions are realised by the
use of morphemes, so an error analysis which analyses the morpheme level is necessary, and this
is the second focus of the chapter.
As mentioned in section 3.3.6.3, the functions investigated in my study include the referential
function, spatial reference, and temporal reference. Each of these areas will be discussed in more
detail below. Discussion of the language produced by the participants has been arranged firstly
in order of year of study (1st, 2nd, 3rd), and secondly in order of amount of exposure to Xhosa (e.g.
Ann, Ben and Claire are all 1st year students but Claire has the most exposure to Xhosa so her
language is discussed third).
4.3.1 Referential function

Perdue (1990, 1991) claims that development of the referential function proceeds from full noun
phrases to pronominal forms (see section 2.6.4.1). These patterns are found here, but the wide
range of pronominal forms in Xhosa means that learning the pronominal system is a complex task.
For example, there are several forms of “it” depending on agreement with different noun classes
in the linguistic context. The referential function may be realised by noun phrases, subject
concords, object concords, absolute pronouns (e.g. yena “he/she”), demonstrative pronouns (e.g.
lo “this one”), quantitative pronouns (e.g. bonke “they all”) and copulatives (e.g. ngu- “it is”).
None of the learners use demonstratives and quantitative pronouns with a pronominal function
since these morphemes only appear with nouns when they use them. Copulatives are always used
with an impersonal function (“it is”). Table 4.3 indicates the range of forms used to perform the
referential function by all learners6.
6
A tick indicates use of the form and a cross indicates that the form does not appear in the data.
109
Table 4.3 Forms used to realise the referential function

Full range of subject Full NPs instead Absolute Object Possessive
concord pronominal of pronominal pronouns concords pronominal stems
forms forms
Ann x % % x x
Ben % % x x x
Claire % % % % %
Pat % % % % %
Karen % % % % %
Sarah % % % % x
The following patterns are observed in the data:

C Ann uses subject concord pronominal forms but only some of these are used correctly.
She uses the “I” (ndi-) form correctly but other pronouns (ni- “you-plural”, si- “we”, ba-
“they”) are used interchangeably.
C All learners have problems supplying the correct subject concords denoting “it” because
the choice of forms is wide.
C Object concords are not used as much as they would be by L1 speakers.
C Possessive pronominal forms are always used with prepositional na- (with) forms; e.g.
nabo (with them), naye (with him/her).
These learners have a range of referential devices available to them and most of these devices take
the form of morphemes rather than pragmatic devices. Although forms are not always correct, it
is necessary to note that the Xhosa pronominal system is complex because the class system used
with nouns demands a range of different agreement forms.
4.3.2 Spatial reference

In Xhosa, spatial reference can be made by means of gestures, adverbs or adverbial clauses,
demonstrative forms with nouns, or by means of locatives which indicate location and
prepositional relationships. All the interviewees make use of these elements at some point during
110
the interviews7. Spatial reference is required in different tasks across all the interviews and this
results in a variety of forms to express this function. Table 4.4 indicates the range of forms used
by these learners to realise spatial reference.
Table 4.4 Forms used to realise spatial reference

Locative Position 1 Position 2 Adverbs Adverb for a
demonstrative demonstrative demonstrative
Ann % % all incorrect % x
Ben % % x % %
Claire % % x % %
Pat % % x % %
Karen % % % % x
Sarah % % % % x
The following patterns are found in the data:

C Ben prefers the prototypical locative morpheme (add e- to the beginning of the word).
This strategy is applied in the following example:
*Ehlobo *thanda ukudlala elangeni (B7C)
In summer like to play in the sun
Ehlotyeni ndithanda ukudlala elangeni.
‘In summer I like to play in the sun.’
C First position demonstratives (this/these) are produced by the learners, but second position
demonstratives (that/those) are not always produced or are substituted by forms such as
phaya and khona (which mean “there”). The use of these adverbs is probably less
cognitively demanding since their form remains constant.
C Pat adds an unnecessary locative (e-) to an adverbial (phandle) which shows the primacy
of the locative as a means of indicating location in her system:
Ebusika *uyabanda kakhulu sinay sinayo ikhephu kusasa *uyabanda kakhulu
kakhulu
In the winter it cold very (hesitancy) we have it snow early it cold very very
*akufuna *ukuye e *ephandle (P7C)
you don’t want to went to outside
Ebusika kuyabanda kakhulu. Sinekhephu kusasa. Kuyabanda kakhulu kakhulu
Akafuni ukuya phandle.
7
Gestures cannot be accurately recorded because of the data-collection methods, but they are
implied in the data, especially in the picture-description tasks.
111
‘In winter it is very cold. We have snow early. It is very very cold. One does not
want to go outside.’
The locative morpheme and adverbial forms seem to be the most usual means of indicating spatial
reference for these learners. A limited range of demonstratives is also employed by all learners.
Table 4.5 indicates the range of adverbs employed by the learners.
Table 4.5 Adverbs used by the learners

phantsi (below/down) % % % % % x
phandle (outside) % % % x % %
phambili (in front) % % x x x x
phezulu (above) % % % x % %
apha (here) x x % % % %
phakathi (inside) x x % % x x
ngasekohlo (left) x x x x % x
ngasekunene (right) x x x x % x
kufuphi (close to) x x x x % x
apho (there) x x x x % %
4.3.3 Temporal reference

Temporal reference can be achieved by pragmatic or syntactic means. Leaving the hearer to infer
time reference from the context, using adverbial forms and conjunctions are examples of
pragmatic means, while the use of morphemes indicating tense is an example of syntactic means.
Both of these methods of temporal reference occur in the data, with noticeable differences which
depend on the proficiency of the interviewee.
Gough et al. (1989:104) say that the present tense is used in Xhosa to show that an “action,
process or state takes place, is taking place, or exists in the present.” There are two forms of the
future tense: the near future “indicates that something will take place in the relatively near future”
(Gough et al. 1989:112), and the remote future “indicates that an action will take place in the
remote future” (Gough et al. 1989:113). The learners in this study use only the near future tense
form. Gough et al. (1989:114) say that the simple past tense (perfect tense) “indicates a completed
action” (e.g. I have read the book) and that the remote past tense is used to show “that an action
112
took place in the relatively remote past” (Gough et al. 1989:118). The near past continuous tense
is used to show “that an action was continuous in the past” (Gough et al. 1989:20). Table 4.6
indicates the forms used to convey the temporal function in the data.
Table 4.6 Forms used to convey the temporal function

Present tense % % % % % %
Adverbs % % % % % %
Near past continuous x % % % % %
Remote past x % % % % %
Future x % % % % %
Simple past (perfect) x % x % % %
xa (when) x % % % % %
The tense used most frequently is the present tense, and the past and future forms are left to the
context in many cases. For example, Ben leaves the near past continuous form to context when
it is introduced by Thembi, although he uses a past tense form in the utterance, e.g.:
T: Ibinjani iholide yakho yePasika? (B2C)
‘How was your Easter holiday?’
B: *Imnandi ja *ndiye eMonti ...
It nice ja I went to East London ...
Ibimnandi bendiye eMonti
‘It was nice I went to East London.’
Tenses are often mixed in a single utterance and much of the time reference is left to the context
because the morphemes do not always make sense in the context. A typical extract from a
conversation by Claire in interview 4 shows the mixing of tense forms:
*Nake *andikwaz’ *k’qubha kakuhle *lee nto bendihlala ekhaya yonke imini efama
? I don’t know it to swim well this thing I was staying at home every day on the farm
*odwa ifama yethu *ndiyahamba ngeenyawo elwandle *emnandi ka ka inyani *imnandi
but farm our I walk with feet in the sea that is nice (hesitancy) really it is nice
eziny’ *imin umama *am *wandithath’ eMonti *bemnandi (C4C)
other days mother my took me to East London it was being nice
Andikwazi ukuqubha kakuhle. Bendihlala ekhaya efama yonke imini. Bendihamba
elwandle. Bekumnandi, nyani, bekumnandi. Ngezinye iimini umama wam ubendisa
eMonti bekumnandi.
‘I don’t know how to swim well. I stayed at home on the farm every day. I walked in the
113
sea which was nice, really, it was nice. Some days my mother took me to East London
and that was nice.’
Negative forms of the past tense are not produced, and present tense forms are substituted, e.g.:
T: Utata lo ukhwele imoto (S3P)
‘This man drove a car.’
S: *Awukhweli what did you say i imoto imoto *ikhwela ukhwela ibhayisekile
he does not ride what did you say car car it rides he rides bicycle
Akakhwelanga imoto, ukhwela ibhayisekile.
He was not driving a car, he was riding a bicycle.
The forms of the future tense seem to be problematic for all learners who use the form. The basic
form of the future tense patterns is: subject concord + za/ya + ku- + verb stem.
Ben usually uses kwazu- + verb stem, sizo- + verb stem (which is a correct contracted form of the
future) and a half future form za- with no ku- component to indicate the future tense. Claire also
uses this latter form. She also uses an uzo ku- future form and half future forms (which may be
a result of omission of ku-, or the contracted form with no coalescence). Pat and Sarah use these
half future forms as well, e.g.:
*Ndizagoduka ... (P5C)
I will go home
Ndiza kugoduka.
‘I will go home.’
Pat also uses uyakwazi (you know it) as a future form. Sarah uses the infinitive (uku-) for the
future form, which may mean that she focusses on the ku- part of the future form. Karen’s future
tense sometimes consists of a double ku- form (kuku-) form in interviews 3 and 6, e.g.:
(Thembi) *uza kuthanda uku ukuya eSpur *kukuyitya ikofu? (K3C)
(Thembi) you will like to to go to Spur going to it eat coffee
Thembi uyathanda ukuya eSpur ukuya kuphunga ikofu?
‘Thembi, would you like to go to the Spur to go and drink coffee?’
A common error in the use of the near continuous tense is the substitution of ibe- for ibi-, e.g.:
*Ibemnandi (referring to iholide) (B6C)
It was nice (referring to a holiday)
Ibimnandi.
‘It was nice.’
Despite these problems with the use of the different tense forms which all need to agree with the
114
noun with which they are co-referenced, these learners are attempting to use morphemes to
convey their meanings. In many cases, the context is the least cognitively demanding means of
indicating a non-present form, but they attempt to use different morpheme forms. Confusing
meanings may emerge from these forms and the hearer is obliged to use the context to discover
meaning at times.
4.4. MORPHEMES EVIDENT IN THE DATA SET8

Development from the pragmatic mode to the syntactic mode is shown partly in the acquisition
of morphemes to perform functions previously realised by means of reliance on the context,
adverbial elements and gestures (Givón 1979). This development can be seen mainly in Ann
because her speech is closest to the pragmatic mode at the beginning of the data-collection period.
Other participants seem to have a fairly high level of syntactization, but they also display examples
of development in this direction, if one views the pragmatic and syntactic modes as extremes on
a continuum. Although by using morphemes the learners can be placed closer to the syntactic
extreme, improved correct performance and greater use of the morphemes means that there is
further movement along the continuum.
Learning the morphemes of Xhosa is a fairly complex task, since learning a morpheme entails
learning two things:
C the function of the morpheme; and
C the form of that morpheme, depending on the class of the noun with which it agrees.
Throughout the data, it is evident that although the function of the morpheme is learned early, the
multiple forms which the morpheme can take cause difficulties for the learner. For most
morphemes, a small number of agreement forms are learnt first and the forms of less commonly
used classes usually appear later.
4.4.1 Noun prefixes

The noun prefix is attached to all nouns and indicates the class membership of the noun. It is
therefore an important part of the noun, which is used to determine the concordial forms of other
8
Appendix F provides a brief explanation of Xhosa structure.
115
morphemes in the utterance.
Table 4.7 shows that in the overwhelming majority of cases (average 89.8%), the correct noun
prefix is included. To a lesser extent, the incorrect noun prefix is included (average 5.4%), and
only in a relatively small number of cases (average 4.8%) is the noun prefix omitted. Ann, Karen
and Sarah reach a 90% correct suppliance level9 while Ben is very close to this level. Claire and
Pat have the poorest levels of correct suppliance. They have both learnt some of their Xhosa from
natural input and this may have influenced their levels of correct use. On the other hand, they may
be less concerned with correctness and more concerned with producing speech, a conclusion
perhaps supported by the fact that they use more nouns than other participants. Claire’s high
number of omissions may be a result of her exposure to natural input which may mean that her
noun prefix system has developed more slowly and that forms without prefixes may have been
automatised. Although Sarah has a high level of correct suppliance, omission levels are higher
than incorrect suppliance levels, which is an unexpected result given her level of exposure to
formal input where this morpheme is strongly emphasized.
Table 4.7 Summary of noun prefix data10

Name Correct Incorrect Omissions Total
Ann 320 (90.9%) 20 (5.7%) 12 (3.4%) 352 (100%)
Ben 412 (89.0%) 26 (5.6%) 25 (5.4%) 463 (100%)
Claire 529 (85.3%) 34 (5.5%) 57 (9.2%) 620 (100%)
Pat 474 (87.9%) 49 (9.1%) 16 (3%) 539 (100%)
Karen 456 (94.2%) 18 (3.7%) 10 (2.1%) 484 (100%)
Sarah 467 (93%) 14 (2.8%) 21 (4.2%) 502 (100%)
Total 2658 (89.8%) 161 (5.4%) 141 (4.8%) 2960 (100%)
Ann’s high level of noun prefix suppliance is perhaps surprising in the light of her early level of
9
I agree with Gass & Selinker (1994) that 100% suppliance should be a better measure of
suppliance since a form should be used correctly in both form and function if it is target-like, but
a 90% suppliance level is widely used to measure use/development of interlanguage in the
literature.
10
A full table of results for each task is included in Appendices G1 and G2.
116
development, and it is necessary to look at which nouns she uses, to find a possible explanation
for this phenomenon. Closer analysis of her nouns reveals that she uses 116 different nouns
(including names and plurals) in all seven interviews. The range of nouns is therefore fairly
restricted and this may mean that the forms are themselves formulaic to some extent. This may
reduce the processing load during production. On the other hand, Ann seems to realise from an
early stage of acquisition that the noun prefix is detachable, although this early realisation might
be a benefit from formal instruction. She detaches concords from stems when Thembi supplies
vocabulary and she sometimes detaches the concord and repeats only the stem, e.g.:
A: UThembi ufunda - what’s English? (A1C)
Thembi she studies - what’s English?
T: IsiNgesi
English
A: *Ngesi
English
In the next section, a number of common patterns of use of the noun prefix by different
participants in the data will be discussed.
4.4.1.1 Patterns of noun prefix use

Table 4.8 is a summary of the patterns of noun prefix use indicating which subjects use particular
patterns. Examples of each type of noun prefix use are presented below.
Table 4.8 Patterns of noun prefix use

Unmarked/marked classes % % % % % %
Class 2 for class 2a x % x % x x
Animacy % % % % % %
Opposite of animacy % x % x x x
Singular/plural forms x x % % % x
Class 2/6 interchangeability x x % % % x
e-/i- confusion % % x % % %
noun prefixes on English words (%)11 (%) (%) (%) (%) (%)
11
A bracketed tick indicates variable use.
117
4.4.1.1.1 Unmarked and marked classes

For all participants, there seem to be unmarked classes (e.g. classes 1, 2, 9 and 10) and marked
classes (e.g. 3, 11 and 15). The unmarked classes tend to contain more lexical items than the
marked classes, so generalisation of the former may be fairly productive. From a semantic point
of view, the unmarked classes seem to be slightly more regular than the unmarked ones as well.
For example, most human referents fall into classes 1 and 2. Most borrowed words and the words
for many everyday items are found in classes 9 and 10. All students regularise some words from
the marked classes, giving them prefixes from common classes, e.g.:
ibisi for ubisi (milk) (Claire)
ilwandle for ulwandle (sea) (Claire)
ikutya for ukutya (food) (Claire and Pat)
icango for ucango (door) (Pat)
inqwazi for umqwazi (hat) (Pat)
imtshayelo for umtshayelo (broom) (Pat)
ifanekiso for umfanekiso (picture) (Ann)
imfanekisi for imifanekiso (pictures) (Sarah)
iculo for imiculo (songs) (Sarah)
4.4.1.1.2 Class 2 for class 2a

Related to the substitution of unmarked for marked forms are the examples from Ben’s and Pat’s
speech, where they use ba (class 2) for oo (class 2a). The former is the usual plural for human
referents, while the latter appears only on a few nouns, e.g.:
*Ababhuti bam bafunda eKingswood (B6C)
Brothers my they study at Kingswood
Oobhuti bam bafunda eKingswood.
‘My brothers study at Kingswood.’
4.4.1.1.3 Animacy cues

Semantic animacy cues may encourage learners to presume that most animate referents fall into
classes 1 and 2. As a result, some animate referents with class 9 noun prefixes are given class 1
noun prefixes by all learners, e.g.:
*Untombazana mhle (C6T)
Girl is pretty
Intombazana intle.
‘The girl is pretty.’
118
4.4.1.1.4 Opposite of animacy

On the other hand, use of animacy cues seems to be a tendency rather than a rule. Correct forms
are produced and, more importantly, the opposite substitution also occurs. For example, there are
class 9 substitutions for class 1 forms which denote animate referents, e.g.:
*Uyathanda *ititshalakazi *elesiNgesi ngoba uyakunceda (C1C)
He/she likes teacher of English because she helps him/her.
Uthanda utitshalakazi wesiNgesi ngoba uyakunceda.
‘He/she likes the English teacher because she helps him/her.’
4.4.1.1.5 Singular and plural substitution

Singular forms of the noun are used where the context demands the plural. In some cases other
concords in the utterance indicate plurality, so learners may be avoiding redundancy. Generally,
the singular form occurs on the noun, with plural concords following on other morphemes, but
the opposite also occurs to a lesser extent. The context may indicate plurality, so that using the
singular form is cognitively less demanding, e.g.:
Ndinayo *utitshala abathathu (P1C)
I have teacher that are three
Ndinootitshala abathathu.
‘I have three teachers.’
4.4.1.1.6 Class 2 and 6 interchangeability

Class 2 and 6 forms are interchanged in some places, possibly because of phonetic similarity and
the fact that both forms denote plurality, e.g.:
uSpotty *uya *unamahlobo abaninzi efameni (K5C)
Spotty he goes he has friends that are many on the farm
uSpotty unabahlobo abaninizi efameni.
‘Spotty has many friends on the farm.’
4.4.1.1.7 E-/i- confusion

The data suggests that there may be an e- class for some learners, as several nouns begin with this
form, although they are not locatives in the context. This may be a result of the phonetic similarity
of i- and e-, or there may be some confusion with the locative, e.g.:
Utitshalakazi unxiba *elokhwe (A6S)
Teacher wears dress
Utitshalakazi unxiba ilokhwe.
‘The teacher wears a dress.’
4.4.1.1.8 Xhosa concords on English words

Noun prefixes are used on some English words when learners code-switch. This shows that they
119
are trying to produce mainly Xhosa forms and that they can adapt English forms to look more like
Xhosa forms so that they are more acceptable in Xhosa, e.g.:
Batya isoup *inyama *uchocolate ja *ichocolate (S3C)
They eat soup meat chocolate ja chocolate
Batya isoup nenyama nechocolate
‘They eat soup, meat and chocolate.’
The above section has looked at different uses of the noun prefix and the next section moves on
to discuss some of the common patterns in the omission of noun prefixes which are found in the
data.
4.4.1.2 Patterns of noun prefix omission

Table 4.9 indicates the patterns of noun prefix omission for each learner. Examples of the
omissions are presented below.
Table 4.9 Patterns of noun prefix omission

No Xhosa prefixes on English words % (%) (%) (%) (%) (%)
English in utterances x % x % x x
Prefix missing when other forms added x x % x x x
Initial vowels omitted x % x % % x
At beginning of utterances x % x x % x
Correct omission on vocative x % x (%) % %
In addition to the instances displayed above, all learners also drop noun prefixes when they are
struggling to retrieve vocabulary, and when processing loads are high.
4.4.1.2.1 No Xhosa prefixes on English words

Although there are many examples of adding Xhosa noun prefixes to English words, there are also
a number of examples where this does not occur, e.g.:
Ndiyafuna *udlala *aerobics (P4C)
I want play aerobics
Ndifuna ukudlala i-aerobics.
‘I want to do aerobics.’
120
4.4.1.2.2 English in utterances

Where English occurs in the utterance, it seems to interfere with the suppliance of morphemes;
perhaps students are shifting between two different language systems, e.g.:
It will be *titshalakazi *unika umfundi ipensile (P5S)
It will be teacher she gives student pencil
Utitshalakazi unikela umfundi ipensile.
‘The teacher gives the student a pencil.’
4.4.1.2.3 Other forms added

Parts of noun prefixes may disappear when other forms are added to them, as is evident from
Claire’s use of ngeXhosa for ngesiXhosa.
4.4.1.2.4 Initial vowels

There is a tendency to omit initial vowels of noun prefixes, e.g.:
Ndibona *bantwana (B4P)
I see children
Ndibona abantwana.
‘I see children.’
4.4.1.2.5 Beginnings of utterances

Nouns prefixes on nouns which occur at the beginning of utterances are sometimes omitted, e.g.:
*Sisi wam *iminyaka *ishumi *nesithandathu ... (K6C)
Sister my years ten with six
Usisi wam uneminyaka elishumi nantandathu.
‘My sister is 16.’
4.4.1.2.6 Correct omission on vocatives

Most learners correctly omit the noun prefix on nouns in the vocative voice, e.g.:
Mama *uThembi lo ... (B2C)
Mother Thembi this
Mama nguThembi lo ...
‘Mother, this is Thembi ...”
4.4.1.3 Conclusion
The noun prefix has the highest levels of correct suppliance, making it one of the earliest acquired
prefixes in the data if one uses the 90% suppliance criterion of acquisition. There are two possible
reasons for this high level of suppliance:
C it is an integral part of the noun and the prefix is not learnt entirely separately from the
stem;
121
C the noun prefix forms the basic unit for agreement in the remainder of the utterance and
it is therefore very important that it be learnt correctly.
4.4.2 Subject concords

The subject is co-referenced with the verb by means of a subject concord which derives its form
from the class of the subject noun. The subject concord has a linking function (mainly a
grammatical function) and a pronominal function (mainly a semantic function) within the sentence
so that the subject is clearly the referent performing the action denoted by the verb. It can also
have a pronominal function when the subject noun is omitted after its initial mention in an
utterance.
Table 4.10 provides a summary of the subject concord use and omission in the data. Only Sarah
reaches the 90% correct suppliance level (Karen only just falls short). Incorrect forms are more
prevalent (15.4% average) than omissions (4.7% average) for all participants, which shows an
awareness that these forms should be included in their utterances. Karen and Sarah have
substantially lower percentages of incorrect uses (7.8% and 8.2% respectively) than the other four
learners, which may indicate that this morpheme is corrected at later stages of development and
benefits from formal instruction to some extent. Omissions are three to four times higher for Ann
(8.1%), Ben (9.4%) and Claire (6.5%) than for the other learners. There are four possible
explanations for this:
C they may be at an earlier level of development than the other three learners;
C they may be less careful about supplying the morpheme;
C they may have higher processing loads for vocabulary retrieval, which leaves less
processing energy for concord retrieval;
C the morpheme has more of a linking function, which is a grammatical function, than a
pronominal function for them, since they do not usually use it without its co-referenced
noun. If their focus is on meaning rather than grammatical correctness, they may omit the
morpheme more frequently.
122
Table 4.10 Summary of subject concord data12

Ann 161 (72.5%) 43 (19.4%) 18 (8.1%) 222 (100%)
Ben 318 (75.4%) 64 (15.2%) 40 (9.4%) 422 (100%)
Claire 506 (70.2%) 168 (23.3%) 47 (6.5%) 721 (100%)
Pat 417 (79.6%) 94 (17.9%) 13 (2.5%) 524 (100%)
Karen 502 (89.5%) 44 (7.8%) 15 (2.7%) 561 (100%)
Sarah 464 (90.6%) 42 (8.2%) 6 (1.2%) 512 (100%)
Total 2368 (79.9%) 455 (15.4%) 139 (4.7%) 2962 (100%)
4.4.2.1 Patterns of use of the subject concord

Table 4.11 indicates patterns of use of the subject concord. Examples of these patterns are
discussed below.
Table 4.11 Patterns of subject concord use

Pronoun confusion % x x x x x
Animacy % % % % % %
Opposite of animacy x x % % % %
Class 2 and 6 confusion x % % % % %
Hidden concords x % % x % %
Singular/plural confusion % x % % % %
Plural interchangeability x x % % % x
Two referents, singular concord % % x x x x
Vowel-commencing verbs x % x x % %
Existential ku- problems % % % % x x
4.4.2.1.1 Pronoun confusion

For Ann, the use of pronominal subject concord forms (e.g. ndi “I”, u “he/she”) and agreement
forms differs. Pronominal forms are usually included, whereas agreement forms are sometimes
12
A full table of results for all tasks is included in Appendices G3 and G4.
123
omitted. The preferred subject concord is u- (class 1) and she overuses this form, e.g.:
*Uhlala phandle egadini (A5C)
He/she/you sit outside in the garden
Sihlala phandle egadini.
‘We sit outside in the garden.’
Ann is also confused about the forms of the “we” (si-) and “they” (ba-) pronouns, e.g.:
*Nihamba eDulcies (A7C)
You-plural walk to Dulcies
Siya eDulcies.
‘We go to Dulcies.’
4.4.2.1.2 Animacy
As they did for noun prefixes, all learners substitute class 1 or class 2 forms for other class
concords when the referents are animate. Class 1 forms are often substituted for class 9 forms,
since many animal names fall into class 9. If direct translation is occurring here, they are probably
translating the pronoun as “he/she” rather than “it” for animals and using “he/she” for human
referents which fall into class 9, e.g.:
Ikati *uyajonga *nempuku (P2P)
Cat he/she looks with mouse
Ikati ijonga impuku.
‘The cat looks at the mouse.’
Indoda *ubulisa umfazi and *ujongisa ihashe (B2P)

Man he greets woman and he makes look horse
Indoda ibulisa umfazi. Ijonga ihashe.
‘The man greets the woman. He looks at the horse.’
4.4.2.1.3 Opposite of animacy

On the other hand, the opposite pattern also occurs where class 9 forms are substituted for class
1 forms (in the same kind of substitution as the noun prefix), e.g.:
Ubhuti *yam uyafunda isiXhosa eSt Andrews *iyathetha kakuhle (C6C)
Brother my he studies Xhosa at St Andrews it talks well
Ubhuti wam ufunda isiXhosa eSt Andrews. Uyathetha kakuhle.
‘My brother studies Xhosa at St Andrews. He speaks well.’
It should be noted, however, that these verbs usually occur relatively far away from the original
referent. Higher levels of processing power are required to find the correct form, since the co-
referent is far away and this may interfere with correct morpheme selection.
124
4.4.2.1.4 Class 2 and 6 confusion

Class 2 and 6 concords are used interchangeably, as they are for noun prefixes, e.g.:
Abafazi *ayathanda *upheka (C2S)
Women they like cook
Abafazi bayathanda ukupheka.
‘The women like to cook.’
Amakwenkwe *baleqa inja (K4P)

Boys they chase dog
Amakwenkwe aleqa inja.
‘The boys chase the dog.’
Nouns from class 6 generally seem to cause problems when agreement morphemes need to be
selected. This may be a result of the marked status of class 6, since fewer lexical items fall into
it compared to some other classes. A further problem may be the multifunctionality of the a-
morpheme, which is used in the noun prefix of class 6 nouns, to form the negative and
qualificative, as well as being the concordial form for class 6 nouns in the possessive.
4.4.2.1.5 Hidden forms

The hidden concords of classes 5 and 11 cause difficulties for learners. They usually choose the
subject concord of classes 9 and 1 respectively for these classes, probably because the surface
forms of the nouns indicate these choices, e.g.:
Ihashe *iyatya ingca *itya ingca (C5T)
Horse it eats grass it eats grass
Ihashe litya ingca.
‘The horse eats grass.’
4.4.2.1.6 Singular and plural confusion

Some learners do not always provide singular subject concords with singular nouns, but instead
often use plural concords with singular nouns. This may indicate a lack of redundancy or that they
only decide on the plural form once they have embarked on the utterance, e.g.:
... intombi bazi *banxiba ilokhwe ... (P1P)
... girl (hesitancy) they wear dress ...
... intombi inxiba ilokhwe ...
‘... the girl wears a dress ...’
The opposite also occurs where a plural noun is used with a singular subject concord, e.g.:
Abahlobo *ufunda eRhodes (A1C)
125
Friends he/she studies at Rhodes

Abahlobo bafunda eRhodes
‘Friends study at Rhodes.’
4.4.2.1.7 Plural forms interchanged

Plural forms are used interchangeably to some extent without regard for agreement, e.g.:
T: Iimini zasebusika? (C7C)
‘The days of winter?’
C: The days in winter zifutshane zona *amanye *ayabanda
The days in winter they are short them some are cold
Zimfutshane zona ezinye ziyabanda.
‘They are short, some are cold.’
4.4.2.1.8 Two referents with a singular subject concord

Two referents are used with a singular subject concord in the data. The singular form of the
subject concord therefore agrees with only one of the referents (usually the second one), e.g.:
*Numzana nenkosikazi *uthatha *uthetha (A5P)
Mister and Miss (hesitancy) he/she talks
Umnumzana nenkosikazi bayathetha.
‘The man and woman are talking.’
4.4.2.1.9 Subject concords on vowel-commencing verbs

Adding subject concords onto vowel-commencing verbs, usually requires extra forms or
coalescence of sounds, e.g.:
T: Uzokwenza ntoni apha eRhodes? (B1C)
‘What will you do at Rhodes?
B: *Zoyenza iBA HMS
Will do BA HMS
Ndenza iBA HMS.
‘I am doing a BA HMS.’
4.4.2.1.10 Existential forms

The ku- existential form seems to be a form acquired later. Although all learners use it, it is
formulaic in several instances, especially when the vocabulary refers to the weather. This
vocabulary is used to introduce the use of the existential form in the formal context, but its
function is not fully grasped by all learners, e.g.:
*Ngeendlebe *kuyagodola (A3S)
With ears it is cold
Iindlebe ziyagodola.
126
‘The ears are cold.’
4.4.2.2 Patterns of subject concord omission in the data

There are several common omission patterns in the data. Subject concords are omitted by all
learners when the processing load seems to be higher, probably because they are struggling to
retrieve vocabulary. Where the focus is on the message rather than the concords, forms may be
omitted. Picture-description tasks seem to be particularly problematic in terms of requiring high
levels of processing power. Basic sentences with common words are usually correct, either
because vocabulary retrieval is easy, which leaves processing power for concordial agreement or
because the forms are automatised (or a combination of these areas). Table 4.12 indicates which
of the learners displays particular omission patterns in the data.
Table 4.12 Subject concord omissions in the data

Retrieval problems % % % % % %
Intervening morphemes x % x x % x
After xa (if, when) x x % % x x
After pronouns x x % x x x
No ndi- (I) x % % x x x
Repetition x % x x x %
Thembi supplies form % x x % x %
4.4.2.2.1 Vocabulary retrieval problems

Vocabulary retrieval problems cause difficulties with the suppliance of subject concords, since
processing loads are higher than when vocabulary is familiar. In the following example, Karen is
telling a fairly complex story in her utterance and she is struggling to retrieve the necessary
vocabulary, e.g.:
I don’t know OK xa uSpotty *ukuya *uzingela emini *hamba *ayithandi ukulala ...
(K5C)
I don’t know OK when Spotty to go hunts/chases in the afternoon walk it does not like
to sleep
uSpotty uya kuzingela. Emini uyahamba. Akathandi ukulala ...
‘Spotty goes to hunt. In the afternoon he walks. He does not like to sleep ...’
127
4.4.2.2.2 Intervening morphemes

Intervening morphemes may cause omissions of the subject concord, perhaps because the subject
concord is too far away from its co-referent, e.g.:
Bhuti uThembi lo *funda eRhodes (B2C)
Brother Thembi this studies at Rhodes
Bhuti, nguThembi lo. Ufunda eRhodes.
‘Brother, this is Thembi. She studies at Rhodes.’
4.4.2.2.3 After xa
The subject concord is often missing after xa (if/when). It may seem implicit in the subordinate
clause since the subject concord is mentioned in the main clause, e.g.:
*Ayatya *nayasela icooldrink *nxa *hamba (C2P)
They eat and drink cooldrink when walk
Bayatya basela nesiselo xa behamba.
‘They eat as well as drink cooldrink when they are walking.’
4.4.2.2.4 After pronouns

Another similar lack of subject concords occurs after the use of a pronominal form. Here the
subject concord would be repetitive and therefore possibly redundant in the eyes of the learner,
e.g.:
Umakhulu wam *naye *hlale eMont’ (C6C)
Grandmother my and she lived in East London
Umakhulu wam yena uhlala eMonti.
‘My grandmother lives in East London.’
4.4.2.2.5 No ndi- form

When pictures are described, the ndi- (“I”) form is sometimes omitted, probably because it is
implicit in the context or mentioned earlier, e.g.;
*Umfanekiso *bona ingca and abantwana *bancinci (B1P)
Picture see grass and children are small
Emfanekisweni ndibona ingca nabantwana abancinci.
‘In the picture I see grass and small children.’
4.4.2.2.6 Repetition
A second verb in the utterance may also lack the subject concord, which would be a repetition of
an earlier use on another verb and may therefore be regarded as redundant, e.g.:
Sibhala *ephepha and *sebenza ebhodini (B1C)
We write on paper and work on the board
128
Sibhala ephepheni sisebenze ebhodini.

‘We write on the paper and work on the board.’
4.4.2.2.7 Thembi supplies the verb

Another omission occurs when Thembi has used the subject concord and only the stem is used
by the learner in the next utterance. Again, the subject concord is probably implicit because of its
previous use, e.g.:
I: Ukhwela ntoni apha? (P5P)
‘What does he/she ride here?’
R: *Khwela *umtshalo pha
Ride broom there
Ukhwela umtshayelo pha.
‘He/she rides a broom there.’
4.4.2.3 Infinitive forms

A separate area which needs discussion is the use and omission of the infinitive form (uku-), since
this morpheme seems to appear later in the interlanguage of the learners than many subject
concord morphemes.
Ann’s first infinitive appears in interview 3, but it seems to be formulaic. In all other cases where
infinitives are required, two verbs are merely juxtaposed. Other learners also use this strategy,
e.g.:
uThembi uthanda *lala (A1C)
Thembi likes sleep
uThembi uthanda ukulala.
‘Thembi likes to sleep.’
Ben, Claire and Karen use part forms of the infinitive with the initial vowel missing, e.g.:
Bafuna *kusela icooldrink (B4T)
They want to drink cooldrink
Bafuna ukusela isiselo.
‘They want to drink cooldrink.’
Some learners use the verb with a subject concord rather than an infinitive, e.g.:
uThembi *uyathanda *utya inyama (P1C)
Thembi likes eat meat
uThembi uthanda ukutya inyama.
‘Thembi likes to eat meat.’
129
4.4.2.4 Conclusion
The subject concord has a fairly high level of suppliance, although the forms used are not always
correct. In the formal context, this is a form introduced in the first two weeks of the course, which
may play a role in its use in the interlanguage of learners in the formal context. Its dual linking and
pronominal function may also entrench its use to a greater extent than if it had only a grammatical
linking function. The infinitive form appears later than the subject concord form, perhaps because
it is an extension of the basic SVO utterance form. On the other hand, it is often necessary if the
learner is translating directly from English, so it appears fairly frequently in the interviews.
4.4.3 The verbal formative -ya-

In Xhosa, the verbal formative -ya- is used after the subject concord of the verb in the indicative
present tense when the verb is positive and intransitive. The other primary use of the morpheme
is between the subject concord and the object concord to separate the two concords. A further
use is between the subject concord and verb root with any adverb, infinitive, adverbial clause, or
object concord + (object) + adverb if the verb is emphasized (translated from du Plessis
1978:115). Generally, this morpheme therefore has a primarily grammatical function, although it
can also be used for emphasis.
The -ya- morpheme seems to be quite salient for these learners and all learners use it appropriately
and superfluously. Table 4.13 is a summary of the use of -ya- in the data for all learners. “Correct
use” indicates use on an indicative, present tense, intransitive verb. “Overuse” indicates use on
transitive, present tense verbs and use on non-present tense forms. “Omissions” are instances
where the -ya- form is not used on intransitive, present tense, indicative verb forms. It does not
seem that the target system of -ya- use is operating for Claire and Pat because of their overuse
of the morpheme on many of their present tense, transitive verbs in the indicative. The low number
of forms for other learners indicates that they usually use transitive verbs (Ann and Ben) or use
verbs which are not in the present tense (Karen and Sarah). The overuse of the form is also
prevalent for these four learners, although lower numbers of verbs make the percentages a less
reliable measure than those for Pat and Claire.
130
Table 4.13 Summary of the use of -ya-13

Name Correct Overuse Omissions Total
Ann 12 (80%) 3 (20%) 0 15 (100%)
Ben 11 (37.9%) 18 (62.1%) 0 29 (100%)
Claire 45 (15.2%) 228 (77.7%) 24 (8.1%) 297 (100%)
Pat 28 (13.6%) 178 (86.4%) 0 206 (100%)
Karen 11 (50%) 10 (45.5%) 1 (4.5%) 22 (100%)
Sarah 4 (44.5%) 3 (33.3%) 2 (22.2%) 9 (100%)
Total 111 (19.2%) 440 (76.1%) 27 (4.7%) 578 (100%)
In Ann’s utterances, this morpheme is usually correctly omitted, which means that her preferred
utterance structure is SVO. Only 12 of her verbs are used intransitively in the positive, indicative
present tense. Ann’s use of this form is the best of all the learners, and this is mainly because she
produces very few intransitive verbs, she produces fewer verbs than other learners, and she rarely
needs to use the form. This does not mean, however, that she understands when the use of this
morpheme is obligatory. Her low level of overuse may actually be inadvertent because she does
not produce forms which require this morpheme.
Ben overuses -ya- in contexts where the verb is not intransitive. However, he self-corrects his use
of the morpheme in some cases and this may indicate knowledge of the system. On the other hand,
some of his correct forms are copies from Thembi, so it is unclear whether he has formulated a
rule for use, although the lack of omissions may indicate a fairly systematic use. His primary
sentence pattern is SVO, so few of his verb forms are intransitive, which lowers the potential for
omissions.
Claire and Pat overuse -ya- on most transitive verbs in the present tense. The emphatic function
of the morpheme does not seem to be understood because it is overused to such a great extent.
Although most of their utterances contain transitive verbs, they include -ya- on the verb. It seems
therefore that their default present tense verb form is: subject concord + -ya- + verb stem, e.g.:
Abazali *bayathenga inyama kodwa abantwana bathenga iilekese. (P2T)
13
A full table of results for overuse of -ya- for all tasks is included in Appendix G5.
131
Parents buy meat but children buy sweets.

