Journal of Speech and HearingResearch, Volume 38, 187-199, February 1995
Sentence Comprehension in
Children With Specific Language
Impairment: The Role of
Phonological Working Memory
James W. Montgomery
Division of Speech
and Hearing Sciences
Clinical Centerfor Development
and Learning
University of North Carolina
at Chapel Hill
This study examined the influence of phonological working memory on sentence comprehension in children with specific language impairment (SLI). Fourteen children with SLI and 13 with
normal language (NL) participated in two tasks. In the first, a nonsense word repetition task
(index of phonological working memory), subjects repeated nonsense words varying in length
from one syllable to four. In a sentence comprehension task, subjects listened to sentences
under two conditions varying in linguistic redundancy (redundant, nonredundant). On the
nonsense word repetition task, between- and within-group analyses revealed that subjects with
SLI repeated significantly fewer 3-syllable and 4-syllable nonsense words. On the sentence
comprehension task, between- and within-group analyses determined that subjects with SLI
comprehended significantly fewer redundant (longer) sentences than nonredundant (shorter)
sentences. A positive correlation was found between subjects' performance on the nonsense
word repetition and sentence comprehension tasks. Results were interpreted to suggest that
children with SLI have diminished phonological working memory capacity and that this capacity
deficit compromises their sentence comprehension efforts.
KEY WORDS: children, specific language Impairment, comprehension, phonological
working memory
Over the past several years numerous investigators have suggested that the
language impairment of children with specific language impairment (SLI) may be
related in part to some kind of memory deficiency (e.g., Ceci, Ringstrom, & Lea, 1981;
Curtiss & Tallal, 1991; Graham, 1980). A substantial body of experimental research
in fact reveals that children with SLI seem to have deficits in nearly every major
function of verbal short-term memory (STM), including lexical knowledge (Kail &
Leonard, 1986), scanning speed (Sinninger, Klatzky, & Kirchner, 1989), retrieval
(Ceci et al., 1981; Kail, Hale, Leonard, & Nippold, 1984), and verbal capacity
(Gathercole & Baddeley, 1990a; Kirchner & Klatzky, 1985). However, one drawback
of previous research has been that the documented memory deficits of children with
SLI have not been linked to particular linguistic performance deficiencies. Consequently, the nature of the relationship between deficient verbal STM and linguistic
processing in children with SLI remains unclear.
Given the documented memory difficulties (as well as other cognitive processing
deficits) exhibited by children with SLI, it has been argued that these children might
better be regarded as processing impaired as opposed to language impaired, that is,
having deficits in representational knowledge (Chiat & Hirson, 1987; Curtiss, Katz, &
Tallal, 1992; Curtiss & Tallal, 1991). Curtiss and Tallal (1991), from an analysis of
their longitudinal data comparing the language acquisition patterns of children with
SLI and normal language, propose that children with SLI use the same rules and
© 1995, American Speech-Language-Hearing Association
187
0022-4685/95/3801-0187
188 Journal of Speech and Hearing Research
principles in their grammar-building efforts as normal children. They further propose that children with SLI and normal
children have comparable language learning mechanisms as
well as representational language knowledge. The language
performance deficits that are exhibited by children with SLI,
as argued by these investigators, are instead a reflection of
one or more processing deficiencies (e.g., auditory processing, memory; however, see Gopnik & Crago, 1991, for a
different view regarding representational deficits underlying
developmental language impairment).
In support of their claim, Curtiss and Tallal (1991) provided
indirect evidence that the comprehension difficulties of some
children with SLI relate to an impairment in STM. They
compared the influence of syntactic redundancy (i.e., short
vs. long sentences of comparable syntactic structure and
meaning) on the sentence comprehension of children with
SLI and two groups of normal children. Interestingly, although the age-matched controls "preferred" the redundant
sentences, as did the language-matched controls once they
got older, the children with SLI (regardless of age) "preferred" the shorter, nonredundant sentences. These authors
proposed that an impairment in verbal STM, not syntactic
processing, was responsible for the poorer processing of the
redundant sentences by the children with SLI. Because this
study was not designed to specifically examine the nature of
the memorial processes underlying sentence comprehension, it remains unclear which specific verbal STM process
(or processes) might relate to the sentence processing
difficulties of children with SLI.
One theoretical framework of STM that is especially pertinent to language processing has been proposed by Baddeley and his colleagues. Baddeley (1986) has conceptualized
STM as a multicomponent system referred to as working
memory. For him, working memory is a resource-limited
system comprising a controlling "central executive" that is
subdivided into two distinct "slave" systems, one being the
articulatory loop system and the focus of the present study.
The articulatory loop is assumed to comprise a capacitylimited phonological short-term store and an articulatory
control process (i.e., subvocal rehearsal) that acts to maintain and refresh speech material in the store. Functionally,
the articulatory loop (i.e., phonological working memory
system) is thought to be responsible for the temporary
storage of verbal information while other cognitive tasks,
such as verbal reasoning or auditory and reading comprehension, are performed.
Although phonological working memory has been found to
be important to such linguistic abilities as vocabulary acquisition in young normally developing children (Gathercole &
Baddeley, 1990b), auditory sentence comprehension in children with reading impairment (e.g., Bar-Shalom, Crain, &
Shankweiler, 1993), and adults with neurological dysfunction
(e.g., Baddeley, Vallar, & Wilson, 1987; Vallar & Baddeley,
1984), it has not been investigated extensively in children
with SLI. The two studies that have examined phonological
memory in children with SLI from a working memory perspective have yielded somewhat mixed results and vastly different
interpretations (Gathercole & Baddeley, 1990a; van der Lely
& Howard, 1993). Using a nonsense word repetition task,
Gathercole and Baddeley (1990a) showed that their small
38 187-199
February 995
sample (n = 6) of subjects with SLI had significant difficulties
repeating 3-syllable and 4-syllable nonsense words relative
to 1-syllable and 2-syllable nonsense words compared to two
groups of normal children. They interpreted these results
(i.e., length effect) to suggest that the children with SLI had
reduced phonological memory capacity. Kamhi, Catts,
Mauer, Apel, and Gentry (1988), also using a nonsense word
repetition task, similarly showed that their sample of subjects
with SLI had significantly greater trouble repeating multisyllabic nonsense words than did a group of NL peers. Although
these authors' study was not designed to examine phonological memory, their results could be interpreted as being
consistent with a memory capacity deficit hypothesis. In
contrast to Gathercole and Baddeley's (1990a) capacity
limitation account, van der Lely and Howard (1993), based
on results of several phonological processing tasks, argue
that children with SLI and NL do have comparable STM
capacity. However, vast differences in stimuli, task design,
and task requirements between the van der Lely and Howard
and the Gathercole and Baddeley (1990a) recall tasks most
likely account for the divergence in "recall" findings.
