| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Contact author: Fabrizio Pizzioli, Unité Cognition et Développement, Université catholique de Louvain, Place Cardinal Mercier, 10, 1348 Louvain-la-neuve, Belgium. E-mail: fabrizio.pizzioli{at}uclouvain.be.
 |
Abstract |
|---|
 Top Abstract References |
|---|
Method: The study examined the influence of the argument-structure complexity of a target sentence on the production of grammatical morphemes in French children with SLI compared with younger children matched for grammatical level in production (GL) and children of the same chronological age (CA). A sentence production task was used where the target sentences varied in terms of argument complexity and length.
Results: The results indicated that children with SLI used articles and auxiliaries in obligatory contexts significantly less often than both the GL and CA control groups: More complex argument structures elicited the highest number of grammatical morpheme omissions; this effect was larger in children with SLI than in the GL group and was independent of the length of the sentences, which failed to show any influence on the production of grammatical morphemes.
Conclusion: These results support the hypothesis that grammatical-morpheme deficit in SLI depends at least in part on limited processing capacities.
KEY WORDS: sentence production, argument structure, specific language impairment (SLI)
Children with specific language impairment (SLI) are characterized by a number of deficits in many aspects of language comprehension and/or production, despite preserved nonlinguistic competences: normal nonverbal abilities, no sign of neurological problems, hearing within normal limits, no evidence of a psychopathological development (e.g., Leonard, 1998). Many investigations that considered language abilities in SLI demonstrated a remarkable limitation in grammatical morphology in production.
Bound and free grammatical morphemes seem to be more vulnerable in children with SLI compared with language peers (e.g., Leonard, Eyer, Bedore, & Grela, 1997; Rice & Wexler, 1996). Verb morphology is especially problematic: Along with errors attaining verbal morphology, marking tense and number, the use of the auxiliary verb is particularly vulnerable in children with SLI, who omit the auxiliary in obligatory contexts significantly more often than language peers. These difficulties have been largely demonstrated in English (e.g., Grela & Leonard, 2000; Hadley & Rice, 1996; Rice, Wexler, & Cleave, 1995) as well as in other languages such as Swedish (Hansson, 1997), Croatian (Arapovic & Andel, 2003), and German (Clahsen, 1989, 1991; Kerschensteiner & Huber, 1975), although the difference failed to emerge in Italian (Leonard & Bortolini, 1998). In French children with SLI, spontaneous speech analysis revealed that they omit the auxiliary verb more often than younger children (Methé & Crago, 1996). In particular, Crago and Paradis (2003a, 2003b) and Paradis and Crago (2001) found that French children with SLI frequently omitted the third person present perfect auxiliary verb a (have), whereas they produced the homophonous preposition à (to) most of the time. Hamann and colleagues (Hamann et al., 2003) also showed that in spontaneous production, French children with SLI between the age of 3;10 (years;months) and 5;0 still produce nonadult root infinitives to an extent comparable to what is found in typically developing children in their third year of life.
The use of articles has also been reported to be problematic: English-speaking children with SLI show an inadequate use of articles compared to language peers (e.g., Beastrom & Rice, 1986), and consistent results have been obtained in Italian (Bortolini, Caselli, & Leonard, 1997; Cipriani et al., 1991; Leonard, Bortolini, Caselli, McGregor, & Sabbadini, 1992), in Spanish (Restrepo & Gutierrez-Clellen, 2001), and in Swedish (Hansson, Nettelbladt, & Leonard, 2003; Leonard, Salameh, & Hansson, 2001). In French, the issue is less clear: Some studies analyzing spontaneous speech show that articles are less affected than other grammatical morphemes such as clitics, and in general, article use seems to be not as troublesome as in other languages. For example, LeNormand analyzed speech samples from children with SLI aged between 4 and 6, which showed that these children used articles in obligatory contexts at the same rate as a younger control group of 3- to 6-year-olds (LeNormand, Leonard, & McGregor, 1993). Paradis, Crago, and Genesee (2003) compared French–English bilingual children with SLI with younger, typically developing bilingual peers, as well as with monolingual French children with SLI and monolingual normal-language peers. Results showed that both bilingual and monolingual children with SLI used a higher percentage of definite articles than object clitics in French, which was comparable to younger normal-language developing children. Nonetheless, Elin Thordardottir and Namazi (2007) reported findings that are contrary to these results in that they fail to show evidence of morphosyntactic errors in the speech of French-speaking children with SLI.
Processing or Representations Deficit?
One purpose of the present research is to bring evidence to bear on a central issue concerning the nature of SLI: Do at least some of the problems with grammatical morphemes in SLI depend on processing factors rather than on a representational deficit? This issue is investigated by examining the influence of the argument-structure complexity of a target sentence on the production of grammatical morphemes in children with SLI and typically developing children.
The relationship between knowledge and processing is complex. Formal descriptions of language such as generative grammar are concerned with competences—the knowledge that a speaker/hearer has of a language and that guides the use of the language (Chomsky, 1968)—but leave open the issue of how these competences are used. The rules incorporated in the grammar are not mirrored exactly in the organization of the linguistic mechanisms that can involve processes that are independent. Here, we want to test the hypothesis that at least some production problems with grammatical morphemes can be accounted for by language processing deficit and by how the knowledge is put to use and is not related to a deficit of the linguistic knowledge in the Chomskyian sense. Despite the effort to distinguish representation from processing, ultimately they are probably not distinguishable (e.g., MacDonald & Christiansen, 2002). The hypothesis of a processing deficit and that of a representational deficit in SLI are not mutually exclusive; rather, it is more likely that they coexist and interact during development. Here, we focus on processing.
Argument Structure Complexity
Sentences vary in argument complexity: Intransitive verbs, or one-place argument structures, require only one argument in subject position (x):
Although apparently simple, the lexical information specified by the verb, which organizes the argument structure, is indeed rather complex and is encoded at a number of different levels of linguistic representation: At the semantic level, the thematic grid states which thematic roles the argument(s) can take (e.g., agent, patient, theme), and the selection restriction information constrains some gross semantic features of the argument(s) (e.g., animate/inanimate, edible/inedible, etc.). At the lemma level, information is stored about the number and type of arguments—known as the subcategorization frame—that a verb can take (e.g., Levelt, 1989).1 In sentences 1 through 3, argument-structure complexity is related to sentence length; however, sentences of different argument-structure complexity can be matched for length by adding a so-called adjunct. Contrary to arguments, the adjuncts are not part of the argument structure and are not specified in the main verb argument structure. Consider the following example:
In sentence 4, newspaper is the internal argument of the verb to buy, whereas shop is an adjunct and is not part of the argument-structure specified by the verb. Adjuncts are facultative, whereas arguments are always mandatory; thus, sentence 4 is still grammatically correct if the adjunct is missing (Donald is buying a newspaper) but not if the argument is missing (*Donald is buying in the shop). For verbs such as to eat, the internal argument is often called the facultative argument, with a descriptive value. In formal linguistics, arguments are always mandatory, but certain verbs have more than one argument structure (in this case, to eat has both a transitive and an intransitive argument structure). The main verb assigns thematic and grammatical roles to the internal arguments but not to adjuncts.
In sentence processing, argument-structure complexity has been shown to be one of the crucial factors that determines sentence complexity (e.g., Grela & Leonard, 2000; Rubin, Newhoff, Peach, & Shapiro, 1996; Shapiro, Brookins, Gordon, & Nagel, 1991; Shapiro, Zurif, & Grimshaw, 1987; Tabossi, Collina, Caporali, Pizzioli, & Basso, 2004). Moreover, the effect of argument-structure complexity can originate from two distinct loci. Focusing on the lexical level (lexical access and lexical retrieval), it is frequently assumed, although often implicitly, that multiple bits of information associated with the verb are exhaustively retrieved each time the verb is accessed, as happens in the case of polysemous words (Rubin et al., 1996; Shapiro et al., 1991; Shapiro et al., 1987). Therefore, when a verb is retrieved from the lexicon, a greater number of arguments or multiple possible argument structures and subcategorizationf frames would increase the amount of information to activate and, consequently, the computational load required.2 On the other hand, considering sentence processing, when a more complex argument structure is processed, a greater number of grammatical functions have to be assigned (Bock & Levelt, 1994) and a greater number of thematic roles have to be considered and mapped into syntactic roles, which also increase the computational load.
