Verb Argument Structure Weakness in Specific Language Impairment Essay (Article Review)

Abstract

An argument is to explain something one believes is true. Each argument has a special construction, which may be to the semantic side or to the syntactic side. The syntax is the way in which a statement is organized or arranged. Verbs play an important role in sentence meaning as they represent the action or movement. Thordardottir and Weismer, 2002, investigated verb argument structure weakness in school-age children with specific language impairment as related to age and utterance length. This work aims at giving a brief account of arguments and argument structure a critical review of the article of Thordardottir and Weismer, 2002 will follow.

One of the main reasons for giving a speech or writing an article or thesis is to explain something one believes is true. In this sense, the argument is different from storytelling and is different from an everyday conversation where we make emphatic statements (as I am allergic to eggs). Emphatic statements (assertions) are based on implicit argument, as colleagues, friends, or members of the family assume, there are reasons for the conclusion. Assertions can also be conclusions of preferences (as I do not like eating eggs) (Pylkkanen, L, 2002). In this work, a brief account of argument structure will precede a review of the article Verb argument structure weakness in specific language impairment in relation to age and utterance length by Thordardottir and Weismer, 2002.

Argument structure

Arguments have a special construction, the simplest is composed of three parts namely; a premise, inference, and conclusion. The premise of an argument is a statement, assumed true, which one puts forward (as it is a sunny day). As not everybody may approve it makes a premise prone to challenge. There may be one or more premises in a communication. The second part is the conclusion, which is the statement one wants others to agree with. What is between premises and conclusion is inference, which are statements that take the audience or readers from premises to conclusion. Here is where logic processes work to relate the main statement with substatements (Goldberg, 2002).

Argument structure has two sides; one is about the meaning or difference between meanings of words or symbols (semantic side). The other side is about the rule or patterns of the ordering and of the relationships between the words (verbs, nouns, adjectives) and other elements that construct a phrase or a sentence ( as hyphenation marks) (syntactic side). The word “syntax” comes from the ancient Greek word “syntaxis”, which means the way in which something is organized or arranged. From a syntactic point of view, argument structure is the ability of a verb to combine grammatically with noun phrases in a given sentence in a way that is suitable to a social situation or to communicate with a particular set of people (Bresnan, 1995).

The conventional perspective of argument structure is that a word or combination of words including the verb, objects, or phrases governed by the verb (predicates) blend or merge directly with arguments forming larger constituents as sentences of the same phrasal category that is a verbal phrase (VP), adjective phrase (AP), prepositional phrase (PP). Whenever merging applies, it fulfills a categorization selection (c-selection) feature. The categorization feature subcategorizes predicates in a way closely related to phrase structure. Again, whenever merging applies, a theta-role should be satisfied (Bowers, 2004). However, what is a theta-role? Many researchers switch, exchange or use the terms theta-grid and thematic relations in place of each other.

A thematic relation describes the last part of a word stem before a change in its form (often an addition at its end as [s] for pleural) that indicates a specific grammatical function. Theta role is a concept of ordering and relationship between words and other structural elements of the phrase or sentence as regards the number, type, and placement of obligatory arguments. There appears some ordering by ranking, organization, or mapping of theta roles into a certain position in a group of words (phrases, clauses, or sentences). Researchers defined this as the nature of theta roles and their ordering or briefly thematic hierarchy. Theta grids are a schematic representation of the argument structure of a word or a combination of words including the verb, objects, or phrases controlled by the verb that make a part of the sentence. They indicate where the verb keeps theta roles (Reinhart, 2002 and 2003).

A theta criterion is a prime tool of the Government and Binding theory, which suggests that all languages share a major part of grammar (universal grammar). In this theory, the theta criterion imposes the complementation between arguments and theta roles, in other words, every argument has one theta role, and every theta role serves one argument. The theta criterion acts as a filter on the underlying structure where vocabulary (lexical) items join (D-structure) (Black, 1999).