Because the present study focused on examining what
role phonological working memory might play inthe sentence
processing abilities of children with SLI, it was reasoned that
the most relevant feature of the phonological working memory system to initially examine was capacity, as it was assumed
that some degree of working memory capacity is required for
comprehension (Baddeley, 1986; Vallar & Baddeley, 1984).
Memory capacity inthe present study was indexed by the ability
to accurately repeat nonsense words varying in length from one
syllable to four. It has been argued that a nonsense word
repetition task is a "purer" measure of phonological memory
than repeating real words (Gathercole & Baddeley,
1990a,1990b; Henry & Millar, 1991) because successful nonsense word repetition requires listeners to invoke various phonological processes (e.g., perception, encoding, storage, retrieval, production) independent of lexical knowledge (e.g.,
knowledge of phonological and syllable structure, syntax,
meaning; however, see Dollaghan, Biber, & Campbell, 1993,
for an account of prosodic influences on nonsense word
repetition performance). Poorer performance on longer nonsense words than shorter nonsense words (i.e., length effect)
presumably reflects the capacity-limited nature of the phonological memory system (Gathercole & Baddeley, 1990b).
It would appear that our understanding of the nature of the
sentence processing difficulties of children with SLI could be
enhanced by appealing to the construct of phonological
working memory. Although previous research has documented independent deficits in phonological STM (Gathercole & Baddeley, 1990a) and sentence processing (Curtiss &
Tallal, 1991) in children with SLI, these studies have failed to
examine whether a relationship exists between these deficits
in these children. The present study was thus designed to
explore the possibility of just such a relationship by obtaining
an independent measure of phonological working memory
and then relating it to variations in sentence-level processing
demands in the same sample of children with SLI. Results
from this study should provide us greater insight into at least
one specific memorial process underlying the language
comprehension difficulties of children with SLI. It may turn out
Montgomery: Comprehensionand Working Memory
that phonological memory plays a vital role in sentence
comprehension, especially if it is assumed that listeners store
complete sentences in memory before completing syntactic
and semantic analyses of the input (Clark & Clark, 1977).
Alternatively, sentence comprehension may not rely on phonological memory at all (Butterworth, Campbell, & Howard,
1986). It might be that listeners are able to complete all
syntactic and semantic analyses and construct sentence
meaning without having to store the input in some kind of
phonological form. Finally, phonological memory may play a
more intermediate role in supporting comprehension in that
only portions of the input are stored phonologically in memory while syntactic and semantic processing takes place
(Baddeley et al., 1987).
Accordingly, the main questions addressed in this study
were (a) Do SLI and language-matched children differ with
respect to their phonological memory capacity? (b) Do SLI
children show greater difficulty processing longer, linguistically redundant sentences than shorter, nonredundant sentences relative to language-matched children? and (c) Is
there a relationship between phonological working memory
capacity and sentence comprehension?
189
TABLE 1. Chronological age (months), PPVT score, receptive
language score (RLS), and expressive language score (ELS) on
the CELF-R, and nonverbal IQ for individual subjects with
normal language (NL) and specific language impairment
(SLI).
CELF-R"
Age
(mos)
PPVT
Score
RLS
ELS
IQb
Mean
SD
81.31
15.38
115.77
6.92
107.38
4.68
108.15
4.21
108.46
7.11
Range
61-110
96-123
103-120
100-114
95-118
Subject
NL
SLI
100.50
74.00
70.71
87.21
Mean
98.40
5.04
10.24
SD
21.20
9.26
5.24
85-118
60-78
63-80
Range 72-134 76-108
aClinical Evaluation of Language Fundamentals (CELF-R); Receptive Language Score (RLS), and Expressive Language Score (ELS)
expressed as standard scores.
bTest of Nonverbal Intelligence.
Method
Subjects
Fourteen children with SLI (mean CA = 98.4 months) and
13 normally developing (NL) children (mean CA = 81.3
months) participated in a nonsense word repetition task and
a sentence comprehension task. Thirteen of the subjects with
SLI had been diagnosed previously as language-impaired by
a certified speech-language pathologist and had received or
were currently receiving speech-language intervention. Subjects with SLI met the following language-specific criteria: (a)
performed at a minimum -1.5 standard deviations below the
mean on at least two of the three subtests on each of the
receptive and expressive portions of the Clinical Evaluation
of Language Fundamentals-Revised (CELF-R; Semel,
Wiig, & Secord, 1987); (b) attained a receptive language
score and an expressive language score that fell at least
-1.5 standard deviations below the mean; and (c) performed
below the 25th centile on the Test of Reception of Grammar
(TROG; Bishop, 1989). The subjects with NL were required
to perform at or above the mean (or 25th centile) on the same
standardized language measures. Relative to the NL subjects, subjects with SLI evidenced significantly poorer receptive and expressive abilities, as demonstrated by their attaining a significantly lower total receptive language score [t(25)
= 17.38, p < .01] and total expressive language score [t(25)
= 20.82, p < .01] on the CELF-R. Although performance on
a measure of receptive lexical knowledge was not part of the
entrance criteria, subjects were administered the Peabody
Picture Vocabulary Test-Revised (PPVT-R; Dunn & Dunn,
1981) for purposes of partialling out potential group differences lexical knowledge might have on subjects' nonsense
word repetition performance. As a group the subjects with
SLI attained a significantly lower mean PPVT standard score
(87.21) than the NL subjects (115.76) [t(25) = 2.47, p < .01].
All subjects evidenced (a) at least normal-range nonverbal
IQ (85-120) on the Test of Nonverbal Intelligence (Brown,
Sherbenou, & Johnsen, 1990), (b) normal-range hearing
sensitivity as determined by audiometric pure tone screening
at 20 dBHL (ANSI, 1973), (c) articulation abilities that fell at or
above the 70th percentile on the Goldman-Fristoe Test of
Articulation (1986)1, (d) no oral structural or motor impairments affecting speech or nonspeech movements of the
articulators (Robbins & Klee, 1987), and (e) normal or
corrected vision. No subject had a history of frank neurologic
impairment or psychological/emotional disturbance or attention deficit disorder (from parent report).
Although it has been argued that language matching
procedures are problematic (Plante, Swisher, Kiernan, &
Restrepo, 1993), individual subjects with SLI and NL were
matched on the TROG using raw scores, thereby minimizing
potential group differences in sentence-level syntacticsemantic knowledge. It was reasoned that by matching on
this variable the interpretation of any group differences on the
experimental sentence comprehension measure, particularly
the redundant sentences, would be facilitated. Two of the
subjects in the group with SLI and one of the subjects in the
NL group were of African American or Hispanic descent.