Abazali bathenga inyama kodwa abantwana bathenga iilekese.
‘The parents buy meat but the children buy sweets.’
Karen and Sarah have low numbers of verbs which would require the -ya-, partly because of the
use of mainly SVO patterns in their utterances and partly because of the use of a variety of non-
present tense forms where the -ya- form is not needed.
The -ya- morpheme may be phonologically salient and its constant form may help to entrench its
use. In particular, this form may be phonologically salient when input is provided in the oral mode.
This may be one explanation for overuse of this form by Claire and Pat, who receive more oral
input from L1 Xhosa speakers than other learners.
Another possible explanation for overuse is that the use of -ya- gives the student more time to
retrieve the appropriate verb. The subject concord can be retrieved from the preceding noun with
which the verb is co-referenced, and -ya- can be included as a constant, but retrieval time for the
variable verb is needed. There was often a pause between the production of the subject concord
+ -ya- and the verb which followed and this may be evidence to support this hypothesis. Also,
when the verb was easily retrievable, this -ya- was sometimes left out, e.g. in the sentence-
manipulation tasks. On the other hand, few students removed or added the -ya- where necessary
in the grammaticality-judgement tests. This may show that there is little conscious knowledge of
the rules of use of this morpheme.
4.4.4 Negative subject concords

Xhosa uses a negative subject concord made up of the negative formative a- and the subject
concord (or negative formative a- + -ka- for class 3) on the beginning of the verb in the present
tense to indicate negation. In addition, the final vowel of the verb stem changes to -i in the present
tense. The learner is thus required to supply the correct negative subject concord agreement
morpheme and to remember to change the final vowel of the verb stem. Table 4.14 is a summary
of the use of negative subject concord agreement morphemes in the data. Only Karen reaches
90% correct suppliance, and Ann’s forms are correct only 13.7% of the time. Claire’s uses of the
negative are substantially higher than those of the other learners, which indicates that she uses
these forms in normal conversation more often than the other learners, who produce roughly the
132
same number of negative forms. Since the negative is specifically elicited in some of the tasks, all
learners are forced to use the form, but it is clear that it is not used spontaneously to any great
extent 14.
The negative subject concord appears as a creative form for the first time in Ann’s second
interview and in Ben’s third interview (he uses the formulaic andiyazi form in interview 1). For
all learners, forming negatives from oral data produces more errors than in the sentence-
manipulation task where the input is written.
Table 4.14 Summary of negative subject concord use

Name Correct Incorrect Total
Ann 7 (13.7%) 44 (86.3%) 51 (100%)
Ben 37 (69.8%) 16 (30.2%) 53 (100%)
Claire 117 (83%) 24 (17%) 141 (100%)
Pat 30 (53.6%) 26 (46.4%) 56 (100%)
Karen 54 (91.5%) 5 (8.5%) 59 (100%)
Sarah 54 (76.1%) 17 (23.9%) 71 (100%)
Total 299 (69.4%) 132 (30.6%) 431 (100%)
Ann’s first use of the negative is to add only the negative formative a-. Otherwise, aku- is the
default negative subject concord for her (a form used 37 out of 51 times). This form may be
derived from aka- which is the class 1 negative subject concord or it may be an overgeneralisation
of the “you are not” aku- form, which she may have encountered frequently in the classroom
input. Other forms which she uses are asi- for class 9, 10, 11, and 12 nouns, azi-, and andi-.
Examples of her negatives include:
Abantwana *akutyi *inyama (A4S)
Children you don’t eat meat
Abantwana abatyi nyama.
‘Children don’t eat meat.’
Ben uses a wider variety of forms than Ann, including aka- (for classes 1, 2, 6 and 9), aba- (for
14
This can also be seen in Appendix G6 which indicates the incorrect use of the negative for each
task.
133
classes 1 and 2), aya- (for class 6), and ayi- (for classes 1 and 9) in interview 3. By interview 6,
he has added ali- (for class 5) to his repertoire and all his negative concord forms are correct in
this interview.
4.4.4.1 Patterns of use of the negative subject concord

The patterns of concord use for negative subject concords coincide with the patterns of use for
positive subject concords in the data. Table 4.15 indicates the patterns of negative subject concord
use in the data.
Table 4.15 Patterns of use of the negative subject concord

Singular/plural substitution x % x % % x
Hidden concords x % x % x %
Animacy x % x % x %
Class 6 problems % % % x % %
Past negatives not produced x x % % % %
4.4.4.1.1 Singular/plural substitution

Plural nouns pattern with singular concords and vice versa in the data, e.g.:
Umfazi *abanxiba *ilokhwe ... (P3P)
Woman does not wear dress
Umfazi akanxibi lokhwe.
‘The woman does not wear a dress.’

Class 9 concords are substituted for class 5 concords, e.g.:
Ihashe *ayitya ingca *akatyi ayitya ingca (P6S)
Horse it does not eat grass he/she does not eat it does not eat grass
Ihashe alityi ngca.
‘The horse does not eat grass.’

Animacy cues override the choice of class 9 and 10 concords for animate referents and class 1 and
2 forms are substituted, e.g.:
Le kati *akafuni ukutya (B3T)
134
This cat he/she does not want food

Le kati ayifuni kutya.
‘This cat does not want food.’
4.4.4.1.4 Class 6 problems

Class 6 nouns may cause problems because the subject concord form and the negative form are
the same: a-.This multifunctionality may cause difficulties for learners following the principle that
there should be only one function for a single form (Andersen 1984). Ann uses only the negative
formative for class 6 nouns and other learners also experience problems with this form, e.g.:
Amahash *ayatyi inyama (B3S)
Horses don’t eat meat
Amahashe akatyi nyama.
‘Horses don’t eat meat.’
4.4.4.1.5 Past negatives

Although the negative subject concord is used correctly in the present tense, past tense negatives
which are indicated by Thembi’s past tense forms in the previous utterance are not produced, e.g.:
T: Oomama baphethe iidasta. (P3P)
‘Mothers carried dusters.’
P: Oomama baya *abaphethi iidasta *abaphetha *umtshayelo
Mothers (hesitancy) don’t carry dusters that they carry broom
Oomama abaphethanga iidasta, baphethe imitshayelo.
‘Mothers didn’t carry dusters, they carried brooms.’
4.4.4.2 Negatives of nouns

Ben and Karen generalise the negative of the verb form to the negative of the noun form. When
the noun referent is not known, Ben uses ayi- and aka- although there is no referent available,
e.g.:
Hayi *ayibhayisekile or *akabhayisekile iteksi (B3P)
No it is not a bicycle or it is not a bicycle taxi
Hayi, asiyobhayisekile, yiteksi.
‘No, it is not a bicycle, it is a taxi.’
Claire realises the negative of the noun as hayi + noun stem in interview 3, but as hayi + akho+
noun stem in interview 6, e.g.:
*Hayo *bhayisekile *iteksi (C3P)
No bicycle taxi
Hayi, asiyobhayisekile, yiteksi.
‘No, it is not a bicycle, it is a taxi.’
135
Pat uses a variety of negative forms on the noun, including a negative subject concord, an
invariant asingo + noun stem form and an invariant asikho + noun stem form in one interview,
e.g.:
*Asivenkile (P4P)
it is not a shop
Asiyovenkile.
‘It is not a shop.’
Hayi *asingokati le inja (P4P)

No it is not a cat this dog
Hayi, asiyokati le, yinja.
‘No, it is not a cat, it is a dog.’
Hayi *asikho umabonakude *inomatoto (P4P)

No it is not here television it radio
Hayi, asingomabonakude, ngunomathotholo.
‘No, it is not a television, it is a radio.’
4.4.4.3 Use of the negative terminative i-

As mentioned earlier, a second indicator of the negative form is the change of the final vowel to
i- in the present tense. An indication of the lack of final negative indicators in the data is given in
Table 4.16. The i-negative ending seems to be fairly salient for all learners except Pat. Its constant
form may help it to be used even though it is a verb-final form and is redundant to some extent.
Table 4.16 Summary of lack of negative terminative -i for negative verbs15

Name Number
Ann 6 (11.8%)
Ben 4 (7.5%)
Claire 8 (5.7%)
Pat 26 (46.4%)
Karen 2 (3.4%)
Sarah 0
Total 46 (10.7%)
15
A full table of results indicating lack of the terminative i- is included in Appendix G7.
136
An example of Pat’s failure to produce the i-ending on negatives follows:

*Eli kati *akafuna ukutya (P3T)
This cat he/she does not want food
Le kati ayifuni kutya.
‘This cat does not want food.’
4.4.4.4 Conclusion
The basic form of the negative would seem to be a fairly easy form to acquire once learners know
the subject concord of the positive form, but this does not seem to be the case for these learners
whose performance is much worse on the negative subject concord than their performance on the
positive subject concord. It seems that they may not have realised that the rule requires the
addition of the negative formative to the subject concord (with a different form for class 1).
Instead, they seem to regard the task as learning a new set of morphemes relatively unrelated to
other morphemes which they have encountered before.
4.4.5 Verbal Extensions

There are several different verbal extensions in Xhosa, including the passive, reciprocal, applied,
causative, stative, reduplicative and intensive (Gough et al. 1989). These extensions are added to
the end of the verb and some show relationships between nouns or extend the meaning of the
verb.
The only forms which are used creatively by these learners are the applied extension (Claire, Pat,
Karen and Sarah) and the reciprocal extensions (by Claire, Karen and Sarah). Judging from the
restricted number of words on which other extensions appear, it seems that other verbal
extensions are seen as part of the stem form. For example, -thenga (buy) and -thengisa (sell) are
two different vocabulary items rather than a single form -thenga (buy) with the causative
extension added. This conclusion is supported by Demuth (1992) who says that many lexical items
are not analysed as verb stems and extensions during early stages of L1 seSotho acquisition. In
addition, for these students, the verbal extensions are only taught in the second year.
4.4.6 Auxiliaries
Seven auxiliary forms appear in the data. Claire uses the “just” form anduka and zange (never).
Sarah uses mana (always) and funeka (must). Karen uses ma- (must) and sa (still). She struggles
137
with dla + ngoku (usually) in interview 2, but produces the correct form in interview 4. Auxiliary
forms add meaning to utterances, but they move beyond the basic utterance structure and this may
explain why they are used only by more advanced learners and appear fairly infrequently.
4.4.7 Participials
Claire, Pat, Karen and Sarah fluctuate in their use of the ordinary participial in the dependent
clause after the conjunction xa (“when”). This mood is not used elsewhere with other
constructions, although L1 speakers have a range of uses for this morpheme. An example of the
omission of the participial follows:
Amadoda *abancokola xa *baya uku fix lungisa *ilungisa *ihlangu (P7P)
Men they don’t talk when they go to fix fix it fix shoe
Amadoda akancokoli xa eya ukulungisa izihlangu.
‘The men don’t talk when they are fixing shoes.’
Claire has learned this from exposure to Xhosa, since this form is not taught in the first-year
course, while Pat, Karen and Sarah would have had formal instruction on some of the uses of the
participial.
4.4.8 Copulatives
Jokweni (1997:110-111) says that “(a) copulative is generally defined in terms of its structural
function, namely, to link the subject and the complement. ... The copulative is said to be an
umbrella term which comprises the subject, the copula and the complement ... .” The copulative
in Xhosa is the equivalent of the English “this is”. The copulative morpheme is attached to the
noun and the form of the copulative morpheme is dependent on the class of the noun. Although
the learners use the copulative as a “this is” form, it is not always clear that they have grasped the
verbal function of the morpheme, and they seem to use it with a deictic function in most cases.
Table 4.17 is a summary of the use of copulatives in the data. As the copulative is often omitted
in answer to a question (and this is not strictly unacceptable), such occurrences have not been
counted as omissions when they occur in conversation, e.g.:
Question: What is this? Answer: (It is) a dog.
The set of figures for omissions therefore refers to omissions in utterances where the form is
necessary for an acceptable utterance or required in translation tasks.
138
The copulative occurs in the data of all the learners, although Ann uses it formulaically until the
third interview and Ben only starts using it in the second interview. Copulatives are used by the
beginner learners directly after they learn them, but their use atrophies later. None of the learners
reach 90% correct suppliance levels, which means that none of them have acquired the morpheme
if the 90% correct usage criterion is applied.
Table 4.17 Summary of copulative data16

Ann 20 (58.9%) 1 (2.9%) 13 (38.2%) 34 (100%)
Ben 12 (50%) 0 12 (50%) 24 (100%)
Claire 46 (79.4%) 6 (10.3%) 6 (10.3%) 58 (100%)
Pat 19 (44.2%) 2 (4.7%) 22 (51.1%) 43 (100%)
Karen 48 (87.3%) 3 (5.5%) 4 (7.2%) 55 (100%)
Sarah 27 (69.3%) 2 (5.1%) 10 (25.6%) 39 (100%)
Total 172 (68%) 14 (5.5%) 67 (26.5%) 253 (100%)
Ann uses the copulative in interview 3, but use seems to be mainly formulaic and it only occurs
with class 1 nouns, e.g.:
T: Ngubani igama lakhe? (A3C)
‘What are their names?’
A: NguJulie noThoko
‘It is Julie and Thoko.’
Later in the same interview she does not use the copulative which may show that she is unsure
of the use of this morpheme, e.g.:
*Ngumhlobo *umhlobo *uJulie *uThoko *Kelly (A3C)
It is friend friend Julie Thoko Kelly
Abahlobo nguJulie noThoko noKelly.
‘Friends are Julie, Thoko and Kelly.’
4.4.8.1 Usage patterns for the copulative

A few patterns of copulative usage are evident in the data and these patterns will be discussed
below.
16
A full table of results is included in Appendices G8 and G9.
139
4.4.8.1.1 Non-copulative forms in copulative position

Although forms of the copulative are used, selection of the correct morpheme is not always
appropriate. For example, Claire uses ku and ngo for ngu, e.g.:
Tata *ku (Thembi) lo kutata wam lo igama lesiXhosa ngoVuyisile (C2C)
Father this is Thembi this this is father my this name of Xhosa is Vuyisile
Tata nguThembi lo. Ngutata wam lo. Igama lakhe lesiXhosa nguVuyisile.
‘Father this is Thembi. This is my father. His Xhosa name is Vuyisile.’

Using the same strategy which was used for subject concord selection, that of relying on the
surface forms of the noun, Sarah uses a class 9 form for a class 5 copulative, e.g.:
... intombi *yixhego I mean yiwitch (S2P)
... girl is old man I mean she is a witch
... intombi ligqwirha
‘... the girl is a witch’
4.4.8.1.3 Opposite of animacy cues

Claire uses the opposite of animacy cues (also found in subject concord selection) to produce the
following example where the class 1 copulative ngu- is substituted by a class 9 form yi- in
-titshalakazi (teacher), e.g.:
Umama *am uyafund *fundisi iMusic esikolweni *uyititshalakazi (C2C)
Mother my she teaches Music at school she is a teacher
Umama wam uyafundisa iMusic esikolweni ungutitshalakazi.
‘My mother teaches Music at school. She is a teacher.’
4.4.8.2 Copulative omission patterns in the data

Table 4.18 indicates the omission patterns in the data and indicates which of the learners use these
patterns. Examples are discussed below.
Table 4.18 Copulative omission patterns in the data

With demonstratives (%) (%) (%) (%) (%) (%)
Context % % % % % %
Verbal function not realised % % % % % %
140
4.4.8.2.1 Omissions with demonstratives

Copulatives which should co-occur with demonstratives are left out, e.g.:
(Thembi) *utata lo *umama lo *usisi lo (A2C)
(Thembi) father this mother this sister this
Thembi ngutata lo, ngumama lo, ngusisi lo.
‘Thembi this is father, this is mother, this is sister.’
4.4.8.2.2 Leave to the context

Using pragmatic means to indicate the copulative, some forms are left to the context, e.g.:
Into *andithando *ires *iduty (P3C)
Thing I don’t like res duty
Into endingayithandiyo ehostele yiduty.
‘The thing I don’t like in residence is duty.’
4.4.8.2.3 Verbal functions

As mentioned above, the verbal function of the copulative form is not realised, which means that
several utterances remain verbless, e.g.:
Usisi *umdala *inesi the other sisi *utitshalakazi (A6C)
Sister is old nurse the other sister teacher
Usisi omdala yinesi, omnye usisi ngutitshalakazi.
‘The older sister is a nurse, the other sister is a teacher.’
4.4.8.3 Conclusion
The copulative appears to be a later-acquired morpheme, although it appears in formulae which
are taught during the early stages of the formal course in Xhosa. It is also interesting to note that
the copulative is the Xhosa form of the verb “to be”, a verb which is often exceptional and usually
difficult to acquire in other languages.
4.4.9 Locatives
The locative form in Xhosa has two important functions. Firstly, it is used to indicate the location
of an object, as its name suggests, and, secondly, it fulfills the role of many forms which would
be prepositions in English. For example, “in the town” is realised as edolophini, which is idolophu
(town) with locative morphemes added to it. The locative appears in all the interviews, although
it is only taught in the second semester to the first-year students, so Ann may use it formulaically
at the beginning.
141
With regard to the functions of the locative, in interview 7 it becomes apparent that Ann does not
fully grasp the prepositional function of locatives when she says in interview 7: “I don’t know
what is write bhala I don’t know how to say on paper”.
Table 4.19 is a summary of the five suppliance patterns of the locative in the data. Only Karen
reaches 90% correct suppliance. The range of possible permutations may inhibit correct use of
this form by other learners.
Table 4.19 Summary of locative morphemes17

Name Correct Omissions Incorrect Initial only Final only Total
Ann 53 (73.6%) 8 (11.1%) 3 (4.2%) 3 (4.2%) 5 (6.9%) 72 (100%)
Ben 119 (78.3%) 6 (3.9%) 5 (3.3%) 20 (13.2%) 2 (1.3%) 152 (100%)
Claire 153 (79.7%) 13 (6.8%) 7 (3.6%) 13 (6.8%) 6 (3.1%) 192 (100%)
Pat 87 (66.9%) 24 (18.5%) 5 (3.8%) 13 (10%) 1 (0.8%) 130 (100%)
Karen 176 (95.2%) 4 (2.2%) 1 (0.5%) 3 (1.6%) 1 (0.5%) 185 (100%)
Sarah 164 (87.3%) 7 (3.7%) 9 (4.8%) 7 (3.7%) 1 (0.5%) 188 (100%)
Total 752 (81.8%) 62 (6.8%) 30 (3.3%) 59 (6.4%) 16 (1.7%) 919 (100%)
There are four patterns of use of the locative morpheme in the data:
1) no locative - only the basic form of the noun is supplied with no locative morphemes.
Context is relied on to convey meaning or the use of the noun indicating a place is deemed
sufficient to convey location, e.g.:
Abafundi bafunda *iklasi (A4S)
Students study/read class
Abafundi bafunda eklasini.
‘The students study/read in class.’
2) prefixed locative - usually the most common locative prefix (e-) is added to the noun,
while the remainder of the noun retains its basic shape. When only the initial form is used
when both are required, it may be related to the general principle that learners pay more
attention to the beginnings of words (Slobin 1985). Also, since the prototypical use of the
17
A full table of results is included in Appendices G10 - G13.
142
locative is to prefix e-, students may regard this as the most productive means of
indicating location. For five of the six students, using only the initial form where both are
required is more common than only suffixing a locative form.
Another possibility which occurs in utterances of the more advanced learners is the
prefixing of ku-, kwa- or nga- to the noun. Ben uses locatives from the first interview, but
nga- as a locative is infrequent at early stages, e.g.:
*Cawa on Sunday yes *Cawa siye *church (B2C)
Sunday on Sunday yes Sunday we went church
NgeCawa siye ecaweni.
‘On Sunday, we went to church.’
The default form of the locative is e-, with or without a form of -ini on the end. This is
especially common where nga- and ku- are needed, e.g.:
... ndiyafuna *uye *egqirha *amazinyo (P5C)
... I want to went to doctor teeth
... ndifuna ukuya kugqirha wamazinyo.
‘... I want to go to the dentist.’
All the learners use ku-, nga- and e- locative forms interchangeably, e.g.:
*Akafuna *lendihamba *kuSpur *kukuhlwa ufuna join me? (C7C)
Don’t he/she want? I walk to Spur to evening you want to join me?
Akufuni ukuya eSpur ngokuhlwa?
‘Don’t you want to go to the Spur tonight?’
Ann uses the locative from the first interview, although it appears to be formulaic or
retrievable from the previous utterance by Thembi, e.g.:
Ndivela eJameson House (A1C)
‘I come from Jameson House.’
The formula ndivela e- ... (I come from ...) is one of the first conversational formulae
taught in the first-year class and the above example is probably based on this form.
3) suffixed locative - in this case, the initial vowel retains its basic shape, but a form of -ini
is suffixed onto the noun. The shape of the -ini suffix is dependent on the final vowel of
the noun and this shape is not always correctly formed in the utterances, perhaps because
the patterns have not yet been established or coalescence rules are not understood. It is
143
also possible that the student realises halfway through the word that a locative function
is needed and adds it to the end of the word.
... *indleleni ndibona ilori ... (B6P)
... in the road I see a lorry ...
... endleleni ndibona ilori ...
‘... in the road, I see a lorry ...’
4) prefixed and suffixed locative - this is usually the correct form of the locative. Some of the
occurrences may be related to formulaic use. For example, the form edolophini (in town)
occurs in Ann’s first interview and this is also a formula used in early classes. Another
example where the prefix and suffix forms are correctly added to the stem is:
... ngoku ndidlala esikolweni. (P4C)
‘... now I play at school.’
In interview 3, Ann seems to be constructing the locative by herself. In the following

example from a translation exercise, she is unsure of the necessary vocabulary but
constructs a locative form from the noun which Thembi supplies, e.g.:
A: I’m not sure what some is either abantu akuhl no ja *akuhlali *indlela
I’m not sure that’s houses no I don’t know (A3T)
I’m not sure what some is either people (hesitancy) you don’t live road
I’m not sure that’s houses no I don’t know
T: It’s close
A: No I don’t know
T: Izindlu
Houses
A: Sorry *ezindlini
Sorry in houses
Ezindlwini
‘In houses’
4.4.9.1 Omission patterns in the locative

Table 4.20 indicates omission patterns of the locative in the data of individuals. Examples are
discussed below.
144
Table 4.20 Omission patterns in the locative

Omit on English words % % % % % %
No “is” form % % x % x (%)
Preposition omitted % % x x x x
4.4.9.1.1 Omissions on English words

Locatives are not always supplied on English forms, e.g.:
Khona abantu abane *Jeep ... (B5C)
There are people that are four Jeep
Kukho abantu abane kwiJeep.
‘There are four people in the Jeep.’
4.4.9.1.2 No “is” form

The use of the “is” form on the locative is variable and this may be because, as with the
copulative, the verb “is” remains problematic for learners, e.g.:
Sithetha isiXhsoa xa siseklasini (S5T)
‘We talk Xhosa when we are in the class.’
I: Intombi iphi apha? (S2P)

‘Where is the girl here?’
R: Intombi *ekitshini
Girl in kitchen
Intombi isekitshini.
‘The girl is in the kitchen.’
4.4.9.1.3 Omit prepositional forms

Other omissions occur where some locative forms require an adverbial followed by a prepositional
form and the preposition is omitted. For example, the kwa- preposition is not used after emva
(after) and phezulu (above).
4.4.9.2 Conclusion
The locative is not taught until fairly late in the first-year course and it is interesting that this form
is needed from the beginning, even by Ann. This is probably because of the need to use the
location thematic role.
145
4.4.10 Demonstratives
The Xhosa demonstrative is used to mean “this” in the same way that English uses a
demonstrative adjective (e.g. this dog) and a demonstrative pronoun (e.g. this). It is also used with
the copulative for the translation equivalent of “this is an X this”. There are three different
positions for the Xhosa demonstrative, depending on proximity of the object to the
conversationalists, corresponding roughly to “this”, “that” and “that over there”. The form of the
demonstrative agrees with the noun with which it is co-referenced.
It is not possible to calculate the omission of the demonstrative in ordinary conversation as it is

not an obligatory form. Baseline data from first-language speakers is also not an accurate measure
as there are not usually obligatory contexts for the use of this part of speech. In addition, Xhosa
allows one to refer to a noun by using the subject concord on the verb, so use of the
demonstrative as a pronoun is not as frequent as it would be in English which does not have this
additional referential device. The omissions can therefore only be calculated for translation
exercises which explicitly require the use of a demonstrative. Table 4.21 is a summary of
demonstrative use in the data. Only Karen reaches the 90% correct obligatory suppliance
criterion. The number of incorrect forms is higher than usual because the translation task requires
demonstratives denoting distance from the speaker, forms which few of the learners can produce
correctly, although they usually know the demonstratives denoting proximity to the speaker.
Table 4.21 Summary of demonstrative data18

Ann 13 (54.2%) 5 (20.8%) 6 (25%) 24 (100%)
Ben 38 (66.7%) 17 (29.8%) 2 (3.5%) 57 (100%)
Claire 56 (64.4%) 27 (31%) 4 (4.6%) 87 (100%)
Pat 33 (60%) 18 (32.7%) 4 (7.3%) 55 (100%)
Karen 39 (90.7%) 3 (7%) 1 (2.3%) 43 (100%)
Sarah 24 (77.4%) 4 (12.9%) 3 (9.7%) 31 (100%)
Total 203 (68.4%) 74 (24.9%) 20 (6.7%) 297 (100%)
18
A full table of results is included in Appendix G14.
146
In interview 3, Ann is required to translate demonstrative forms soon after she has been taught
these forms. However, in interview 7 she claims to have forgotten all demonstrative forms in the
translation task which repeats the task in interview 3. She attempts distinctions in proximity in
interview 3, e.g.:
*Aboya bantu *aboya *babantu *hlala ... (A3T)
Those people those people live ...
Abo/abaya bantu bahlala...
‘Those people live ...’
4.4.10.1 Patterns of use of the demonstrative

Table 4.22 indicates the patterns of use of the demonstrative, while examples are discussed
below.
Table 4.22 Patterns of use of the demonstrative

Class 2 and 6 confusion x x % x x %
Overgeneralised forms % % % % x x
Animacy cues x x x % x x
Interchangeable plural forms x x x % % x
Distinction in proximity (%) x x x % %

Class 2 and 6 concords continue to be used interchangeably when producing the demonstrative,
e.g.:
*La bantu *ayahlala ... (C3T)
These people they live ...
Aba bantu bayahlala ...
‘These people live ...’
4.4.10.1.2 Overgeneralised forms

Pat uses a class 5 form for a number of classes throughout the interviews, a strategy which is less
cognitively demanding than retrieving the correct demonstrative form for each noun, e.g.:
*Eli bantwana abanayo *incwadi (P3T)
This children do not have book
Aba bantwana abanancwadi.
‘These children do not have books.’
147
Incorrect choices seem to follow some of the same patterns as noun prefixes and subject
concords. For example, class 9 demonstrative concords are overgeneralised, e.g.:
Ndiyafunda eRhodes *le nyaka (C1T)
I study at Rhodes this year
Ndifunda eRhodes kulo nyaka.
‘I study at Rhodes this year.’

Class 9 animate nouns are supplied with a class 1 animate concord, e.g.:
Ikati *lo (P5P)
Cat this
Ikati le
‘This cat’
4.4.10.1.4 Plural forms are used interchangeably

As for the subject concord, there is interchangeability of ama/izi/aba (class 6/10/2) plural forms,
e.g.:
Amehlo *ezi (P5P)
Eyes these
Amehlo la
‘These eyes’
4.4.10.1.5 Proximity distinctions not made

No distinctions are made between “this/that/these/those” demonstratives in interviews 3 and 7
where these forms are tested in the translation task. Adverbials like apho (there) and phaya (there)
may be used instead of demonstrative forms denoting distance from the speaker in interview 7:
Abantwana apho *abanayo ezi ncwadi phaya (P7T)
Children there they don’t have these books there
Abo bantwana abanazo ezo ncwadi.
‘Those children don’t have those books.’
4.4.10.1.6 Correct usage

With regard to correct usage, Karen is the only interviewee to use demonstratives consistently in
answers to questions of the yintoni le (what is this) type, e.g.:
T: Yintoni le? (K2P)
‘What is this?’
K: Yingca le.
‘This is grass.’
148
The initial vowel of the noun following the demonstrative is usually correctly omitted by all
learners.
4.4.10.2 Conclusion
The demonstrative is elicited to some extent in the interviews because translation tasks demand
that these forms be used in interviews 3 and 7. The use of the morpheme may thus be fairly
restricted in general conversation for these learners. Only one learner can produce the morpheme
correctly over 90% of the time and this morpheme may be a later-acquired one, possibly because
it is an extra form which adds meaning to basic utterances. Other adverbs can also be used to
convey the same meanings and they may be preferred because of their constant form.
4.4.11 Possessive concord

The possessive concord is used to indicate a possessive function and it is prefixed to a noun or
to a pronominal stem. The form of the possessive concord depends on the class of the noun which
is the possessee; e.g. to say “my book”, one looks at the class membership of “book” (incwadi),
which is from class 9 and attaches the class 9 possessive concord ya- to -m, which is the
pronominal stem for the first person, singular pronoun.
Table 4.23 indicates the use of the possessive concord in the data. All interviewees except Ann
use this morpheme from the first interview, with Ann using it formulaically in interviews 1 and 4
and only using it independently from interview 5. The range of correct suppliance is 50% to
78.4% which shows that none of the learners has mastered all the forms of the morpheme,
although omission rates are fairly low (except for Pat).
149
Table 4.23 Summary of possessive concord data19

Ann 10 (55.6%) 8 (44.4%) 0 18 (100%)
Ben 46 (78%) 11 (18.6%) 2 (3.4%) 59 (100%)
Claire 72 (58.6%) 48 (39%) 3 (2.4%) 123 (100%)
Pat 47 (50%) 27 (28.7%) 20 (21.3%) 94 (100%)
Karen 92 (77.3%) 22 (18.5%) 5 (4.2%) 119 (100%)
Sarah 62 (78.4%) 16 (20.3%) 1 (1.3%) 79 (100%)
Total 329 (66.9%) 132 (26.8%) 31 (6.3%) 492 (100%)
Ann’s first independent possessive forms are usually incorrect. There are several correct formulaic
forms, and they are generalised to new contexts. In the following example, she uses lakho (your)
for lakhe (her) and this form is probably copied from the formulaic form ngubani igama lakho
(what is your name):
Igama *lakho nguLinda ngusisi wam igama *lakho *uNitha (A5C)
Name your is Linda is sister my name your Nitha
Igama lakhe nguLinda. Ngusisi wam. Igama lakhe nguNitha.
‘Her name is Linda. It is my sister. Her name is Nitha.’
The next example also uses lakho (your), but this time it is for lehashe (of the horse):
Igama *lakho ihashe *Black Beauty (A5C)
Name your horse Black Beauty
Igama lehashe nguBlack Beauty.
‘The name of the horse is Black Beauty.’
In another example in interview 5, she tries to form the possessive on a noun by using ne- instead
of e-:
*Igamehlo *nehashe *ezinkulu (A5C)
Eyes and horse that are big
Amehlo ehashe makhulu.
‘The eyes of the horse are big.’
In the fourth example, she overuses the “my” pronominal stem to indicate the possessive function
because she is unable to use the possessive on the noun.
19
Full tables of results are included in Appendices G15 and G16.
150
A: My boyfriend’s family (A5C)

T: Usapho
Family
A: Usapho *lwam
Family my
Usapho lomhlobo wam.
‘My friend’s family.’
This overuse of the “my” form appears elsewhere in the data and may be an overgeneralised way
to show the possessive function when a range of forms is not available.
4.4.11.1 Patterns of possessive concord use

Table 4.24 indicates common patterns of use of the possessive concord across learners. Examples
are discussed below.
Table 4.24 Patterns of possessive concord use

Animacy cues x % % x % %
Class 2 and 6 confusion x x % % x x
Singular/plural interchangeability x % % % % %
Marked classes % % % x x %
Hidden forms x x % x % %
Class 1 overgeneralisation x x % x x x
Overuse of ka- x x x % x %
Generalised e- x x x % x x

Forms which may pattern with “he/she” are used with animate referents. This may indicate the
prevalence of animacy as a cue to selecting forms, as found for other morphemes, e.g. -khe
(his/her) for animate forms. A typical example is the following:
Inja igama *lakhe nguMissy (C5C)
Dog name of him/her is Missy
Inja, igama layo nguMissy.
‘Dog, its name is Missy.’
151

Throughout the interviews, Claire and Pat produce am (class 6) with utata (father) and umama
(mother), e.g.:
Utata *am uyasebenza eMonti (C6C)
Father my works in East London
Utata wam usebenza eMonti.
‘My father works in East London.’
This may be a routinised form since it is only used with these two words.
Similarly, am (class 6) may be used instead of bam (class 2) for utata nomama (father and
mother). This may be a class 2 and 6 confusion or a result of not marking plurality since the am
(class 6) form is used with each of these referents when they occur individually, e.g.:
Umama notata *am *ayahamba eBhayi namhlanje (C3C)
Mother and father my walk to Port Elizabeth today
Umama notata bam baya eBhayi namhlanje.
‘My mother and father go to Port Elizabeth today.’
4.4.11.1.3 Singular/plural interchangeability

Not all possessive concords agree in terms of singular or plural forms, which may indicate a more
general principle of not marking morphemes redundantly, a pattern found with other morphemes
as well, e.g.:
Ndiyathanda ukuthetha nabahlobo *wam (P1C)
I like to talk with friends my
Ndiyathanda ukuthetha nabahlobo bam.
‘I like to talk to my friends.’
4.4.11.1.4 Marked classes

There are problems with supplying the possessive concords of marked classes, e.g.:
*Ndiyathanda ukutya *wakhe (C5C)
I like food of her/him
Ndithanda ukutya kwakhe.
‘I like his/her food.’
The choice in the example above seems to be based on the initial vowel rather than the whole
noun prefix of class 15.
152

There are errors with nouns from class 5 where part of the noun prefix is hidden. Class 9 forms
are used instead of class 5 possessive concords, a choice based on the visible part of the noun
prefix. Again, this is a pattern found in the selection of other concords. A typical example is:
Igama lam *yesiXhosa *Sindiswa (C2C)
Name my of Xhosa Sindiswa
Igama lam lesiXhosa nguSindiswa.
‘My Xhosa name is Sindiswa.’
4.4.11.1.6 Class 1 overgeneralisation

The class 1 concord is overgeneralised to a class 1(a) noun with a possessive concord consisting
of a vowel, e.g.:
... umhlobo *wamama *am ... (C5C)
... friend of mother my ...
... umhlobo kamama wam ...
‘... my mother’s friend ...’
Few other concords make distinctions between class 1 and class 1(a) and it appears that Claire
is using this general feature in the above example.
4.4.11.1.7 Overuse of ka-

On the other hand, Pat overgeneralises the ka- form to a class 1(a) noun with a possessive
concord which consists of a consonant and vowel. It appears that she has not fully grasped the
rule for the use of ka- here, e.g.:
Abazali *kaThembi bayasebenza (P1C)
Parents of Thembi work
Abazali bakaThembi bayasebenza.
‘Thembi’s parents work.’
A similar lack of understanding of the rule is evident in the overuse of the ka- possessive with
nouns which do not fall into class 1(a), e.g.:
... xa ibirthday *kaumntu ... (P4C)
... when birthday of person ...
... xa umhla wokuzalwa womntu ...
‘... when a person’s birthday ...’
4.4.11.1.8 Generalised e- form

A generalised e- possessive form is used in several examples (rather than the correct possessive
153
concord), e.g.:
Umqhubi *emoto *akancoka (P7S)
Driver of car does not chat
Umqhubi wemoto akancokoli.
‘The driver of the car does not chat.’
4.4.11.2 Patterns of omission

There are a few similar omission patterns across learners in the data. Firstly, the pragmatic means
of indicating possession is to place nouns next to each other and hope that the hearer derives the
meaning from the context, e.g.:
Ndiza kuthatha imoto *umama wam (C3C)
I will take car mother my
Ndiza kuthatha imoto kamama wam.
‘I will take my mother’s car.’
For Pat and Sarah, possessives are often missing from infinitive forms, probably as a result of
direct translation from English, e.g.:
... izinto *ukupheka (P4C)
things to cook
... izinto zokupheka
‘... things to cook’
Transfer from English also causes the omission of some possessive markers, e.g.:
*Ikrisimesi iholide ide kakhulu (P5C)
Christmas holiday long very
Iholide yeKrisimesi inde kakhulu.
‘The Christmas holiday is very long.’
4.4.11.3 Conclusion
Although possessive forms appear in the data and possession seems to be a function necessary in
the interlanguage of these learners, they are not always able to select the correct form of the
morpheme.
4.4.12 Qualificatives
Two classes of words in Xhosa perform the equivalent of the English adjectival function, i.e.
adjectives and relatives. The adjective class is fairly restricted and includes the words for the
numbers one to six, -bi (ugly), -dala (old), -de (long, tall), -fuphi (close), -futshane (short), -hle
154
(beautiful), -khulu (big), -ncinane (small), -ncinci (small), -ninzi (many), -ngaphi (how many),
and -tsha (new). All other qualificatives are in the relative class. Adjectival and relative concords
are fairly similar, but learners have to learn two sets of morphemes to be sure that the correct
morpheme is used. However, I have chosen to present the data of both morphemes together in
Table 4.25 since their function is the same. I have not included in this table examples where the
qualificative formative a- has been added or should be added, since these examples are discussed
in section 4.4.13.
All learners use the adjective and relative concords, although Ann only starts using them in the
third interview. These forms are additional to the basic utterance and they modify elements of the
basic utterance. This may mean that learners regard them as less necessary for conveying the basic
message and so they may be learnt later. They are also presented towards the end of the second
semester in the first-year class. Four learners have correct suppliance of under 50% for these
morphemes, as indicated in Table 4.25.
Table 4.25 Summary of qualificative concord data20

