Descriptive Grammar of Fanti
Descriptive Grammar of Fanti
Descriptive Grammar of Fanti
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact [email protected].
Linguistic Society of America is collaborating with JSTOR to digitize, preserve and extend access to Language.
http://www.jstor.org
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PREFACE
The study and analysis of Fanti, and the preparation of this grammar and
vocabulary, were made possible by a generous grant from the American Council
of Learned Societies. The work was part of the Intensive Language Program
of that Council, and part of the project of supplying implementation for the
study of African languages at the University of Pennsylvania. My deepest
gratitude is due the American Council of Learned Societies, and Dr. J. M.
Cowan, the Director of the Intensive Language Program.
It is hard to express my indebtedness also to Mr. Francis N. K. Nkrumah and
Mr. Kobina-Ahin Ahumanu Mbura, on whose speech this grammar is based.
Both proved to be patient and cooperative informants, and their appreciation
of the work that was being done made them much more than that. Both are
remarkable in their ability to express in precise English the meaning of a Fanti
expression. Work with them was no chore, but a pleasure.
Dr. Zellig S. Harris, of the faculty of the Graduate School of Arts and Sciences
of the University of Pennsylvania, is responsible in large measure for the accom-
plishment of the task of analyzing and describing Fanti. He shared in the
initial phonological work, maintained a constant interest in the rest of the
work, and contributed many valuable suggestions and, above all, an unfailing
inspiration. I owe him my profoundest thanks.
Grateful acknowledgement should also be made of the entire class of eight
which took the intensive course in Fanti offered at the University of Penn-
sylvania during the summer of 1942, when work on Fanti had just been begun.
Their help in obtaining material can hardly be measured, and their patience
with the unstable materials and methods of intensive instruction is not forgotten.
Particular mention should be made of Mr. Kalman Silvert, who first recognized
several Portugese loan-words in Fanti.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
TABLE OF CONTENTS
(Subsections are listed only if they are of particular grammatical significance)
page
1-3 .............
INTRODUCTION, . ................................ 7
I. THE PHONEMES................................................. 8
0.0-2. Introductory . ........................................ 8
1.0-2. The list of phonemes ............... .... ......... 8
2.0-7. The positional variants of consonants .................... 9
3.0-4. The positional variants of consonants or vowels........... 13
4.0-14. The positional variants of vowels ....................... 15
5.0-10. Junctures . ........................................... 20
6.0-3. Suprasegmental phonemes .......................... 23
7.0-7. The distribution of phonemes ..................... ... 26
8.0-6. Orthographies ................................ ..... 27
II. THE M ORPHEMES
. . ............................................ 30
0. Introductory ......................................... 30
1.0-21. Positional alternants of morphemes ...... .............. 30
2.0-12. Morpheme classes ........3........ ................... 37
2.1. Verbal stem s .... ................................... 37
2.2. Free non-verbal stems ... . ..........................
. 37
2.3. Non-verbal stems bound to non-verbal prefix........... 38
2.4. Non-verbal stems bound to non-verbal suffix........... 38
2.5. Non-verbal stems bound to stems ..................... 39
2.6. Reduplication . . .................................... 39
2.7. Nominalizing tone . .
................................ 39
2.8. Verbal prefixes ................................... 39
2.9. Non-verbal prefixes . .... .. ...... ............. ....... 41
2.10. Verbal suffix . ...................................... 41
2.11. Non-verbal suffixes ...................... .......... 41
2.12. Utterance-final particles ............... .............. 42
3.0-3. The phonemic structure of morphemes ................... 43
III. THE SYNTACTICELEMENTS...................................... 45
0.0-5. Introductory ............. .. .. ................... 45
1.0-2. Interjections ......................................... 45
2. Utterance-final particles ...............4.............. 46
3.0-2. Conjunctions ............... ... .. ............... . 46
4.0-2. Adjectives ........................... ....... ...... .. 46
5.0-2. Adverbs ........................................... 47
6.0-4. N ouns ................... .. ... .... ................... 47
6.10-16. Affixes of segments between two breaks............ 47
6.20-26. Bases of segments between two breaks ............. 49
6.30-37. Subclasses ..................................... 50
6.40-46. Structure of segments including breaks ............ 52
7.0-3. Verbs ..... .................. .............. 553
7.1. Bases of segments between two breaks ............. 53
5
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
6 TABLE OF CONTENTS
page
7.20-23. Affixes and structure of such segments ............. 54
7.3. Segments including breaks ....................... 56
IV. THE SYNTAX .................................. ............ 57
0. Introductory ......................................... 57
1.0-2. Exclamations ......................................... 57
2.0-2. Questions ............................................ 57
3. Nominal Sentences .................................... 58
4.0-3. Simple Stative Verbal Sentences ........................ 59
5.0-3. Simple Active Verbal Sentences ......................... 60
6. Verbal constructions preceded by dee . ............ . 62
7.0-13. Verbs in series ........................................ 63
8.0-5. Adverbs and Temporal Noun expressions . .......... 66
9.0-8. Conjunctive Constructions ............................. 68
APPENDIX. Texts ................................................. 73
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
INTRODUCTION
1. Fanti is the principal language of the Gold Coast Colony, which forms the
southern part of the Gold Coast, British West Africa. It is the native language
of perhaps a million people, and is used as a second language by many more.
To a large extent it appears to be mutually intelligible with Twi, which is spoken
in the Ashanti province to the north. It is not, however, mutually intelligible
with GA, the language of Accra, the Gold Coast capital. Fanti is a member
of the Akan group of languages, and falls into the general classification of Sudanic
languages.
2. Speakers of Fanti recognize the dialect of the coastal town of Anumabo
as the "best" Fanti, although that dialect has no official standing. The town of
Cape Coast appears to be a more active center of Fanti education and writing,
but this description follows as closely as possible the dialect of Anumabo.
3. The material on which this description was first based is the speech of Mr.
Francis N. Nkrumah. Work was done with him during the summer of 1942,
at Philadelphia, while he was a graduate student at the University of Penn-
sylvania. Since that time, however, it has appeared that his speech was not as
consistently that of Anumabo as had been thought; in fact, there appeared to
be a considerable amount of dialect mixture and perhaps even of language
mixture. Consequently, this description has since been adjusted to the dialect
of Anumabo by a study of the speech of Mr. Kobina-Ahin Ahumanu Mbura, a
native of Anumabo; this study was made in part-time work in August-October,
1944. The basic analysis remains the same, but omissions and inconsistencies
have to some extent been eliminated. Mr. Mbura is responsible also for the
texts which have been appended, and for valuable contributions to the accuracy
and completeness of the vocabulary.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
I. THE PHONEMES
0.0. This chapter is essentially a revision, with numerous corrections and
additions, of the article "The Phonemes of Fanti", by William E. Welmers and
Zellig S. Harris, in JAOS 62.4 (1942). Besides many material changes, a
basic change has been made in the form of presentation. In that article, the
sound types are listed first, and are then grouped into phonemes on the criterion
of complementary distribution. This chapter reverses that procedure; the
phonemes are listed first, and then their positional variants are described. No
notes are made of changes or additions; for all purposes except a study of method-
ology, this chapter supplants the earlier article.
0.1. In describing the positional variants of the phonemes, two procedures
are possible; all the variants of each phoneme may be listed together, following
the list of phonemes, or variants of several phonemes which show a common
characteristic may be described together. The latter course seems preferable
in this case, and has been followed.
0.2. The use of a great number of phonetic symbols has been avoided; in cases
where a commonly used symbol is readily understandable, it is used, but other-
wise a brief description of the sound is given.
1.0. The following are the phonemes of Fanti; they are divided into two
groups, segmental and suprasegmental, and further subdivided into phoneme
classes on the basis of distribution, and finally, for convenience sake, into phonetic
classes.
1.1. The segmental phonemes consist of thirteen consonants (three of which
are written as clusters), four consonants or vowels, five vowels, and five junctures.
A phoneme is a consonant if it never occurs with a tone; it is a consonant or a
vowel if it sometimes occurs with a tone; it is a vowel if it always occurs with a
tone. The names given to the junctures are not intended to be descriptive of
their phonetic character; the criteria for establishing and distinguishing them are
discussed below in 5.0-10.
Consonants:
Voiceless stops: p t k kw
Voiced stops: b d g gw
Voiceless spirants: f s h hw
Voiced spirants: y
Consonants or vowels:
Voiced spirants: w r
Nasal spirants: m n
Vowels:
Front Central Back
High i u
Middle e o
Low a
8
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 9
Junctures:
Stop: (period) /./
Pause: (comma) /,/
Break: (space or) / ~ /
Open juncture: (hyphen) /-/
Close juncture: (no space or mark between phonemes).
1.2. The suprasegmental phonemes all have reference to vowels only. They
are classified according to the number and arrangement of vowels with which
they occur.
Simultaneous with a limited series of vowels not necessarily adjacent, and
written after the last /i/ or /u/ of the series:
Heightening of vowels: /'/.
Simultaneous with one vowel or with two immediately adjacent vowels, and
written above the second of the two:
Nasalization: //.
Simultaneous with only one vowel: High tone, Mid tone, Low tone. These
tones are written according to the following rules:
One or more initial low tones: unmarked.
The first of one or more high tones, occurring initially, after low tone,
or after another high tone if open juncture (hyphen) intervenes: /'/.
The first of one or more low tones after a high tone: /'/.
The first of one or more mid tones, which occurs only after a high tone
and with no open juncture (hyphen) intervening: /'/.
The reference of any tone mark, including no mark initially, extends to
the next break (space), passing the bounds of open juncture (hyphen). The
following common sequences illustrate these rules:
.. '..
' low.. .high... aburentie 'young man'
.. .' low... high low mirikdsd 'I am speaking'
..." low .. high mid niankiupon 'God'
..." low .. high low mid asupddtir 'shoes'
'-...' high-. . . high md-agi'd 'my father'
2.0. The positional variants of consonants are described in this section,
including the variants of /w, r, m, n/ when they are used as consonants, having
no tone.
2.1. It is convenient to speak of the "normal" phonetic value of consonant
phonemes as those variants which occur in all positions except as noted below.
For some consonants, the "normal" variant is limited to the occurrences of the
consonant before non-nasalized /a, o, u/; in other cases, there are more positions;
in the case of /gw, hw/, no occurrences exist except as listed below; and /r/
varies even within this limited position, as described below. The "normal"
variants of the remaining consonants /p, t, k, kw, b, d, g, f, s, h, w, y, m, n/ are
not appreciably different from the values of the corresponding English phonemes
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
10 LANGUAGE DISSERTATION NO. 39, 1946
in initial position; voiceless stops are aspirated fortis, voiced consonants are
lenis. Examples of these consonants with their "normal" variants:
/t, dl are affricated with sibilant release, resulting in [ts, dz], before /i/ only.
E.g.,
ti hear di' partake of, eat
tintin long adikedi gift
The stop attack of [dz] is very light in some cases, and in rapid speech is often
scarcely audible; e.g., orididi' 'he is eating'. In the case of a few exceptions,
the normal stop variants of these phonemes occur; [d] occurs when the next
consonant (after close juncture) is /h/, the only example of which is /odihi/
'nobleman, rich man', and in the case of the one place name /edind/ 'Cape
1 Some of these variants, particularlyin the labial consonants,did not occur in Mr
Nkrumah'sspeech.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 11
Coast';2 [t] often occurs in /tind/ 'live at'. [dz] also occurs before [a] in the
one form [dza-];this is analyzed as /di N aa/, see 4.12 below.
/s/ retains its normal variant with no audible alternation before /i, e/. How-
ever, by comparison with the variants of /t, d/, it may be said that /s/ is similarly
affected in this position, but that the effect equals zero because /s/ is alreadv a
sibilant.
/r/ is treated comprehensively in 2.5 below.
/n/ is palatalized with [y]release, resulting in [ny], before /i, e/. E.g.,
Several cases of [n] before /i/ and [ny] before [a] require special analysis and
statement, as follows:
/n/ retains its normal variant [n] before /ii/; the only case of this is /anil/
'west'.
Initial [ni], always followed by open juncture (hyphen), is analyzed as /nu/.
This does not contrast with other occurrences of /nu/, and it happens also to
represent the morpheme which has the form [nu], also analyzed as /nu/, in other
positions. Compare [ni] in /nu-ddn/ 'his house' with [nu] in /6nu/ 'he, him'
and /iddn-nu/ 'the house'.
Final [ni, ni], always with low tone, are analyzed as /ni, ni'/. This does not
contrast with [nyi, nyi] in any case; the latter occur finally only with high tone.
Morphologically, this always represents the past tense of a verb whose stem ends
in /n/. E.g., /migwini/ 'I thought'; compare /migwzn/ 'I think', and [ny] in
/oni/ 'with'.
The common sequence [nya] is analyzed as /ni'a/, unless the vowel harmony
demands /nia/ (see 6.1 below). The variant of /i/ in such cases is zero (see
4.12 below), written to represent the variant of the preceding consonant, and
often also because of the vowel harmony; other consonants require the same
analysis. E.g., /6ni'c/ 'he gets', /6ni'in/ 'he wakes up', /niami/ 'God'.
This palatalized variant of /n/ must be carefully distinguished from phonemic
/ny/, and from /nn/ before /i, e/, which is phonetically [nny]. In both cases
there is a syllable division between the two phonemes, and, if the cluster is
initial, /n/ is a vowel; in addition, the tongue contact for /nn/ is longer than that
for /n/. E.g., /oni/ 'with', /onye dem/ 'it isn't so', /onni hd/ 'he isn't here'.
/k, g, h, y/ have the variants [c , ,z, before /i, e/, except that /h/ has this
variant only before non-nasalized /i, e/; the variant of /h/ before nasalized
2 This
may be a borrowedword, or an archaic form now largely supplanted by the English
form. Mr. Nkruma uses the normal variant also in /adikii/ 'morning' and /adisaa/ 'eve-
ning'. Some dialects of Fanti have [ts, dz] also before /e/ regularly, paralleling other
consonants. In Mr. Mbura's speech, this phenomenon occurs freely in a few words, and
regularly in a few others, which were possibly learned from speakers of dialects in which
that alternation is regular; such forms are for the present assumed to be with [t] and [d]
in Anumabo, until more complete evidence is available.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
12 LANGUAGE DISSERTATION NO. 39, 1946
/i, e/ is palatalized [h]. [z] has very little friction, and is often scarcely dis-
tinguishable from [y]. E.g.,
kiw fry iyi' this
nkin salt ye do, make, be described as
ke give, present hi'a need ([hy])
gi'nd dwell ohin chief ([hy])
hird earthen water-pot ([s]) ihen boat ([hy])
he he located at (out of sight) ([s])
There are several specific exceptions to this general statement. The normal
variant of /k/ occurs when the next consonant is /t, s/ (/g, h, y/ do not occur in
this position, but would presumably follow the same rule), and also before /ira/,
in the second as well as the first /k/ of /ki'etke/ 'train', and in /k/ before a
consonant plus low-tone vowel or before a consonant plus high-tone vowel when a
low tone follows that. In the last case, the first vowel after /ke/ is /i, e, a/,
and /ke/ is morphologically a verbal prefix (see Chapter II, 2.8 below), but its
positions are phonemically distinctive. E.g., the normal variants in /kisd/
'chew', /ki'td/ 'polish', /kesi/ 'big', /kira/ 'soul', /mirikedd/ 'I'm going to sleep'.