In order to make specific predictions about the effect of argument-structure complexity, we used the sentence production model put forward by Bock and Levelt (1994). Relevant characteristics of this model are that it specifies two temporally and functionally separate stages: In the first stage, lexical items are selected, and grammatical functions are assigned to them following the argument-structure specified by the verb. Later, at the positional level, grammatical morphemes—which are called for by the syntactic structure associated with the verb—are retrieved from a potentially separate store and are later inserted and integrated within the prosodic structure. Thus, the assignment of argument structure and morphosyntax occur in sequence, with argument structure being dealt with first; it brings to the prediction that because of slow processing at the argument-structure level, morphosyntactic assignment ends up being adversely affected. For instance, processing would take too long, and before grammatical morphemes are retrieved and/or assigned to the correct position, a sentence that misses grammatical morphemes could be spelled out. Remember: We assume that limited resources or slower processing affect grammatical morphemes processing as a consequence of the caracteristic that functional and positional stages are sequential and not parallel, so that grammatical morphemes are processed after thematic and grammatical role assignment. Behaviorally, the rate of omissions of grammatical morphemes in children with SLI is likely to be influenced by argument-structure complexity, with more omissions occurring during attempts to produce transitive rather than intransitive sentences: This effect is expected to be more extensive in children with SLI than in younger children with normal language.
Theoretical Accounts of SLI
Several hypotheses try to account for the difficulties with grammatical morphology shown by children with SLI. They can be roughly placed in two main general schools of thought. According to the first account, children with SLI have developed a poor or deficient representation of grammatical morphology. Implicit grammatical rules deficit (Gopnik, 1990a, 1990b; Gopnik & Crago, 1991), restriction to the range of contexts to which rules are applied (Ingram & Carr, 1994), problems with the acquisition of functional categories (Eyer & Leonard, 1995) and representational deficit of dependent relationship (van der Lely, 1996) are some of the explanations advanced. All these accounts have in common the assumption that problems and weaknesses of morphosyntax in children with SLI originate from incomplete or deficient "knowledge of particular rules, principles, or constraints" (Leonard, 1998, p. 213), which would result in an occasional omission and erroneous use of grammatical morphemes. If the problem with grammatical morphemes depends on an incomplete or deficient knowledge of particular rules, principles, or constraints, the use of grammatical morphemes should be relatively constant and independent from processing factors: Following the representational hypothesis, argument-structure complexity of the target sentence is not expected to influence the number of grammatical morpheme errors and omissions. Within the framework of the representational deficit approach, it has been claimed that children with SLI have trouble with representing dependent relationships such as those in the argument structure (van der Lely, 1994, 1996), so that noncanonical, semantically reversible sentences and sentences that involve complex syntactic operations such as movement are problematic for these children. This hypothesis does not predict that structural complexity has an impact on other subsequent stages of language processing but only that the representation of structural relationship is not correctly or fully specified. It does predict that when the relationship between elements depends on the appreciation of structural relationships in the sentence alone, children with SLI will fail to represent the relationships between elements. Note, however, that in the case of simple structural relationships in semantically irreversible, unambiguous active sentences such as those used in the present experiment, the representation deficit of dependent relationship hypothesis does not even predict difficulty in unfolding the structural relationships between elements. Thus, if in general this hypothesis predicts that structural relationships are not correctly represented, it does not predict that the rate of grammatical morphemes varies following the structural complexity.
The second approach stresses the role of processing and the computational cost associated with language comprehension and production. In this view, children with SLI will dispose of limited processing capacities that interfere with language acquisition and actual language processing (Ellis Weismer, 1994; Ellis Weismer & Hesketh, 1996; Johnston, 1991, 1994).3 In children with SLI, complex linguistic operations may overwhelm the system's capacity, resulting in competition for resources among different stages of language processing and thus generating a computational trade-off that would benefit early stages of language processing and burden later ones. According to this view, the argument-structure complexity influences the number of omissions of grammatical morphemes in children with SLI. If children with SLI dispose of limited processing capacities, the effect of argument-structure complexity is expected to be larger in children with SLI than in younger control children with normal language.
SLI and Argument-Structure Complexity
Psycholinguistic data (e.g., Shapiro et al.,1987, 1991); electrophysiological measures (Rubin et al., 1996); neuropsychological data (Collina, Marangolo, & Tabossi, 2001; Collina, Pizzioli, Caporali, Basso, & Tabossi, 2002; Tabossi, Collina, Caporali, Pizzioli, & Basso, 2005; Thompson, Lange, Schneider, & Shapiro, 1997; Thompson, Shapiro, Li, & Schendel, 1995); and computational modeling simulation (Haarmann, Just, & Carpenter, 1997) support the claim that argument complexity is associated with greater cognitive demand; more complex argument structures would require greater amounts of time and/or engage higher resources consumption. Literature on children with SLI is in line with these results. Roberts, Rescorla, and Borneman (1994) first demonstrated that children with SLI omit more obligatory arguments than children of the same age. In Grela and colleagues (Grela, 2003; Grela & Leonard, 1997), children with SLI were asked to complete a story using a sentence of varying argument complexity. Results indicated that both children with SLI and the language-matched control group omitted the subject argument more often when the sentences had a more complex argument structure. In addition, compared with language-matched peers, more children with SLI omitted the subject argument as linguistic complexity increased.
Grela and Leonard (2000) also investigated the influence of argument-structure complexity on the use of the auxiliary in a group of children with SLI, compared with a mean length of utterance (MLU) control group and a chronological-age control group. The children were asked to complete a story using sentences whose target verb was provided and that varied in argument complexity (intransitive, transitive, and ditransitive sentences) and length; in the long sentences, an indirect optional argument was added. The results demonstrated that (a) children with SLI omitted more "be" auxiliaries than did both the age and MLU controls; (b) argument structure affected the ability to use the "be" auxiliary (the more complex the argument structure, the greater the auxiliary omission); (c) argument complexity also affected the use of other linguistic elements, such as articles, subjects, progressive inflection, and adjuncts; (d) argument complexity had a numerically larger effect in children with SLI than in younger normal language children but nonetheless, the Number of Arguments x Groups interaction was not significant; and (e) phrase length had no effect. These results provide only partial support for the limited processing capacities theory, which predicts that argument-structure complexity has a larger effect on the performance of children with SLI than on that of the MLU control group, which was not observed. Moreover, in this study, it was not possible to exclude the effect of length from argument-structure complexity because the more complex argument structures were systematically longer than simpler ones.
Difficulties with more complex argument structures were also observed in spontaneous speech samples analysis. Elin Thordardottir and Ellis Weismer (2002) analyzed speech samples from a group of 50 school-age children with SLI in comparison with a group of children of the same chronological age. Twenty-five of the 50 children with SLI were also compared to a group of children with the same MLU. They considered the number of omissions of arguments—internal and external—and the repertoire of arguments and argument structures used. They found that although children with SLI did not omit obligatory arguments more often than age-matched peers, they differed from age-matched controls, demonstrating fewer argument types and argument structure types and a less flexible use of verb alternations. "Findings indicate that inordinate difficulty with complex argument structures persists in school-age children [with SLI] and they also indicate that children with SLI and normal language differ significantly in this respect even when utterance length is controlled" (Elin Thordardottir & Ellis Weismer, 2002, p. 245). Nonetheless, from these results, it is still unclear whether children with SLI have a less complete representation of the verb's complexity or a processing limitation that interferes with the production of more complex structures.