Ann 4 (30.8%) 7 (53.8%) 2 (15.4%) 13 (100%)
Ben 12 (40%) 13 (43.3%) 5 (16.7%) 30 (100%)
Claire 46 (75.4%) 10 (16.4%) 5 (8.2%) 61 (100%)
Pat 11 (42.3%) 14 (53.9%) 1 (3.8%) 26 (100%)
Karen 9 (69.2%) 4 (30.8%) 0 13 (100%)
Gill 12 (48%) 10 (40%) 3 (12%) 25 (100%)
Total 94 (56%) 58 (34.5%) 16 (9.5%) 168 (100%)
Ann’s earliest qualificative forms appear in interview 3. Some of the examples in this interview
show that she merely places the root of the qualificative after the word it modifies, relying on
pragmatic means to convey her message, e.g.:
*Mpahla *shushu *shushu (A3C)
Clothes hot hot
Impahla eshushu
20
155
‘Clothes that are hot’
Also in this interview, there are examples where she adds ku- onto the qualificative, e.g.:
Kushushu ikofu iti (A3C)

It is hot coffee tea
Zishushu ikofu neti.
‘The coffee and tea are hot.’
She encountered the existential form ku- on -shushu (hot) in the weather vocabulary which she
had learned just before this interview and this may have resulted in transfer of these forms.
Another example which may be transfer of ku- or a correct form using knowledge of the
qualificative concord system is the following:
Ukutya *kumbi (A3C)
Food is bad
Ukutya kubi.
‘Food is bad’
If the example above is an example of transfer of ku-, then Ann’s early development of the
qualificative concord is to place the qualificative next to the noun which it modifies or to use the
ku- form on the qualificative. Differentiation of forms may occur at a later stage.
4.4.12.1 Patterns of use of the qualificative

There are a number of common patterns of use of the qualificative in the interviews. Firstly, a
productive strategy which indicates the relevant function but is not too cognitively demanding is
to use only one or two forms of a morpheme. Ann uses only the omnye (“one” for class 1) form
and it is probably formulaic. Sarah’s default form is enye (class 9) with a few examples of omnye
(class 1). Pat replaces enye (class 9) with omnye (class 1) in some places. e.g.:
... *indoda *abubini *bayancokola *omnye *uyatshaya ... (P7P)
... man that are two they chat one he smokes
... amadoda amabini ayancokola. Enye iyatshaya ...
‘... two men are talking. One is smoking ...’
Claire, Karen and Sarah again interchange class 2 and 6 concords, e.g.:
Abantu *ayasebenza *isitalatu *amanye *ayathengisa *ikutya abanye ikafe abanye
impahla (C4P)
People work street some sell food others coffee others clothes
Abantu bayasebenza esitalatweni. Abanye bayathengisa ukutya, abanye ikofu abanye
impahla.
156
‘People work in the street. Some sell food, some coffee, some clothes.’
4.4.12.2 Omission patterns of the qualificative in the data

Table 4.26 indicates the omission patterns in the data. Examples are presented below.
Table 4.26 Omission patterns of the qualificative in the data

On numerals x % x % x %
Stems with lu- x % % x x x
No coalescence % % x x x x
4.4.12.2.1 On numerals
Qualificatives are not always used on numerals, perhaps because they redundantly show plurality,
e.g.:
T: Zingaphi iklasi onazo evekini? (B1C)
‘How many classes do you have per week?’
B: I think *hlanu
I think five
Ndicinga zintlanu.
‘I think five.’
4.4.12.2.2 Stems with lu-

Qualificative forms are not always added to stems which begin with lu-. This is the subject
concord for class 11 and learners may think that words beginning with lu- already have concords
so they do not need to add another one, e.g.:
Umthi *luhlaza (B6T)
Tree green
Umthi uluhlaza.
‘The tree is green.’
4.4.12.2.3 Lack of coalescence

The -hle adjective form, which requires coalescence when an adjective concord with a nasal is
added, is not always coalesced, e.g.:
Intombazana *ihle (B6T)
Girl is pretty
Intombazana intle.
‘The girl is pretty.’
157
4.4.12.3 Conclusion
As noted above, correct suppliance of the qualificative morpheme is relatively poor and this may
be a result of:
C the range of forms which need to be learnt;
C the less urgent need to learn forms which relate to optional elements of the utterance;
C the fact that these forms fulfill mainly an agreement function rather than a semantic
function, since the semantic function is carried by the stem of the qualificative.
4.4.13 Qualificative a- morpheme

The qualificative a- morpheme is added to the verb or to the qualificative (as discussed above).
In effect, a subordinate clause is created with this morpheme, so one would expect that forms
would be omitted or incorrect since the utterance structure moves beyond the basic form. Instead,
a comparison of tables 4.25 and 4.27 shows that for four of the six learners, the qualificative a-
is used correctly more often than the qualificative morpheme by itself. However, the level of
omission of the qualificative a- morpheme is higher for five of the six learners and this is probably
a better reflection of their abilities to use the morpheme. Pat has the lowest correct suppliance of
all, although she creates the most opportunities for use by using complex utterances but leaving
out the form. The qualificative a- form is used more often than the adjective and relative concords
by themselves and this embedded form is probably necessary because the learners translate directly
from English in many cases. Table 4.27 is a summary of the use of the qualificative a- concord in
the data.
Table 4.27 Summary of qualificative a- concord21

Ann 11 (45.8%) 5 (20.9%) 8 (33.3%) 24 (100%)
Ben 34 (47.9%) 12 (16.9%) 25 (35.2%) 71 (100%)
Claire 49 (60.5%) 15 (18.5%) 17 (21%) 81 (100%)
Pat 20 (30.8%) 16 (24.6%) 29 (44.6%) 65 (100%)
Karen 69 (75%) 13 (14.1%) 10 (10.9%) 92 (100%)
Sarah 70 (76%) 11 (12%) 11 (12%) 92 (100%)
Total 253 (59.6%) 72 (16.9%) 100 (23.5%) 425 (100%)
21
158
4.4.13.1 Patterns of use of the qualificative a- morpheme

Many of the errors made when selecting qualificative a- forms are the same as those made when
choosing other qualificative forms, since the qualificative a- form is added to the qualificative
morpheme and coalesced where necessary, e.g.:
Isikolo sihle (qualificative concord)

‘The school is beautiful.’
Isikolo esihle (qualificative a- + qualificative concord + coalescence of forms)
‘The school which is beautiful’
Pat overuses the e- form on several qualificatives, e.g.:

Kwi-exam time abantu*eninzi abantu bayafuna kufu bafuna ukufunda ... (P3C)
At exam time people that are many people want (hesitancy) they want to learn ...
Ngexesha lemviwo abantu abaninzi bafuna ukufunda ...
‘At exam time many people want to study ...’
4.4.13.2 Omission patterns of the qualificative a- morpheme

There are several omission patterns for individuals in the data and these patterns will be discussed
below.
4.4.13.2.1 Processing constraints

Ben omits the qualificative a- when he produces longer utterances which may cause processing
problems, e.g.:
*Omnye ine what’s a roof dak ... *phahla *omnye *nephahla *ebomvu ne *necango
One has what’s a roof roof ... roof one has roof that is red and and door
*ubomvu nefestile *eblauw *udonga *lumhlophe ... (B6P)
red and window that is blue wall is white
Enye inophahla olubomvu nocango olubomvu nefestile eluhlaza nodonga olumhlophe
...
‘One has a red roof and a red door and a blue window and a white wall ...’
4.4.13.2.2 No subordinate clause marking

Sarah combines three sentences with no relative clause markers, e.g.:
Ndinenja *ihlala ekhaya *yam *imhlophe (S2S)

I have dog it stays at home my it is white
Ndinenja emhlophe ehlala ekhayeni lam.
I have a dog that is white that stays at my home.
4.4.13.2.3 Stems beginning with lu-

159
Sarah uses no qualificative a- forms on colour terms which start with lu-, e.g.:
Utitshala unxiba ibhulukwe ende eblauw nehempe *lubhelu (S1P)
Teacher wears pants that are long that are blue and shirt yellow
Utitshala unxiba ibhulukwe ende eluhlaza nehempe elubhelu.
‘The teacher wears long blue pants and a yellow shirt.’
Omissions usually relate to a failure to understand that the qualificative a- is used to indicate an
embedded clause.
4.4.13.3 Verbal relatives

Because the qualificative a- can also be added to verbs, it is necessary to examine the use of the
-yo morpheme in the data. Gough et al. (1989:265) say that “(t)he relative suffix -yo is added to
the verbal relative when there is no adjunct, object or any other extension ...”.
Pat, Claire, Sarah and Karen usually use the -yo form, although there are omissions by all of them,
e.g.:
T: Ngowuphi umdlalo owuthandayo? (P4C)
‘Which sport do you like?’
P: Ndiya izidlalo *ndiyathanda ...
I go sports I like
Izidlalo endizithandayo ...
‘The sports that I like ...
Sarah also uses existential forms for the qualificative a- + -yo form in interview 7, e.g.:
... sinxiba impahle *kupholile (S7C)
... we wear clothes it is cool
... sinxiba impahla epholileyo
‘... we wear cool clothes’
4.4.13.4 Conclusion
An utterance containing an embedded clause is more complex than the basic utterance and these
learners show that they do not always indicate the presence of an embedded clause by means of
the correct morpheme. It should also be noted that when analysing spoken utterances it is not
always clear whether an utterance is meant to be embedded or whether the speaker is starting a
new utterance which uses only the basic qualificative form. Levels of omission may therefore not
be accurately reflected in Table 4.27.
4.4.14 Na-/nga-: “with”

160
Sections 4.4.14, 4.4.15 and 4.4.16 discuss three different morphemes which take the same form,
i.e. na-, but which have different functions and names.
The first morpheme is used to denote “with”. The morpheme has two forms: nga- when used
with human referents and locatives and na- with all other referents. A summary of the use of both
forms of the morpheme is included in Table 4.28. Average use (correct + incorrect uses) of the
morpheme reaches the 90% suppliance level, but the uses are not always correct, so it cannot be
said that any of the learners have acquired the morpheme fully.
Table 4.28 Summary of na-/nga-

Ann 14 (82.4%) 0 3 (17.6%) 17 (100%)
Ben 34 (58.6%) 20 (34.5%) 4 (6.9%) 58 (100%)
Claire 82 (83.7%) 13 (13.3%) 3 (3%) 98 (100%)
Pat 39 (81.3%) 6 (12.5%) 3 (6.2%) 48 (100%)
Karen 64 (88.9%) 5 (6.9%) 3 (4.2%) 72 (100%)
Sarah 53 (68.8%) 9 (11.7%) 15 (19.5%) 77 (100%)
Total 286 (77.3%) 53 (14.3%) 31 (8.4%) 370 (100%)
Ben, Claire, and Sarah usually distinguish the nga- “with” and the na- “with” forms, although
there are some errors, e.g.:
Ndihlala *nebhola phandle (B2S)
I live(play) with ball outside
Ndidlala ngebhola phandle.
‘I play with the ball outside.’
The omission of nga- is particularly prevalent on forms which denote temporality, since this form
is not a direct translation from English, e.g.:
*Ndiyaqubha kakuhle *izinye imini (C5C)
I swim nicely one day
Ndiqubhe kakuhle ngezinye imini
‘One day I swam nicely.’
When na- and nga- are used purely to mean “with” as they would be used in English, these forms
are usually correct. However, the problematic usage is when nga- is added to a form which
161
denotes temporality. The “with” function is not particularly obvious to the English-speaking
person here, since the translation of these forms into English does not require the use of “with”.
Transfer may therefore be a source of errors in this case.
4.4.15 Na-: Associative Copulative

The morpheme na- can be used to indicate “have” when it is attached to the noun or pronoun.
Table 4.29 summarises the use and omission of the morpheme and indicates that omission levels
are fairly low, although except for Karen, correct usage does not reach the 90% correct
suppliance.
Table 4.29 Summary of the associative copulative

Ann 1 (25%) 0 3 (75%) 4 (100%)
Ben 16 (55.2%) 10 (34.5%) 3 (10.3%) 29 (100%)
Claire 16 (84.2%) 2 (10.5%) 1 (5.3%) 19 (100%)
Pat 36 (87.8%) 4 (9.8%) 1 (2.4%) 41 (100%)
Karen 20 (95.2%) 1 (4.8%) 0 21 (100%)
Sarah 29 (85.3%) 5 (14.7%) 0 34 (100%)
Total 118 (79.7%) 22 (14.9%) 8 (5.4%) 148 (100%)
4.4.15.1 Patterns of use of the associative concord

4.4.15.1.1 Lack of correct negative forms
One of the main sources of incorrect forms of “have” is that the correct negative forms are not
always produced, e.g.:
Aba bantu *abanencwadi *neencwadi (S3T)
These people they don’t have book have books
Aba bantu abanancwadi.
‘These people don’t have books.’
4.4.15.1.2 Formal form

Pat uses only a very formal associative form: pronoun + na- + -yo + noun. Although this form is
not incorrect, it is regarded as very formal in the context of everyday conversation. Pat uses a
form which is invariant regardless of the need for agreement of the pronoun with the noun which
162
follows it, e.g.:

Ndinayo izinto *izininzi to do (P4C)
I have things many to do
Ndineezinto ezininzi zokwenza.
‘I have many things to do.’
4.4.15.1.3 Lack of coalescence
Coalescence of the initial vowel of the prefix and the associative copulative causes problems,
although the morpheme is used, e.g.:
... uThabo *unemakhulu (B1C)
... Thabo he has grandmother
... uThabo unomakhulu.
‘... Thabo has a grandmother.’
4.4.15.2 Conclusion
Incorrect coalescence of positive forms and the lack of non-coalescence of negative forms seem
to be the two major areas affecting the use of correct forms of the associative copulative. The
function of this morpheme is therefore used in most cases, although the forms of the morpheme
remain problematic for most learners.
4.4.16 Na-: Additive conjunction

The final use of the na- form is as an additive conjunction. Other conjunctions which take the
form of individual words are discussed in section 4.2.2. The additive conjunction is used or
required fairly widely in the data and since it is a morpheme which is attached to stems, I have
chosen to discuss it separately from the other conjunctions.
Table 4.30 Summary of use of the additive conjunction

Name Correct Incorrect Omit Total
Ann 0 0 9 (100%) 9 (100%)
Ben 24 (39.3%) 28 (45.9%) 9 (14.8%) 61 (100%)
Claire 7 (43.8%) 0 9 (56.2%) 16 (100%)
Pat 29 (47.5%) 2 (3.3%) 30 (49.2%) 61 (100%)
Karen 33 (64.7%) 5 (9.8%) 13 (25.5%) 51 (100%)
Sarah 38 (70.4%) 0 16 (29.6%) 54 (100%)
163
Total 131 (52%) 35 (13.9%) 86 (34.1%) 252 (100%)

4.4.16.1 Patterns of additive conjunction use
Table 4.31 indicates the patterns of use of the additive conjunction for different learners.
Examples are discussed below.
Table 4.31 Patterns of additive conjunction use

Not on lists % % % % % %
Discourse marker x % % x x x
On infinitives x x x x (%) x
On verbs x x % % x x
4.4.16.1.1 Not on lists

The additive conjunction is not always used, especially in lists of English and Xhosa words, e.g.:
Ndibona amaqanda necephe nebhotolo nepleyiti *bowl *ikeki *iscale *iswekile ... (S5P)
I see eggs and spoon and butter and plate bowl cake scale sugar
Ndibona amaqanda necephe nebhotolo nepleyiti nebowl nekeyiki nescale neswekile.
‘I see eggs, spoon, butter, plate, bowl, cake, scale and sugar.’
4.4.16.1.2 Discourse markers

The English word “and” may be used as a discourse marker in utterances, e.g.:
... ngoobhuti bam aba *Jonathan *and uWalter (B5C)
... these are brothers my these Jonathan and Walter
... ngoobhuti bam aba nguJonathan noWalter
‘... these are my brothers Jonathan and Walter.’
4.4.16.1.3 On infinitives
The na + uku (additive conjunction + infinitive) form to indicate “and” on an infinitive is used
variably, e.g.:
Ndithanda ukukhwela *intaba *kuya elwandle (S3C)
I like to climb the mountain to go to the sea
Ndithanda ukukhwela entabeni nokuya elwandle.
‘I like to climb the mountain and go to the sea.’
4.4.16.1.4 On verbs
The additive conjunction is used incorrectly on verbs, e.g.:
... watshata *nabuya izolo (V2C)
164
... she married and returned yesterday

... watshata sibuye izolo.
‘... she married and we returned yesterday.’
4.4.16.2 Conclusion
There are many omissions of the additive conjunction in the data. English patterns may affect the
use of the morpheme since English does not require the use of “and” between each noun in a list.
4.4.17 Object concords

One function of the object concord in Xhosa is to show definiteness, as Xhosa does not have an
article system which is equivalent to English. A second function is a pronominal function
operating both cataphorically and anaphorically. The object concord is attached to the verb
preceding the object or to a verb which refers back to an object in a different utterance. The shape
of the object concord is dependent on the class of the object noun to which it refers.
The omission of the object concord is difficult to calculate, as its use is not obligatory and it is
difficult to detect from the context when definiteness is necessary and when it is not. Baseline data
for the conversations is not an adequate measure of obligatory use, because the conversations
differ from person to person. Table 4.32 provides a summary of object concord use in the data.
The object concord is not used by Ann, and Ben uses one incorrect form. The levels of use of the
morpheme by other learners are very low and it is clear that this is one of the later-acquired
morphemes.
Table 4.32 Summary of the use of object concords

Ann 0 0 0 0
Ben 0 1 (33.3%) 2 (66.7%) 3 (100%)
Claire 14 (37.8%) 4 10.8%) 19 (51.4%) 37 (100%)
Pat 5 (62.5%) 0 3 (37.5%) 8 (100%)
Karen 9 (60%) 4 (26.7%) 2 (13.3%) 15 (100%)
Sarah 3 (50%) 2 (33.3%) 1 (17.7%) 6 (100%)
Total 31 (45%) 11 (15.9%) 27 (39.1%) 69 (100%)
165
For Claire, only the pronominal function of the object concord is realised and this is variable.
Pronominal forms are not always used to refer to previous utterances and the referent is left to
the context or repeated, e.g.:
... zinde zishushu and *ndiyathanda kakhulu (C7C)
... they long they hot and I like very much
... zinde zishushu ndiyazithanda kakhulu.
‘... they (days) are long and hot and I like them very much.’
Pat does not use object concords forms in interviews 1 to 4, even when these forms are expressly
required in a sentence-manipulation exercise. An example of a turn requiring object concords is
the following:
Ndiyathanda ukufunda iincwadi *thetha nabahlobo nabahlobo bam exercise amthambo
I like to read books talk with friends with friends my exercise exercise
*ndiyathanda (P2C)
I like
Ndiyathanda ukufunda iincwadi nokuthetha nabahlobo bam. Umthambo ndiyawuthanda.
‘I like to read books and to talk to my friends. Exercise, I like it.’
Sarah does not use object concords in interviews 1 and 2. There are no definite object concord
markers in 1, 3 and 5, although object concord pronominal forms are used. The object concord
appears in the sentence manipulation in 2 and 5 but without a -ya- between the subject and object
concord, e.g.:
Utitshalakazi *usinika *usinika ipensile ekhabhatini (S5S)
Teacher she us give she us give pencil from the cupboard
Utitshalakazi uyasinika ipensile evela ekhabhatini.
‘The teacher gives us a pencil from the cupboard.’
Karen’s object concord pronominal forms are omitted in interviews 1, 2 and 5, although there is
one example of an object concord pronominal when she builds her own sentence in interview 1,
e.g.:
Xa ndifuna uku *kudlala nenja yam ndiyayibiza *ngegama *lakho (K1S)
When I want to to play with dog my I it call with name of you
Xa ndifuna ukudlala nenja yam, ndiyayibiza ngegama layo.
‘When I want to play with my dog, I call it by name.’
Karen also uses the object concord form correctly in the sentence-manipulation exercise where
this form is required.
These learners seem to regard the object concord as optional since it has a pronominal function
166
rather than a full semantic function. In addition, its function of showing definiteness is not
understood by these learners. The article system of a language is often learnt later and this may
be one of the reasons why this morpheme is not used by these learners. In addition, English uses
a separate word to show this function, whereas Xhosa uses a morpheme in the middle of the word
(a non-salient position for language learners, according to Slobin 1985).
4.4.18 Njenga-
The Xhosa form njenga- translates as “like” and is attached to the noun or verb. Only Karen and
Sarah use this morpheme correctly, e.g.:
Bathanda uku ukutya ukutya okushushu njengesoup soup *njani *njani. (K3C)
They like to to eat food that is hot like soup soup etcetera etcetera.
Bathanda ukutya ukutya okushushu njengesoup njalo njalo.
‘They like to eat hot food like soup etcetera etcetera.’
Pat and Ben omit this morpheme, e.g.:

Ndiyathanda ukudlala *amathambo isport ihoki (P2S)
I like to play sport sport hockey
Ndithanda ukudlala umdlalo njengehoki.
‘I like to play sport like hockey.’
Pat uses an uncoalesced form in a later interview, e.g.:

Ndithanda ukutya *ikutya *kushushu *njenge isop (P7C)
I like to eat food hot like soup
Ndithanda ukutya ukutya okushushu njengesoup.
‘I like to eat hot food like soup.’
This morpheme occurs rarely in the data, but it is not needed in many cases since the sentence can
be changed to avoid the form.
4.5 CONCLUSION
This chapter has presented the data of my study from a functional and formal perspective.
Throughout the chapter, it is evident that these learners are able to perform a range of functions
with their interlanguage. Although the forms they use are not always correct in the linguistic
context and the hearer carries a heavy burden of trying to work out what is meant, they manage
to convey a range of meanings. It is clear, however, that these learners have moved beyond only
using pragmatic means to convey their meanings and they can manipulate a wide range of
morphemes.
167
In the next chapter, common patterns of acquisition across morphemes and learners are
investigated. In addition, the development of morphemes for individuals is discussed and a possible
sequence of development is proposed.
168
CHAPTER 5
ACQUISITION PATTERNS IN THE INTERLANGUAGE OF LEARNERS OF XHOSA
5.0 INTRODUCTION
The functions and morphemes represented in the data were presented in chapter 4 and it is clear
that there are numerous common patterns of morpheme realisation. This chapter will analyse
common acquisition patterns and suggest possible explanations for them. Further areas of
discussion include individual developmental sequences of particular morphemes, discussion of the
developmental sequence of morphemes in the data, and an overview of the learners’ performance
on different tasks.
5.1 PRINCIPLES AND PATTERNS IN MORPHEME USE AND OMISSION

The results discussed in chapter 4 show that there are common patterns of morpheme use and
common principles of morpheme choice across a range of morphemes which have an agreement
component. Since the form of all agreement morphemes relates to the class of the co-referenced
noun, it might be expected that similar principles would be applied when making choices across
these morphemes and wherever possible, parallels are drawn between my study and the findings
of other studies. A number of patterns of suppliance and non-suppliance of morphemes are
discussed below. It is important to note that some examples may have several explanations, since
different cognitive principles may interact.
5.1.1 Singular forms for plural forms and vice versa

In several places (e.g. noun prefixes, subject concords, negative subject concords, possessives),
singularity/plurality may be marked only on some stems in the utterance. For example, a noun with
a singular noun prefix may be co-referenced with a plural subject concord and vice versa. There
are two possible explanations for this pattern. Firstly, the context often clarifies whether singular
or plural forms are necessary, so the learner may deem it unnecessary to mark number distinctions
more than once in the utterance. Secondly, the pattern may be a result of a strategy to avoid
redundancy, so that plurality is marked only once in the utterance.
169
In a study of first-language acquisition1 of siSwati2 noun classes, Kunene (1986) found that
children (aged 2-3) produced a bare noun stem first. However, they used agreement forms which
marked number on the possessive and the demonstrative forms which followed bare noun stems.
This means that these children acquire agreement forms before they acquire the noun prefix. This
finding is not true for the L2 learners in my study, since they acquire the noun prefix before other
agreement forms. However, an interesting hypothesis is that L2 learners are behaving similarly by
marking singular/plural forms only after the noun rather than on the noun prefix. This may mean
that number distinctions are acquired later, although learners realise that morphemes should be
supplied on co-referenced stems before number distinctions can be made (as for the child).
A further possibility for number acquisition is that plural noun forms are derived from singular
noun forms. However, this would mean that if singular forms are learnt as wholes, one could
expect plural forms to be added to the singular forms, e.g. *abaumntwana (class 2 ba- plus class
1 u-). This pattern is not apparent in the data3. Instead, there appears to be a productive rule
applied to noun stems. Further evidence of rule application is discussed in section 5.1.2.
Nevertheless, it is not possible to rule out the possibility that nouns and noun prefixes are learnt
as wholes and automatised (rather than being the result of a process of rule application), although
the memorisation of hundreds of singular and plural noun forms is cognitively wasteful.
Nevertheless, Kunene (1986:37) says that learning forms as wholes may be the way L1
acquisition proceeds:
One possible explanation as to why Cimcim learned plural prefixes with some nouns and
singular prefixes with others has to do with the frequency of the noun. If the noun more
commonly occurs with its plural prefix, then the child might learn the plural prefix of that
noun before learning its singular prefix. [and vice versa]
1
First- and second-language acquisition is comparable in many ways and similarities have been
found in several studies (e.g. Dulay & Burt 1974a, b; Felix 1978; Mace-Matluck 1979; Suzman 1982,
1999; Wode 1978).
2
Because of the lack of studies of second-language acquisition of African languages, several
first-language acquisition studies are referred to in this chapter. Examples from studies of siSwati,
isiZulu and seSotho are included here. These languages share many of the typological characteristics of
Xhosa and I have made generalisations where they can reasonably be applied to Xhosa because of
typological similarities.
3
Kunene (1986) also finds that this pattern is not upheld in L1 data.
170
This may be a strategy employed by L2 learners, especially at early stages of acquisition when the
extent of the noun class system is not yet fully understood.
5.1.2 Mix forms

When several forms have a common attribute, these forms may be used interchangeably. For
example, forms which denote plurality (classes 2, 6, 10) are used interchangeably for noun
prefixes, subject concords and demonstratives. Since learners would not have heard incorrect
forms in the input, this suggests creative application of forms to stems (rather than learning a stem
as a whole, as discussed above in 5.1.1). It also suggests that semantics rather than phonology
may form the basis of morpheme choice to some extent because, although many agreement forms
are phonological copies of the noun prefix, the learners do not seem to always use this cue. This
finding is in contrast to the findings for L1 acquisition of seSotho since Demuth (1992:630)
concludes that:
There is no evidence from the Sesotho data, nor from the other studies mentioned, that
children learning Bantu languages rely on semantics to help them into the noun class and
agreement system. Rather, there is support for the fact that they rely on phonological
information. In Demuth (1988b), I propose that this holds up crosslinguistically. Access
to the semantics of the system becomes available only at later stages of development,
whereas early overgeneralizations are normally of a phonological nature. We know little
about how and when Sesotho-speaking children do acquire access to the semantics of the
noun class system, such as productively deriving human nouns from non-human nouns.
The cognitive differences between young L1 learners and L2 adult learners may explain this
difference, since semantic choices also seem to play a role in the choices discussed in the next
section.
5.1.3 Animacy cues

In the data, agreement forms of class 9 and 10 animate noun forms are replaced by class 1 and 2
agreement forms respectively for noun prefixes, subject concords, negative subject concords,
demonstratives, and possessives. These substitutions seem to be based on semantic distinctions,
and semantic distinctions have been found to play a role in the acquisition of a range of different
morphemes (Bybee 1991; Carroll 1999; Young 1991). From a semantic perspective, many human
referents fall into classes 1 and 2, while many non-human referents fall into classes 9 and 10. In
addition, the pronoun for “he/she” is u- and the pronoun for “they” is ba-, and the learner may be
using the pronoun “he/she/they” rather than “it” if they are translating from English. The
171
“he/she/they” pronominal forms are the same as the subject concord forms for classes 1 and 2
respectively (although there are tonal differences in pronunciation of pronominal forms and
subject concords which these learners do not realise in their production). Together, these two
factors seem to encourage the selection of class 1 and 2 concords for animate referents which
normally fall into classes 9 and 10. This explanation supports Perdue’s (1991) findings that the
forms for personal pronouns appear before forms for reference to “it”.
Kunene (cited in Demuth 1992:599) found similar overgeneralisations for human referents which
are found in nonhuman classes for children aged 4;6-5;9:
Once the lexical noun and corresponding subject marker were initially used, children’s
sentences would undergo pro-drop and they would switch to class 1/2(the human class)
subject markers. Once these older children deleted the head noun, semantics - at least for
the human class - may have begun to play a role.
Similarly, Suzman (1999:143) found that

(w)hen first year university students learning Zulu as a second language were asked to
compose sentences using ‘human’ nouns in classes other than NC1 [Noun Class 1] or 1a
... they often used agreeing prefixes from NC1 ... . Students have obviously used meaning
rather the form as the basis of agreement.
The difference between these findings and the findings of my study is that the learners in my study
focussed on animacy rather than humanness. So, they extended these substitutions to animals as
well as humans in classes 9 and 10.
5.1.4 Opposite of animacy

However, the principle operating in 5.1.3 is not applied consistently, since the opposite
substitution also occurs. Animate referents which fall into classes 1 and 2 are sometimes given
agreement forms from classes 9 and 10 on noun prefixes, subject concords, negative subject
concords and copulatives. A possible explanation for this principle is that class 9 and 10 forms are
overgeneralised because they have the highest number of lexical items. Frequency of class 9 and
10 forms in the input may encourage selection of these forms (as Musau 1995 predicts).
5.1.5 Marked and unmarked

As mentioned in section 4.4.1.1.1, there seem to be marked and unmarked classes for these
learners when choosing noun prefixes, subject concords, negative subject concords,
172
demonstratives, possessives, qualificatives and qualificative a- morphemes. Classes 1, 2, 9 and 10

seem to be unmarked and learners sometimes substitute forms from one of these classes for forms
from other classes. These classes may be deemed unmarked because:
C they contain the most lexical items;
C most borrowed lexical items are found in these classes;
C they have agreement morphemes which match noun prefixes phonologically; and
C they pattern fairly regularly across morphemes.
Other classes may have few lexical items, hidden concord forms (e.g. class 5 has a hidden li- form
in its prefix and class 11 has a hidden lu- form in its prefix), and unpredictable forms (e.g. class
3 singular to class 4 plural changes).
A more pragmatic reason for overuse of classes 1 and 2 may be that they appear first in the chart
of morpheme inflections and these forms may be the first forms learnt and therefore the forms
which receive the most attention (although this does not explain the use of these forms by Claire
who has had less exposure to formal input).
Musau (1995) found that learners of Swahili overgeneralise class 9 and 10 forms and he links this
to Andersen’s (1984) One-to-One Principle where one form is used to show one function. This
strategy reduces the cognitive processing load, because learners use only a few of the available
forms.
5.1.6 Beginnings of words

In a few morphemes (e.g. locatives and verbal relatives), morphemes may be formed by a prefix
and a suffix. A comparison of the suppliance of prefixes only and suffixes only (when prefixes and
suffixes are required) shows that prefixes are supplied more often than suffixes (cf. Richards
1985). In the case of the locative, the use of only the initial forms is strengthened by the
prototypicality of adding only the e- form to the stem when forming the locative.
Another example of omitting parts of a morpheme occurs when a prepositional form has two
parts. The second part is attached to the noun which follows the first part, and the second part
may be omitted by these learners (e.g. phambi kwa- “before” has omissions of the kwa-
preposition).
173
Double marking of a single function is redundant and this may strengthen the tendency not to
supply both morphemes when one form conveys the intended meaning.
However, there is an exception to this tendency. The negative terminative -i morpheme is usually
produced, but this is probably because it is constant in the present tense, while the initial
morpheme is variable according to the noun class with which it agrees.
5.1.7 Overgeneralised forms

Overgeneralisation can take two forms. Firstly, a single form is overgeneralised to convey one
function in a variety of agreement contexts. Secondly, a morpheme category can be
overgeneralised to new contexts. Overgeneralizations are found in a number of studies (e.g.
Dittmar 1981, 1984; Schumann 1978; Tomasello & Herron 1988).
Examples of the first form of overgeneralisation include overuse of:

C aku- as a negative subject concord;
C class 1 copulatives;
C first-position demonstratives;
C qualificatives which add only ku-;
C agreement forms from classes 1, 2, 9 and 10;
C the e- form of the qualificative a-.
Findings by Suzman (1982 for L1 acquisition of Zulu; 1999 for L2 acquisition of Zulu), show that
the Zulu equivalents of Xhosa classes 1 and 9 are learned early by L1 learners and classes 1, 5 and
9 are learnt early by adult learners. The learning of class 5 at an early stage is a finding not upheld
in my data and it is a surprising finding, because the hidden form of class 5 (i(li)) makes choosing
the correct agreement form more difficult. Reasons for overgeneralisation of classes 1 and 9 have
been discussed in section 5.1.5.
The second type of overgeneralisation occurs when learners overgeneralise verbal negative
morphemes to the noun negative (see section 4.4.4.2) and use the additive conjunction on the verb
(see section 4.4.16.1.4). These overgeneralisations may indicate either a lack of understanding of
restrictions on where forms can occur, or a lack of knowledge of the morphemes needed in the
context. In order to communicate, the learners overgeneralise what they already know, based on
a common function.
174
5.1.8 Redundancy
Several studies have confirmed that there is a tendency to avoid redundancy when learning an L1
and L2 (Bybee 1991; Schumann 1987; Slobin 1985), although Young (1991) found that his
learners marked plurality redundantly. Redundancy is probably avoided to reduce the cognitive
processing load. A lack of redundancy is found in several places in my study, e.g.:
C the use of singular noun prefixes with plural concordial forms (and vice versa);
C subject concord omissions when the form is implicit or repeated;
C the lack of demonstratives with copulatives;
C using only one of the locative morphemes;
C leaving forms to the context; and
C the omission of the qualificative form on numerals.
5.1.9 Restricted functions of morphemes

Although a wide range of morphemes appears in the data, the target-language range of functions
of the morphemes is not always exploited. Many morphemes are used with a single function or
perhaps two functions and less prototypical functions are not used by the learners. Instead learners
focus on a particular function (sometimes based on transfer from English) and use the morpheme
in a restricted range of contexts. Wolf-Quintero (cited in Towell and Hawkins 1994) describes
this as the principle of conservatism, since learners are reluctant (or unable) to employ the full
range of functions of a morpheme initially. For example, object concords are only used with a
pronominal function rather than as definiteness markers as well, and locatives are not always
understood to have prepositional functions. The functions of the Xhosa copulative listed by
(Jokweni 1997) are not exploited since these learners use the copulative mainly as the copula
would be used in English.
Another example of restricted function of morphemes is the case where learners have difficulties
with the non-present-tense forms of some morphemes (e.g. subject concords, negative subject
concords and demonstratives). They learn the present-tense form in the first year of instruction,
although for second- and third-year students and those with exposure to natural input, the most
frequent tenses they are exposed to are probably not present tense forms.
5.1.10 Distance from the noun

Agreement morphemes are co-referenced with the noun and if they occur relatively far away from
the noun they are more likely to be incorrect than those right next to the noun. This is probably
175
a result of processing constraints. The learner continually has to refer back to the noun in the
utterance to produce the correct agreement form at a later point in the utterance. Subject
concords are particularly prone to this strategy since they are forms which can occur far away
from the noun. Many of the other agreement forms occur within the noun phrase (e.g.
possessives, qualificatives), but they also follow this pattern if there is a question or comment
from Thembi between the noun produced by the learner and the agreement form produced after
Thembi’s turn. This pattern links to Wolf-Quintero’s (cited in Towell & Hawkins 1994) principle
of continuity, which says that adjacent forms are more likely to be combined.
5.1.11 Processing problems

McLaughlin (cited in Gass & Selinker 1994:154) says that “(p)rocessing energy is limited and
must be distributed economically, if communication is to be efficient.” Vocabulary retrieval seems
to require high levels of processing energy and morphemes such as subject concords and
qualificative a- morphemes are omitted or used incorrectly if there are difficulties with vocabulary
retrieval. Processing constraints are particularly obvious for Ann, because she has many problems
retrieving vocabulary.
Adding additional forms to the noun (e.g. na-/nga- “with”) may interfere with suppliance of the
noun prefix, because this increases the processing load, especially when there is a need for vowel
coalescence. Furthermore, initial vowels may be left off noun prefixes, especially at the beginning
of utterances. It is possible that processing loads may be fairly high at the beginning of utterances
where the learner is shifting from English to Xhosa and thinking about the message to be
produced. All participants seem to struggle with adding agreement morphemes to words
containing clicks, perhaps because of the extra processing power needed to produce these non-
English sounds.
5.1.12 Discreteness of forms

To some extent it is possible that words are viewed as fairly separate entities by these learners and
that preceding or subsequent words do not affect words. Examples of forms in my study which
do not show an awareness of relationships between morphemes and other words in the utterance
include the lack of:
C initial vowel dropping on a noun which follows a negative;
176
C the subjunctive on successive verbs in the present tense; and

C object concords.
The one place where initial vowels are dropped on the following noun is with the demonstrative.
This may be a result of the phonological incompatability of successive vowels of the
demonstrative and the initial vowel of the noun (e.g. lo (u)mntu).
5.1.13 Influence of English

When English is used in an utterance, it seems to interfere with the use of morphemes on noun
prefixes and locatives. At a cognitive level, the learner is moving between two different underlying
systems which may interfere with each other.
Transfer from English is evident on a range of morphemes (e.g. additive conjunctions, omission
of possessives, lack of prepositional forms, no nga- “with” forms on locatives). Positive transfer
may occur for word order, but generally little is directly transferable because of the difference in
typology of English and Xhosa. Transfer tends to be more noticeable at the discourse level, where
there may be over- or underuse of emphasis, disturbed topic-focus patterns and overuse of noun
phrases where pronouns are more appropriate.
5.1.14 Structure of words