Group cognitive and language data are displayed in Table 1,
and individual subject data appear in Appendix A.
'It should be noted that all subjects' phonetic inventories contained those
phonemes present in the nonsense word stimuli used in Experiment 1.
190 Journalof Speech and Hearing Research
TASK 1: REPETITION OF NONSENSE
WORDS
Nonsense Word Stimuli
Twelve nonsense words were created at each of four
syllable lengths (1, 2, 3, 4), for a total of 48 stimuli. Stimulus
words did not contain any word-initial consonant clusters or
the phonemes /r/, /I/, or I/th/ in word-initial position, constraints
that were intended to minimize complex articulatory/output
demands. Half of the items contained an initial stop consonant; the other half a nonstop consonant (e.g., fricative,
affricate, nasal). Stimulus items conformed to the following
phonetic structures: CVC, CVCVC, CVCVCVC, CVCVCVCVC.
Appendix B displays the experimental nonsense words.
Stimulus Tape Generation Procedures
High-quality audiocassette recordings were made of one
male adult speaker producing each nonsense word in an
acoustically isolated booth. A SONY HX Pro audiocassette
recorder and a 1070-B Realistic microphone were used for
recording. Stimulus items were produced at a normal rate
and with the following stress patterns according to word
length: 1-syllable = strong (e.g., CAID); 2-syllable = strongweak (e.g., SHUdep); 3-syllable = strong-weak-weak (e.g.,
BOfudish); 4-syllable = variable (e.g., zoPANishful, CONishament). Each recorded item was low-pass filtered (f =
4.5 KHz) using a Wavetech variable analog filter, digitized
(10 Ksamples/sec) using an analog-to-digital converter (Data
Translation-DT2820 series), and stored on disk (CompuAdd
316SX Laboratory computer). From a digital representation,
each nonsense word was edited interactively using a custom-written waveform editing routine (written with ASYST
software) to eliminate any noise preceding and/or following
the nonsense word. Acoustic analyses of the stimulus items
revealed that overall duration increased significantly as the
number of syllables increased. Digitized items were then
output (10 ksamples/sec) and low-pass filtered (fc = 4.5 kHz)
to audiotape. Recorded stimulus items were controlled for
overall relative intensity (1.5 volts) using a Techtronics
analog oscilloscope. A 3-sec interstimulus interval separated
each stimulus item on the tape.
TASK 2: COMPREHENSION OF
NONREDUNDANT AND REDUNDANT
SENTENCES
Sentence Comprehension Stimuli
Two sets of 20 sentences each were created corresponding to a set of linguistically redundant (longer) sentences and
a set of linguistically nonredundant (shorter) sentences (e.g.,
Curtiss & Tallal, 1991). The redundant set consisted of four
sentence types, three of which were used by Curtiss and
38
187-199 Febrary
995
Tallal (1991): (a) sentences containing double marking of
number (e.g., "Point to the picture of the three cats"); (b)
semantically reversible sentences with a single embedded
subject relative clause (e.g., "The girl who is smiling is
pushing the boy"); (c) semantically reversible sentences with
a double embedded subject and object relative clause (e.g.,
"The little boy who is standing is hitting the little girl who is
sitting"); and (d) active sentences with adjectival/adverbal
material modifying the subject and/or object noun (e.g., "The
dirty little boy climbs the big fat tree"). The nonredundant and
redundant sentences were nearly identical structurally and
encoded essentially the same semantic information. The only
difference between the sentence conditions was that the
redundant cues and modifying adjectival/adverbial lexical
items were absent in the nonredundant sentences, thereby
making these sentences shorter (e.g., "Point to the picture of
the cats." "The girl smiling is pushing the boy." "The little boy
standing is hitting the little girl sitting." "The little boy climbed
the fat tree."). Although the redundant sentences in sentence-type 4 were not redundant in a syntactic sense as in
the other sentence types, they were defined as such because
the added verbiage served as redundant (nonessential) cues
to sentence interpretation. The mean number of words
contained in the redundant and nonredundant sentences
was 11.20 and 7.95, respectively. The experimental sentences appear in Appendix C.
A high-quality cassette recording of the stimulus sentences
was made of the same male speaker reading each sentence
at a normal conversational rate with normal prosodic variation. Across the 40 experimental trials, redundant and nonredundant sentences appeared randomly.
Picture Stimuli
For each of the 40 stimulus-sentence pairs (i.e., nonredundant and redundant item), four color pictures were created,
one matching the stimulus sentence and three foils. Foil
pictures differed from the target picture along only one or two
relevant semantic dimensions (e.g., gender or size of the
sentence's subject/object; reversed agency of subject and
object; size, color, or number of object). A stimulus bookletcontaining 50 pre-experimental pictures (corresponding to
the nouns, verbs, adjectives, adverbs contained in the experimental sentences), 6 practice pictures, and then the 40
experimental pictures-was created. Target pictures appeared equally often in each quadrant of the stimulus page.
Experimental Procedures and Scoring
Nonsense word repetition task. Subjects received in
random order the 48 experimental items (plus 4 practice
items) binaurally via headphones (Denon AH-D100) at a
comfortable listening level. Subjects were asked to listen to
some "pretend" words and to repeat each one immediately
after hearing it. Subjects were given 5 sec to respond. If
needed, they were also given one stimulus repetition and two
opportunities to produce the item. Subject responses were
tape recorded and later broadly phonetically transcribed for
accuracy and judged correct or incorrect. Each subject's final
Montgomery: Comprehensionand Working Memory
production was the response that was scored. This procedure was intended not to penalize subjects whose performance may have been affected by output constraints. A
second listener (graduate student in speech pathology) transcribed 50% of SLI and 50% of the NL tapes (chosen at
random). Point-to-point transcriptions yielded 97% and 94%
agreements inscoring decisions for the subjects with NL and
SLI, respectively.
Sentence comprehension task. Subjects received the 40
experimental sentences via headphones at a comfortable
listening level. While listening to each sentence, subjects
were shown an array of four pictures. After hearing each
sentence, they were asked to point to the picture corresponding to the sentence. Subjects were allowed one additional
presentation of each stimulus sentence. Subject responses
were scored as correct or incorrect. Before experimental
testing, a pretest was administered to assess subjects'
knowledge of the nouns, verbs, adjectives, and adverbs
contained in the experimental sentences. All subjects performed with 100% accuracy.
191
Results
.