As already stated previously, the first aim of the present study was to test whether children with SLI are limited in linguistic processing capacities compared to language-matched peers by looking at the effects of argument complexity on the omission of grammatical morphemes in sentence production with transitive and intransitive verbs. This study also aims to investigate whether this structural complexity is a primary factor, independent of the length of the sentence, following the hypothesis that the relevant cognitive load is determined by the greater number of grammatical and thematic roles to assign in transitive compared with intransitive sentences; it is expected that in sentences that are not awkwardly long, length, per se, should not influence grammatical morpheme use once the structural complexity and syntactic structure are kept constant. Finally, considering that argument-structure complexity possibly has different origins, this study tries to tease apart the lexical level (lexical access and lexical retrieval) from the sentence level (role assignment) as possible origins of the argument-structure complexity effect. Before testing sentence production, single-verb comprehension and production were evaluated. This evaluation showed whether the processing load associated with transitive and intransitive verbs is a relevant factor at the lexical level when the verb has to be produced or understood in isolation, without the sentence context.
Two independent comparisons were planned in order to separate the effect of argument complexity from the effect of length. In the first comparison, sentences of varying argument-structure complexity (intransitive [V1] and transitive [V2]), but of the same length, were used (cf. Sentence 5 vs. Sentence 6) as follows:
In order to have sentences of different argument complexity but comparable length, a so-called adjunct has been added to the intransitive sentence (on the armchair in Sentence 5). The two groups of sentences thus differed in verb type (transitive or intransitive) and internal complement status (adjunct or argument).
The second comparison focused on the effect of length: For each level of argument complexity, sentences of the same argument complexity and variable length were contrasted (cf. Sentence 7 vs. Sentence 8, both of which show an example of a transitive verb).
The long sentences were created by adding a modifier to the two noun phrases (subject and internal complement); this modifier does not alter the syntactic structure of the sentence.
Article and auxiliary verb omissions were adopted as dependent variables. Auxiliary omission was considered because it is common in French, as shown by results in the literature (e.g., Crago & Paradis, 2003a, 2003b; Methé & Crago, 1996; Paradis and Crago, 2001), and it is also demonstrated to be a valid dependent variable to test the effect of structural complexity (Grela & Leonard, 2000). Moreover, we decided to include the use of articles for three main reasons. First, from the theoretical picture presented previously, it follows that the computational load deriving from the processing of more complex argument structures should affect all grammatical morpheme processing and not just the processing of the auxiliary verb. In particular, the article of the postverbal complement—in which the thematic and grammatical role are assigned to the internal argument—could be especially critical, as its position is one in which the cognitive load is particularly high. Second, there is an ongoing debate about the reason why French children with SLI are not particularly bad at article use and whether children with SLI of other languages show an impairment (LeNormand, Leonard, & McGregor, 1993; Paradis et al., 2003). Further evidence can be helpful. Third, considering the auxiliary verb omission as the only dependent variable can be somewhat problematic in French for the following reasons: The typical and more frequent tense that asks for the auxiliary is the passé compose—the present perfect in English (cf. Bassano, Maillochon, Klampfer, & Dressler, 2001)—and this tense is formed by the present form of the auxiliary plus the past participle of the main verb.
Method
Participants
Ten children with SLI and 20 typically developing children were included in the study. Children with SLI were selected from two different special schools for learning disabilities in Wallonie, the French-speaking region of Belgium, and ranged from 8;1 to 13;0 (years;months), M = 9;11, SD = 1;7. Eight children were boys, and 2 were girls. All children had been diagnosed with SLI by clinicians and speech-language therapists and were receiving language intervention in the special school at the time of the study. Selection was based on the speech-language therapists' description of every child, and children with severe primary phonological disorders were not included. All children were given two standardized tests of language, and only the children who scored at least 1.5 SD below average were included in the study. All children who were selected on the basis of the therapists' suggestions were included in the experimental session, and no child was excluded in the course of testing. The two standardized tests of oral language for French were the Evaluation du Langage Oral (ELO; Evaluation of Oral Language; Khomsi, 2001) and the Test des Compétences Verbales et Métalinguistiques (NBTL; Anglade, Ravard, & Ravard, 1993). The ELO test includes subtests for lexical reception, lexical production, sentence comprehension, sentence production, and story narration. The NBTL test includes five subtests: articulation, sentence comprehension, sentence repetition, narrative from a sequence of pictures, and a test of sequential images. All of the children were at least 1.5 SD below the performance of age-matched peers in sentence comprehension (M = 2.4, SD = 1.3) and production (M = 5.1, SD = 2.9). Raw scores obtained in the sentence production subtest of the ELO (M = 11.1, SD = 4.6) were used to match children with SLI with a group of children of the same linguistic level.
Concerning nonlinguistic capacities, all children met the exclusionary criteria for SLI: Only children with SLI who had an IQ above 85 were included, without regard to the gap between their language and intelligence scores (children were tested with the Wechsler Intelligence Scale for Children–Revised [WISC-R; French version; Grégoire, 1992, 1999 ; mean nonverbal IQ, M = 94.7, SD = 7.9, range = 85–108). Children who were included had hearing within normal limits, which was verified by clinicians using pure-tone hearing screening (20dB HL at 500, 1000, 2000, and 4000 Hz), and cases of otitis media with effusion were not reported in any of the children's medical histories. No neurological dysfunctions were signaled in the clinical history of any of the children. None of the children had malformations of the oral structure (e.g., a cleft palate). None of the children presented any sign of psychopathology. Finally, none of the children were bilingual.
A group of 10 children matched for sentence production level (LPM) was selected on the basis of the raw scores for the Sentence Production subtest of the ELO test, M = 11.7, SD = 2.26. These children were younger than the children with SLI and ranged from 5;6 to 6;4 (M = 5;11). The age control group was in the range between 9;2 and 11;8 (M = 10;0, SD = 1;0) and included 4 girls and 6 boys. For both control groups, no hearing problems were reported by parents. The selection of control children was made with the help of teacher reports and included only children whose school achievement was not exceptionally high or low. Again, bilingual participants were excluded. Concerning socioeconomic status, children included in the study were from middle class families; the schools and the families did not come from particularly disadvantaged environments.
Materials
Participants were administered three tasks: a single verb production task, a single verb comprehension task, and a sentence production task. In each task, argument-structure complexity was manipulated by presenting either transitive or intransitive verbs. In the sentence production task, length was also independently manipulated by adding modifiers.
Two groups of verbs were selected and two sets of sentences were built for each verb: Eighteen intransitive (V1) and 18 transitive ones (V2; see Table 1).4 Intransitive sentences were made up of the same subject (e.g., le canard [the duck]) and for all sentences, an intransitive verb and an optional adjunct. The transitive sentences were composed of a subject, the transitive verb, and its mandatory argument.
|
So as not to confuse argument-structure complexity with other possible factors, a number of variables that could have played a role in sentence production were controlled by selecting among a much larger set of words two sets matched for lexical characteristics. In this way, the two sets of sentences (V1 vs. V2) were indistinguishable for length, in terms of number of phrases (three phrases per sentence), mean number of phonemes (V1: M = 17.72, SD = 1.41; V2: M = 16.56, SD = 3.01; t(34) = 1.488, p = .146), and mean time needed to utter the sentence in ms (V1: M = 1293, SD = 117; V2: M = 1234, SD = 162; t(34) = 1.261, p = .216).