[ga] is analyzed as /gi'a/ (cf. /ni'a/ above); /kid, ki'd/ occur, but with the normal
variants of the vowels; see 4.12 below. Note also the normal variant of /k/ in
/kikd/ as stated above.
/kw, gw/ have as variants before /i, e/ palatal affricates similar to [c, #], but
with bilabial and palatal release. These sounds are similar to the common
variants of English /t, d/ in true, drew, but there is no retroflex release with the
Fanti sounds. E.g.
kwT pull awokwi eight gwi quiet down
The voiceless member of this pair occurs also before [a]; the combination is
analyzed as /kwi'a/. The voiced form occurs also before /u/, usually when
/m/ follows. Since /gw/ has no "normal" variant, no further statement is
necessary; however, it is apparent that most or all of the cases of /u/ are second-
ary changes from /i/. E.g., /gwuu/ 'lice' (perhaps originally /gwiw/); /agwuima/
'work', compare /agwi'n-fi/ 'craftsmen'.
/hw/ is a close, palatalized bilabial spirant, produced with the lips pursed
almost as for whistling; it occurs only before /i, e/. E.g.
hwi beat hwe look at mirihwihwe I want
2.3. Many consonants have variants with similar characteristics when they
occur before phonemically nasalized vowels or before vowels which are non-
phonemically nasalized before /m, n/.
/p, t, k, kw/ have variants with nasalized aspiration in this position. This is
true also of their palatalized variants before /i, e/. E.g.,
pi' scatter ka speak, touch
apim a thousand kan count
dapdn a week iku'n husband
tu bake nkin salt
tan hate kwi pull
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 13
/h, hw, w/ are nasalized throughout in this position; in addition, /w/ has a
very short /n/ onset. E.g.,
hu' see wen to guard
ahwia sand wdnwa wonderment
This variant of /w/ with [n] onset must be distinguished from /nw/, where
there is a syllable division; e.g., /nwurdba/ 'stars', both sounds in /wdnwa/.
2.4. /p, b, k, h/ have a light bilabial release before /on, or/;the release is
nasalized before /on/, according to 2.3 above. E.g.,
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
14 LANGUAGE DISSERTATION NO. 39, 1946
above section so far as articulatory data are concerned. The conditions under
which these phonemes are consonants or vowels are considered in this section.
3.1. /m, n/ are vowels when they occur between a break (space) and a con-
sonant, or between two consonants. E.g.,
mpd mat, bed But mi-mpd my bed
ntd twins But minti I don't hear
dindinndin very difficult (the only occurrences between consonants
are in this type of triplicated form).
3.2. /m, n, r/ are sometimes vowels after other vowels, with a different tone
from the preceding vowel. In these cases a tone mark is used which indicates
both the tone and the fact that the phoneme is a vowel. There are only a few
cases of final vocalic /n/, and all are rare as vowels medially, before a consonant.
E.g.,
dem' thus busaum moon
ddmi checkers or chess ahwir sugar cane
iki'n drum bam-ki'm the state umbrella
bir become ripe furdhkd flag
bur beat fafirdhtd butterfly
3.3. In some forms, final /w, r, m, n/ are phonetically vowels, with low tone
preceded by a high tone; in very slow and careful speech, the vowel /i/ is added
to these forms, and the phonemes in question become consonants. These forms
always occur medially in an utterance; in sentence final position the forms with
final /i/ always occur, even in rapid speech. In the light of these facts, and
of the fact that no contrast with other forms in final /'wi/ etc. exists, it is possible
to write the slow-speech alternant always; the vocalic /w, r, m, n/ is then an
alternant of the combination /wi, ri, mi, ni/ in this particular tone pattern. E.g.,
osumi na-agi'a 'he served his father'
oburi ni-nu'a 'he beat his brother'
otoni rmpua 'he sold bananas'
okiwi ndm 'he fried fish'
osumi (with consonantal /m/, normal vowel) 'he served'
3.4. In many forms, final /w, m, n/ are vowels, with low or high tone, preceded
by either a consonant or a vowel, and by either the same or a different tone;
these forms are heard only in rapid speech, while in slow or moderately slow
speech a vowel is added and the phoneme in question becomes a consonant.
These forms occur medially or finally in an utterance. Open juncture (hyphen)
precedes /w, m, n/ in these cases; in fact, the statements below are the criterion
for establishing this juncture in these cases. When the juncture is preceded by a
vowel, and when the same tone precedes, the /w, m, n/ in rapid speech is at the
borderline between customary definitions of a consonant and a vowel. In these
cases, however, the normal variant of the vowel (see 4.1, 2 below) is used, rather
than the pre-consonantal variant (see 4.7 below); consequently the /w, m, n/
after the juncture is to be considered a second vowel. All of these forms are
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 15
written, however, with the vowel that follows them in slower speech; the vowel
variant of /w, m, n/ is then an alternant of the consonant-vowel sequence.
The final sequences that occur are /-mi, -mu', -nu, -wu/. In all recorded cases,
the phonemic distinction between /-mi/ and /-mu'/ is apparent even when both
are reduced to vocalic [m] in rapid speech. The tone of /-mu'/ and its variant
[m]is often high, while that of /mi/ is always low; vowel harmony often demands
/-mu'/ and makes /-mi/ impossible; and the vowels /i, e, a/ are always rounded
before /-mu'/. If none of these criteria applies, there would presumably be
ambiguity in rapid speech, or, just as likely, the speech would be slowed down a
trifle for the full form. No such instances have been recorded, and the mor-
phology allows for very few possibilities of that kind, so it seems profitable to
write the slow-speech form in all cases; in any case, the slower form is easy to
elicit from a speaker of the language. E.g.,
otdn-wu he hates you nokwdr-nu the truth
ontdn-mi he doesn't hate me ka-mu' shout
mihi'-wu I see you bura-mu' come in!
ohi'-mi he sees me os/i'-mu' he set out
orikiw-nu he is frying it oti ddn kesi-mu' he lives in a large
abir-nu that time, then house.
The most common usage of /-mu'/ is in constructions such as /owo ddn-nu
nu-mu'/ 'he is in the house'; /-mi/ never occurs in this position.
4.0. The positional variants of vowels other than /w, r, m, n/ are described in
this section.
4.1. It is convenient to speak of the "normal" phonetic value of vowel pho-
nemes as those variants which a non-nasalized vowel has before a juncture or
before a consonant which is followed by another vowel. All vowels occur in this
position. The phonetic values are those usually assigned to the symbols used,
except as noted. The closest English comparisons are given for the sake of
convenience.
/i/ is [I] or slightly lower. E.g. /si/ 'say'. Cf. sit.
/e/ is [e]. E.g. /pe/ 'like, want'. Cf. pet.
/a/ is [a]. E.g. /da/ 'lie down, sleep'. Cf. ma, dot.
/o/ is [o]. E.g. /do/ 'to love'. Cf. for, fought; /o/ is between.
/u/ is [u] or slightly lower. E.g. /bu/ 'get drunk'. Cf. put.
4.2. The "normal" value of vowel phonemes under the influence of /'/, which
indicates a higher and more relaxed series of vowels, is as follows; all the vowels
given are under the influence of /'/, whose scope will be described below in 6.1:
/i'/ is [i]. E.g., /si'/ 'build, happen'. Cf. seat.
/'el is [e]. E.g., /si'e/ 'bury'. Cf. ate. /e/ occurs only after /'/ unless
it is separated from /'/ by an open juncture (hyphen) or break (space). When
it is followed immediately by hyphen or space, it is always preceded by one of the
palatalized variants of a consonant; the sequence /ke-mu'/ is possible, with the
variant [c] for /k/, but the sequence /te-mu'/ is impossible, because /t/ does
not have a palatalized variant before /e/. The sound [e] occurs before /'/ with
only close junctures following, and after /t, d, s/ with hyphen or space imme-
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
16 LANGUAGE DISSERTATION NO. 39, 1946
diately following, but in these cases it is phonemically analyzed as /a/; see below.
/'a/ is [ae] or slightly lower after /i/ when no consonant intervenes. E.g.
/hi'a/ 'meet'. It is [a] after /u/ when no consonant intervenes. E.g., /bu'd/
'reply'. It is [a] after anything else. E.g., /pi'ra/ 'hurt', su'md 'hide'.
/a/ before /'/ is fe]. This does not contrast with /e/ in this position,
because /e/ occurs in this position only after consonant alternants which never
precede /a/. The variants of the two vowel phonemes as such are the same,
but it can be determined which is which phonemically by the preceding con-
sonant. E.g., /adu'/ 'ten', /biraafi'/ 'wicker basket (for holding fish)', /ha-mu'/
'forest, bush country' (here /h/ is [h]; compare /he-mu'/ 'put in', where /h/ is
[g], though the rest of the form is identical phonetically).3
/'o/ is [o]. E.g. /ofu'n/ 'corpse'. Cf. hope.
/u'/ is [u]. E.g. /tu'/ 'dig, fly'. Cf. toot.
4.3. /i, e, a/ before /-mu'/, as illustrated above, are also rounded, becoming
respectively [ii], [5], and (the same) [6].
4.4. The vowel of initial /nu/ has the same phonetic value as /i/ has elsewhere,
as stated above in 2.2. This vowel and the vowels of initial /mi, mi, wu/ and
of initial or non-initial /ri, ri/ are only partially influenced by heightening
(/'/), forming exceptions to the statement of 4.2. The following rules cover the
cases:
/i/ is regularly influenced before/yi'/. E.g. /miyi'/ 'I take away'.
/i/ is [u] before /hu'/. E.g. /mihu'/ 'I see'.
/i/ is [u] before /hu/. E.g. /mihi/ 'myself', /nuhii/ 'himself'.
/i/ in other cases, and /u/ in all cases, are partially assimilated to the
following vowel. They are slightly higher before a higher vowel, lower before a
low vowel; /i/ is farther back before a back vowel, /u/ is farther forward before
a front vowel. There are as many different variants as there are vowels. E.g.,
/mido/ 'I love', /wu-szkdn/ 'your knife'. This alternation is even more easily
3This analysis of [e] as /e/ and /a/ in complementary environments is required by the
very existence of a contrast such as /ha-mu', he-mai'/. It is strengthened by a consideration
of morphophonemics which should be stated here. In a large number of the positions in
which [el is assigned to the phoneme /a/, it represents all or part of the same morpheme as
[a] does when /'/ is not influencing the vowel. Thus a widespread morphophonemicchange
is covered with no injustice to the facts by a phonemic statement. E.g., /madsi/ is for
[madsi] 'I have said'; /madsV/ is for [meesi] 'I have built'.
By the same morphophonemic principle, both Itel and /ta/ are actualized as [tel when
they come under the influence of /'/, and the same is true when the consonant is /d, s/,
since these consonants alone have no special variants before /e/. Here a phonemic writing
of both forms as /ta/ is necessary, although it obscures the distinction between morphemes;
after all, that distinction is obscured in speech as well. Such a procedure is far preferable
to analyzing [e] as /e/ in all cases and thus obscuring real phonemic distinctions such as are
given above.
A /similar morphophonemic principle changes /a/ to /o/ between labials, especially
/
/w/, and -mu'/ ; e.g. /kw6 -mUi/ 'bush country' from /kwA/ 'farm', f6-mt'/ or /fd-mi'/ 'the
ground, down' from /fd/ 'earth, swish'; compare /wo-mi'/ 'be inside' from /wo/ 'be at'.
Here the phonemic analysis parallels the phonetic facts; /o/ is [o] in each case. But the
analysis of [e] as /a/ is not a violation of any phonetic data; [el as a variant of /a/ and [e]
as a variant of /e/ occur in complementary distribution, and other phonetic data, the
variants of preceding consonants, demand the analysis.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 17
heard than that between a following short and unstressed /u/ and /i/; the
clearer of the two vowel distinctions in /mi-kurd/ 'my calabash' and/mi-kird/
'my soul' is in the first vowel, not the second.
4.5. By a somewhat similar process, /o/ has a light palatal off-glide before a
consonant followed by /i'/. E.g., /wosi'/ 'they build'.
4.6. Nasalized vowels have the same Fosition as those which are not nasalized.
Vowels are nasalized non-phonemically before consonantal but not before vocalic
/m, n/ when a juncture or another consonant follows; this nasalization determines
the allophones of the preceding consonants (see 2.3 above), but is itself determined
by the following nasal spirant. Non-phonemic nasalization occurs also after
/m, n/, with some freedom as to its degree. Only /i, u, a/ are phonemically
nasalized. The nasalization includes the immediately preceding vowel if there is
one. E.g.,
ft' dirty ku fight
kwi pull hi'a need
kd speak, touch su'a study
ku' kill adikif morning
4.7. Vowels before final consonants, which are only /w, r, m, n/, or before
/m, n/ followed by another consonant, have variants that are somewhat shorter,
and assimilated partly in the direction of the midcentral tongue position; all the
vowels, are however, still distinguishable in this position. After /t, d, s/, and
before /m, n/, /a/ is [ae] or slightly lower. E.g.,
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
18 LANGUAGE DISSERTATION NO. 39, 1946
the preceding consonant has no palatalized variant before /i/, as is the case of
/k/ before /ira/, the distinction between /i/ and /u/ is audible even in rapid
speech, though it takes careful listening to hear it, even for a native speaker; in
slow speech the distinction is perfectly clear. It is convenient also to note that
in all dubious cases except the one pair /kurd/ 'calabash' and /kird/ 'soul', the
vowel in slower speech turns out to be /u/. These very short vowels are often
voicelss. The following examples show the criteria for distinctions that are
easier to hear than the mere distinction between the two vowels.
opi'rd he hurts (palatalized /p/, high vowels)
opird he sweeps (palatalized /p/, low vowels)
opu'ro he stumbles (normal /p/, high vowels)
pu'rko6 pig (normal /p/, high vowels)
otur6 he spreads (normal /t/, low vowels)
furd call (normal /f/, low vowels)
kurd calabash (vowel is [u] when heard carefully)
kird soul (vowel is [i], /k/ perhaps slightly farther
forward than in /kurd/.)
4.10. Double writing of vowels indicates length. This is true also when a
juncture is written between the two vowels. E.g.,
pii many, much mbd ahin how many children?
nkda debt md-agi'd my father
tuu throw, shoot md-awufu my parents
buuku' book obaa hd he came here
4.11. There are a few cases in which double (i.e., long) vowels occur before
final /n/. In these cases, the whole sequence indicates a long nasalized vowel
which is higher toward the end. When the vowel is /i/, the rise of the tongue
seems to approach a palatalized [n]. When the vowel is /o/, the rise of the
tongue seems to approach rV]. E.g.,
dii'n silent su'oon stinking
d6nkuroon deep akunkurdan raven
4.12. /i, u/ have zero variants in the following cases:
/i, u/ after /w, r, m, n/ when the combination has the variant of a vocalic
/w, r, m, n/, as stated fully in 3.3-4 above.