.___.._
Nonsense Word Repetition Task
The dependent variable was the number of correctly
imitated nonsense words at each stimulus length. Figure 1 is
a plot of both groups' nonsense word repetition as a function
of stimulus length. Inspection of Figure 1 suggests that,
compared to NL subjects, the subjects with SLI performed
more poorly on the 3-syllable and 4-syllable items but not on
the 1- and 2-syllable items. Between-group analyses of
variance substantiated these impressions by revealing that,
compared to NL subjects, the subjects with SLI performed
significantly less well on the 3-syllable nonsense words
[F(1,25) = 19.78, p < .001] and 4-syllable nonsense words
[F(1,25) = 35.20, p < .001]. On the 1-syllable and 2-syllable
items, however, no group differences were found.
It can also be seen from Figure 1 that the nonsense word
repetition of the subjects with SLI was influenced to a greater
12
11
U)
10
9
a)
0
8
7
C
a)
6
5
U)
C
4
Z
3
a
2
L..
.-
0
1
O
1-Syllable
2-Syllable
3-Syllable
4-Syllable
Stimulus Length
FIGURE 1. Mean number of correct nonsense word repetitions by word length for the subjects with normal language (NL) and
for those with specific language Impairment (SLI).
192 Journal of Speech and Hearing Research
38
TABLE 2. Mean number of nonsense words repeated correctly
by word length for the subjects with NL and SLI (total possible
correct at each word length Is 12).
1-Syll
2-Syll
3-Syll
4-Syll
Grand
Mean
NL
Mean
SD
Range
11.53
.52
11-12
10.69
.85
9-12
9.61
1.19
8-11
8.69
1.43
7-12
40.53
2.87
36-45
SLI
Mean
SD
Range
11.00
.96
9-12
10.21
1.36
8-12
6.42
2.31
4-11
4.35
2.23
2-8
32.00
5.02
27-41
degree by stimulus length, a finding supported by a 2-way
repeated-measures ANOVA, Group (2) x Word Length (4),
that yielded a significant Group x Word Length interaction
[F (15,200) = 23.03, p < .001]. Figure 1 reveals that the
performance decrement of the subjects with SLI was precipitous beginning with 3-syllable items, whereas NL subject's
performance diminished quite gradually. To evaluate the
influence of stimulus length on each subject group's nonsense word repetition, within-group post hoc ANOVAs and
Tukey analyses (p < .05) were performed. For the subjects
with SLI, nonsense word repetition indeed decreased sharply
as word length increased [F(3,52) = 42.07, p < .001]. The
subjects with SLI repeated significantly fewer 3-syllable
stimuli than 2-syllable stimuli, and significantly fewer 4-syllable stimuli than 3-syllable stimuli. Although NL subject's
repetition accuracy also deteriorated across word length
[F(3,48) = 17.89, p < .001], their performance decrement
was gradual. They repeated significantly fewer 3-syllable
items compared to 1-syllable items and significantly fewer
4-syllable items than 2-syllable items. No other comparisons
reached significance for these subjects. Group means, standard deviations, and ranges appear in Table 2.
The length effect was clearly of greater magnitude across
individual children with SLI relative to individual NL children
(see Appendix D for the number of nonsense words correctly
repeated at each stimulus length by individual subjects with
NL and SLI). The nonsense word repetition of most of the
children with SLI (11/14) markedly decreased for both 3- and
4-syllable stimuli. In contrast, the nonsense word repetition of
NL subjects gradually decreased as stimulus length increased. In fact, only three NL subjects produced two or
more additional errors for 3-syllable nonsense words relative
to 2-syllable nonsense words, and only four subjects produced two or more additional errors for 4-syllable stimuli
relative to 3-syllable stimuli.
To determine whether group differences in receptive lexical knowledge contributed to the inferior nonsense word
repetition of the subjects with SLI, a second 2-way repeatedmeasures ANOVA was performed using PPVT scores as the
covariate. As before, a significant Group (2) x Word Length
(4) interaction was obtained [F(17,198) = 19.68, p < .001],
indicating that the poorer nonsense word imitation of the
subjects with SLI was not attributable to group differences in
receptive lexical knowledge.
187-199 February 995
Sentence Comprehension Task
The children with SLI and NL demonstrated different
patterns of performance across sentence-type conditions.
Figure 2 is a plot of both groups' sentence comprehension
performance by sentence-type condition. Inspection of Figure 2 suggests that compared to the NL subjects, subjects
with SLI performed more poorly on the redundant sentences
but not on the nonredundant sentences. Indeed, betweengroup analyses of variance supported these observations by
indicating that subjects with SLI performed significantly more
poorly on the redundant sentences [F(1,25) = 24.55, p <
.0001], whereas on the nonredundant sentences the groups
were found to perform comparably. Inspection of Figure 2
further suggests that each subject group's sentence comprehension was differentially affected by linguistic redundancy, a
finding supported by a 2-way repeated measures ANOVA,
Group (2) x Sentence-type (2), that yielded a significant
interaction [F(1,50) = 5.47, p < .02]. Results of within-group
post hoc analyses of variance revealed that subjects with SLI
comprehended a significantly fewer number of redundant
sentences than nonredundant sentences [F(1,26) = 9.80,
p < .01], whereas NL subjects performed comparably across
sentence types. Group means and standard deviations per
sentence-type condition are displayed in Table 3.
Linguistic redundancy clearly had a greater negative influence on each of the subjects with SLI than it had on individual
NL subjects (see Appendix E for the number of correct
20
~I
* NL
E SLI1
~~~~I
i
6o
c
II
(DO 11
n
Oa
a a)10
O_
rr
_
A k__
v
Nonredundant
Redundant
Sentence Type
FIGURE 2. Mean number of correct responses per sentencetype condition (nonredundant, redundant) by the subjects with
normal language (NL) and by those with specific language
impairment (SLI).
Montgomery: Comprehension and Working Memory
lation was computed for nonsense word repetition performance and performance on the sentence comprehension
task. A positive correlation was found (r = .62, p < .001).
Figure 3 graphically presents the relation between nonsense
word repetition ability and sentence comprehension by each
subject. Generally, those subjects attaining lower scores on
the nonsense word repetition task also yielded lower scores
on the sentence comprehension measure (i.e., subjects with
SLI depicted by squares). Conversely, higher scores on the
sentence comprehension task were typically obtained by
those subjects who obtained higher scores on the nonsense
word repetition task (i.e., NL subjects depicted by triangles).
TABLE 3. Mean number of correct responses by sentence-type
condition (nonredundant, redundant) for the subjects with NL
and SLI (total possible correct per condition Is 20).
