A Comparative Study of Inferential Comprehension Strategies Between A Language-Learing Disabled and Non Language-Learning Disabled Child

The aim of this study was to analyse and compare the inferential comprehension strategies used by a language-learning disabled and non language-learning disabled subject, as well as to investigate their performance on areas related to inferential comprehension. For these purposes, tests were either constructed or modified in order to allow for qualitative analysis of the subjects' responses and the strategies used. The non language-learning disabled subject was found to utilize efficient inferential strategies, suggestive of cognitive-linguistic integrity, whereas the language-learning disabled subject was found to use inefficient inferential strategies and to be deficient on several areas related to inferential comprehension. These findings are interpreted as being reflective of a breakdown in the interactional dynamics between cognition and language.

Inferential comprehension refers to the listener's ability to use his real world knowledge, in combination with linguistic information to go beyond the explicitly stated information.Thus, a person's ability to engage in inferential comprehension depends on active participation, as well as utilization of a wide variety of stored information, pragmatics and author-reader conventions (Thorndyke, 1976).Inferential ability is an important component of comprehension and communication.Furthermore, it is an important prerequisite for dealing with information in many academic tasks (Klein-Konigsberg, 1984).The principles of inferential ability are important in reading comprehension (Kail et al. 1977).Furthermore, language demands placed on students in the academic setting require them "to understand and follow the teacher's directions and to focus and derive main ideas from the teacher's lecture, to organize and store these facts for retrieval" (O'Connor and Eldredge, cited by Nelson, 1986).
1 Both cognitive and linguistic hypotheses have been proposed to explain inferential disability (Crais and Chapman, 1987;Ellis Weismer, 1985) but only recently has inferential comprehension become an area of interest in the field of language pathology and learning disabilities (Freston and Drew, cited by Crais and Chapman, 1987;Ellis Weismer, 1985;Klein-Konigsberg, 1984).Until recently, most studies of language-disabled children have focused primarily on syntax and morphology (Wiig and Semel, 1981).The recent shift toward the study of this area -previously the domain of psychological research (Bransford and Franks, 1971;Johnson, Bransford and Solomon, 1973;Paris and Carter, 1973;Paris and Lindauer, 1976) -holds significant theoretical and practical implications relevant to the field of language problems.These implications could have an impact on our current conceptualization of cognitive and linguistic processes that operate in various language disorders, and on the ensuing course of intervention.
Previous research on inferential ability has revealed some interesting trends.For example, it was noted that younger children lacked a deliberate strategic approach when engaging in inferential comprehension tasks (Paris and Lindauer, 1976).Younger children experienced difficulty integrating premises and manipulating linguistic information in memory, and their schemata were fragmented (Danner & Mathews, 1980).Older children were found to employ more deliberate metamemorial strategies to integrate premises (Paris & Lindauer, 1976).Their verbal representational skills were superior and their schemata richer and more comprehensive (Danner & Mathews, 1980).In the light of this developmental perspective it was hypothesized that perhaps language-learning disabled children would use earlier developing strategies which are less goal-orientated and task-appropriate.
There has been little research into deficient inferential abilities.As with the research on the normal inferential process, these studies were flawed by the lack of an adequate conceptual model or systematic delineation of the abilities examined.Despite the scarcity of research on inferential disability in the language-learning disabled population, some trends have emerged from the literature.For example, Snyder (cited by Ellis Weismer, 1985) noted that languagelearning disabled children made fewer inferences as do younger children, and also that their performance is characterized by difficulties with simultaneous analysis and synthesis of information.Some writers (Carroll, 1986), have suggested that the language-learning disabled children have poorly integrated schemata, whilst Crais and Chapman (1987) attributed inferential disability to poor verbal comprehension.
Recent literature on the nature of learning disabilities reflects an orientation which considers strategic inefficiency to be central to academic under-achievement (Reed & Hresko; Torgeson; cited by Wiig and Secord, 1985).Therefore, the primary aim of this study (Tombak, 1987) was to analyse and compare the inferential comprehension strategies used by a language-learning disabled and non languagelearning disabled child.A strategy-based approach was used as it was considered to be a valid method for qualitative investigation of linguistic cognitive processes.With this in mind, the writer (Tombak, 1987) conceptualised a model of inferential comprehension in which the inferential comprehension process may be viewed as an interaction between the text and what the listener brings with him to the comprehension process, resulting in the product or inference (see figure 1).

