(Download PDF) Language and Reading Disabilities 3Rd Edition Allyn Bacon Communication Full Chapter PDF
(Download PDF) Language and Reading Disabilities 3Rd Edition Allyn Bacon Communication Full Chapter PDF
(Download PDF) Language and Reading Disabilities 3Rd Edition Allyn Bacon Communication Full Chapter PDF
https://ebookmass.com/product/etextbook-978-0137021062-clinical-
phonetics-the-allyn-bacon-communication-sciences-and-disorders-
series/
https://ebookmass.com/product/teaching-reading-to-students-who-
are-at-risk-or-have-disabilities-enhanced/
https://ebookmass.com/product/top-50-sat-reading-writing-and-
language-skills-3rd-edition-brian-leaf/
https://ebookmass.com/product/reading-picture-books-with-infants-
and-toddlers-learning-through-language-jane-torr/
Vagueness and Thought Andrew Bacon
https://ebookmass.com/product/vagueness-and-thought-andrew-bacon/
https://ebookmass.com/product/500-sat-reading-writing-and-
language-questions-to-know-by-test-day-third-edition-inc/
https://ebookmass.com/product/elsevier-weekblad-
week-26-2022-gebruiker/
https://ebookmass.com/product/critical-care-eeg-basics-rapid-
bedside-eeg-reading-for-acute-care-providers-
feb-29-2024_1009261169_cambridge-university-press-jadeja/
https://ebookmass.com/product/language-intervention-strategies-
in-aphasia-and-related-neurogenic-communication-disorders-5th-
edition-ebook-pdf-version/
Language and Reading Disabilities
Alan G. Kamhi Hugh W. Catts
Third Edition
Pearson Education Limited
Edinburgh Gate
Harlow
Essex CM20 2JE
England and Associated Companies throughout the world
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted
in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without either the
prior written permission of the publisher or a licence permitting restricted copying in the United Kingdom
issued by the Copyright Licensing Agency Ltd, Saffron House, 6–10 Kirby Street, London EC1N 8TS.
All trademarks used herein are the property of their respective owners. The use of any trademark
in this text does not vest in the author or publisher any trademark ownership rights in such
trademarks, nor does the use of such trademarks imply any affiliation with or endorsement of this
book by such owners.
Table of Contents
I
This page intentionally left blank
Language and Reading:
Convergences and Divergences
I
t is now well accepted that reading is a language-based skill. At that time, the idea that most reading
disabilities were best viewed as a developmental language disorder was an emerging one. A develop-
mental language perspective of reading disabilities was the major theme of our original book and
continues to be the major theme of the present text. This view rests, in part, on the fact that there are nu-
merous similarities between spoken and written language. Reading shares many of the same processes
and knowledge bases as talking and listening. Reading, however, is not a simple derivative of spoken lan-
guage. Although spoken language and reading have much in common in terms of the knowledge and
processes they tap, there are also fundamental, nontrivial differences between the two. Knowledge of the
similarities and differences between spoken language and reading is critical for understanding how chil-
dren learn to read and why some children have difficulty learning to read. In this chapter, we begin by
defining language and reading. This is followed by an in-depth comparison of the processes and knowl-
edge involved in understanding spoken and written language. Other differences between spoken and
written language are then discussed.
DEFINING LANGUAGE
Definitions of language are broad based and highly integrative. An example of such a definition is of-
fered by the American Speech-Language-Hearing Association (ASHA, 1983):
Language is a complex and dynamic system of conventional symbols that is used in various modes
for thought and communication. Contemporary views of human language hold that: (a) language
evolves within specific historical, social, and cultural contexts; (b) language, as rule-governed be-
havior, is described by at least five parameters—phonologic, morphologic, syntactic, semantic, and
pragmatic; (c) language learning and use are determined by the interaction of biological, cognitive,
psychosocial, and environmental factors; and (d) effective use of language for communication
requires a broad understanding of human interaction including such associated factors as nonverbal
cues, motivation, and sociocultural roles. (p. 44)
From Chapter 1 of Language and Reading Disabilities, Third Edition. Alan G. Kamhi, Hugh W. Catts.
Copyright © 2012 by Pearson Education, Inc. All rights reserved.
1
Language and Reading: Convergences and Divergences
As reflected in the definition, it is generally agreed that there are five parameters of lan-
guage. These parameters are described briefly in the next section.
Phonology
Phonology is the aspect of language concerned with the rules that govern the distribution and
sequencing of speech sounds. It includes a description of what the sounds are and their compo-
nent features (phonetics), as well as the distributional rules that govern how the sounds can be
used in various word positions and the sequence rules that describe which sounds may be com-
bined. For example, the // sound that occurs in the word measure is never used to begin an
English word. Distributional rules are different in different languages. In French, for example,
the // sound can occur in the word-initial position, as in je and jouer. An example of a sequence
rule in English would be that /r/ can follow /t/ or /d/ in an initial consonant cluster (e.g., truck,
draw), but /l/ cannot.
Semantics
Semantics is the aspect of language that governs the meaning of words and word combinations.
Sometimes semantics is divided into lexical and relational semantics. Lexical semantics involves
the meaning conveyed by individual words. Words have both intensional and extensional mean-
ings. Intensional meanings refer to the defining characteristics or criterial features of a word. A
dog is a dog because it has four legs, barks, and licks people’s faces. The extension of a word is
the set of objects, entities, or events to which a word might apply in the world. The set of all real
or imaginary dogs that fit the intensional criteria becomes the extension of the entity dog.
Relational semantics refers to the relationships that exist between words. For example, in
the sentence The Panda bear is eating bamboo, the word bear not only has a lexical meaning, but
it also is the agent engaged in the activity of eating. Bamboo is referred to as the “patient”
(Chafe, 1970) because its state is being changed by the action of the verb. Words are thus seen as
expressing abstract relational meanings in addition to their lexical meanings.
Morphology
In addition to the content words that refer to objects, entities, and events, there is a group of
words and inflections that conveys subtle meaning and serves specific grammatical and prag-
matic functions. These words have been referred to as grammatical morphemes. Grammatical
morphemes modulate meaning. Consider the sentences Dave is playing tennis, Dave plays ten-
nis, Dave played tennis, and Dave has played tennis. The major elements of meaning are similar
in each of these sentences. The first sentence describes an action currently in progress, whereas
the next sentence depicts a habitual occurrence. The last two sentences describe actions that have
taken place sometime in the past. What differentiates these sentences are the grammatical mor-
phemes (inflections and auxiliary forms) that change the tense and aspect (e.g., durative or per-
fective) of the sentences.
Syntax
Syntax refers to the rule system that governs how words are combined into larger meaningful
units of phrases, clauses, and sentences. Syntactic rules specify word order, sentence organiza-
tion, and the relationships between words, word classes, and sentence constituents, such as noun
phrases and verb phrases. Knowledge of syntax enables an individual to make judgments of
2
Language and Reading: Convergences and Divergences
well-formedness or grammaticality. For example, all mature English speakers would judge the
sentence The boy hit the ball as well formed and grammatical. In contrast, the sentence Hit the
boy ball the would be judged as ungrammatical. It should be apparent that knowledge of syntax
plays an important role in understanding language.