Subject
Nonredundant
Sentences
Redundant
Sentences
Grand
Mean
NL
Mean
SD
17.76
1.43
17.69
1.64
35.46
2.90
Range
15-20
15-20
31-40
SLI
Mean
SD
Range
16.50
1.78
14-19
14.14
2.17
11-17
30.64
3.73
27-36
Discussion
-------
responses obtained by each subject with SLI and NL under
each sentence-type condition). Thirteen out of the 14 subjects with SLI attained a lower score in the redundant
condition than in the nonredundant; the remaining subject
performed comparably across conditions. Moreover, compared to his/her NL control, 13 of the 14 subjects with SLI
obtained lower scores in the redundant condition. In contrast,
only three NL subjects attained lower scores inthe redundant
condition than the nonredundant, whereas the remaining
subjects either performed comparably in both conditions
(n = 7) or attained higher scores in the redundant condition
(n = 3).
This study was designed to evaluate what role phonological working memory might play in the sentence comprehension of children with SLI. Subjects completed a nonsense
word repetition task and a sentence comprehension task.
The results of the nonsense word repetition task (index of
phonological working memory) revealed that, relative to the
NL children, the children with SLI performed more differently
than similarly. The one similarity was that for both groups
accuracy of nonsense word repetition decreased as stimulus
length increased. Despite this general similarity, however,
the children with SLI were found to perform significantly more
poorly on the longest items than the NL children.
The present findings are consistent with previous reports
of difficulties by children with SLI repeating multisyllabic
nonsense words (Gathercole & Baddeley, 1990a; Kamhi et
al., 1988). With respect to the Gathercole and Baddeley
(1990a) findings, the subjects with SLI in the present study
Relation Between Phonological Memory
and Sentence Comprehension
To examine the relation between phonological memory
and sentence processing, a Pearson product-moment corre48
47
..
46
44
_
a)
71 3
0
C 3..
- -
------
---
-----
..
--
26
C 37
------ ------
.
.~j,1 40
---------A----....----.-----
. ..-----
.
..
SLI----1-1------------C--- ---
43
° co42
4
..
....
U
3
35
345_ .-..
.
... .-
.- .
----9
....-.
.
A
-.
.
.
....
........ ..
...
......................
--------
Z
26
22
21
-
-
--
193
--
---------
____
--------------------------------
-
- ------
.--
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
Sentence Comprehension
(Total possible correct = 40)
FIGURE 3. Plot representing the relationship between nonsense word repetition and sentence
comprehension for subjects with normal language (NL) and for those with specific language
impairment (SLI).
194 Journal of Speech and Hearing Research
also demonstrated a strong length effect, showing a precipitous performance decrement when repeating 3- and 4-syllable nonsense words relative to 1- and 2-syllable nonsense
words. However, unlike the NL subjects in the Gathercole
and Baddeley study who showed no length effect, the NL
children in the present study did, albeit a weak one. Differences between the two studies in subject number, sampling,
matching procedures, and experimental stimuli most likely
account for this variance in findings.
The present findings also extend those of Gathercole and
Baddeley by demonstrating that nonsense word repetition
deficits also seem to be characteristic of older children with
SLI. The oldest children with SLI studied in the Gathercole
and Baddeley study were under 9 years of age, whereas the
present study included five children with SLI between the
ages of 9 and 11. It is interesting to note that even the
youngest NL subjects outperformed the oldest subjects with
SLI on the 3- and 4-syllable nonsense words (see
Appendix D).
The inferior nonsense word repetition of the children with
SLI suggests the interpretation that these children had a
limitation in phonological memory capacity (Gathercole &
Baddeley, 1990a). In support of this interpretation, recall that
the children with SLI performed especially poorly on the 3and 4-syllable nonsense items relative to (a) their own
performance on the 1- and 2-syllable nonsense words and
(b) the NL controls. Additional support for this interpretation
derives from the fact that the magnitude of the length effect
appeared to be markedly greater for the children with SLI
than the NL children. Finally, the fact that a robust group
difference in nonword repetition remained even after group
differences in receptive lexical knowledge were partialled out
lends further support to the interpretation that a memory
deficit most likely accounted for the poor nonsense word
repetition of the children with SLI.
It has been argued that successful imitation of nonsense
words is dependent upon a number of phonological abilities
(e.g., perception, encoding, rehearsal, articulatory/output
abilities), in addition to an intact temporary storage system. A
deficit in one or more of these processes may have contributed in part to the inferior nonsense word repetition performance of the children with SLI in the present study. However,
results from other studies examining these phonological
processing abilities in children with SLI appear to weaken this
argument. For instance, Gathercole and Baddeley (1990a)
compared the perceptual, encoding, and rehearsal processes in their samples of children with SLI and NL and
found no group differences for any of these abilities. Similarly, van der Lely and Howard (1993) found that their
subjects with SLI and NL performed comparably on a phonological encoding task. Likewise, the possibility of group
differences in articulatory-motor abilities is weakened given
that (a) Gathercole and Baddeley (1990a) found that their
subjects with SLI and NL demonstrated comparable rates of
articulation for 1- and 3-syllable word productions, (b) Stark
and Montgomery (1994) similarly found that their subjects
with SLI and NL did not differ in rate of articulation when
repeating strings of nonsense syllables, (c) all stimulus words
in the present study were designed to minimize complex
articulatory demands, and (d) all subjects in the present
38 187-199 February 995
study had normal-range articulation and oral-motor abilities.
Considering these findings/factors together, it might be argued that these phonological processes probably did not
contribute (at least not strongly) to the observed group
differences in nonsense word repetition. Instead, a limitation
in phonological storage capacity was the most likely source
of difficulty for the children with SLI. However, a comparison
of the perceptual, encoding, and articulatory abilities of a
subset of the same children with SLI and NL who participated
in this study is presently under investigation as a means to
provide replication and extension of the Gathercole and
Baddeley (1990a) findings in particular.
One final caveat should be mentioned concerning a capacity limitation interpretation. The metaphonological processes of segmentation and blending have also been proposed as being critical to successful nonsense word
repetition (Snowling, Chiat, & Hulme, 1991), although there
are no data yet supporting this hypothesis. Given that children with SLI typically show delayed acquisition of metalinguistic skills, particularly syllable and phoneme segmentation
abilities, the children with SLI in the present study may
indeed have done more poorly repeating nonsense words
because of metalinguistic difficulties. This possibility too
might be weakened, however, by the fact that the subjects
with SLI in the present study were generally older (mean CA
= 8.2 years; range = 6.0 to 11.2 years) than the children with
SLI in previous metalinguistic studies (e.g., Kamhi & Catts,
1986; Kamhi & Koenig, 1985). Nonetheless, the possibility
that reduced metaphonological abilities relate to the poor
nonsense word repetition of children with SLI warrants examination.