Yael Tombak and Glenda Shapiro
The contextual basis of inferential ability was considered in terms of the 'microstructure' (Carroll, 1986), and the linguistic factors which may affect comprehension.Listener variables were considered in terms of macrostructure and schematic knowledge.These refer to the listener's knowledge of the standard arrangement of information and generalized experiential knowledge respectively (Moates and Schumacher, 1980).Problem-solving strategies were also taken into account.Lastly, the final inference product was analysed in terms of inferential subject matter and the role of an interceding inference.
This model attempted to delineate the main factors involved in the inferential comprehension process.It also aimed to provide a method of systematic task analysis, in order to establish and interpretive basis for the subjects' inferential performance and the strategies used.

AIMS
The primary aim of this study was to analyse and compare the inferential comprehension strategies used by a languagelearning disabled and non language-learning disabled child.The specific aims of the study were to describe the different inferential strategies used by these two children in a systematic, qualitative manner and to investigate the influence of selected task factors on the strategies used.
SUBJECT DESCRIPTION SI was a 13.0 year old learning-disabled male.He was a standard 5 pupil at a remedial school which he had been attending for the past 2 years.On the Wechsler Intelligence

MICROSTRUCTURE
• TEXT ORGANIZATION   Wiig, 1980) 2. Story recall "The stork in the wheat" (Berry, 1969) 3. Narrative tasks:-Sequence story, Poster Picture and Perso-• nal Narratives (Westby, 1984) The test battery administered was considered under the following three categories:ja) Criterion tests: These were.tests utilized to assess the subject's fulfilment of preselection criteria.b) Correlation tests: These tasks were used to evaluate the subjects' performance in areas related to inferential comprehension.. c) Main test: These constituted the primary tests of inferential comprehension.

Tests of Auditory Memory
Clinical Evaluation of Language Functions (C.E.L.F., Semel and Wiig, 1980) Subtest 6 -Processing Spoken Paragraphs.
Story recall (Berry, 1969) A test story, "The stork in the wheat", was used to measure retention and recall of salient sequential information in a story.
Expressive Language Tasks (Westby, 1984) Expressive language was evaluated using three narrative language tasks: a sequence story narrative, poster picture narrative and personal narrative.Responses to the stimuli were evaluated in terms of 'spontaneous inferencing', or spontaneously generated inferential elaboration.

b) CORRELATION TESTS
Adequate grammatic comprehension, receptive vocabulary and schematic knowledge are considered to be critical for linguistic comprehension.These areas were tested to examine the relationship between these factors and inferential ability.This subtest evaluated the ability to make causal inferences based on existing event chains in which one or more causal links were missing.Where appropriate, the subjects were probed for linguistic explanation, justification and elaboration of their responses.

Inferential Comprehension Test 1 (see Appendix 2)
This was the first of two tasks devised by the experimenter, specifically for this study, in order to evaluate inferential strategies.Items consisted of sentence pairs followed by a question, and were organized according to the conceptual model outlined earlier (see figure 1).The inferential abilities required for these tasks were primarily based on real world knowlege.The tasks were linguistically simple to minimize the influence of linguistic factors.

Inferential Comprehension Test 2 (see Appendix 3)
This test used a paragraph format and was more reliant on linguistic factors for comprehension.Each of the factors considered was represented by a pair of thematically similiar paragraphs, one of which was systematically varied with regard to that factor.Each paragraph was followed by questions and probing.

CRITERION AND CORRELATION TESTS
These were scored according to test manual instructions, or specially devised scoring systems for each test, which were constructed for the purpose of the study.

MAIN TESTS Evaluation of Spontaneous Inferencing
The number of spontaneous inferences generated by the subjects was calculated and the nature of the inferences qualitatively described.Divergent shift analysis -This analysis considered the subjects' ability to make divergent conceptual shifts.

Test'of Language
Congruency analysis -This analysis considered the subjects' attention to semantic congruency between response alternatives.

Test of Inferential Comprehension 1
The responses were qualitatively analysed in terms of strategies.