Verb roots which begin with vowels seem to be marked forms which are irregular when
concordial elements are attached to them. Usually an extra form needs to be added to link the
vowel of the concord to the initial vowel of the verb. Monosyllabic verbs are also marked in terms
of the way in which concords are added, since they may also include extra forms. These
difficulties conform to Wolf-Quintero’s (cited in Towell & Hawkins 1994) principle of
generalisation which says that learners avoid exceptions.
5.1.15 Phonological constraints

The confusion of class 2 (ba-) and class 6 (a-) forms (on noun prefixes, subject concords, negative
subject concords, demonstratives and qualificatives) may be the result of phonological constraints.
The a- vowel is common to both classes in the agreement forms, which may influence selection
of the appropriate morphemes. A second example is the substitution of e- and i- on noun prefixes,
which may also be a result of phonological perception problems.
177
5.1.16 Multiple functions for one form

Multiple functions for one form seem to cause problems as Andersen’s One-to-One Principle
(1984) suggests they might. The learner’s preference is usually for one form per function. The
form which seems to cause the most problems for these learners is the morpheme a-. This form
is used as the class 6 subject concord, qualificative a-, negative formative, and possessive of class
6. The learners in my study also use it as a default subject negative form on its own possibly
because it is the negative formative. When choosing forms for class 6, the multifunctionality
causes problems and learners seem to hesitate to supply agreement forms which only consist of
a-.
The ku- form is also multifunctional. Visser (1989:155) lists a range of forms which use ku-,
including the infinitive, the class 15 noun, and agreement forms with class 15. The confusion of
forms such as ukutya (to eat/food) may stem from the multifunctionality of the uku- form at the
beginning of this word.
Wolf-Quintero’s (cited in Towell & Hawkins 1994) principle of uniqueness is relevant here, since
learners prefer one form to have one meaning. Young (1991) also notes that a pattern of one form
to one function which is easy to learn is preferred.
5.1.17 Rule application

Rule application appears before automatisation (after a period of formulaic use which is
automatised). This principle is shown in coalescence, where learners do not alway produce a
correctly coalesced form. The errors made for locatives and noun prefixes also show a preference
for rule application rather than automatisation (although the incorrect form may have been
automatised).
5.1.18 Conclusion
In the above section, several common patterns of morpheme use and explanations for these
patterns have been discussed. It is evident that most of the above principles have been found in
other studies of interlanguage, which shows that they are probably general language-learning
principles, regardless of the typology of the languages involved. In the next section, the levels of
development of the learners in my study will be investigated.
178
5.2 LEVELS OF DEVELOPMENT

Levels of development have been very broadly defined in the literature because of the wide range
of factors which needs to be taken into account when assessing development in interlanguage
(IL). A particular problem in measuring development in an agglutinative language is that most
words have more than one morpheme, which means that presenting noun:verb ratios or utterance
length measured in number of words in the data is problematic. Even in L1 development, where
the standard measure of development is the Mean Length of Utterance (MLU), there is criticism
of this measure for agglutinative languages (Demuth 1992). However, I have applied some of the
measures of development suggested by Givón (1979) and Corder (1977) to the data and I will
show where these measures fail. It should be noted that none of these measures of development
can give an accurate picture on its own and that the interaction of the measures should be
considered for a more reliable assessment.
It should also be noted that different tasks influence the level of language produced. So, in some
tasks, learners’ interlanguage appears to be more developed, while high task demands result in
backsliding to earlier forms in other tasks4. Givon’s (1979) and Corder’s (1977a) characteristics
of early IL will be investigated below.
5.2.1 Givón’s pragmatic-syntactic continuum

The first level of development which I will discuss is that of Givón (1979). He says that learners
progress from pragmatic to syntactic levels of development and, for my data, it is necessary to see
this model as a continuum. The pragmatic mode is characterised by inter alia:
C loose coordination;
C lower noun:verb ratios;
C simple verbs; and
C no use of grammatical morphology.
For many of the measurements, the level of development is a ranking of each of the six learners
against each other, since there are no other available standards against which these results could
be accurately measured. However, ranking these learners shows how they might be placed on
Givón’s (1979) continuum in relation to each other.
4
Appendices G1 to G20 provide an impression of the suppliance and non-suppliance of
individual morphemes for each task in the data.
179
5.2.1.1 Conjunction use

The first area of investigation is the use of conjunctions in the data. Table 5.1 indicates that
coordinating conjunctions are not the only conjunctions in the data, as Givón (1979) suggests they
should be at early levels of development.
Table 5.1 Conjunction use in the data

Name Number of Number of subordinate Conjunction total Number of different
coordinate conjunctions conjunctions used
conjunctions
Ann 3 16 19 2
Ben 6 55 61 7
Claire 47 145 192 11
Pat 36 45 81 4
Karen 63 141 204 13
Sarah 72 121 193 9
None of these learners employ only loose coordination to join elements of utterances. In fact, use
of subordinate conjunctions is higher than use of coordinate conjunctions for all learners. This
means that all learners have moved beyond the pragmatic level of development in their
conjunction use. Using the total use of conjunctions as a measure of the level of development,
the learners5 are ranked from lowest to highest in the following way:
Ann, Ben, Pat, Claire, Sarah, Karen
Another way to look at conjunction use is to look at the number of different conjunctions
employed. This may give a better indication of conjunction use, since a large repertoire of forms
can show an ability to produce more sophisticated meaning relations between parts of the
utterance. Using this measure of development, the following ranking (lowest to highest) is found:
Ann, Pat, Ben, Sarah, Claire, Karen
The problem with measuring development by looking at the number and range of conjunctions
in this data is that learners are producing oral data for the measurement. Using conjunctions is to
5
A reminder of the background of each learner is included in Appendix B.
180
some extent optional, since suprasegmental features can be used to separate utterances which
might be joined by conjunctions in writing. Also, speech is usually less complex with regard to the
use of conjunctions, other than fairly simple additive conjunctions like “and”, “but”, etc. This is
because of the processing constraints involved in the use of conjunctions which express more
complex meaning relations. Task constraints may also influence the number of conjunctions used
because, where more processing energy is needed (e.g. in picture descriptions where there are
vocabulary retrieval difficulties), simple utterances may be preferred.
5.2.1.2 Noun:verb ratio

The noun:verb ratio indicates the number of subjects and objects per verb. Early development is
characterised by a low noun:verb ratio (often 1:1), although very early stages have many nouns
and very few verbs. The noun:verb ratio is fairly problematic when measuring an agglutinative
language, since, although many morphemes are added to noun or verb stems, the result is not a
noun or a verb in terms of form or function. Forms like the copulative are added to the verb to
form a verb-noun complex and selecting the forms to count is therefore problematic. Nevertheless,
a rough guide to the noun:verb ratio may be obtained by comparing the number of noun stems to
the number of verb stems. The following ratios are evident:
Ann 1.4:1
Ben, Pat 1.3:1
Karen, Sarah 1.1:1
Claire 0.9:1
In an agglutinative language, the ratio of stems with morphemes to stems without morphemes
might be a more accurate measure of development, since the use of uninflected stems indicates
a lower level of development (although there are a few forms which may be correctly omitted, e.g.
the subject concord on an imperative). Results of this measurement are presented in the discussion
of the use of grammatical morphology in 5.2.1.4.1.
5.2.1.3 Verb complexity

Verb complexity is another measure used by Givón (1979) to measure level of development.
Bardovi-Harlig (1994), Schumann (1987) and van Els et al. (1984) also mention that verbal
morphology indicates higher levels of development. They look at the auxiliary and modal forms
which are used with the verb to measure development. In Xhosa, the auxiliary forms do not have
the function of indicating tense as English forms do. For this reason, it is necessary to extend the
181
category of forms counted to include tense markers. Auxiliaries and verbal extensions are counted
because both of these categories are additions to the verb stem. The range of forms (number of
different types of morphemes) used by the learners is shown in Table 5.2.
Table 5.2 Range of forms added to the verb

Name Non-present tense markers Auxiliary forms Verbal extensions Total
Ann 0 0 0 0
Ben 4 0 0 4
Claire 3 2 2 7
Pat 4 0 1 5
Karen 4 3 2 9
Sarah 4 2 2 8
The table indicates that a fairly small range of forms which increase the complexity of the verb
phrase is used by these learners. Ranking of learners from lowest range of forms to highest range
of forms is as follows:
Ann, Ben, Pat, Claire, Sarah, Karen
The use of tense markers is over 90% for all learners, so verbal morphology is used (although not
always correctly or extended beyond the present tense). These learners therefore have a highly
syntacticised interlanguage (IL) in Givón’s (1979) terms.
5.2.1.4 Use of grammatical morphology

According to Sato (1988), the use of grammatical morphology is one of the most widely
researched features of L2 levels of development. Givón (1979) claims that the absence of
morphology is a reliable indicator of early language development and different formulae have been
used to measure the use of grammatical morphology.
5.2.1.4.1 Morpheme suppliance

Firstly, there is a measure of suppliance of morphemes. This shows whether morphemes are
absent in the IL. Pica (cited in Gass & Selinker 1994:45) says that a reliable measure of suppliance
should give 2 points to the suppliance of a correct form and 1 point for suppliance of an incorrect
182
form, as well as an indication of the total number of obligatory contexts. Her formula is the
following:
(number of correct suppliance x 2) + (number of misformations)
total obligatory contexts x 2
Using this formula across 15 morphemes discussed in chapter 4, the following ranking is obtained
for the six learners (the percentage of morphemes supplied is indicated in brackets):
Pat (74%), Claire (77%), Ann and Ben (82%), Sarah (90%), Karen (91%)
It is clear from these figures that these learners do not lack morphology. Even Ann, who has only
learnt Xhosa for a short period of time, does not omit a very large number of morphemes. There
are several possible explanations for this:
C this formula may not be an accurate measurement of development for learners of
agglutinative languages because of the predominance of morphemes which carry meaning;
C these learners are not early learners;
C all these learners have been influenced by the formal context so morpheme use may be
higher than for learners in natural contexts. The formal context may help learners to realise
that the use of morphemes is vital for Xhosa and they may therefore pay more attention
to them (Schmidt 1990; Tomlin & Villa 1994). Measuring morpheme use may not
therefore be an accurate way to measure the development of these learners;
C most morphemes in Xhosa do not have only grammatical functions, but also have semantic
functions. Omitting morphemes therefore reduces the meaning which can be conveyed.
5.2.1.4.2 Target-like use

Another similar measure of morpheme use relates to target-like use of morphemes in the data
(Pica cited in Gass & Selinker 1994:45). In this formula, the number of forms which are supplied
in non-obligatory contexts is included in the calculation. This number is included because learners
may use forms where they are not required, which indicates non-target-like use of the morpheme.
A weakness of measuring only target-like use is that learners may be using a morpheme to
perform a function, but this is not reflected in the calculation (Long & Sato 1984). The formula
is as follows:
number of correct suppliances in obligatory contexts
number of obligatory contexts + number of suppliances in non-obligatory contexts
The following ranking is found for the learners (the percentage of correct suppliance is provided
in brackets):
183
Pat (61%), Claire (64%), Ben (73%), Ann (74%), Sarah and Karen (86%)
Comparing the figures for morpheme suppliance and target-like use, it is clear that use of
morphemes is always higher than target-like use of morphemes. The difference between the two
figures is higher for less proficient learners (8%-13%) and lower for more proficient learners (4%-
5%). Although learners are using forms of the morphemes to convey a range of functions, they
are not always able to provide the correct form of the morpheme.
5.2.1.4.3 Ninety-percent suppliance

Related to the previous measurement, an indication of the number of different morphemes which
reach 90% suppliance for each learner reveals the following ranking:
Pat (none)
Ann (noun prefix)
Ben (i-ending on negatives)
Claire (i-ending on negatives)
Sarah (noun prefix, subject concord, i-ending on negatives)
Karen (noun prefix, subject concord, i-ending on negatives, locatives, demonstratives,
associative copulative)
This is a low number of morphemes which can be said to be used in nearly target-like ways, which
shows that these learners do not have target-like competence in many morphemes, although they
may use them to convey the functions they need in their interlanguages.
5.2.1.4.4 Number of different morphemes in the data

A different way of looking at levels of development with regard to grammatical morphology is
to look at the range of morphemes used by each learner. The range of morphemes is obtained by
counting the number of different morpheme types (rather than forms of each morpheme) which
appear in the data. It should be noted, however, that some morphemes may not be used
productively to any great extent. The tenses which are indicated are non-present tenses, since the
present tense is used by all learners. Using this ranking, the following pattern emerges:
Ann (15 different types of morphemes)
Ben (17 different types of morphemes and 4 different tenses)
Pat (19 different types of morphemes and 4 different tenses)
Claire (20 different types of morphemes and 3 different tenses)
Karen (21 different types of morphemes and 4 different tenses)
Sarah (21 different types of morphemes and 4 different tenses)
184
The problem with this approach is that learners may know other morphemes but choose to avoid
the use of these morphemes in speech (this is supported in the discussion of the morphemes used
in written work in section 5.7). Furthermore, they may have alternate ways of producing the
morpheme, e.g. using the adverb, phaya (there), instead of the demonstrative. Avoidance of
morphemes may not mean that they cannot produce them, but that they choose not to produce
these forms. These learners are certainly using only a subset of the full set of morphemes available
to the L1 user, so even though 21 different types of morphemes may seem impressive, there are
many forms which they do not use and conjugations which they do not employ (e.g. auxiliaries,
past tense forms, negatives). It should also be noted that the study is based on a small sample of
speech which may not elicit the full range of forms and functions these learners know.
5.2.1.4.5 Range of morphemes used to realise the referential, spatial and temporal functions
The number of different morphemes used to realise the functions investigated in the data also
supplies a useful measure of development. Table 5.3 indicates the number of different morphemes
used to realise the functions discussed in chapter 4.
Table 5.3 Number of morphemes used to realise each function

Name Referential function Spatial function Temporal function
Ann 4 4 2
Ben 3 3 7
Claire 5 3 6
Pat 6 3 7
Karen 6 4 7
Sarah 6 4 7
This measurement ranks learners against each other, since these forms are only subsets of the
number of morphemes available to Xhosa speakers who wish to realise the same functions. The
rankings for each of the functions follows, from lowest to highest (the number of morphemes is
shown in brackets):
Referential function: Ben (3)

Ann (4)
185
Claire (5)
Pat, Karen, Sarah (6)
Spatial function: Ben, Claire, Pat (3)

Ann, Karen, Sarah (4)
Temporal function: Ann (2)

Claire (6)
Ben, Pat, Karen, Sarah (7)
More advanced learners usually use a wider range of morphemes to express particular functions.
Overuse of a few forms is characteristic of early IL, since learners use a small range to accomplish
as much as possible in terms of communication.
5.2.1.5 Reliability of rankings

The rankings which result from these measures seem to be fairly consistent. Karen is consistently
the most advanced learner according to the ranking. Sarah is second to her on all but one measure
(range of conjunctions). The lowest ranking is occupied by Ann in 5 instances, by Pat in 3
instances and by Ben in 3 instances. Where range of forms is ranked, Pat is usually higher than
Ann and Ben, but where accuracy is measured, she usually scores below them. Pat and Claire
occupy the two lowest rankings for morpheme suppliance and target-like use and this may be a
result of their high levels of naturalistic input. The formal context which the other four learners
have experienced may have played a role in their use of morphemes, although the range of forms
they use may develop slowly over a long period of time.
5.2.2 Corder’s characteristics of simplified systems

Corder’s (1977) list of characteristics of simplified systems includes characteristics which are not
mentioned by Givón (1979). According to Corder (1977), the learner’s simplified system is
characterised by inter alia:
C a simple or virtually non-existent morphological system;
C more-or-less fixed word order;
C a simple personal pronoun system;
C a small number of grammatical function words and grammatical categories;
C the absence of an article system;
C some deictic words; and
C syntactic relations expressed by word order.
186
The presence of a morphological system has been discussed in section 5.2.1.4.1. It should also
be noted that Corder (1997) notes that semantic functions are usually realised by lexical means
rather than morphemes in the learner’s interlanguage. However, this cannot be entirely true of
learners of agglutinative languages because of the greater reliance on morphology to mark
semantic functions in these languages.
5.2.2.1 Fixed word order

The word order used by these learners is usually SVO, but this is the word order of both English
and Xhosa, and so positive transfer should occur. Nevertheless, word order in Xhosa can deviate
from the SVO order for emphasis in discourse. Gough (1992) says that Xhosa is a SV(O)
language, but VS also occurs because the subject concord attached to the verb links cataphorically
to the subject. This VS structure is used for foregrounding events and is normal for oral narratives
where events are important. These learners do not use this VS structure, and prefer to use the
SVO pattern, probably because of transfer from English and lack of knowledge of the discourse
effects of the choice of VS patterning.
5.2.2.2 Simple personal pronoun system

A simple personal pronoun system can be looked at in two ways. Firstly, there can be a small
subset of forms from a larger set of forms of a particular morpheme type. Secondly, only a small
range of morphemes employed as pronominals in the target language (TL) may be produced in
the IL.
If one considers the personal pronoun system to consist of morphemes for “I/we/he/she/you-
sg/you-pl/they”, then only Ann is unable to produce all of these forms correctly. She is able to
produce singular forms of these pronouns, but confuses plural forms in some places. None of the
learners produce the “it” form correctly all the time in the data, and this is probably because of
the complexity of the system, which has a range of different agreement forms.
The learners also only produce some of the range of pronominal forms available in the TL. Of the
five types of pronouns (absolute, demonstrative, quantitative, emphatic absolute and possessive
pronominal stem) discussed by Gough et al. (1989) and du Plessis (1978), the learners produce
the absolute pronoun and the possessive pronominal stem. The demonstrative (this/that/that over
187
there) and the quantitative (all) are not used as pronominal forms, but occur only with the noun.
The only exception is the use of the quantitative stem -odwa (only), which occurs a few times in
an inflected form, ndedwa (by myself). The emphatic absolute pronoun is not used at all by these
learners, probably because its function is peripheral to their needs. The possessive pronominal
stem is usually -m (I), and there are problems with other forms of this morpheme (e.g. -khe
“he/she”).
Several studies of pronouns show that they can be confused at early stages (Felix 1981), and that
proper nouns are preferred to pronouns. Myles et al. (1999) found that pronouns occur first in
formulaic chunks, then they are realised as full NPs and then the pronoun is used correctly. Perdue
(1991) found that singular forms appear before plural forms (which seems to be true for Ann),
pronouns for animates before inanimates (which can be extrapolated as use of classes 1 and 2
before other classes, which is true only to some extent), and definitely referring NPs before overt
pronouns. This last point is realised in several ways in the data, with the use of full object NPs
instead of object concords6, and full subject noun phrases instead of subject concords.
Furthermore, where pronominal forms could appear on their own (e.g. demonstratives), these
forms usually occur with the noun.
The use of the NP instead of a pronoun can be explained in two ways. Firstly, the learners use
fairly short utterances and the utterances are not always linked together by means of pronominal
forms. Each utterance is seen as a new message and the NP is therefore repeated. A second
explanation may be that they are avoiding the pronoun, preferring to use the NP which may be
easier to retrieve or require less processing power because agreement forms do not have to be
accessed.
From a discourse point of view, it is important to note that pronouns are probably not used by
these learners as often as L1 users use them, because they use the NP where the pronoun is used
by L1 speakers. Baseline data is not available to test this hypothesis, but Gough (1992) says that
the VS order where the subject concord is added to the verb is a normal foregrounding move for
narratives. The first element of a sentence/utterance is most prominent and emphasized and
6
Suzman (1982, 1999) notes similar late acquisition of object concords for L1 and L2 Zulu
acquisition.
188
overuse of the NP by these learners may create non-target-like discourse effects.
Suzman (1982:54) found that Zulu “children first use concord markers as pronominal elements
which arise in discourse or in non-linguistic context” and this is also true of this data, where
pronominal subject concord forms are used before agreement forms, and object concords are used
with pronominal functions before they acquire definiteness functions.
5.2.2.3 Few grammatical function words or grammatical categories

These learners use between 15 and 21 different morphemes, but many of these morphemes have
semantic rather than purely grammatical functions. There are very few morphemes which have
only a grammatical function in Xhosa, because much of the meaning is carried by morphemes. The
only purely grammatical morpheme used by these learners is -ya-.
5.2.2.4 Absence of article system

The article system is shown primarily by the use of object concords in Xhosa, since the use of
these forms denotes definiteness. Since object concords are used very rarely in the IL of these
learners, there is no fully-fledged article system.
Suzman (1982) found in her study of L1 Zulu acquisition that subject-verb agreement occurs
before object-verb agreement because object-verb agreement is optional and indicates
definiteness. Subject-verb agreement is obligatory and word initial, while object-verb agreement
is less salient because of phonetic structure. Objects are realised as full noun phrases in her study,
as they are in my study.
5.2.2.5 Some deictic words

There are some deictic words in the IL of the learners in my study because demonstratives and
adverbs are used to indicate spatial relationships. The functions of demonstratives for these
learners seem to differ from those of L1 speakers. Gough (1992) says that demonstratives
function in discourse as definiteness markers and thematic prominence markers rather than spatial
deixis indicators for L1 speakers. However, the spatial deixis function is the primary function of
demonstratives in my data.
189
5.2.2.6 Syntactic relations shown by word order

In Xhosa, many syntactic relations are shown by the use of morphemes and these morphemes are
usually employed, as shown by the calculations in 5.2.1.4. The common SVO order of English and
Xhosa also helps to encode correct syntactic relations, although the subject is not replaced by the
subject concord as in TL use (Gough 1992).
5.2.2.7 Conclusion
In conclusion, it appears that the IL of these learners does not conform to all the characteristics
of simplified systems mentioned above, which may mean that:
C they are not early learners; and/or
C this is not a good set of criteria for measuring development of learners of agglutinative
languages.
5.3 SEQUENCES OF MORPHOLOGICAL DEVELOPMENT

A number of studies have shown that there are fairly common sequences of development for
morphemes as a whole. Cazden (1968) lists four periods of development for L1 development
which seem to be generalisable to L2 development:
C no inflection;
C formulaic;
C communication with many errors and overgeneralisation;
C 90% correct use of a form.
Within the development of individual morphemes, this pattern is evident, although it cannot be
said to hold true for the IL as a whole because it consists of morphemes at different levels of
development. Only a few morphemes reach the 90% correct suppliance level, as indicated in
section 5.2.1.4.3. Ann uses formulaic forms before forms with no inflection at early stages of
development to some extent, and other studies support her sequence (e.g. Myles et al. 1998;
1999).
A second sequence, specifically for L2 development, is presented by Skiba & Dittmar (1992):
C words are put next to each other but they are not related;
C syntax starts to develop;
C IL and TL converge and syntax and morphology is fairly target-like.
These levels of development are fairly broad and all three levels are found if one looks at
190
individual morphemes. Looked at from a broader perspective, the IL of learners has examples of
all three levels, because morphemes are at different levels of development.
Suzman’s (1999) study of L1 Zulu acquisition found that the early sequence of development is:
Stage 1: singular forms of class 1a, 5 and 9, the subject marker, possessive, demonstratives
and pronouns in the first stage (2 years);
Stage 2: other singular classes 1 and 7, the object concord, relative and adjective and the
plural forms of noun classes (2, 6, 8, 10). This second stage is at about 2;6 years;
Stage 3: complex sentences with pronominals, negative morphemes, verbal extensions
including questions, some passives, relative clauses, cleft constructions and
expletive/impersonal constructions.
To some extent, the learners in my study follow this pattern. They learn class 1, 2, 9 and 10 forms
first (although the lack of number marking is not apparent for them, possibly due to advanced
cognitive skills). The possessive and the demonstrative forms are later forms for these learners.
Many of the forms from stage 3 are learnt later or appear only in the speech of more advanced
learners, as they do for Suzman’s subjects.
The sequence of development in Ann’s IL is the following:

1) Use a formula wherever possible in the context;
2) Use a form correctly if it has been recently taught and omit it if not recently taught;
3) Use the root form only;
4) Learn one form and apply it to several contexts (usually class 1 or 9);
5) Use the correct and incorrect forms variably, depending on the retrievability of the
vocabulary and the correct morphemes.
Ann’s sequence conforms roughly to the three sequences mentioned above, but the influence of
the formal learning context needs to be taken into account, as in level 2. Level 2 also supports the
findings of the influence of formal instruction which show that if the learner is not ready to acquire
the morpheme, it may appear in the interlanguage (IL) immediately after it has been taught but
will not remain in the IL (Pienemann 1984; Towell & Hawkins 1994). Levels 3 to 5 concur with
Wode’s (1981) study of German learners of English.
Ben’s sequence of development clearly shows the use of class 1 forms first with extension to a
few other classes (e.g. omnye (one), negatives and some subject concords). The agreement
morphemes seem to develop from agreement in phrases, to agreement in utterances, to agreement
191
across turns.
For the referential function, Claire develops from using mainly noun phrases to subject concords
to pronominal forms to object concords, a sequence also found by Suzman (1999) for L1 learners
of Zulu.
Generally, it seems that these learners follow the morphological development patterns found in
other studies, which probably means that general language-learning principles apply in this study.
5.3.1 Order of teaching

Although research (discussed in section 2.7) has established that formal teaching probably has
very little influence on the construction of the IL system, it is nevertheless necessary to indicate
the order in which morphemes were presented to learners in the first year. It is assumed that
second- and third-year learners have received formal instruction in all the morphemes presented
to first-year learners, as well as additional instruction on other elements of the grammatical
system. This does not necessarily imply, however, that these forms have been incorporated into
their interlanguages. In particular, it is noticeable that although forms appear briefly in the
learners’ interlanguages soon after they have been taught, their use may atrophy over time (e.g.
demonstratives appear in interview 3 and they are forgotten by interview 7 for Ann). At later
stages, the production probably relies on an internalised system rather than the formally taught
system, although the acquisition of the internalised system may be hastened by the demands of the
formal situation. The positive influence of formal instruction on the rate of acquisition is seen in
Karen and Sarah whose performance on many morphemes is better than Claire’s, even though
Claire has been learning Xhosa for a longer period of time and probably has had more exposure
to it than any of the other learners.
In Table 5.4, the order of teaching, the order of appearance of morphemes and the order of
accuracy of morphemes are contrasted for Ann. She seems to be the learner most likely to be
influenced by the formal context since she has no input from naturalistic contexts. “Order of
teaching” refers to when the forms are first taught, although some forms are revised and/or
retaught after the first teaching. “Order of accuracy” refers to the average accuracy of use of a
morpheme over the course of the interviews (although this obscures development).
192
Table 5.4 Comparison of order of teaching, appearance and accuracy for Ann
Order of teaching Order of appearance Semantic/ Order of accuracy
(date in brackets) in interviews (date in brackets) grammatical
morpheme
noun prefix interview 1 mainly 1

(23/2) (16/3) grammatical,
partly semantic
subject concord interview 1 semantic and 5

(9/3) (16/3) grammatical
-ya- (9/3) interview 1 (16/3) grammatical 3

na-/nga- (with) interview 3 semantic 2
(16/3) (18/5)
copulative interview 3 semantic and 6
demonstrative, 1st interview 2 semantic and 8

position (23/3) (20/4) grammatical
negatives interview 2 semantic and 12
demonstrative, interview 3 semantic and never correct
2nd position (4/5) (18/5) grammatical
demonstrative, 3rd never semantic and not used
position (11/5) grammatical
na- (have) (12/5) interview 7 (19/10) semantic 11
qualificative interview 3 semantic and 10

qualificative a- interview 4 semantic and 9
possessive interview 5 semantic and 7
absolute pronouns interview 4 semantic and used only 3 times,
(24/8) (27/7) grammatical so ranking is not
reliable
locatives interview 3 semantic 4
(5/10) (prototypical forms) (18/5)
future tense (19/10) never semantic not used

193
From the above data, it is not possible to say that morphemes which fulfill semantic functions are
always learned before morphemes which fulfill grammatical functions. Firstly, many morphemes
with semantic functions have grammatical functions as well. Secondly, the only morpheme with
a purely grammatical morpheme which appears in the data, -ya-, occurs before several semantic
morphemes. On the other hand, the learners may not produce some of the other possible
grammatical morphemes and this would mean that these forms are acquired later than semantic
forms. For example, some of the functions of the subjunctive and the participial are primarily
grammatical, and these forms are rarely realised in the data, and only appear in the ILs of more
advanced learners.
It is clear from Table 5.4 that the order of teaching does not necessarily correlate with the order
of appearance7 or the order of accuracy, since locatives appear in the data long before they are
formally taught.
5.4 GENERAL SEQUENCES OF DEVELOPMENT FOR MORPHEMES

As found in the morpheme studies (see Wode 1981 for an overview), it is not possible to work
out an exact sequence of development of morphemes found in this study. Instead, it is only
possible to indicate which morphemes appear earlier and which appear later (cf. Andersen 1978).
This could be interpreted as a weakness since it does not tell us enough about the sequences of
development. However, I feel that even fairly tentative indications of a sequence of development
can be useful. Firstly, the sequence tells us about the functions necessary to communicate at early
levels of development. Secondly, it tells us which forms learners feel they need to pay attention
to when they are learning to speak the new language. Thirdly, comparisons can be made between
the interlanguages of different languages by looking at the functions performed by the morphemes,
and generalisations can be made from this type of comparison. Fourthly, it can serve as a guide
as to whether morphemes should be taught earlier or later, so that the natural acquisition sequence
can be maximally utilised.
Finding a sequence of development across learners is difficult because there is a difference

between a sequence based on order of appearance and one based on accuracy. A further
7
The forms recently taught are included in the elicitation tasks so the order of appearance may
not be a reliable indicator of natural development.
194
problematic feature is avoidance, where learners are able to use a form in their class work and
tests, but it never appears in the oral IL of the interviews. These morphemes cannot be captured
in the sequence of development which I present because it is based on the oral IL. Another
possible problem is that some forms occur or occur more frequently in the oral data because they
are required in the elicitation tasks. Nevertheless, an analysis of the use of forms shows that all
forms appear in all tasks at some point in the interviews, although frequency and number of errors
may differ8. It is also important to note that some studies (Ellis 1987b; Krashen et al. 1976) claim
that orders are disrupted for formal elicitation tasks.
5.4.1 Order of appearance

Using a measure of order of appearance (based mainly on Ann’s forms at early stages since other
learners have already acquired most of the forms), the following sequence is found:
Early acquisition (appear in 1st interview for Ann): noun prefix

subject concord
-ya-
Later acquisition (appear in 2nd/3rd interview for Ann): locatives in the prototypical form
na-/nga- (with)
qualificative
demonstrative
negative subject concord
adverb
copulative
Even later acquisition (only from 4th interview for Ann): conjunction
qualificative a-
absolute pronoun
possessive
associative copulative (have)
Latest acquisition (never in some learners): non-prototypical locative

additive conjunction
object concord
participial
extensions on verbs
njenga- (like)
8
It should be noted that several morphemes occur too infrequently (below 50 occurrences) for
analysis to determine reliable ordering.
195
In order to support the findings on development, comparisons of the interviews which were
repeated (interviews 2 and 5, and 3 and 7)can show that some forms appear later in the IL in
equivalent contexts. It is clear from a comparison of interviews for Ann that there is development
in her IL, mainly in terms of the range of morphemes which is available to express particular
meanings. The morphemes which do not appear in 2, but which are used in 5, include copulatives
and possessives. Demonstratives and locatives are found in both interviews, as is a preference for
the forms ndi- (I) and u- (you/he/she). There are also similar errors in both interviews, e.g.:
Ihashe *ukutya ingca (A2T)
Ihashe *ukutya ingca (A5T)
Horse to eat/food grass
Ihashe litya ingca.
‘The horse eats grass.’
This example shows continued difficulty with the multifunctional ku- form, which can be a class
15 agreement form or an infinitive.
An example which shows a wider range of morphemes in an equivalent utterance follows:

Amehlo *umkhulu (A2C)
Eyes are big
Amehlo amakhulu
‘Eyes that are big’
*Igamehlo *nehashe *ezinkulu (A5C)

Eyes and horse that are big
Amehlo ehashe amakhulu
‘Eyes of the horse that are big’
In the first example, Ann uses a qualificative agreement form to indicate the qualificative and the
utterance contains no qualificative a-. In interview 5, although the forms are incorrect, she has
added a possessive and the qualificative a- form.
Another example shows how Ann’s vocabulary increases over the course of the interviews:
inyama (A2T)
meat
Abazali *basenga inyama kodwa abantwana *basenga (A5T)

Parents they milk meat but children milk
Abazali bathenga inyama kodwa abantwana bathenga (iilekese).
‘Parents buy meat but children buy (sweets).’
196
This example shows development from what is usually a word or two (nouns and verbs) to
utterances which use the locative and some qualificatives. The utterance length therefore
increases.
Ben’s development can be measured in terms of:

C complexification (wider range of morphemes used in an utterance, e.g. demonstratives,
na-/nga- (with), infinitive added);
C increased accuracy when using the morphemes (e.g. demonstratives, possessives,
coalescence); and
C an increase in the number of times morphemes are used (e.g. copulatives, future tense).
Claire, Pat, Karen and Sarah do not show the same types of development. Their use of forms
seems to be fairly fossilised in terms of complexification, accuracy and number of times
morphemes are used.
5.4.2 Morpheme suppliance

Using measures of suppliance and accurate suppliance, Table 5.5 shows the values and rankings
of the top fifteen morphemes across the learners. The “average suppliance” value shows how
often each morpheme is used (correctly and incorrectly) in the data. The “average for accurate
use” value shows how often the morpheme is supplied correctly in the data. Rankings are given
for each of these values as an indication of which forms are used and acquired first. It should be
noted that these are average values and that there is some deviation from these sequences for
individual learners.
197
Table 5.5 Suppliance of top fifteen morphemes

Morpheme Suppliance average Ranking Average for accurate Ranking
use9
Noun prefix 92.5% 1 89% 1
i-ending on negative 89.3% 2 89.3% 2
Subject concord 87.6% 3 79.9% 4
Locative 87.5% 4 81.4% 3
Associative copulative 87.2% 5 79.2% 5
Negative subject concord 84.7% 6 69.4% 7
na-/nga- (with) 84.5% 7 77.3% 6
Possessive 80.2% 8 66.7% 10
Demonstrative 80.1% 9 68.4% 8
Qualificative 73.2% 10 55.3% 12
Copulative 70.8% 11 68% 9
Qualificative a- 68% 12 59.5% 11
Additive conjunction 58.9% 13 52% 13
-ya- 57.3% 14 19.2% 15
Object concord 52.9% 15 44.9% 14
Suppliance sequences and order of appearance sequences are slightly different, but this may be
partly because of the influence of Ann’s data on the order of appearance list (see 5.4.1).
5.4.3 Comparison with English sequences of development

Comparison with English is highly problematic, because in Xhosa many morphemes mark
functions which are encoded lexically in English. Comparing Xhosa morphemes, with their
strongly semantic functions, and English morphemes, which do not always have such a strong
semantic function, is not particularly productive. Indeed, even the functions of forms in the two
9
Some of these values are slightly different from those in the tables in chapter 4, because of the
inclusion of unnecessary use which is not always indicated in the tables in chapter 4.
198
sequences do not appear to match. A comparison with the English sequence of development is
made in Table 5.6.
Table 5.6 Comparison of English and Xhosa sequences of development
English morpheme in Function of English Comment on Xhosa form and order of
order of morpheme acquisition
appearance10
-ing marks aspect Xhosa aspect is marked by forms of the
subject concord, but aspectual marking is
not an early form in my data.
plural -s marks number Number marking occurs on the noun prefix
in Xhosa and these morphemes occur early,
as they do for English.
copula form of the verb The copulative is acquired late in my data.
auxiliary marks temporality, These functions are not marked by
tentativeness auxiliaries until fairly late in my data, and
(may/might/could), some non-temporal functions are not
ability (can/could), marked at all for some learners.
intention
(shall/should)
article marks definiteness The equivalent form, the object concord, is
or indefiniteness one of the latest acquired forms for learners
in my study.
irregular past marks temporality Past forms are learnt relatively late, as they
seem to be in the English sequence.
regular past marks temporality same as above
third-person singular purely grammatical There is no equivalent morpheme in Xhosa.
agreement function
possessive marks possession This form is marked fairly early for learners
in my study.
5.5 EXPLANATION OF THE SEQUENCE OF DEVELOPMENT IN MY STUDY

Several explanations of the sequence of development seem plausible, although a range of
explanations is needed to account for all occurrences and several may be applicable to some
morphemes.
10
This is the English order of acquisition reported by Krashen (cited in Zobl 1995).
199
5.5.1 Morphemes as foundations in utterances

The noun prefix seems to be the first form learnt, probably because of the central role it plays in
the selection of morphemes in the remainder of the utterance. The importance of this morpheme
as the foundation for further morphological choices may be the reason for its high level of
suppliance.
Practice seems to mean that forms can become automatised and retrieved more quickly (Robinson
1997a, b). Some morphemes (e.g. noun prefixes) occur very often and may therefore be
memorised fairly quickly. Verbal forms are also foundations of utterances and the most simple
associated morphology may therefore appear early.
5.5.2 Links to thematic roles

Robert’s (1989) view of the interlanguage (IL) system is that it consists of a computational system
and a conceptual system (discussed in section 2.5.3). Thematic relations are important in the
conceptual system. To some extent, this may be true of the IL produced by these learners, since
functional constraints seem to be important determiners of the use of morphemes.
Radford (1988:373) mentions eight thematic roles: theme/patient, agent/actor, experiencer,

benefactive, instrumental, locative, goal and source. An analysis of the thematic roles in my data
shows that the use of the agent, patient/theme and locative functions (the location, goal and
source functions are realised by the locative morpheme) is predominant. Since these three
thematic roles are primary, the required morphemes appear early. The benefactive and
instrumental functions occur occasionally, with the latter usually shown by nga-/na- (with) or left
to the context. The benefactive function is usually shown by the verb which has two arguments
(direct object and indirect object) or an applicative extension -ela, which only appears in the
utterances of more advanced learners. The experiencer role is not apparent in the data, perhaps
because of the elicitation tasks which have been used.
5.5.3 Agreement functions vs. Semantic functions