/u/ after several consonants before /'o/; the combination /u'o/ hasthe variant
[o] in these cases. The same combination does not occur after any consonants
with any other variants. The analysis of [o] in these cases as /u'o/ instead cf
/i'/ is made for one reason only; [o] is so rare as the only vowel in an utterance
that it is preferable to say that it is always (as it obviously is in most cases) an
alternant of /o/ under the influence of /'/. In the vast majority of cases /u'/ or
/i'/ with its normal variant occurs elsewhere in the vicinity; in the few cases
where this is not true, /u'/ is written before /o/ and /u/ is said to have a zero
variant. Compare the similar analysis of an occasional /i'e/ below. Examples
of /u'o/ as [o] are /asu'on/ 'seven', /pu'o/ 'roar, howl'.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 19
/i/ after /g, kw, n/ before /'a/, and after /g, n, k, hw, y/ before I'el; the
combination /i'a/ in such cases is [a], the combination /i'e/ is [e]. E.g. /gi'a/ or
/gi'e/ 'leave', /kwi'a/ 'cut', /ni'a/ 'get', /ni'an/ 'wake up', /ki'et6ke/ 'train',
/hwi'ew/ 'sip from the hands', /yi'g/ 'be good'. Note that phonemic /ki'a/
'greet' does not have the zero variant of /i/; however, it is not included in the
above statement, and the statement includes no contrasts or contradictions.
/i/ after /d, n/ before /a/. E.g. Idi aa/ 'that which', /niami/ 'God', /niank1/
'friend'. In all of these cases, /i/ is demanded instead of /i'/ because of the
vowel harmony. If the vowel harmony is indecisive, /i'/ is written, as in /ni'am/
'mash, shine'. However, in some other occurrences of this sequence with the
same meaning, /'/ cannot be written; /oriniam/ 'he is mashing' contrasts with
/orini'dm/ 'he is shining'.
/i/ after /s/ before /'a/ in some tonal contexts but not in others. This
statement is not complete, but there is little doubt that an adequate statement
can be made with additional evidence. The /i'/ is demanded by vowel harmony
in every case. The worst that can be said for this statement is that it gives only a
unilateral one-to-one correspondence, from speech to writing but not from writing
to speech; the best that can be said is that a more careful study of the tone
contexts will clarify the confusion. In the following cases /i/ has the zero
variant: /osi'dn dee/ 'because', /isi'dm/ 'flour', /mdnsi'd/ 'strong as three men
(said of a machine, somewhatslangy)' or the name given to the third male child
in the family (compare /mdnsa/, the third female child's name), and perhaps
also /si'an/ 'untie, let down, carve'.4 The following forms have the normal
pre-vocalic variant of /i/: /asi't/ 'six', /basi'a/ 'woman'.
/i/ after /b/ before /'a/ in the one form /ebi'dsa/ 'three'. This does not
contrast with other cases of /bi'a/, but compare /bid/ 'place', with the normal
variant of /i/. The vowel harmony demands /'/ in /ebi'dsd/, and /i/ must be
written because /'/ never occurs after vowels other than /i, u/ in other cases.5
There are two reasons for this type of phonemic analysis in these cases; both
reasons are present in many cases, but sometimes only one. /i/ is demanded
with a zero variant in many cases to explain the palatalized variant of the
preceding consonant; e.g., /kwi'a/ 'cut', /niami/ 'God'. /'/ is demanded in
most cases to explain the vowel harmony, and since it does not occur apart from
/i, u/ in the vast majority of cases, it is written with /i, u/ in these cases as well;
e.g. /esu'on/ 'seven', /isi'dm/ 'flour'. Both reasons are present in cases such as
/okwi'd/ 'he cuts', /oni'dn/ 'he wakes up'. The analysis is happily confirmedin
many cases by dialectical differences.
A similar analysis might be used in cases of phonetic [tse, tse], writing them as
/tie, ti'e/; however, the vowel harmony is never decisive in these cases, and it
4 It is significant that, according to Mr. Mbura,/osi'dn de /does not have the zero variant
but the normal variant of /i/ in the Cape Coast dialect. A small Fanti dictionary by
J. Delaney Russell (1910) lists the form as osande, but lists the word for 'flour' as isTam,
indicating the presence of the normal alternant of /i/ in that form also. The evidence for
,si'an/ is rather scarce, but it seems likely that the zero alternant occurs in some forms of
the verb, as in /orisi'ln/ 'he is untieing', but not in others as /osi'dn/ 'he unties'.
6Cape Coast has the normal variant of /i/ in /ebi'dsd/, and the same is regular in Mr.
Nkrumah's speech.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
20 LANGUAGE DISSERTATION NO. 39, 1946
seems extremely likely that these are forms from another dialect; they are
therefore analyzed as /te/ for the time being.
4.13. The sequences /u'e, ze/ represent, respectively, long [6, 5]. They occur
only in /ku'er/ 'weigh' and /suer/ 'arise'.
4.14. The phoneme /u/ has the variant [i] in the one form /owui'/ 'he died';
compare /wadwu'/ 'he has died, they have died' and /wu'/ 'to die'. The writing
of /u/ in this case is mandatory because /w/ has its normal variant, not the
palatalized variant which occurs regularly before /i/.
5.0. The determination of phonemic junctures is not completely clear, but
many essential distinctions are stated in this section. There are three reasons
for the inadequacy of this statement at present. First, the amount of material
recorded limits the number of pertinent examples of the more difficult cases, to
say nothing of the questions of accuracy in recording that arise. Second, there
is in the nature of the case a comparatively small number of cases where juncture
is the only phonemic contrast in a pair. Third, even the best informant may say
that two things are the same when they are different, or different when they are
the same, particularly in this type of material.6 Two general facts are clear: a
phonemic juncture may easily be set up in a vast number of cases on grounds
such as the difference between /bd-ru/ (see 3.4 above) 'the child' and /bdn/
'a wall'; on the other hand, there are at least some dubious cases, such as /mitina
asi/ 'I live down below' and /miti-nu asi/ 'I understand him'. Mr. Mbura says
that the latter two forms are identical in sound; however, the former is suspect
because it was elicited under very artificial circumstances and is not a common
expression. In the case of /k&-nu/ 'touch it' and /kan/ 'count', Mr. Mbura also
says that the two are the same in rapid speech, but in this case it is safe to dis-
tinguish them at least on the basis of the slower form of the first, which has the
normal rather than zero variant of the final vowel.
It must also be pointed out that the establishment of phonemic junctures is
independent of any considerations as to morpheme boundaries. The statements
that follow have to do only with the phonemic interpretation of phonetic data.
The choice of one interpretation over another, of calling something a juncture
rather than, for example, setting up a whole new series of vowels, may have been
suggested by knowledge of the morpheme boundaries; but it is not a true phone-
mic analysis unless it can also be justified on purely phonemic grounds. The fact
that most if not all of the junctures established occur at morpheme boundaries
is a fortunate, and indeed probably reasonable, coincidence; however, it does not
make or break the analysis as such. As a matter of fact, many morpheme
boundaries are not phonemically marked according to the following statements;
the form /mirihwihwe/ 'I am looking for, I want' contains four morphemes, but is
no different juncturally from /mbirikisi'/ 'thicket', which contains two so far as
is known, or from /fuzrdaikd/'flag', which certainly contains only one. On the
6 This is, of course, no reflection on any informant for any language; many a trained
linguist has raised his eyebrows the first time the junctural distinction was pointed out
between the second and last words in the sentence Your highness [height] astounds me,
Your Highness.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 21
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
22 LANGUAGE DISSERTATION NO. 39, 1946
and /'... y ...'.../. E.g., /oni'i sanku' bo/ 'he knows how to play the organ';
compare /ope sanku'-b6/ 'he likes organ playing', and the high-mid sequence in
/denkem/ 'crocodile'.
5.4. Break is sometimes determined only by stress; an alternative possibility
would be to consider stress phonemic, but the free use of a momentary pause in
slower speech makes the analysis of stress as a determiner of juncture preferable,
to say nothing of the morphemic consideration that morpheme boundaries are
always present. Break occurs in such an analysis between a syllable with high
tone and a stressed syllable with low tone, in contrast with close juncture between
a syllable with high tone and and unstressed syllable with low tone. E.g.,
/nde da/ 'yesterday'; compare /miribd/ 'I am coming'. Break by this criterion
always occurs before the interrogative particle /a/.
5.5. Break is determined by stress also between a syllable with low tone and a
syllable with high tone. If the first syllable, with low tone, has zero stress,
it is followed by break; if it has a stress (weaker than that of the following high
tone, but not zero), it is followed by close juncture. This statement probably
covers more cases of break than any other statement; the presence of the weaker
stress as opposed to zero stress is hard to hear except in cases of minimal contrast,
but at present the criterion seems to be universally valid. E.g., /bo ddm/ 'go
crazy' is phonetically [bo'ddm], while /bodom/ 'dog' is phonetically [Ibo'dom].
Here is a case where knowledge of morpheme boundaries may determine the
choice of analyses. One might normally expect the analysis to be the reverse,
close juncture after zero stress and open juncture or break after secondary stress,
as in English. The unexpected analysis is preferable in Fanti in the light of
morphemic considerations; but the fact that a junctural analysis is called for is
determined purely by phonetic data.
5.6. Open juncture occurs before final /mi, wu, nu, mu'/ according to the
criteria stated above in 3.4 for rapid speech; they represent vocalic /m, w, n/,
preceded by a consonant or by the normal unshortened and non-nasalized or
phonemically nasalized alternants of vowels. In slower speech these four syl-
lables and also /hen, hum, hon/ are preceded by open juncture on criteria basically
the same as those used to establish break in 5.5 above; they are either low tone
and stressed, or, in the case of /-mu'/, sometimes high tone but lower than a
preceding high series. Morphemically, all of these except /-mu'/ are personal
pronouns. E.g.,
iddn-nuthe house
kd-nu touch it; compare /kan/ 'count', which may be identical in
rapid speech.
bd-nu the child; compare /bdn/ 'wall', which is never the same.
mido-wu I love you; compare /midow/ 'I plow'.
5.7. Open juncture occurs after initial /mi, wu, nu, h6n, hum, h6n/ when they
are stressed and when a stress occurs after them. These are the only initial
syllables that occur in this environment; they were not chosen out of many
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 23
possibilities for morphemic reasons. /mi/ is the only one which also occurs
stressed without a following stress. E.g.,
mi-dan my house mi-wofa my uncle
wu-ti'r your head wu-hundm-mu' e how are you?
hen-kdsd our language But niuhu himself (one stress)
There are no cases of complete minimal contrast to illustrate this juncture, but
the following forms have only one stress also: /mido/ 'I love', /okdsd/ 'he speaks'.
Morphologically, this statement indicates that all but a very few nominal forms
with pronouns have a juncture, while verbal forms with pronouns never do.
5.8. Open juncture occurs between two occurrences of the same vowelmedially,
if the first vowel is preceded by initial /m, n, w/ and if both vowels have the
same tone with the second stressed; if other consonants precede, break occurs.7
This statement excludes forms in which the two vowels have different tones,
or in which the first is stressed. E.g.,
md-awufu my parents mbd ahin how many children?
wd-agi'd your father mitina asi I live down below
Forms such as the following are not included in this statement: /nkda/ 'debt',
/nkdd/ 'remembrance', /madgi'd/ 'I have left'. Morphologically, as in 5.7
above, nominal forms with pronouns have juncture, while verbal forms do not.
5.9. Open juncture occurs between consonants that are not homorganic,
unless tonal considerations stated in 5.3 above demand a break; in slow speech
and sometimes even in rapid speech a release of breath occurs at the point of
juncture. E.g.,
ahum-gwii peace ahum-ka happiness
ani'm-gu'-dsi shame eni'm-ni'dm glory
5.10. Open juncture occurs in segments between two breaks at the point where
the influence of /'/ is interrupted (see 6.1 below). E.g.,
awire-fi' forgetfulness eni'm-gu'-dsi shame
aka-tu'a reward, pay sanklc'-b6 organ playing
akire-suza dates (here there is no known morpheme boundary at
the point of phonemic juncture).
6.0. The suprasegmental phonemes must be described as to their phonetic
nature and their range of influence.
6.1. Heightening of vowels, //, is written after the last /i/ or /u/ in the series
it influences. The variants of vowels under its influence have been described in
4.2 above. The phoneme is established as an analysis of certain phonetic data
to cover a type of vowel harmony which is characteristic of Fanti. All the
vowels within certain limits belong to either the high series of vowels, [i, e, o, u]
7 This is a case in which the morphemic consideration has determined the
choice of open
juncture in some cases and break in others, on the basis of the same data; however, the
phonemic contrast is present between the two in terms of the preceding consonants.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
24 LANGUAGE DISSERTATION NO. 39, 1946
and [a] only after [i, u], or the low series of vowels, [I, e, o, u] and [a] with no
limitations. All of the low vowels occur as the only vowel in an utterance, but
of the high vowels only [i, u] occur as the only vowel, with the exception of rare
cases of [e, o, a]. The low series of vowels is phonemically analyzed as /i, e, o,
u, a/; the high series is analyzed as the same vowels with /'/, and /'/ is written
after the last 'i' or /u/ of the series; the rare cases of [e, o, a] as the only vowel
in an utterance but of the high series are analyzed as /i'e, u'o, i'a/, usually with
other reasons contributing to the analysis (see 4.2 above). This description of
vowel harmony makes it possible to write the same morpheme in the same way
even when it has different phonetic forms. E.g., /6si'/ 'he builds' and /6si/
'he says', where 'he' is phonetically [o] in the first case and [o] in the second.
This analysis also obviates a special statement as to the high series in forms such
as /okwi'd/ 'he cuts', where, phonetically, [o] precedes [a] in contrast with the
normal [o] before [a] as in /odd/ 'he sleeps'. The range of influence of /'/ is
covered in the two following statements:
All vowels after /'/ to the next juncture are under its influence. E.g.
mpapaa-mu' chapters, divisions (Notice, the first /a/ is [a] here, while
/aa/ is [e].)
odo-mu' he is out ([o..o..u])
6si'-mu' he sets out ([o..i..u])
6si-mi' ([o..i..u]) (a hypothetical case elicited from Mr. Mbura to
show the contrast in a minimal pair; he had no hesita-
tion about the pronunciation).
O6kfi'e he goes home ([o..o..ie])
mburd-ni' lawyer ([..e..i])
ipufu-ni' fisherman ([i..u..u..i])
ofi' mbirikisi' ko mbirikisi' from everlasting to everlasting (archaic;
all high).
These cases illustrate also the necessity of analyzing a vocalic /m/ in rapid
speech as /mu'/, even apart from the fact that the slower variant is just that.
6.2. Nasalization, /~/, is self-explanatory as far as its phonetic character is
concerned. Its range is a single vowel or two adjacent vowels; in the latter
case the symbol is written over the second. It is always final.