Test of Inferential Comprehension 2
The paragraph pairs were analysed in terms of strategy patterns, divergence, and the effect of factor variation on performance.

RESULTS AND DISCUSSION
The subjects' performance on the criterion, correlation and main tests were analyzed quantitatively and qualitatively, and compared.Their responses on the main tests were considered to reflect the inferential strategies utilized.

a) CRITERION TESTS
Both subjects scored above the revised pass-fail criterion of 14 on the C.E.L.F.-Subtest 6 (Semel & Wiig, 1980) indicating the presence of adequate auditory memory for paragraph material.On the story memory task (Berry, 1969), both subjects scored above 70%, calculated using a scoring system devised by the examiner (Tombak, 1987).The scores were interpreted as reflective of adequate auditory memory for complex'story material.Both subjects fulfilled the criterion of adequate verbal expression on the narrative task battery.However, although both subjects were able to verbally express their ideas in a clear manner, their expressive abilities differed along the dimensions of spontaneous inferencing and coherence in favour of S2.
b) Correlation Tests (see

) MAIN TESTS Evaluation of Spontaneous Inferencing
There was a marked difference between the subjects' spontaneous inferencing abilities, both quantitatively and qualitatively.
Spontaneous inferencing was measured quantitatively in units, each unit being an inferential proposition or item of information.This scoring system was developed for the purpose of the study (Tombak, 1987).Unit scoring was carried out by the examiner.SI generated a total of 4 units (picture Narrative -2 units, Poster Narrative -2 units), in contrast to a total of 30 units produced by S2 (Picture Narrative -12 units; Poster Narrative -18 units).
Si's narratives lacked elaboration, and were concrete and highly stimulus-bound, and there was inadequate thematic content.S2's inferences were complex narrative sequences elaborating on preceding and consequent events, motivations and internal states.

Test of language competence
Quantitative analysis:-SI experienced great difficulty with inferential reasoning, and selected a low proportion (43%) of correct inferential responses, while S2 selected a high proportion (87%) of correct responses.
Qualitative analysis:-This analysis was devised to compare and contrast the nature of the deficient strategies used by the 2 subjects.As stated earlier, the responses were organized by the examiner into 12 categories of deficient strategies (see figure 2).
The first ten of these were grouped into four strategy clusters, designated by the letters (A) -(D).Each cluster corresponded to a number of ineffective ways of implementing a constructive strategy.This category occurred frequently in Si's responses as opposed to S2's, where no instances were noted.SI relied heavily on linguistic context, reflecting deficient hypothetical reasoning.In contrast, S2's responses were hypotheticodeductive.

Conceptual syncretism
Conceptual syncretism is a Piagetian concept which refers to a tendency found in preoperational and concrete operational thought -to juxtapose rather than synthesise logical and causal relations (Piaget, 1927).A syncretic response is characterised by an indiscriminate linkage of two verbal statements with no consideration of their conceptual relationship.
eg. T.O.L.C Trial Item (SI) E: They had to go to eat at a restaurant because... S: They had a turkey at home and they wanted to eat at a restaurant.
This strategy occurred exclusively in Si's responses.Due to Si's inability to go beyond th€ linguistic context into the realm of hypotheses, he adopted the strategy of randomly juxtaposing any two salient concepts presented simultaneously, and forming a syncretic connection.
S2's reasoning, in contrast, was highly divergent and hypothetical.Si's deficit was characteristic of preoperational difficulty with constructing relationships that reflect attention to and simultaneous retention of all critical relevant information.Previous investigations have also documented difficulties in simultaneous analysis and synthesis of information in learning-disabled children (Crais and Chapman, 1987;Gerber, 1981;Klein-Konigsberg, 1984).

Lexical syncretism
This term refers to the tendency to juxtapose any two lexical items indiscriminately, with no apparent conceptual basis for this fusion.This tendency was a unique feature observed in Si's responses, and appeared to be a primitive, and possibly deviant manifestation of conceptual syncretism, which resulted in distortion of the relationships depicted.SI resorted to this preoperational strategy when he experienced a comprehension breakdown.A similar finding was also reported by Klein-Konigsberg (1984) who found that language-learning disabled children tended to attend to smaller sentence constituents when applying integrational strategies.