The sentences were recorded and digitized. In addition, other lexical variables that could have played a role were also controlled: imageability, frequency, and familiarity. Because databases of lexical frequency, familiarity, and imageability do not exist in French for the developmental age, 24 judges evaluated the frequency, the familiarity, and the imageability of a selected list of 260 words—nouns and verbs—on a scale from 0 to 10, where 0 corresponded to the least frequent (familiar or imageable) and 10 corresponded to the most frequent (familiar or imageable). The judges were selected from among caregivers, teachers, and mothers, all of whom had significant experience with children between the ages of 6 and 11 years. Selecting from among the list of words, it was possible to match the constituents—the verb and the internal phrase—for lexical characteristics across conditions. Thus, verbs in V1 were indistinguishable from verbs in V2 for (a) mean phonological length, V(V1): M = 644, SD = 80; V(V2): M = 640, SD = 89; t(34) = 0.168, p = .868; (b) subjective frequency, V(V1): M = 7.61, SD = 1.23; V(V2): M = 7.26, SD = 1.41; t(34) = 0.803, p = .428; (c) familiarity, V(V1): M = 8.48, SD = 1.37; V(V2): M = 8.11, SD = 1.52, t(34) = 0.751, p = .458; and (d) imageability, V(V1): M = 8.28, SD = 1.12; V(V2): M = 7.73, SD = 1.28. t(34) = 0.880, p = .176. The internal phrase was also matched for the same characteristics: (a) subjective frequency, N(V1): M = 7.31, SD = 1.13; N(V2): M = 7.74, SD = 1.43; t(32) = 0.943, p = .353; (b) familiarity, N(V1): M = 8.31, SD = 1.09; N(V2): M = 8.44, SD = 1.12; t(32) = 0.363, p = .719; and (c) imageability, N(V1): M = 9.04, SD = 0.60; N(V2): M = 8.94, SD = 1.08; t(32) = 0.322, p = .750.
To investigate the effect of length itself, each sentence was built in a short (S) and long condition (L). In the long condition, a modifier was added to the subject noun phrase and to the internal complement; the information carried by the modifier was not relevant to the general meaning of the sentence and was of great simplicity, frequency, and familiarity. Sentences in this condition were comparable to the short (S) sentences for syntactic and argument-structure complexity, except for length (see Table 1).
For the production tasks, a child-friendly colored picture was drawn for each verb (single verb production task) or sentence (sentence production task), depicting a well-known cartoon character (Donald Duck) performing the action. For the single-verb comprehension task, three pictures were presented for each trial; in each picture, Donald Duck performs an action. The three images were presented on a touchscreen—one picture was the target word, and the other two were foils, one foil semantically related to the target and one foil unrelated (e.g., rire [to laugh]; pleurer [to cry]; marcher [to walk]).
The groups of pictures were rated by 25 judges for how well they represented the sentence on a scale from 0 to 10 (0 = no relation between image, 10 = perfectly clear representation). Transitive and intransitive sentences did not differ on this dimension: quality of the representation, V1: M = 8.91, SD = 0.73; V2: M = 8.64, DS = 1.10; t(34) = 0.860, p = .396.
Procedure
Single Verb Production
The child was asked to name the action performed by Donald Duck in a picture presented on a computer screen. If the first answer did not match the target word, a second answer was encouraged and, if correct, accepted. The dependent variable was the rate of incorrect answer; a correct answer was scored when the verb corresponded to the target. In addition, certain answers that did not match the target verb were accepted if the child produced a strict synonym comparable for semantic complexity and specificity. For instance, we accepted peigner (to comb) for coiffer (to do one's hair); tousser (to cough) for éternuer (to sneeze); tuer (to kill) for fusiller (to shoot); and casser (to break) for détruire (to destroy). All other answers were scored as errors: verbs that were semantically related but not close synonyms, visually related verbs, unrelated verbs, and no answers. The order of presentation was randomized. Ten practice items were used to familiarize the child with the task, and feedback was provided during the practice but not during the experimental task.
Single-Verb Comprehension
All 10 children with SLI were tested on sentence production scores, as were the 10 children matched for chronological age and 10 children matched with children with SLI. A picture-pointing task was used. The children were instructed to point to the picture depicting the action of the verb they heard through the headphones as quickly as possible. The touchscreen recorded both the type of answer and response time (for the correct answer only), which were computed as dependent variables. Reaction times were taken from the onset of the orally presented verb. The order of presentation was randomized. Ten practice items were used to familiarize the child with the task, and feedback was provided during the practice but not during the experimental task.
Sentence Production
For each of the 72 trials, the child was presented with a picture on the computer screen. The picture was left on the screen until the beginning of the next trial, which was preceded by a short pause of variable length. When the picture was presented, the experimenter read three or five words in a random order—the order was never the correct subject–verb–complement order—and the child had to repeat all the words; in cases where the child failed to repeat all the words, the experimenter read the words again and the child was given a second or a third chance to repeat the words. The orally presented words were all open-class words corresponding to the subject and the object noun phrases, and a verb—in the infinitive form—of the target sentence, e.g., courir (to run), canard (duck), champ (field). In the long condition, the subject and the complement nouns were read with the corresponding adjectives (e.g., canard jaune [yellow duck], grand champ [big field]). The child was instructed to use all the words that were given to describe the action in the picture; all of the children were able to do so. Children never added materials, and most produced the target sentences: The subject phrase was correctly used in 97.8% of the sentences by children with SLI, and in 99.8% of sentences by language control children and was not dependent on the argument complexity of the target sentence. The internal argument was used in 89% of sentences by children with SLI and in 96% of sentences by the control group. In the long condition, adjectives were used in most of the cases (SLI, 93.5% of produced phrases; language controls, 95.2% of produced phrases). The high percentage of adjective use was probably because the nouns were always given with their adjective (e.g., "big field," "green duck"), and the children had to repeat them; if they did not correctly repeat all the words, they were not asked to produce the sentence. Moreover, the adjectives were extremely simple. When the child correctly repeated all the words, the experimenter asked a question to elicit the target sentence—for example, Qu'est qu'il a fait hier le canard? (What did the duck do yesterday?)
The question was formulated using the passé compose in order to elicit an answer with the same tense; the passé compose is likely to be used since it is very frequent in spoken French and is the first tense to be acquired in French after the present (Bassano et al., 2001). Stimuli were presented randomly. Ten training trials were provided before the experimental materials were presented. Three pauses were allowed during the experiment. Children were tested individually in a quiet room at the school they attended. Testing took about 20 min, which was well within the attention span of both participants with SLI and younger normal controls.
Scoring and Reliability
All sentences were scored for the presence and absence of auxiliary verbs and articles. Since there were two articles in each sentence, one for the subject and one for the complement, the position (first vs. second position) was considered as a further variable and entered in the analysis of variance as a within-subject factor. A rate of article omission was computed for each participant. Only the following sentences were included in the analysis: (a) sentences where the subject, verb, and internal argument were present; and (b) sentences that preserved the same argument structure as the target verb. Thus, for example, an intransitive verb produced in place of a transitive verb was a condition for exclusion, but this kind of error was actually extremely infrequent (six times out of 1,440). Close synonyms were accepted. Using these criteria, 19% of the sentences were excluded for participants with SLI and 18% of sentences for language-matched peers, and 3% for the age control group.
Scoring reliability has been evaluated on a subgroup of six children (three SLI and three normal language controls), for whom two different judges scored the presence or absence of articles. The accord was almost total (99%); in the very few cases of disagreement, the incongruity depended on wrong encoding of cases that were clear, per se.
Results
Single Verb Production (Naming Action)
The 10 children with SLI were compared with younger normal-language–matched peers (NLC) for the mean rate of errors in the two conditions of argument-structure complexity (transitive vs. intransitive). The age control group was not included in the statistical analysis, as these participants performed at a ceiling level. Rather, they served as the control for the appropriateness of the task, as their performance confirmed that the pictures were easy enough to name: They correctly named 94.3% of the pictures when the second answer was considered. A mixed-model analysis of variance (ANOVA) of the rate of the errors by participants was computed. Subject groups (SLI and NLC) served as the between-subjects factor, and transitivity served as the within-subject factor.