6.3. Tones, written as described in 1.2 above, are relative in their precise levels;
low tone, however, is always at the pitch at which a maximum comfortable
relaxation of the vocal cords is possible for the individual speaker. Even in single
isolated sounds, there is a clear difference between high and low tones; most
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 25
utterances of one syllable, especially when citations, are of high tone. There is
no coordination between the actual pitch level and the particular vowel with
which the tone occurs; however, in citing a list of forms of one syllable with vowels
in the sequence /i', i, 'e, e, a, o, 'o, u, u'/, both Mr. Nkrumah and Mr. Mbura
did a striking performance of going first down and then up the scale. In minimal
forms, the interval between a low tone and a high tone is normally about a musical
fifth, and the interval between high and mid a minor third; e.g.,/odihi/'nobleman,
rich man' has pitch levels such as C-G-E. In longer sequences, the following
statements apply:
All the tones in a series of the same phonemic level are also of the same absolute
level, except for one instance noted below. E.g., /abirdntie/ 'young man' has
only two levels of pitch, low and high.
A second series of high tones in an utterance is slightly lower than the first,
and each succeeding series is slightly lower than the one preceding it. E.g.,
/minni ntiumi' nnum nsd/ 'I won't be able to drink liquor' has four levels of
non-low tones, all of which are phonemically high.
A mid tone within a succession of high tones is lower than the high tone follow-
ing it. E.g. /dnkem kesi/ 'a large crocodile'.
In a series of nothing but low tones occuring in isolation or between breaks,
the next to the last tone is slightly higher than the others. E.g. /wofa/ 'uncle',
/pii'/ 'many, much', /bokoo/ 'slowly'; the first vowel of /wofa/ is higher in tone
than the second, but not as high as the first vowel of /wofd/ 'they take', while
both low tones have the same absolute level in /mi-w6fa/ 'my uncle'.
A phonemically low vowel after a break is freely, and after an open juncture
preceded by a single syllable (see 5.7-8 above) is always, of the same level as a
high tone immediately preceding; this level is higher than that of a following
phonemically high tone. E.g., the vowels after junctures in the following cases:
/mbd ahin/ 'how many children?', /edu' enu'm/ 'fifteen', /md-awufu/ 'my
parents', /mt-mbd/ 'my children'.
In 5.4, 5.5, 5.7, 5.8 above, stress was used as a criterion for determining
junctures. In other cases, stress is automatically determined by tone. Except
as noted in those sections, stress occurs with the following tones:
The first high tone of a series of one or more, except that initial /mi, i, o, ye,
hum, wo/ are unstressed when they are the first of a series of two or more high
tones. E.g., /aburentie/ 'young man', /donkuroon/ 'deep'; but stress on the
second syllable in /mibad-nu/ 'when I came'. This exception is confined mor-
phologically to verbal forms only of this type.
The first low tone of an initial series of two or more, and an initial single low
tone except on an initial vowel. E.g., /mankai/ '(a kind of) yam', /iwu'rd/
'dirt, sweepings', /hemd/ 'canoe'; but not in /iddn/ 'house'. The stress in these
cases is of a lesser degree than that on a first high tone.
The second low tone of a series of low tones preceded by a high tone. The
stress here may be even weaker. E.g., /mi-wofa/ 'my uncle', has a weak stress
on the last vowel.
In general, the actual volume of stressed vowels appears to decrease throughout
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
26 LANGUAGE DISSERTATION NO. 39, 1946
an utterance as does the pitch of high tones. Notice also that mid tones always
have zero stress, except that a mid tone following a low tone has a weak stress,
as does a second low tone; e.g., /lbiuku'/ 'book'.
7.0. The following are some of the more significant features of the distribution
of phonemes within an utterance, apart from the adequate statements already
made about tones and /'/.
7.1. The distribution of /N/ is as follows:
// occurs only with /a, i, u/.
// occurs only after voiceless consonants.
// occurs only finally, i.e., before a break, pause, or stop, or open juncture,
not before consonants.
The fact that /~/ does not occur before consonants has certain interesting
implications. The only final consonants are /m, n, w, r/. It would therefore be
possible to interpret /m, n/ as nasalized /w, r/, or better still, to write nasalization
with vowels before /w, r/ and say that the consonants equal [m, n] in this position.
This could be extended to all occurrences of /m, n/; initially before vowels the
following vowel could be written as nasalized, and initially before consonants the
phoneme /7/ is sufficient with no distinction as to articulatory position, since
/m, n/ are neutralized in that position. By such an analysis, /m, n/ could be
entirely eliminated as phonemes.
7.2. As just pointed out, /n, n/ occur in complementary distribution before
consonants in close juncture; /m/ occurs only before /p, b, m/, and /n/ occurs
before all other consonants. Apart from the analysis suggested above, by which
both would be written in this position simply as /-/, it is possible in any case to
write either /m/ or /n/ for all pre-consonantal occurrences. The choice of
writing used has been determined only by considerations of the phonetic char-
acter of the phonemes.
7.3. The only phonemes that occur in final position are /m, n, w, r/ and the
vowels. Medially, only /m, n/ occur before other consonant phonemes, but
neither occurs before /r/.
7.4. /kw, gw, hw, y/ have limited distribution. /hw/ occurs only before
/i, e/; /kw, y/ occur before /i, e, a/; /gw/ occurs before /i, e, a, u/.
7.5. The phoneme /r/ never occurs initially, i.e., after a pause or stop. It
does occur after a break within an utterance, but only in the combination /ri/,
which is a morpheme indicating continuous action in present time or action just
about to take place.
7.6. On the basis of evidence not yet adequately studied, it seems possible that,
by a complicated statement involving tone and juncture, /r/ and /y/ may turn
out to be in complementary distribution with each other. This would make an
even neater picture of the phoneme list and certain distributional phenomena.
7.7. The phonemic structure of all utterances can be summarized in a formula;
however, the formula would be complicated, and the phonemic structure of
individual morphemes is a more noteworthy feature of the language; this will be
described in Chapter II, 3.0-3.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 27
8.0. The standard orthography of Fanti used in the Gold Coast differs in many
important respects from the phonemic orthography established in this chapter.
Fanti has been a written language for perhaps a century. By about 1870 the
orthography had been pretty well standardized, and was used in publications
and grammars. Along with many other African languages, Fanti was given a
new, reformed orthography in the late 1920's, largely through the instrumentality
of the International Institute of African Languages and Cultures (now the
International African Institute) and its representative Professor D. Westermann.
The new orthography differs from the old primarily in the writing of vowels.
This "new script" or "phonetic script", as it is commonly called in the Gold
Coast, is now used generally in schools and publications. It will be compared
with the phonemic orthography of this description in the following sections, with
a few parenthetical notes about the old orthography, and a few historical and
dialectical observations. The comparison is not complete in every detail; no
intensive study of either the old or the new orthography has been made; but
sufficient comparison is made to make it possible to use either orthography
with little difficulty.
8.1. Tones have never been consistently written in either accepted orthog-
raphy. Tone marks are occasionally found in works written by Europeans about
the language, but these works all attempt an explanation of their own tone
marks, which are apparently as unsatisfactory as those used in this grammar.
The recent Grammarof the Fante-Akan Language, by W. T. Balmer and F. C. F.
Grant (London, 1929), gives essentially the same system of tone writing as that
described above in 1.2 above and used herein, with one minor difference, that a
low tone with stress before a high tone (see 6.3 above) is marked with /'/ This
system was arrived at independently for the present grammar, but was one of two
alternative systems for a time; the choice between the two was made after finding
the same system in Balmer and Grant, in order that there might be at least some
recognizable contact between this analysis and previous work on the language.
However, after a very brief section in which the tone system is explained and a
few examples of tonal contrasts are given, Balmer and Grant do not mark tones
in the rest of the grammar except in a few isolated cases, mostly paradigms.
Since many tone patterns within segments between two breaks are syntactically
determined, this is not too serious for practical purposes for a native speaker of
the language. In addition, many sequences of segmental phonemes occur with
only one tone pattern, so that the tone pattern is easily recognizable from the
segmental phonemes to one who knows the language. However, such omissions
are not faithful to the phonemic situation, and, even with the help of lists of
syntactic forms with well-defined tone patterns, there are many forms whose
tones are not syntactically but only lexically determined, and such at the very
least ought to be written.
8.2. Consonants are written with some subphonemic distinctions, particularly
for the palatalized alternants. The following table compares the two native
orthographies with the phonemic analysis:
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
28 LANGUAGE DISSERTATION NO. 39, 1946
PHONEMENORMALVARIANT VARIANT
PALATALIZED
(IF DIFFERENTLY WRITTEN)
P P
t t ts
k k ky
b b
d d dz (Old script: ds)
g9 gy
f f
s s
h h hy
w w (Old script sometimes: w)
r r
Y Y
m m
n n ny (also for phonemic /ny/)
kw kw tw
gw - dw
hw - hw (Old script sometimes: fw)
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
PHONEMES 29
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
II. THE MORPHEMES
0. The description of the morphemes of Fanti, the smallest sequences of
phonemes to which meaning can be assigned, will consist of three parts: first, a
study of morpheme alternants and their analysis as members of morpheme units;
second, the classification of all the morphemes into morpheme classes on the
basis of their distribution in relation to each other; and third, a statement of the
phonemic structure of morphemes.
1.0. The primary subject of this section is the positional alternation of mor-
phemes, or, roughly speaking, morphophonemics. However, there are also a
few pairs of free morpheme alternants that deserve mention. The two members
of each pair of morpheme alternants have the same meaning, and occur in the
same, not complementary, environments. They are therefore essentially
different morpheme units, or synonyms; however, their close similarity to each
other, and the fact that the differences between them follow certain patterns in
some cases, makes it preferable to refer to them as "free morpheme alternants".
It is possible that the alternants may represent some dialectical variation, but
there is at present no way of ascertaining that fact with any certainty. In each
case the two members of the pair are probably historically identical. All
recorded cases of free morpheme alternants will be noted in a list of lexical
elements ;8 some typical cases are as follows:
wuma and muwa leather, book, learning
nti'r and nti' reason, source (cf. iti'r head)
aben and amen horn
ibdn and imdn herring
kurd and gwird drinking calabash
sider and sidd or sometimes sird (in fast speech) shilling
If two or more sequences of phonemes having the same meaning occur in
complementary environments, they are positional alternants of one morpheme
unit. In some cases, the phonemic differences apply to every occurrence of a
given phonemic pattern; these illustrate a regular phonological alternation
between phonemes. In other cases, the phonemic differences apply to only some
occurrences of a given phonemic pattern; these illustrate true morphophonemic
alternation. In still other cases, the phonemic differences apply to only one case;
these illustrate morpholexical alternation between phonemes, or, if the alternants
are not phonemically similar, they are cases of suppletion. In the following
sections, all known cases of regular phonological alternation are given, and they
apply regularly to all comparable cases. All known types of morphophonemic
alternation are also listed, but the illustrations given do not exhaust the lists of
8
The list of lexical elements referred to here and elsewhere in these pages is still in
manuscript. It was originally intended as an appendix to the present work; the de-
cision to withhold it is based on the hope that it may be published separately in the
not-too-distant future, in a more complete and useful form than the present version.
30
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
MORPHEMES 31
cases to which each alternation applies; additional cases are noted in the list of
lexical elements. The cases of morpholexical alternation and suppletion are
presumably complete. Morpheme units, apart from alternant forms, are cited
in braces {...}.
1.1. /n/ and /m/ as the final or only phoneme in a morpheme alternate
regularly before close juncture and a following consonant, but do not alternate
before open juncture; /m/ occurs before /p, b, m/, /n/ before all other consonants.
E.g.,
mbd children ma o6tba let him come
ntd twins ma onko let him go
mimpe-nu I don't like it But ahum-gwii peace
minti-nu I don't hear it ntem-tem quickly
1.2. Final /e/ has the alternant /a/ before open juncture or break followed by
/'/, when the preceding consonant is /t, d, s/. In the cases of other consonants
preceding, /e/ remains phonemically as determined by the palatalized alternant
of the c3nsonant. After these three consonants, whose palatalized variantsdo
not occur before /e/, the sound [e] in this position must be analyzed as /a/ (see
Chapter I, 4.2). There are very few statistical possibilities of this alternation,
but two adequate illustrations are:
opetd yi' this vulture
op6td-mu' in a vulture. Both from {o-pet6} 'vulture'.
1.3. Final /a/ usually has the alternant /o/ before open juncture followed by
/mu'/, when the preceding consonant is /kw/ or /w/, or when the preceding vowel
is/u/; the alternation sometimes occurs after /f/, and occasionally after /p, b, m/.
E.g.,
kwo-mu' or sometimes kwd-mu' the bush country. From {kwd} 'farm'.
fo-mu' but usually fd-mW'on the ground, down. From *{fd} 'clay'.
mpapaa-mu' chapters, divisions. From {paa} 'split, divide'.
1.4. Morphemes which end in /w, r, m/ have alternants with /i/ added before
/nV/ or /-ni'/. E.g.,
ofari-ni' fisherman. From {far} 'fishermen's quarter (of a town)'.
niamisumi-ni' servant of God. From {niami} 'God' plus [sum} 'serve'.
But owEn-ni'guide. From {wen} 'to guard'.
1.5. The tones of stem morphemes (see 3.1-2 below) all become low before
close or open juncture followed by another stem; i.e., when the stem is the first
element of a compound. E.g.,
beenku'm-ndn the left foot. Cf. beenku'm'left'.
denkembu diamond. Cf. d&nktm'crocodile'.
fantikdsd the Fanti language. Cf fanti 'Fanti'.
nkati-nkwdn peanut soup. Cf. nkatt 'peanuts'.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
32 LANGUAGE DISSERTATION NO. 39, 1946
This statement does not apply, however, when the second stem is {di 'a thing'
or {bd 'child, young, small'. In such cases the tone of a preceding stem usually,
but not always, remains the same. E.g.,
akirkw-di writing implement. Cf. kirgw 'write'.
iku'r6ba village. Cf. iku'ro 'city'.
But siradi ointment, oil. Cf. sird 'rub on ointment'.
1.6. Nasalization drops before close or open juncture, except before final pro-
nouns and /-mu'/. E.g.,
nsadfu' palm wine. Cf. nsa 'liquor' and afi' 'white'.
akuh6n navy. Cf. ku 'fight' and hUn'ship'.
ni'ansa-nsgm wise sayings. Cf. ni'dnsa 'wisdom'.
But mihA-wu I see you.
1.7. Final lew, owl have the alternant /o/ before close juncture followed by
/i/. E.g.,
akiroi writing, writings. Cf. kirMw'write'.
furoi a stew. Cf. furow 'make a stew'.
1.8. The tone sequence high-mid has the alternant low-high, and the tone
sequence high-low has the alternant low-low in all morphemes when they occur
after a prefixed pronoun with high tone (see 1.16 below). E.g.,
hen-kdsd our language. Cf. kdsd 'language'.
mi-sikdn my knife. Cf. sikdn 'knife'.
nd-awu'ra his master. Cf. a-wu'rd 'master, Mr., sir'.
1.9. The /w/ of final /ow/ drops in some cases before open juncture. E.g.,
apo-mu' (in) the joints. Cf. pow 'knot, joint'.