Psychological explanation
This refers to the preoperational tendency to provide a psychological explanation when a casual event explanation would be more appropriate (Piaget, 1927).Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012) Comparative Study of Inferential Comprehension Strategies instances when he did attempt to diverge beyond the linuistic context but it did not occur in S2's responses.This strategy was used when the linguistic context was not fully understood, due to fragmented lexical schemata.Poor assimilation and accommodation strategies may account for such deficient schemata.

Literal comprehension difficulty
This refers to difficulty in comprehending the basic relationships depicted, due to vocabulary deficits.eg.T.O.L.C. Trial item (SI) E: Do you have any idea why ... they weren't able to eat at home?S: Because the house was trimming.
Literal comprehension deficits were rare in Si's responses and absent from S2's responses on this test.
Strategy cluster Β (4, 5, 6) represented deficits in linguistic convergence.SI failed to perceive several lexical implications and their interactions simultaneously, reflecting a preoperational tendency to attend to only one feature or property at a time (Wiig & Semel, 1984).S2, on the other hand was able to attend to the full scope of the linguistic information presented, resulting in "lawful divergence" (Phillips, 1971).Difficulties in simultaneous analysis and synthesis have been noted in language-learning disabled children (Crais and Chapman, 1987;Ellis Weismer, 1985;Klein-Konigsberg, 1984) as well as younger normal children (Paris and Lindauer, 1976).

C. Strategy cluster C
This cluster was related to ineffective utilization of real world knowledge.

,,•Illogical inference in terms of real world knowledge
This feature refers to the formulation of inferences which are implausible in terms of real world knowledge.eg.T.O.L.C. Item 5 (SI) !E: They talked to a policeman because ... S: ... Because they rode on a crowded bus.
Only SI formulated inferences that were implausible in terms of real world knowledge.It seemed that SI had not acquired comprehensive schemata, as indicated by his performance on the Schema Test.His schemata appeared to lack refinement, and resisted absorption of new information.This deficit could also be related to inadequate accessing as well as induction of schemata (Carroll, 1986).

Non-optimal inference in terms of real world knowledge
This feature is characterised by the formulation of plausible, but unlikely inferences due to inefficient utilization of real world knowledge.SI tended to make illogical inferences, in contrast to S2 whose inferences were always plausible, although not always probable.
Strategy cluster C (7, 8) was concerned with the utilization of schematic knowledge in generating logical inferences.Fragmented schemata result in implausible or non-optimal inferences.This strategic failure could be caused by poorly induced schemata or failure to access the appropriate schemata timeously.

D. Strategy cluster D
This strategy was related to causal reasoning.

Inversed cause-effect reasoning
This feature refers to reasoning which is characterised by inadequate distinction between and sequencing of cause and effect events.Inversed causal reasoning is characteristic of preoperational and concrete operational thinking and is associated with egocentrism and a tendency towards syncretic perception (Phillips, 1971).The expression of the causal relationship was inversed by SI and an effect was formulated instead of a cause.Si's responses were characterised by deficient causal reasoning as his concept of 'because' was associated with a sequential meaning, rather than a true concept of causality.S2's reasoning was characterised by explicit delineation of causeeffect relationships and he was able to distinguish clearly between cause and effect.E: Why do you think that's a good answer?S: Because he didn't leave the waiter a tip.E: and c), 'The food and service were excellent?S: Yes, the food and service were excellent.
Egocentric reasoning was a common feature of Si's reasoning while only one instance was observed in S2's responses.Egocentric explanations are characteristic of preoperational reasoning, and are caused by a lack of concern for the listener's perspective (Piaget, 1923).

Die Suid-Afrikaanse Tydskrif vir Kommunikasieafwykings, Vol. 35, 1988
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012) Strategy cluster D (9, 10) was concerned with the role of causal reasoning in inferencing.Si's reasoning was characterised by poor comprehension of cause-effect relationships and egocentrism, reflective of a cognitive developmental lag (Phillips, 1971).S2 was able to engage in formal-operational explicit causal reasoning.