Linguistic expression includes a connection between an ordered set of sounds and a word, clause, or sentence meaning. This expression depends on grammar, of which syntax is an important component, for transference. Proper verbal construction and the word arrangement in a sentence are two important factors in deciding the meaning of a word or a statement (utterance). There are two types of verbal arguments; first, factual arguments of the verb, and second, supplementary, or additional arguments of the verb (do not belong to the fundamental argument of the verb). Argument structure theories, in the core, explain how to introduce these supplementary verbs. Researchers argue over the mechanisms by which verb arguments develop. The emphasis of the research was on languages like English where verbs have a grammatical form case that identifies the source, agent, or instrument of action of the verb and accordingly affects other sentence elements like nouns, pronouns, and adjectives. Opinions varied, as far as children’s learning is concerned; some believe that learning children have early access to the semantics of different verb classes. An opinion opposed by those who believe that learning children are somewhat reluctant or cautious and use verbs in the syntactic context as they heard them and slowly develop the syntax of verb argument (Sag and Wasow, 2001).

Article review

The research question

The aim of the study of Thordardottir and Weismer, 2002 was twofold:

1. To investigate the argument structure use by children at school age who are having specific language impairment (SLI). In addition, they aimed to evaluate argument structure alternations, indirectly, by knowing the ability to adapt (flexibility) in using the different argument structures of those verbs.
2. To investigate the number of argument places by examining the groups to find out similarities and differences in their use of multiple places and other argument structures.

The hypothesis

The authors provided a comprehensive review of the literature to the following points:

Viewpoints of verbs difficulties in children with SLI

The authors showed that verb difficulties for children with SLI are mainly in the grammatical category that considers qualities of action independent of tense. The difficulties may include verb inflection, which is a change in the form of a verb, often an addition at the end of it, which indicates a particular grammatical function. Verb difficulties also include occasional learning of new verbs and verb variations in the spontaneous language (this includes dependence on a group of frequently used verbs or general all-purpose verbs abbreviated as GAP verbs).

How verbs are important in current modern grammar, and types of verb arguments

Thordardottir and Weismer, 2002 suggested that verbs are central in modern current grammar as they are words used to show that an action is taking place or to indicate the existence of a state or condition. Therefore, verb arguments relate to the stem of the verb (thematic relationship). On the other hand, the argument structure of the verb mandates the presence of certain constituents in the sentence as an agent, the goal of verb action, and the receiver of verb action or motion (beneficiary). The authors also showed that a verb argument can be obligatory when the verb allows only one argument structure or optional when the verb allows two or more argument structures.

Verb’s argument structure in SLI children

The authors reviewed the literature that examined verb’s argument structure in SLI children and suggested:

1. 1. Children with SLI are likely to overlook a verb that requires two objects (ditransitive) when used with another verb to indicate person, number, mood, tense, or aspect (auxiliary verb).
   2. The mean length of the style of expression (utterance) matches that of normal control children. The authors assumed, therefore, that argument structure complexity in children with SLI is, perhaps, a source of added processing load for these children.

Thordardottir and Weismer, 2002, also reviewed the literature where others investigated argument structure in individuals with agrammatic aphasia. They agreed with the conclusion of Shapiro and others (after Thordardottir and Weismer, 2002) that verbs, which permit one argument structure, take less to progress than those, which permit two or more argument structures.

Studies of argument development in children

Based on their review of literature, the authors suggested that acquiring the complete range of argument structure choices for complex verbs continues into adolescence.

Studies of language weaknesses

The authors assumed that language impairment in SLI children includes simplifications or a failure to formulate the choices available in the practice area in the language.

Thus, Thordardottir and Weismer furnished the hypothesis that verb argument structure can be a problem of special interest in children with SLI, this is due to the central role of verbs in sentence structure, and because this role includes linking meaning with syntax. Moreover, even when clear errors are absent, limited use of argument structure available choices indicates:

1. Partial learning of the verb’s options, or
2. Limitations in performance that influence language production, or
3. Both A and B produce limited use of argument structure available choices.

The test of the hypothesis

Participants

The work of Thordardottir and Weismer, 2002, included 100 school-age children divided into two groups. The first group included 50 children with SLI and the second group included 50 children with normal language (NL). Both groups matched on age mean, 25 children were selected from each group (as subgroups) to match the mean length of utterance (MLU).