Pragmatics
Pragmatics concerns the use of language in context. Language does not occur in a vacuum. It is
used to serve a variety of communication functions, such as declaring, greeting, requesting infor-
mation, and answering questions. Communicative intentions are best achieved by being sensitive
to the listener’s communicative needs and nonlinguistic context. Speakers must take into account
what the listener knows and does not know about a topic. Pragmatics thus encompasses rules of
conversation or discourse. Speakers must learn how to initiate conversations, take turns, maintain
and change topics, and provide the appropriate amount of information in a clear manner.
Different kinds of discourse contexts involve different sets of rules (Lund & Duchan, 1993;
Schiffrin, 1994). The most frequent kinds of discourses children encounter are conversational,
classroom, narrative, and event discourses.
DEFINING READING
Reading, like spoken language, is a complex cognitive activity. Gates (1949), for example, de-
fined reading as “a complex organization of patterns of higher mental processes . . . [that] . . .
can and should embrace all types of thinking, evaluating, judging, imagining, reasoning, and
problem-solving” (p. 3). A view of reading that emphasizes higher-level thinking processes is a
broad view of reading (Perfetti, 1986). Thinking guided by print is another way to characterize
a broad view of reading. Reading ability defined in this way is associated with skill in compre-
hending texts. Although this is a widely accepted view of reading, particularly among practition-
ers, there are both practical and theoretical problems with this broad definition.
The fundamental problem with the broad view of reading is that it conflates two very dif-
ferent abilities—word recognition (word-level reading) and comprehension. Word recognition
involves a well-defined scope of knowledge (e.g., letters, sounds, words) and processes (decod-
ing) that can be systematically taught. Comprehension, in contrast, is not a skill with a well-
defined scope of knowledge; it is a complex of higher-level mental processes that includes
thinking, reasoning, imagining, and interpreting (see Kamhi, 2009a). With a broad definition of
reading, a theory of reading necessarily becomes a theory of inferencing, a theory of schemata,
and a theory of learning (Perfetti, 1986). The problems with the broad view of reading led Gough
and his colleagues (Gough & Tunmer, 1986; Hoover & Gough, 1990) to propose the Simple
View of Reading. The central claim of the Simple View is that reading consists of two compo-
nents: decoding and linguistic comprehension. Decoding refers to word recognition processes
that transform print into words. Linguistic comprehension (i.e., listening comprehension) is de-
fined as the process by which words, sentences, and discourses are interpreted (Gough &
Tunmer, 1986).
The Simple View of Reading has appealed to many researchers and practitioners. Some re-
searchers, however, prefer restricting the definition of reading to just the decoding component
(e.g., Crowder, 1982). One advantage of a narrow view of reading is that it delineates a restricted
set of processes to be examined (Perfetti, 1986). Crowder (1982), who advocates a narrow view
of reading, made the following analogy between the “psychology of reading” and the “psychology
3
Language and Reading: Convergences and Divergences
of braille.” The psychology of braille does not include such topics as inferences and schema ap-
plication. These abilities involve broad-based cognitive-linguistic processes. Crowder argued
that it was superfluous to make the study of these higher level processes part of the study of
braille. The study of braille is necessarily restricted to the decoding process, or how a reader de-
codes braille to language. By analogy, the study of reading should also be restricted to the decod-
ing process.
Kamhi (2009a, b) recently suggested that embracing the narrow view may provide a solu-
tion to the reading crisis in the United States. The basic argument was that it is possible to elim-
inate reading failure if reading is defined narrowly as decoding abilities. Reading proficiency
levels should reach 90 percent, at a minimum, given the numerous research-supported instruc-
tional programs that have been shown to effectively teach word-level reading (National Reading
Panel [NRP], 2000; Simmons et al., 2007). As Catts (2009) pointed out, a narrow view of read-
ing promotes a broad view of comprehension that recognizes its complexity. Not only are there
different levels of understanding (e.g., literal, analytic, creative), but comprehension also de-
pends on thinking and reasoning processes that are domain and content specific rather than do-
main general (cf. Kintsch, 1998). This is why the best predictor of comprehension is often
familiarity with content knowledge domains (Hirsch, 2006; Willingham, 2006).
It should be apparent that the way one defines reading will have a significant impact on
how reading is measured and taught. We encourage educators to embrace a view of reading that
clearly distinguishes word recognition processes from the reasoning and thinking processes in-
volved in comprehension.
4
Language and Reading: Convergences and Divergences
well as higher-level linguistic and conceptual knowledge to be good readers. Whereas bottom-up
and top-down models emphasize sequential processing, interactive models allow for parallel or
simultaneous processing to occur. Later stages could thus begin before earlier stages have been
completed. Although more complex than serial processing models, parallel processing models
better reflect the types of processing that occur in complex tasks such as reading.
Connectionist models have also been used to explain how children learn to recognize
words (e.g., Seidenberg, 1995; Seidenberg & McClelland, 1989). With this approach, the lexicon
is viewed as an interactive network of connections among different layers of processing. Instead
of depicting different routes (top-down or bottom-up) to access meaning, Seidenberg and
McClelland (1989) propose two different layers of units, orthographic and phonological, that
connect with each other and another layer of units that represents meaning. Because activation
levels are input driven, word frequency has a significant impact on word recognition because the
more often a particular set of units is activated together (e.g., phonological, orthographic, con-
ceptual), the greater the strength of the pathway associated with the particular word (cf. Whitney,
1998). A detailed review of parallel processing models of spoken and written language process-
ing is beyond the scope. For our purposes, it is sufficient to note that simplistic serial processing
models, whether bottom-up or top-down, cannot adequately capture the complex interactions
that occur within and between different processing levels.
Auditory Auditory
Input Analysis
(speech) Phonological
Representation
Word Sentence/Text
Comprehension
Meaning Processing
Visual Visual
Visual
Analysis Representation
Input
(print)
Lexicon
Perceptual Analysis Word Recognition Discourse-Level Processes
FIGURE 1 A Model of Spoken and Written Language Comprehension
5
Language and Reading: Convergences and Divergences
Perceptual Analyses
The input to the perceptual analysis is speech or print. For this input to be recognized, it must be
detected and analyzed. The sensory mechanisms involved in the detection of speech and print are
distinctive; the ear is used to detect speech, and the eye is used to detect print. Sensory deficits
involving hearing or vision place a child at risk for spoken and written language problems.
Children born deaf cannot detect the speech signal through the auditory modality and, as a result,
have considerable difficulty developing intelligible speech. Individuals who are blind cannot de-
tect print through the visual modality. Braille, which relies on the tactile modality, is one way to
bypass the visual deficit. An intact auditory system provides the blind another avenue to access
text material by way of tape recordings.
Once the input has been detected, the segmental and suprasegmental features of spoken
and written words are analyzed. In speech, the processes underlying phonetic discrimination
and phonemic identification are involved. Phonetic discrimination refers to the ability to hear
the difference between two sounds that differ acoustically and phonetically. For example, the
initial t in the word tap is phonetically different from the final t in the word bat. Phonetic differ-
ences that do not affect meaning are often referred to as allophonic variations. If the t sounds in
the preceding words were changed to k sounds, this would change the meaning of the words.
Tap would become cap, and bat would become back. The phonetic differences between /t/ and
/k/ are thus also phonemic differences because they change the meaning of the word. The task
for the young child learning language is to determine which differences between sounds make a
difference in meaning.