Most morphemes do not have only an agreement function, but encode semantic information and
perform core discourse functions as well. It is therefore necessary to use morphemes to convey
semantic and discourse functions even if the forms are incorrect.
200
If selection of a morpheme is based more on grammar than on semantics, the form may not be
used. For example, the verbal relative -yo has both a grammatical and a semantic function, but the
semantic function can also be encoded at the beginning of the stem, so the -yo form is not always
produced. The form -ya- has a grammatical function, and although it appears early, it is not
produced accurately by some learners.
5.5.4 Invariant forms

Some invariant forms (e.g. i-ending on negatives) seem to be among the earliest forms learnt. The
invariant forms seem to be cognitively easier to acquire and may therefore appear fairly early,
perhaps especially so for these learners who have to learn so many agreement forms. Furthermore,
formal instruction may be successful if rules are simple and consistent (de Graaff 1997; DeKeyser
1995; Robinson 1997a, b).
5.5.5 Intraphrasal forms

Suzman (1999:139), in a discussion of L1 Zulu acquisition, notes that “(a)s a general observation,
agreement markers with primarily pronominal function overgeneralize while agreement markers
within the noun phrase are accurate.” Here she is referring to morphemes such as the qualificative.
However, this is clearly not true of my data, since qualificative forms are acquired fairly late and
agreement forms are not always correct.
5.5.6 Extension of the basic utterance

Forms which extend the core message are acquired later in my data. The learners’ focus may be
on basic nouns, verbs and location forms at the beginning, because these forms are important for
the message which they convey. Once learners have some measure of control over these forms,
they may be willing to expand their utterances with qualificatives, demonstratives, etc.
5.5.7 Lexical equivalence to English

The locative encodes forms which would be prepositions and adverbs in English and the
associative copulative encodes the English lexical form “have”. In order to translate successfully
from English, these forms need to be learned early and it might explain their early appearance in
the sequence of development.
201
5.5.8 Embedded forms

Forms which expand the utterance to a complex or compound structure are learned later (Wode
1981). The qualificative a- and the additive conjunction fall into this category, and these forms
appear late in the sequence of development.
5.5.9 Free vs. Bound forms

The forms discussed in the sequence of development of morpheme suppliance (section 5.4.2) are
bound forms, but some forms like adverbs appear in the middle stages of the order of appearance
sequence (section 5.4.1). These forms are free lexical forms and this may explain their relatively
early development (Wode 1981).
5.6 AVOIDANCE PATTERNS IN THE DATA

Avoidance of morphemes is an important strategy to take into account in IL analysis (Kleinmann
1977; Schachter 1974). Gass & Selinker (1994:26) say that “one cannot interpret data only on
the basis of what is present, because we do not know if absence of forms means lack of
knowledge of forms.” Avoidance takes three forms in the data of my study:
C omissions of forms, with reliance on context for interpretation;
C substitution of other forms for a morpheme; and
C restructuring of utterances to avoid the production of a difficult morpheme.
Examples of these patterns include:

C avoidance of demonstratives which show distance from the speaker and replacement by
adverbs (phaya, apho “there”);
C avoidance of non-present tense forms. In these cases, learners rely on the context to
indicate the necessary tense of the utterance;
C overuse of some subject concord pronominal forms instead of other subject concord forms
in the sentence-building task;
C avoiding forms like njenga- (like), participials and conjunctions by producing different
utterances rather than joining parts of utterances to form more complex forms;
C avoiding object concords and the definiteness which their use would imply and leaving
interpretation of definiteness to the context;
C avoiding pronominal forms by repeating the noun phrase;
C overusing one form to express particular functions; e.g. overuse of adverbs rather than
tense indicators and locatives;
C choosing to produce only simple utterances using forms that the learner feels he/she can
control; and
C overgeneralising one agreement form.
Avoidance seems to result from an attempt to decrease processing loads by choosing lexical items
202
or single forms rather than agreement forms, reliance on the context to infer meaning, and a lack
of knowledge of forms.
5.7 PERFORMANCE ON DIFFERENT TASKS

A comparison of the performance of these students on different spoken and written tasks is
necessary to provide a more complete picture of their development and capabilities.
As mentioned in section 3.2.3.3.2, the use of class work as an indicator of development is

problematic because students may have access to charts which contain the correct forms of
morphemes. Incorrect forms may therefore occur less frequently if the learner has grasped the
function of the morpheme and knows where it should be used. A further limitation is that some
class tasks call for performance on a particular, specified morpheme so that their use may not be
a good indicator of knowledge of the function of the morpheme. DeKeyser (1997) also notes that
although learners may be able to produce forms in the classroom situation where the task is
narrowly defined and the permutations are limited, they may not be able to apply the forms to
other contexts.
Nevertheless, there is evidence in the test and examination scripts that learners can produce a
wider range of forms than they produce in the oral data. This may point to avoidance in some
contexts or to the ability to produce forms in only one context. The level of learned metalinguistic
knowledge which can be applied in the test situation should also not be discounted.
In the oral interviews, exercises which give the learner the greatest level of control over the
content and subject matter seem to be supplied most fluently and correctly (e.g. building sentences
from words). They are not listening to the interviewer, struggling to comprehend the message and
trying to find vocabulary which meets the demands of Thembi’s questions all at once. Instead,
they control the topic and they can select more routinised, familiar patterns.
Ann
In the first few interviews, most spoken utterances are about two to three words long and when
they are longer than this, problems appear in choosing and including the correct concordial
element. Ann’s written work is generally better than her oral data. Her tests show problems with
203
class 4, 5, 6, and 9 subject concords, negative concords (although these forms are more often
correct than in the oral data), and qualificative concords. She only uses the existential with
weather terms, which is similar to her oral data. Nevertheless, some forms are better in tests than
in oral data, and she uses ka- possessive forms and correct copulatives in tests. The positive effect
of learning for a test may explain the observation, but test effects are usually temporary and have
little effect on the IL system. Some forms therefore appear correctly in the written work or in tests
where they are incorrect in the oral data.
Ann’s lack of vocabulary interferes with her performance on the dictation tests, although her
dictation texts begin to resemble Xhosa texts by the end of the interview period. In the
grammaticality judgements, she notices coalescence problems and some agreement problems
(although they are usually wrongly corrected until interview 7).
Ben
Ben’s range of morphemes is slightly wider in his written work (e.g. he uses absolute pronouns
and demonstratives in all three positions), but many of the errors are similar to those in his spoken
language. In the written work, negative forms are correct, the hamba/ya (go/go towards)
confusion remains, marked classes have more errors and non-prototypical locatives are sometimes
incorrect. The qualificative a- also shows errors. Coalescence is better in the written work,
possibly because of the time available to consider the correct form.
Ben’s dictation tests resemble Xhosa texts from the first interview, although several forms are
omitted. Familiarity with the vocabulary of the tests seems to influence correctness of morpheme
forms, which may indicate that extra processing energy is needed to produce morphemes.
Agreement forms and coalescence are usually appropriately corrected in the grammaticality-
judgement tasks.
Claire
Claire’s written work reproduces many of the same errors which she makes in her spoken data,
but the range of morphemes she uses is smaller (probably because she is not usually required to
produce spontaneous writing on a scale larger than the sentence). Although her written work
appears to be slightly better than her spoken data in terms of frequency of errors, the types of
204
errors remain the same (e.g. overuse of -ya-, lack of agreement). She appears to have more
qualificative a- problems in her writing than in her oral data, which may mean that many of the
forms she uses in oral data are automatised. Other problematic morphemes are absolute pronouns,
possessives, and the associative copulative. Her written and spoken performances seem to be
quite similar, except that there is a much wider range of morphemes in her spoken data.
Claire seems to find the dictation tests difficult and often produces stems without agreement
morphemes. Her spoken ability appears to be superior to performance on this task. One reason
for this finding may be that she is unfamiliar with written forms of Xhosa and may not know how
to encode forms with which she is familiar. In the grammaticality-judgement task, coalescence is
usually corrected and her performance on correcting incorrect agreement forms improves over
the course of the interview period.
Pat
Pat’s range of morphemes in her written class work is wider than the range of forms she uses in
oral data. She adds to her written repertoire passives, some object concord pronominal forms,
more conjunctions, a few subjunctive forms and negative forms of a range of tenses. However,
she fails all her tests because she makes many similar errors to the ones in the oral data. As in her
oral data, she makes many concord errors, she overuses -ya-, and she uses the subject concord
+ nayo form for the associative copulative. Her test forms and her oral forms are therefore fairly
similar, while some written essay work appears to be better than her oral data (probably because
of the resources available to find agreement forms).
For Pat, the first two dictations are fairly poor with many forms left off. Subject concords cause
problems and there are interchangeable forms for classes 2 and 6, as in the oral data. The present
tense indicative form is preferred and clicks cause perception problems. From interview 3,
dictations are better, perhaps because the vocabulary is more familiar. Where interviews are
repeated, the dictation is markedly better in the repeat interview. In the grammaticality
judgements, Pat wrongly corrects many agreement forms, but coalescence problems are usually
corrected. What is interesting is that in repeated interviews, Pat makes almost exactly the same
corrections, which may point to stability in her underlying IL system.
205
Karen
In the interviews, Karen uses correct forms when she has a high level of control over the content
(e.g sentence-builiding tasks) and translation is particularly good. Grammaticality judgements are
usually excellent and the dictation tests are nearly perfect towards the end of the data-collection
period. The vocabulary is presumably very familiar for these tasks and the morphemes do not
seem to cause many problems. For the grammaticality-judgement tasks, the extra time available
to consider forms and the written nature of the input may help her to make correct judgements.
The written work which Karen presents in class and in examinations shows a higher level of
development than the forms she produces in the oral interviews. She uses a wider range of
morphemes and these are usually used correctly. Some of the errors which she makes in the oral
data hardly ever appear in her written work (e.g. incorrect agreement forms for subject concords,
possessives, etc.). She also uses a range of auxiliary morphemes, passives, past and future tense
forms, and negative forms more often than these appear in the oral data (although these may be
a result of the tasks set for the class and may not reflect natural use of the forms). She makes
errors with some tenses, and object concords are not always used where necessary. The time
available to her seems to influence her performance, as found in several studies of the effect of
planning time on fluency, accuracy and complexity (Crookes 1989; Foster & Skehan 1996; Ortega
1999).
Sarah
Sarah uses a slightly wider range of morphemes in her written class work and examinations than
in her oral data, adding the use of the potential mood and a wider range of conjunctions.
However, her written abilities seem to be quite close to her spoken abilities and she makes the
same kind of errors in both contexts. It is probably fair to say that her spoken IL is quite close to
her written IL and that the interviews are a fair reflection of her Xhosa ability. Many of the written
errors she makes are similar to those in her spoken data (e.g. incorrect choices of agreement
forms, using the terminative -e form on verbs in the remote past tense, and using the same tenses
as she uses in the oral data).
Sarah struggles with dictation tests when vocabulary is unfamiliar, but tests with fairly basic
vocabulary show a knowledge of the agreement system and correct use of most morphemes. Most
206
incorrect agreement forms are appropriately corrected in the grammaticality-judgement tests,

although grammatical forms like the subjunctive are not supplied where necessary.
5.8 CONCLUSION
This chapter has discussed common patterns of morpheme choice which are evident in the data
and has compared these patterns to those found in other studies. Many of the patterns and
principles are common to my study and other studies of L1 and L2 acquisition. This may point
to common language-learning principles which operate when learning a range of different
languages. However, it is also evident that there are acquisition patterns which can be attributed
to the complexities of the agglutinative structure and that these patterns are not found in
typologically different languages. These patterns create difficulties for comparison with other
studies and, in particular, the methods used to measure levels of L2 development.
In the concluding chapter, the findings of my study are compared to the findings relating to the
Basic Variety.
207
CHAPTER 6
COMPARISONS WITH THE BASIC VARIETY AND CONCLUSION
6.0 INTRODUCTION
In this final chapter, comparisons are made in selected areas between the findings of the European
Science Foundation (ESF) study (Klein & Perdue 1993, 1997; Perdue 1993b) and the findings of
my study. Suggestions for further research are also discussed in this chapter.
6.1 COMPARISONS WITH THE ESF STUDY

Since one of the main findings of the ESF study is the existence of a learner variety called the
Basic Variety (BV), this comparison will focus on the BV and whether selected characteristics
of the BV are evident in my data. Since my study has focussed mainly on the forms which appear
in the interlanguage (IL) of learners of Xhosa and the morphological development evident in the
data, the aspects of the BV which relate to these two areas will be the focus of this section.
6.1.1 Why might there be points of similarity?

Before comparing the findings, it is necessary to ask why there might be similarities between the
findings of the two studies. The main reason is that, on the basis of the literature reviewed in
chapter 2 and the common learning principles found in my data and other studies, it seems
reasonable to hypothesise that there might be similarities between the BV and the learner language
produced in my study.
The second point is that the ESF study is a cross-linguistic study which found points of similarity
between the ILs produced by learners of a range of languages. On the basis of these similarities,
it might be expected that there would be points of comparison with the IL found in my study.
A third point is that formal instruction seems to have a limited effect on the IL (see section 2.7)
and, as a result, the BV and the IL of the learners of my study may be comparable. Klein &
Perdue (1997) hesitate to generalise the BV beyond naturalistic contexts of learning, although
they also say that the BV is observed outside the classroom for learners who learn L2s in formal
contexts.
208
Finally, the learners in my study are fairly early learners if one looks at the amount of input they
have received. Learners who have had only formal input have spent roughly 100 - 130 hours in
formal classes per year1. If one compared the learners in the ESF study (who are immersed in the
language in an immigrant situation) with the learners of my study, one would probably find that
the ESF learners are exposed to more of the target language and use it much more than the
learners in my study.
6.1.2 Why might there be differences between the BV and the learner language in my study?
Klein & Perdue (1997) may be correct in their hesitancy to generalise the BV beyond naturalistic
contexts of learning, especially if learners have been exposed only to formal input, where the focus
is on morphology. Most learners in my study have had very little exposure to Xhosa in naturalistic
contexts and much of their input is written. Written input allows more time to focus on
morphology, since one can move beyond looking only for the meaning of stems to find the main
message (as learners in my study say they do when listening to oral input). Studies discussed in
section 2.7 also point to the earlier acquisition of forms when learners are exposed to input in
formal contexts, which means that learners may move beyond the BV fairly quickly. Nevertheless,
this does not mean that the learners do not use a BV at any stage in their development, although
they may use it for only a short period of time.
The typological differences between the languages learned in the ESF study and the language
learnt in my study may lead to differences in findings (see also Giacalone Ramat, 1992, who found
that learners of Italian learnt morphology fairly early because of the importance of morphology
in that language). The agglutinative structure of Xhosa may override some of the constraints on
learning morphology found in languages where morphology does not play such an essential
semantic role. In Xhosa, the learner cannot ignore morphology because it is too semantically
important. Indeed, many morphemes in Xhosa are the equivalent of lexical items in isolating or
analytic languages. In order to produce equivalent messages, one would need to learn the
morphology.
1
However, it should be noted that most of the class time is spent talking in English because
explanations are provided almost exclusively in English. Learners may also spend some time doing
homework and learning for tests and examinations, but, as mentioned in 3.2.1.1, not a lot of time is
devoted to Xhosa outside the classroom by most learners in my study.
209
Another reason why the ESF study and my study may not be comparable is that the learners’
language in my study may be too far developed; i.e. they may not be at sufficiently early levels of
development. Development can be related to amount of exposure, but, as noted above, the
amount of input to which they have been exposed is fairly low and some of the input has been
sporadic or occurred a long time ago. Development can also be measured according to the type
of language produced (e.g. use or omission of morphology, subordination, etc.). In this regard,
learners in my study may have moved beyond the BV because of the input from formal contexts
to which they have been exposed.
A final reason why the two studies may not be comparable is that many of the forms produced in
my study may be routinised forms which include both stems and morphemes (e.g. noun stems may
be routinised as noun prefix + noun stem). In the mind of the learner, some of the morphology
may not in fact be additional morphology, but may be integrally related to the stem so that the two
are indistinguishable. The overgeneralisation of some forms may also support this point.
There are, therefore, reasons for expecting similarities and reasons for expecting differences
between the findings of the two studies. In the next section, a comparison between the BV and
the IL of learners in my study is made.
6.1.3 Comparison of Basic Variety forms and forms in the IL of learners in my study
6.1.3.1 Linguistic elements in the BV and my study
Table 6.1 is a comparison of the linguistic elements found in the BV (Carroll & Dietrich 1985;
Klein & Perdue 1997; Perdue 1993b) and those found in my study.
210
Table 6.1 Comparison of linguistic elements in the BV and my study

Linguistic form In the BV? In my study? Comment
Proper names, noun-like o o
words, verb-like words
Adverbial-like words o o
(temporal and spatial)
Numerals o o
Personal pronouns o o
Anaphoric pronouns x o the form for “it” is problematic, but
for inanimate objects it is supplied
Word for negation o (o) some use of hayi (no), but usually
by means of negative morphemes
Determiners which are o o demonstratives have spatial
demonstratives rather reference function, object concord
than articles system not fully developed
Formulaic utterances o o
Copula x (o) copulative morpheme is used, but
function is not always known
Prepositions x o some free forms, mainly realised by
the common locative morpheme
Invariant morphological o (o) some invariant forms, most have a
forms restricted range rather than
invariant form
Morphemes with purely x (o) only one (-ya-)
grammatical functions
Finite morphology x (o) tense is marked, but aspect is not
marking tense and aspect usually marked
Agreement in number x o
Agreement in gender x o central to the Xhosa class system
and it appears early2
Agreement by x o central to the Xhosa agglutinative
morphology structure and it appears early
2
The masculine, feminine, (neuter) gender system of many Indo-European languages is replaced
by the class system in Xhosa.
211
A comparison of the linguistic elements in the BV and the IL of the learners in my study shows
that they appear to have developed beyond the BV. Their IL contains the elements of the BV, but
it also contains elements which show that it is probably more advanced than the Basic Variety
level (e.g. in terms of morphology, anaphoric pronouns for inanimate objects, prepositions,
copula). Some morphemes, such as tense, are supplied variably or realised by a restricted range
of forms by the learners in my study. Nevertheless, morphology is used in their IL.
Should one conclude then that the learners in my study have moved beyond the BV level? One
argument in favour of this conclusion is that they produce forms which Klein & Perdue (1997)
say show that learners have moved beyond Basic Variety level of development (e.g. morphology,
prepositions). However, a closer analysis of these forms shows that their presence can be closely
linked to the language typology of Xhosa, where morphology is the usual means of indicating
negation, many prepositions, tense, anaphoric pronouns for inanimate objects, copula forms,
agreement, number and gender. In fact, the present tense, anaphoric pronouns for inanimate
objects, agreement, number and gender can all be indicated by the use of one type of morpheme:
the subject concord. This means that as soon as learners are able to manipulate the subject
concord, they are able to perform this wide range of functions. I would therefore hesitate to claim
that the presence of morphology indicating these forms means that learners have moved beyond
the Basic Variety level. Instead, I would argue that because morphology is so important in Xhosa,
learners would need to produce a far wider range of forms in order to be said to be approaching
target-like usage (as Klein & Perdue, 1997, claim learners are doing when they move beyond the
BV). It might, in fact, be more useful to posit a slightly different type of BV for agglutinative
languages.
6.1.3.2 Phrasal constraints

Klein & Perdue (1997) find that the usual word order is Agent-Verb-Patient in the ESF study.
They also mention that this relates to the SVO pattern of fully-fledged languages, but they hesitate
to give elements of the BV status as subject, verb or object. This is because subjects and objects
may be noun-like forms rather than nouns, and verbs may not be finite forms. Furthermore, they
find that subjects and objects are not the primary means of organising the BV. Instead, it is
organised by means of semantic principles with elements in topic and focus position.
212
In the IL of the learners in my study, the SVO pattern is used and the Agent-Verb-Patient pattern
is also common. Nevertheless, other thematic roles are also assigned and, in particular, the
location role appears early. The nouns which occupy subject and object positions may also be
expanded by the use of descriptive elements such as qualificatives. English and Xhosa both have
the SVO structure, so positive transfer can occur.
According to Bierwisch (1997), the basic phrasal pattern found in the BV is:
(NP)-V-(XP-(XP)) where X = Noun, Adjective or Preposition, depending on the
subcategorization of V.
Adverbial forms may also occur before or after the basic pattern.
This pattern is found in the IL of learners of my study, but they also expand this pattern because
some of them use conjunctions and subordinating forms. Subordination is not found in the BV
(Perdue 1993b) and, as a result, it appears that the learners in my study are indeed more advanced
than the BV level (although their subordination patterns do not reach target-like levels of use).
6.1.3.3 Levels of development

Klein & Perdue’s (1993:25) levels of development are “from nominal to infinite to finite utterance
organisation”. At the nominal level, learners produce mainly noun-like forms and there is no
functional morphology, case role assignment or argument structure. At the infinite level, verb-like
forms appear but there is no “distinction between the finite and non-finite component of the verb”
(Klein & Perdue 1993:25). The finite level includes verbs which make the finite/infinite distinction.
If one looks only at the markers of finiteness in Xhosa, one has to conclude that the learners in
my study are at the finite level of development (although there is some backsliding when they omit
these forms). From the “summary of subject concord data” (Table 4.10 in section 4.4.2), it is clear
that omission of the subject concord (and therefore finiteness markers) is relatively rare and all
six learners reach the 90% suppliance criterion for this morpheme. In fact, this is one of the
earliest acquired morphemes, rather than a late form as the BV predicts.
Furthermore, because of the range of morphology which appears in Xhosa, I find the use of a
measure which looks mainly at verbal morphology problematic. As indicated in section 6.1.3.1,
213
one set of verbal morphology would suffice to move the learner beyond the infinite level if only
verbal morphology were taken into account in the measurement. Instead of this narrow view, I
believe that one needs to consider morphology which occurs on noun stems, qualificative stems,
etc. Without this morphology, the learner is not approaching target-like use in agglutinative
languages, as the finite level of development suggests in non-agglutinative languages.
Based on the findings of my study, I would suggest that the following levels of development might
prove useful when measuring development of Xhosa interlanguage:
Stage 1: use of nouns, verbs and locatives;
Stage 2: expansion of the noun phrase and verb phrase to include qualificatives,
demonstratives, possessives, etc.;
Stage 3: complexification of verbal morphology to increase the range of tenses, aspect,
and verbal extensions, as well as the development of subordination;
Stage 4: expansion of the functions of the forms which appear in earlier stages, and
development of target-like discourse norms relating to use of morphology.
I suggest that morphology appears from stage 1, although its use may be variable. The number
and types of different morphemes which appear in the IL is important when measuring
development in a language which relies extensively on morphology to encode meaning. Increases
in the use of morphology go hand-in-hand with the development of the range of functions which
can be expressed in the IL. Increasingly target-like use of morphemes can be measured by an
increase in accuracy and an increase in the target-like functions of morphemes, as shown in stage
4.
6.1.4 Conclusion
It appears from the above points that the findings of the two studies are different in several ways.
The learners of my study use the forms found in the BV, but they also use forms which do not
appear in the BV. However, concluding that this means that learners in my study have moved
beyond the BV is problematic. Instead, it appears that the BV may not be an adequate description
of the level of development of early learners of Xhosa. The use of morphemes appears to be too
widespread in Xhosa for one to rely mainly on a measure of use of verbal morphology to indicate
that learners are approaching target-like use. Furthermore, the effect of the formal context on the
214
IL of the learners of my study is unclear. Exposure to the formal context probably hastened the
acquisition of morphology and it may have meant that these learners proceeded very quickly
through the early stages of development.
6.2 SUGGESTIONS FOR FURTHER RESEARCH

As a result of the conclusions reached above, there are a number of areas which need further
research in the field of acquisition of forms in languages which are typologically different from
inflectional and analytic languages.
Firstly, more studies of the second-language acquisition of morphological forms in other African
languages are needed in order to compare findings with those of my study. The order of
acquisition may be strongly related to the order of teaching in my study, although studies of the
influence of formal input suggest that order of teaching may have little influence on acquisition.
Although the order of acquisition in my study corresponds roughly to the order of acquisition in
L1 acquisition studies of other typologically similar languages, the differences in cognitive
development of learners make comparisons problematic.
There is also a need for studies of learners who are exposed only to naturalistic input from the
very beginning. Although my study found that the BV could not be applied successfully to the IL
produced by learners in my study, it may be more easily applied to learners in naturalistic
situations (as Klein & Perdue, 1997, suggest). This kind of study would also indicate the effect
of formal instruction on morphological development.
Further studies in the area of L2 acquisition of Xhosa (and other typologically related languages)
would shed more light on appropriate measures of levels of development. The number and/or
types of different morphemes which need to be learned in order to claim that a learner has reached
particular milestones needs to be determined by examining the ILs of many more learners.
Further research into the teaching of Xhosa as a second language needs to be conducted. For
example, the effect of using a syllabus based on the order of appearance of morphemes in the IL
in this study could be tested. It appears, for example, that the locative should be taught quite early
as learners seem to need to perform the location function early. In addition, there seems to be a
215
need to introduce a more communicative way of teaching so that learners practise oral skills and
actually use their interlanguage to perform useful communicative functions (see also Duvené de
Wit & Ntuli 1994).
6.4 CONCLUSION
This study investigated morphological development in the IL of English learners of an
agglutinative language. In addition, it considered some of the commonly investigated functions
which appear in the interlanguage of these learners. It showed that the learners are able to express
temporality, spatial and pronominal reference. They can use morphemes (and other means) to
convey the functions necessary to communicate the messages required of them in the interviews.
One of the aims of the study was to compare the findings of my study with the findings of other
IL studies and, in particular, the findings of the European Science Foundation study. Comparisons
have shown that there are both similarities and differences in the findings of my study compared
with these other studies. Firstly, there were many similarities in terms of learning principles which
the learners use to make sense of the input and produce language in the interview situation.
Similarities were also found between the findings of L1 acquisition studies of African languages
and my study with regard to learning principles and some parts of the sequence of development
of morphemes. It proved impossible to compare the sequences of development found in English
and Xhosa, primarily because of the differences in the way in which forms are realised in the two
languages.
The most marked differences between the findings of my study and other studies appeared to be
the extensive use of morphology by the learners in my study. This is not predicted by other studies
of IL which conclude that morphology is acquired late and only after extensive exposure to the
target languages. Typological differences therefore appear to be very important when drawing
conclusions about which forms will appear in the early IL of learners (as Slobin, 1992, also
concludes for L1 acquisition). The measures of development which are used to assess learners of
Indo-European languages therefore appear to be inadequate measures of the level of development
of learners of an agglutinative language such as Xhosa.
216
In conclusion, it appears that similar language-learning and production principles are employed
by learners of typologically different languages. However, there are differences in which forms
appear early in the ILs of learners of non-analytic and non-inflectional languages. Where
morphemes are important, as in an agglutinative language, learners of these languages pay
attention to the need for morphology and learn it early.
217
REFERENCES
Adamson, H. D. (1989). Does the bioprogram affect second language acquisition? In M. Eisenstein
(Ed.), The dynamic interlanguage: Empirical studies in second language variation (pp. 33-47).
New York: Plenum Press.
Adjémian, C. (1976). On the nature of interlanguage systems. Language Learning, 26(2), 297-320.
Alexander, R. (1979). Elements of a theory of second language learning. Frankfurt: Verlag Peter
Long.
Allwright, R. L. (1984). The analysis of discourse in interlanguage studies: The pedagogical evidence.
In A. Davies, C. Criper, & A. P. R. Howatt (Eds.), Interlanguage (pp. 204-219). Edinburgh:
Edinburgh University Press.
Altman, R. (1997). Oral production of vocabulary. In J. Coady & T. Huckin (Eds.), Second language
vocabulary acquisition (pp. 67-97). Cambridge: Cambridge University Press.
Andersen, R. W. (1977). The impoverished state of cross-sectional morpheme acquisition/accuracy

methodology. Working Papers on Bilingualism, 14, 49-82.
Andersen, R. W. (1978). An implicational model for second language research. Language Learning,
28(2), 221-282.
Andersen, R. W. (1983). Pidginization and creolization as language acquisition. Rowley: Newbury

House.
Andersen, R. W. (1984). The One to One Principle of interlanguage construction. Language

Learning, 34(4), 77-95.
Andersen, R. W. (1991). Developmental sequences: The emergence of aspect marking in second

language acquisition. In T. Huebner & C. A. Ferguson (Eds.), Crosscurrents in second
language acquisition and linguistic theories (pp. 305-324). Amsterdam: John Benjamins.
Antinucci, F., & Miller, R. (1976). How children talk about what happened. Journal of Child
Language, 3(2), 167-189.
Bahns, J., Burmeister, H., & Vogel, T. (1986). The pragmatics of formulas in L2 learner speech: Use
and development. Journal of Pragmatics, 10, 693-723.
Bailey, K., Madden, C., & Krashen, S. (1974). Is there a “natural sequence” in adult second language
learning? Language Learning, 24(2), 235-243.
218
Bailey, N. (1989). Discourse conditioned tense variation: Teacher implications. In M. Eisenstein

(Ed.), The dynamic interlanguage: Empirical studies in second language variation (pp. 279-
295). New York: Plenum Press.
Baker, C. (1979). Syntactic theory and the projection problem. Linguistic Inquiry, 10(4), 533-581.
Bardovi-Harlig, K. (1994). Anecdote or evidence? Evaluating support for the hypotheses concerning
the development of tense and aspect. In E. E. Tarone, S. M. Gass, & A. D. Cohen (Eds.),
Research methodology in second-language acquisition (pp. 41-60). Hillsdale: Lawrence
Erlbaum Associates.
Bardovi-Harlig, K., & Bofman, T. (1989). Attainment of syntactic and morphological accuracy by
advanced language learners. Studies in Second Language Acquisition, 11(1), 17-34.
Beck, M., Schwartz, B. D., & Eubank, L. (1995). Data, evidence and rules. In L. Eubank, L.
Selinker, & M. Sharwood Smith (Eds.), The current state of interlanguage: Studies in honor
of William E Rutherford (pp. 177-195). Amsterdam: John Benjamins.
Bell, A. (1984). Language style as audience design. Language in Society, 13(2), 145-204.
Bhardwaj, M., Dietrich, R., & Noyau, C. (1988). Temporality. Strasbourg: European Science
Foundation.
Bialystok, E. (1990). Communication strategies: A psychological analysis of second language use.

Oxford: Basil Blackwell.
Bialystok, E. (1998). Coming of age in applied linguistics. Language Learning, 48(4), 497-518.
Bialystok, E., & Frohlich, M. (1980). Oral communication strategies for lexical difficulties.
Interlanguage Studies Bulletin, 5(1), 3-30.
Bialystok, E., & Sharwood Smith, M. (1985). Interlanguage is not a state of mind: An evaluation of
the construct for second language acquisition. Applied Linguistics, 6(2), 101-116.
Bickerton, D. (1982). The roots of language. Ann Arbor: Karoma.
Bickerton, D. (1983). Comments on Valdman’s ‘Creolization and second language acquisition’. In

R. Andersen (Ed.), Pidginization and creolization as language acquisition (pp. 235-240).
Rowley: Newbury House.
Bickerton, D. (1990). Language and species. Chicago: Chicago University Press.
Bierwisch, M. (1997). Universal Grammar and the Basic Variety. Second Language Research, 13(4),
348-366.
219
Birdsong, D. (1989). Metalinguistic performance and interlinguistic competence. New York:

Springer.
Bley-Vroman, R. (1983). The comparative fallacy in interlanguage studies: The case of systematicity.
Language Learning, 33(1), 1-17.
Bokamba, E. G. (1977). The impact of multilingualism on language structures: The case of central
Africa. Anthropological Linguistics, 19(5), 181-202.
Broeder, P. (1991). Talking about people: A multiple case study on adult language acquisition.
Amsterdam: Swets & Zeitlinger.
Broeder, P., Extra, G., & van Hout, R. (1993a). Richness and variety in the developing lexicon. In
C. Perdue (Ed.), Adult language acquisition: Cross-linguistic perspectives: Vol. 1. Field
methods (pp. 145-163). Cambridge: Cambridge University Press.
Broeder, P., Extra, G., van Hout, R., & Voionomaa, K. (1993b). Word formation processes in
talking about entities. In C. Perdue (Ed.), Adult language acquisition: Cross-linguistic
perspectives: Vol. 2. The results (pp. 41-72). Cambridge: Cambridge University Press.
Brown, H. D. (1998). Some further perspectives on the development of applied linguistics: A

response to Bialystok. Language Learning, 48(4), 519-526.
Bybee, J. (1985). Morphology. Amsterdam: John Benjamins
Bybee, J. L. (1991). Natural morphology: The organization of paradigms and language acquisition.
In T. Huebner & C. A. Ferguson (Eds.), Crosscurrents in second language acquisition and
linguistic theories (pp. 67-91). Amsterdam: John Benjamins.
Carroll, M. (1984). Cyclical learning processes in second language production. Frankfurt: Peter Lang.
Carroll, M. (1990). Word order in instructions in learner languages of English and German.
Linguistics, 28(5), 1011-1037.
Carroll, M., & Becker, A. (1993). Reference to space in learner varieties. In C. Perdue (Ed.), Adult
language acquisition: Cross-linguistic perspectives: Vol. 2. The results (pp. 119-149).
Cambridge: Cambridge University Press.
Carroll, M., & Dietrich, R. (1985). Observations on object reference in learner languages.
Linguistiche Berichte, 98, 301-337.
Carroll, S. E. (1999). Input and SLA: Adults’ sensitivity to different sorts of cues to French gender.
220
Cazden, C. (1968). The acquisition of noun and verb inflections. Child Development, 39(2), 433-448.
Cazden, C., Cancino, E., Rosansky, E., & Schumann, J. (1975). Second language acquisition in
children, adolescents and adults. Final report. Washington, DC: National Institute of
Education.
Chaudron, C. (1983). Research on metalinguistic judgments: A review of theory, methods, and

results. Language Learning, 33(3), 343-377.
Clahsen, H. (1987). Connecting theories of language processing and (second) language acquisition.
In C. Pfaff (Ed.), First and second language acquisition processes (pp. 103-116). Cambridge:
Newbury House.
Clahsen, H. (1995). German plurals in adult second language development: Evidence for a dual-
mechanism model of inflection. In L. Eubank, L. Selinker, & M. Sharwood Smith (Eds.), The
current state of interlanguage: Studies in honor of William E Rutherford (pp. 123-137).
Amsterdam: John Benjamins.
Clark, E. (1985). The acquisition of Romance with special reference to French. In D. Slobin (Ed.),
The crosslinguistic study of language acquisition Vol. 1. (pp. 687-782). Hillsdale: Lawrence
Erlbaum Associates.
Comrie, B. (1984). Why linguists need language acquirers. In W. Rutherford (Ed.), Language
universals and second language acquisition (pp. 11-29). Amsterdam: John Benjamins.
Comrie, B. (1997). On the origin of the Basic Variety. Second Language Research, 13(4), 367-373.
Connors, K. (1988). Morphological markedness in second language acquisition. International Review

of Applied Linguistics, 26(2), 101-113.
Cook, V. J. (1991). Second language learning and language teaching. London: Edward Arnold.
Cook, V. J. (1993). Linguistics and second language acquisition. Basingstoke: Macmillan.
Cook, V. J. (1996). Universal Grammar in second language acquisition. In P. Jordens & J. Lalleman
(Eds.), Investigating second language acquisition (pp. 167-185). Berlin: Mouton de Gruyter.
Cooreman, A., & Kilborn, K. (1991). Functionalist Linguistics: Discourse structure and language
processing in second language acquisition. In T. Huebner & C. A. Ferguson (Eds.),
Crosscurrents in second language acquisition and linguistic theories (pp. 195-224).
Corder, S. P. (1967). The significance of learners’ errors. International Review of Applied

Linguistics, 5(4), 161-170.
221
Corder, S. P. (1971). Idiosyncratic dialects and error analysis. International Review of Applied
Corder, S. P. (1975). Error analysis, interlanguage and second language acquisition. Language
Teaching and Linguistics: Abstracts, 8, 201-218.
Corder, S. P. (1977). ‘Simple codes’ and the source of the second language learner’s initial heuristic
hypothesis. Studies in Second Language Acquisition, 1(1), 1-10.
Corder, S. P. (1981). Error analysis and interlanguage. Oxford: Oxford University Press.
Corder, S. P. (1992). A role for the mother tongue. In S. M. Gass & L. Selinker (Eds.), Language
transfer in language learning (rev. ed.), (pp. 18-31). Amsterdam: John Benjamins.
Crookes, G. (1989). Planning and interlanguage variation. Studies in Second Language Acquisition,
11(4), 367-383.
Crookes, G. (1990). The utterance, and other basic units for second language discourse analysis.
Applied Linguistics, 11(2), 183-199.
Crystal, D. (1997). The Cambridge encyclopedia of language. Cambridge: Cambridge University

Press.
De Beaugrande, R., & Dressler, W. (1981). Introduction to text linguistics. London: Longman.
De Graaff, R. (1997). The Experanto experiment: Effects of explicit instruction on second language
acquisition. Studies in Second Language Acquisition, 19(2), 249-276.
DeKeyser, R. M. (1995). Learning L2 grammar rules: An experiment with a miniature linguistic

system. Studies in Second Language Acquisition, 17(3), 379-410.
DeKeyser, R. M. (1997). Beyond explicit rule learning: Automatizing second language

morphosyntax. Studies in Second Language Acquisition, 19(2), 195-221.
DeKeyser, R., & Sokalski, K. (1996). The differential role of comprehension and production practice.
Dekydtspotter, L., Sprouse, R. A., & Thyne, R. (1998). Evidence for full UG access in L2 acquisition
from the interpretive interface: Quantification at a distance in English-French interlanguage.
In A. Greenhill, M. Hughes, H. Littlefield, & H. Walsh (Eds.), Proceedings of the 22nd annual
Boston University Conference on language development Vol. 1 (pp. 141-152). Somerville:
Cascadilla Press.
222
Demuth, K. (1992). The acquisition of seSotho. In D. I. Slobin (Ed.), The crosslinguistic study of
language acquisition Vol. 3 (pp. 557-638). Hillsdale: Lawrence Erlbaum Associates.
Dickerson, L. J. (1975). The learner’s interlanguage as a system of variable rules. TESOL Quarterly,
9(4), 401-407.
Dietrich, R. (1989a). Nouns and verbs in the learner’s lexicon. In H. Dechert (Ed.), Current trends
in European second language acquisition research (pp. 13-22). Clevedon: Multilingual
Matters.
Dietrich, R. (1989b). Communicating with few words. An empirical account of the second language
speaker’s lexicon. In R. Dietrich & C. Graumann (Eds.), Language processing in social
context (pp. 233-276). Amsterdam: North-Holland.
Dietrich, R., Klein, W., & Noyau, C. (1995). The acquisition of temporality in a second language.
Dittmar, N. (1981). On the verbal organization of L2 tense marking in an elicited translation task by
Spanish immigrants in Germany. Studies in Second Language Acquisition, 3(2), 136-164.
Dittmar, N. (1984). Semantic features of pidginised learners of German. In R. Andersen (Ed.),

