of Developmental MORRIS
University of Toronto Nineteen children with developmental language impairments were given tests which measured syntactic and phonological skills in both comprehension and production conditions, semantic ability, syllable sequencing, and digit span. The results of discriminant function analysis show that the children could be divided into two groups. Group 1, expressive, was characterized primarily by deficits in the production of syntax and phonology. Group 2, expressive-receptive, was more impaired on measures of phonological discrimination, digit span, and semantic ability in addition to showing global syntactic deficits. An auditory-perceptual basis for the language impairments was not supported.
Two principal approaches have been taken to the study of languageimpaired (LI) children. They have been compared with normal children; implicit in this approach is that the LI group is homogeneous and that a single mechanism can account for the language disorder. But this may not be so. Another approach distinguishes subgroups of developmental language impairment. Researchers who have compared LI children with normal children have differed in their sampling criteria. Tallal (1973, 1974, 1975, 1976) selected children who were impaired in both production and comprehension of syntax. She found these children inferior to normal controls in the perception of brief speech sounds (1973) and in discriminating transitional auditory information (1975). She proposed an auditory-perceptual deficit as the basis for the language disorder. Morehead and Ingram (1973) and Leonard, Bolder, and Miller (1976) compared LI with normal children at similar stages of linguistic development as determined by mean length of utterance. The production of language was the principal criterion for selection. On the basis of their results, these authors proposed that the underlying conceptual development of LI children is also delayed. The authors thank N. Axelrod, B. Drewry, S. Henderson, M. A. Ladanyi, Salloum, J. Santa-Barbara, G. Wallach, J. Weber, and W. Wescottfor providing who participated in the study. Reprint requests should be addressed to Morris Erindale College, University of Toronto, Mississauga, Ontario, Canada, LSL 152 0093-934x/80/030152-20$02.00/0 [email protected]
~t @I MUI by Academic Press, Inc. AU rights of reproduction in any form reserved.
S. Page, S. the children Moscovitch, lC6.
Thus perceptual dysfunction and conceptual delay have both been proposed to account for developmental language impairment. However, the samples may have been drawn from different populations. In Tallal’s sample, comprehension and production of language were inadequate whereas in the other two studies only production difficulties were described. Ajuriaguerra (1966) distinguished two subgroups, on the basis of clinical experience only: a group with poor comprehension and production of language and a group with better comprehension than production abilities. Aram and Nation (1975) found individual variation along several dimensions in a sample of LI children. They administered speech and language tests to LI children and factor analysis of the data yielded six factors. These differed primarily in whether comprehension or production was impaired, in severity of the impairment, and in the aspect (i.e., phonology and syntax, or semantics) of language which was most disturbed. Although LI children are generally regarded as language-impaired because they have delayed acquisition of syntax production, their syntax comprehension and semantic ability are also relevant in making this diagnosis. The main purpose of this study was to identify subgroups of LI children. More specifically, it was asked whether a group of children who demonstrate deficits in syntax and/or semantic ability could be divided into (A) one group with syntax production deficits and another group with syntax production and comprehension deficits, or (B) a group which is semantically normal and another group which is semantically impaired. Second, we intended to investigate differences in their expressive phonology which could be related to different types of LI; a third aim was to determine if auditory-perceptual deficits could account for the language impairment. METHOD Subjects The subjects were 20 children between the ages of 4.3 and 7.5 years with a mean age of 5.7 years. They had been diagnosed by a speech pathologist as having specific language disorders. The children were tested in the schools or institutions they were attending. Audiometric screening indicated that their acuity was within the normal range; they were free of oral abnormalities sufficient to interfere with speech. General mental deficiency was ruled out in all the children: 16 out of the 20 subjects were within the normal range on the Block Design subtest (of the Wechsler Intelligence Scale for Children (WISC) or the Wechsler Preschool and Primary Scale of Intelligence); this subtest was administered by the experimenter. Their scaled scores on this subtest ranged from 8 to 15, with a mean of 10.8. Two of the remaining children had been tested on the Leiter International Performance Scales and achieved IQ scores of 92 and 98. The other two children had been tested on the Stanford-Binet Intelligence Scale and achieved IQ scores of 90 and 102.
Procedure Perceptual-cognitive and linguistic tasks were chosen for administration to the LI children. These tests consisted of measures of syntactic comprehension and production, semantic ability, digit span, syllable sequencing, and phonological discrimination and production.