The language a child is learning determines which phonetic differences are phonemic. In
Japanese, for example, the differences between /r/ and /l/ are allophonic. In English, however, the
phonetic differences between /r/ and /1/ make a difference in meaning. In French, the front
rounded vowel /y/ is phonemically different from the back rounded /u/. An American who does
not make this distinction will not be able to differentiate between the words tout (all) and tu
(you). These examples are meant to illustrate that learning phonemic categories requires knowl-
edge of the language being learned. The acquisition of phonological knowledge about language
necessarily involves higher-level conceptual processes. Low-level perceptual processes, such as
detection and discrimination, do not lead to knowledge about phonemic categories. In light of
these points, it is important to note that in most listening situations, individuals seldom have to
make distinctions between minimal phoneme pairs (e.g., p/b in the words pin and bin) that are
common stimuli on tests of discrimination. In many instances, lexical and higher-level language
knowledge often eliminate the need for phonemic-level identification.
In reading, just as with speech, discrimination and identification processes are involved. In
reading, discrimination refers to the ability to see the visual differences between letters.
Identification requires knowledge of the correspondences between letters and phonemes. For ex-
ample, the child who confuses the letters b and d in words such as bad and dad is often said to
have a visual discrimination problem. It is more likely, however, that the child can perceive the
visual differences between the letters b and d but has not learned that the letter b is associated
with the phoneme /b/ and the letter d is associated with the phoneme /d/. In other words, the child
has not learned the phoneme–letter correspondences for these two sounds.
To illustrate the difference between low-level visual discrimination ability and higher-
level conceptual (identification) ability, consider the following analogy. In teaching large
classes, it is common to confuse students. The first author once called a girl named Aimee,
Anna. Although Aimee and Anna were both 20-something female graduate students, they could
6
Language and Reading: Convergences and Divergences
Word Recognition
Reading and spoken language begin to share similar knowledge domains and processes in the
word recognition stage. Until this point, the processing of print and speech involves different
sensory and perceptual processes. In the word recognition stage, the features identified in the
previous perceptual stage are used to access the mental lexicon. The words heard or seen must
activate or be associated with previously stored concepts in the individual’s mental lexicon.
These stored concepts in the mental lexicon represent one’s vocabulary. Importantly, the content
and structure of the mental lexicon is essentially the same for both reading and spoken language.
The content of the lexicon includes information about the word’s phonological or visual form as
well as information about the word’s meaning and how the word relates to other words.
Consider, for example, the kind of conceptual information that might appear in the mental lexi-
con for the word pencil.
It refers to an instrument used for writing or drawing; it is a manmade physical object, usually
cylindrical in shape; and it functions by leaving a trail of graphite along a writing surface. . . .
A pencil is one of a class of writing instruments and a close relative of the pen, eraser, and
sharpener. (Just & Carpenter, 1987, p. 62)
The mental lexicon also includes syntactic and semantic information that indicates part of speech
(e.g., noun, verb, or adjective) and possible syntactic and semantic roles. For example, the syn-
tactic information about pencil might indicate that it is a noun that functions semantically as an
instrument (“She wrote the letter with a pencil”) or as a patient (“Peggy bought a pencil”).
The structure of the mental lexicon has received considerable research attention during the
past 30 years. Network models consisting of nodes corresponding to concepts and features have
been a popular way to depict the structure of the lexicon (Collins & Loftus, 1975; Collins &
Quillian, 1969). Early network models were hierarchical in nature, with the ordering in the hier-
archy defined by set inclusion relations. For example, higher-order concepts such as animal in-
cluded lower-order concepts such as bird and sparrow. Other network models have been referred
to as heterarchical, reflecting concepts from ill-structured domains (Just & Carpenter, 1987).
Although theorists might differ in their portrayal of the content and structure of the mental lexi-
con, they generally agree that the mental lexicon is the same for language and reading. The way
in which word meanings are accessed can differ, however, in spoken language and reading.
In processing speech, word meaning is accessed through a word’s phonological represen-
tation. The output of the perceptual analysis is a representation of a word’s acoustic and phonetic
7
Language and Reading: Convergences and Divergences
features. These acoustic–phonetic representations of speech input are used by the listener to
activate or instantiate a word’s phonological representation in the lexicon. This may involve the
listener in attempting to match acoustic–phonetic representations with phonological representa-
tions. Phonological representations are directly linked to a word’s meaning because this informa-
tion is stored together for each word in the mental lexicon.
Phonological representations of words stored in the mental lexicon can take one of sev-
eral forms. Words may contain discrete phonetic and phonemic segments or syllable segments,
or be represented as whole words or short phrases (e.g., “it’s a” as “itsa” and “did you know” as
[dIdəno]). Although young children’s phonological representations begin to contain more dis-
crete phonetic and phonemic information as they progress through the preschool years, the abil-
ity to access this information may not develop until age 5 or later, depending on early literacy
experiences and formal instruction. Studies of young children’s speech perception (e.g.,
Nittrouer, Manning, & Meyer, 1993) have found that there is a gradual shift in the acoustic cues
used to make phonological decisions. Nittrouer and colleagues hypothesize that as children
gain experience with a native language, they become more sensitive to phonetic structure. In a
more recent study, Nittrouer (1996) showed that this shift is related to children’s developing
phonemic awareness. It seems that early exposure to reading as well as developmental changes
in speech perception both contribute to young children’s ability to represent speech as discrete
phonemic segments.
In contrast to speech, in which there is only one way to access a word’s meaning, in read-
ing there are two ways: indirectly, by way of a phonological representation, or directly, by way of
a visual representation (see Figure 1). Use of a visual representation to access the lexicon is var-
iously referred to as the direct, visual, look-and-say, or whole-word approach. In accessing the
lexicon in this way, the reader locates the word in the lexicon whose visual representation con-
tains the same segmental and/or visual features as those identified in the previous perceptual
analysis stage. In other words, a match is made between the perceived visual configuration and a
visual representation that is part of the mental lexicon for the particular word.
Word meaning can also be accessed through a phonological representation. With this
indirect or phonological approach, the reader uses knowledge of phoneme–letter correspondence
rules to recode the visually perceived letters into their corresponding phonemes. Individual
phonemes are then blended together to form a phonological sequence that is matched to a simi-
lar sequence in the lexicon. The phonological approach is particularly important in the develop-
ment of reading. The ability to decode printed words phonologically allows children to read
words they know but have never seen in print. Reading by the phonological approach also causes
the child to attend to the letter sequences within words. The knowledge gained about letter se-
quence makes the child’s visual representations more precise.
Reading by the phonological route is thus similar to speech recognition in that a word is
recognized by way of its phonological representation. There is one important difference, how-
ever, in using phonological representations to access meaning in comprehending spoken and
written language. To successfully use the phonological route in reading, one must have explicit
awareness of the phonological structure of words, specifically, the knowledge that words consist
of discrete phonemic segments (Liberman, 1983). These segments are not readily apparent to
young children because the sound segments of speech are blended together in the acoustic signal.
For example, the word cat is one acoustic event; its sound segments do not correspond exactly to
its three written symbols. Although preschool children might show some phonological aware-
ness, much explicit instruction and practice is usually required for a child to become efficient in
using the phonological approach.