1.10. Final /r/ in some morphemes only drops before open juncture. E.g.,
ayifdr dowry. Cf. yir 'wife'.
os6fi priest, minister. Cf. sor 'pray'.
But afar-fiu fishermen. Cf. fdr 'fishermen's quarter (of town)'.
1.11. Final /r/ in at least two cases and final /n/ in at least one case occur
only when the morphemes to which they belong occur in combination with other
morphemes. E.g.,
mikori I went. Cf. ko 'go'.
mikori h6 I went there.
mihuni'-nu I saw him. Cf. hi' 'see'.
kikir tie up. Cf. ki 'catch'.
ayifdr dowry. Cf. yir 'wife' plus fa 'take'.
1.12. Three morphemes used as verbal prefixes (see 2.8 below) have different
alternants before different vowels following the next consonant. Of these three
morphemes, the forms {ko} 'go' and {fa} 'take' are used in other environments
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
MORPHEMES 33
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
34 LANGUAGE DISSERTATION NO. 39, 1946
stems, before noun stems, and after nominal prefixes. All of the alternants
which end in a vowel before a consonant must also be compared with alternants
before close or open juncture followed by a vowel; in this case, the final vowel
of the pronoun morpheme is the same as the following vowel, which is /a/ in all
but a very few cases. In the following list of alternants, the assumption is made
that a following vowel is /a/; such forms are listed second.
Nothing is said about tones in the above list. The forms before verb stems
usually have high tone immediately before a one-syllable verb stem with no
other prefixes and no suffix; elsewhere they have low tone, with exceptionsas
noted in 3.2. below. It will be noted also that the forms of the third person
singular and plural before a vowel followed by verb stem are identical. It
should be noted also that the second person plural pronoun is not very commonly
used. In most contexts, either the speaker is included and the first person
plural is used, or both the speaker and the listener are excluded and the third
person plural is used. The second person does exist for certain specific occasions,
but does not occur with the same frequency and under all of the same circum-
stances as it does in many languages. Examples of these alternants of pronouns
are as follows:
iti-mi asi a Do you understand me?
miti-wu asi I understand you.
md-agi'd nui-ponko my father's horse.
owo ho He is there or She is there or It is there.
ma yenko Let's go.
wadko He has gone or She has gone or It has gone or
They have gone.
woriye iben adi What are they doing? or What are you doing?
na humddidi a Have you eaten?
owo ddn-nu nu'-mu' He is in the house (He is at the house its inside).
hon-kdsc onye din kuraa Their language isn't difficult at all.
imi d It's me.
iwu wo hin Where are you?
onu wo hd He's here. or She's here. or That one is here.
ihen We, us.
ihum You (pl.)
ohon They, them.
1.17. A suffix which forms the past tense of verbs has different alternants for
verb stems of different phonemic pattern, and different alternants when in
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
MORPHEMES 35
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
36 LANGUAGE DISSERTATION NO. 39, 1946
rising tone occurs with some other words, usually seeming to add emphasis to
the word; in still other cases a vowel is lengthened without rise of tone. This,
however, does not seem to be a comparable morphological process, but merely a
stylistic lengthening for emphasis. Thus the vowel of /pii'/ 'many, much',
which is always long, is sometimes lengthened almost indefinitely, especially in
story-telling. Similarly /kokoo/ 'red', /bokoo/ 'slowly'.
1.19. A morpheme which nominalizes verb stems consists simply of a change
of tone. Its alternants are not completely predictable on the grounds of phonemic
structure of the stems, but certain statements can be made. For verbs whose
stems have but one vowel, the tone change is from low to high. For verbs
whose stems have two vowels with the tone pattern low-high, the tone change
seems to be usually to high-high. For verbs whose stems have two vowels
with the tone pattern high-low, the tone change seems to be usually to high-mid.
Exceptions to these statements are listed in the lists of lexical elements at the end
of the grammar. This is not a common morpheme, but the following examples
suffice to describe it:
od6 love; mi-do my love; cf. mido I love.
du'a tree; cf. du'd to plant.
kdsd language; hen-kcds our language; cf. kdsd to speak.
1.20. There are several forms of reduplication. With non-verbal stems
reduplication is common, but in most cases the unreduplicated stem does not
occur independently; the most common cases of this are in color terms. With
verbal stems, reduplication may indicate intensive, repeated, habitual, continual,
or causative action. A complete list of all known reduplicated forms is given in
the lists of lexical elements at the end of the grammar. These show the following
alternants; to some extent they are phonemically determined, but between two
groups of alternants there is only lexical determination:
The initial consonant of the stem plus /i/, when the first stem vowel is /i, e, a/:
/didi'/ 'eat, /hwihwe/ 'look for', /kikd/ 'bite habitually, be vicious', /fifi'r/
'perspiration'.
The initial consonant of the stem plus /u/, when the first stem vowel is /o, u/:
/huhur/ 'wash', /fufur/ 'new'.
The initial consonant of the stem plus /in/, when the first stem vowel is
/i, e, a/ and is nasalized or followed by /m, n/: /kinkdn/ 'read', /tintin/ 'long'.
The initial consonant of the stem plus /un/, when the first stem vowel is /o, u/
and is nasalized or followed by /m, n/: /ttntunm/ 'black'.
All of the above alternants occur in complementary environments, but not
with the next, which is the initial consonant of the stem plus the first vowel
of the stem; many cases of the above forms, which have /i/ or /u/ as the stem
vowel, could also be assigned to this alternant; since they have already been
listed, this includes only forms whose first stem vowel is /e, a, o/: /kdkaa/
'toothache', /totootor/ 'chicken pox'.
Also including some of the above forms, but others as well, is the alternant
which consists of repetition of the entire stem: /siasi'e/ 'get ready', /nt6m-tem/
'quickly', /purompurom/ 'prosperous'.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
MORPHEMES 37
There are also other cases of apparent reduplication, including some of medial
or final reduplications. These do not fit into the above patterns, and in every
case the unreduplicated form does not occur independently; in such cases of
doubt, the lexical elements are listed in their occurring forms; possibilities of
reduplication among them are obvious from their appearance. E.g. /fafurdtad/
'butterfly', /fiifi'n/ 'chest, middle'; in connection with the former it may be
pointed out, lest prejudice arise, that some other insect names do have normal
reduplication, but again the stems are bound.
Reduplication must be distinguished from repetition, which contains a phonemic
break, and in which the entire form is repeated, including prefixes. E.g., ndiema
ndiema 'many things, things of all kinds, stuff and things'. Compare the
cases of repetitive reduplication.
1.21. Triplication occurs in a few adjectival forms; in each case, the stem is
monosyllabic, and the final phoneme of the second of the three occurrences of
the stem is repeated. The tone is high throughout. E.g. /dindinndin/ 'terribly
hard'; /tintinntin/ 'awfully long'; /f6feefe/ 'gorgeous'; /dedeede/ 'sugar sweet'.
The meaning is always intensive.
2.0. The morphemes of Fanti may be classified on the basis of their distribution
in relation to other morphemes into twelve classes. The most basic dichotomy of
morpheme classes is between free morphemes and bound morphemes. Free
morphemes are either verbal stems (2.1) or non-verbal stems (2.2). Bound
morphemes are either non-verbal stems (2.3-5) or affixes of various classes
(2.6-12). In the description of the classes given below, illustrations are given of
each of the five types of stems; in the case of the seven types of affixes, the com-
plete list is given. There is some intersection of the classes; at least three of the
verbal prefixes occur also as free verbal stems.
2.1. Verbal stems, all of which are free, are distinctive in their distribution in
that they alone occur before the verbal suffix (2.10), after verbal prefixes (2.8),
after the pre-verbal alternants of the pronouns (see 1.16 above), and before the
other alternants of the pronouns. Not every verbal stem has as complete a
distribution as this, but every one partakes of at least a part of this description,
distinctively from other morphemes. E.g.,
da 'sleep'; oridd 'he is sleeping'; middi 'I slept'.
bo 'strike'; oribo-hon 'he is striking them'.
fund 'get tired'; madfund de 'I'm awfully tired'.
pdsdr 'take a walk'; oriko kepdsdr 'he's going to take a walk'.
2.2. Free non-verbal stems are the class which comprises all other free mor-
phemes; some of them occur with non-verbal prefixes and suffixes (2.9, 2.11),
many occur with other stems, forming compounds, and some occur only inde-
pendently. These facts do not suffice to determine morpheme sub-classes, but
are related to the sub-classification of syntactic elements.
However, there is one sub-classification here. Most of the morphemes in this
class occur sometimes as complete utterances. Others, however, are not free
in this strict sense; there are three types of such morphemes. First, some
stems are bound to a prefix when they occur in isolation or initially in an utter-
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
38 LANGUAGE DISSERTATION NO. 39, 1946
ance, but between breaks within an utterance they occur independently, and
are used in all the environments common to the rest of the class; this group
includes all of the pronouns. Second, there are two stems which do not occur in
isolation or initially, even with a prefix, but whose distribution is comparable to
that of the rest of the class in other respects; these two morphemes are {mu'}
'theinside' and {du} 'the top'. Third, a fewmorphemes donot occur inisolation
nor with prefixes or suffices, but occur between breaks in an utterance; these are
syntactically conjunctions, as {ad} 'who, which', {ni'a} 'he who', {se} 'if', {na}
'but, and then'; this group will be more fully described in connection with the
syntax, Chapters III and IV.
Examples of completely free non-verbal stems and stems which occur with a
prefix only initially are as follows:
denkem 'crocodile'; denkem-nu 'the crocodile'; denkem-bu'diamond'.
kon (the) neck'; nu-kon 'his neck'.
fdr 'the fishermen's quarter (of a town)'; afar-fu 'fishermen'.
bod6m 'dog'; mi-bodom 'my dog'; mbodom 'dogs'.
kesi 'big'; kesinara 'very big'.
ipuin 'table'; nu-pun 'his table'; oriye pun 'he is making a table'.
imi 'I'; mi-nd 'my mother'; ma-mi biuku'-nu 'give me the book'.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
MORPHEMES 39
2.5. Bound non-verbal stems of the third type are those which are bound to
other stems to form compounds. There are several reasons for setting up this
classification as distinct from the classifications of prefixes and suffixes. All
of the prefixes and suffixes are living formations, which may be applied to newly
invented forms; bound sterns, which occur in essentially the same environments,
are distinct in that they are limited to a very few cases and cannot be freely used
to make up new compounds. The prefixes and suffixes are also relational in
meaning, while the bound stems are derivational. The precise meaning of
many of the bound stems in compounds is not known; a sequence of phonemes
is analyzed as a bound stem if it is not a prefix or suffix as defined below in 2.9,
2.11, and if it occurs with a known stem morpheme or with a bound morpheme
whose identity is attested by other occurrences. E.g.,
*{kata} : akatd-si'a 'girl, young lady'; compare basi'a 'woman', bani'n 'man'.
*{si'a}: compare the above forms.
*{kire}: akireba 'sister'; compare {ba in the above forms.
*{ta}: gi'atd 'lion', atdgwi 'tiger nut'; compare gi'a-hin 'lion', ohin 'chief',
agwi 'palm kernel'.
2.6. The remaining seven classes of morphemes are all affixes. The first two
classes, with only one member in each, are derivational; the remaining five are
relational. The first class of derivational affixes consists of the morpheme of
reduplication. The alternants and meaning of this morpheme are described in
1.20 above. It occurs immediately before verbal and non-verbal stems; stems so
reduplicated occur in the same environments as non-reduplicated stems.
2.7. The second class of derivational affixes consists of the morpheme described
in 1.19 above, a change of tone which nominalizes verb stems. This is the only
morpheme in Fanti which never contains any segmental phonemes.
2.8. Relational affixes are divided into two main groups, prefixes and suffixes.
Each of these groups contains verbal and non-verbal classes, and in addition
there is a class of utterance-final particles. The first of these classes of relational
affixes is verbal prefixes. A complete list of these follows, with a statement of
their meanings, tones, and distribution.
{ri} indicates continuous action in present time, or action about to take place.
Its tone is high before a monosyllabic stem, low before all other stems. It
usually occurs immediately before the stem, occasionally before the prefixes
{bo} and {k6}, never before the negative prefix {n }. E.g., oridd 'he is sleeping';
miriko 'I'm going' (at the moment or in just a moment); orikdsd 'he's speaking';
mirikddi' gu'd 'I'm going shopping'.
{d indicates an action or state that has been done or maintained in the past,
with no specification as to whether it is complete or as to the present state of
affairs; in the affirmative it is translated by the English "perfect tense"; in the
negative it is translated by a negative "past tense". Its tone is high beforeall
stems, but low before the negative prefix {n . It occurs immediately before the
stem, before the negative prefix {n}, and occasionally before the prefix {k6}.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
40 LANGUAGEDISSERTATIONNO. 39, 1946
E.g., wadbd 'he has come, they have come'; madkasd 'I have spoken'; madfund
'I'm tired, I have gotten tired'; maandidi' 'I didn't eat'; maanko 'I didn't go'.
{a} indicates an action that takes place after and in relation to a preceding
action already mentioned, or an action that is the purpose of a preceding action
already mentioned. This prefix is apparently not used in the negative. It
always has low tone, distinguishing it from {d} above which always has high
tone in the affirmative; if this prefix does occur before the negative prefix {n},
it is probably phonemically identical with {d} above, which has low tone in that
position. In the first meaning, indicating the second of two consecutive actions,
forms with this prefix are always preceded by a verb and the conjunction {na},
and the prefix occurs immediately before the stem. In the second meaning,
indicating a purposed action, forms with this prefix are always preceded by a
verb; the prefix occurs before the verbal prefix {k6} or before the stem. A
monosyllabic stem following this prefix has high tone, rather than its normal
low tone; the tones of other stems remain the same. E.g., miriko akadidi' na
maabd 'I'm going to go eat, and I'll be right back'.
{n}, with the alternants /n, m/ depending on the consonant that follows,
indicates the negative. Its tone is always low, and a monosyllabic stem following
it has high tone instead of its normal low tone. It occurs immediately before
the stem, and may occur after the prefix {d} or in forms vith the verbal suffix
of the past tense (see 2.10 below). It does not occur with other prefixes; for the
formation of negatives of such forms, see Chapter III, 7.22. E.g., mimpe
nkati-nkwdn 'I don't like peanut soup'; minfunc ntem-tem 'I don't get tired
quickly'; waambd 'he (or she, or it, or they) didn't come'; ombda'he hasn't come.
{n}, with the alternants /n, m/ depending on the consonant that follows,
indicates a cohortative meaning, or an optative meaning for the third person.
Its tone is always high, and the tone of a following monosyllabic stem is high.
It occurs immediately before the stem, or rarely before {bo}, and perhaps before
{ko}. It is usually preceded in an utterance by {ma}, a verb meaning 'give',
but here with a meaning something like 'permit it, arrange it'. E.g., ma yenko
'let's go'; ma onko 'let him go, have him go'; ma wu-nuia inko'let (or have) your
brother go'.