A phonological test was developed to enable comparison of the ability to produce and discriminate phonological features. Six features were selected: voicing, place of articulation, consonant cluster, syllable arrest, diphthongization, and syllable sequencing (in real words). Impaired sequencing of nonsense syllables has typically been related to articulatory disorders. Many LI children have difficulty sequencing consonant blends and multisyllabic words. Therefore it was of interest to determine whether similar difficulty in the sequencing of nonmeaningful material would be associated with language impairment. Sequencing of one-, two-, and three-syllable units was measured as there were no a priori guidelines for selecting an appropriate level of difficulty.
Part A: Procedure for Standardized Tasks The Northwestern Syntax Screening Test (NSST) was used to obtain measures of syntax comprehension and production. The Auditory Association and Auditory Sequential subtests of the Illinois Test of Psycholinguistic Abilities (ITPA) were selected to evaluate semantic ability and digit span, respectively. The Auditory Association subtest was selected to assess the competence of the subjects to deal with the meanings of words, i.e., the semantic aspects of words, for the following reasons: The ability to detect relationships among verbal items reflects (a) abstract reasoning and (b) competence in using words in terms of the meanings they convey rather than in terms of their syntactic functions. The Similarities subtest of the WISC and the Auditory Association subtest of the ITPA both measure this ability. Since the subjects were of normal intelligence, it was assumed that the reasoning component of the tasks would not be the source of diiculty for subjects with low scores; semantic difficulties would be indicated. The Auditory Association subtest was preferred over Similarities because it is more complex, and hence more likely to discriminate between semantically normal and semantically impaired subjects.
Part B: Procedure for Syllable-Sequencing and Phonological Tasks Sequencing of nonsense syllables was assessed by measuring the number of one-, two-, and three-syllable units a child could repeat in IO sec. This interval was chosen to allow sufficient time to obtain a valid measure. Items repeated were /PA/, ltnl, /p&4, and /P&&A/. An average score for /PA/ and /tn/ repetition was calculated for each child. This average score represented one syllable-sequencing rate in subsequent analyses. The number of repetitions of /pr\tA/ represented two syllable-sequencing rate and the number of repetitions of /pat,&%/ represented three syllable-sequencing rate. The ability of the children to discriminate and produce the following phonological features in words was tested: voicing, place of articulation, consonant cluster, syllable arrest, diphthongization, and syllable-sequencing. The production condition was given first in all cases to ensure, that subjects were familiar with the lexical items to be used during the discrimination condition. Naming errors were corrected during the production task. This was desirable because the discrimination condition was intended to investigate whether subjects could distinguish lexical items on the basis of phonological differences and not their knowledge of the lexicon. The stimuli were pictures consisting of colored drawings mounted on 3 x 5-in. cards. The pictures depicted objects or events which were intended to be familiar even to the youngest subjects. The stimuli were organized in six sets corresponding to each of the phonological features. In the production condition, the subjects were asked to identify the pictures. When necessary, a sentence completion format was used to elicit responses. If achild was not able to name a picture, an imitative response was obtained. Following the production condition, the six discrimination subtests were administered in random order. Taped word stimuli recorded in the experimenter’s voice were presented concurrently with corresponding picture stimuli.
For three of the subtests, the subjects were required to select one item from a pair of phonologically contrastive items in a discrimination format. Two pairs of picture stimuli depicting two sets of contrasts were mounted on 8 x 1l-in. sheets; that is, four pictures were presented together and the child had to select one of these pictures in response to a word stimulus. Two rather than one pair of contrastive pairs were presented together in order to reduce the probability of selecting the correct picture by chance from 50 to 25%. Subtest 1 contained 10 pairs of voicing contrasts, subtest 2 contained 8 pairs of place contrasts, and subtest 3 contained 8 pairs of consonant cluster contrasts (Appendix A). One word from each contrastive pair was selected at random for initial presentation and the presentation of pairs was counterbalanced. The pictures were placed in front of the child and he was asked to point to the picture which showed the word he heard on tape. For the remaining three subtests, the targets were individual pictures. A taped word stimulus was presented along with each of the pictures and the subjects were required to judge whether they sounded right or wrong. In these subtests, half of the items were selected at random to be recorded with phonemes correctly produced and the other half with incorrect production. In subtest 4, syllable arrest, there were 20 CVC or VC words recorded correctly or with the final consonant omitted. In subtest 5, diphthongization, 20 CVC, VC, or CV words were recorded correctly or with the vowel unit reduced. In subtest 6, syliablesequencing, a set of five multisyllabic words was presented four times through, giving a total of 20 words in which 10 were produced with syllables in incorrect sequence. For the right/wrong discrimination tasks, training trials using nonexperimental stimulus pictures were given; each of the phonological contrasts was demonstrated. The child was given feedback about the correctness of his response during training but not during actual testing.