Both SLI and NLC groups correctly named the actions most of the time (correct responses of language-matched control children: V1 = 90.91%, V2 = 89.38%; correct responses of participants with SLI: V1 = 90.13%, V2 = 88.95%). The two groups do not significantly differ with respect to the rate of correct answers, F < 1. The main effect of transitivity was not significant, F < 1. Typically developing children committed slightly more errors with V2 than with V1, but the difference was not significant, t(1, 9) = 1.801, p = .105.
Single Verb Comprehension (Picture Pointing)
The performance of the 10 children with SLI was compared with that of NLC as to the mean percentage of errors and latencies of answers for the comprehension task. A mixed-model ANOVA of the percentage of errors was computed. Subject groups (SLI and NLC) served as between-subject factor, while transitivity served as within-subject factor. The age control group was not included in the statistical analysis to avoid distortion due to their performance at a ceiling level; they pointed at the correct picture in 97% of trials. Globally, the children performed at a good level (correct responses of participants with SLI: V1 = 86.67%, V2 = 85.56%; correct responses of language control participants: V1 = 87.50%, V2 = 85.56%). The two groups did not significantly differ in the percentage of correct answers, F < 1. Transitivity did not significantly influence performance, F < 1. With respect to latencies (participants with SLI: M = 2093, SD = 289; NLC participants: M = 2303, SD = 340), the two groups pointed at the correct picture with similar delays F (1, 18) = 1.373, p = .256. Argument structure did not have a significant influence on latency, F < 1.
Sentence Production
Auxiliary Verb Omission
The rate of auxiliary verb omissions was computed for each subject in the three groups tested: children with SLI, children matched for sentence production, and children matched for age. The mean percentage of auxiliary omission by group can be seen in Table 2. The age-matched children performed at a ceiling level with 100% of auxiliary use in obligatory context, so they have been excluded from the statistical analysis to avoid distortions due to the ceiling effect.
|
As expected, results revealed that children with SLI omitted the auxiliary more often than did younger controls, F(1, 18) = 6.132, p = .023; in the analysis by item, the difference remained significant, F(1, 34) = 88.177, p = .000. Transitivity was also statistically significant by subject, F(1, 18) = 4.599, p = .047, and by item, F(1, 34) = 11.134, p = .002. In spite of a numerically larger difference between transitive and intransitive conditions in children with SLI than in the control group (see Table 2), the Group x Transitivity interaction failed to reach the significance level, F < 1 (
2 = .054, power = .121.) Finally, length did not affect the use of the auxiliary, F < 1 (2 = .001, power = .053.)
Article Omission
We analyzed the percentage of omissions of articles in obligatory contexts in the three groups tested: participants with SLI, participants matched on sentence production, and participants matched for age. Variability in the use of articles by the two groups of normal-language participants was rather reduced; language controls omitted the second article 2.5% of the time on average, while the age control group never omitted articles. As long as the age control group presented no variability and a ceiling performance, we did not include this group in the analysis of variance. Figure 1 presents a summary of the performance of participants with SLI and the control group.
|
None of the main variables had a significant effect: Argument complexity was not significant, F(1, 18) = 1.365, p = .258 (2 = .71, power = .198), nor was the effect of length, F < 1 (2 = .035, power = .120), or position, F < 1 (2 = .001, power = .053). On the contrary, the Transitivity x Position interaction, F(1, 18) = 7.462, p = .014, and the second-level Transitivity x Position x Group interaction, F(1, 18) = 9.102, p = .007, were significant. Both interactions were close to significance in the analysis by item: Transitivity x Position, F(1, 18) = 3.581, p = .067, and Transitivity x Position x Group, F(1, 18) = 3.992, p = .054. To identify the origin of the second-level interaction, we computed an analysis of variance considering separately the rate of utilization of the article of the external argument—the subject—and that of the internal complement. This analysis revealed that argument structure had an effect on article omission only in the second position—internal complement—for children with SLI but not for younger normal language children (see Figure 1).
Concerning the article of the first noun phrase, the subject of the sentence, the group of children with SLI omitted the article more frequently than did the control group, but the difference was not statistically significant, F < 1 (2 = .157, power = .411), and the use of the article was not influenced by argument complexity, F(1, 18) = 1.762, p = .201 (2 = .089, power = .242), or length, F < 1 (2 = .003, power = .056). On the contrary, the article of the internal argument was omitted more frequently by children with SLI than by normal language controls, F(1, 18) = 5.117, p = .035, and the difference was significant in the analysis by item, as well, F(1, 18) = 27.874, p = .000. The main effect of argument complexity was significant by subject, F(1, 18) = 4.760, p = .043, and by item, F(1, 18) = 5.839, p = .021; the article of the internal complement was omitted more often with transitive than intransitive sentences. Importantly, for the article of the internal complement, argument complexity differently affected the use of the article in the two groups; this difference surfaced in the statistically significant Argument Complexity x Group interaction, F(1, 18) = 6.191, p = .023, which remained significant in the analysis by item, F(1, 18) = 6.476, p = .016. Children with SLI more often omitted the article in transitive than in intransitive sentences, whereas the number of omissions in the language control group did not vary significantly across argument-structure type. Length did not show a significant effect, F(1, 18) = 2.730, p = .116, (2 = .132, power = .346). Although one could argue that the effect of length failed to emerge because of the reduced number of participants, it should be noted that participants tended to omit slightly more articles with short rather than long sentences; thus, the argument of a limited statistical power as the basis of these results can hardly be made.
Although the main focus was on the use of the grammatical morphemes, post hoc it was decided to analyze the use of internal and external complements because results showed that, for example, the subject argument can be influenced by argument-structure complexity (Grela, 2003; Grela & Leonard, 1997). Results showed that the language-impaired children did not present problems in this respect (external phrase: children with SLI: V1 = 98.3%, V2 = 98.6%; language control children: V1 = 99.4%, V2 = 99.7%; internal phrase: children with SLI: V1 = 84.8%, V2 = 92.1%; language control children: V1 = 94.4%, V2 = 96.9%). They correctly used the subject phrase in 97.8% of the sentences, which was comparable to younger peer control children who used the subject phrase in 99.8% of the sentences, F(1, 18) = 2.136, p = .161. Subject-argument omission did not depend on the argument complexity, F < 1 (2 = .017, power = .082), or length, F < 1 (2 = .000, power = .050). The use of the internal argument did not reveal any difference between the group of language-impaired children and younger peers; the first used the internal argument in 89% of the sentences, whereas the control group used the internal argument in 96% of the sentences, F(1, 18) = 2.136, p = .161 (2 = .106, power = .283). Nonetheless, there was a general tendency for children in both groups to omit the adjuncts more frequently than the arguments, F(1, 18) = 4.812, p = .042, which shows that, similar to younger language peers, children with SLI treated the adjuncts as optional elements.
Discussion
From the tasks with single verbs—action naming and picture pointing tasks—it was first observed that three groups of children named transitive and intransitive verbs in isolation at similar and high percentages, showing no bias toward either of the two argument types. Therefore there is no reason to think that one set of verbs—either transitive or intransitive—was easier or better known by our groups of children; moreover, in the task used, no evidence emerged that would support the claim that at the lexical level, argument-structure complexity exerts any influence on verb processing (although a more sensitive test might show a difference). Comprehending and producing a verb with a more complex argument structure in isolation does not appear to be associated with a higher computational load, in children with SLI and in younger normal children, as is also shown by similar response times in verb comprehension.
Overall, the sentence-production task showed that children with SLI omitted the article and the auxiliary verb in obligatory contexts more often than younger language peers, which is in line with several previous findings reported in the literature (e.g., Beastrom & Rice, 1986; Bortolini, et al., 1997; Crago and Paradis, 2003a, 2003b; Hadley & Rice, 1996; Leonard et al., 1997; Methé & Crago, 1996; Paradis and Crago, 2001; Rice & Wexler, 1996).