8
Language and Reading: Convergences and Divergences
The recognition that there were two possible routes to word recognition led to the popular-
ity of dual-route models of word recognition (cf. Stanovich, 1991). Although early proponents of
dual-route models agreed that there were two routes to word recognition, they differed in as-
sumptions about the various speeds of the two access mechanisms and how conflicting informa-
tion was resolved. The size of the sound–letter correspondences in the phonological route also
differed from model to model (e.g., sound-by-sound, syllables, word level). Discussions of the
different variations of these models can be found in Coltheart, Curtis, Atkins, and Haller (1993),
Patterson and Coltheart (1987), and Rayner and Pollatsek (1989).
Questions about the nature of the print-to-sound conversion have recently taken a new turn
due to the increasing popularity of parallel-distributed processing models that contain no word-
level representations or lexicon in the network (Share & Stanovich, 1995). Regardless of how the
print-to-sound conversion takes place, there is recent evidence that this conversion is essential
for the large numbers of low-frequency words that cannot be recognized on a visual basis (Share
& Stanovich, 1995). In contrast, high-frequency words seem to be recognized visually with min-
imal phonological recoding even in the very earliest stages of reading acquisition (Reitsma,
1990). The more exposure a child has to a word, the more likely a visual approach will be used.
The use of a visual versus a phonological approach to word recognition depends on the fre-
quency of the word rather than the particular reading stage a child is in.
Discourse-Level Processes
Up to this point, we have considered the processes involved in recognizing words. Spoken and
written language, however, consists of longer discourse units, such as sentences, conversations,
lectures, stories, and expository texts. Psycholinguistic studies carried out in the 1960s and
1970s (cf. Carroll, 1994; Clark & Clark, 1977) explored the role that syntactic, semantic, and
world knowledge played in comprehending larger units of spoken and written discourse. By
focusing on the independent contribution these different types of knowledge made toward mean-
ing, these early studies were limited in what they could tell us about the interaction of different
types of knowledge and whether different discourse types are processed the same way by listen-
ers and readers. Despite these limitations, it is useful to consider how structural, propositional,
and situation or world knowledge can be used to construct meaning.
STRUCTURAL KNOWLEDGE. A variety of structural cues are used by listeners and readers in
comprehending speech and text. These cues include word order, grammatical morphemes, and
function words such as relative pronouns, conjunctions, and modals. Listeners and readers often
use syntactic and morphologic cues to figure out the meaning of unknown words. Grammatical
morphemes, for example, provide information about word classes. Adverbs are signaled by the
inflections -ly and -y, whereas adjectives are marked by the suffixes -able and -al. Verbs are sig-
naled by the inflections -ed, -ing, and -en. Nouns are marked by definite and indefinite articles,
plural and possessive markers, and suffixes such as -ment and -ness. The reason why readers are
able to make any sense at all out of a sentence like “Twas brillig and the slithy toves did gyre and
gimble in the wabe” is that inflections (y and s) and syntactic markers (the and did) provide cues
about grammatical form class.
Clark and Clark (1977) provide an excellent review of studies that demonstrate the influ-
ence syntactic and morphologic knowledge have on sentence comprehension. It has been shown,
for example, that listeners use function words to segment sentences into constituents, classify the
9
Language and Reading: Convergences and Divergences
constituents, and construct meanings from them (e.g., Bever, 1970; Fodor & Garrett, 1967).
Consider the following two sentences, one with relative pronouns and one without:
1. The pen that the author whom the editor liked used was new.
2. The pen the author the editor liked used was new.
Fodor and Garrett (1967) found that listeners had more difficulty paraphrasing sentences like (2)
than sentences like (1). More recent studies have continued to attempt to prove that the initial
segmentation of a sentence (i.e., parsing) is performed by a syntactic module that is not influ-
enced by other kinds of knowledge (e.g., Frazier, 1987).
WORLD KNOWLEDGE. Structural and propositional knowledge are crucial for constructing
meaning, but an individual’s knowledge of the world or what has come to be called situation
model representations also plays an important role in comprehension. Consider, for example,
how world knowledge makes the sentence Jake ate the ice cream with relish unambiguous,
10
Language and Reading: Convergences and Divergences
whereas a similar sentence, Jake ate the sausage with relish, is ambiguous (Just & Carpenter,
1987). We know that relish is normally not eaten with ice cream. Such information is not specific
to language; instead, it reflects general knowledge about the tastes of foods to assign with relish.
World knowledge can be divided into knowledge of specific content domains and knowl-
edge of interpersonal relations. Specific content domains would include academic subjects such as
history, geography, mathematics, and English literature; procedural knowledge such as how to fix
a car, tie a shoelace, and play tennis; and scriptlike knowledge of familiar events. Interpersonal
knowledge involves such things as knowledge of human needs, motivations, attitudes, emotions,
values, behavior, personality traits, and relationships. It should be evident how these kinds of
world knowledge play an important role in processing spoken and written language.
Because world knowledge can be so broad, psychologists have focused attention on the sit-
uation-specific world knowledge that listeners and readers use to construct meaning (e.g., van
Dijk & Kintsch, 1983). The assumption is that as we process discourse, we construct a mental or
situational model of the world as described by the discourses.
MODELS OF DISCOURSE PROCESSING. To understand larger units of spoken and written dis-
course, it is necessary not only to construct representations that consider structural, proposi-
tional, and situational information, but also to relate these representations to one another. One
must also use this information to make inferences about meaning and make decisions about
which information should be remembered. Given the variety of knowledge types and cognitive
processes involved in discourse processing, no one model can expect to capture all these facets of
discourse processing. It is useful, however, to consider the kinds of models that have been pro-
posed. Although these models deal primarily with how readers construct meaning from texts,
their basic principles can be applied to spoken language discourse as well.
Kintsch and van Dijk’s initial model of text comprehension proposed that multiple levels
of representation were needed to construct meanings based on different kinds of knowledge
(Kintsch & van Dijk, 1978; van Dijk & Kintsch, 1983). Three levels of representation correspond
to the three knowledge types: structural, propositional, and knowledge of the situation/world.
This initial model of comprehension relied on schema-driven, top-down processing to
build the knowledge of the world (i.e., situation model) representation. Kintsch (1988), however,
felt that these notions were not adaptive to new contexts, were too inflexible, and could not ac-
count for how schemas were initially constructed. His most recent theory, called construction in-
tegration theory, acknowledges that many elements enter into the comprehension process
(Kintsch, 1998). These include perceptions, concepts, ideas, images, or emotions. A crucial con-
sideration in the theory is where these elements come from—from the world via the perceptual
system or from the individual in the form of memories, knowledge, beliefs, body states, or goals.
For Kintsch, the heart of the theory “is a specific mechanism that describes how elements from
these two sources are combined into a stable mental product in the process of comprehension”
(Kintsch, 1998, p. 4).