General Procedure Each child was individually tested on all tasks over one to three sessions. Sessions were terminated if subjects were no longer attending or if there was any indication of fatigue. The child was told that he and the experimenter would look at some pictures and talk about them. Pegs or pieces of LEG0 which the child could put in a pegboard or later assemble were given for each response on the tasks in which subjects appeared disinterested. Brief play intervals were introduced as needed between tasks to younger subjects. Responses from the production portion were recorded either on a Sony TR 96 or a Bell and Howell Educator Cassette Recorder. Transcription and scoring were carried out by the experimenter. Responses were designated correct or incorrect only on the basis of whether the target features of phonemes were produced. Other expressive errors were transcribed but not scored as incorrect responses. During the discrimination tasks, the responses were immediately scored as correct or incorrect.
RESULTS Subgroups of Language Impairment One child was omitted because he was not able to follow the instructions. Of the remaining 19 children, 18 were more than 2 SD below the normative mean on the expressive syntax measure and 8 of these were more than 2 SD below the mean on the receptive syntax measure as well. The remaining child was within the normal range in syntactic ability but below age level on the semantic measure (Auditory Association); he was not included in the first analysis since he could not be classified on the basis of expressive or receptive syntax impairment. Seven out of the
nineteen subjects were more than 1 SD below the mean on the semantic test. Appendix B shows the individual data for the entire group of 20 subjects. In order to determine if the sample of LI children could be divided more readily into subgroups differing in syntactic ability or semantic ability, two discriminant function analyses were carried out. The following variables were selected for use in stepwise discriminant function analysis (Dixon, 1967; Nie, Hull, Jenkins, Steinbrenner, & Brent, 1975) because they represent a range of linguistic and perceptual-cognitive abilities considered to be relevant to the proposed groupings. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
Age Syntax production Syntax comprehension Digit span Voicing production Place production Consonant cluster production Syllable arrest production Diphthong production Production of three-syllable words
11. Voicing discrimination 12. Place discrimination 13. Consonant cluster discrimination 14. Syllable arrest discrimination 15. Diphthong discrimination 16. Discrimination of syllable sequence in three-syllable words
The analysis selects out variables contributing to a function which best discriminates between designated groups. Analysis 1: Expressive vs. expressive-receptive syntax impairment. In this analysis, the 18 subjects with syntactic impairment were divided into two groups. Group 1 consisted of 10 children who were below 2 SD on only the expressive portion of the syntax test and group 2 consisted of 8 children who were below 2 SD on both the expressive and receptive portions. Variables 2 and 3 were omitted from the analysis since they were used as criteria for grouping. The discriminant function analysis is a method for determining which combination of variables shows large differences in group means. Whether these means are significantly different can be determined by means of an F test. In this analysis, the approximate final F value demonstrated that the groups were significantly different (Table 1). Standard discriminant function coefficients represent the relative contributions of variables selected by the program to a function which discriminates between the groups. The program selected semantic ability and consonant cluster discrimination, indicating that children with expressive impairments can best be differentiated from children with expressivereceptive impairments by these two variables in combination. However, both groups were performing at ceiling levels on cluster discrimination
Variables Semantic ability Cluster discrimination Approximate final F
- .95” .68” 11.06, & = 2, 15,p < .ool
Discriminant function Canonical correlation U-Statistic X2
37 40 13.59, df = 2, p < .ool
Jacknifed classification Expressives Expressive-receptives Total
80% 88% 83%
discriminant function coefficients.
(the mean scores were 90 and 94% correct for the expressives and expressive-receptives, respectively), and this was the only discrimination subtest in which the expressives scored lower than the expressivereceptives. Therefore this result is probably an artifact. The contribution of semantic ability was high and considerably greater than that of cluster discrimination. The function could therefore be labelled “semantic ability.” The U-statistic is a measure of the discriminating power of the function. It is inversely related to the magnitude of the strength of the relationship. The smaller the value of the U-statistic, the greater the discriminating power of the function; .40 represents a moderately high ability to differentiate the two groups and was significant by x2. The program classified each subject as expressive or expressivereceptive on the basis of the function. Eight out of the ten expressives (80%) and seven out of the eight expressive-receptives (88%) were classified correctly. In total, 15 out of the 18 subjects (83%) were correctly classified. Therefore, the function was fairly effective in classifying LI children as expressive or expressive-receptive. The means for the two subgroups are shown in Table 2. The expressive-receptive group, although 0.8 year (10 months) older than the expressive group, had lower mean scores on both the digit span and semantic subtests. The scaled score representing the normative mean on these tests is 36 and 1 SD cutoffs are at 26. The expressive-receptive group, therefore, falls at the borderline of 1 SD on digit span and considerably lower on the semantic subtest. They had lower mean scores than the expressives on five out of the six discrimination tasks, as well. However, on the phonology production tasks they were better on three out of the six. In other words, the expressive-receptive group tended to have more difficulty with most of the tasks except for the phonology
AND KINSBOURNE 2
Age (years) Digit span (scaled score) Semantic ability (scaled score) Correct production (%) of Voicing Place Consonant cluster Syllable arrest Diphthongs Three-syllable words Correct discrimination (%) of Voicing Place Consonant cluster Syllable arrest Diphthongs Syllable sequence in three-syllable words
5.3 34.2 35.6
6.1 29.8 23.8
79.5 87 58.9 82.5 88.5 63
77.5 86.1 75.5 67.5 93.1 67.5
88 87.8 90.8 90 91
83.1 83.6 94.8 82.5 82.5
a A scaled score of 36 represents the normative mean; 1 SD is + 6.