In the present study, we first addressed whether and how argument complexity affects the omission of grammatical morphemes in order to test opposite predictions drawn from the two main theoretical accounts of grammatical disorders in SLI. The hypothesis of limited-processing capacities predicts that (a) processing factors and, in particular, argument-structure complexity should influence the rate of omission of grammatical morphemes and (b) the effect of argument-structure complexity should be greater on children with SLI than on children with normal language. On the other hand, representational-deficit hypotheses state that the use of grammatical morphemes should be relatively constant and independent from processing factors.
In this study, we found clear evidence that the rate of omission of grammatical morphemes varies as a function of argument complexity. Considering the use of the auxiliary verb, we replicated the results reported by Grela and Leonard (2000) that the use of the auxiliary varies as a function of argument complexity. This effect was significant in children with SLI but not in the younger control group—although the Group x Argument Complexity interaction was not significant. Concerning article omission, results follow the same pattern, but the relevant Argument Complexity x Group interaction was statistically significant. Children with SLI omitted the article of the internal complement, but not the article of the subject, more frequently with transitive than intransitive structures, whereas this was not the case for control children. The higher rate of omission of the article of the internal complement compared with the rate of omission of the article of the subject—and a significant argument-complexity effect for the former but not for the latter—may be explained by a greater computational load just after the verb. It may also be due to the thematic and grammatical processing required for the subject or because the subject and its article were the same in all sentences, although the internal argument changed.
The problematic use of articles shown by children with SLI in part conflict with results of studies of French-speaking children with SLI. For example, LeNormand et al. (1993) showed that children with SLI whose ages range from 4 to 6 years use articles in obligatory contexts at the same rate as the younger control group (3;0–3;6). Our data are congruent with these results concerning the use of the article of the subject-noun phrase but not for the article of the internal complement, which was omitted significantly more often by children with SLI than by language-matched children. One possible reason may be the different ages of the participants in the two experiments: Our group was much older than the group (9;11) in LeNormand et al. (1993), as was our control group (M = 5;10, SD = 0.2). As a consequence, it could be the case that although normal-language children rapidly master article use or come close to ceiling performance, children with SLI remain longer in the below-mastery level. Note that our older group of children with SLI used articles in about 85% of obligatory contexts, similarly to what Le Normand et al. (1993) reported. Paradis and colleagues (2003) also reported no differences between school-age children with SLI and normal-language peers in the use of articles; in this case, too, both groups were substantially younger than our groups. A second difference concerns the method: The studies focusing on French children with SLI reported here analyzed spontaneous speech samples (with the exception of Grela and Leonard's studies), whereas we used a structured sentence-elicitation task that tapped more on verbal working memory, compared with spontaneous production because children have to remember the words and use them. This more demanding task may produce more morphological errors and evidence problems that elsewhere were not detected.
On the whole, these results have implications for the current explanations of grammatical deficit in children with SLI and are worth discussion. The evidence that grammatical-morpheme omission varies with argument-structure complexity clearly supports the processing-limitation hypothesis, which assumes that children with SLI are slow or inefficient linguistic processors. The hypothesis of a deficient representation of grammatical morphology does not directly account for the fluctuation of the omission following argument-structure complexity (Eyer & Leonard, 1995; Gopnik, 1990a, 1990b; Gopnik & Crago, 1991; Ingram & Carr, 1994).
The processing-limitation hypothesis is also supported by the fact that in the present study, children always used the grammatical morphemes correctly. The underspecified or troublesome representation of grammatical paradigms should sometimes result in the wrong use of grammatical morphemes. Children with SLI and with normal language consistently used the correct morphemes; they correctly used auxiliary verbs and distinguished between être (to be) and avoir (to have). Both groups also used the correct articles for gender and number and used definite and indefinite articles correctly, as well.
Nonetheless, problems in the interpretation of the results may arise from the fact that the rate of omission of grammatical morphemes for the control group matched on sentence production was much lower than for children with SLI; it may be that if the control children were more vulnerable to omission and were at the same article proficiency level as the children with SLI, argument complexity would have had an effect and the Group x Transitivity interaction would not have occurred. In further studies, researchers could use a control group matched on article proficiency. It should also be noted that we are not claiming that children with SLI do not have any representational deficit but that at least part of the problem with grammatical morphology in children with SLI depends on a processing limitation.
The second aim of this study was to determine whether structural complexity is a primary factor, independent of the length of the sentence. Results show that sentence length cannot account for the argument-complexity effect, as we compared two sets of sentences (V1 and V2) with the same length. In addition, we controlled for other variables such as frequency, familiarity, and imageability, so these variables cannot explain our results. Moreover, length by itself did not reveal any significant influence on the use of grammatical morphemes—this finding is consistent with that of Grela and Leonard (2000) but is in contrast with those of Bloom (1990, 1993) and Montgomery (1995, 1996, 2003). Bloom reported an effect of length on the use of the subject argument, but he tested much younger children than we did; the participants in his study were 3 typically developing children between 1 and 2 years of age, whereas the typically developing children in our study were between 5 and 6 years of age and the children in our SLI group were between 8 and 13 years of age. Because of the age differences, language proficiency in these two studies is not comparable. Montgomery's study focused on comprehension and manipulated length in a different way by creating syntactic redundancy, morphological redundancy, and lexical redundancy. For example, in Redundant Adjectival and Adverbial Items (C1), "The girl chases the horse" is nonredundant; "The pretty little girl quickly chases the big fast horse" is redundant. Even in the case of the sentences in our study that were most similar to his, he used five or six adjectives per sentence to increase sentence length, whereas we used just two. It is possible that in the present experiment, the computational load associated with length was too weak and not challenging enough; if we had added five or six words in the long condition, as in Montgomery (1995), the effect of length might have emerged. However, a structured sentence-elicitation task in which words have to be remembered and used seems very hard to use because children can hardly remember more words than those that were given in this study. The long condition appears to be challenging and not a trivial task for language-impaired children; this corroborates the hypothesis that the experimental manipulation of length is sizeable and, consequently, that length is not a relevant processing factor for children with SLI or typically developing children at the ages considered in this study. The idea that the effect of length failed to emerge because of statistical power limitations can also be ruled out, considering the fact that children with SLI omitted slightly more articles for short sentences than for long sentences, as did the control group, and there was no tendency toward higher omissions with long sentences compared to short sentences.
Finally this study tried to tease apart the lexical level from the sentence level as a possible origin of argument-structure complexity effect. As previously mentioned, no evidence emerged that argument-complexity plays a significant role at the lexical level, considering all groups of children performed similarly with transitive and intransitive verbs in isolation, in comprehension, and in production. In future research, it would be desirable to keep these two dimensions separated; the comparison of the effect of argument complexity in sentence and single-verb comprehension could be informative in this regard.