{b6}, with the alternants /bc, be, bo/ depending on the following vowel (see
1.12 above) indicates the future tense. The morpheme may be identical with
the verbal stem {ba} 'come', but this is extremely dubious and, at best, of merely
historical interest. This prefix always has high tone. It occurs before the stem,
occasionally after {ri} or { }, and occasionally before the verbal suffix indicating
past tense (!) when the form is preceded by the verb {ba} 'come' with that suffix.
The tone of a following monosyllabic stem is high in some cases, low in others; this
difference seems to be lexically determined, but there also seems to be some
freedom of usage. This grammar does not include a statement of which mono-
syllabic stems have high tone, and which low tone, after this prefix; the material
at present does not justify a complete statement. E.g., obeba 'he'll come';
obeye hiw 'it's going to be hot'; mib6bud-wTi'I'll help you'; obdd hd bedai 'he
came here to sleep'.
{k6}, with the alternants /kd, ke, k6/ depending on the following vowel (see
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
MORPHEMES 41
1.12 above) indicates that the actor 'goes' to perform the action indicated by the
stem. This morpheme is certainly the same as the verb stem {ko} 'go'. Its
tone as a verbal prefix is precisely the same as that of {bo} above. It occurs
before a stem, occasionally after {ri, d, a, n} as described above, and occasionally
before the past tense suffix. E.g., mikddi' gu'd 'I go shopping, I'm going
shopping'.
{di} and {fd}, the latter of which has the alternants /fd, ft, f6/ depending
on the following vowel, indicate that the actor uses, holds, or takes an object
when he performs the action indicated by the stem. These morphemes are
certainly the same as the verb stems {di, fa} 'take'. Their tones as verbal
prefixes are the same as those of {b6, k6 . They occur before a stem, occasionally
before the past tense suffix, and usually in forms which are preceded by another
verb form with the stem {di} or {fa}, which form is followed by the object to be
used, held, or taken. E.g., fa sikdn fokwi'd ndm-nu 'cut the meat with a knife';
odi buiuku'-nudimad-mi 'he gave me the book'.
2.9. Non-verbal prefixes occur before non-verbal stems, both free and bound,
and before verbal stems plus the tonal morpheme which nominalizesthem.
The non-verbal prefixes are {i, a, o, n, an}. They almost invariably have low
tone, but before pronoun stems and some others they have high tone. Many
stems never occur with a prefix; some occur with two different prefixes in different
meanings; some occur with two prefixes with no difference in meaning; some with
prefixes only in certain positions; and some always with a prefix. There are
some traces of meaning in the various prefixes, but there is no well-defined
limitation to any of the definitions, and there is considerable irregularity. In
general these prefixes, including perhaps the absence of any, seem to indicate
certain classes of nouns; however, the classes overlap in many places. A rough
statement of common parallels of prefixes with certain meanings may be given,
but the exceptions are perhaps more common than the rule.
{o} has often indicates persons; however, many personal nouns have other
prefixes, and many forms with {o} are not personal.
{n}, with the alternants /n, m/ depending on the following consonant, often
indicates plurals or abstracts. This is by no means consistent, but many nouns
which occur with other prefixes or no prefix form the plurals by the use of the
prefix {n}.
{a} often indicates plurals, especially of forms which have the prefix {o} in
the singular; in many cases it is used with forms which refer to an action or state,
as nouns formed from verbal stems. However, many other forms also occur.
This is by far the most common prefix, and, apart from some cases with {n}
and a mere handful with {o} and {i}, is the typical prefix of forms that never
occur without a prefix.
{i} seems to cover no particular class at all, though it is common with forms
that occur only initially with a prefix.
2.10. One verbal suffix exists; it indicates the past tense. Its alternants and
illustrations of its use have been described above in 1.17. It occurs after a verbal
stem, and with some verbal prefixes (see 2.8 above).
2.11. Non-verbal suffixes are six in number:
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
42 LANGUAGE DISSERTATION NO. 39, 1946
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
MORPHEMES 43
final particles. {e} and te} are not as well attested as the others, and may turn
out to be the same, but the following meanings and examples seem clear:
{a} is a question particle, and may be added to any utterance to make a
question out of it. E.g., iriko a 'are you going?', imi a 'Is it I?'
{d} is a particle of identification, used apparently only after nouns. E.g.
imi d 'It's me', agu'd d 'it's a chair, here's a chair'.
{e} is a question particle, used apparently only after nouns, with a meaning
something like 'how about?' E.g., na onu e 'how about him?', wd-apo-mu' e or
wu-hundm-mu' e 'How are you?' (lit., 'how about your body?')
{e} is an exclamatory particle used after some greetings and imperatives, most
commonly in nantiw yi'e e 'Goodbye!' (lit., 'walk well!').
{u'o} is a similar exclamatory particle, much more commonly used. E.g.,
nantiw yi'e u'6 'Goodbye!', mrbu' u'6 'that's the stuff!'
3.0. The phonemic structure of morphemes is a pertinent subject for discussion
because a knowledge of the facts helps in the understanding of many of the tonal
features of Fanti and also in the recognition of morphemes as such. Although
every vowel in Fanti has a definable tone, and although the writing system for
tones relates them to segments between two breaks, the determining factor in the
tonal sequences of an utterance is the morpheme unit. Each morpheme has its
own tones, some of them conditioned by the segmental phonemes of the mor-
pheme, others determined only lexically. The tones of some morphemes undergo
morphotonemic changes in positions adjacent to certain other morphemes.
These changes have been covered in the above sections, with one important
exception which is determined by the syntax (see Chapter III, 7.23 below).
The inherent tones and segmental structure of the morphemes according to their
classes is discussed in this section.
3.1. Verb stems are of three basic tonal types, having the patterns low, low-
high, and high-low. All verb stems which contain only one vowel (including
those with a second vowel that has a zero variant) have low tone, with whatever
changes are stated in 2.8 and 2.10 above. The second type, with tones low-high,
include stems of the structure consonant-vowel-vowel, when the vowels are
different; the first vowel in such cases is always /i/ or /u/. Also included in this
second type are all other stems whose first vowel is /i/ or/u/ followed by /w,r,
m, n/. A final consonant may occur with any stem of this type. The third type
of verb stems, with tones high-low, include all other stems with two vowels.
Stems with more than two vowels, and reduplicated stems, except/hwihwe/
'look for, want' and /didi'/ 'eat', follow either the second or third type depending
on the portion of the stem which includes the last two vowels. Illustrations of
these three types of stems are as follows:
I II III
fa take bud 'help' kdsd 'speak'
sum serve bir ripen pdd split
kwi'a cut pi'rd hurt bi'sd ask
hwi'ew sip from hands kirew write ndntiw walk
ku fight siasi'e get ready kinkdn read
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
44 LANGUAGEDISSERTATIONNO. 39, 1946
No unreduplicated verb stem contains more than two vowels with the possible
exception of {doosiu}'be more than enough'; this may be a series of two verbs,
by comparison with {su} 'be big, be enough'. The above statements, then,
suffice to show that the entire verbal system has automatic tones. It is still
necessary to mark tones phonemically, however, because some of the prefixes
are distinguished only by tone, and because some verb forms are distinguished
from some non-verbal forms only by tone. With a complicated statement of
where such possible ambiguities do not exist, the writing of tones with a large
number of verb forms could be eliminated for practical purposes in a morpho-
phonemic orthography.
In general, it will be noted that all verbal stems, as well as non-verbal stems,
begin with a consonant, with one exception noted below. None begins with /r/.
3.2. Non-verbal stems, whether they are free or bound, show many similarities
in their segmental structure to verbal stems. The only case of a non-verbal
stem beginning with a vowel is the case of the conjunction {aa} 'which, who'.
Stems that contain no more than two vowels have the same segmental structure
as the verb stems noted above; however, many non-verbal stems contain more
than two vowels; usually three, sometimes four, rarely if ever more. Longer
stems may be described as to their segmental structure by stating that they
consist of any combinations of the structure of one- and two-vowel stems. The
tones of non-verbal stems, however, are not determined by the segmental struc-
ture, but lexically, except for the pronouns; see the lists of lexical elements at the
end of the grammar. The commonest tone sequence is one or more low tones
followed by a single high tone, or a single high tone with a monosyllabic stem.
Other patterns, in approximate order of frequency, are two final high tones,
final high-mid, three final high tones, final high-low-mid, and final high-low. A
few stems have only low tones. These patterns can be best illustrated by a
few cases of minimal contrasts:
duta tree; du'd tail; adu'd beans; cf. du'd to plant.
nkda debt; nkdd reminder.
buisum a deity; busumf moon, month.
The tones of the pronoun stems are high after a nominal prefix, high before
non-verbal stems, low after verbal stems, and high or low before verbal stems.
Before verbal stems which have only low tone, and no prefixes, they are usually
high. Before other verbal stems with no prefixes they are low. They are low
also before the verbal suffix of the past tense, before {d, a, n, ri, bo} and other
prefixes except {n}; before {n} they are high. A few additional notes will be
made on this subject in Chapter III, 7.23.
3.3. No prefix or suffix contains more than three phonemes, and most contain
only one or two. All prefixes except the verbal prefixes that occur or perhaps
occur elsewhere as free stems consist of a single vowel, with the further exception
or the non-verbal prefix {an} and the verbal prefix {ri}; even these forms, how-
ever, have a segmental structure different from that of any stem. The suffixes
show a wider range of structure. The tones of prefixes and suffixes have already
been fully discussed for each individually in 2.8-12 above.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
III. THE SYNTACTIC ELEMENTS
0.0. The syntactic elements are the immediate constituents of the constructions
discussed below in Chapter IV. They may be classified in such a way that each
member of a class may be substituted for another without changing the syntactic
relations of the construction.
0.1. The criterion for such classification of syntactic elements is the existence
of at least one distinctive position in a construction in which the members of one
class occur and in which the members of other classes do not occur. The state-
ment of these distinctive positions does not exhaust all the positions for each
class, but only suffices to distinguish the classes. The statement of all the
positions of each class is the syntax of the language, which is discussed in Chapter
IV.
0.2. The classes of syntactic elements must not be considered as "meaning"
classes. To a large extent they happen to parallel meaning distinctions, but
meaning is in no case a criterion for classification.
0.3. The syntactic elements of Fanti vary in length. No attempt is made to
describe them in terms of "words" or "phrases", although such names may be
practical in many cases. The elements may consist of a single morpheme,
often with no phonemic mark to set them apart from adjacent elements; they are
often equivalent to a segment between two breaks, in which case they might
conveniently be called "words"; and in many cases they are segments including
one or more breaks, to which the term "phrases" might conveniently apply,
though it cannot be used in a phonemic sense.
0.4. The following sections state the distinctive position in which each class of
syntactic elements is set up, and the internal structure of each class in terms of its
immediate constituents down to the level of morphemes. The discussion of the
internal structure of the elements involves the setting up of sub-classes in some
cases.
0.5. Syntactic elements are not necessarily identical with lexical elements.
The lexical elements included in the classified lists at the end of the grammar
do not include breaks, except in a very few cases; forms with initial pronouns are
not included, and forms with relational affixes such as the verbal prefixes are not
included. In fact, the relational affixes may be considered as lexical elements in
themselves.
1.0. Interjections are elements which always occur after a stop or pause, and
which always occur before a stop or pause or before an utterance-final particle
followed by a stop or pause. That is, they always occur in isolation, except for
the utterance-final particle that may follow them. All other elements differ from
interjections in that they do not always occur in isolation.
1.1. In segments between breaks, interjections consist of a single stem or a
stem with a non-verbal prefix. E.g.,
du'e. That's too bad! or Take it easy!
hdi'. Hey!
ku'se. Excuse me! (Said after an accidental action.)
45
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
46 LANGUAGE DISSERTATION NO. 39, 1946
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTACTIC ELEMENTS 47
4.2. In segments including one or more breaks, adjectives after a pause consist
of one of the above, except forms ending in {-nui}, followed by {bi'} or {yi'}; in
other positions, they consist of one of the above followed by dudu 'too much,
too', or of two or more of the above joined together by the conjunction {na}.
E.g.,
kesi bi' a big one (referring to something already mentioned)
oye kesi na tuntgi'm it's big and black
kesi dudu too big
5.0. Adverbs are elements which do not occur after pause at all, except in
isolation. All the remaining elements occur after pause with this limitation
or the limitation stated for adjectives.
5.1. In segments between two breaks, adverbs have the same construction as
adjectives (see 4.1 above), except that {-nu} is never added. E.g.,
dem thus mpu'o even
ntem or ntem-temor ntemara quickly
5.2. In segments including one or more breaks, one of the above may be
followed by dudu 'too much, too', as in the case of adjectives, though not all
adverbs occur in this position. Temporal noun groups (see 6.33 below) also
occur in adverbial position, but are not adverbs because they share other positions
with nouns and do not share the limitations of adverbs as defined in 5.0 above.
The syntax of these temporal complexes is discussed more fully in comparison
with adverbs in Chapter IV, 8 below.
6.0. Nouns are elements which do not occur between a pause and an adverb
followed by pause; that is, no utterance consists of only a noun followed by an
adverb. The remaining elements do occur in this position.
6.10. In segments between two breaks, nouns consist of a base which is com-
posed of one or more stems, or Such a base plus non-verbal affixes.
6.11. Starting from the end of such a noun, the last morpheme that occurs is the
definitizing suffix {-nu}. This is usually translated by the English definite
article "the"; however, it is more accurately a remote demonstrative, sometimes
demanding the translation "that", while "the" is sometimes expressed in Fanti
by iyi' 'this'. {-nu} also occurs sometimes after forms which have an initial
pronoun morpheme. Syntactically, it is in a class with the demonstrative noun
iyi', which occurs in non-initial position as yi' (see 6.36 below). It is discussed
here only because of its junctural situation; it is undoubtedly also the same
morpheme as the stem in onu 'that one, he, she, it'. E.g.,
buuitku'-nu the book, that book
nu-buiuku'-nu that book of his
ipufi-ni'-nu the sailor
6.12. The suffixes {num} and {dra} also occur in final position only; however,
they do not occur after {fu } or {n'}, and not freely with all nouns. E.g.,
iyi'-num these
obi'ara anyone
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
48 LANGUAGE DISSERTATION NO. 39, 1946
6.13. The suffixes {fu} and {ni'} and {i} (see Chapter II, 2.11) occur finally
or before {-nu}; the combination {fi-ni'} also occurs (ibid.).
6.14. The arrangement of these suffixes is summarized in the following chart.
Read horizontally, including no more than one member of a vertical column in
one form, and omitting any vertical column optionally.
nu
fu | nit
BASE
nitm
dra
6.15. Starting from the beginning of nouns in segments between two breaks,
the first morpheme that occurs is a pronoun. This combination expresses a
possessive relationship. E.g.,
mi-ddn my house
md-awufu my parents
6.16. Any one of the nominal prefixes {i, a, o, n, an} may occur initially, and
{a, n, an] and very occasionally {i, o} occur after pronouns. These prefixes
are more fully discussed in Chapter II, 2.9 above.