On the sentence comprehension task, the children with SLI
were found to perform more poorly than their NL counterparts. Significantly, the group difference was solely due to the
poorer performance of the children with SLI on the redundant
sentences; on the shorter, linguistically nonredundant sentences the groups performed comparably. Most important,
however, relative to their own performance on the nonredundant sentences, the children with SLI comprehended significantly fewer of the redundant sentences. The NL children, in
contrast, performed comparably across sentence conditions.
Together, these findings suggest the interpretation that the
poor comprehension of the redundant sentences by the
children with SLI was not attributable to a lack of sentencelevel syntactic-semantic knowledge, but instead to difficulty
managing the increased demands on phonological working
memory. As sentence length increased, so did the demands
for concurrently storing longer sequences of words while
processing new input. These findings, coupled with the
findings of Task 1, could be interpreted to suggest that a
capacity limitation in phonological working memory compromised the comprehension efforts of the children with SLI.
Additional support for this interpretation derives from the
positive correlation found between subjects' performances
on the nonsense word repetition and sentence comprehension tasks. It is important to point out that although limited
phonological memory capacity appears to be the strongest
candidate explanation of the sentence processing difficulties
of the children with SLI, it may be but one potential factor.
Because the correlation between phonological memory and
Montgomery: Comprehensionand Working Menuory
sentence comprehension was only +.62, other sources of
variance (e.g., perceptual processing, phonological encoding, rehearsal) may also contribute to these children's comprehension difficulties.
The findings from the sentence comprehension experiment are both consistent and at variance with the findings of
Curtiss and Tallal (1991). The findings showing that the
children with SLI in the present study performed more poorly
on the longer sentences than the shorter sentences are
consistent with the findings reported by Curtiss and Tallal.
Recall, however, that the age-matched and older languagematched NL subjects in the Curtiss and Tallal study were
found to "prefer" the longer, more redundant sentences than
the shorter sentences. By contrast, the NL subjects in the
present study showed no preference for redundant sentences over nonredundant sentences; their performance in
both sentence conditions was essentially identical.
Taken together, the findings from these two experiments
provide preliminary empirical evidence of a link between
impaired memory functioning and a specific linguistic processing deficit in children with SLI. Consistent with the claim
of Curtiss and Tallal (1991), the sentence processing deficits
of at least some children with SLI do not appear to be related
to linguistic-specific knowledge deficits but rather to an
impairment in verbal memory.
How might phonological working memory relate to the
process of sentence comprehension? It seems clear that
sentence comprehension requires that previous information
be stored temporarily while new, incoming information is
processed. Two extreme positions have been proposed.
Clark and Clark (1977) have proposed that phonological
memory is critical to comprehension because listeners presumably store entire sentences in a phonological input store
until all syntactic and semantic analyses have been completed. On such a view, it might be hypothesized that the
children with SLI should have had equal difficulty comprehending the redundant and nonredundant sentences because their memory capacity deficit would have prevented
them from storing a sentence of any length. This clearly was
not the case, suggesting that what was stored in phonological memory was not an entire sentence. In direct opposition
to this view, Butterworth et al. (1986) have argued that
phonological memory plays no role at all in sentence comprehension. If this were the case, the subjects with SLI might
have been expected to perform equally well across sentence
conditions and perhaps comparably to the NL subjects in
both conditions (given that they were matched on sentencelevel syntactic-semantic knowledge) because their storage
deficit should not have entered into play during comprehension. The present findings also are clearly at variance with
this view, given that the children with SLI comprehended
significantly fewer redundant than nonredundant sentences.
An alternative and intermediate view of the relationship
between phonological working memory and sentence comprehension has been put forward by Baddeley et al. (1987).
They argue that the phonological store of the articulatory loop
functions as a "mnemonic window" in which sequences of
incoming lexical items are held (presumably in some sort of
literal form that maintains serial order of information) while
the items within the sequences are processed and inter-
195
preted, thereby allowing "mental models or representations"
(i.e., semantic representation) of the input to be constructed.
The "mnemonic window" proposal generally corresponds to
models of sentence comprehension in which phonological
memory is involved in the pre-parsing stages of comprehension (e.g., Waters, Caplan, & Hildebrandt, 1987) during
which lexical items, along with their corresponding syntactic
categories, are maintained in the phonological store (i.e.,
"look-ahead buffer") before any syntactic structures are
constructed. According to this view, it might have been
predicted that, relative to themselves and the NL subjects,
the subjects with SLI should have had greater difficulty
processing longer sentences than shorter ones. This prediction follows from the assumption that if the window in children
with SLI is reduced they should be less able to represent
longer sequences of words in memory, thus preventing them
from constructing as accurate a mental model of what they
hear. Tentative support for this view derives from the finding
that the subjects with SLI did indeed have greater difficulty
comprehending the longer, redundant sentences than the
shorter, nonredundant ones.
Summary
The findings from this study provide preliminary evidence
of a processing deficit account of the sentence comprehension difficulties of at least some children with SLI. The
sentence comprehension difficulties of some children with
SLI indeed appear to be related, at least in part, to a capacity
deficit in phonological working memory. The findings from
this study, however, in no way preclude the existence of
other verbal memory difficulties potentially contributing to
poor sentence comprehension. The present interpretation is
consistent with recent claims that some children with SLI
might be better conceptualized as processing impaired as
opposed to specifically language impaired (e.g., Chiat &
Hirson, 1987; Curtiss & Tallal, 1991). As phonological working memory has been shown to be important to a variety of
cognitive performances (Baddeley, 1986), future research
with children with SLI and NL might not only examine more
systematically how and when phonological memory exerts its
influence during spoken language comprehension, but also
during the performance of a range of cognitive-linguistic
problem-solving activities. Results from such efforts should
provide us greater insights into how children with SLI manage their cognitive resources in the service of languagerelated processing activities. Clearly, more sensitive treatment paradigms could be developed if we could determine in
what ways processing limitations contribute to the language
difficulties of children with SLI.
Acknowledgments
The author thanks Tim Brown for his assistance with data analysis
and Cheryl Hunter for preparing the figures. This project was
supported by a New Investigators Award from the American Speech,
Language, and Hearing Foundation and by grants awarded to the
Center for the Study of Development and Learning by the Child
196
Journal of Speech and Hearing Research
Health Bureau (#MCJ-379154-02-0) and the Administration on
Developmental Disabilities (#90DD0207).
References
American National Standards Institute (1973). American National
Standards Specification for Audiometers (ANSI S3.6-1969,
R1973). New York: ANSI.
Baddeley, A. (1986). Working memory and comprehension. In D.
Broadbent, J. McGaugh, M. Kosslyn, N. Mackintosh, E. Tulving,
& L. Weiskrantz (Eds.), Working memory(pp. 54-108). New York:
Oxford University Press.