The three main results reported here form a cohesive picture within the sentence production model put forward by Bock and Levelt (1994) that can be helpful to guide further research. The relevant feature of this model is represented by the two functionally and temporally separate stages: (a) the function assignment and (b) the subsequent grammatical-morphemes processing. This configuration implies that thematic and grammatical roles are assigned before functional words are retrieved and put in the right slot. Under normal conditions, typically developing children assign thematic roles in transitive and intransitive sentences without any problem, and argument-structure complexity has no detectable effect on grammatical-morpheme processing. On the contrary, if the grammatical encoding or all of language processing is slower than normal (see Bock & Levelt, 1994), as is assumed on the part of children with SLI (Ellis Weismer, 1994, Ellis Weismer & Hesketh, 1996; Johnston, 1991, 1994), sentences—both transitive and intransitive—could sometimes be spelled out before closed class elements are fully retrieved and assigned to the corresponding slots. This would explain the higher rate of article and auxiliary verb omissions by children with SLI than by language peers. Moreover, as the production of transitive sentences is computationally more expensive than that of intransitive sentences (Collina et al., 2001; Collina et al., 2002; Rubin et al., 1996; Shapiro et al., 1987; Shapiro et al., 1991; Tabossi et al., 2004; Thompson et al., 1995, 1997), children with SLI would omit the article more often in transitive than in intransitive sentences, whereas typically developing children, whose resources suffice to complete all stages of linguistic processing, would perform comparably in the two conditions. On the basis of these results, the deficit with grammatical morphemes typically shown by children with SLI can be explained—at least in part—by a processing approach rather than by a classical representational deficit hypothesis. Concerning the issue of specificity, this study does not allow any clear conclusion as extra-linguistic cognitive domains were not investigated in depth and there were limitations in the paradigm used. Still these children do not have any gross nonverbal impairment; otherwise, they would not have met the criteria for SLI. It is possible that the processing deficit is specific to some components of language: for example, a limited syntactic processing deficit seems plausible. Again for the limited language domains under investigation and the logic of the paradigm used, we think that the conclusions cannot be pushed too far to specify which component is affected.
The results also have some clinical implications and show that is important to consider language from the point of view of processing resources and not simply from that of language knowledge: even very simple active structures (transitive vs. intransitive) may be a source of an important cognitive load and may present a remarkable source of difficulty for children with SLI. The ability to use grammatical morphemes varies following the argument structure of the sentence to be produced; it suggests that when a new morpheme is introduced in the rehabilitation program, it should first appear in the simplest structures. Afterward, when the child shows a good degree of mastery, the intervention can include more complex structures.
In summary, the results reported here demonstrate that at sentence level, and independent of length, argument-structure complexity influences the use of grammatical morphemes in children with SLI but not in normal language developing children, and these data can be better accounted for by the processing limitation hypotheses than by the representational deficit approaches.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Multiple subcategorization frames are allowed—for example, by so-called alternate dative verbs such as to give something to someone (NP_PP) or togive someone something (NP_NP).
3 In the predominant view, resources and limitation in SLI are conceived interms of speed of processing. Nonetheless, speed, space, and energy aretightly linked when the temporal dimension is considered: If a system israpid, it could perform more operations in a unit of time, and the energyrequired to perform the same operation would be greater if the system isslow/inefficient.
4 We decided not to include ditransitive verbs because it was not possible toselect a meaningful number of ditransitive verbs matched with the othertwo categories for lexical characteristics.
Received December 15, 2005
Revision received December 18, 2006
Accepted September 12, 2007
|
References |
|---|
|
TopAbstractReferences |
|---|
Anglade, J. C., Ravard, J. F., & Ravard, J. C. (1993). Test des compétences verbales et métalinguistiques [Test of verbal and metalinguistics skills]. Issy-les-Moulineaux, France: Editions Scientifiques et Psychologique.
Arapovic, D., & Andel, M. (2003). Morfoloske pogreske u diskursu djece s PJT [Morphological errors in discourse of children with SLI]. Hrvatska Revija za Rehabilitacijska Istrazivanja, 39, 11–16.
Bassano, D., Maillochon, I., Klampfer, S., & Dressler, W. U. (2001). L'acquisition de la morphologie verbale en français et en allemand autrichien. II: L'épreuve des faits [The acquisition of verbal morphology in French and Austrian German: II. The race facts]. Enfance, 53, 117–148.
Beastrom, S., & Rice, M. (1986, November). Comprehension and production of the articles a and the. Paper presented at the Annual Convention of the American Speech-Language-Hearing Association, Detroit, MI.
Bloom, P. (1990). Subjectless sentences in child language. Linguistic Inquiry, 21, 491–504.
Bloom, P. (1993). Grammatical continuity in language development: The case of subjectless sentences. Linguistic Inquiry, 24, 721–734.
Bock, K., & Levelt, W. J. (1994). Language production: Grammatical encoding. In M. A. Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 945–984). San Diego, CA: Academic Press.
Bortolini, U., Caselli, M. C., & Leonard, L. B. (1997). Grammatical deficits in Italian-speaking children with specific language impairment. Journal of Speech, Language, and Hearing Research, 40, 809–820.
Cipriani, P., Chilosi, A., Bottari, P., Pfanner, L., Poli, P., & Sarno, S. (1991). Uso della morfologia grammaticale nella disfasia congenita [Use of grammatical morphology in congential dysphasia]. Giornale Italiano di Psicologia, 18, 765–779.
Clahsen, H. (1989). Grammatical characterization of developmental dysphasia. Linguistics, 27, 897–920.
Clahsen, H. (1991). Child language and developmental dysphasia. Amsterdam: John Benjamins.
Collina, S., Marangolo, P., & Tabossi, P. (2001). The role of argument-structure in the production of nouns and verbs. Neuropsychologia, 39, 1125–1137.[CrossRef][Medline]
Collina, S., Pizzioli, F., Caporali, A., Basso, A., & Tabossi, P. (2002). The production of action nouns: A case study. Brain and Language, 83, 109–112.
Cook, V., & Newson, M. (1996). Chomsky's universal grammar: An introduction. Oxford: Blackwell.
Crago, M., & Paradis, J. (2003a). Two of a kind? Commonalities and variation in languages and language learners. In Y. Levy & J. Schaeffer (Eds.), Language competence across populations: Towards a definition of specific language impairment (pp. 97–110). Mahwah, NJ: Erlbaum.
Crago, M., & Paradis, J. (2003b). Language impairment in children: Cross-linguistic studies. In R. Kent (Ed.), The MIT encyclopedia of communication sciences and disorders (pp. 327–329). Cambridge, MA: The MIT Press.
Elin Thordardottir, & Ellis Weismer, S. (2002). Verb argument structure weakness in specific language impairment in relation to age and utterance length. Clinical Linguistics & Phonetics, 16, 233–250.[CrossRef][Medline]
Elin Thordardottir, & Namazi, M. (2007). Specific language impairment in French-speaking children: Beyond grammatical morphology. Journal of Speech, Language, and Hearing Research, 50, 698–715.
Ellis Weismer, S. (1994, June). Factors influencing novel word learning and linguistic processing in children with specific language impairment. Paper presented at the Symposium on Research in Child Language Disorders, Madison, WI.
Ellis Weismer, S., & Hesketh, L. (1996). Lexical learning by children with specific language impairment: Effects of linguistic input presented at varying speaking rates. Journal of Speech and Hearing Research, 39, 177–190.[Medline]
Eyer, J., & Leonard, L. B. (1995). Functional categories and specific language impairment: A case study. Language Acquisition, 4, 177–203.[CrossRef]
Gopnik, M. (1990a, April 19). Feature blind grammar and dysphasia. Nature, 344, 715.[Medline]
Gopnik, M. (1990b, September 6). Genetic basis of grammar defect. Nature, 347, 26.[Medline]
Gopnik, M., & Crago, M. (1991). Familial aggregation of developmental language disorder. Cognition, 39, 1–50.[CrossRef][Medline]
Grégoire, J. (1992). Evaluer l'intelligence de l'enfant [Evaluate the child intelligence]. Liège, Belgium: Mardaga.
Grégoire, J. (1999, June 29, July 2). Construct validity of the French adaptation of the WISC-III. Paper presented at the Joint European Conference of the International Association of Cross-Cultural Psychology and the International Test Commission, Graz, Autriche, France.
Grela, B. (2003). The omission of subject arguments in children with specific language impairment. Clinical Linguistics & Phonetics, 17, 153–169.[CrossRef][Medline]
Grela, B., & Leonard, L. B. (1997). The use of subject arguments by children with specific language impairment. Clinical Linguistics & Phonetics, 11, 443–453.[CrossRef]
Grela, B., & Leonard, L. B. (2000). The influence of argument-structure complexity on the use of auxiliary verbs by children with SLI. Journal of Speech, Language, and Hearing Research, 43, 1115–1125.