Criteria for children selection were strict and included: normal physical, emotional, and motor development. The authors performed tests to ensure normal hearing, normal visual acuity, and normal phonological skills. A certified speech-language pathologist made the diagnosis of SLI in the first group of children. All children with SLI either were enrolled in language intervention in their schools or had a multidisciplinary evaluation, which showed they need intervention. All 100 children showed normal cognitive (non-verbal) abilities. Children with SLI scored lower on each of the language measures. All children were monolingual native English speakers.

Procedure

The authors used narrative language samples (describe a book, a movie, how to play your favorite sport). A single examiner collected these samples, then copied and tabulated results as regards correct spelling and grammar (orthographic transcription). Coding destined for the smallest meaningful element of speech or writing in language samples was according to the systematic analysis of language transcription (SALT) as described by Miller and Chapman, 1993 (after Thordardottir and Weismer, 2002).

Analysis of argument structure, complexity, and correctness

The authors performed argument structure analysis by adapting Thompson and others, 1995 model of investigating argument structure in individuals suffering from aphasia (after Thordardottir and Weismer, 2002). The authors made one modification that was analyzing the middle 50 utterances from each sample (instead of 25 by Thompson et al, 1995). This resulted in the augmentation of the number of utterances and verbs in the present study.

The authors coded all verbs as regards number and type of possible argument structures. Then they coded each used verb for the actually used argument and argument structure. Thus, they measured in their study the actual verb use rather than the possible maximum complex argument structure of the verb. Supportive secondary verbs and those expressing grammatical mood (such as possibility or necessity) were not included in the study. The past participle was considered a verb if it only expresses an actual action portrayed by the verb. The authors coded subject omissions when they were acting as modifying nouns, adjectives, or adverbs within a sentence as well as subjects of verbs expressing a command or a request and subjects of infinitive verbs.

Thordardottir and Weismer, 2002, obtained the following measures for each participant:

1. Number of different arguments used.
2. Number of types of argument structures used.
3. Number of verbs used with more than one argument structure.
4. Number and type of argument structure errors.

Reliability

The authors reviewed copied tabulated results and the smallest meaningful element of speech or writing codes in 15 % of the total number of samples collected to ensure the reliability of the results. This showed 97.5% agreement with the initial results.

Findings

Omission of obligatory arguments

The authors used a one-way analysis of variance (one-way ANOVA) test to, statistically, express their results.

1. For chronological age-matched groups of children: Children with SLI showed a statistically significant difference in the extent to which they included all obligatory verb arguments.
2. For MLU matched subgroups, the difference was not statistically different.
3. The types of argument omission in children with SLI were distributed as follows: 82% for subject argument, 8% for theme argument, 4% for verbs that link the subject of a sentence with an adjective or noun phrase complement relating to it, and 3% for the goal argument. In normal language children, the distribution was 68%, 14%, 10%, and 7% respectively.

Range of arguments and argument structures

To report on the range of argument and argument structure, Thordardottir and Weismer, 2002 used the following parameters:

1. Number of different argument types used.
2. Number of argument structure types used.
3. Number of verbs used for multiple argument structures.

To test for statistical significance, the authors used analysis of variance test (ANOVA) and Post Hoc Fischer LSD (least significant difference) test for matching groups.

1. For groups matching in chronological age: Children with SLI differed significantly from normal language children in all three measures (A, B, and C). As regards the number of verbs used for multiple argument structures, children with SLI used fewer types for three-place argument structures than did the normal language children.
2. For subgroups matching in MLU: Children with SLI showed a significant difference in item B than did the normal language children. As regards item C, children with SLI showed a significant difference in the number of three placed argument structure types used.

Interpretation of results

Thordardottir and Weismer, 2002, investigated argument structure in school-age children in two ways: first, as omissions (errors) of obligatory arguments and second, as the catalog of range of arguments and arguments structures used.