6.20. What remains of nouns in segments between two breaks when the
affixes are accounted for is the base, which in many cases occurs with no affixes
at all. The base consists of a single non-verbal stem, a reduplicated stem, a
verbal stem plus the morpheme of nominalizing tone, occasionally a reduplicated
verbal stem with the same morpheme, a simple or reduplicated verbal stem with-
out nominalizing tone if {i} follows, or a group of two or more stems. All of
these constructions are perfectly simple or have been fully illustrated in previous
sections, except the last; such a group of two or more stems in a noun base is a
compound base. Several types of compound bases occur, as described in the
following sub-sections.
6.21. Compound bases which consist of two non-verbal stems, each of which
occurs elsewhere as the base of a noun or one or both of which is a bound stem,
are by far the most common type. There seems to be no regularity as to which
of the elements is the head and which is the modifier, though perhaps the order
modifier-head is more common. The second element is very commonly {bd}
'child, young, small', {di} 'thing', or {asi} 'the under part'; compounds with
mu' } 'the inside of' may be freely constructed for any object that has an inside,
and similarly for {du} 'the top'. However, that is only a statement of frequency;
there is no special limitation on what second elements occur. In a few cases,
non-verbal prefixes occur with the second element, regardless of whether there is
one before the first. E.g.,
iku'roba village; cf. iku'r6 city; bd child, young.
ahindi kingdom, reign; cf. ohin chief; adi thing.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTACTIC ELEMENTS 49
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
50 LANGUAGE DISSERTATION NO. 39, 1946
ani'da-du hope; cf. ani' the eye; da to lie; dzithe top, on; miwo ani'da-du
dee mibokoI have hope that I can go; I hope that I can
go; the compound seems to mean literally 'the eye
lying on it'.
ani'm-gu'-dsi shame; cf. ani'm face; gu' pour or drop; asi the under part,
down; perhaps 'the falling of the face'.
6.26. A few compound bases composed of two verbal stems have been recorded.
E.g., with the suffix {fu}, wudi'-fi murderer; cf. wu' die; di' partake of, consume
(a stem with a wide range of meaning and many special meanings in certain
collocations). However, this form may have to be included under 6.23 above.
Cf. waddi' ow' 'he has committed murder', where Iwu'} occurs with nominaliz-
ing tone.
6.30. Before describing the structure of nouns in elements which include one or
more breaks, it is necessary to set up certain subclasses of nouns which are
differently used in such segments. As in the case of syntactic elements as a
whole, the criterion for the establishment of these sub-classes is the existence of a
distinctive position in which the members of each class occur. Here also there
is a parallel with "meaning" classes, but again meaning is not a classificatory
criterion.
6.31. Locative nouns are those which do not occur in any segments which
include a break, except that {ho6 'there' and {hd} 'here' occur after personal
nouns. A possible exception to this is the very occasional use of a locative
followed by another noun in an origin-object construction; e.g., iho nsu' 'the
water of that place', aburo-ki'r ki'ente 'foreign cloth'. These, however, seem to
be technical or literary expressions which, judging from the speech of Mr. Mbura,
are not living constructions. They are perhaps used in titles, advertising, and
trade language, which are special linguistic types similar to newspaper English
and do not form a basic part of a descriptive grammar of the spoken language.
Locatives consist of place names and a few other words. E.g.,
6.32. Personal nouns are those which do not occur after a pronoun nor before a
demonstrative, a numeral, or an adjective. They include the pronouns them-
selves and names of persons. The pronouns are not a distinctive sub-class on
any criterion of distribution, but have been identified and named as a distinct
group of morphemes because of their morpheme alternants (see Chapter II,
1.16). Personal names are an interesting subject of discussion by themselves;
Fanti has names for males and females depending on which day of the week they
were born, and also has names given for the order of birth. However, this is
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTACTIC ELEMENTS 51
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
52 LANGUAGE DISSERTATION NO. 39, 1946
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTACTIC ELEMENTS 53
6.43. Any abstract or unit noun may be followed by an adjective; and at least
unit nouns may be followed by a series of two or more adjectives, or by a series of
adjectives joined to each other by (na) (see 4.2 above). E.g.,
bodom tuntu'm (na) kesi kur bi' a (certain) big black dog
ikom kesinara a great famine, great hunger
Adjectives joined by {na} in this construction seem to occur in any order.
Series without {na} are infrequent, and no statement can be made at present as
to the order in which they occur. However, there seems to be a definite order,
perhaps color first, then size, then other characteristics.
6.44. Personal nouns other than pronouns, unit nouns, and probably also
abstract nouns, may be followed by a unit noun with the initial pronoun {nu-}
'his, her, its', or {hon-} 'their'; this expresses a possessor-object relationship.
Included in this category are the very common expressions of local relationship,
in which the second noun is mu' 'the inside', du 'the top', asi 'the bottom', nk&n
'the vicinity', ani'm 'the front', aki'r 'the back', hu 'the circumference, the body',
etc.; these also occur in ordinary series as described in 6.42 above, but with less
frequency; in this construction, preceded by {nu-}, the first noun commonly has
the suffix {-nu} or a prefixed pronoun. E.g.,
md-agi'd nu-ponko my father's horse
buuki' yi nu-di'n the name of this book
owo ddn-nu nu-mu' he's in the house
ohe du'a-nu nd-aki'r he's behind the tree
mi-fi'e wo nu-fi'e nui-nken my house is near his house
oye-nu wo nku'ro-fu ni'ndara hon-ani'm he does it before all the people
6.45. A series of nouns may be joined by the conjunction oni (often ni non-
initially) 'with, and'. Pronouns preceding ni have their pre-verbal alternants.
This construction is used only before active verbs (see Chapter IV, 5 below)
of nouns that accompany each other in the action. E.g.,
mini md-agi'd kori anumabu. My father and I went to Anumabo.
mipe dee mini-wu k6. I'd like to go with you.
mi-ni'a ni md-agi'd riye agwu'ma. My brother is working with my father.
6.46. A series of nouns may be joined by the conjunction na 'and' in other
zonstructions; in series with more than two members, a pause usually precedes
each na, and na may be omitted entirely. E.g.,
md-agi'd na mi-nd ti ho. AMyfather and mother live there.
mihuni' wd-agi'd na wu-nd. I saw your father and mother.
yewo mbodom, (na) mponko, We have dogs, horses, pigs, and goats.
(na) mpu'reko, (na) aponkt.
7.0. The remaining syntactic elements are verbs; they are those which, unlike
all others, sometimes occur between a pause and an adverb followed by a pause;
that is, an utterance may consist of only a verb and an adverb.
7.1. In segments between two breaks, verbs consist of a base alone or of a base
plus verbal affixes (Chapter II, 2.8, 10). A base consists of a single verbal
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
54 LANGUAGE DISSERTATION NO. 39, 1946
stem or a reduplicated verbal stem, and nothing more. Pronouns in their appro-
priate alternants (see Chapter II, 1.16 above) precede verbal forms with close
juncture, or follow them with open juncture; however, they are syntactically
nouns, and not part of the verbal form at all.
7.20. Verbal bases in relation to affixes are of two types; some are extremely
limited in their occurrence with affixes, others are unlimited. The first group is
syntactically a class of stativeverbs(see Chapter IV, 4 below); the second group is
syntactically a class of activeverbs (see Chapter, 5 below).
7.21. Some stative verbs occur with none of the verbal affixes; these all happen
to be monosyllabic, and pronouns before them are always low in tone instead of
the normal high; in one case the verbal stem itself has high tone. A few other
stative verbs occur with only the future prefix {bo}; all of these happen to mono-
syllabic stems with high tone also, and there are no other stems of this kind.
The verb stem {ye} 'be described as' also belongs in this class; it occurs with the
future prefix {b6} and the perfect prefix {d}; it deserves special mention because
it also occurs as an active verb in the meaning 'do, make', with all affixes. E.g.,
wo (alternant in negative: ni) 'have, be located at'.
owo si'ka. He has money.
owo ddn-nu nu-mu'. He is in the house.
For other tenses, compare ni'a 'get', ko 'go to', ba 'come to'.
ni 'be identical with'.
mi-ke'wni yi'. This is my hat.
md-agi'd ni 6nu. He is my father, or That is my father.12
ti 'live at, be located at'.
miti ho. I live there.
For other tenses, compare tind 'live at, be located at'.
di 'take, use, hold'.
odi sikdn. He has a knife.
midi sikdn dikwi'd ndm. I use a knife to cut meat.
For other tenses, compare fa 'take'.
ni'mh 'know'. For other tenses, compare hif 'see'.
yite 'be good'.
mu'6 'be bad'.
sA 'be large'.
ye 'be described as'.
oye kew. It's a hat. Compare oye kew. He makes a hat.
12 The old translation of the Bible into Fanti renders I John
4:8b as niami ni od6 ' God is
(identical with) love', which may mean also 'Love is God'. It is reported, however, that
most ministers cite the verse as niami ye od6 'God is (to be described as) love', and it is
also reported that the new translation of the Bible about to be or recently published has the
latter form. The first is good Christian Science teaching, the latter is the only acceptable
orthodox translation. One wonders whether the early translators, who were Europeans
assisted by native informants, were quite aware of this distinction in Fanti. The new
translation, in the hands of natives to a much larger extent, may show many other improve-
ments over the old, besides the orthographic change which was apparently the impetus for
the retranslation.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTACTIC ELEMENTS 55
(none) wofure-wu den. What's your name? onfund. He doesn't get tired.
...i mikirew kurddtd. I wrote a letter. minfindi. I'm not tired.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
56 LANGUAGE DISSERTATION NO. 39, 1946
7.23. The above forms show the usual tone of verbs. There is, however,
another entire set of verb tones which is used in subordinate constructions (see
Chapter IV, 9). This second tone pattern is completely regular in its relation
to the normal one assumed above, and can be very simply stated: any low tones
before the first high tone in the above forms, including the tones of pronouns,
change to high; any low tones following the first high tone change to mid; there
are no low tones in any form in this pattern. This raised pattern need not be
given for the entire list, but only illustrated:
mipe dee miboko. I want to go.
miwo ani'da-duid6e oribd hd. I hope he's coming.
ma mirikd yi', ... As the story goes (in folk stories).
abir dad bad hd-nu, ohuni' wd-agi'd. When he came here, he saw your father.
7.3. In segments which include one or more breaks, verbs include primarily
the negative forms included in the list above. Other sequences of two or more
stems are more conveniently described as members of different syntactic con-
structions, and are discussed as verbs in series in Chapter IV, 7, below.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
IV. THE SYNTAX
0. The syntax is the statement of all the positions in which each class of
syntactic elements occurs. This is a more complete statement than the descrip-
tion of one distinctive position in which each class occurs, which was used to
differentiate the classes, unless a given class occurs in only that one position.
In describing the syntax, it would be possible to list all the positions of each
class in order, with sub-classes treated under each class; however, such a descrip-
tion would involve a great deal of duplication, and would not be conducive to
ready understanding of the features of arrangement. Accordingly, a different
procedure is followed in this chapter. Complete utterances with the smallest
number of elements and the simplest structure are described first, according to
their different sentence types, and elements which are added to those sentence
types in different ways are discussed later. Sections 1-5 are descriptions of the
basic sentence types; sections 6-9 are descriptions of variations on these themes
by the use of added elements.
1.0. Exclamations are utterances which consist of only one element, or of one
element plus an utterance final particle. They are not necessarily "exclama-
tions" in the traditional grammatical sense of the word, for they represent a
wider range of usage than mere interjections.
1.1. Interjections, with or without utterance-final particles, are all exclama-
tions, for they always occur in isolation. See Chapter III, 1.0-2.
1.2. Any other syntactic element except a verb with affixes, an utterance-
final particle, or a conjunction, may occur in isolation as an exclamation. Some
of these are exclamations in the more traditional sense of the word, and may be
followed by the exclamatory utterance-final particles; others with the interroga-
tive utterance-final particles form questions; and others, without utterance-final
particles, are used primarily as citations or as answers to questions. E.g.,
afipd (u'o). Happy New Year!
ayir-pd (u'6). Congratulations (on your marriage)!
imi d. It's me.
onu d. It's him.
6nu e. How about him?
nsu' a. Water?
wdna. Who?
md-agi'd na mi-nd. My father and mother. (e.g., in answer to "Who?")
2.0. Questions, including some of the above exclamations, are sentences of any
type discussed in this chapter which contain either the utterance-final particle
a, the conjunction an&ein final position, or a noun which never occurs before
the particle a; such nouns may be referred to as interrogatives.
2.1. Any affirmative sentence may be made into an interrogative sentence
or a question by the addition of a. In many cases the conjunction na is also
used at the beginning of the question, with a free variant and in this construc-
tion only; occasionally a is omitted when these occur. Questions of choice may
be formed by adding to a statement the conjunction anee 'or...?' (sometimes
57
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
58 LANGUAGE DISSERTATION NO. 39, 1946
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTAX 59
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
60 LANGUAGE DISSERTATION NO. 39, 1946
4.3. Noun-Verb-Noun constructions occur with all the other stative verbs.
E.g.,
owo ho. It's over there.
omdnhin nu-ddn wo hin. Where is the paramount chief's house?
owo aki'r aki'r. It's far away.
onni hd. It isn't here.
nku'ro akesi pWi'nni si'ka-mbiu-du. There aren't many big cities on the
Gold Coast.
miti mpuanu. I live at the shore.
oti bakd-nu nd-aki'r. He lives on the other side of the lake.
na wd-abusu'd-fi ni'ndara woti asi a. Is your family all living?
wd-agi'd ti asi a. Is your father living?
nyi ew. oti nu-mpo-mu. Yes, he's well.
koto nuhu ti den. What does a crab (or lobster) look like?
5.0. Active verbal sentences are by far the commonest sentence type in Fanti.
They are of three kinds, each of which uses a different sub-class of verbal stems.
There is a considerable amount of intersection between these classes, however,
Many verbal stems are used in two or more of the constructions, and it is likely
that all active verbs may be used in the first construction, and that all used in
the third may also be used in the second. Any of these constructions may be
made into a command by omitting the actor noun and using a verb with no
affixes; such a construction is rare in the first type, however, and perhaps never
occurs with verbs that are not used in the other two constructions.
5.1. Actor-Action constructions are formed by a noun and a verb alone.
Many of the verbs used in this construction are translated into English by
"get" or "become" plus an adjective in the present, and by "be" plus an adjective
in the perfect. It is important to remember, however, that these are not stative
verbs in Fanti, even though they are translated by stative constructions in
English. E.g.,
madfund. I'm tired.
adi dsa. It's evening.
akutu'-nu mbirl. The orange isn't ripe.
busum-nu dpu'ee. The moon is out.
awi'-mu' rimu'nd. It's getting cloudy.
wadfon. He's gotten thin.
abir dst. or abir dben. Time's up.
wadbin ma. He's very clever.
nsi' dwtw. The water has dried up.
adi'bdn dbin. The food is done.
kur ti dk&d. There's only one left.
orihi'm. He's trembling. (as from a chill).
nsui-nu rihu'r. The water is simmering. (at the boiling point.)
nsu'-nu rihuhu'r. The water is boiling. (violently.)
md-ani' dkum. I'm sleepy.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTAX 61
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
62 LANGUAGE DISSERTATION NO. 39, 1946
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTAX 63
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
64 LANGUAGE DISSERTATION NO. 39, 1946
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTAX 65
na ?ib6tu'm ?asi' mpuaniu?fi' tdkoradi Can you go along the shore from Tako-
Oak6nkiran a. radi to Accra?
kwasi' mpu'o ?tut'm?ye. Even Kwesi can do it.