Kintsch goes on to provide a brief synopsis of the theory: One starts with a reader who has
specific goals, a given background of knowledge and experience, and a given perceptual situa-
tion, such as printed words on a page of text. The propositional idea units created from these
words are then linked to the reader’s goals, knowledge, and experiences to create an interrelated
network of idea units. Unlike Kintsch’s earlier schema-driven models in which context was used
to construct meaning, the construction of the network of idea units is viewed as an entirely
bottom-up process, unguided by the larger discourse context. The initial context-insensitive con-
struction process is followed by “a constraint-satisfaction, or integration, process that yields if all
11
Language and Reading: Convergences and Divergences
goes well, an orderly mental structure out of initial chaos” (Kintsch, 1998, p. 5). The constraint-
satisfaction process involves selectively activating those elements from the network of idea-units
that fit together and deactivating the rest.
If it all sounds pretty complicated, it is because complicated models and theories are
needed to explain how text information is integrated with a reader’s background knowledge and
experiences to construct meaning. Simplistic bottom-up and top-down models are too general to
explain how meaning is actually constructed, but some of the notions from these models, such as
scripts and schemas, still work well for understanding how children construct meaning for cer-
tain prototype forms of discourse such as familiar events and stories. A schema is generally
thought of as a structure in memory that specifies a general or expected arrangement of a body of
information. Familiar events, for example, are well captured by scripts, which are a particular
type of schema. Scripts contain slots for the components of an event, such as the main actions,
participants, goals, and typical position of each action. Scripts make it easier to process familiar
events by providing individuals with a coherent structure into which they can insert new informa-
tion. Scripts also allow individuals to add necessary information that might be omitted in spoken
or written discourse. For example, familiarity with a restaurant script allows listeners and readers
to anticipate some mention of the menu. If no mention of the menu is made, but information
about the kind of restaurant is given (e.g., Italian), one can infer the contents of the menu.
Certain types of discourse, such as stories, seem to have a consistent structure or grammar.
This was recognized years ago when researchers proposed that stories had a common story
grammar or schema. A story schema can be viewed as a mental framework that contains slots for
each story component, such as a setting, goal, obstacle, and resolution. Story grammars represent
a slightly different characterization of the knowledge of story structures. Story grammars specify
the hierarchical relations among the components more directly than a story schema (Mandler &
Johnson, 1977; Stein & Glenn, 1979). Story grammars attempt to specify the structural organiza-
tion of stories in the same way that syntactic grammars specify the structural organization of sen-
tences (Just & Carpenter, 1987, p. 231). The main structural components of a story are a setting
and an episode. The setting introduces the characters and the context of the story. Episodes can
be further divided into an initiating event, internal response, attempt, consequence, and reaction.
Knowledge of the structure and function of stories, like knowledge of scripts, can facilitate com-
prehension of spoken and written language (Just & Carpenter, 1987; Perfetti, 1985).
OTHER METAPHORS FOR DISCOURSE AND TEXT UNDERSTANDING. The notion that multi-
ple sources of knowledge or representations are involved in processing discourse and text is an
important one for understanding what is involved in comprehension. Other notions about compre-
hension, however, are important as well. Graesser and Britton (1996) have found that five
metaphors capture the essence of the various ways of thinking about text comprehension. The first
metaphor, understanding is the assembly of a multileveled representation, has already been dis-
cussed. Speech-language pathologists and other educators are familiar with at least two of the
other metaphors: understanding is the process of managing working memory and understanding is
inference generation. The two metaphors we may not be familiar with are understanding is the
construction of a coherent representation and understanding is a complex dynamical system. To
these five metaphors, we will add a sixth: understanding is a metacognitive ability. Although
Graesser and Britton apply these metaphors to text understanding, in most cases they can be ap-
plied to spoken discourse as well. Each of these last five metaphors will be discussed briefly next.
Understanding Is the Management of Working Memory. Most psychologists and educators
are comfortable with the assumption that comprehension is managed in a limited-capacity working
12
Language and Reading: Convergences and Divergences
memory. Every educator has had firsthand experience with this metaphor. For example, when the
demands of comprehension exceed the limitations of working memory, students’ comprehension
decreases dramatically. Students with low working memory spans often experience difficulty
when comprehension components tax working memory. Poor comprehenders have also been
shown to have problems suppressing irrelevant information from working memory
(Gernsbacher, 1996).
Understanding Is Inference Generation. The ability to construct meaning requires more
than interpreting explicit propositions. It involves accessing relevant world knowledge and gen-
erating inferences that are needed to make sentences cohere (local coherence) and to relate text
to world knowledge (global coherence). A number of different systems exist to classify infer-
ences. In several studies comparing inferencing abilities in good and poor readers (e.g., Kucan &
Beck, 1997; Laing & Kamhi, 2002; Trabasso & Magliano, 1996), inferences were classified as
either predictions, associations, or explanations. A predictive inference speculates about events
or actions that may occur based on what has already occurred in a story or text. For example, a
predictive inference for the sentence She played hard everyday would be She probably will be in
good physical shape. An associative inference is a statement that makes generalizations about
characters, actions, objects, or events in a story or text. Associative inferences can also be speci-
fications of procedures or responses to wh-questions. An associative inference for the sentence
He ate ice cream would be that He likes ice cream or He was hungry. An explanatory inference
provides causal connections between actions and events in a story or text. They are usually re-
sponses to “why” questions that provide explanations for a state, event, or action. For example,
in a story about a child who wants a faster computer, an explanatory inference might be He was
not very happy because he wanted a new computer. The proportion of explanatory inferences
generated has been found to be significantly related to comprehension performance (e.g.,
Trabasso & Magliano, 1996). This is not surprising because explanatory inferences require re-
trieving and remembering causal information that serves to unite propositions in a story.
Inferences can also be distinguished according to whether they are derived from the con-
tent of activated world knowledge structures (e.g., scripts and schemas) or whether they are
novel constructions that are needed to construct the situation model. Inferences that are gener-
ated from existing world knowledge tend to be generated “online.” Graesser and Britton (1996)
argue that a satisfactory model of text understanding should be able to accurately predict infer-
ences that are quickly or automatically made during comprehension as well as those that are time
consuming. Inferences generated online include those that address readers’ goals, assist in estab-
lishing local or global coherence, and are highly activated from multiple information sources
(e.g., Long, Seely, Oppy, & Golding, 1996). Inferences that are more time consuming may be
caused by minimal world knowledge about the topic or by contradictions, anomalies, or irrele-
vant propositions in the text. Readers attempt to generate explanations and justifications to re-
solve the contradictions and anomalies. The process of generating these “elaborative inferences”
is necessarily time consuming and may not be used by readers with low motivation (Graesser &
Britton, 1996, p. 350).
Understanding Is the Construction of Coherent Representations. The basic notion with this
metaphor is that the more coherent the discourse or text, the easier it is to understand. A text is
fully connected if every proposition is conceptually connected to one or more other propositions.
Some theorists, following Kintsch (1974), believe that noun-phrase arguments are critical for
connecting propositions and establishing coherence. More recent research, however, has shown
that argument overlap is neither a necessary nor sufficient condition for establishing coherence;
13
Language and Reading: Convergences and Divergences
instead, it is merely one type of connection (cf. Graesser & Britton, 1996). Other types of con-
nections that have been considered include the connections between predicates of propositions
(Turner, Britton, Andraessen, & McCutchen, 1996), causal connections and goals of story char-
acters (van den Broek, Risden, Fletcher, & Thurlow, 1996), and the connections that tie deep
metaphors to lexical items and explicit expressions (Gibbs, 1996).