tasks, which the expressive
group found relatively
Analysis 2: Normal semantic vs. impaired semantic ability. The 19 subjects were grouped into 11 who scored within 1 SD of the mean on the semantic subtest and 8 who scored more than 1 SD below the mean. Variable 5 was omitted from the analysis because it was used to group the subjects. The F value indicates that the two groups were significantly different (Table 3). The U-statistic was significant by x2; therefore the discriminant function was able to differentiate the two groups. The standardized discriminant function coefficients show that the function is most readily characterized as a “syntax comprehension ability” since this variable made the largest contribution. The variables which provide further discriminating power to the function are the ability to produce three-syllable words, age, and digit span; the semantically normal children tended to be younger and had more difficulty producing three-syllable words; semantically impaired children were worse at syntax comprehension and had lower digit spans. These four variables are the best combination for discriminating between the two groups. The function correctly classified 7 out of the 8 impaired children (88%) and 8 out of the 11 normal children (73%). Total correct classification was 15 out of 19 children (7%), indicating that in combination, the variables
Variables Digit span Three-syllable word production Syntax comprehension Age Approximate final F
+.39a -.52 +.70 -so 4.19, df = 4, 14, p < .02
Discriminant function Canonical correlation U-statistic x2
.74 .46 11.81, df = 4, p < .019
Jacknifed classification Normal semantic Impaired semantic Total
73% 88% 7%
discriminant function coefficient.
were moderately effective in classifying the LI children as semantically normal or semantically impaired. The means for the two groups are shown in Table 4. Although the semantically impaired group was 0.6 year (7 months) older than the semantically normal group, the impaired group performed less well on syntax comprehension, syntax production, digit span, and the six discrimination subtests; the normal group had lower scores on the six phonology production tasks. These results indicate that the children with semantic difficulty had more general linguistic deficits and the children who were semantically normal had more restricted linguistic deficits. Comparison of analyzer 1 and 2. The function discriminating the expressive and expressive-receptive syntax groups was best characterized as “semantic ability” and the function discriminating the semantically normal and impaired groups as “syntax comprehension ability.” Consequently, the correspondence between the classification of subjects by the two functions was assessed. Table 5 shows the results of the crosscomparison. Seven children in the expressive syntax group were also in the semantically normal group; two were in the semantically deviant group. Eight children in the expressive-receptive group were in the semantically deviant group and one was in the semantically normal group. Only 3 out of the 18 children were classified differently by the functions. This indicates that there is considerable overlap in the classification of the children by the two discriminant functions.
TABLE 4 MEAN SCORESFOR SEMANTICALLY NORMAL AND IMPAIRED GROUPS Normal
AgeWears) Syntax comprehension (percentage correct) Syntax production (percentage correct) Digit span (scaled score) Correct production (%) of Voicing Place Consonant cluster Syllable arrest Diphthongs Three-syllable words Correct discrimination (%) of Voicing Place Consonant cluster Syllable arrest Diphthongs Syllable sequence in three-syllable words
77.3 84.3 62.6 76.8 90 60
82.5 91.5 75.5 76.9 92.5 76.3
89.1 88.3 93.9 89.1 90.5
83.1 84.5 91.6 85 84.4
a A scaled score of 36 represents the normative mean; 1 SD is 26.