Interceding inference
When a final inference relies on an intermediate step or inference, the task is complicated since a faulty interceding inference may lead to the induction of a faulty final inference.SI consistently offered a paraphrase of previous explanations as an alternative inference, retaining the same conceptual basis for all interpretations (Wiig & Secord, 1985).S2's responses were characterised by alternative interpretations and a more divergent orientation.Si's responses reflected deficits in divergent conceptual shifting, resulting in overreliance on linguistic context.This finding was in agreement with that of Snyder (cited by Ellis Weisman, 1985) who found that language-disordered children made fewer inferences than age peers.Difficulty with conceptual shifting has also been documented in the learning disabled population (Wiig & Semel, 1976).
The difference in divergence abilities appeared to be diagnostically significant.This deficit appeared to be central to Si's inferential disability.

Divergent shift analysis
Sl'.s responses were characterised by few conceptual shifts (Total = 3 conceptual shifts) and a low proportion (30%) of matching inferences.S2's responses were divergent, with a high level of conceptual shifting (Total = 10 conceptual shifts) and a high proportion of matching inferences (90%).(2) Because his uncle let him ride it.
Number of matching inferences in above example = 0/2.Neither of the above inferences matched the correct responses listed for item 6 (b, d).
eg. Good conceptual shifting Item 1 (Subject 2) E: Jack didn't leave a tip because -S: (1) Because the restaurant wasn't good.The food was bad and the service was bad."" (2) He never had any small change on him.
Number of matching inferences in above example = 2/2.These 2 inferences above matched the correct responses listed for item 1 (b, d).

Congruency analysis
Si tended to ignore congruency between response alternatives, reflecting difficulty with simultaneous analysis and synthesis, suggesting a cognitive lag.This difficulty has been noted in the language-learning disabled population (Crais and Chapman, 1987;Ellis Weismer, 1985), as well as in younger normal children (Paris and Lindauer, 1976).

TEST OF INFERENTIAL COMPREHENSION 1
Both subjects performed adequately on this test, which relied primarily on schematic knowledge, rather than linguistic context.However, S2's responses were more divergent, and contained more conceptual shifting.

TEST OF INFERENTIAL COMPREHENSION 2
This test was analysed in terms of the effect of factor variation on inferential ability (see table 3).It was found that the factors or forced reliance on schematic knowledge, reversible embedding and passivization, and unclear intersentential relationships, had an adverse effect on only Si's performance.Both subjects' performance was adversely affected by complex vocabulary level and the presence of an interceding inference.However, Si's problems resulted in more extensive utilization of the deficient strategies described in the T.O.L.C. analysis.Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012) Non-reversible passivization and embedding had no effect on the subjects' comprehension, since the paragraph content was highly predictable due to the presence of semantic constraints.
Si's breakdown in inferential reasoning, precipitated by some of the factors listed above, paralleled his inferior results on the correlation tests.This confirmed previous findings of such deficits in the learning-disabled population (Wiig & Semel, 1976;1984 With regard to the interceding inference factor, SI made an incorrect interceding and final inference.S2's deduction of the interceding inference was influenced by residual preoperational difficulty with distinguishing real world from imaginary events (Phillips, 1971).The relationship between imagination and inferential ability has not been addressed in the literature, but may be of interst for further investigation.