Second languages: A cross-linguistic perspective (pp. 243-270). Rowley: Newbury House.
Doughty, C. (1991). Second language instruction does make a difference: Evidence from an empirical
study of SL relativization. Studies in Second Language Acquisition, 13(4), 431-469.
Douglas, D., & Selinker, L. (1994). Research methodology in context-based second-language

research. In E. E. Tarone, S. M. Gass, & A. D. Cohen (Eds.), Research methodology in
second-language acquisition (pp. 119-131). Hillsdale: Lawrence Erlbaum Associates.
Duffield, N., Montrul, S., Bruhn de Garavito, J., & White, L. (1998). Determining L2 knowledge of
Spanish clitics on-line and off-line. In A. Greenhill, M. Hughes, H. Littlefield, & H. Walsh
(Eds.), Proceedings of the 22nd annual Boston University Conference on language
development Vol. 1 (pp. 177-188). Somerville: Cascadilla Press.
Dulay, H. C., & Burt, M. K. (1973). Should we teach children syntax? Language Learning, 23(2),
245-257.
Dulay, H. C., & Burt, M. K. (1974a). Natural sequences in child second language acquisition.
Dulay, H. C., & Burt, M. K. (1974b). A new perspective on the creative construction process in child
second language acquisition. Language Learning, 24(2), 253-278.
223
Du Plessis, J. A. (1978). isiXhosa 4. Goodwood: Oudiovista.
Duyvené de Wit, H. E., & Ntuli, R. M. (1994). Die bevordering van natuurlike diskoers en
kommunikatiewe kompetensie in die Afrikatale. South African Journal of African Languages,
14(1), 1-14.
Eckman, F. R. (1994). The competence-performance issue in second-language acquisition theory: A

debate. In E. E. Tarone, S. M. Gass, & A. D. Cohen (Eds.), Research methodology in
second-language acquisition (pp. 3-15). Hillsdale: Lawrence Erlbaum Associates.
Eckman, F. R., Bell, L., & Nelson, D. (1988). On the generalization of relative clause instruction in
the acquisition of English as a second language. Applied Linguistics, 9(1), 1-20.
Edge, J. (1989). Mistakes and corrections. London: Longman.
Ellis, N. C. (1996). Sequencing in SLA: Phonological memory, chunking, and points of order. Studies
in Second Language Acquisition, 18(1), 91-126.
Ellis, N. C., & Schmidt, R. (1997). Morphology and longer distance dependencies: Laboratory
research illuminating the A in SLA. Studies in Second Language Acquisition, 19(2), 145-171.
Ellis, R. (1984). Classroom second language development. Oxford: Pergamon.
Ellis, R. (1985a). Sources of variability in interlanguage. Applied Linguistics, 6(2), 118-131.
Ellis, R. (1985b). Understanding second language acquisition. Oxford: Oxford University Press.
Ellis, R. (Ed.). (1987a). Second language acquisition in context. London: Prentice-Hall.
Ellis, R. (1987b). Contextual variability in second language acquisition and the relevancy of language
teaching. In R. Ellis (Ed.), Second language acquisition in context (pp. 179-194). London:
Prentice-Hall.
Ellis, R. (1987c). Variability in interlanguage systems. In R. Ellis (Ed.), Second language acquisition
in context (pp. 67-71). London: Prentice-Hall.
Ellis, R. (1989). Are classroom and naturalistic acquisition the same? A study of the acquisition of
German word order rules. Studies in Second Language Acquisition, 11(3), 305-328.
Ellis, R. (1990). Instructed second language acquisition. Oxford: Basil Blackwell.
Ellis, R. (1992). Learning to communicate in the classroom: A study of two language learners’
requests. Studies in Second Language Acquisition, 14(1), 1-23.
224
Ellis, R. (1993). The structural syllabus and second language acquisition. TESOL Quarterly, 27(1),
91-113.
Ellis, R. (1994). The study of second language acquisition. Oxford: Oxford University Press.
Ellis, R., & Roberts, C. (1987). Two approaches for investigating second language acquisition in
context. In R. Ellis (Ed.), Second language acquisition in context (pp. 3-29). London:
Prentice-Hall.
Extra, G., & van Hout, R. (1996). Second language acquisition by adult immigrants: A multiple case
study of Turkish and Moroccan learners of Dutch. In P. Jordens & J. Lalleman (Eds.),
Investigating second language acquisition (pp. 87-113). Berlin: Mouton de Gruyter.
Faerch, C., & Kasper, G. (1983). Plans and strategies in foreign language communication. In C.
Faerch & G. Kasper (Eds.), Strategies in interlanguage communication (pp. 21-57). London:
Longman.
Faerch, C., Haastrup, K., & Phillipson, R. (1984). Learner language and language learning. London:
Multilingual Matters.
Fathman, A. (1978). ESL and EFL learning: Similar or dissimilar? In C. Blatchford & J. Schachter
(Eds.), On TESOL ‘78: Policies, programs and practices (pp. 213-223). Washington, DC:
TESOL.
Felix, S. (1978). Some differences between first and second language acquisition. In N. Waterson &
C. Snow (Eds.), The development of communication (pp. 469-479). New York: Wiley.
Felix, S. W. (1980). Second language development. Trends and issues. Tubingen: Gunter Narr
Verlag.
Felix, S. W. (1981). The effect of formal instruction on second language acquisition. Language
Learning, 31(1), 87-112.
Felix, S. W. (1984). Maturational aspects of Universal Grammar. In A. Davies, C. Criper, & A. P.

R. Howatt (Eds.), Interlanguage (pp. 133-161). Edinburgh: Edinburgh University Press.
Felix, S. W. (1995). Universal Grammar in L2 acquisition: Some thoughts on Schachter’s

Incompleteness Hypothesis. In L. Eubank, L. Selinker, & M. Sharwood Smith (Eds.), The
current state of interlanguage: Studies in honor of William E Rutherford (pp. 139-151).
Felix, S., & Hahn, A. (1985). Natural process in classroom second-language learning. Applied
225
Flynn, S. (1991). Government-Binding: Parameter-setting in second language acquisition. In T.

Huebner & C. A. Ferguson (Eds.), Crosscurrents in second language acquisition and linguistic
theories (pp. 143-167). Amsterdam: John Benjamins.
Foster, P., & Skehan, P. (1996). The influence of planning and task-type on second language
performance. Studies in Second Language Acquisition, 18(3), 299-323.
Fries, C. C. (1945). Teaching and learning English as a foreign language. Michigan: University of
Michigan Press.
Gass, S. M. (1979). Language transfer and universal grammatical relations. Language Learning,
29(2), 327-344.
Gass, S. M. (1984a). A review of interlanguage syntax: Language transfer and language universals.
Gass, S. M. (1984b). The empirical basis for the Universal Hypothesis in interlanguage studies. In A.
Davies, C. Criper, & A. P. R. Howatt (Eds.), Interlanguage (pp. 3-24). Edinburgh: Edinburgh
University Press.
Gass, S. M. (1994). The reliability of second-language grammaticality judgments. In E. E. Tarone,

S. M. Gass, & A. D. Cohen (Eds.), Research methodology in second-language acquisition
(pp. 303-322). Hillsdale: Lawrence Erlbaum Associates.
Gass, S. M. (1995). Universals, SLA, and language pedagogy: 1984 revisited. In L. Eubank, L.
Selinker, & M. Sharwood Smith (Eds.), The current state of interlanguage: Studies in honor
of William E Rutherford (pp. 31-42). Amsterdam: John Benjamins.
Gass, S. M., & Selinker, L. (Eds.). (1992). Language transfer in language learning (rev. ed.).
Gass, S. M., & Selinker, L. (1994). Second language acquisition: An introductory course. Hillsdale:
Lawrence Erlbaum Associates.
George, H. (1972). Common errors in language learning. Rowley: Newbury House.
Ghrib, E. M. (1987). A longitudinal study of a Tunisian adolescent’s learning of English morphemes

in a classroom context. International Review of Applied Linguistics, 25(1), 41-54.
Giacalone Ramat, A. (1992). Grammaticalization processes in the area of temporal and modal
relations. Studies in Second Language Acquisition, 14(3), 297-322.
Giacobbe, J. (1992). A cognitive view of the role of L1 in the L2 acquisition process. Second
Language Research, 8(3), 232-250.
226
Givón, T. (1979). On understanding grammar. New York: Academic Press.
Givón, T. (1984). Universals of discourse structure and second language acquisition. In W.

Rutherford (Ed.), Typological universals and second language acquisition (pp. 109-136).
Givón, T. (1985). Function, structure and language acquisition. In D. Slobin (Ed.), The
crosslinguistic study of language acquisition Vol. 1 (pp. 1005-1028). Hillsdale: Lawrence
Erlbaum Associates.
Gough, D., Jones, J., Mfusi, M., Podile, K., Finlayson, R., Louw, J. A., & Satyo, S. C. (1989). Xhosa
study guide 3 for XHA100-F, XHB100-K. Pretoria: University of South Africa.
Gough, D. H. (1992). Demonstratives and word order: Aspects of discourse reference in Xhosa
narrative. South African Journal of African Languages, 12(1), 1-4.
Gough, D. H. (1995). Xhosa beyond the textbook: An analysis of grammatical variation of selected
Xhosa constructions. South African Journal of African Languages, 15(4), 156-161.
Gregg, K. R. (1989). Second language acquisition theory: The case for a generative perspective. In
S. M. Gass & J. Schachter (Eds.), Linguistic perspectives on second language acquisition (pp.
15-40). Cambridge: Cambridge University Press.
Hakuta, K. (1974). A preliminary report on the development of grammatical morphemes in a

Japanese girl learning English as a second language. Working Papers on Bilingualism, 3, 18-
43.
Hakuta, K. (1976). A case study of a Japanese child learning English as a second language. Language
Learning, 26(2), 321-351.
Hatch, E. (1978). Acquisition of syntax in a second language. In J. Richards (Ed.), Understanding

second and foreign language learning: Issues and approaches (pp. 34-70). Rowley: Newbury
House.
Hatch, E. (1979). Apply with caution. Studies in Second Language Acquisition, 2(1), 123-143.
Holm, J. (1990). Features in the noun phrase common to the Atlantic creoles. Linguistics, 28(4), 867-
881.
Hölscher, A., & Möhle, D. (1987). Cognitive plans in translation. In C. Faerch & G. Kasper (Eds.),
Introspection in second language research (pp. 113-134). Philadelphia: Multilingual Matters.
Huang, J., & Hatch, E. M. (1978). A Chinese child’s acquisition of English. In E. M. Hatch (Ed.),
Second language acquisition: A book of readings (pp. 118-131). Rowley: Newbury House.
227
Huebner, T. (1979). Order of acquisition vs. dynamic paradigm: A comparison of methods in

interlanguage research. TESOL Quarterly, 13(1), 21-28.
Huebner, T. (1983). A longitudinal analysis of the acquisition of English. Ann Arbor: Karoma.
Huebner, T. (1991). Second language acquisition: Litmus test for linguistic theory. In T. Huebner &
C. A. Ferguson (Eds.), Crosscurrents in second language acquisition and linguistic theories
(pp. 3-22). Amsterdam: John Benjamins.
Hulstijn, J. H. (1989). A cognitive view of interlanguage variability. In M. Eisenstein (Ed.), The

dynamic interlanguage: Empirical studies in second language variation (pp. 17-31). New
York: Plenum Press.
Hyltenstam, K. (1977). Implicational patterns in interlanguage syntax variation. Language Learning,

27(2), 383-411.
Jagtman, M., & Bongaerts, T. (1994). Report-COMOLA: A computer system for the analysis of
interlanguage data. Second Language Research, 10(1), 49-83.
Jake, J. L. (1998). Constructing interlanguage: Building a composite matrix language. Linguistics,

36(2), 333-382.
James, C. (1980). Contrastive analysis. Essex: Longman.
James, C. (1990). Learner language. Language Teaching, 23(4), 205-213.
James, C. (1994). Don’t shoot my dodo: On the resilience of contrastive and error analysis.
International Review of Applied Linguistics, 32(3), 179-200.
James, C. (1998). Error in language learning and use: Exploring error analysis. New York: Longman.
Johnson, D. M. (1992). Approaches to research in second language learning. New York: Longman.
Johnson, K. (1988). Mistake correction. English Language Teaching Journal, 42(2), 89-97.
Jokweni, M. W. (1997). Identifying, descriptive and associative copulatives in Xhosa: Structure and
function. South African Journal of African Languages, 17(3), 110-113.
Jordens, P. (1988). The acquisition of word order in L2 Dutch and German. In P. Jordens & J.
Lalleman (Eds.), Language development (pp. 149-180). Dordrecht: Foris.
Jordens, P. (1996). Input and instruction in second language acquisition. In P. Jordens & J. Lalleman
(Eds.), Investigating second language acquisition (pp. 407-449). Berlin: Mouton de Gruyter.
228
Jordens, P. (1997). Introducing the Basic Variety. Second Language Research, 13(4), 289-300.
Kanno, K. (1998). The stability of UG principles in second-language acquisition: Evidence from

Japanese. Linguistics, 36(6), 1125-1146.
Kaplan, T. I. (1998). General learning strategies and the process of L2 acquisition: A critical
overview. International Review of Applied Linguistics, 36(3), 233-246.
Karmiloff-Smith, A. (1979). A functional approach to child language. Cambridge: Cambridge

University Press.
Kellerman, E. (1977). Towards a characterisation of the strategy of transfer in second language

learning. Interlanguage Studies Bulletin, 2(1), 58-145.
Kellerman, E. (1979). Transfer and non-transfer: Where we are now. Studies in Second Language
Acquisition, 2(1), 37-57.
Kellerman, E. (1984). The empirical evidence for the influence of the L1 in interlanguage. In A.
Davies, C. Criper, & A. P. R. Howatt (Eds.), Interlanguage (pp. 98-122). Edinburgh:
Kellerman, E. (1986). An eye for an eye: Crosslinguistic constraints on the development of the L2
lexicon. In E. Kellerman & M. Sharwood Smith (Eds.), Crosslinguistic influence in second
language acquisition (pp. 35-48). Oxford: Pergamon.
Kellerman, E., & Sharwood Smith, M. (Eds.). (1986). Crosslinguistic influence in second language
acquisition. Oxford: Pergamon.
Kessler, C., & Idar, I. (1979). Acquisition of English by a Vietnamese mother and child. Working
Papers on Bilingualism, 18, 65-77.
Klein, W. (1986). Second Language Acquisition. Cambridge: Cambridge University Press.
Klein, W. (1991). SLA theory: Prolegomena to a theory of language acquisition and implications for
theoretical linguistics. In T. Huebner & C. A. Ferguson (Eds.), Crosscurrents in second
language acquisition and linguistic theories (pp. 169-194). Amsterdam: John Benjamins.
Klein, W. (1993). The acquisition of temporality. In C. Perdue (Ed.), Adult language acquisition:
Cross-linguistic perspectives: Vol. 2. The results (pp. 73-118). Cambridge: Cambridge
University Press.
Klein, W. (1998). The contribution of second language acquisition research. Language Learning,
48(4), 527-550.
229
Klein, W., Dietrich, R., & Noyau, C. (1993). The acquisition of temporality. In C. Perdue (Ed.),
Adult language acquisition: Cross-linguistic perspectives: Vol. 2: The results (pp. 73-118).
Klein, W., & Dittmar, N. (1979). Developing grammars. Berlin: Springer.
Klein, W., & Perdue, C. (1993). Utterance Structure. In C. Perdue Adult language acquisition:
Cross-linguistic perspectives: Vol. 2. The results (pp. 3-40). Cambridge: Cambridge
University Press.
Klein, W., & Perdue, C. (1997). The Basic Variety (or: Couldn’t natural languages be much simpler?)
Second Language Research, 13(4), 301-347.
Kleinmann, H. (1977). Avoidance behavior in adult second language acquisition. Language Learning,
27(1), 93-108.
Kolk, H., & Heeschen, C. (1992). Agrammatism, paragrammatism and the management of language.
Language and Cognitive Process, 7(2), 89-130.
Krashen, S. D. (1981). Second language acquisition and second language learning. New York:
Pergamon.
Krashen, S. D. (1982). Principles and practice in second language acquisition. Oxford: Pergamon.
Krashen, S. D. (1985). The input hypothesis: Issues and implications. New York: Longman.
Krashen, S. D., Houck, N., Guinchi, P., Bode, S., Birnbaum, R., & Strei, G. (1977). Difficulty order
for grammatical morphemes for adult second language performers using free speech. TESOL
Quarterly, 11(3), 338-341.
Krashen, S. D., & Scarcella, R. (1978). On routines and patterns in second language acquisition and
performance. Language Learning, 28(2), 283-300.
Krashen, S. D., Sferlazza, V., Feldman, L., & Fathman, A. (1976). Adult performance on the SLOPE
test: More evidence for a natural sequence in adult second language acquisition. Language
Learning, 26(1), 145-151.
Kumpf, L. (1984). Temporal systems and universality in interlanguage: A case study. In F. Eckman,
L. Bell, & D. Nelson (Eds.), Universals of second language acquisition (pp. 132-143).
Kunene, E. C. L. (1986). Acquisition of siSwati noun classes. South African Journal of African
Languages, 6(1), 34-37.
230
Lado, R. (1957). Linguistics across cultures. Applied linguistics for language teachers. Ann Arbor:
University of Michigan Press.
Lalleman, J. (1983). The principle of elimination: Establishing word order regularities in the Dutch
of foreign workers. Linguistiche Berichte, 87, 40-63.
Lalleman, J. (1996). The state of the art in second language acquisition research. In P. Jordens & J.
Lalleman (Eds.), Investigating second language acquisition (pp. 3-69). Berlin: Mouton de
Gruyter.
Lardiere, D. (1998). Dissociating syntax from morphology in a divergent L2 end-state grammar.

Second Language Research, 14(4), 359-375.
Larsen-Freeman, D. (1975). The acquisition of grammatical morphemes by adult ESL students.

TESOL Quarterly, 9(4), 409-430.
Larsen-Freeman, D. (1991). Second language acquisition research: Staking out the territory. TESOL
Quarterly, 25(2), 315-350.
Larsen-Freeman, D. (1998). On the scope of second language acquisition research: “The learner
variety” perspective and beyond - a response to Klein. Language Learning, 48(4), 551-556.
Larsen-Freeman, D., & Long, M. H. (1991). An introduction to second language acquisition

research. London: Longman.
Lennon, P. (1991). Error: Some problems of definition, identification and distinction. Applied
Linguistics, 12(2), 180-196.
Lightbown, P. M. (1980). The acquisition and use of questions by French L2 learners. In S. Felix
(Ed.), Second language development: Trends and issues (pp. 151-175). Tubingen: Gunter
Narr Verlag.
Lightbown, P. M. (1983). Exploring relationships between developmental and instructional sequences

in L2 acquisition. In H. Seliger & M. Long (Eds.), Classroom oriented research in second
language acquisition (pp. 217-245). Rowley: Newbury House.
Lightbown, P. M. (1984). The relationship between theory and method in second language
acquisition research. In A. Davies, C. Criper, & A. P. R. Howatt (Eds.), Interlanguage (pp.
241-252). Edinburgh: Edinburgh University Press.
Lightbown, P. M., & Spada, N. (1990). Focus-on-form and corrective feedback in communicative
teaching: Effects on second language learning. Studies in Second Language Acquisition,
12(4), 429-448.
231
Long, M. H. (1983). Does second language instruction make a difference? TESOL Quarterly, 17(3),
359-382.
Long, M. H. (1988). Instructed interlanguage development. In L. Beebe (Ed.), Issues in second

language acquisition: Multiple perspectives (pp. 115-142). Rowley: Newbury House.
Long, M. H. (1991). A design feature in language teaching methodology. In K. de Bot, R.

Grimsberg, & C. Kramsch (Eds.), Foreign language research in cross-cultural perspective (pp.
39-52). Amsterdam: John Benjamins.
Long, M. H., & Sato, C. J. (1984). Methodological issues in interlanguage studies: An interactionist
perspective. In A. Davies, C. Criper, & A. P. R. Howatt (Eds.), Interlanguage (pp. 253-279).
Edinburgh: Edinburgh University Press.
Lyons, J. (1968). An introduction to theoretical linguistics. Cambridge: Cambridge University Press.
Mace-Matluck, B. J. (1979). Order of acquisition: Same or different in first- and second-language

learning? Reading Teacher, 32(6), 696-703.
Makoni, S. B. (1996). Variation in unplanned discourse. International Review of Applied Linguistics,

34(3), 167-181.
Marshall, H. W. (1989). The colloquial preterit: Language variation and the ESL classroom. In M.
Eisenstein (Ed.), The dynamic interlanguage: Empirical studies in second language variation
(pp. 297-315). New York: Plenum Press.
McGroarty, M. (1998). Cognitive and constructivist challenges for applied linguistics. Language
Learning, 48(4), 591-622.
McLaughlin, B. (1990). Restructuring. Applied Linguistics, 11(2), 114-128.
Meisel, J. M. (1987). Reference to past events and actions in the development of natural second
language acquisition. In C. Pfaff (Ed.), First and second language acquisition processes (pp.
206-224). Cambridge: Newbury House.
Meisel, J. M. (1997). The L2 Basic Variety as an I-language. Second Language Research, 13(4),
374-385.
Meisel, J. M., Clahsen, H., & Pienemann, M. (1981). On determining developmental stages in natural
second language acquisition. Studies in Second Language Acquisition, 3(2), 109-135.
Mitchell, R., & Myles, F. (1998). Second language learning theories. London: Arnold.
232
Munnich, E., Flynn, S., & Martohardjono, G. (1994). Elicited imitation and grammaticality judgment
tasks: What they measure and how they relate to each other. In E. E. Tarone, S. M. Gass, &
A. D. Cohen (Eds.), Research methodology in second-language acquisition (pp. 227-243).
Hillsdale: Lawrence Erlbaum Associates.
Musau, P. M. (1995). Communicative strategies of Swahili learners: The one to one principle.
Myhill, J. (1991). Typological text analysis: Tense and aspect in creoles and second languages. In T.
Huebner & C. A. Ferguson (Eds.), Crosscurrents in second language acquisition and linguistic
theories (pp. 93-121). Amsterdam: John Benjamins.
Myles, F., Hooper, J., & Mitchell, R. (1998). Rote or rule? Exploring the role of formulaic language
in classroom foreign language learning. Language Learning, 48(3), 323-363.
Myles, F., Mitchell, R., & Hooper, J. (1999). Interrogative chunks in French L2: A basis for creative
construction? Studies in Second Language Acquisition, 21(1), 49-80.
Naiman, N., Frohlich, M., Stern, H. H., & Todesco, A. (1978). The good language learner. Research
in Education Series 7: The Ontario Institute for Studies in Education.
Nakuma, C. K. (1998). A new theoretical account of “fossilization”: Implications for L2 attrition

research. International Review of Applied Linguistics, 36(3), 247-256.
Nattinger, J. R., & DeCarrico, J. S. (1992). Lexical phrases and language teaching. Oxford: Oxford
University Press.
Nemser, W. (1971). Approximative systems of foreign language learners. International Review of

Nickel, G. (1989). Some controversies in present-day error analysis: ‘Contrastive’ vs. ‘non-
contrastive’ errors. International Review of Applied Linguistics, 27(4), 293-305.
Nunan, D. (1996). Issues in second language acquisition research: Examining substance and
procedure. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of second language acquisition
(pp. 349-374). San Diego: Academic Press.
Odlin, T. (1989). Language transfer: Cross-linguistic influence in language learning. Cambridge:

Cambridge University Press.
Oller, J. W. (1979). Language tests at school: A pragmatic approach. London: Longman.
O’Malley, J., & Chamot, A. (1990). Learning strategies in second language acquisition. New York:
233
Ortega, L. (1999). Planning and focus on form in L2 oral performance. Studies in Second Language
Acquisition, 21(1), 109-148.
Paradis, J., Le Corre, M., & Genesee, F. (1998). The emergence of tense and agreement in child L2
French. Second Language Research, 14(3), 227-256.
Pawley, A., & Syder, F. (1983). Two puzzles for linguistic theory: Nativelike selection and nativelike
fluency. In J. Richards & J. Schmidt (Eds.), Language and communication (pp. 191-266).
London: Longman.
Perdue, C. (1990). Complexification of the simple clause in the narrative discourse of adult language
learners. Linguistics, 28(5), 983-1009.
Perdue, C. (1991). Cross-linguistic comparisons: Organizational principles in learner languages. In

T. Huebner & C. A. Ferguson (Eds.), Crosscurrents in second language acquisition and
linguistic theories (pp. 405-422). Amsterdam: John Benjamins.
Perdue, C. (Ed.). (1993a). Adult language acquisition: Cross-linguistic perspectives: Vol. 1. Field
methods. Cambridge: Cambridge University Press.
Perdue, C. (Ed.). (1993b). Adult language acquisition: Cross-linguistic perspectives: Vol. 2. The
results. Cambridge: Cambridge University Press.
Perdue, C. (1996). Pre-basic varieties: The first stages of second language acquisition. In E.
Kellerman, B. Weltens, & T. Bongaerts (Eds.), EUROSLA 6: A selection of papers (pp. 135-
149). Amsterdam: VU Uitgeverij.
Perdue, C., & Klein, W. (1992). Why does the production of some learners not grammaticalize?
Studies in Second Language Acquisition, 14(3), 259-272.
Perdue, C., & Klein, W. (1993). Concluding Remarks. In C. Perdue (Ed.), Adult language
acquisition: Cross-linguistic perspectives: Vol. 2. The results (pp. 253-272). Cambridge:
Perkins, K., & Larsen-Freeman, D. (1975). The effects of formal language instruction on the order
of morpheme acquisition. Language Learning, 25(2), 237-243.
Pfaff, C. (1987). Functional approaches to interlanguage. In C. Pfaff (Ed.), First and second language
acquisition processes (pp. 81-102). Cambridge: Newbury House.
Pica, T. (1983). Adult acquisition of English as a second language under different conditions of
exposure. Language Learning, 33(4), 465-497.
234
Pica, T. (1998). Second language learning through interaction: Multiple perspectives. In V. Regan
(Ed.), Contemporary approaches to second language acquisition in social context (pp. 9-31).
Dublin: University College Dublin Press.
Pienemann, M. (1984). Psychological constraints on the teachability of languages. Studies in Second

Language Acquisition, 6(2), 186-214.
Pienemann, M. (1987). Psychological constraints on the teachability of languages. In C. Pfaff (Ed.),

First and second language processes (pp. 143-168). Cambridge: Newbury House.
Pienemann, M. (1989). Is language teachable? Psycholinguistic experiments and hypotheses. Applied

Pienemann, M. (1992). COALA - a computational system for interlanguage analysis. Second

Pinker, S. (1984). Language learnability and language development. Cambridge: Harvard University
Press.
Posthumus, L. C. (1994). Word-based versus root-based morphology in the African languages. South
African Journal of African Languages, 14(1), 28-36.
Radford, A. (1988). Transformational grammar: A first course. Cambridge: Cambridge University

Press.
Richards, J. C. (1985). The context of language teaching. Cambridge: Cambridge University Press.
Robert, J-M. (1989). Intralanguage, agrammatism and the conceptual system. International Review
Roberts, C., & Simonot, M. (1987). This is my life: How language acquisition is interactionally
accomplished. In R. Ellis (Ed.), Second language acquisition in context (pp. 133-148).
London: Prentice-Hall.
Robinson, P. (1996a). Consciousness, rules and instructed second language acquisition. New York:
Peter Lang.
Robinson, P. (1996b). Learning simple and complex second language rules under implicit, incidental,
rule-search, and instructed conditions. Studies in Second Language Acquisition, 18(1), 27-67.
Robinson, P. (1997a). Generalizability and automaticity of second language learning under implicit,
incidental, enhanced, and instructed conditions. Studies in Second Language Acquisition,
19(2), 223-247.
235
Robinson, P. (1997b). Individual differences and the fundamental similarity of implicit and explicit
adult second language learning. Language Learning, 47(1), 45-99.
Rosansky, E. (1976). Methods and morphemes in second language acquisition. Language Learning,
26(2), 409-425.
Rutherford, W. E. (1984). Description and explanation in interlanguage syntax: State of the art.
Rutherford, W. E. (1987). Second language grammar: Learning and teaching. London: Longman.
Rutherford, W. E., & Sharwood Smith, M. (1985). Consciousness raising and Universal Grammar.
Sajjadi, S., & Tahririan, M. H. (1992). Task variability and interlanguage use. International Review
Sanchez, L., & José Giménez, M. (1998). The L2 acquisition of definite determiners from null to
Sá
overt. In A. Greenhill, M. Hughes, H. Littlefield, & H. Walsh (Eds.), Proceedings of the 22nd
annual Boston University Conference on language development Vol. 2 (pp. 640-650).
Somerville: Cascadilla.
Sato, C. J. (1988). Origins of complex syntax in interlanguage development. Studies in Second

Sato, C. J. (1990). The syntax of conversation in interlanguage development. Tubingen: Gunter Narr
Verlag.
Satyo, S. (1998). Soft harmonisation and cross-fertilising vocabularies: The case of Nguni languages.
In K. Kwaa Prah (Ed.), Between distinction and extinction: The harmonisation and
standardisation of African languages. Johannesburg: Witwatersrand University Press.
Schachter, J. (1974). An error in error analysis. Language Learning, 24(2), 205-214.
Schachter, J. (1989). A new look at an old classic. Second Language Research, 5(1), 30-42.
Schachter, J. (1992). A new account of language transfer. In S. M. Gass & L. Selinker (Eds.),
Language transfer in language learning (rev. ed.), (pp. 32-46). Amsterdam: John Benjamins.
Schachter, J. (1998). Recent research in language learning studies: promises and problems. Language
Learning, 48(4), 557-583.
Schachter, J., & Celce-Murcia, M. (1977). Some reservations concerning error analysis. TESOL
Quarterly, 11(4), 441-451.
236
Schenning, S., & van Hout, R. (1994). Dimensional spatial relations in adult language acquisition.
In R. Bok-Bennema & C. Cremers (Eds.), Linguistics in the Netherlands (pp. 235-246).
Schmidt, R. (1990). The role of consciousness in second language learning. Applied Linguistics,
11(2), 129-158.
Schumann, J. H. (1975). Affective factors and the problem of age in second language acquisition.
Schumann, J. H. (1978). The pidginization process: A model for second language acquisition.
Schumann, J. H. (1979). The acquisition of English negation by speakers of Spanish: A review of the
literature. In R. Andersen (Ed.), The acquisition and use of Spanish and English as first and
second languages (pp. 3-23). Washington, DC: TESOL.
Schumann, J. H. (1982). Simplification, transfer, and relexification as aspects of pidginization and

early second language acquisition. Language Learning, 32(2), 337-366.
Schumann, J. H. (1987). The expression of temporality in basilang speech. Studies in Second

Schwartz, B. D. (1993). On explicit and negative data affecting competence and linguistic behavior.
Schwartz, B. D. (1997). On the basis of the Basic Variety. Second Langauge Research, 13(4), 386-
402.
Schwartz, B. D., & Sprouse, R. (1994). Word order and nominative case in non-native language
acquisition: A longitudinal study of (L1 Turkish) German interlanguage. In T. Hoekstra & B.
Schwartz (Eds.), Language acquisition studies in generative grammar: Papers in honor of
Kenneth Wexler for the GLOW 1991 Workshops (pp. 317-368). Amsterdam: John
Benjamins.
Scovel, T. (1988). A time to speak. Rowley: Newbury House.
Seliger, H. W. (1979). On the nature and function of language rules in language teaching. TESOL
Quarterly, 13(3), 359-369.
Seliger, H. W., & Long, M. H. (Eds.). (1983). Classroom-oriented research in second language
acquisition. Rowley: Newbury House.
237
Seliger, H. W., & Shohamy, E. (1989). Second language research methods. Oxford: Oxford
University Press.
Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics, 10(3), 209-231.
Selinker, L. (1984). The current state of IL studies: An attempted critical summary. In A. Davies, C.
Criper, & A. P. R. Howatt (Eds.), Interlanguage (pp. 332-343). Edinburgh: Edinburgh
University Press.
Selinker, L. (1989). CA/EA/IL: The earliest experimental record. International Review of Applied
Linguistics, 27(4), 267-291.
Selinker, L. (1992). Rediscovering Interlanguage. New York: Longman.
Selinker, L., & Douglas, D. (1985). Wrestling with ‘context’ in interlanguage theory. Applied
Selinker, L., & Lakshamanan, U. (1992). Language transfer and fossilization: The multiple effects
principle. In S. M. Gass & L. Selinker (Eds.), Language transfer in language learning (rev.
ed.), (pp. 197-216). Amsterdam: John Benjamins.
Selinker, L., & Lamendella, J. (1978). Two perspectives on fossilization in interlanguage learning.
Interlanguage Studies Bulletin, 3(2), 144-191.
Sharwood Smith, M. (1979). Strategies, language transfer and the simulation of the second language
learner’s mental operations. Interlanguage Studies Bulletin, 4(1), 66-83.
Sharwood Smith, M. (1991). Speaking to many minds: On the relevance of different types of
language information for the L2 learner. Second Language Research, 7(2), 118-132.
Sharwood Smith, M. (1993). Input enhancement in instructed SLA. Studies in Second Language
Acquisition, 15(2), 165-179.
Sharwood Smith, M. (1994). Second language learning: Theoretical foundations. London: Longman.
Shaughnessy, M. (1977). Errors and expectations. Oxford: Oxford University Press.
Silva-Corvalán, C. (1991). Cross-generational bilingualism: Theoretical implications of language

attrition. In T. Huebner & C. A. Ferguson (Eds.), Crosscurrents in second language
acquisition and linguistic theories (pp. 325-345). Amsterdam: John Benjamins.
Singh, R., d’Anglejan, A., & Carroll, S. (1982). Elicitation of Inter-English. Language Learning,
32(2), 271-288.
238
Skiba, R., & Dittmar, N. (1992). Pragmatic, semantic, and syntactic constraints and
grammaticalization. Studies in Second Language Acquisition, 14(3), 323-349.
Slobin, D. (1982a). The cross-linguistic study of language acquisition. Hillsdale: Lawrence Erlbaum
Associates.
Slobin, D. (1982b). Universal and particular in the acquisition of language. In E. Wanner & L.
Gleitman (Eds.), Language acquisition: The state of the art (pp. 128-170). Cambridge:
Slobin, D. (1985). The crosslinguistic study of language acquisition Vol. 2: The theoretical issues.
Hillsdale: Lawrence Erlbaum Associates.
Slobin, D. (1992). Introduction. In D. Slobin (Ed.), The crosslinguistic study of language acquisition
Vol. 3 (pp. 1-13). Hillsdale: Lawrence Erlbaum Associates.
Sorace, A. (1996). The use of acceptability judgements in second language acquisition research. In
W. Ritchie & T. Bhatia (Eds.), Handbook of second language acquisition (pp. 375-409). San
Diego: Academic Press.
Spolsky, B. (1989). Conditions for second language learning. Oxford: Oxford University Press.
Stauble, A. (1984). A comparison of the Spanish-English and Japanese-English interlanguage

continuum. In R. Andersen (Ed.), Second language: A crosslinguistic perspective (pp. 323-
353). Rowley: Newbury House.
Stenson, N. (1974). Induced errors. In J. Schumann & N. Stenson (Eds.), New frontiers in second
language learning (pp. 54-70). Rowley: Newbury House.
Stevick, E. (1989). Success with foreign languages. New York: Prentice Hall.
Suzman, S. M. (1982). Strategies for acquiring Zulu concord. South African Journal of African
Languages, 2(3), 53-67.
Suzman, S. M. (1999). Learn Zulu the way children do. South African Journal of African Languages,
19(2), 134-147.
Swain, M., Dumas, G., & Naiman, N. (1974). Alternatives to spontaneous speech: Elicited
translation and imitation as indicators of second language competence. Working Papers on
Bilingualism, 3, 68-79.
Swan, M. (1987). Non-systematic variability: A self-inflicted conundrum? In R. Ellis (Ed.), Second

language acquisition in context (pp. 59-66). London: Prentice-Hall.
239
Tarone, E. E. (1982). Systematicity and attention in interlanguage. Language Learning, 32(1), 69-84.
Tarone, E. E. (1983). On the variability of interlanguage systems. Applied Linguistics, 4(2), 143-163.
Tarone, E. E. (1987). Methodologies for studying variability in second language acquisition. In R.