Discrimination and production of phonological features. In order to determine whether auditory-perceptual abilities were related to the ability to produce phonological features, the score from each phonological discrimination subtest was correlated with its production counterpart, partialling out the effect of age. Of the six correlations, only syllable arrest was significant (r = S3, p < .02). This indicates that the ability of the children to decide whether a closing consonant in a CVC or VC word is present or absent is correlated with the ability to produce closing consonants. TABLE 5 CORRJZSPONDENCE BETWEEN CLASSIFICATIONS 1 AND 2 (1) Syntactic deficits Expressive (2) Semantically normal Semantically impaired
Expressive-receptive 1 8
The mean scores (Appendix B) on the phonological discrimination subtests are generally high, indicating that the children did not find the discriminations difhcult. The errors frequently seemed attributable to lapses in attention or uncertainty about picture labels. The latter was demonstrated when the children selected pictures (in the four picture sets) which were not contrastive with the target responses but were associated: e.g., target word: bear; stimulus pictures: bear, pear, goat, boat; response: goat. Given the lack of correspondence between discrimination and production and the relatively high scores on discrimination, it is unlikely that difficulty with discrimination is the source of difficulty in production. Syntax comprehension correlated with semantic ability and phonology.
Partial correlations controlling for age were carried out between syntax comprehension and each of the phonological measures, as well as the semantic test (Table 6). Mean phonology scores were obtained by averaging the phonology production scores and the phonology discrimination scores for each of the 19 subjects; these scores were included in the analysis. Syntax comprehension was significantly correlated with semantic ability , corroborating the results of the discriminant function analyses. It was significantly correlated with four out of the six discrimination tasks (voicing, place, syllable arrest, and syllable sequences) and with the mean discrimination score as well. It was correlated with only one production task, syllable arrest. Therefore, syntax comprehension is primarily TABLE CORRELATIONSOF
Syntax comprehension Discrimination of Voicing Place Consonant cluster Syllable arrest Diphthongs Syllable sequence in three-syllable words
Syntax comprehension .60** .a* - .02 .53* .53*
.32 -42 -.Ol .62** -.34 .08
Mean phonology discrimination score
* p < .05. ** p < .Ol. *** p < .OOl.
Production of Voicing Place Consonant cluster Syllable arrest Diphthongs Three-syllable words Mean phonology production score
related to measures of phonological discrimination and to semantic ability. Syntax production
correlated with semantic ability and phonology.
Partial correlations between syntax production and the phonological measures resulted in significant correlations with one discrimination task, diphthong, and one production task, consonant cluster. Each of the two mean scores was also significantly correlated with syntax production (Table 7). The mean diphthong production score (91% correct) approaches ceiling. The lower score on the discrimination subtest (87% correct) indicates that the conceptual rather than the perceptual demands of the task were difficult for the children. It may therefore be inferred that the correlation between syntax production and diphthong discrimination is spurious. On the whole, a weak, and possibly artifactual, association between syntax production and the discrimination tasks was found. A somewhat stronger association was obtained between syntax production and phonological production. The correlation between syntax production and semantic ability was nonsignificant. Rapid repetition of nonsense syllables correlated with syntax and phonology. Partial correlations controlling for age demonstrated relation-
ships between the ability to repeat two-syllable sequences (path) and syntax production, five out of the six phonological production tasks, and the mean phonology production score (Table 8). In contrast with this, none of the sequencing tasks were related to measures of syntax comTABLE
.40 .39 -.09 .38 .47*
Production of Voicing Place Consonant cluster Syllable arrest Diphthongs Three-syllable words
.35 .42 .58** .41 .32 .44
Mean phonology discrimination score
* P < .05. ** p -=c.Ol. *** p < .ool.
Syntax production Discrimination of Voicing Place Consonant cluster Syllable arrest Diphthongs Syllable sequence in three-syllable words
Mean phonology discrimination score
CORRELATIONS OF ONE, Two, AND THREE NONSENSE SYLLABLE REPETITION WITHSYNTAXPRODUCTIONANDPHONOLOGICALPRODUCTIONTASKS Mean of lpAl<d one syllable Syntax
Production of Voicing Place Consonant cluster Syllable arrest Diphthongs Three syllable words Mean phonology production score
/patAl two syllable
IpAtakd three syllable
.33 .59** .45 .32 .33 .42
.52* .56* .58** .49* .24 .49*
.54* .22 .ll .38 .09 -.Ol
* p < .05. ** p < .Ol. *** p < ml.
prehension or any of the phonological discrimination tasks. Thus, rapid syllable-sequencing appears to be a skill related to the production aspects of language but not to the comprehension and discrimination aspects. Commenrs. Four out of the twenty children could not be trained to make right/wrong decisions (subjects 3, 4, 12, and 18 in Appendix B); these children also tended to be more severely impaired than the rest on several other tests. Several children were distractible and impulsive. Long latencies in naming and associated naming responses were indicative of word-finding difficulties. DISCUSSION
The results show that language-impaired children are not a homogeneous population. The first analysis defined one group, the expressives, who had difficulty with syntax production but normal comprehension of syntax, and another group, the expressive-receptives, who had difficulty with both the comprehension and production of syntax. The expressive group had normal semantic ability, normal digit spans, and were better than the expressive-receptives on the phonological discrimination tasks. They did, however, have more difficulty than the expressive-receptives in producing consonant clusters in words, words containing three syllables, and diphthongs in words. This suggests that children with syntax production deficits also have difliculty with the productive aspects of phonology.