Haarmann, H. J., Just, M. A., & Carpenter, P. A. (1997). Aphasic sentence comprehension as a resource deficit: A computational approach. Brain and Language, 59, 76–120.[CrossRef][Medline]
Hadley, P., & Rice, M. (1996). Emergent uses of BE and DO: Evidence from children with specific language impairment. Language Acquisition, 5, 209–243.[CrossRef]
Hamann, C., Ohayon, S., Dubé, S., Frauenfelder, L., Rizzi, L., Starke, M., & Zesiger, P. (2003). Aspects of grammatical development in young French children with SLI. Developmental Science, 6, 151–158.[CrossRef]
Hansson, K. (1997). Patterns of verb usage in Swedish children with SLI: An application of recent theories. First Language, 17, 195–217.
Hansson, K., Nettelbladt, U., & Leonard, L. (2003). Indefinite articles and definite forms in Swedish children with specific language impairment. First Language, 23, 343–362.[Abstract]
Ingram, D., & Carr, L. (1994, November). When morphological ability exceeds syntactic ability: A case study. Poster presented at the Annual Convention of the American Speech-Language-Hearing Association, New Orleans, LA.
Johnston, J. R. (1991). Questions about cognition in children with specific language impairment. In J. Miller (Ed.), Research on child language disorders (pp. 299–307). Austin, TX: Pro-Ed.
Johnston, J. R. (1994). Cognitive abilities of children with language impairment. In R. Watkins & M. Rice (Eds.), Specific language impairments in children, 4, pp. 107–121). Baltimore: Brookes.
Kerschensteiner, M., & Huber, W. (1975). Grammatical impairment in developmental aphasia. Cortex, 11, 264–282.[Medline]
Khomsi, A. (2001). ELO: évaluation du langage oral [ELO: Assessment of oral language]. Paris: ECPS.
LeNormand, M. T., Leonard, L. B., & McGregor, K. K. (1993). A cross-linguistic study of article use by children with specific language impairment. European Journal of Disorders of Communication, 28, 153–163.[CrossRef][Medline]
Leonard, L. B. (1998). Children with specific language impairment. Cambridge, MA: The MIT Press.
Leonard, L. B., & Bortolini, U. (1998). Grammatical morphology and the role of weak syllables in the speech of Italian-speaking children with specific language impairment. Journal of Speech, Language, and Hearing Research, 41, 1363–1374.
Leonard, L. B., Bortolini, U., Caselli, M. C., McGregor, K., & Sabbadini, L. (1992). Morphological deficits in children with specific language impairment: The status of features in the underlying grammar. Language Acquisition, 2, 151–179.[CrossRef]
Leonard, L. B., Eyer, J., Bedore, L., & Grela, B. (1997). Three accounts of the grammatical morpheme difficulties of English-speaking children with specific language impairment. Journal of Speech, Language, and Hearing Research, 40, 741–753.
Leonard, L. B., Salameh, E. K., & Hansson, K. (2001). Noun phrase morphology in Swedish-speaking children with specific language impairment. Applied Psycholinguistics, 22, 619–639.[CrossRef]
Levelt, W. J. M. (1989). Speaking: From intention to articulation. Cambridge, MA: The MIT Press.
Mervis, C. B., & Klein-Tasman, B. P. (2004). Methodological issues in group-matching designs: Alpha levels for control variable comparisons and measurement characteristics of control and target variables. Journal of Autism and Developmental Disorders, 34, 7–17.[CrossRef][Medline]
Methé, S., & Crago, M. (1996, June). Verb morphology in French children with language impairment. Paper presented at the Symposium on Research in Child Language Disorders, Madison, WI.
Montgomery, J. W. (1995). Sentence comprehension in children with specific language impairment: The role of phonological working memory. Journal of Speech, Language, and Hearing Research, 38, 187–199.
Montgomery, J. W. (1996). Sentence comprehension and working memory with specific language impairment. Topics in Language Disorders, 17, 19–32.
Montgomery, J. W. (2003). Working memory and comprehension in children with specific language impairment: What we know so far. Journal of Communication Disorders, 36, 221–231.[CrossRef][Medline]
Paradis, J., & Crago, M. (2000). Tense and temporality: Commonalities and differences in French-speaking children with specific language impairment and French second language speakers. Journal of Speech, Language, and Hearing Research, 43, 834–848.
Paradis, J., & Crago, M. (2001). The morphosyntax of specific language impairment in French: Evidence for an extended optional default account. Language Acquisition, 9, 269–300.[CrossRef]
Paradis, J., Crago, M., & Genesee, F. (2003). Object clitics as a clinical marker of SLI in French: Evidence from French-English bilingual children. In B. Beachley, A. Brown, & F. Conlin (Eds.), Proceedings of the 17th Annual Boston University Conference on Language Development (pp. 638–649). Sommerville, MA: Cascadilla Press.
Restrepo, M. A., & Gutierrez-Clellen, V. F. (2001). Article use in Spanish-speaking children with SLI. Journal of Child Language, 28, 433–452.[CrossRef][Medline]
Rice, M. L., & Wexler, K. (1996). Toward tense as a clinical marker of specific language impairment in English-speaking children. Journal of Speech and Hearing Research, 39, 1239–1257.[Medline]
Rice, M. L., Wexler, K., & Cleave, P. (1995). Specific language impairment as a period of extended optional infinitive. Journal of Speech and Hearing Research, 38, 850–863.[Medline]
Roberts, J., Rescola, L., & Borneman, A. (1994, June). Morphosyntactic characteristics of early language errors: An examination of specific expression language delay. Poster presented at the Symposium on Research in Child Language Disorders, Madison, WI.
Rubin, S. S., Newhoff, M., Peach, R. K., & Shapiro, L. P. (1996). Electrophysiological indices of lexical processing: The effects of verb complexity and age. Journal of Speech and Hearing Research, 39, 1071–1080.[Medline]
Shapiro, L. P., Brookins, B., Gordon, B., & Nagel, N. (1991). Verb effects during sentence processing. Journal of Experimental Psychology: Learning, Memory, and Cognition, 17, 983–996.[CrossRef][Medline]
Shapiro, L. P., Zurif, E., & Grimshaw, J. (1987). Sentence processing and mental representation of verbs. Cognition, 27, 219–246.[CrossRef][Medline]
Tabossi, P., Collina, S., Caporali, A., Pizzioli, F., & Basso, A. (2004, March). Speaking of actions: The case of CM. Poster presented at Trends in European Cognitive Psychology, Naples, Italy.
Tabossi, P., Collina, S., Caporali, S., Pizzioli, F., & Basso, A. (2005). Speaking of actions: The case of CM. Manuscript submitted for publication.
Thompson, C. K., Lange, K. L., Schneider, S. L., & Shapiro, L. P. (1997). Agrammatic and non-brain-damaged subjects' verb and verb argument structure production. Aphasiology, 11(4–5), 473–490.[CrossRef]
Thompson, C. K., Shapiro, L. P., Li, L., & Schendel, L. (1995). Analysis of verbs and verb argument structure: A method for quantification of aphasic language production. In P. Lemme (Ed.), Clinical aphasiology (pp. 121–140). Austin, TX: Pro-Ed.
van der Lely, H. K. J. (1994). Canonical linking rules: Forward versus reverse linking in normally developing and specifically language impaired children. Cognition, 51, 29–72.[CrossRef][Medline]
van der Lely, H. K. J. (1996). Language modularity and grammatically specific impaired children. In M. Aldridge (Ed.), Child Language (pp. 188–201). Avon, United Kingdom: Multilingual Matters.
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| All ASHA Journals | AJA | AJSLP | JSLHR | LSHSS |