?wontu'mini-hon ?nye hwii. You can't do anything with them.
Cf. on'mhsikdn siw ma. He's an expert at sharpening knives.
7.6. Series of which the first member is di 'take, hold, use', or fa 'take', are
perhaps the most common verbal series in Fanti. The stem di is stative, fa is
active; there seem to be other differences as well, roughly that di refers to picking
up or using something that is one's own or near at hand, while fa refers to taking
something offered or named especially with the purpose of moving it. This is
not an ironclad definition, however.
na iyi' ?wodi ?ye iben adi. What does one do with this?
?midi ma-ani'wa abi'en yi' ?huni'. I saw it with my own two eyes.
?wodi biru' of drdta ?ba hd. They bring coal here from Lagos
(Nigeria).
se Oidiki'etke ofi'tadkoradi
?rik6nkiran If you take a train from Takoradi to
a, Oikwi'd-mu?wotakwd,kumdasi, Accra, you pass through Tarkwa,
ani'nd-asi, ansa na idu'r nkiran. Kumasi, and Enyinasi before you
get to Accra.
?odi nfoni'n ?dsdn bdn-nu nuihu. He's hung a picture on the wall.
?midi akedi ?kemadmi-nianku. I took a present to my friend.
ddtir-nu ?di adikikiri ?tu mi?-ndndu. The doctor put a bandage on my foot.
?fa igu'd-nu fdsi' pun-nu nu-nken. Put the chair at the table.
?fa nsu'-nu "foku' kwansd-nu nu'-mt'. Put the water in the pot.
?fa ?tuu ha. Put it here.
?wadfdnihu "ddi'. He's free.
?fa dddgui-nutuu dbuu-nu nu-mu'. Padlock the door.
?minni ntui'm ?nfd ?mbd. I can't bring it.
7.7. The verb ker 'to be a long time' occurs commonly as the second verb in a
series; it also occurs alone, but less frequently.
?otind hd ?kerl. He stayed here a long time.
Ookori?dker. He went a long time ago.
?miburz?dkerma. It's been a long time since I've been swimming.
na nsu'-nu ?wadto?dkera. Has it been raining long?
?minni ntu'm ?nkwen?nker. I can't wait any longer.
7.8. The verb form kepim 'go up to, until' is used only as the second element
in a series.
?mikasddhdusa ?kepimabir dd midi' nfi awokwi. I spoke Hausa until I was eight.
7.9. The verb-noun dpd hi 'has passed' is used only as the second element in a
series, of time only.
?wadbonnon-esi'd ?dpa hu miniti adu'asd. It's 63:0.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
66 LANGUAGE DISSERTATION NO. 39, 1946
7.10. The verb form beye 'will be' is used with the meaning 'about, approxi-
mately' as the second element of a series.
?wadk6?bgyedapen. He's been gone about a week.
?waddi'nfi ?beyeadu'. He's about ten.
7.11. The verb-noun da hu is used in the meaning 'still' as the first element
of a series.
?mida hi ?kkoh6 ddbaa. I still go there often.
?miwo akireba baasa ?wodahu ?ti asi. I have three sisters still living.
na se iki'na ?oda hi ?ye-mi ydd bi'o a, If it still gives me pain tomorrow, I'll
?mib6kl ?maak6hf' ddtir bi'o. go see the doctor again.
7.12. The verbs kin and sin 'surpass' are used as the second element of a
series in the meaning 'than'.
mihu ?ye din ?kin-nu. I'm stronger than he is.
7.13. A few series of verbs are composed of elements so commonly used
together, and not often with other syntactic elements separating them, thatthey
at least have the status of separate lexical elements.
?ka ?kire-nu. Tell him.
minni as6m bl'ara dd ?mibekd?akire-nu. I don't have anything to tell him.
?okd4wiuhuasem ni'ndara ?kire-mi. He told me all about you.
?tuu "kin. Throw it away.
?tuu kurddtd-nu?kin. Throw the paper away.
?mingi'g ?ndi'. I don't believe it.
Oso?hwe. Taste it (to see if it's allright).
?oris6 ?ahw6se nkwdn dbin a. She's tasting to see if the soup is done.
?oriso nkwdn ?ahwe. She's tasting the soup.
8.0. Adverbs and temporal noun expressions, along with a few other construc-
tions, are added to the above constructions initially or finally as described below.
8.1. True adverbs are of two kinds; the first are used after any word for
emphasis. E.g.,
nu-bu ye din didu. It's too expensive.
nsi' diud wo-mu'. There's too much water in it.
mipe-nu dudu. I like it very much.
se mini'dd adagir mhpu'oa, nke minni nye. Even if I had had the chance, I
wouldn't have done it.
ondii' nfi dui mipu'6. He's not even ten years old.
onu mpu' tiu'mye. Even he can do it.
8.2. Other adverbs, along with a few adjectives and noun constructions,
occur after a verbal construction. Final nominal constructions included in the
sections below are not included here.
ose-nu kuraa. It's just like him.
ony6 din kuraa. It isn't hard at all.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTAX 67
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
68 LANGUAGE DISSERTATION NO. 39, 1946
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTAX 69
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
70 LANGUAGE DISSERTATION NO. 39, 1946
mini-wu b6ko, na mipe dee mididi' wi'e. I'll go with you, but I want to eat first.
itu'i' kinkdn na kirew fanti a. Can you read and write Fanti?
wadsiid fanti nfi abi'en, ?a su ontiu'm He's been studying Fanti for two years,
nkdsd. but he still can't speak it.
madye ndiemba pii' dmd-wu, na owo I've done a lot for you, so you ought
dee wu-su iye yt' ma-mi. to do this for me.
mi-n't'a kur ti fi'e, na kur-nu suiti gu'ad.One of my brothers lives at home,
the other in Cape Coast.
oyee agwz'ma diindnndin, na mbu'om He worked hard, but he still couldn't
wantit'rhaner, nd-abusu'd. support his family.
miwo ani'da-du dee obeba,na su mi-wire I hope he'll come, but I'm not sure.
nni'nm.
igi'd-nu ddui'm, na mbu'om oda hu The fire has gone out, but it's still
ripu'm wuS,uI. smoking.
iku'ro yi' yi'e, na mbuo'm wonni seni This town is O.K., but they don't have
wo hd. any movies.
minantiwi donhwir kur, na nd-aki'r-nu I walked for an hour, and then I
migi'ee md-ahum. rested.
wowuu2-miwo si'ka-mbiu du, na mindo- I was born on the Gold Coast, and
m.u'da. I've never been anywhere else.
9.4. The conjunction anee (or freely anda) 'or' is used to coordinate single
elements or whole constructions. With whole constructions it is used usually
if not always in questions. The combination anee su 'or else, or also' also occurs.
owo addn abi'dsd anee andn. He has three or four houses.
minni si'ka anee su adi'bdn. I have neither money nor food.
mibemd-wulmpua anee su akutu'. I'll give you a banana or an orange.
na opi'rd-wi bi'o anee. Does it still hurt?
na oye abofurdanda oye pani'n. Is he young or old?
na oye bibini' anee (oye) buro-ni'. Is he an African or a European?
9.5. The subordinating conjunction ma usually expresses purpose or result,
but is used also after constructions expressing a lack. It always occurs after the
head construction. A following verb often has the prefix {a-}. There is little
doubt but that this conjunction is morphemically the verb {ma} 'to give', but
it is considered a conjunction because of its special uses.
dkd akurd ma waadu'r ho. He's almost there.
dkd akurd ma abir aben. Time's almost up.
dkd akurd ma maawi'e. I'm almost finished.
abir-nu dd mini kwaku' atd kobur-nu, When I went swimming with Kweku-
nke dkd kitikiti ma ipu afd-mi. Atta, I almost drowned.
onni nker b'ara ama ma hen-hu aye din We'll soon be stronger than they are.
akin-hon.
oyge ma waani'a si'ka. He did it to make money.
oyee agwi'ma ma waatiu'rh waan&h He worked to support his family.
nd-abusu'd.
mipa-wu kew kire-mi ma mibasi' dd Please tell me how to get to your home.
miboko wufi'e.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
SYNTAX 71
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
72 LANGUAGE DISSERTATION NO. 39, 1946
ni'a wadko sukiu'r-nuowo kwdn pii'dra A man who has been to school has many
dd obesi'-mu' wadni'd nuhu.i opportunities to get rich.
9.8. Subordinate constructions placed either before or after the head construc-
tion are introduced by the conjunctions se ... a 'if, when', kdnsa ... a 'even if,
although', osi'dn dee 'because', and ansd na 'before'. {se} is sometimes omitted,
with the final {a} remaining. Compare also the initial temporal construction
beginning with abir da and ending with the suffix {-nu}, in which abir dd may
be omitted.
se 6bd a, nib6kofi'e. When he comes I'll come home.
se aambd a, nu-bu b6fu'. If you don't come, he'll be angry.
se miwi'e a, miko fi'e. When I'm finished, I go home.
se ini'a adagir a, iye den. What do you do in your spare time?
se mini'a adagir a, miboko. I'll go as soon as I can.
mididi' wi'e a, mini-wu boko. I'll go with you when I'm through
eating.
se ikeri miniti adu' a, nke inni ntuu'-mi. If you had come ten minutes later,
you wouldn't have met me.
owo degemiko6fi'd-adi, na se miwi'e a, First I have to go to the store, and
mirikobur. when that's over I'm going swim-
ming.
na se 6d1' nfi anu'm a, ob6ko suku'r. (And) when he's five, he's going to go
to school.
kdnsa waambdrMpu'6a, mib6ko. Even if he doesn't come, I'll go.
mipe dee mitind si'ka-mbiu du, osi'dn I prefer to live on the Gold Coast,
dee oye awo duduiwo aburo-ki'r. because it's too cold in Europe.
oni'a nuhufpii', osi'dh dee oybeagwu'ma He got rich because he worked hard.
dindinndin.
mihuni'-nu ansd na orihu'-mi. I saw him before he saw me.
ansd na mibad hd-nu, mitfi aburo-ki'r. Before I came here I lived abroad.
ansd na iboko-nu, mipe dee mini-wu di' I want to talk to you before you go.
nkom6.
na 6ker ansd na idu'r h6a. How long does it take you to get there?
or Does it take you long to get
there?
owo dee iwi'4 adi da iriye ansd na aako. You have to finish what you're doing
before you go.
nki'etek6-nu 6ker ansd na waadu'r ho a. How long does the train take to get
there? or Does it take long to get
there by train?
wohu' besi'a hu ansd na wotuu mpd-nu. They (make sure they) see the woman
before they make the bed. (A
proverb, "Don't count your chick-
ens before they're hatched".)
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
APPENDIX
TEXTS
2. Elmina castle
The following conversation was held with Mr. Mbura in an imaginary situation
in which he talks with a visitor to the Gold Coast. Mr. Mbura's speech is
marked by the initial M.; the visitor's remarks, checked for accuracy with
Mr. Mbura, are marked by the initial W.
M. ana adht' idind abdn-nu a.
W. 6o'ho. kd niuhu nsem kir-mi.
M. osA u'o. oye abasd-mu' b6ye aduasi'd tintin, nu-tdtar-mu' su oye abasd-mu'
b6yeadu'asd, na addn pii' wo-mu', na suipird pii' sisi' abdn nu'-du. Portugese-
fu aburofu-nu na wosti' nff beye ahd andn dbesin yi' nu-mi'. wodi abu sii'. osi'
mpuanu ho dra, na se idi wusu'-h6n ba si'ka-mbiu du a, ib6tu'm ahu' demhabdn
yi' wo pui-mu'ho.
W. na woda hi ye ndiemba wo-mu' a.
M. niyi'ew. Postoffice ddn-nu wo-mu', nd su siasi'aara oye ahohuddn ma
ngirasi' aburofi-nu. ngirasi' aburofu-nu nu-mu' bi'-num, se woma-hon kwdn a,
wokdgi'd hon-ahum wo idind abdn-nu nu-mi'.
73
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
74 LANGUAGE DISSERTATION NO. 39, 1946
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
TEXTS 75
And when this is done, what comes next is that they are 'introduced'.6When this
is done, then the young people are husband and wife.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
76 LANGUAGE DISSERTATION NO. 39, 1946
except himself. So Kweku Ananse began to think how he would go about taking
the wisdom out of his country so that only he would be a wise one there. And
finally this thought came into his head: "Why don't I take this wisdom and put
it in a pot, and seal the mouth and bury it?" Well, Kweku Ananse made a pot
and tied up all the wisdom and put it inside, and sealed the mouth of the pot.
And then he changed his mind, and took a canoe instead, and set out to go to
the coconut palm that was in the middle of the river. His idea was this: "If
I take this wisdom and put it in the ground, perhaps someone will stumble over
it and get the wisdom; but if I take it and hang it in the branches of the coconut
palm, no one will ever touch it." So he took a rope and tied it to the neck of
the pot, and began to climb the tree. Well, Kweku took (the pot) and he'd be
climbing a little and slipping a little. And finally, when he stretched out his
hand to tie the pot to the branches, he slipped so that he almost fell to the ground.
So he quickly grabbed hold of the coconut tree with both his hands, and when he
did that, the pot slipped from his hand and fell into the canoe and broke. And
the wisdom that was in the pot spread over all the world. That is why today
there is widom in the world. And Kweku Ananse, when he saw what he had
done, was terribly ashamed. [So he jumped up on the celing to hide,]8 and that
is why today the spider sticks to the ceiling. The end.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
TEXTS 77
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions
78 LANGUAGE DISSERTATION NO. 39, 1946
through, he told them to get up, and they couldn't. So he had them try again
to get up, and still they couldn't. So Kweku Ananse began to laugh again and
to tease them, and he began to make fun of them. After this, he sat down and
proceeded to eat. Well, Kweku Ananse ate away for three days, and when he
was through chewing on the bones, he threw them in their faces. And sometimes
Kweku Ananse would take some soup and pour it over their faces, and made
them a very pitiable sight indeed. And after a while Kweku set out to go back
to his land. And then it began to rain terribly hard, so that the animals were
able to get up, and right away they swore that if they got hold of the spider, they
would tear him limb from limb. So they began to chase the spider. It wasn't
long before they caught him, and the wolf began to tear him up. The spider
began to laugh again, and told them that if they wanted him to be quiet they
should throw him up in the air three times and let him fall. And the fools
believed (him) and did it. When they threw Kweku up, he would fall down, but
meanwhile the spider was looking for a place to run and hide. So when they
threw the spider up again, he never did fall down, but stuck to the ceiling. That
is why today the spider sticks there. The end.
This content downloaded from 130.15.241.167 on Thu, 17 Dec 2015 01:23:40 UTC
All use subject to JSTOR Terms and Conditions