For omissions of obligatory arguments, their findings suggested that considering all types of errors, children with SLI at school age show a statistically significant difference. The commonest argument involved in both groups was the subject argument. They assumed that the omission of obligatory arguments is not the foremost source of the difficulty. Their findings came in agreement with many previous studies.

For argument and argument structure, children with SLI differed from age-matched normal language children significantly. Children with SLI showed fewer argument types as well as fewer argument structure types and showed reduced flexible use of verb variations. They assumed that school-age children with SLI show delay in argument structure development.

The MLU matched subgroup of children with SLI showed significantly lesser types of multiple place argument structures (mainly three-place). The authors assumed that complex argument structures (as three-place argument structures) stand for an important area of difficulty for children with SLI even if subgroups matched for mean length of utterance.

Personal Assessment

The introduction section of the study of Thordardottir and Weismer, 2002, provided a comprehensive review of literature in the area of verb argument structure in children with SLI. Their review covered the areas of verbal problems in general and to what extent this affects children with SLI. They reviewed the literature as regards the role of verbs in grammar, argument structure of verbs, and types of verb arguments and explained why verb difficulties can result in a marked effect on argument structure. They also reviewed the literature as to aspects of verb uses in spontaneous speech, and in individuals with agrammatic aphasia. Finally, they reviewed studies on argument structure development in children.

The authors were prompt in their criteria for the selection of participants and made every possible effort to minimize the effect of different variables. They also included an equal number of normal language children as a control group. However, the authors did not achieve an ethnic match between the two groups, which may predispose to some social impact on their results. Whereas the test group was composed of, white Americans (36/50), African Americans (9/50), three Asian, a Hispanic, and a native American, the control group was composed predominantly of white Americans (48/50) and only two African Americans.

The authors used the test model of Thompson et al, 1995 (after Thordardottir and Weismer, 2002) which Thompson and colleagues designed for individuals suffering from aphasia. Aphasia, by definition, is a speech disorder characterized by a defect in the ability of expression secondary to a brain disorder. This disorder affects the comprehension and expression of meaning and is a disorder of the use of symbols in speech. Specific language impairment is a mere language disorder not caused by developmental or neurological causes. This again is different from dysarthria, which is a disorder of the motor mechanisms of articulation (Brain and Walton, 1969). The choice of a test designed for aphasic patients may shadow the fitness of the test to normal children or to children with SLI especially when excluding all neurological and expressive language disorders as the authors did for the test participants.

According to the definition Thordardottir and Weismer, 2002 gave to verb argument, which is not without difficulty; they investigated the syntactic part of such an argument and did not look at the semantic part, which may not coincide with the syntactic part.

Reliability is to investigate how much the test results are dependable. It has two aspects, repeatability, and reproducibility. In the methods section, one speech-language pathologist certified SLI and one examiner collected narrative language samples and all other tests were the same for all children who participated in the work. Results are then expected to be repeatable; however, the authors did not go for any of the statistical tests for reliability as sum scale, Cronbach’s alpha, or split-half reliability. Thordardottir and Weismer, 2002 did not comment on testing the reproducibility of their results.

Although the authors investigated a statistically good number of participants (more than 30), they only examined statistical significance and did not provide additional statistical data as regards correlation or regression to elicit the relationship between different variables (dependable or not) included in the study. In addition, post hoc tests are many and designed to compare two means in order to find out if these two means are different. Of all post hoc tests, the Fischer LSD test, used by the authors, is the least recommended, as it is comparable to multiple t-tests (Everitt, 2006)

The author’s results covered their hypothesis questions as their results suggested that children with SLI did not significantly differ from children with normal language as regards error of argument structure types, especially when considering the sensible realistic (pragmatic) context. On the other hand, their results suggested that children with SLI used fewer argument structure types (statistically significant) and verb alternations for argument. As regards MLU (mean length of utterance), Thordardottir and Weismer, 2002, found a significant difference between both groups involving mainly multiple (three-place) argument structure.

An overlook on the results suggests that Children with SLI exhibit, generally, acceptably accurate yet less complicated verb argument structure and this is not because of utterance length (as a production limitation). They suggested that this might be attributed to incomplete argument structure representation.