Despite the challenge of identifying the specific types of connections that tie texts together,
the “understanding-as-coherence” metaphor makes a large number of predictions about compre-
hension performance. Most of these predictions are generally intuitive. For example, a proposition
has a greater likelihood of being recalled when it has more connections to other propositions in the
text, and reading time increases when there is a break in coherence. However, some are counterin-
tuitive. For example, Mannes and St. George (1996) found that there are more connections (or
stronger ones) between text and world knowledge if there is a discrepancy between an outline and
text content. The discrepancy causes improved problem solving, though recall for the text suffers.
Summary
We have attempted in this section to provide a way of thinking about the knowledge and processes
involved in understanding spoken and written language. Although the emphasis has been primarily
on the similarity of knowledge and processes, some important differences in the word recognition
processes were acknowledged. In our discussion of discourse comprehension processes, we tended
to treat research as if it applied both to spoken and written language comprehension when, in fact,
it rarely did. Our assumption here was that a model of comprehension that is sufficiently dynamic,
flexible, and multifaceted would apply equally well to both spoken and written discourse. Although
the six metaphors discussed were meant to illuminate the different aspects of comprehension, per-
haps they made a complete muddle of comprehension for some. Graesser and Britton (1996)
thought that after reading through their book on text understanding with all its different models and
14
Language and Reading: Convergences and Divergences
views of comprehension, someone might ask, “What is text understanding?” Readers might won-
der the same thing about our view of comprehension. With a slight modification to include
discourse as well as text comprehension, the definition of comprehension Graesser and Britton
suggest provides a good answer to the question:
Text [and discourse] understanding is the dynamic process of constructing coherent represen-
tations and inferences at multiple levels of text and context, within the bottleneck of a limited-
capacity working memory. (p. 350)
Having emphasized the similarities between spoken and written language up to this point, in the
next section we consider some of the differences between the two.
15
Language and Reading: Convergences and Divergences
example, speeches and lectures can be planned much like writing, radio talk lacks a visual di-
mension and contextual support, and tape recordings are durable.
Physical Differences
Whereas speech consists of temporally ordered sounds, writing consists of marks made on a sur-
face (e.g., paper) in a two-dimensional space. As such, writing is relatively durable; it can be read
and reread. Speech, unless it is recorded, is ephemeral. It has no existence independent of the
speaker. The durability of writing gives the reader control over how fast or slow to read. Certain
texts can be savored, whereas others can be skimmed. The listener, in contrast, is tied to the fleet-
ing speech of the speaker. Missed words or sentences will be lost if clarification is not requested.
Perera (1984, p. 161) noted that readers often have the benefit of a whole range of visual
cues, such as running headlines, different-size type, color, and summaries or abstracts. In addi-
tion, a device such as the footnote allows the writer to provide additional information without in-
terrupting the main thread of the text. Such devices allow the reader to decide the level at which
he or she will read. The listener, in contrast, is completely dependent on the speaker’s selection
of material. Note, however, that the listener could choose not to listen to the speaker’s message.
Situational Differences
The most frequent type of spoken language is face-to-face communication. Conversations are
often interactive exchanges between two or more individuals. Questions are followed by an-
swers, requests by responses, and statements by acknowledgments. When a listener does not un-
derstand something, a clarification is requested. Careful planning is not the rule in conversational
discourse. When speakers pause too long before talking, they will usually be interrupted. Despite
this time pressure to speak, misunderstandings are infrequent; when they occur, they are easily
resolved by repeating or rephrasing the message. Nonverbal communication acts, such as ges-
tures, facial expressions, and body postures, can help to clarify messages. Speakers and listeners
also share the same nonlinguistic setting. People and objects that are visible can be referred to by
pronouns rather than by noun phrases (even without prior reference), and many adverbials and
prepositions can be expressed by here, there, and like this.
In contrast, writing and reading are often individual endeavors. The writer receives no
prompting about what to write and no immediate feedback on the clarity of the writing. But the
writer is generally under less severe time constraints and can thus take more time to search for
the best way to express a message. The writer can also correct and revise a text until a final copy
is produced. Such care and precision is necessary in writing because there are no contextual and
nonverbal cues to aid comprehension. The written text thus has to bear the whole burden of com-
munication, which is one reason why writing is usually more precise than talking.
Functional Differences
One of the earliest needs to generate a writing system was to retain accurate records of property,
commercial transactions, and legal judgments. A Chinese proverb holds that “The palest ink is
better than the best memory.” Writing has enabled the knowledge of centuries to accumulate,
thus allowing each new generation to build on the ideas, discoveries, and inventions of the gen-
eration or generations before. Many academic subjects, such as history, geography, the physical
sciences, and social sciences, owe their very existence to writing (Perera, 1984, p. l64). Another
function not served by speech is labeling. Although speech is used to label objects in a referential
16
Language and Reading: Convergences and Divergences
sense, written labels serve more of an information function. Consider such labels as street
names, signposts, nameplates on theaters and public buildings, brand labels, and danger warnings.
Written language can also serve a variety of communicative functions, such as relating stories,
events, and experiences or sharing information and making requests. Finally, a specialized function
of writing is found in literature. Societies have oral literatures, but oral literatures are restricted to
a few types, such as ballads, epic poetry, drama, folk stories, and myths. Essays, novels, diaries,
and memoirs are some of the genres that are particular to writing.
Perera (1984) has suggested that the most basic uses of writing involve the recording of
facts, ideas, and information. Although speech also has an informative function, an equally im-
portant function of speech is the role it plays in establishing and maintaining human relation-
ships. A large part of everyday speech with friends, acquaintances, and other individuals serves
social-interpersonal functions rather than intellectual ones. E-mail and instant messaging now
serve this role as well.
One advantage writing has over speech, according to Perera (1984, p. l65), is that it allows
ideas to be explored at leisure and in private. Writing can thus become a means of extending and
clarifying one’s thinking and ideas. Often in conversation when a controversial topic is raised,
there is a tendency for opinions to polarize. Someone who tries to take both sides of a issue might
be pressed to select one particular view. In writing, however, one can take time to develop a line of
thought, weigh opposing arguments, notice errors in reasoning, and develop new lines of thinking.
Form Differences
The most obvious difference in form is that speech consists of sounds, whereas written language
consists of letters. As indicated earlier, this would not be so much of a problem if speech sounds
(i.e., phonemes) stood in one-to-one correspondence with written letters. Form differences be-
tween spoken and written language are not limited to the discrete segments (i.e., phonemes and
letters) that make up speech and text. Spoken and written language also differ in how they repre-
sent suprasegmental, paralinguistic, and prosodic features. Paralinguistic features include pitch
and timbre differences that distinguish male and female voices; general voice quality, such as
breathiness, hoarseness, or nasality; and the general manner of how an utterance is produced,
such as shouted, whispered, or spoken. Perera has pointed out that these features do not usually
affect the actual meaning of an utterance; however, they may reflect the speaker’s attitude about
what is being said.