Ellis (Ed.), Second language acquisition in context (pp. 35-46). London: Prentice-Hall.
Tarone, E. E. (1988). Variation in interlanguage. London: Edward Arnold.
Tarone, E. E. (1989). On chameleons and monitors. In M. Eisenstein (Ed.), The dynamic

interlanguage: Empirical studies in second language variation (pp. 3-15). New York: Plenum
Press.
Tarone. E. E. (1994). A summary: Research approaches in studying second-language acquisition or

“if the shoe fits...”. In E. E. Tarone, S. M. Gass, & A. D. Cohen (Eds.), Research
methodology in second-language acquisition (pp. 323-336). Hillsdale: Lawrence Erlbaum
Associates.
Tarone, E. E., Cohen, A., & Dumas, G. (1976). A closer look at some interlanguage terminology:
A framework for communication strategies. Working Papers on Bilingualism, 9, 76-90.
Tomasello, M., & Herron, C. (1988). Down the garden path: Inducing and correcting
overgeneralization errors in the foreign language classroom. Applied Psycholinguistics, 9(3),
237-246.
Tomasello, M., & Herron, C. (1989). Feedback for language transfer errors: The garden path
technique. Studies in Second Language Acquisition, 11(4), 385-395.
Tomlin, R., & Villa, V. (1994). Attention in cognitive science and second language acquisition.
Towell, R. (1987). Variability and progress in the language development of advanced learners of a
foreign language. In R. Ellis (Ed.), Second language acquisition in context (pp. 113-127).
Towell, R., & Hawkins, R. (1994). Approaches to second language acquisition. Clevedon:
Multilingual Matters.
Trahey, M., & White, L. (1993). Positive evidence and preemption in the second language classroom.
Trévise, A. (1987). Toward an analysis of the (inter)language activity of referring to time in

narrative. In C. Pfaff (Ed.), First and second language acquisition processes (pp. 225-251).
Cambridge: Newbury House.
240
Vainikka, A., & Young-Scholten, M. (1996). The early stages in adult L2 syntax: Additional
evidence from Romance speakers. Second Language Research, 12(2), 140-176.
Valdman, A. (Ed.). (1977). Pidgin and creole. Indiana: Indiana University Press.
Valdman, A. (1989). Classroom foreign language learning and language variation: The notion of
pedagogical norms. In M. Eisenstein (Ed.), The dynamic interlanguage: Empirical studies in
second language variation (pp. 261-277). New York: Plenum Press.
Van Els, T., Bongaerts, T., Extra, G., Van Os, C., & Janssen-van Dieten, A. (1984). Applied
linguistics and the learning and teaching of foreign languages (2nd ed.). London: Edward
Arnold.
VanPatten, B. (1984). Processing strategies and morphological acquisition. In F. R. Eckman, L. H.

Bell, & D. Nelson (Eds.), Universals of language acquisition (pp. 88-98). Rowley: Newbury
House.
Váradi, T. (1980). Strategies of target language learner communication: Message adjustment.

Veronique, D. (1987). Reference to past events and actions in narratives in a second language:
Insights from North African workers’ French. In C. Pfaff (Ed.), First and second language
acquisition processes (pp. 252-272). Cambridge: Newbury House.
Visser, M. (1989). The syntax of the infinitive in Xhosa. South African Journal of African Languages,
9(4), 154-184.
Von Stutterheim, C., & Klein, W. (1987). A concept-oriented approach to second language studies.
In C. Pfaff (Ed.), First and second language acquisition processes (pp. 191-205). Cambridge:
Newbury House.
Wagner-Gough, J. (1975). Comparative studies in second language learning. CALERIC/CLL

Series on languages and linguistics 26.
Wardhaugh, R. (1970). The contrastive analysis hypothesis. TESOL Quarterly, 4(2), 123-130.
Weinert, R. (1987). Processes in classroom second language development: The acquisition of

negation in German. In R. Ellis (Ed.), Second language acquisition in context (pp. 83-99).
Weinert, R. (1995). The role of formulaic language in second language acquisition: A review. Applied
Linguistics, 16(2), 180-205.
241
Weist, R. (1986). Tense and aspect: Temporal systems in child language. In P. Fletcher & M.
Garman (Eds.), Language acquisition: Studies in first language development (pp. 356-374).
White, L. (1977). Error-analysis and error-correction in adult learners of English as a second

language. Working Papers on Bilingualism, 13, 42-58.
White, L. (1991). Adverb placement in second language acquisition: Some effects of positive and
negative evidence in the classroom. Second Language Research, 7(2), 133-161.
White, L. (1992). Universal Grammar: Is it just a new name for old problems? In S. M. Gass & L.
Selinker (Eds.), Language transfer in language learning (rev. ed.), (pp. 217-232). Amsterdam:
John Benjamins.
White, L. (1998). Introduction: The implications of divergent outcomes in L2 acquisition. Second

White, L., Spada, N., Lightbown, P., & Ranta, L. (1991). Input enhancement and L2 question
formation. Applied Linguistics, 12(4), 416-432.
Whitman, R., & Jackson, K. (1972). The unpredictability of contrastive analysis. Language Learning,
22(1), 29-41.
Williams, J. (1989). Variation and convergence in nonnative institutionalized Englishes. In M.

Eisenstein (Ed.), The dynamic interlanguage: Empirical studies in second language variation
(pp. 117-135). New York: Plenum Press.
Wode, H. (1976). Developmental sequences in naturalistic L2 acquisition. Working Papers on

Wode, H. (1978). The L1 vs. L2 acquisition of English interrogation. Working Papers on

Wode, H. (1979). Operating principles and ‘universals’ in L1, L2, and FLT. International Review of
Wode, H. (1981). Learning a second language: An integrated view of language acquisition.

Tubingen: Gunter Narr Verlag.
Yang, L. R., & Givon, T. (1997). Benefits and drawbacks of controlled laboratory studies of second
language acquisition: The Keck second language learning project. Studies in Second
242
Young, R. (1988). Variation and the interlanguage hypothesis. Studies in Second Language
Acquisition, 10(3), 281-302.
Young, R. (1991). Variation in interlanguage morphology. New York: Peter Lang.
Yuan, B. (1997). Asymmetry of null subjects and null objects in Chinese speakers’ L2 English.
Zobl, H. (1979). Systems of verb classification and cohesion of verb-complement relations as

structural conditions on interference in a child’s L2 development. Working Papers on
Zobl, H. (1980). The formal and developmental selectivity of L1 influence on L2 acquisition.

Zobl, H. (1982). A direction for contrastive analysis: The comparative study of developmental
sequences. TESOL Quarterly, 16(2), 169-183.
Zobl, H. (1984). Cross-generalizations and the contrastive dimension of the interlanguage hypothesis.
In A. Davies, C. Criper, & A. P. R. Howatt (Eds.), Interlanguage (pp. 79-97). Edinburgh:
Zobl, H. (1992). Prior linguistic knowledge and the conservatism of the learning procedure:
Grammaticality judgments of unilingual and multilingual learners. In S. M. Gass & L. Selinker
(Eds.), Language transfer in language learning (rev. ed.), (pp. 176-196). Amsterdam: John
Benjamins.
Zobl, H. (1995). Converging evidence for the ‘acquisition-learning’ distinction. Applied Linguistics,
16(1), 35-56.
Zotwana, S. Z. (1991). Xhosa in context: From novice to intermediate. Cape Town: Perskor.
243
APPENDIX A: BIOGRAPHICAL/ATTITUDINAL QUESTIONNAIRE
1. How old are you now?
2. When did you start learning Xhosa?
3. Why did you decide to study Xhosa at university?
4. What would you like to do with your Xhosa?
5. Do you talk to L1 speakers outside the classroom?

If yes - how often do you do this?
6. What kind of Xhosa classroom exercises or activities do you enjoy most?
7. How would you rate yourself on:

a) speaking ability
b) understanding of spoken language
c) reading ability
d) writing ability
8. How important do you think it is to pronounce Xhosa correctly?
9. How important do you think it is to produce correct grammar when you are speaking?
10. What do you think a Xhosa course at university should teach you?
11. How many hours a week do you think you spend on Xhosa outside the classroom?
12. Do you intend to do Xhosa next year?
13. Would you encourage other people to learn Xhosa?
14. Do you enjoy learning Xhosa?

244
APPENDIX B: A SHORT SUMMARY OF THE BACKGROUND OF THE

PARTICIPANTS
Ann
Ann was a beginner Xhosa learner at the start of the data-collection period. She was exposed only
to input in the formal classroom and never interacted with L1 Xhosa speakers, except to greet
them. Her interlanguage is at a very basic level at the beginning, although she uses a range of basic
morphemes from the beginning of the interview period.
Ben
Ben had learned some Xhosa in primary school and for the first two years of high school. He was
in the first-year class during the data-collection period. He rarely spoke to L1 speakers, although
he had had extensive exposure to Xhosa at a young age. His interlanguage is fairly fluent,
although the range of forms is limited.
Claire
Claire grew up on a farm where she learnt Xhosa, but she had never learned to read and write the
language. She was in the first-year class so that she could learn to read and write Xhosa. She is
a fluent speaker who uses a range of forms, although analysis of her interlanguage shows many
errors.
Pat
Pat had had some exposure to Xhosa at home on a farm and she was willing to communicate with
L1 speakers. She was in the second-year class at the time of the interviews. She made many
errors, although she used a wider range of forms than Ann and Ben.
Karen
Karen had studied Xhosa at school and was in the second-year Xhosa class. She produced a wide
range of forms which are usually correct, although she spoke fairly slowly and appeared to be self-
monitoring to a large extent.
245
Sarah
Sarah had studied Xhosa at school and was in the third-year Xhosa class. She produced a range
of forms, although there were still errors in several basic forms. She felt that her ability to speak
had deteriorated over the years.
246
APPENDIX C: INTERVIEW MATERIAL

Interview 1
1. Have a conversation with X about where he/she comes from, his/her surname, his/her
family, his/her friends, why he/she is at Rhodes, when he/she came to Rhodes, etc.
2. Have a conversation with X about what X usually does in his/her Xhosa class.
3. Make up a story about Thembi. Thembi says a few sentences, X says a few sentences.
About two or three sentences at a time. Carry on until you have exhausted the subject.
4. Thembi reads a passage to X. X writes down as much as he/she can. Thembi reads once
through, then twice more, fairly slowly.
Passage:
Abafundi bahlala ezidesikeni. Idesika ihlala umfundi omnye. Abafundi
bangamashumi amabini anesihlanu eklasini yooThembi. Ngoko ke kukho iidesika
ezingamashumi amabini anesihlanu. Idesika zineziciko. Abafundi bayazivula bafake
iincwadi nezinto zokubhala. Abanye abafundi bayathandana ukubhala iidesika.
(From Zotwana 1991:96)
5. Translate these sentences into Xhosa.
Hello, my name is Thembi.

I am studying at Rhodes this year.
I live in town in a house.
I would like to learn Xhosa to talk to other people.
Open the door.
I am going to PE this weekend.
6. Translate these sentences into English.
Sisitulo sikatitshalakazi. Utitshalakazi usondeza isitulo etafileni ahlale. Abantwana babeka

iincwadi etafileni phambi kukatitshala. Utitshalakazi uyazivula, ajonge umsebenzi
wabantwana, alungise iziphoso. Usebenzisa usiba olubomvu. Ubanika amanqaku
abantwana. Uwabhala encwadini yamanqaku. Ngezinye iimini utitshalakazi ubanika
umsebenzi wasekhaya abafundi.
7. X describes the picture in as much detail as possible.
8. X builds sentences with the following words:
vula; biza; beka; bona; thanda

247
Interview 2 (repeated in interview 5)
1. Have a conversation with X about his/her family. Who they are, what they do, how old
they are, what they like to do, what they like to eat, etc.
2. Have a conversation with X about what he/she did during the vac.
3. Make up a story about animals (ihashe/ingonyama/inja). Thembi says a few sentences, X

says a few sentences. About two or three sentences at a time. Carry on until you have
exhausted the subject.
Passage:
Abasebenzi basenga iinkomo ederi. Bavuka ekuseni ngenj'ixukuxa, basenge lide
liphume ilanga kuba iimazi zininzi. Abasengi ngezandla, basenga ngoomatshini
bokusenga. Olunye ubisi lugalelwa ezikalini lusiwe edolophini ngelori: olunye
luyajijwa kwenziwe ibhotolo ngalo: olunye lunikwa abasebenzi. Ixibhiya inikwa
izinja neehagu. Efameni kukho iigusha zoboya kunye neeseyibhokhwe.
Ziyachetywa qho ngoSeptemba. Uboya buthunyelwa emalikeni yoboya eMonti.
I have a headache.
Grandfather lives in town.
My brother is studying.
The parents buy meat but the children buy sweets.
They like milk.
We speak Xhosa in class.
The horse eats grass.
Utitshalakazi ubhala ebhodini. Ubhala ngetshokhwe. Ubhala imibuzo yomsebenzi

waseklasini ikanye yovavanyo. Ushwankathela zonke izifundo ebhodini. Sakuphela
isifundo utitshalakazi uthatha idasta asule ibhodi. Ngamanye amaxesha abafundi bayayisula
ibhodi. UThembi uyathanda ukusula ibhodi. Kodwa akafiki phezulu kuba mfutshane.
Ibhodi imnyama.
Abafundi babhala ngeentsiba, nangeepensile. Iintsiba zabafundi zisebenzisa i-inki. I-inki

isebhotileni ye-inki. Abanye abafundi basebenzisa i-inki emnyama. Utitshalakazi akafuni
abafundi basebenzise i-inki emnyama; ufuna basebenzise eblowu.
(From Zotwana 1991:98-99)
7. X describes the picture in as much detail as possible. Elicit data by asking: Yintoni
le/Ziintoni ezi and Wenza ntoni?
248
inja; thanda; thetha; abafazi; dlala; tya; vala; ubhuti; hlala; bhala
9. Use the following words to make grammatical Xhosa sentences. You may change the
form of the words and add extra elements to form a grammatical sentence.
a) tya umntwana ukutya

b) thanda abafazi thenga idolophu inyama
c) utitshalakazi nika si iipensile ikhabhathi
10. Look carefully at the following sentences and correct any errors in them.
a) Isikolo sikaThembi singena ngo-8.00, kodwa intsimbi zikhala ngo-7.55.

b) Abantwana bayaphuma eziklasini baya phandle, baphumla imizuzu elishumi elinisihlanu,
sidlala.
c) Ihashe ibaleka iplasini.
d) Sibhala ngipensile.
e) Sivala zifestile nacango.
249
Interview 3 (repeated in interview 7)
1. Have a conversation with X about the weather and what X likes to do in different kinds
of weather. You could also discuss typical weather for different seasons.
2. Role-play a situation where X asks you out for an evening. Talk about where you will
meet, what you will do, etc.
3. Make up a story about living in res. Thembi says a few sentences, X says a few sentences.
About two or three sentences at a time. Carry on until you have exhausted the subject.
Passage:
Ngoku masingene egumbini lokutyela. Igumbi lokutyela lakuthi lincinci. Kukho
itafile nezitulo. Izitulo zithandathu. Phezu kwetafile kukho ilaphu letafile. Kukho
isayibhodi. Umama ufaka izitya neeglasi neekomityi esayibhodini. Ezinye izitya
zihlala ekhabhathini ekitshini. Isayibhodi ineedrowa. Ezidroweni umama ufaka
amacephe, iimela, iifolokhwe namatispuni. Amanye amatispuni namacephe
neemela neefolokhwe zihlala ekhabhathini ekhitshini.
We drink this water.

They don’t like meat.
This cat doesn’t want food.
Those people live in the forest.
Some people don’t live in houses.
When does that bus arrive?
These children don’t have those books.
Nasi isitovu. Sisitovu sombane. Umama uphekela esitovini. Isitovu sakuthi sinewoma
neondi. Iipleyiti zesitovu sakuthi zine. Iipleyiti ezintathu zinkulu. Ipleyiti encinci inye.
Esitovini siyapheka, siyafrayisha. Sipheka ukutya; sifrayisha amaqanda okanye inyama.
Ngamanye amaxesha umama uyabhaka esitovini, e-ondini. Ubhaka isonka okanye iikeyiki.
Mna notata asikwazi ukubhaka, kodwa sifuna ukufunda. Umama uthi uza kusifundisa
ukubhaka.
7. X describes the picture in as much detail as possible. Elicit negative sentences by asking
questions which require negative answers.
iincwadi; iindlebe; kushushu; imphelaveki; amaphepha; ukutya; amahashe; abantwana;

250
ixesha; kusasa
9. Use the following words to make grammatical NEGATIVE Xhosa sentences. You may
change the form of the words and add extra elements to form a grammatical sentence.
a) bhala umfundi incwadi

b) ikati phandle lala
c) amahashe inyama tya
10. Change the following sentences into the negative

a) Umama uya ngetreyini edolophini.
b) Ndithanda ukubhala ngepensile.
c) Ibhasi ifika ngo-2:00 ekuseni.
a) La mntu uyafunda Monti.

b) Lamahashe la abaleka.
c) Abantwana akafuni kuya eBhayi.
d) Ezi inja sitya.
e) Oomama aziseli ubisi.
251
Interview 4
1. Have a conversation with X about what he/she intends to do this term, e.g. work,
recreation, sport, etc.
2. Build a story about each other's vacs. X says two or three sentences, Thembi says two or
three sentences.
3. Role-play a situation where X wants to buy some items and Thembi is shop assistant e.g.
questions on size, price, where to find it, how many, what colour, etc.
Passage:
Namhlanje ndifuna ukunixelela ngeYunivesiti yaseKapa. Ndifuna ukunixelela
ngeyunivesiti kuba ndiyayithanda. Ndiyithanda kakhulu. Yiyunivesiti enkulu
kakhulu. Inabafundi abaninzi. Abafundi apha bavela kwiimbombo zone zomhlaba.
Abanye bavela phesheya. Ngamanye amaxesha abadlali bezinye iindawo bayeza
apha. Kumnandi kakhulu apha ngempelaveki kodwa mna andidlali kulo nyaka.
The children play outside.

They want to drink cooldrink.
We don't like meat.
Where is the shop?
This tree is green.
Those men drink coffee, but these women drink tea.
They don't eat bread on weekends.
6. Translate this passage into English:
uMatthew usengxakini kakhulu namhlanje. Ufuna ukufunda isiXhosa kodwa iincwadi

zakhe azikho; zisesikolweni. Akakwazi ukuzifumana kuba nguMgqibelo namhlanje.
Ootitshala bayayitshixa igeyithi esikolweni. Ngoko ke abafundi abakwazi ukungena
esikolweni ngempelaveki. Ngoku uMatthew ufuna ukusebenzisa ifowuni athethe noJohn
Patterson, umhlobo wakhe. Ikhaya likaJohn liseMowbray. UMatthew uthetha nomama
wakhe. Uthi: “Mama, ndicela ukusebenzisa ifowuni.”
7. X describes the picture in as much detail as possible. Ask “Yintoni le/Ziintoni ezi”
questions and ask some questions that require negative answers when X has run out of
things to say about the picture.

252
hlala; bona; iklasi; funda; idolophu; inkwenkwe; igama; namhlanje; khulu; izindlu
9. Formulate questions which would require the following answers:

a) Ndihlala eRhini.
b) Sitya izonka.
c) Abafazi bayalala.
d) Izinja zitya ukutya ngo-6.
a) tya abantwana inyama

b) utitshalakazi ibhodi bhala
c) idolophu hamba amakwenkwe
a) Umalume banabantwana athathu.

b) Oobhuti bafundi iincwadi.
c) Lo abantu bahamba idolophu.
d) Ndisika isonka imela.
e) Iikati silala.
253
Interview 6
1. Have a conversation with X about what he/she did during the vac.
2. Build a story about each other’s families. X says two or three sentences, Thembi says two
or three sentences. (Describe who is in the family, what they look like, etc).
3. Role-play a situation where Thembi asks X to go out, but X doesn’t want to go out.
Discuss where, when, why, etc. Could also try to make a date for another night.
through, then twice more fairly slowly.
Passage:
Utitshala usebenza esikolweni. Ufundisa abantwana. Iiprofesa zifundisa
eyunivesithi. Umsebenzi weeprofesa noweetitshala kukufundisa. Abanye abafundi
bafuna ukuba ngoosomashishini. Umsebenzi kasomashishini kukuthengisa.
Bathengisa izinto ngezinto. Abanye bathengisa iibhasi; abanye bathengisa iimoto,
iibhaki, izithuthuthu neelori. Abanye bathengisa iibhayisekile. Abanye baneeteksi.
The tree is green.

The girl is beautiful.
They live in a big house.
The dogs eat meat, but the horses eat grass.
The ugly dogs play in the brown houses.
The car doesn’t stay in the garage.
This is a red bicycle.
6. Translate this passage into English:
Ngamanye amaxesha ndiya elayibri, ndiye kuboleka iincwadi okanye ndifundele khona.
Ezinye iincwadi aziphumi elayibri. Ilayibri ivula imini yonke ivala ngo-10 ebusuku phakathi
evekini, kodwa ngoMgqibelo ivala ngo-12 emini; ayivuli ngeCawe. Bonke abafundi
banamakhadi. Ikhadi lomfundi linefoto yakhe. Abafundi bayafota bafumane amakhadi
ekuqaleni konyaka. Elayibri ukhupha ikhadi lakho ukuba ufuna ukuboleka incwadi,
uphume nayo. Zininzi iilayibri apha; zinkulu. Zonke zineencwadi ezininzi.
7. X describes the picture in as much detail as possible. Ask “Yintoni le/Ziintoni ezi”
questions and ask some questions that require negative answers when X has run out of
things to say about the picture.
ivenkile; thenga; bhala; nini; enza; utitshalakazi; hamba; uMgqibelo; hle; ninzi
254
9. Formulate questions which would require the following answers:

a) Ndiya edolophini ngoku.
b) Ndifuna isonka.
c) Sihamba yonke imihla.
d) Abafazi bahlamba iimpahla.
a) ihashe tya ingca

b) iimpahla abantwana thenga
c) umntwana incwadi bhala
a) Lo mntu isebenza.
b) Aba mthi luhlaza.
c) Abanta akafuni izonka esimhlophe.
d) La ihashe atya ingca iluhlaza.
e) Ubuso bakaThandi buhle.
255
APPENDIX D: NUMBER OF UTTERANCES IN THE DATA
Interview number 1 2 3 4 5 6 7 Task

Total
Ann
Conversation 22 10 20 6 16 17 20 111
Picture description 2 10 17 14 10 19 18 90
Translation 6 6 7 6 7 6 6 44
Sentence building 3 13 14 18 16 22 16 102
Ben
Conversation 25 18 23 14 13 20 18 131
Translation 6 0 7 7 7 7 7 41
Claire
Conversation 29 18 23 21 15 24 22 152
Translation 6 7 7 7 7 9 7 50
Pat
Conversation 26 16 20 30 17 19 18 146
Translation 6 7 6 7 7 7 7 47
Karen
Conversation 23 18 18 14 12 13 14 112
Translation 6 7 7 7 7 7 7 48
Sarah
Conversation 31 16 20 16 11 11 17 122
Translation 6 7 7 7 7 7 7 48
Interview Total 264 290 335 303 279 337 314 2122
256
APPENDIX E: COUNTING MORPHEMES
Utterance: Utitshalakazi wam nguB nguBulelwa andiyazi ifani yakhe

teacher my is Bulelwa I don’t know surname her
*olundiyamthanda yhu
that I her like whew
‘My teacher is Bulelwa. I don’t know her surname. I like her a lot.’
Morphemes: u+titshala+kazi wa+m

noun prefix+stem+noun suffix possessive concord+possessive stem
(nguB) ngu+Bulelwa
(hesitant form not counted) copulative formative+noun stem
a+ndi+y+az+i
negative formative+negative subject concord+linking consonant+root+terminating
vowel
i+fani ya+khe
noun prefix+stem possessive concord+possessive stem
o+lu+ndi+ya+m+thand+a yhu
qualificative formative+basic prefix+pronoun+verbal formative+object
concord+root+terminating vowel interjective
= 24 morphemes
257
APPENDIX F: XHOSA STRUCTURE

The Xhosa class system
Xhosa nouns are each assigned a grammatical gender which manifests itself as a class system.
Each noun is a member of a class and the class of the noun is marked by a class prefix (du Plessis
1978).
The form of an agreement morpheme is dependent on the class of the noun with which the
morpheme needs to agree1. Fifteen different noun classes are recognised in Xhosa, although, when
agreement is marked, only thirteen different forms of each agreement morpheme are used. Each
pair in classes 1 and 1(a), and classes 2 and 2(a) shares the same form of the morpheme when it
is brought into agreement with the noun. Effectively this means that the language learner has to
learn the function of the morpheme and up to thirteen different forms of the morpheme if it is to
be supplied correctly in a particular linguistic context.
Noun prefixes
A noun prefix always appears on a noun stem in Xhosa, although the initial vowel is dropped on
vocatives (e.g. bantwana “children”), after negatives (e.g. andifuni bantwana “I don’t want the
children”) and after demonstratives (e.g. lo tata “this father”). There are fourteen different noun
prefixes in Xhosa, as classes 1 and 3 share a common prefix. The prefix of the noun shows
whether the root noun to which it is attached is in the singular or plural form. Nouns that fall into
classes with odd numbers indicate the singular form and nouns that fall into the classes with even
numbers indicate the plural form (see the table overleaf). The plurals of nouns in class 11 are
found in class 10, while classes 14 and 15 have only one form as they are usually mass or abstract
nouns and infinitive forms respectively (which do not make singular/plural distinctions). The noun
prefix table over the page gives examples of nouns in different classes.
1
It should be noted that a study by Satyo (1998) found that L1 Xhosa speakers do not always
produce correct agreement forms. He cites examples of class 11 nouns which appear with class 5
agreement forms, where phonological similarities seem to influence morpheme choice.
258
Noun prefixes
Class number Prefix Example

1 um- umntwana
child
1a u- utata
father
2 aba- abantwana
children
2a oo- ootata
fathers
3 um- umlilo
fire
4 imi- imililo
fires
5 ili- i(li)xesha/ilifu
time/cloud
6 ama- amaxesha/amafu
times/clouds
7 isi- isitya
dish
8 izi- izitya
dishes
9 in-/i- indlu/ibhokwe
house/goat
10 izin-/ii- izindlu/iibhokwe
houses/goats
11 ulu- uluvo/u(lu)cango
opinion/door
14 ubu- ubuso
face
15 uku- ukutya
food
Adapted from Zotwana (1991) and du Plessis (1978)
Other morphemes
Many other morphemes agree with the class of the noun with which they are co-referenced.
259
Examples of morphemes which appear in the data and which take agreement forms are:
e) subject concords
f) negative subject concords
g) participials
h) demonstratives
i) copulatives
j) possessives
k) qualificatives
l) object concords
Any learner learning to use the morphemes listed above would need to learn multiple forms of the
morphemes, since they would need to know which form to use depending on the noun with which
the morpheme needs to be co-referenced.
The following morphemes which appear in the data do not agree with nouns, but usually have
single forms which are added to noun or verb stems, with or without coalescence of the vowels:
a) ya-
b) verbal extensions
c) auxiliaries
d) locatives
e) qualificative a-
f) associative copulative
g) nga-/na- “with” morphemes
h) additive conjunctions
i) njenga- (“like”)
j) absolute pronouns
k) pronominal forms for personal pronouns
For most of these morphemes, the learners need to learn only a single form and the function of
the form. They would also need to know the coalescence rules which regulate the sound changes
which occur when two vowels appear next to each other. For example, if the qualificative a- is
to be added to a qualificative form which begins with i-, the a- and the i- are coalesced to form
e- and the form which is added to the stem is therefore e-.
Example sentences
The following would be a typical Xhosa sentence using a variety of morphemes:
Abantwana abaninzi abazifundi iincwadi esikolweni.
Children that are many don’t read them the books at school.
‘Many children don’t read the books at school.’
In the above example, abantwana (children) is a noun with a class 2 noun prefix.
260
Abaninzi (that are many) consists of a qualificative a- morpheme followed by a relative concord,
ba-. The form ba- is derived from the class 2 noun abantwana (children), with which the adjective
stem -ninzi (many) is co-referenced.
The next word, abazifundi (they don’t read them), is a verb in the negative with an object
concord. It begins with a negative formative a- which is followed by the subject concord ba-. Ba-
is selected as the subject concord because it is co-referenced with abantwana (children), a class
2 noun. The zi- form is an object concord which is co-referenced with the object iincwadi
(books). The object concord in this case denotes definiteness. The verb stem ends in terminative
-i because verbs in the present tense which are also in the negative always end in -i.
The form esikolweni (at school) is a locative derived from the noun isikolo (school). The initial
vowel changes to e- and the suffix -ini is added to the stem. Coalescence of the final vowel -o and
the first vowel of the suffix -i, results in the change to we-.
261
APPENDIX G1: INCORRECT NOUN PREFIXES
The table indicates the number of times an incorrect noun prefix has been supplied by each of the
learners for each of the tasks investigated. The number of times that a noun prefix has been
incorrectly supplied for each task is presented in the Task Total column on the right-hand side of
the table, and the number of times that a noun prefix has been incorrectly supplied in each
interview is presented in the Int Total column at the bottom of the table. The percentages refer
to the percentage of incorrect suppliance as a percentage of the total number of obligatory uses
in each task or interview.
Int No 1 2 3 4 5 6 7 Task
Total
Ann
C2 1 (5%) 0 3 (25%) 0 1 (7%) 2 (20%) 4 (36%) 11 (14%)
PD 0 1 (4%) 0 0 0 0 0 1 (2%)
T 0 0 0 0 0 1 (11%) 0 1 (1%)
SB 0 0 3 (17%) 2 (15%) 1 (5%) 1 (6%) 0 7 (6%)
Ben
C 1 (3%) 0 1 (6%) 0 1 (5%) 4 (27%) 1 (6%) 8 (6%)
PD 1 (8%) 3 (7%) 1 (7%) 3 (11%) 3 (8%) 2 (25%) 0 13 (7%)
T 1 (20%) 0 1 (11%) 0 0 0 1 (11%) 3 (6%)
SB 0 0 1 (7%) 0 0 1 (8%) 0 2 (2%)
Claire
C 3 (7%) 5 (7%) 2 (7%) 1 (9%) 3 (6%) 1 (3%) 2 (9%) 17 (6%)
PD 6 (27%) 1 (3%) 0 1 (3%) 2 (5%) 1 (6%) 0 11 (6%)
T 0 0 0 1 (10%) 0 2 (20%) 0 3 (5%)
SB 0 1 (6%) 2 (14%) 0 0 0 0 3 (3%)
Pat
C 2 (5%) 3 (9%) 0 3 (11%) 5 (11%) 4 (24%) 3 (13%) 20 (9%)
PD 1 (6%) 6 (24%) 1 (4%) 2 (13%) 2 (7%) 2 (11%) 4 (17%) 18 (12%)
T 1 (20%) 0 0 1 (10%) 0 0 0 2 (3%)
SB 1 (25%) 3 (21%) 1 (6%) 1 (7%) 3 (14%) 0 0 9 (9%)
Karen
C 1 (2%) 7 (10%) 0 1 (7%) 1 (3%) 2 (6%) 0 12 (6%)
PD 1 (13%) 2 (11%) 0 0 0 1 (6%) 2 (12%) 6 (5%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Sarah
C 1 (4%) 3 (8%) 1 (5%) 1 (5%) 1 (3%) 0 0 7 (4%)
PD 1 (6%) 2 (5%) 0 2 (6%) 1 (3%) 0 1 (5%) 7 (4%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Int 22 (6%) 37 (6%) 17 (5%) 19 (5%) 24 (4%) 24 (6%) 18 (5%) 161 (5%)
Total
2
C = Conversation, PD = Picture description, T = Translation, SB = Sentence building and Sentence
manipulation
262
APPENDIX G2: NOUN PREFIX OMISSIONS
The table indicates the number of times a noun prefix has been omitted by each of the learners
for each of the tasks investigated. The number of times that a noun prefix has been omitted for
each task is presented in the Task Total column on the right-hand side of the table, and the
number of times that a noun prefix has been omitted in each interview is presented in the Int Total
column at the bottom of the table. The percentages refer to the percentage of omissions as a
percentage of the total number of obligatory uses in each task or interview.
Int No 1 2 3 4 5 6 7 Task
Total
Ann
C 2 (9%) 0 1 (8%) 0 1 (7%) 0 0 4 (5%)
PD 1 (14%) 1 (4%) 2 (17%) 0 3 (16%) 0 0 7 (6%)
T 0 0 0 0 0 1 (11%) 0 1 (2%)
SB 0 0 0 0 0 0 0 0
Ben
C 8 (23%) 1 (4%) 0 0 3 (14%) 3 (20%) 0 15 (11%)
PD 3 (23%) 0 1 (7%) 0 0 2 (7%) 2 (9%) 8 (4%)
T 0 0 0 0 1 (10%) 0 0 1 (2%)
SB 0 0 0 1 (7%) 0 0 0 1 (1%)
Claire
C 3 (7%) 8 (11%) 1 (4%) 2 (18%) 5 (9%) 4 (10%) 4 (18%) 27 (10%)
PD 2 (9%) 9 (25%) 0 1 (3%) 4 (11%) 2 (11%) 0 18 (9%)
T 1 (17%) 1 (10%) 0 0 2 (18%) 0 0 4 (6%)
SB 3 (50%) 0 1 (7%) 2 (14%) 2 (13%) 0 0 8 (9%)
Pat
C 8 (19%) 0 0 1 (4%) 2 (5%) 0 0 11 (5%)
PD 0 0 0 0 2 (7%) 0 1 (4%) 3 (2%)
T 0 0 0 0 0 0 0 0
SB 0 1 (7%) 0 0 1 (5%) 0 0 2 (2%)
Karen
C 3 (7%) 3 (4%) 0 0 1 (3%) 1 (5%) 0 8 (4%)
PD 0 0 1 (13%) 0 1 (16%) 0 0 2 (2%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Sarah
C 1 (4%) 0 0 3 (14%) 5 (15%) 2 (8%) 3 (15%) 14 (8%)
PD 0 2 (5%) 1 (11%) 0 4 (11%) 0 0 7 (4%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Int 35 (10%) 26 (5%) 8 (2%) 10 (3%) 37 (7%) 15 (4%) 10 (3%) 141 (5%)
Total
263
APPENDIX G3: INCORRECT SUBJECT CONCORDS
The table below indicates the number of times an incorrect subject concord has been supplied by
each of the learners for each of the tasks investigated. The number of times that a subject concord
has been incorrectly supplied for each task is presented in the Task Total column on the right-hand
side of the table, and the number of times that a subject concord has been incorrectly supplied in
each interview is presented in the Int Total column at the bottom of the table. The percentages
refer to the percentage of incorrect suppliance as a percentage of the total number of obligatory
uses in each task or interview.
Int No 1 2 3 4 5 6 7 Task Total
Ann
C 2 (12%) 1 (10%) 1 (6%) 0 3 (60%) 0 2 (11%) 9 (11%)
PD 0 1 (17%) 5 (71%) 1 (50%) 2 (40%) 0 1 (33%) 10 (43%)
T 1 (14%) 1 (25%) 0 3 (50%) 3 (38%) 3 (75%) 1 (50%) 12 (34%)
SB 0 4 (33%) 3 (30%) 1 (7%) 3 (20%) 0 1 (9%) 12 (15%)
Ben
C 4 (16%) 1 (3%) 6 (18%) 6 (32%) 7 (28%) 7 (25%) 4 (13%) 35 (18%)
PD 4 (36%) 8 (22%) 0 0 4 (19%) 2 (12%) 1 (8%) 19 (15%)
T 0 0 0 3 (50%) 0 0 0 3 (10%)
SB 0 1 (6%) 1 (11%) 0 4 (25%) 0 1 (17%) 7 (9%)
Claire
C 11 (14%) 17 (24%) 19 (34%) 6 (21%) 8 (14%) 11 (17%) 21 (36%) 93 (22%)
PD 4 (17%) 6 (16%) 3 (43%) 10 (40%) 5 (13%) 10 (50%) 9 (32%) 47 (26%)
T 3 (38%) 2 (29%) 1 (14%) 1 (20%) 2 (22%) 1 (20%) 0 10 (24%)
SB 0 6 (35%) 3 (23%) 3 (19%) 3 (16%) 1 (7%) 2 (17%) 18 (18%)
Pat
C 2 (6%) 2 (5%) 16 (33%) 7 (15%) 7 (13%) 8 (31%) 11 (30%) 53 (19%)
PD 2 (33%) 9 (38%) 6 (33%) 0 1 (6%) 3 (33%) 5 (36%) 26 (25%)
T 1 (17%) 1 (13%) 0 1 (20%) 1 (14%) 3 (75%) 0 7 (19%)
SB 0 0 1 (9%) 3 (21%) 2 (11%) 2 (13%) 0 8 (8%)
Karen
C 3 (7%) 6 (7%) 4 (12%) 0 4 (8%) 1 (3%) 0 18 (6%)
PD 0 7 (39%) 1 (17%) 5 (28%) 4 (20%) 0 3 (20%) 20 (20%)
T 0 2 (29%) 0 0 0 0 0 2 (5%)
SB 0 0 1 (6%) 0 2 (10%) 1 (5%) 0 4 (3%)
Sarah
C 1 (3%) 0 2 (4%) 0 3 (8%) 6 (14%) 4 (10%) 16 (6%)
PD 0 10 (30%) 0 2 (9%) 2 (8%) 2 (13%) 4 (20%) 20 (15%)
T 1 (11%) 0 0 0 1 (14%) 1 (25%) 0 3 (7%)
SB 0 0 1 (6%) 1 (7%) 0 0 1 (6%) 3 (3%)
Int 39 (10%) 85 (15%) 74 (20%) 53 (14%) 71 (14%) 62 (16%) 71 (18%) 455 (15%)
Total
264
APPENDIX G4: SUBJECT CONCORD OMISSIONS
The table indicates the number of times a subject concord has been omitted by each of the learners
for each of the tasks investigated. The number of times that a subject concord has been omitted
for each task is presented in the Task Total column on the right-hand side of the table, and the
number of times that a subject concord has been omitted in each interview is presented in the Int
Total column at the bottom of the table. The percentages refer to the percentage of omissions
as a percentage of the total number of obligatory uses in each task or interview.
Ann
C 4 (24%) 3 (30%) 3 (19%) 0 0 0 1 (6%) 11 (13%)
PD 0 0 0 0 1 (20%) 0 0 1 (4%)
T 0 0 1 (50%) 3 (50%) 0 0 0 4 (11%)
SB 0 0 0 0 1 (7%) 0 1 (9%) 2 (2%)
Ben
C 3 (12%) 3 (11%) 4 (12%) 1 (5%) 4 (16%) 1 (4%) 2 (7%) 18 (9%)
PD 0 8 (22%) 0 4 (21%) 2 (10%) 2 (12%) 2 (12%) 18 (14%)
T 0 0 0 1 (17%) 0 0 0 1 (3%)
SB 0 1 (6%) 0 0 2 (13%) 0 0 3 (4%)
Claire
C 6 (8%) 2 (3%) 0 2 (7%) 6 (10%) 3 (5%) 7 (12%) 26 (6%)
PD 0 2 (5%) 0 0 10 (26%) 1 (5%) 4 (14%) 17 (9%)
T 0 0 0 0 1(11%) 0 0 1 (2%)
SB 0 1 (6%) 1 (8%) 0 0 1 (7%) 0 3 (3%)
Pat
C 0 3 (8%) 2 (4%) 0 2 (4%) 0 0 7 (3%)
PD 0 1 (4%) 1 (6%) 0 2 (11%) 0 0 4 (4%)
T 0 0 0 0 0 0 0 0
SB 0 1 (5%) 0 0 0 0 1 (8%) 2 (2%)
Karen
C 2 (4%) 6 (7%) 1 (3%) 1 (4%) 1 (2%) 0 0 11 (4%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 1 (6%) 0 0 0 3 (18%) 4 (3%)
Sarah
C 0 0 2 (4%) 0 1 (3%) 1 (2%) 0 4 (1%)
PD 0 0 0 0 0 0 2 (10%) 2 (2%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Int 15 (4%) 31 (6%) 16 (4%) 12 (3%) 33 (6%) 9 (2%) 23 (6%) 139 (5%)
Total
265
APPENDIX G5: SUPERFLUOUS USE OF -YA-
The table indicates the number of times that -ya- is used when it should not be for each task. The
number of times that the -ya- form is supplied superfluously is presented in the Task Total column
on the right-hand side of the table, and the number of times that -ya- is used superfluously in each
to the percentage of overuse as a percentage of the total number of obligatory uses in each task
or interview.
Ann
C 0 0 0 0 0 0 0 0
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 2 (100%) 0 0 2 (100%)
SB 0 0 0 0 1 (100%) 0 0 1 (10%)
Ben
C 3 (75%) 3 (100%) 2 (50%) 0 1 (50%) 1 (100%) 0 10 (71%)
PD 0 4 (80%) 0 2 (67%) 1 (100%) 1 (100%) 0 8 (67%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Claire
C 29 (85%) 20 (87%) 12 (71%) 6 (100%) 14 (93%) 18 (82%) 13 (81%) 112 (84%)
PD 4 (80%) 16 (67%) 4 (100%) 11 (79%) 16 (100%) 1 (50%) 6 (50%) 58 (82%)
T 3 (100%) 4 (80%) 1 (100%) 3 (100%) 5 (83%) 3 (100%) 0 19 (90%)
SB 2 (100%) 8 (89%) 7 (78%) 8 (80%) 7 (100%) 3 (100%) 4 (50%) 39 (89%)
Pat
C 10 (71%) 11 (100%) 18 (67%) 9 (100%) 19 (100%) 4 (100%) 10 (91%) 81 (85%)
PD 2 (100%) 15 (94%) 3 (50%) 4 (100%) 10 (77%) 3 (100%) 6 (67%) 43 (81%)
T 1 (100%) 4 (100%) 3 (100%) 9 (90%) 3 (75%) 1 (100%) 2 (100%) 23 (91%)
SB 4 (100%) 7 (100%) 1 (100%) 4 (100%) 9 (90%) 2 (100%) 4 (100%) 31 (97%)
Karen
C 1 (100%) 0 2 (100%) 2 (100%) 0 0 0 5 (71%)
PD 0 0 0 0 1 (50%) 0 0 1 (20%)
T 2 (100%) 0 0 0 0 0 0 2 (50%)
SB 2 (67%) 0 0 0 0 0 0 2 (33%)
Sarah
C 0 1 (100%) 0 0 0 0 0 1 (50%)
PD 0 1 (50%) 0 0 0 0 0 1 (50%)
T 0 0 1 (100%) 0 0 0 0 1 (33%)
SB 0 0 0 0 0 0 0 0
Int 63 (76%) 94 (81%) 54 (65%) 58 (84%) 89 (88%) 37 (88%) 45 (69%) 440 (79%)
Total
266
APPENDIX G6: INCORRECT NEGATIVE SUBJECT CONCORDS
The table indicates the occurrence of incorrect negative subject concords in the data. The number
of times that the incorrect negative subject concord is supplied for each task is presented in the
Task Total column on the right-hand side of the table, and the number of times that the incorrect
negative subject concord is supplied in each interview is presented in the Int Total column at the
bottom of the table. The percentages refer to the percentage of incorrect uses as a percentage of
the total number of obligatory uses in each task or interview.
Ann
C 0 0 0 2 (100%) 0 0 0 2 (50%)
PD 0 0 6 (100%) 6 (100%) 0 4 (100%) 4 (100%) 20 (100%)
T 0 0 3 (100%) 1 (50%) 0 0 3 (75%) 7 (88%)
SB 0 0 4 (100%) 3 (100%) 0 3 (100%) 5 (83%) 15 (79%)
Ben
C 1 (100%) 0 0 0 0 0 1 (100%) 2 (67%)
PD 0 0 6 (60%) 2 (50%) 0 0 2 (50%) 10 (53%)
T 0 0 2 (50%) 0 0 0 1 (50%) 3 (33%)
SB 0 0 1 (14%) 0 0 0 0 1 (5%)
Claire
C 1 (9%) 2 (22%) 0 2 (18%) 1 (67%) 0 3 (18%) 9 (15%)
PD 0 0 2 (33%) 2 (67%) 0 1 (17%) 2 (40%) 7 (16%)
T 0 0 2 (29%) 1 (10%) 0 0 0 3 (20%)
SB 0 1 (50%) 1 (17%) 1 (33%) 0 1 (33%) 1 (17%) 5 (25%)
Pat
C 0 0 2 (100%) 0 0 2 (67%) 2 (50%) 6 (50%)
PD 1 (100%) 0 1 (25%) 4 (100%) 0 0 2 (67%) 8 (57%)
T 0 0 1 (33%) 1 (50%) 0 0 1 (100%) 3 (50%)
SB 0 0 4 (50%) 3 (75%) 0 1 (33%) 1 (20%) 9 (45%)
Karen
C 0 0 0 0 2 (67%) 0 0 2 (22%)
PD 0 0 2 (25%) 0 0 0 1 (100%) 3 (18%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Sarah
C 0 0 0 0 2 (67%) 0 0 2 (9%)
PD 0 1 (100%) 4 (67%) 0 0 2 (33%) 2 (67%) 9 (47%)
T 0 0 1 (33%) 0 0 0 0 1 (11%)
SB 0 0 1 (40%) 0 0 1 (33%) 3 (50%) 5 (24%)
Int 3 (11%) 4 (18%) 43 (38%) 28 (33%) 5 (42%) 15 (22%) 34 (32%) 132 (31%)
Total
267
APPENDIX G7: LACK OF I-ENDING ON NEGATIVES
The table indicates when the final i-ending is not used for the negative verb. The number of times
that no i-ending is supplied for each task is presented in the Task Total column on the right-hand
side of the table, and the number of times that no i-ending is supplied in each interview is
presented in the Int Total column at the bottom of the table. The percentages refer to the
percentage of omissions as a percentage of the total number of obligatory uses in each task or
interview.
Ann
C 0 0 0 0 0 1 (100%) 0 1 (25%)
PD 0 0 2 (33%) 0 0 2 (50%) 0 4 (20%)
T 0 0 0 0 0 0 1 (33%) 1 (13%)
SB 0 0 0 0 0 0 0 0
Ben
C 0 0 0 0 0 0 0 0
PD 0 0 3 (30%) 0 0 0 1 (25%) 4 (21%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Claire
C 0 0 0 0 0 1 (11%) 4 (24%) 5 (8%)
PD 1 (11%) 1 (33%) 0 0 0 0 1 (14%) 3 (7%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Pat
C 0 2 (100%) 2 (100%) 2 (50%) 0 0 1 (100%) 7 (58%)
PD 0 0 6 (100%) 0 0 0 2 (67%) 8 (57%)
T 0 0 2 (67%) 1 (50%) 0 1 (100%) 2 (67%) 6 (55%)
SB 0 0 2 (50%) 0 0 1 (33%) 2 (40%) 5 (25%)
Karen
C 0 0 0 0 0 0 0 0
PD 0 0 2 (25%) 0 0 0 0 2 (12%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Sarah
C 0 0 0 0 0 0 0 0
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Int Total 1 (4%) 3 (14%) 19 (17%) 3 (4%) 0 6 (9%) 14 (13%) 46 (11%)