Baddeley, A., Vallar, G., & Wilson, B. (1987). Sentence comprehension and phonological memory: Some neuropsychological
evidence. In M. Coltheart (Ed.), Attention and performance XII:
The psychology of reading (pp. 507-529). London: Erlbaum
Associates.
Bar-Shalom, E.G., Crain, S., & Shankweller, D. (1993). A comparison of comprehension and production abilities of good and poor
readers. Applied Psycholinguistics, 14, 197-227.
Bishop, D.V.M. (1989). Test of Reception of Grammar. Manchester,
England: University of Manchester.
Brown, L., Sherbenou, R., & Johnsen, S. (1990). Test of Nonverbal Intelligence (2nd ed.). Austin, TX: PRO-ED.
Butterworth, B., Campbell, R., & Howard, D. (1986). The uses of
short-term memory: A case study. Quarterly Journal of Experimental Psychology, 38A, 705-738.
Ceci, S., Rlngstrom, M., & Lea, S. (1981). Do language learning
disabled children have impaired memories? In search of underlying processes. Journal of Learning Disabilities, 14, 159-173.
Chlat, S., & Hirson, A. (1987). From conceptual intention to
utterance: A study of impaired output in a child with developmental
dysphasia. British Journal of Disorders of Communication, 22,
37-64.
Clark, H., & Clark, E. (1977). Psychology and language. New York:
Harcourt Brace Jovanovic.
Curtiss, S., Katz, W., & Tallal, P. (1992). Delay versus deviance in
the language acquisition of language-impaired children. Journal of
Speech and Hearing Research, 35, 373-383.
Curtlss, S., & Tallal, P. (1991). On the nature of the impairment in
language-impaired children. In J. Miller (Ed.), Research on child
language disorders: A decade of progress. Austin, TX: PRO-ED.
Dollaghan, C., Biber, M., & Campbell, T. (1993). Constituent
syllable effects in a nonsense-word repetition task. Journal of
Speech and Hearing Research, 36, 1051-1054.
Dunn, L. M., & Dunn, L. M. (1981). Peabody Picture Vocabulary
Test-Revised. Circle Pines, MN: American Guidance Service.
Gathercole, S., & Baddeley, A. (1990a). Phonological memory
deficits in language disordered children: Is there a causal connection? Journal of Memory and Language, 29, 336-360.
Gathercole, S., & Baddeley, A. (1990b). The role of phonological
memory in vocabulary acquisition: A study of young children
learning new words. British Journal of Psychology, 81, 439-454.
Goldman, R., & Fristoe, M. (1986). Goldman-Fristoe Test of Articulation. Circle Pines, MN: American Guidance Service.
Gopnlk, M., & Crago, M. (1991). Familial aggregation of a developmental language disorder. Cognition, 39, 1-50.
Graham, N. (1980). Memory constraints in language deficiency. In F.
Jones (Ed.), Language disability in children (pp. 69-84). Baltimore: University Press.
38
187-199
Februay 995
Henry, L., & Millar, S. (1991). Memory span increase with age:
A test of two hypotheses. Journal of Experimental Child Psychology, 51, 459-484.
KalI, R., Hale, C., Leonard, L., & Nippold, M. (1984). Lexical
storage and retrieval in language impaired children. Applied Psycholinguistics, 5, 37-49.
Kall, R., & Leonard, L. (1986). Word finding abilities in language
impaired children (ASHA Monographs 25). Rockville, MD: American Speech-Language-Hearing Association.
Kamhl, A., & Catts, H. (1986). Toward an understanding of developmental language and reading disorders. Journal of Speech and
Hearing Disorders, 51, 337-347.
Kamhl, A., Catts, H., Mauer, D., Apel, K., & Gentry, B. (1988).
Phonological and spatial processing abilities in language- and
reading-impaired children. Journal of Speech and Hearing Disorders, 53, 316-327.
Kamhl, A., & Koenig, L. (1985). Metalinguistic awareness in normal
and language-disordered children. Language, Speech, and Hearing Services in Schools, 16, 199-210.
Kirchner, D., & Klatzky, R. (1985). Verbal rehearsal and memory in
language disordered children. Journal of Speech and Hearing
Research, 28, 556-564.
Plante, E., Swisher, L., Kleman, B., & Restrepo, M. (1993).
Language matches: Illuminating or confounding? Journal of
Speech and Hearing Research, 36, 772-775.
Robbins, J., & Klee, T. (1987). Clinical assessment of oropharyngeal motor development in young children. Journal of Speech and
Hearing Disorders, 52, 271-277.
Semel, E., Wilg, E., & Secord, W. (1987). Clinical evaluation of
language fundamentals-Revised. The Psychological Corporation.
Slnnlnger, Y., Klatzky, R., & Klrchner, D. (1989). Memory scanning speed in language disordered children. Journal of Speech
and Hearing Research, 32, 289-298.
Snowling, M., Chlat, S., & Hulme, C. (1991). Words, non-words
and phonological processes: Some comments on Gathercole,
Willis, and Baddeley. Applied Psycholinguistics, 12, 369-373.
Stark, R. E., & Montgomery, J. W. (1994). Rapid syllable production in specifically language impaired children. In N. Jordan and J.
Goldsmith-Phillips (Eds.), Leaming disabilities: New directions for
assessment and intervention (pp. 101-122). Boston: Allyn and
Bacon.
Vallar, G., & Baddeley, A. (1984). Phonological short-term store,
phonological processing and sentence comprehension: A neuropsychological case study. Cognitive Neuropsychology, 1, 121141.
van der Lely, H. K., & Howard, D. (1993). Children with specific
language impairment: Linguistic impairment or short-term memory
deficit? Journal of Speech and Hearing Research, 36, 1193-1207.
Waters, G., Caplan, D., & Hlldebrandt, N. (1987). Working memory
and written sentence comprehension. In M. Coltheart (Ed.), Attention and performance XII: The psychology of reading (pp. 531555). London: Erlbaum Associates.