Thordardottir and Weismer, 2002, interpreted their results by comparison to previous studies. Their finding that argument omissions include mainly the subject argument came in agreement with previous studies except for those studies, which investigated only complement omissions. This study is different from some earlier studies in that it focused on school-age children, not preschool children who are younger in age and thus, have lower levels of language development. To compare their results on the variety of arguments and argument structures, the authors compared their results with those of Naigles et al, 1992 (after Thordardottir and Weismer, 2002).

The authors omitted at least four important studies (Oetting, 1999, Budwig, 1998, Joannise and Seidenberg, 1998 and Bastiaanse and Bol, 2001), which collectively address the grammatical aspects of verb retention and interpretation. The previous studies suggest that children with SLI may fail to get hold of age-appropriate language skills, despite normal development, because of either impairment in the inherent knowledge of grammar or impaired information processing that interferes with language learning. Thordardottir and Weismer, 2002, used a mean length of utterance (MLU) as a grouping variable and for assessment of complex argument structure. A recent study by Rice and colleagues, 2006 demonstrated that MLU is valid for both situations in the range of 3-10 years, which coincides with the age range of children investigated in the present study.

Although Thordardottir and Weismer, 2002, did not provide new methods of research and their results were mostly matching other previous results, yet the authors reviewed a reasonably good number of articles and provided research results comparing a statistically good sample of children with SLI to a control group of normal language children. This article also provided unresolved questions, which may be the basis of future research.

References

Thordardottir, E.T, and Weismer, S.E (2002). Verb argument structure weakness in specific language impairment in relation to age and utterance length. Clinical Linguistics & Phonetics, 16 (4), 233-250.

Pylkkanen, L. (2002). Introducing arguments. Unpublished doctoral dissertation, Massachusetts Institute of Technology.

Goldberg, A.E. (2002). Pragmatics and argument structure. In Horn, L. and Ward, G. (Ed.), Handbook of Pragmatics (Blackwell series of handbooks in linguistics). Oxford: Blackwell.

Bresnan, J. (October 12, 1995). Lexicality and argument structure. Paris: Syntax and Semantics conference.

Bowers, J (2004). Towards a Unified Theory of Argument Structure and Grammatical Function Changing Morphology. Workshop on Argument Structure. Tromso: University of Tromso.

Reinhart, T (2002). The Theta System – An Overview. Theoretical Linguistics, 28 (3), 229-290.

Reinhart, T (2003). Thematic Arity Operations and Parametric Variations. Web.

Black, C.A. (1999). A step-by-step introduction to the Government and Binding theory of syntax. Web.

Sag, I.A. and Wasow, T. (2001). Syntactic theory: A formal introduction. Stanford: CSLI Publications.

The Late Lord Brain and Walton, J.N (1969). Brain’s diseases of the nervous system. Oxford: The English Language Book Society and Oxford University Press.

Everitt, B.S. (2006). The Cambridge Dictionary of Statistics. Cambridge: Cambridge University Press.

Oetting, J.B. (1999). Children With SLI Use Argument Structure Cues to Learn Verbs. Journal of Speech, Language and Hearing Research, 42, 1261-1274.

Budwig, N (1998). How far does a construction grammar approach to argument structure take us in understanding children’s language development? J. Child Lang., 25, 443-484.

Joanisse, M.F. and Seidenberg, M.S. (1998). Specific language impairment: a deficit in grammar or processing? Trends in Cognitive Sciences, 2 (7), 240-247.

Bastiaanse, R. and Bol, G. (2001). Verb Inflection and Verb Diversity in Three Populations: Agrammatic Speakers, Normally Developing Children and Children With Specific Language Impairment (SLI). Brain and Language, 77, 274-282.

Rice, M.L., Redmond, S.M. and Hoffman, L. (2006). Mean Length of Utterance in Children With Specific Language Impairment and in Younger Control Children Shows Concurrent Validity and Stable and Parallel Growth Trajectories. Journal of Speech, Language and Hearing Research, 49, 793-808.