268
APPENDIX G8: INCORRECT COPULATIVE
The table indicates use of the incorrect form of the copulative. The number of times that the
incorrect form of the copulative has been supplied for each task is presented in the Task Total
column on the right-hand side of the table, and the number of times that the incorrect form of the
copulative is supplied in each interview is presented in the Int Total column at the bottom of the
table. The percentages refer to the percentage of incorrect uses as a percentage of the total
number of obligatory uses in each task or interview.
Ann
C 0 0 0 0 0 1 (11%) 0 1 (3%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Ben
C 0 0 0 0 0 0 0 0
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Claire
C 0 4 (29%) 0 0 1 (11%) 0 0 5 (13%)
PD 1 (25%) 0 0 0 0 0 0 1 (6%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Pat
C 0 0 0 0 0 1 (50%) 0 1 (4%)
PD 0 0 1 (33%) 0 0 0 0 1 (9%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Karen
C 2 (14%) 0 0 0 0 0 0 2 (5%)
PD 0 0 0 0 1 (25%) 0 0 1 (7%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Sarah
C 0 0 0 0 0 0 0 0
PD 0 1 (33%) 0 0 1 (50%) 0 0 2 (20%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Int Total 3 (6%) 5 (8%) 1 (6%) 0 3 (5%) 2 (5%) 0 14 (5%)

269
APPENDIX G9: COPULATIVE OMISSIONS
The table indicates the number of times the copulative is omitted in the data for each task and
learner. The number of times that the copulative is omitted for each task is presented in the Task
Total column on the right-hand side of the table, and the number of times that the copulative is
omitted in each interview is presented in the Int Total column at the bottom of the table. The
percentages refer to the percentage of omissions as a percentage of the total number of obligatory
Int No 1 2 3 4 5 6 7 Task
Total
Ann
C 0 4 (100%) 1 (33%) 0 4 (36%) 3 (33%) 0 12 (38%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 1 (100%) 0 0 0 0 0 1 (100%)
Ben
C 2 (67%) 1 (33%) 0 1 (100%) 0 2 (100%) 0 6 (43%)
PD 0 1 (50%) 2 (67%) 0 0 0 0 3 (60%)
T 0 0 0 0 0 1 (100%) 0 1 (50%)
SB 0 0 1 (100%) 1 (50%) 0 0 0 2 (67%)
Claire
C 0 2 (14%) 0 0 2 (22%) 1 (14%) 0 5 (13%)
PD 0 0 1 (100%) 0 0 0 0 1 (6%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Pat
C 3 (43%) 2 (29%) 2 (100%) 2 (100%) 4 (80%) 1 (50%) 0 14 (54%)
PD 0 1 (100%) 1 (33%) 2 (100%) 0 0 1 (50%) 5 (45%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 1 (100%) 2 (100%) 0 0 3 (60%)
Karen
C 0 3 (20%) 0 0 1 (17%) 0 0 4 (10%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Sarah
C 1 (14%) 1 (17%) 0 0 2 (25%) 0 0 4 (15%)
PD 0 0 1 (100%) 0 0 2 (67%) 0 3 (30%)
T 2 (100%) 0 0 0 0 1 (100%) 0 3 (100%)
SB 0 0 0 0 0 0 0 0
Int Total 8 (16%) 16 (25%) 9 (53%) 7 (50%) 15 (25%) 11 (26%) 1 (9%) 67 (26%)
270
APPENDIX G10: LOCATIVE OMISSIONS
The table indicates the number of times that locatives have been omitted for each task. The total
for each task is presented in the Task Total column on the right-hand side of the table, and the
number of times that the locative has been omitted in each interview is presented in the Int Total
column at the bottom of the table. The percentages refer to the percentage of omissions as a
percentage of the total number of obligatory uses in each task or interview.
Ann
C 0 3 (50%) 0 0 0 0 0 3 (11%)
PD 0 0 0 0 0 1 (33%) 1 (33%) 2 (25%)
T 0 0 1 (50%) 0 0 0 0 1 (8%)
SB 0 0 0 1 (20%) 1 (20%) 0 0 2 (8%)
Ben
C 0 2 (13%) 0 1 (20%) 0 0 0 3 (4%)
PD 0 0 0 1 (17%) 1 (20%) 0 0 2 (7%)
T 0 0 0 0 0 0 0 0
SB 0 1 (14%) 0 0 0 0 0 1 (3%)
Claire
C 2 (7%) 1 (5%) 1 (8%) 0 0 0 0 4 (3%)
PD 0 0 0 2 (20%) 0 2 (20%) 0 4 (11%)
T 2 (33%) 0 0 0 0 0 0 2 (13%)
SB 0 0 0 2 (33%) 0 1 (33%) 0 3 (12%)
Pat
C 1 (14%) 1 (14%) 1 (9%) 3 (23%) 0 2 (25%) 2 (20%) 10 (14%)
PD 2 (100%) 1 (50%) 0 0 0 0 0 3 (20%)
T 2 (40%) 0 0 0 0 1 (50%) 0 3 (20%)
SB 0 0 2 (50%) 2 (50%) 1 (20%) 2 (67%) 1 (20%) 8 (29%)
Karen
C 1 (5%) 0 1 (13%) 0 1 (4%) 0 0 3 (3%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 1 (100%) 0 0 0 0 0 0 1 (3%)
Sarah
C 1 (4%) 0 1 (7%) 0 0 2 (15%) 1 (11%) 5 (5%)
PD 0 0 0 0 0 0 1 (25%) 1 (3%)
T 0 0 0 0 0 0 0 0
SB 0 1 (13%) 0 0 0 0 0 1 (2%)
Int Total 12 (8%) 10 (6%) 7 (7%) 12 (11%) 4 (3%) 11 (8%) 6 (5%) 62 (7%)
271
APPENDIX G11: INCORRECT LOCATIVES
The table indicates the number of times that a locative has been incorrectly supplied for each task.
The total number of incorrectly supplied forms is presented in the Task Total column on the right-
hand side of the table, and the number of times that the locative has been incorrectly supplied in
each interview is presented in the Int Total column at the bottom of the table. The percentages
refer to the percentage of incorrect forms as a percentage of the total number of obligatory uses
in each task or interview.
Ann
C 0 0 0 0 1 (25%) 0 0 1 (4%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 1 (33%) 1 (50%) 2 (17%)
SB 0 0 0 0 0 0 0 0
Ben
C 0 0 0 0 0 0 0 0
PD 0 0 1 (100%) 1 (20%) 0 0 0 2 (7%)
T 1 (17%) 0 1 (50%) 0 0 0 0 2 (13%)
SB 0 0 1 (20%) 0 0 0 0 1 (3%)
Claire
C 2 (7%) 1 (5%) 0 0 1 (4%) 0 0 4 (3%)
PD 0 1 (17%) 0 0 0 1 (6%) 0 2 (6%)
T 1 (17%) 0 0 0 0 0 0 1 (7%)
SB 0 0 0 0 0 0 0 0
Pat
C 0 0 1 (9%) 0 0 0 1 (10%) 2 (3%)
PD 0 0 0 0 0 1 (50%) 0 1 (7%)
T 0 0 1 (50%) 0 0 0 1 (33%) 2 (13%)
SB 0 0 0 0 0 0 0 0
Karen
C 0 0 0 0 0 0 0 0
PD 1 (25%) 0 0 0 0 0 0 1 (8%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Sarah
C 0 0 0 2 (17%) 0 0 1 (11%) 3 (3%)
PD 0 2 (25%) 0 0 1 (20%) 1 (14%) 0 4 (14%)
T 0 0 0 0 0 0 0 0
SB 0 1 (13%) 1 (13%) 0 0 0 0 2 (5%)
Int Total 5 (3%) 5 (3%) 6 (6%) 3 (3%) 3 (2%) 4 (3%) 4 (3%) 30 (3%)
272
APPENDIX G12: ONLY INITIAL PART OF LOCATIVE
The table indicates the number of times that only the initial part of the locative has been supplied
for each task. The total is presented in the Task Total column on the right-hand side of the table,
and the number of times that only the initial part of the locative is supplied in each interview is
percentage of only initial suppliance as a percentage of the total number of obligatory uses in each
task or interview.
Int No 1 2 3 4 5 6 7 Task
Total
Ann
C 0 0 0 0 0 0 0 0
PD 0 0 0 0 1 (100%) 1 (33%) 0 2 (25%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 1 (25%) 0 1 (4%)
Ben
C 2 (17%) 2 (13%) 0 0 2 (25%) 0 1 (7%) 7 (9%)
PD 0 1 (14%) 0 1 (17%) 1 (20%) 1 (33%) 0 4 (14%)
T 0 0 1 (50%) 0 0 3 (100%) 2 (40%) 6 (38%)
SB 1 (3%) 0 0 2 (33%) 0 0 0 3 (9%)
Claire
C 1 (3%) 0 0 0 0 0 0 1 (1%)
PD 0 0 1 (100%) 1 (10%) 1 (33%) 3 (30%) 1 (33%) 7 (19%)
T 0 0 0 0 0 1 (33%) 0 1 (7%)
SB 0 0 1 (25%) 1 (17%) 1 (25%) 0 1 (20%) 4 (16%)
Pat
C 2 (29%) 1 (14%) 0 0 0 0 0 3 (4%)
PD 0 0 1 (33%) 0 2 (100%) 1 (50%) 0 4 (27%)
T 0 0 0 0 1 (100%) 0 0 1 (7%)
SB 0 0 0 1 (25%) 1 (20%) 1 (33%) 2 (20%) 5 (18%)
Karen
C 0 1 (3%) 0 0 0 0 0 1 (1%)
PD 0 0 0 1 (17%) 0 0 0 1 (8%)
T 1 (20%) 0 0 0 0 0 0 1 (6%)
SB 0 0 0 0 0 0 0 0
Sarah
C 1 (4%) 0 0 1 (8%) 0 0 0 2 (2%)
PD 0 1 (14%) 0 0 0 2 (29%) 0 3 (10%)
T 0 0 0 0 0 0 0 0
SB 1 (100%) 0 0 0 0 0 1 (20%) 2 (5%)
Int Total 9 (6%) 6 (4%) 4 (4%) 8 (7%) 10 (7%) 14 (10%) 8 (7%) 59 (6%)
273
APPENDIX G13: ONLY FINAL PART OF LOCATIVE
The table indicates the number of times that only the final part of the locative has been incorrectly
supplied for each task. The total is presented in the Task Total column on the right-hand side of
the table, and the number of times that only the final part of the locative is supplied in each
to the percentage of suppliance of final parts of the locative as a percentage of the total number
of obligatory uses in each task or interview.
Ann
C 0 0 0 0 0 0 0 0
PD 0 0 0 0 0 0 1 (50%) 1 (13%)
T 0 0 0 0 0 1 (33%) 0 1 (8%)
SB 0 0 0 2 (40%) 0 1 (25%) 0 3 (12%)
Ben
C 0 0 0 0 0 0 0 0
PD 0 0 0 0 0 1 (33%) 0 1 (3%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 1 (17%) 0 0 0 1 (3%)
Claire
C 1 (3%) 1 (5%) 0 0 2 (9%) 1 (6%) 0 5 (4%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 1 (50%) 1 (7%)
SB 0 0 0 0 0 0 0 0
Pat
C 0 0 0 0 1 (6%) 0 0 1 (1%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Karen
C 0 0 0 0 0 0 0 0
PD 0 0 0 0 0 0 0 0
T 0 0 1 (50%) 0 0 0 0 1 (6%)
SB 0 0 0 0 0 0 0 0
Sarah
C 0 0 0 0 0 0 0 0
PD 0 0 0 1 (25%) 0 0 0 1 (3%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Int Total 1 (1%) 1 (1%) 1 (1%) 4 (4%) 3 (2%) 4 (3%) 2 (2%) 16 (2%)
274
APPENDIX G14: INCORRECT DEMONSTRATIVES
The table indicates the use of incorrect demonstrative concords in the data. The number of times
that the demonstrative is incorrectly supplied for each task is presented in the Task Total column
on the right-hand side of the table, and the number of times that the demonstrative is incorrectly
supplied in each interview is presented in the Int Total column at the bottom of the table. The
percentages refer to the percentage of incorrect forms as a percentage of the total number of
obligatory uses in each task or interview.
Int No 1 2 3 4 5 6 7 Task
Total
Ann
C 0 0 0 0 0 0 0 0
PD 0 0 1 (100%) 0 0 0 0 1 (25%)
T 0 0 4 (100%) 0 0 0 0 4 (100%)
SB 0 0 0 0 0 0 0 0
Ben
C 1 (50%) 1 (13%) 0 0 1 (33%) 1 (50%) 0 4 (21%)
PD 0 0 0 0 0 0 0 0
T 1 (50%) 0 4 (67%) 3 (100%) 0 0 4 (100%) 12 (92%)
SB 0 0 0 1 (100%) 0 0 0 1 (33%)
Claire
C 1 (50%) 2 (13%) 1 (20%) 0 0 0 1 (100%) 5 (15%)
PD 1 (50%) 5 (56%) 0 3 (100%) 0 0 5 (42%) 14 (39%)
T 1 (100%) 0 1 (50%) 1 (100%) 0 0 3 (60%) 6 (55%)
SB 0 0 0 1 (100%) 0 0 1 (100%) 2 (100%)
Pat
C 0 1 (100%) 1 (50%) 3 (100%) 1 (20%) 0 0 6 (46%)
PD 0 0 0 1 (100%) 2 (40%) 0 0 3 (12%)
T 1 (100%) 0 3 (60%) 1 (25%) 0 1 (100%) 2 (50%) 8 (67%)
SB 0 0 0 1 (100%) 0 0 0 1 (50%)
Karen
C 0 1 (14%) 0 0 0 1 (50%) 0 2 (15%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 1 (33%) 0 0 0 1 (6%)
SB 0 0 0 0 0 0 0 0
Sarah
C 1 (100%) 1 (50%) 0 0 0 0 0 2 (67%)
PD 0 0 0 1 (25%) 0 0 0 1 (11%)
T 0 0 0 1 (25%) 0 0 0 1 (8%)
SB 0 0 0 0 0 0 0 0
Int Total 7 (44%) 11 (20%) 15 (27%) 18 (49%) 4 (11%) 3 (17%) 16 (25%) 74 (26%)
275
APPENDIX G15: INCORRECT POSSESSIVE CONCORDS
The table presents the number of times that a possessive concord has been incorrectly supplied.
The total for each task is presented in the Task Total column on the right-hand side of the table,
and the number of times that a possessive concord has been incorrectly supplied in each interview
is presented in the Int Total column at the bottom of the table. The percentages refer to the
percentage of incorrect forms as a percentage of the total number of obligatory uses in each task
or interview.
Ann
C 0 0 0 0 8 (73%) 0 0 8 (62%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Ben
C 1 (17%) 1 (9%) 2 (25%) 2 (40%) 1 (13%) 0 0 7 (18%)
PD 2 (50%) 0 0 0 0 0 0 2 (29%)
T 0 0 0 0 0 0 0 0
SB 0 0 2 (50%) 0 0 0 0 2 (18%)
Claire
C 7 (47%) 15 (45%) 2 (22%) 2 (50%) 7 (35%) 10 (43%) 3 (60%) 46 (42%)
PD 0 1 (33%) 0 0 0 0 0 1 (9%)
T 0 1 (33%) 0 0 0 0 0 1 (17%)
SB 0 0 0 0 0 0 0 0
Pat
C 3 (30%) 1 (8%) 0 3 (23%) 4 (22%) 7 (41%) 3 (75%) 21 (28%)
PD 0 0 1 (100%) 0 2 (100%) 0 0 3 (75%)
T 0 0 0 0 0 0 0 0
SB 0 1 (33%) 0 1 (50%) 0 1 (50%) 0 3 (27%)
Karen
C 2 (15%) 5 (14%) 0 0 2 (20%) 0 0 9 (11%)
PD 0 4 (67%) 0 1 (33%) 5 (83%) 0 2 (100%) 12 (71%)
T 0 0 0 0 0 0 0 0
SB 1 (33%) 0 0 0 0 0 0 1 (6%)
Sarah
C 2 (20%) 3 (21%) 1 (50%) 1 (25%) 2 (18%) 0 0 9 (16%)
PD 0 2 (100%) 0 3 (100%) 0 0 0 5 (83%)
T 0 0 0 0 0 0 0 0
SB 0 1 (25%) 1 (33%) 0 0 0 0 2 (15%)
Int 18 (25%) 35 (26%) 9 (26%) 13 (28%) 31 (29%) 18 (25%) 8 (27%) 132 (27%)
Total
276
APPENDIX G16: POSSESSIVE CONCORD OMISSION
The table indicates the number of times that a possessive concord has been omitted for each task.
The total number of omissions is presented in the Task Total column on the right-hand side of the
table, and the number of times that a possessive concord has been omitted in each interview is
percentage of omissions as a percentage of the total number of obligatory uses in each task or
interview.
Ann
C 0 0 0 0 0 0 0 0
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Ben
C 0 0 1 (8%) 0 0 0 0 1 (3%)
PD 0 1 (7%) 0 0 0 0 0 1 (14%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Claire
C 0 1 (3%) 1 (11%) 1 (25%) 0 0 0 3 (3%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Pat
C 3 (30%) 2 (17%) 0 4 (33%) 6 (33%) 0 1 (25%) 16 (21%)
PD 0 0 0 0 0 1 (100%) 0 1 (25%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 1 (50%) 0 0 2 (67%) 3 (27%)
Karen
C 0 0 0 1 (50%) 1 (10%) 1 (8%) 0 3 (4%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 2 (50%) 2 (12%)
Sarah
C 0 0 0 0 0 0 1 (7%) 1 (2%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Int Total 3 (4%) 4 (3%) 2 (6%) 7 (15%) 7 (6%) 2 (3%) 6 (20%) 31 (6%)
277
APPENDIX G17: INCORRECT QUALIFICATIVE CONCORDS
The table indicates the occurrence of incorrect qualificative concords in the data. The number of
times that the qualificative concord is incorrectly supplied for each task is presented in the Task
Total column on the right-hand side of the table, and the number of times that the qualificative
concord is incorrectly supplied in each interview is presented in the Int Total column at the
bottom of the table. The percentages refer to the percentage of incorrect forms as a percentage
of the total number of obligatory uses in each task or interview.
Int No 1 2 3 4 5 6 7 Task
Total
Ann
C 0 1 (100%) 1 (50%) 0 1 (33%) 0 1 (50%) 4 (44%)
PD 0 0 0 0 0 2 (67%) 0 2 (67%)
T 0 0 0 0 0 1 (100%) 0 1 (100%)
SB 0 0 0 0 0 0 0 0
Ben
C 0 1 (33%) 0 2 (50%) 0 1 (50%) 2 (67%) 6 (29%)
PD 0 0 0 0 0 3 (75%) 0 3 (60%)
T 0 0 0 0 0 1 (100%) 0 1 (100%)
SB 0 0 0 1 (100%) 0 1 (50%) 1 (100%) 3 (44%)
Claire
C 2 (50%) 0 0 0 2 (40%) 0 3 (60%) 7 (30%)
PD 0 0 0 0 2 (40%) 0 0 2 (11%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 1 (100%) 0 1 (25%)
Pat
C 0 0 0 2 (67%) 2 (100%) 1 (100%) 0 5 (50%)
PD 0 1 (100%) 0 1 (50%) 2 (100%) 0 1 (33%) 5 (45%)
T 0 0 0 0 0 1 (100%) 0 1 (100%)
SB 0 0 0 0 1 (100%) 1 (100%) 1 (100%) 3 (75%)
Karen
C 0 0 0 1 (100%) 0 1 (50%) 1 (50%) 3 (33%)
PD 0 0 0 0 0 0 1 (100%) 1 (25%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Sarah
C 0 1 (33%) 1 (100%) 1 (50%) 1 (100%) 0 0 4 (42%)
PD 0 1 (33%) 0 0 1 (50%) 0 0 2 (33%)
T 0 0 0 0 0 1 (50%) 0 1 (50%)
SB 0 1 (100%) 0 0 0 1 (50%) 1 (100%) 3 (67%)
Int Total 2 (18%) 6 (44%) 2 (29%) 8 (62%) 12 (55%) 16 (47%) 12 (55%) 58 (35%)
278
APPENDIX G18: QUALIFICATIVE CONCORD OMISSIONS
The table indicates the number of omissions of the qualificative concord in the data. The number
of times that the qualificative concord is omitted for each task is presented in the Task Total
column on the right-hand side of the table, and the number of times that the qualificative concord
is omitted in each interview is presented in the Int Total column at the bottom of the table. The
percentages refer to the percentage of omissions as a percentage of the total number of obligatory
Ann
C 0 0 1 (50%) 0 0 1 (50%) 0 2 (22%)
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Ben
C 1 (100%) 0 1 (25%) 0 1 (33%) 0 0 3 (14%)
PD 0 0 0 0 0 1 (25%) 1 (100%) 2 (40%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Claire
C 0 1 (50%) 0 0 0 1 (50%) 0 2 (13%)
PD 0 0 0 0 0 2 (67%) 0 2 (11%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 1 (50%) 0 0 0 1 (14%)
Pat
C 0 0 0 0 0 0 0 0
PD 0 0 0 0 0 0 1 (33%) 1 (9%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Karen
C 0 0 0 0 0 0 0 0
PD 0 0 0 0 0 0 0 0
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Sarah
C 0 2 (33%) 0 0 0 0 0 2 (17%)
PD 0 0 0 0 0 0 1 (100%) 1 (17%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Int Total 1 (9%) 3 (19%) 2 (14%) 1 (8%) 1 (9%) 5 (15%) 3 (15%) 16 (10%)
279
APPENDIX G19: INCORRECT QUALIFICATIVE A- CONCORDS
The table indicates the occurrences of incorrect suppliance of qualificative a- concords in the
data. The number of times that the qualificative a- concord is incorrectly supplied for each task
is presented in the Task Total column on the right-hand side of the table, and the number of times
that the qualificative a- concord is incorrectly supplied in each interview is presented in the Int
Total column at the bottom of the table. The percentages refer to the percentage of incorrect
forms as a percentage of the total number of obligatory uses in each task or interview.
Int No 1 2 3 4 5 6 7 Task
Total
Ann
C 0 0 0 0 1 (33%) 0 1 (50%) 2 (20%)
PD 0 0 0 1 (25%) 1 (100%) 0 0 2 (22%)
T 0 0 0 0 0 1 (67%) 0 1 (20%)
SB 0 0 0 0 0 0 0 0
Ben
C 0 1 (20%) 0 0 1 (17%) 0 2 (33%) 4 (14%)
PD 1 (14%) 0 0 1 (25%) 2 (33%) 2 (25%) 1 (20%) 7 (21%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 1 (100%) 1 (33%)
Claire
C 1 (20%) 1 (20%) 2 (29%) 1 (20%) 2 (50%) 0 0 7 (22%)
PD 0 1 (20%) 0 1 (20%) 1 (50%) 1 (33%) 2 (40%) 6 (23%)
T 0 0 0 1 (100%) 1 (33%) 0 0 2 (33%)
SB 0 0 0 0 0 0 0 0
Pat
C 0 0 2 (40%) 1 (25%) 2 (50%) 1 (25%) 1 (11%) 7 (19%)
PD 0 1 (14%) 3 (50%) 0 0 1 (50%) 2 (45%) 7 (31%)
T 0 0 0 0 0 1 (33%) 0 1 (17%)
SB 0 0 0 0 0 1 (100%) 0 1 (50%)
Karen
C 1 (17%) 2 (25%) 2 (22%) 0 0 0 1 (25%) 6 (16%)
PD 1 (20%) 4 (50%) 0 0 0 0 2 (33%) 7 (26%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Sarah
C 2 (33%) 1 (33%) 0 1 (25%) 0 0 1 (25%) 5 (16%)
PD 0 0 0 0 2 (100%) 2 (22%) 0 4 (19%)
T 0 0 0 0 0 1 (33%) 0 1 (17%)
SB 0 0 1 (25%) 0 0 0 0 1 (5%)
Int Total 6 (11%) 11 (23%) 10 (15%) 7 (14%) 13 (18%) 11 (15%) 14 (24%) 72 (17%)
280
APPENDIX G20: QUALIFICATIVE A- CONCORD OMISSIONS
The table indicates the omission of the qualificative a- concord. The number of times that the
qualificative a- concord is omitted for each task is presented in the Task Total column on the
right-hand side of the table, and the number of times that the qualificative a- concord is omitted
in each interview is presented in the Int Total column at the bottom of the table. The percentages
refer to the percentage of omissions as a percentage of the total number of obligatory uses in each
task or interview.
Ann
C 0 0 1 (100%) 0 0 2 (50%) 1 (50%) 4 (40%)
PD 0 0 0 1 (100%) 0 2 (100%) 0 3 (33%)
T 0 0 0 0 0 1 (33%) 0 1 (20%)
SB 0 0 0 0 0 0 0 0
Ben
C 0 1 (20%) 2 (33%) 0 2 (50%) 4 (67%) 2 (33%) 11 (35%)
PD 2 (29%) 0 0 0 2 (33%) 6 (75%) 1 (20%) 11 (32%)
T 1 (100%) 0 0 0 0 2 (100%) 0 3 (100%)
SB 0 0 0 0 0 0 0 0
Claire
C 2 (40%) 1 (20%) 1 (14%) 0 0 1 (25%) 1 (50%) 6 (19%)
PD 4 (67%) 2 (40%) 1 (17%) 0 1 (50%) 0 3 (60%) 11 (34%)
T 0 0 0 0 0 0 0 0
SB 0 0 0 0 0 0 0 0
Pat
C 0 3 (100%) 1 (20%) 3 (75%) 1 (25%) 2 (50%) 5 (56%) 15 (48%)
PD 3 (100%) 2 (29%) 0 1 (100%) 2 (100%) 1 (50%) 2 (40%) 11 (42%)
T 1 (100%) 0 0 0 0 1 (33%) 0 2 (33%)
SB 0 0 1 (100%) 0 0 0 0 1 (50%)
Karen
C 1 (17%) 3 (38%) 0 1 (20%) 0 1 (17%) 0 6 (14%)
PD 0 0 0 0 1 (100%) 2 (100%) 0 3 (11%)
T 0 0 0 0 0 0 0 0
SB 1 (100%) 0 0 0 0 0 0 1 (5%)
Sarah
C 0 0 1 (25%) 1 (17%) 1 (17%) 0 1 (25%) 4 (13%)
PD 1 (17%) 0 0 1 (25%) 0 2 (22%) 1 (33%) 5 (16%)
T 0 0 0 0 0 0 0 0
SB 0 2 (40%) 0 0 0 0 0 2 (9%)
Int 16 (31%) 14 (23%) 8 (12%) 8 (16%) 10 (13%) 27 (42%) 17 (24%) 100 (24%)
Total

Yes Downloadpdf145050886.pdf 2

Uploaded by

Copyright:

Available Formats

Yes Downloadpdf145050886.pdf 2

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Yes Downloadpdf145050886.pdf 2

Uploaded by

Copyright:

Available Formats

MORPHOLOGICAL DEVELOPMENT IN

A thesis submitted in fulfilment of the

CAROL BONNIN HOBSON

of Xhosa. A quasi-longitudinal research design is used to trace development in the oral

range from fairly unstructured conversation tasks to highly structured sentence-manipulation

exposed to input mainly in formal contexts of learning.

inflectional languages. However, this study analyses the interlanguage of learners of an

inadequate as a measure of early development in the interlanguage of learners of a language such

of learners of Xhosa cannot be an indicator of advanced language development.

LIST OF APPENDICES iii

CHAPTER 2: LITERATURE REVIEW 5

CHAPTER 4: FORMS AND FUNCTIONS IN THE DATA 94

CHAPTER 5: ACQUISITION PATTERNS IN THE INTERLANGUAGE

CHAPTER 6: COMPARISONS WITH THE BASIC VARIETY AND

TABLE 3.1 INTERVIEW SCHEDULE 70

TABLE 4.1 NUMBER OF UTTERANCES AND MORPHEMES IN THE DATA 95

TABLE 5.1 CONJUNCTION USE IN THE DATA 179

TABLE 6.1 COMPARISON OF LINGUISTIC ELEMENTS IN THE BV AND

This thesis is dedicated to my grandfather, Donald Bonnin Hobson.

L2 [second-language] acquisition research is concerned with investigating the processes

1.1 BACKGROUND TO THE STUDY

1.3 OUTLINE OF THE THESIS

2.1 CONTRASTIVE ANALYSIS

Three main theoretical claims result from CA:

Studies taking a CA perspective therefore focussed primarily on transfer phenomena and

2.1.2 Problems with contrastive analysis

2.1.3 The relevance of contrastive analysis

2.2 ERROR ANALYSIS

2.2.2 Criticisms of error analysis

2.2.3 Errors vs. Mistakes

2.3 MORPHEME STUDIES

The main findings of these studies include:

2.3.2 Criticisms of morpheme studies

2.4 NATURAL ROUTES

(1996:20) concludes that:

2.4.3 Influence of elicitation tasks

2.4.4 Influence of input

2.5.2 Key concepts in interlanguage

Adjémian (1976) contributed to the concept of IL by emphasizing its permeability to influence

2.5.3 Composition of the IL system

Second-language acquisition development in this view requires increased knowledge and/or

2.5.3.1 Universal elements

2.5.3.2 Formulaic elements

2.5.4 Features of interlanguage

Table 2.1 Features of the pragmatic and syntactic mode

1. Topic-comment structure 1. Subject-predicate structure

2.5.5 Interlanguage development

2.5.5.3 Development of the interlanguage system

The focus on L1 and developmental errors is a combination of a contrastive analysis (CA)

Fillmore (cited in Johnson 1992) sees L2 development as comprising three stages:

Cazden (1968) offers a different classification into four periods in L1 acquisition:

2.5.5.3.1 Strategies for development

2.5.5.3.2 Morphological development

The following acquisition patterns are also found:

Morphological development is usually found to be a fairly late development in the acquisition of

2.5.5.3.3 Utterance development

2.6 BASIC VARIETY

2.6.1 Theoretical background