Received March 10, 1994
Accepted August 5, 1994
Contact author: James W. Montgomery, PhD, Center for Development and Learning, CB #7255, BSRC, University of North Carolina, Chapel Hill, NC 27599-7255. E-mail: jim=montgomeryo°/faculty%
[email protected]
Montgomery: Comprehensionand Working Memory
Appendix A
Appendix B
Chronological age (months), PPVT score, receptive language
score (RLS), and expressive language score (ELS) on the
CELF-R, and nonverbal IQ for Individual subjects with normal
language (NL) and specific language Impairment (SLI)
Experimental Nonsense Words
Subject
Svlahb II annth
1
CELF-Ra
Gender
Age
(mos)
PPVT
Score
RLS
ELS
IQ
M
F
M
F
F
M
M
M
F
M
M
M
F
72
110
100
86
77
72
67
74
63
68
61
86
107
119
123
97
96
123
122
115
118
110
116
111
118
117
107
120
103
110
103
107
107
104
105
106
110
103
111
110
110
100
112
112
108
110
107
108
109
106
100
114
111
103
95
108
111
102
118
98
111
109
118
114
112
b
1
7
8
9
10
11
12
13
SLI
F
80
92
78
78
111
F
78
93
77
70
102
M
89
81
78
76
103
M
110
81
72
67
110
M
88
80
63
60
85
6
M
134
76
77
72
99
7
F
100
76
63
70
87
8
M
132
81
75
72
95
9
M
75
86
74
72
88
10
M
109
86
72
64
92
11
M
72
108
77
76
106
12
M
89
90
78
67
99
13
M
112
90
76
76
112
14
M
112
91
73
70
118
'Clinical Evaluation of Language Fundamentals-Revised (CELFR); Receptive Language Score (RLS), and Expressive Language
Score (ELS) expressed as standard scores.
bTest of Nonverbal Intelligence.
1
2
3
4
5
2
3
4
Stops
Dep
Bift
Caid
Pood
Gud
Tob
Pennish
Batum
Dishuck
Tannod
Gobush
Cubop
Koppefate
Bofudish
Deshondum
Pidocate
Tocumish
Gimaning
Gommecitate
Banifamine
Dopaniful
Puzanium
Tiventiful
Conishament
NonStops
Mave
Noke
Zipe
Sep
Shup
Hin
Maudim
Nanpeed
Zupud
Sipish
Shudep
Hampent
Mikidish
Nitandum
Zaydiful
Sekiding
Shaculting
Hipovent
Misonokich
Nupanobic
Zopanishful
Sopeniment
Shudopadate
Haydomiden
NL
2
3
4
5
6
197
198 Journal of Speech and Hearing Research
38
187-199
Appendix C
Experimental Sentences
Double Marking
1. Point to the
1. Point to the
2. Point to the
2. Point to the
3. Point to the
3. Point to the
4. Point to the
4. Point to the
5. Point to the
5. Point to the
of Number
picture of the hat.
picture of the one hat.
picture of the dogs.
picture of the three dogs.
picture of the big apples.
picture of the two big apples.
picture of the little red balls.
picture of the three little red balls.
picture of the little kitty.
picture of the one little kitty.
Subject Relative Clause Items: Single Embedding
6. The little girl smiling is pushing the boy.
6. The little girl who is smiling is pushing the boy.
7. The boy kicking the girl is tall and skinny.
7. The boy who is kicking the girl is tall and skinny.
8. The little boy falling is kicking the big girl.
8. The little boy that is falling is kicking the big girl.
9. The big clown walking is pulling the little girl.
9. The big clown that is walking is pulling the little girl.
10. The big boy pulling the girl is crying.
10. The big boy who is pulling the girl is crying.
Subject and Object Relative Clause Items: Double Embeddings
11. The girl crying is pushing the boy smiling.
11. The girl who is crying is pushing the boy who is smiling.
12. The boy smiling is kissing the girl laughing.
12. The boy who is smiling is kissing the girl who is laughing.
13. The little boy standing is hitting the little girl sitting.
13. The little boy who is standing is hitting the little girl who is sitting.
14. The girl smiling is kissing the boy hugging the clown.
14. The girl who is smiling is kissing the boy who is hugging the clown.
15. The boy laughing is kissing the clown kicking the girl.
15. The boy who is laughing is kissing the clown who is kicking the girl.
Redundant Adjectival and Adverbial Items
16. The girl chases the horse.
16. The pretty little girl quickly chases the big fast horse.
17. The little boy climbs the tree.
17. The dirty little boy climbs the great big tall green tree.
18. The little car is going to hit the train.
18. The little old blue car is going to hit the big speeding train.
19. The big brown dog is chasing the little cat.
19. The big brown furry dog is quickly chasing the little yellow and black cat.
20. The big pig is running after the little horse.
20. The big dirty pig is running after the little brown and white horse.
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
(NR)
(R)
February 1995
Montgomery: Comprehension and Working Memory
199
Appendix D
Appendix E
Number of nonsense words repeated correctly by word length
for Individual subjects with NL and SLI
Number of correct responses by sentence-type condition
(nonredundant, redundant) by each subject with NL and SLI
1-Syllable
(n = 12)
2-Syllable
(n = 12)
3-Syllable
(n = 12)
4-Syllable
(n = 12)
Subject
NL
1
2
3
4
5
6
12
11
11
12
12
11
12
11
11
10
10
11
8
11
10
9
9
9
7
8
8
12
10
7
8
9
12
12
10
10
9
10
11
12
13
10
9
11
12
12
11
11
8
8
11
11
11
9
11
11
10
10
8
9
8
10
8
8
8
SLI
1
2
3
4
5
6
7
8
9
10
11
12
13
14
10
12
11
11
11
12
10
12
11
12
10
12
11
9
11
9
9
10
8
10
10
11
10
12
8
11
12
12
5
7
9
11
4
7
5
10
4
6
5
5
4
8
5
4
8
7
2
5
2
8
2
6
3
3
3
4
NL
1
2
3
4
5
6
7
8
9
10
11
12
13
SLI
1b
2 a,b
3 a,b
4 a,b
5 a,b
6 a,b
7 a,b
8a
9 a,b
10 a,b
11 a,b
12 a,b
13 a,b
14 a,b
Nonredundant
Sentences
Redundant
Sentences
(n = 20)
(n = 20)
18
20
17
19
19
18
18
17
16
15
16
18
19
18
20
20
19
19
18
17
18
16
16
15
18
18
17
17
17
16
17
14
18
15
14
12
19
17
18
15
16
14
15
12
16
10
16
11
15
14
17
15
19
16
Note. a = subject with SLI performing more poorly on the redundant
sentences than nonredundant sentences relative to him/herself. b =
subject with SLI performing more poorly on the redundant sentences
relative to his/her NL counterpart.
Sentence Comprehension in Children With Specific Language Impairment: The
Role of Phonological Working Memory
James W. Montgomery
J Speech Hear Res 1995;38;187-199
This article has been cited by 3 HighWire-hosted article(s) which you can access for
free at:
http://jslhr.asha.org/cgi/content/abstract/38/1/187#otherarticles
This information is current as of February 12, 2013
This article, along with updated information and services, is
located on the World Wide Web at:
http://jslhr.asha.org/cgi/content/abstract/38/1/187