Although on the average 10 months older, the expressive-receptives were inferior to the expressives in semantic ability and phonological discrimination; they also had lower digit span scores. This indicates that their deficits are more general than those of the expressives. On the other hand, their ability to produce phonological features was as good or better than the expressive group’s. The second analysis yielded two groups differing on a semantic dimension. A semantically impaired group showed more general deficits than a semantically normal group with the exception that the impaired group performed better on the phonological production tasks. The semantically normal group, although 7 months younger, were superior to the impaired group on expressive and receptive syntax, digit span, and the phonological discrimination tasks. The division of the sample into those with only impaired syntax production and those with impaired comprehension as well was more effective than when the sample was divided along a semantic dimension. The classification of children was better in the syntactic grouping and the function discriminating the two groups was more powerful in the syntactic analysis. However, cross-comparison of the classifications of the children showed a high degree of correspondence between the two groupings. Furthermore, each function was best described by the grouping dimension of the other, that is, the children placed into syntactic groups were best discriminated by afunction to which semantic ability contributed the most, and the children placed in semantic groups were best discriminated by a function to which syntax comprehension contributed the most. It can be concluded that the semantically normal group represented the same set of children as the expressive group, and the semantically impaired group represented the same set of children as the expressive-receptive group but the two types of children are most readily distinguished on the basis of their syntactic abilities. Combining the results of the two analyses, the children could be described as follows: (1) a group with syntax production difficulty and (2) a group with deficits in both syntax production and comprehension, as well as impaired semantic ability. Children who are impaired in the production and not the comprehension of syntax perform normally on tests of semantic ability and digit span. They have relatively more difIiculty with phonology production than the second group. Children who are impaired in the production and comprehension of syntax, in addition to the semantic deficit, also have reduced digit spans and difRculty with phonological discrimination tasks. They tend to be relatively less impaired in the production of phonemes and words than in other linguistic and conceptual tasks but this may be a function of their ages; given that they were older, it is possible that phonological skills improve relatively faster with age than other linguistic abilities.
These findings lend quantitative support to the delineation of two subgroups, expressive and receptive, such as Ingram (1969) has described clinically. They are also consistent with the findings of Aram and Nation (1975). The two subgroups map onto patterns described in their study. The results of the study do not support an auditory-perceptual basis for developmental language impairments. The ability to discriminate phonological features was not correlated with the ability to produce the same features, apart from syllable arrest. Generally, performance on the discrimination tasks was good. The positive correlation obtained on the syllable arrest subtests could be considered evidence for a perceptual-deficit hypothesis. However, the subject data do not support such an interpretation. For the 12 out of the 19 children who scored between 85 and 100% correct (Table 9) on the production task, there is no need to consider a perceptual deficit. Since they were producing closing consonants consistently, it can be inferred that they could discriminate them as well. Of the remaining children who scored below 85% correct on production, two were able to carry out the discrimination task very well; subjects 7 and 8 scored 45 and 75% correct on the production task yet both scored 95% correct on the discrimination task. Clearly, a perceptual deficit could not account for their production difficulties. Of the remaining five children, three (subjects 3, 4, and 18) could not make right-wrong discriminations, even on features they were able to produce. Therefore, their low scores on syllable arrest discrimination do not necessarily reflect perceptual deficits. They seem to have been limited, instead, by the conceptual demands of the task. The discrimination scores of the remaining two children (subjects 1 and 6) were in line with their production scores but it is doubtful that their data alone could have resulted in the positive correlation between the discrimination and production counterparts of syllable arrest. Added weight may have been given to the correlation by the scores of subjects 3, 4, and 18. These children had low scores on both production and discrimination but it has been pointed out that their poor performance on the discrimination task was probably due to factors unrelated to discrimination per se. Thus, the present finding cannot be considered compelling evidence that a perceptual deficit underlies the failure of some children to produce closing consonants. We offer two suggestions to account for the association between syntax comprehension and several of the phonological discrimination tasks. First, children who performed poorly on the syntax comprehension test were the ones with more general deficits. Such children have been described as impulsive and distractible (Berry, 1969; Johnson & Myklebust, 1967). Perhaps, in addition to the linguistic delay, children in the expressive-receptive subgroup also have immature attentional skills. Their poor performance on tasks requiring them to select one target
of closing consonants.