CONCLUSIONS
The findings of this study suggested that the nature of learning-disability could be conceptualised in terms of a verbalcognitive interactional model.The study of inferential abili-ty^permits generalizations about verbal and cognitive skills, since both verbal and cognitive knowledge are required to engage in inferential reasoning.The main conclusion of this study was that the performance discrepancies documented, pointed to deficiencies in verbal-cognitive ability or 'verbal thought' (Vygotsky, 1962) (see figure 3).The language-learning disabled child presented with:--Cognitive deficits that interfered with the growth of formal verbal reasoning and the acquisition of higher level linguistic concepts.This would be in agreement with Bryen (1981) who asserted that "... this linguistic delay is not caused by inability to acquire certain linguistic symbols, but rather by reduction or delays in the acquisition of logical structures that determine their meaning".
-Linguistic deficits that possibly interfered with the development of higher levels of logical thinking and conceptual thought, as asserted by Bruner (cited by Lerner, 1976), andPiaget (1967).
However, the exact interactional dynamics between cognition and language in learning disabilities is still a topic of controversy as reflected in the following conclusion:-"The degree to which language delay contributes to or is a function of ... cognitive delay must still be determined" (Gerber, 1981).However, it must be noted that a limitation of the study was the use of only 2 subjects which restricted a generalisability of the study.The study should be replicated on a larger sample in order to verify results.In addition, future research should address the issue of subjective evaluation by making use of more than one rater, and ensuring inter-rater reliability." This study highlighted the fact that comprehension is a multi-faceted process and that various aspects of this ability should be considered in therapy.The results of this study, when considered within the theoretical framework proposed earlier, suggest that therapeutic intervention could be systematically organized to focus on various components considered by the model of inferential ability described.This study outlined certain areas of particular importance for therapy with a patient with inferential deficits and strategic breakdown.Therapy should consider such areas as schematic induction and accessing, a higher level linguistic deficits.Therapy could focus on the comprehension of causal relations and other connective relationships.Therapy should also stress comprehension of anaphoric devices and memory strategies (Gerber, 1981;Klein-Konigsberg, 1984).Therapy should emphasise lawful divergence and formulation of plausible hypotheses.Finally, the therapist should stress the importance of the listener's perspective and encourage explicit reasoning and an analytical orientation.
This study generated some implications for further research.Developmental studies of inferential abilities in both normal and learning-disabled populations, are needed.Other factors of the inferential comprehension model outlined earlier, may be investigated, such as the influence of level of abstraction, text organization or other categories of inference, on inferential strategies.The relationship between inferential ability and imagination has not been addressed in the literature, and constitutes an interesting area for investigation.The relationship between inferential comprehension and academic performance also needs to be considered in more depth, and the relationship between cognitive and linguistic deficits in learning disability warrants more attention.Finally the construction of assessment tools for evaluating inferential comprehension skills, is required.
The study of inferential ability holds interesting theoretical and practical implications for both speech therapists, educationalists and researchers.The continued investigation of this area is important, in view of its potential to clarify many controversial issues related to language pathology.Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012)

Figure 1 :
Figure 1: Model of inferential comprehension strategies -The T.O.L.C. responses were analised qualitatively by the examiner.The responses were organized by the examiner into 12 major categories of deficient strategies (see Results Section).

STRATEGY
inference i.t.o.Real world knowledge 8. Non-optimal inference i.t.o.Real world knowledge STRATEGY CLUSTER B: Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012) A. Strategy cluster A This cluster was related to difficulty with divergent thinking.1.1 Concretism Concretism refers to the inability to go beyond the linguistic context presented.A concrete answer lacks hypotheticodeductive reasoning.eg.T.O.L.C Item 5 (SI) (E = Examiner; S = Subject).E: They talked to a policeman because ... S: Because Bob had bad luck.
eg. T.O.L.C. Trial item (SI) E: Do you have any idea why ... they weren 't able to eat at home?S: -Because the house was trimming eg. T.O.L.C. Item 1 (SI)9 E: Jack didn't leave a tip because ... S: He didn't feel like it ...He was unkind.
eg. T.O.L.C. Item 5 (SI) E: They talked to a policeman because ... S:(selects d:) Bob lost his money sometime before they got to the mall... because they couldn't pay for the bus.
10. Egocentric reasoning Egocentric explanations are characterised by presupposition of the listener's knowledge of the speaker's internal reasoning process.Egocentric reasoning is a characteristic of preoperational thought.eg.T.O.L.C. Item 1 (SI) E: He didn't leave a tip because ... S: (selects a) The restaurant closed when he arrived.
eg. Poor conceptual shifting Item 6 (Subject 1) E: Eric was grateful to his Uncle Fred because ... S: (1) Because his uncle could let him use the bike ...
. = No effect on performance (-) = Adverse effect on performance The South African Journal of Communication Disorders, Vol.35, 1988 Figure 3: Interactional model of verbal-cognitive dynamics

Table 2 )
Si's performance was below average on the following tasks: ι -

T.O.A.L. -Subtest 2 (Hammill et al. 1980) I -Embedding Test I -Test of Linguistic Concepts -Passive Concept (Wiig &
The South African Journal of Communication Disorders, Vol.35, 1988Reproduced by Sabinet Gateway under licence granted by the Publisher(dated 2012)