Prosodic features include intonation, stress, and rhythm. Perera (1984) presented four
functions of prosodic features: (1) to enable the communicative intent of an utterance to differ
from its grammatical form (e.g., He’s lost it versus He’s lost it?), (2) to group words into infor-
mation units, (3) to place emphasis, and (4) to convey the speaker’s attitude. These functions dif-
fer in the extent to which they can be reflected in writing. Whereas punctuation effectively
changes the communicative intent of an utterance, it is not so effective in signaling which words
belong together in information units. Italics, underlining, and the use of capital letters are some
ways to distribute emphasis throughout a written utterance. But heavy use of these devices in for-
mal writing is usually discouraged. Expressing attitudes in writing is clearly difficult. Perera
(1984, p. l78) provided an example of how much attitudinal information is conveyed by prosodic
features in the following quote of a journalist who listened to one of the Watergate tapes:
Once you hear the tapes, and the tone in which he (Nixon) uttered the comments which previ-
ously have only been available in a neutral transcript, any last shred of doubt about his guilt
must disappear.
17
Language and Reading: Convergences and Divergences
Perera (1984) goes on to consider the extent to which the writing system represents the
segmental and suprasegmental aspects of speech. Among other things, she pointed out that
graphemes represent the “citation” (well-spoken) form of words rather than the degraded
productions that often occur in fast speech (e.g., compare “did you know” to [dIdəno]).
Punctuation can signal the grammatical function of a sentence and mark some prosodic bound-
aries. The writer, however, has no conventional way to express voice quality, volume, rate of
speech, rhythm, and into national patterns.
Vocabulary Differences
One would expect that there would be differences in the vocabulary used in spoken and written
language because writing provides more time and, therefore, more resources to select words. The
additional time allows writers to choose words that communicate meanings clearly. Clear, unam-
biguous writing is necessary to ensure that the author’s intended meaning is derived. Readers,
unlike listeners, don’t have the luxury of requesting clarification when the message is unclear. In
contrast, conversational speech provides little opportunity to consider alternative word choices,
and though one can revise a word choice once it is spoken, too much fumbling detracts from ef-
fective communication (Chafe & Danielewicz, 1987). With writing, time is often not a factor;
one can spend seconds, minutes, hours, or days finding the appropriate word or expression, and
even after a selection is made, the writer is free to revise without anyone noticing. Word process-
ing programs do not keep track of earlier drafts.
The consequence of these differences is that the vocabulary of spoken language tends to be
more limited in variety. A simple way to demonstrate this is by calculating type-token ratios
(TTRs) for spoken and written language. TTR is calculated by dividing the number of different
words by the total number of words in a sample. Chafe and Danielewicz (1987) show that TTRs
for spoken language are consistently lower than for written language. Interestingly, the ratio for
academic lectures is about the same as in conversations (.19/.18), indicating that rapid production
of spoken language produces less varied vocabulary, regardless of the kind of speaking involved.
The frequent use of nonspecific terms (e.g., thing, whatever, “doohickey,” “whatchamacallit”),
hedges (sort of, kind of), and maze behaviors (interjections, disfluencies, false starts, repetitions)
are all indications of the processing demands of spoken language. Chafe and Danielewicz also
show how spoken language has less referential explicitness than written language. Nonspecific
third person pronouns (it, this, that) are used frequently in spoken language and are one of the
factors that differentiate good from poor writing.
The lexicon that speakers and writers choose from is also not the same for writing and
speaking. There is a literate lexicon (Nippold, 2007) that writers draw from in formal writing.
For example, conjunctive adverbs like thus, therefore, hence, and accordingly are rarely used
in spoken language. Chafe and Danielewicz (1987) suggest that spoken language compensates
for its restricted lexical variety by assigning a premium to freshness. Speakers must stay cur-
rent. Cool may have been cool 10 or 20 years ago, but not now. Freshness of vocabulary is less
important in writing, where there is more of a premium on choosing the right word to convey
a particular meaning. Not surprisingly, conversations and academic papers differ considerably
in their use of literary and colloquial vocabulary (Chafe & Danielewicz, 1987). Academic pa-
pers had only 1 instance per 1,000 words of colloquial vocabulary (e.g., kid bike, figure out),
whereas conversations had 27. Lectures and letters fall somewhere in the middle, indicating
that there is nothing in the nature of speaking that prevents a speaker from using literary
vocabulary and nothing in the nature of writing that prevents a writer from using colloquial
18
Language and Reading: Convergences and Divergences
vocabulary. Lectures are thus more literary than conversations, whereas letters are more con-
versational than academic papers.
Grammatical Differences
Samples of spoken language uncover relatively high frequencies of coordination, repetition, and
rephrasing. Conversational discourse is typically low in lexical density and high in redundancy.
Lexical items are spaced out and separated by grammatical words, and a high number of total
words are used to convey a relatively small amount of information. Written language, in contrast,
is high in lexical density and low in redundancy. This results from the use of grammatical struc-
tures that decrease redundancy and increase lexical density.
Studies (e.g., Chafe & Danielewicz, 1987) have shown that in conversation it is more com-
mon to provide smaller amounts of information at a time. Most written language, by contrast, is
denser lexically as well as propositionally. Conversations, because of their interactive nature, are
generally less coherent than writing. Speakers are free to change the subject at almost any point
in a conversation. Topics need not be related in any logical way. In writing, however, an overall
theme is necessary. Topic changes must be justified and explicitly made. Writing also has pre-
scribed rules for organizing content. These rules cover the use of topic sentences, paragraph
structure, and introductory and concluding statements.
Processing Differences
Earlier in this chapter, we talked about top-down processing models, discourse-level comprehen-
sion processes, and the higher-level knowledge schemas that contribute to comprehension of
spoken and written language. The focus in these sections was on the commonalities between un-
derstanding speech and text. There are very important differences, however, in the contribution
higher-level processes make to spoken and written language comprehension. The role of higher-
level processes or context effects in reading has received considerable research attention and
caused much confusion. One reason for this confusion is that researchers often fail to distinguish
between the use of context to facilitate word recognition and the use of context to facilitate text
comprehension. Context plays an important role in facilitating text comprehension; it plays a
very limited role, however, in facilitating word recognition in good readers.
Support for the limited role of higher-level processes in word recognition comes from eye-
movement experiments. Research using various eye-movement methodologies has been consistent
in finding that the vast majority of content words in text receive a direct visual fixation (Just &
Carpenter, 1987; Rayner & Pollatsek, 1989). Short function words may be skipped, but even
many of these receive a direct visual fixation. The span of effective visual information during a
fixation is thus quite small, meaning that text is sampled in a very dense manner, even when the
words are highly predictable (Balota & Chumbley, 1985).
Based in part on evidence from these eye-movement studies, most models of reading have
expectancy-based, top-down processes functioning after words have been recognized
(Seidenberg, 1985; Till, Mross, & Kintsch, 1988). Higher-level contextual information plays
more of a role in speech perception or language processing because of the well-documented am-
biguity in decontextualized speech. For example, isolated words from normal conversation are
often not recognized out of context. This is not the case, however, for written language. Fluent
readers can identify written words out of context with near-perfect accuracy. As Stanovich
(1986) has noted, the physical stimulus alone completely specifies the lexical representation in
writing, whereas this is not always true in speech. It is more important in reading, therefore, for
19
Language and Reading: Convergences and Divergences
the input systems involved in word recognition to deliver a complete and accurate representation
of words to higher level processes. Paradoxically, then, poor readers who have difficulty accu-
rately decoding words must rely more on contextual information than good readers who have
proficient word recognition skills.