Note. Scores are percentage correct for each of the 20 children obtained on the production
picture out of a set, as three of the phonological discrimination tasks did, may reflect inconsistent application to the task. For example, subject 10 received scores of 95, 100, and 100% correct on these three production tasks and only 60, 88, and 94% correct on the discrimination counterparts of these tasks. Clearly, this child’s ability to produce phonemes with correct voicing and place of articulation, and words containing consonant clusters was good but his performance on the discrimination subtests was relatively poor. Alternatively, the expressive-receptive children may be delayed in the acquisition of certain task-related skills. The three children who had the most difficulty making right-wrong discriminations (subjects 3,4, and 18) did not have comparable difficulty on the tasks requiring them to select pictures for words. For example, subject 4’s scores on discrimination tasks 1, 2, and 3 (Table 10) show that he was able to select pictures in response to target words with some degree of success but in discrimination tasks 4, 5, and 6, which required right-wrong discriminations, he selected a response, e.g., “right,” and applied it to the entire task; since half the target words were correct productions, he received a score of 50% but it was evident that he did not understand what to do. It can be concluded that for some children in the expressive-receptive group, the conceptual demands of the discrimination task caused them difficulty and other expressive-receptive children seemed to be unable to attend to the relevant phonological distinctions. The rates at which one, two, and three nonsense syllable units (PA, tA, pAtA, pAtAkA) were repeated were exclusively related to the production of language and speech. There were no significant correlations between syllable repetition rate and either syntax comprehension or phonological discrimination. The mean number of three-syllable repetitions (phtAk,Q produced by this sample of children was considerably lower than that obtained from a sample of normal children in the same age range (Prins, 1%2). Sequencing was a factor in real word production as well; the children had the most difiiculty with the production of three-syllable words and consonant clusters, both of which entail sequencing. It is, TABLE 10 SULBJECT~'SSCORESONTHEPHONOLOGICALTASKS
Production (% correct) Discrimination (% correct) ’ l-voicing, bk words.
2-place, 3-consonant cluster, Csyllable
however, unclear whether the sequential disorder is causally related to the production difficulties. The present findings may have implications for the assessment and treatment of childhood language disorders. With respect to assessment, unless both comprehension and production are adequately tested, diagnostic procedures cannot be considered complete. Results of such testing would indicate whether treatment should focus on the comprehension or production of language; expressive types would require more training in the production of language while receptive types would require both comprehension and production training. Given the relative independence of the ability to decode and encode the same phonological material, therapists cannot expect instruction in discrimination to facilitate the acquisition of the syntactic or semantic aspects of language. For instance, although some children should be taught to understand meanings associated with different syntactic structures and to organize verbal material, this goal could not be achieved by training in phoneme discrimination. Nor would discrimination training facilitate phonology production. The results also have theoretical implications. The ability to apply appropriate rules for extracting meaning from verbal input does not guarantee the ability to use these rules in production. Presumably the brain bases for these two applications of given rules are different. The association of lower digit span scores with poor performance on the syntax comprehension and semantic measures supports the notion that digit span is not a measure of information processing capacity but rather reflects the individual’s ability to organize verbal information (Olson, 1973). In future investigations of developmental language impairment, subjects should be documented so as to clarify whether one or both subtypes described in this study have been included. SUMMARY
Children with developmental language impairments can be classified into (1) a group with dficulties in the production of syntax and phonology but not in the comprehension of syntax or in semantic ability, and (2) a group characterized by a global syntactic deficit and semantic impairments who perform more poorly on a variety of perceptual and linguistic tasks. An auditory-perceptual deficit was not supported as a causal factor in developmental language impairment.
APPENDIX Picture (1) Voicing
boat pig cat cup frog bed leg hat truck em mop cake crib feet book dog flag tub bread skate
Tasks (3) Cluster
boat/goat back/bat pear/tear soup/suit pail/tail pan/can ball/doll glue/blue
+ + + + + + + + + +
(2) Place of articulation
pear/bear ice/eyes doll/tall lock/log coat/goat cab/cap tear/deer pig/pick bee/pea sing/sink
(4) Syllable arrest
(5) Diphthongization + + + + + + + + + + +
Note. + indicates correct production production of taped stimulus word.
rope water whale bike boys house light cage cloud bow pie onion cow kite knife whistle snake boy cake swing
top/stop key/ski tick/stick wing/swing nail/snail pot/spot pill/spill talking/stocking
Tasks (6) Syllable sequencing + + + + + + + + + + + -
elephant elevator animal helicopter telephone elephant elevator animal helicopter telephone elephant elevator animal helicopter telephone elephant elevator animal helicopter telephone
of taped stimulus word. - indicates incorrect
B: SUBJECT DATA
Subject 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
4.3 4.4 4.4 4.9 4.9 5.2 5.3 5.3 5.4 5.7 5.7 6.1 6.2 6.2 6.3 6.5 6.7 6.8 7.1 7.5
13* 18* 0* 0* 35* 0* 8* 23* 0* 13* 50* 25* 30* 0* 33* 55* 60 0* 52* 50*
50 70 43* 22* 80 75 60 73 88 48* 70 70 88 55* 53* 88 80 55* 70* 85*
63 77 52 51 80 70 60 71 84 % 95 91 89 78 93 89 98 66 93 95
80 94 71 65 91 80 92 94 89 83 82 69 95 90 93 90 100 75 97 99
30 41 29 36 37 33 25 39 29 23 31 28 34 27 43 43 36 25 29 26
38 44 31 19 38 34 40 42 30 14 17 26 42 23 30 33 25 8 29 36
Note. The Northwestern Syntax Screening Test was used to obtain measures of syntax production and comprehension: (*) indicates that score is below second SD. Data from subject 12 were omitted from all analyses. Data from subject 17 were omitted from the analysis of syntactic subgroups. Digit span was measured by the Auditory Sequential and semantic ability by the Auditory Association subtests of the Illinois Test of Psycholinguistic Abilities. A scaled score of 36 represents the normative mean; 1 SD is & 6.
REFERENCES Aram, D. M., & Nation, J. E. 1975. Patterns of language behavior in children with developmental language disorders. Journal of Speech and Hearing Research, 18, 229-241. Arthur, G. 1952. The Arthur adaptation of the Leiter Znternational Performance Scale. Washington, D. C.: Psychological Service Center Press. Berry, M. F. 1969. Language disorders in children: The basis and diagnosis. New York: Appleton-Century-Crofts. de Ajurtaguerra, J. 1966. Speech disorders in childhood. In C. Carterette (Ed.), Brain function: Speech, language, and communication. Los Angeles, California: University of California Press. Vol. 3. Dixon, N. J. 1967. BMD: Biomedical computer programs. Berkeley/Los Angeles: Univ. of California Press. 2nd ed.
Ingram, T. S. 1969. Developmental disorders of speech. In P. J. Vinken & G. W. Bruyn (Eds.), Handbook of clinical neurology. Amsterdam: North-Holland. Vol. 4. Johnson, D. J., & Myklebust, H. R. 1967. Learning disabilities: Educational principles and practices. New York: Grune & Stratton. Kirk, S. A., McCarthy, J. J., & Kirk, W. D. 1974. The Illinois Test of Psycholinguistic Abilities. Urbana: Univ. of Illinois Press. Lee, L. L. 1969. Northwestern Syntax Screening Test. Evanston, IL: Northwestern Univ. Press. Leonard, L. B., Bolder, J. G., & Miller, J. A. 1976. An examination of the semantic relations reflected in the language usage of normal and language-disordered children. Journal of Speech
Morehead, D. M., & Ingram, D. 1973. The development of base syntax in normal and linguistically deviant children. Journal of Speech and Hearing Research, 16,330-352. Nie, N., Hull, C., Jenkins, J., Steinbrenner, K., & Brent, D. 1975. SPSS: Statistical package for the social sciences. New York: McGraw-Hill. Olson, G. M. 1973. Developmental changes in memory and the acquisition of language. In T. E. Moore (Ed.), Cognitive development and the acquisition of language. New York: Academic Press. Prins, T. D. 1962. Analysis of correlations among various articulatory deviations. Journal of Speech and Hearing Research, 5, 152-160. Shane, S. A. 1973. Generative phonology. Englewood Cliffs, NJ: Prentice-Hall. Tallal, P., & Piercy, M. 1973. Developmental aphasia: Impaired rate of non-verbal processing as a function of sensory modality. Neuropsychologia, 11, 389-398. Tallal, P., & Piercy, P. 1974. Developmental aphasia: Rate of auditory processing and selective impairment of consonant perception. Neuropsychologia, 12, 83-93. Tallal, P., & Piercy, P. 1975. Developmental aphasia: The perception of brief vowels and extended stop consonant. Neuropsychologia, 13, 69-74. Tallal, P., & Stark, R. 1976. Relation between speech perception and speech production impairment in children with developmental dysphasia. Brain and Language, 3, 305317. Wechsler, D. 1949. The Wechsler Intelligence Scale for Children (WZSC). New York: The Psychological Corp. Wechsler, D. 1967. The Wechsler Preschool and Primary Scale of Intelligence. New York: The Psychological Corp.