20
Another random document with
no related content on Scribd:
I hear within our house?
The ground shakes, | and the home of Gymir
Around me trembles too.”
Gerth spake:
16. “Bid the man come in, | and drink good mead
Here within our hall;
Though this I fear, | that there without
My brother’s slayer stands.
[113]
Skirnir spake:
Gerth spake:
Skirnir spake:
Gerth spake:
Skirnir spake:
Gerth spake:
Skirnir spake:
[118]
[119]
Gerth spake:
Skirnir spake:
Gerth spake:
41. “Tell me, Skirnir, | ere thou take off the saddle,
Or farest forward a step:
What hast thou done | in the giants’ dwelling
To make glad thee or me?”
[120]
Skirnir spake:
Freyr spake:
43. “Long is one night, | longer are two;
How then shall I bear three?
Often to me | has a month seemed less
Than now half a night of desire.”
[107]
[Contents]
NOTES
[108]
Prose. Freyr: concerning his father, Njorth, and the race of the
Wanes in general, cf. Voluspo, 21 and note. Snorri thus describes
Njorth’s family: “Njorth begat two children in Noatun; the son was
named Freyr, and the daughter Freyja; they were fair of aspect and
mighty. Freyr is the noblest of the gods; he rules over rain and
sunshine, and therewith the fruitfulness of the earth; it is well to call
upon him for plenty and welfare, for he rules over wealth for
mankind. Freyja is the noblest of the goddesses. When she rides to
the fight, she has one-half of the slain, and Othin has half. When she
goes on a journey, she drives her two cats, and sits in a cart. Love-
songs please her well, and it is good to call on her in love-matters.”
Hlithskjolf: Othin’s watch-tower; cf. Grimnismol, introductory prose.
He said: both manuscripts have “Then Skathi said:” (Skathi was
Njorth’s wife), but Bugge’s emendation, based on Snorri’s version, is
doubtless correct.
1. My son: both manuscripts, and many editors, have “our son,”
which, of course, goes with the introduction of Skathi in the prose. As
the stanza is clearly addressed to Skirnir, the change of pronouns
seems justified. The same confusion occurs in stanza 2, where
Skirnir in the manuscripts is made to speak of Freyr as [109]“your
son” (plural). The plural pronoun in the original involves a metrical
error, which is corrected by the emendation.
4. Elfbeam: the sun, so called because its rays were fatal to elves
and dwarfs; cf. Alvissmol, 35.
16. Brother’s slayer: perhaps the brother is Beli, slain by Freyr; the
only other references are in Voluspo, 53, and in Snorri’s paraphrase
of the Skirnismol, which merely says that Freyr’s gift of his sword to
Skirnir “was the reason why he was weaponless when he met Beli,
and he killed him bare-handed.” Skirnir himself seems never to have
killed anybody. [113]
19. Apples: the apple was the symbol of fruitfulness, and also of
eternal youth. According to Snorri, the goddess Ithun had charge of
the apples which the gods ate whenever they felt themselves
growing old. [114]
21. Ring: the ring Draupnir (“Dropper”) was made by the dwarfs for
Othin, who laid it on Baldr’s pyre when the latter’s corpse was
burned (cf. Voluspo, 32 and note, and Baldrs Draumar). Baldr,
however, sent the ring back to Othin from hell. How Freyr obtained it
is nowhere stated. Andvari’s ring (Andvaranaut) had a similar power
of creating gold; cf. Reginsmol, prose [115]after stanza 4 and note.
Lines 3 and 4 of this stanza, and the first two of stanza 22, are
missing in the Arnamagnæan Codex.
27. Eagle’s hill: the hill at the end of heaven, and consequently
overlooking hell, where the giant Hræsvelg sits “in an eagle’s guise,”
and makes the winds with his wings; cf. Vafthruthnismol, 37, also
Voluspo, 50. The second line is faulty in both manuscripts;
Hildebrand’s emendation corrects the error, but omits an effective
touch; the manuscript line may be rendered “And look and hanker for
hell.” The Arnamagnæan Codex breaks off with the fourth line of this
stanza.
30. In Regius and in nearly all the editions the first two lines of this
stanza are followed by lines 3–5 of stanza 35. I have followed
Niedner, Sijmons, and Gering. The two words here translated vile
things are obscure; Gering renders the phrase simply “Kobolde.”
32. No gap indicated in the manuscript; Niedner makes the line here
given as 4 the first half of line 3, and fills out the stanza thus: “with
which I will tame you, / Maid, to work my will.” The whole stanza
seems to be either interpolated or out of place; it would fit better after
stanza 25.
34. Most editors reject line 3 as spurious, and some also reject line
6. Lines 2 and 3 may have been expanded out of a single line
running approximately “Ye gods and Suttung’s sons.” Suttung:
concerning this giant cf. Hovamol, 104 and note. [118]
35. Most editors combine lines 1–2 with stanza 36 (either with the
first two lines thereof or the whole stanza), as lines 3–5 stand in the
manuscript after line 2 of stanza 30. Hrimgrimnir (“The Frost-
Shrouded”): a giant not elsewhere mentioned. Line 5, as a repetition
of line 4, is probably a later addition.
36. For the combination of this stanza with the preceding one, cf.
note on stanza 35. The scribe clearly did not consider that the stanza
began with line 1, as the first word thereof in the manuscript does not
begin with a capital letter and has no period before it. The first word
of line 3, however, is so marked. Line 5 may well be spurious.
37. Again the scribe seems to have been uncertain as to the stanza
divisions. This time the first line is preceded by a period, but begins
with a small letter. Many editors have made line 2 [119]into two half-
lines. A charm: literally, the rune Thurs (þ); the runic letters all had
magic attributes; cf. Sigrdrifumol, 6–7 and notes.
Introductory Note
The Harbarthsljoth is found complete in the Codex Regius, where it
follows the Skirnismol, and from the fourth line of stanza 19 to the
end of the poem in the Arnamagnæan Codex, of which it occupies
the first page and a half.
The poem differs sharply from those which precede it in the Codex
Regius, both in metrical form and in spirit. It is, indeed, the most
nearly formless of all the Eddic poems. The normal metre is the
Malahattr (cf. Introduction, where an example is given). The name of
this verse-form means “in the manner of conversation,” and the
Harbarthsljoth’s verse fully justifies the term. The Atli poems
exemplify the conventional use of Malahattr, but in the Harbarthsljoth
the form is used with extraordinary freedom, and other metrical
forms are frequently employed. A few of the speeches of which the
poem is composed cannot be twisted into any known Old Norse
metre, and appear to be simply prose.
Owing to the chaotic state of the text, and the fact that none of the
editors or commentators have succeeded in improving it much, I
have not in this case attempted to give all the important emendations
and suggestions. The stanza-divisions are largely arbitrary.
[Contents]
[123]
Thor spake:
Thor spake:
[124]
Thor spake:
Thor spake:
Harbarth spake:
[126]
Thor spake:
13. “Great trouble, methinks, | would it be to come
to thee,
To wade the waters across, | and wet my middle;
Weakling, well shall I pay | thy mocking words,
If across the sound I come.”
Harbarth spake:
Thor spake:
[127]
Harbarth spake: