contentserver-6

15
J Psycholinguist Res (2014) 43:45–58 DOI 10.1007/s10936-013-9238-6 Pragmatic Language Development in Language Impaired and Typically Developing Children: Incorrect Answers in Context Nuala Ryder · Eeva Leinonen Published online: 14 February 2013 © Springer Science+Business Media New York 2013 Abstract This study focussed on young children’s incorrect answers to pragmatically demanding questions. Children with specific language impairment (SLI), including a sub- group with pragmatic language difficulties (PLD) and typically developing children answered questions targeting implicatures, based on a storybook and short verbal scenarios. Ninety- seven children participated in this study: 30 children with SLI of whom 12 had PLD, 32 typically developing children aged 5–6 years and 35 aged 7–11 years. The incorrect answers produced by the children with SLI were similar in their use of context to those of the 5–6 year old, suggesting developmental delay. The children with PLD produced significantly more irrelevant answers than both the language impaired children without PLD and the typically developing groups and had most difficulty when the context was presented solely verbally. Results are discussed in relation to a cognitive theory of communication and the clinical implications. Keywords Specific language impairment · Pragmatic language impairment · Incorrect answers · Implicatures · Relevance theory Introduction Answering questions about a given content is prevalent in the lives of children. The ability to create meaning on the basis of the integration of the relevant information from the available context, conceptual information, and previously acquired information (knowledge) is seen to develop between the ages of 3 and 6 years (Ryder and Graves 1998; Winer et al. 2001; Sperber and Wilson 2002). Around the age of 3years, children’s incorrect answers to questions are N. Ryder (B ) Department of Psychology, University of Hertfordshire, College Lane, Hatfield, Herts AL10 9AB, UK e-mail: [email protected] E. Leinonen School of Psychology, The University of Wollongong, Northfields Ave, Wollongong, NSW 2522, Australia e-mail: [email protected] 123

Upload: felipe-barbosa

Post on 17-Sep-2015

214 views

Category:

Documents


2 download

TRANSCRIPT

  • J Psycholinguist Res (2014) 43:4558DOI 10.1007/s10936-013-9238-6

    Pragmatic Language Development in Language Impairedand Typically Developing Children: Incorrect Answersin Context

    Nuala Ryder Eeva Leinonen

    Published online: 14 February 2013 Springer Science+Business Media New York 2013

    Abstract This study focussed on young childrens incorrect answers to pragmaticallydemanding questions. Children with specific language impairment (SLI), including a sub-group with pragmatic language difficulties (PLD) and typically developing children answeredquestions targeting implicatures, based on a storybook and short verbal scenarios. Ninety-seven children participated in this study: 30 children with SLI of whom 12 had PLD, 32typically developing children aged 56 years and 35 aged 711 years. The incorrect answersproduced by the children with SLI were similar in their use of context to those of the 56 yearold, suggesting developmental delay. The children with PLD produced significantly moreirrelevant answers than both the language impaired children without PLD and the typicallydeveloping groups and had most difficulty when the context was presented solely verbally.Results are discussed in relation to a cognitive theory of communication and the clinicalimplications.

    Keywords Specific language impairment Pragmatic language impairment Incorrect answers Implicatures Relevance theory

    Introduction

    Answering questions about a given content is prevalent in the lives of children. The ability tocreate meaning on the basis of the integration of the relevant information from the availablecontext, conceptual information, and previously acquired information (knowledge) is seen todevelop between the ages of 3 and 6 years (Ryder and Graves 1998; Winer et al. 2001; Sperberand Wilson 2002). Around the age of 3 years, childrens incorrect answers to questions are

    N. Ryder (B)Department of Psychology, University of Hertfordshire, College Lane, Hatfield, Herts AL10 9AB, UKe-mail: [email protected]

    E. LeinonenSchool of Psychology, The University of Wollongong, Northfields Ave, Wollongong, NSW 2522, Australiae-mail: [email protected]

    123

  • 46 J Psycholinguist Res (2014) 43:4558

    suggested to reflect strategies such as preferentially using knowledge from memory, guessingoutcomes or answering on the basis of a key word (Paul 1990; Ryder and Leinonen 2003). Forsome children i.e. children with specific language impairment (SLI) or children with primarilypragmatic language difficulties (PLD), the use of such compensatory strategies continues forlonger suggesting an impaired ability in integrating relevant information in a given context(Bishop and Adams 1992; Paul 1990; Bishop 2000; Bishop et al. 2000; Leinonen et al. 2003;Botting and Adams 2005). The theoretical grounding of these strategies is important in theclinical setting and Relevance Theory (Sperber and Wilson 1995) provides a framework inwhich to explain this development in typically developing children, and children with SLIwith and without pragmatic difficulties.

    Children with SLI have a delayed developmental language profile in the absence of anyexplanation (normal cognitive, neurological and sensory functioning). On verbal languagetasks, 711 year old children with SLI perform similarly to 4 year old children who are stilldeveloping the ability to integrate information and process implicatures (the outcome of thisintegration). Implicatures from a Relevance Theory perspective are not inferences, but theoutcome of combining information from context (i.e. perceptual information, knowledgein memory including conceptual knowledge, knowledge of previously generated inferencesand experiential knowledge). Both groups have been found to be competent at inferringsemantic and referential meaning (Leinonen et al. 2003; Bishop and Adams 1992; Botting andAdams 2005). Children diagnosed with SLI comprise a heterogeneous group within whichare a sub-group of children whose primary difficulties are in pragmatic areas of languagereferred to as Pragmatic Language Impairment (PLI). There is much debate over the diagnosisof this impairment but it is generally held that these children speak in fluent and well-formed utterances with adequate articulation but have pragmatic language difficulties (PLD)characterised by over-literal interpretation, the odd content and language use, and may havedifficulty maintaining a topic or be poor at turn-taking. The responses of these children onverbal comprehension tasks suggest an impaired ability in understanding meaning in contextor difficulty in integrating relevant information (Bishop and Adams 1992; Ryder et al. 2008).

    In typically developing 2 and 3 year old children a preference or reliance on world knowl-edge/experience is evident in their responses. Paul (1990) for example, found that wherechildren may have only partial understanding of principal semantic and/or syntactic struc-tures of scenarios and questions in language tasks, they use their world knowledge/experienceor a probable reasoning strategy in preference to the given information. World knowledgeincludes schemas, personal knowledge/experiences and knowledge about the specific com-municative situation. Three and four year old children have similarly been found to use theirworld knowledge in preference to the meaning of the linguistic expression (Strohner andNelson 1974; Marinac and Ozanne 1999) or the speakers intended meaning (Robinson andWhittaker 1987) when answering questions. Explanations of these strategies have includedsuggestions that childrens incorrect answers reflect a literal interpretation of the questionsor discourse (Marinac and Ozanne 1999; Winer et al. 2001) and a lack of abstract think-ing. Winer et al. (2001) suggested abstract thinking facilitates the ability to think about theintention of the speaker in asking a question and that this ability develops between the agesof 3 and 4 years. Bishop et al. (2000) also suggested that children with pragmatic languageimpairment do not appear to understand the speakers intention, but further suggest that theirliteral interpretations are due to an inability to utilise the available context effectively. Theyfound that the childrens answers were irrelevant in the given context and suggested the chil-dren had a lack of understanding of the speakers intention. However, intentions were notspecifically investigated and are difficult to empirically examine. An alternative explanationfrom Tomasello (1999) and Nelson (2006:185) suggests that during a childs development

    123

  • J Psycholinguist Res (2014) 43:4558 47

    they come to understand that meaning can only be interpreted in the context of particularuses and in the early years gradually become more competent at utilising the relevant contextin interpretation.

    This use of relevant contextual information has also been investigated in studies of readingability as it is a necessary skill in reading comprehension. Older children (78 years old) withvaried reading abilities, have been found to have difficulties in integrating information fromthe text with relevant information from their knowledge base (knowledge which had beenlearned and understood before reading the text) especially if the text was perceived as difficult(Perrusi et al. (2005). Non-skilled comprehenders were found to have particular difficultiesin selecting relevant information from the text.

    From a relevance theory viewpoint (Sperber and Wilson 1995), it is metarepresenta-tional development which facilitates the ability to utilise relevant context and integrate thiswith conceptual or previous information from memory. They argue humans have an inher-ent ability to inference and this is guided by a cognitive system which is geared towardsrelevance and what develops is the ability to process implicatures (the integration of rel-evant information) which occurs gradually as the child interacts with others in the earlyyears. Implicatures are defined as being derived from integrated implicit premises (combin-ing of conceptual information from memory, perceptual information, previously generatedinferences and so on). They are meanings which a speaker conveys, implies or suggestswithout literally expressing them. On this view, as the development of the ability to processimplicatures proceeds children are able to combine the relevant information available tothem, in order to reach a conclusion or provide an answer. This processing may under-lie the ability to consider the competence of the speaker and their intention in asking thequestion (though it is debatable whether recognising and intention is a process/ability orthe result of the processing of implicatures). Childrens comprehension of meaning in con-text is therefore dependent on the processes of inferencing and deriving implicatures. Theexplicitness of the language expression and the available cues from the context governprocessing. So for example, questions targeting literal meaning may require inferencingbut not the recovery of implicatures. That is, they may require referents to be assignedand semantic meaning to be enriched but they do not require meanings to be combined toyield an implicated conclusion. For example, answering why questions, or understand-ing indirect answers to questions requires going beyond the inferences that can be gener-ated on the basis of the words alone. On this account, when children are developing theability to process implicatures they may preferentially infer meaning when interpretingquestions which require the recovery of implicatures. What is not clear from research iswhether childrens incorrect answers to questions reflect a reliance or preference for infer-ring meaning (i.e. semantic meaning or knowledge from memory) and further, whether theiranswers reflect difficulty in processing implicatures or whether erroneous implicatures aregenerated.

    Relevance Theory also suggests the available context should be considered in interpre-tation. Children with SLI were found to perform similarly to their peers when answeringquestions targeting implicatures on a task where they point to the pictorially given answer.Their performance was also found to improve if a verbally presented scenario was accom-panied by gestures and when giving verbal answers to questions based on storybooks withpictorial support (but answers not available from the pictures) and when gestures accompa-nied verbal presentation of short scenarios. However, where the context was solely verbal(short scenarios followed by a question) and no pictures or gestures were available, childrenwith SLI performed significantly less well (Ryder et al. 2008; Kirk et al. 2011). Taken togetherthese studies suggest symbolic gestures and pictures act as external cues which are then more

    123

  • 48 J Psycholinguist Res (2014) 43:4558

    readily combined with the verbal information distributing the pragmatic load (Ryder et al.2008).

    On the basis of relevance theory, questions can be constructed to target the use of contextuse and childrens incorrect answers can be examined in terms of whether they reflects infer-ring semantic meaning using world knowledge, or whether they reflect integrating knowledgeand given information from context. Childrens incorrect answers are expected to reflect adevelopmental trend in utilising relevant context to understand the focus of the question.On the basis of previous research suggesting 711 year old children with SLI are delayed inlanguage development, we expect their answer strategies to be similar to the 56 year oldtypically developing children. We hypothesis that the children with SLI and the youngestchildrens answers will rely on inferring meaning when answering implicature questionswhereas the 711 year old children are expected to utilise the relevant story or scenario infor-mation. We also explore whether the answers given by children with primarily pragmaticlanguage difficulties reflect an inability or difficulty with implicatures.

    In this study we examine childrens incorrect answers to questions based on short scenar-ios and a short story. We predict that in both tasks incorrect answers will reflect a relianceon world knowledge, inferring meaning on the basis of the words in the scenario and thequestion. In the storybook task we additionally expect incorrect answers to reflect the utilisa-tion of previously generated inferences based on the story. The incorrect answers of childrenwith pragmatic difficulties and the youngest children are expected to be based on worldknowledge which has not been appropriately integrated with the scenario information. Morespecifically:

    1. Do childrens incorrect answers reflect an inability to integrate relevant information fromthe scenario when answering pragmatically demanding questions?

    2. Do the incorrect answer strategies of children with SLI reflect developmental delay (withsimilar answers to those of younger typically developing children)?

    3. Within the SLI group, do the incorrect answers strategies of children with pragmatic lan-guage difficulties reflect more specific difficulties in utilising relevant context comparedto children without pragmatic difficulties?

    Method

    Participants

    A total of 97 children participated in this study (see Table 1). There were 30 children withSLI aged between the ages of 7 and 11 years of age. Within this SLI group were 12 childrenwith primarily pragmatic language difficulties (PLD). There were 67 typically developingmonolingual children with no known language difficulties from similar socio-economic back-grounds. There were 2 age groups: 32 children aged between 5;0 and 6;0 and 35 childrenaged between 7;0 and 11;0. All the children attended one of two schools in the same town.The children with SLI had a diagnosis of SLI on the basis of no evidence of neurological,sensori-motor, non-verbal cognitive or social emotional deficits and a score of 1SD belowthe norm on at least one standard language test. They attended speech and language therapyunits within the schools, receiving therapy for their language difficulties. The children withpragmatic language difficulties were identified from the rest of the children with SLI (with-out pragmatic language difficulties) on the basis of their performance on standard languageassessments as part of their speech and language therapy and on the basis of their performance

    123

  • J Psycholinguist Res (2014) 43:4558 49

    Table 1 Age, gender, numberand grouping of participants Group N Age range Average age Gender

    M F

    SLI 30 6;811;3 8.7 19 115-6 yrs 32 5;26;11 5.7 13 19711 years 35 7;111;2 8.6 13 22Total 97 45 52

    Table 2 Language assessment scores for the children with SLI sub groups

    SLI (n = 30) Subgroups of SLI childrenNo pragmaticlanguage difficulties(SLI) (n = 18)

    With pragmaticlanguage difficulties(PLD) (n = 12)

    Age M(SD) 8.7 (1.23) 8.5 (1.10) 7.5 (1.18)Age range 7;211;3 6;8-10;01Male/female 9 / 9 11 / 1TROG percentile 40.8 (28.4) 42.0 (25.7) 38.9 (32.9)RAPT grammarZ score (SD) 0.01 (1.0) 0.15 (0.8) 0.02 (1/2)RAPT InformationZ score (SD) 0.03 (1.0) 0.29(1.0) 0.14 (0.9)TOWK receptiveM standard score 6.9 (2.4) 7.4 (2.6) 6.7 (2.1)TOWK expressiveM standard score 8.1 (2.7) 8.2 (2.1) 8.4 (3.3)ITPA auditory associationM standard score 29.8 (6.9) 31.5 (6.9) 27.5 (7.0)ITPA sequential memoryM standard score 23.6 (6.1) 25.3 (6.8) 21.5 (5.4)

    in answering pragmatically demanding questions targeting implicatures (in a previous studyby Ryder et al. 2008).

    Standard language tests were administered to the children with SLI. Inclusion in thestudy required a receptive grammar score of at least age five equivalent on the TROG (Testfor the Reception of Grammar version 2, Bishop 2003a). In addition expressive grammar(Renfrew Action Picture Test, Renfrew 1988), auditory memory for sequential numbers,auditory association (Illinois Test of Psycholinguistic Ability, Kirk et al. 1968) and expressiveand receptive vocabulary were assessed and in the normal range (Test of Word Knowledge,Wiig and Secord 1992). The researcher also tested understanding of vocabulary used in thestudy before tasks were undertaken to ensure the tasks were within the childs capabilities.The speech and language therapists working with the children completed the ChildrensCommunication Checklist (version 2, Bishop 2003b) which gives an indication of generalcommunicative abilities and social behaviour. Results are presented in Table 2.

    123

  • 50 J Psycholinguist Res (2014) 43:4558

    The SLTs were asked to identify the children (if any), with primarily pragmatic languagedifficulties (PLD) in their professional opinion. The SLTs made this judgement on the basisof their knowledge of the child and language and behavioural assessments carried out for thepurpose of managing the childs therapy. Thirty two children with SLI were tested, and twowere excluded on the basis of their CCC-2 profile (Childrens Communicative Checklist)which suggested autistic spectrum disorder. Twelve of the remaining 30 SLI children werediagnosed as PLD.

    Procedure and Task Design

    Each child was seen individually in a quiet corner of the classroom. Each task was explainedbefore commencing and practice items were given to ensure the child could succeed inthe task. Incorrect answers from each of the three tasks types were coded according to thecontext used in the answer. Two tasks were used with different contexts. The first was a verbalscenario including two short sentences read aloud followed by a question. No other contextualinformation was available. In line with previous studies focusing on understanding pragmaticlanguage expressions, short scenario rather than longer stories were used in to ensure theyoungest childrens focus of attention was engaged. The second task was a storybook taskwhich included pictures of the characters and the background scenes. The story repeatedcharacters names and the central theme. As in the verbal scenario condition, the story wasread aloud. Both tasks required combining the scenario/story information and generating animplicature to answer successfully.

    (1) Verbal Only (Verbal): Ten A5 cards had a number on one side and a verbal scenario onthe other. The cards were presented number side up and the child pointed to their chosencard. The researcher then read aloud the scenario on that card to the child and asked aquestion targeting implicature on the basis of the short scenario. The child was requiredto answer verbally. The answers require integrating the information from both sentencesand utilising world knowledge (see Appendix). The context is verbal only.

    (2) Storybook (Strybk): A short story was read aloud to the child and eight A4 colour picturesaccompanied the story but answers to questions were not in the pictures. The storybookwas based on a book published for the 35 year age group and sentences closely fol-lowed the same grammatical structure designed to be suitable for children aged 35 years(Waddell and Granstrm 1997). Seven implicature questions were asked intermittentlythroughout the story. The questions in the storybook were asked immediately after therelevant text was read, and were designed to keep interruption to the flow of the storyto a minimum. The questions were kept as grammatically simple as possible using thesame structures as the storybook text. This was to minimise memory load, and to ensurethat the youngest developing children had every chance of processing the grammaticalstructures involved. In addition all the SLI children scored above age 5 equivalent forreceptive grammar test. The child was required to give a verbal answer (see Appendix).

    In both tasks the children could not see any text. The wording of the questions closelyfollowed the scenario/storybook wording and one repetition was allowed. Both tasks werevideo-recorded.

    In this study we focus on childrens incorrect answers and how they utilise context whenattempting to answer pragmatically demanding questions. Childrens incorrect answers foreach task were therefore coded according to the context utilised in the answer. Five categorieswere apparent: 1. Story/scenario context, 2. World Knowledge/experience, 3. Irrelevant, 4.Dont know, 5. No response/tautological (see Appendix for examples). The data were

    123

  • J Psycholinguist Res (2014) 43:4558 51

    transcribed orthographically from videotaped sessions. Two researchers coded 12 childrens(from all groups) incorrect answers to 12 questions (chosen randomly) into one of the fiveincorrect answer categories. The percentage of agreement between the coders was 94 %.

    The childrens incorrect answers were analysed into five categories. Examples of eachanswer type are given in the Appendix:

    (1) Story/scenario context: There were answers where the child had utilised the storybookinformation or the scenario information but they did not focus on the essence of thequestion appropriately. The child shows evidence in the answer of comprehending thetheme of the story based on what they have heard. In Relevance theory terms, the childhas considered what is meant by the question and the comprehended information hasbeen utilised in the response made to the question. There might be some integration ofwhat has been explicitly said in the previous text and enrichment of the words in thequestion (see example in Appendix). The answer is not a repetition of text previouslyread out, but includes previous storybook context.

    (2) World Knowledge: The question has triggered the child to use world knowledge andexperience of similar situations but does not address the focus of the question. That is,they have utilised knowledge from memory based on the words in the text or the question.

    (3) Irrelevant: The child gives an answer which is not relevant in the given context and doesnot appear to be the result of world knowledge/experience or reflect an ability to considerthe focus of the question.

    (4) Dont know: The child says they dont know the answer.(5) No response/tautological: Either no response at all or child repeats words verbatim from

    question or scenario.

    Results

    The distributional features of the dependent variables (i.e. average percent of incorrectanswers) were examined for normality, and parametric or non-parametric statistical pro-cedures were employed accordingly. Overall, there were only a few children who gave noresponse (four children) or tautological answers (three children), so these answers were com-bined with the dont know answer category. A frequency analysis by subject revealed thatall the children were using different types of answers when responding to the questions onboth tasks. That is, each child did not use, or have a strong preference for, one answer type.For each child the number of incorrect answers for each incorrect answer type was convertedinto a percentage score. Table 3 shows the average percentages incorrect depending on age,incorrect answer type and task.

    Overall, for all groups, the verbal task elicited more incorrect answers than the storybooktask but the differences were non significant. Figure 1 shows the groups mean percentageof answers by incorrect answer type. We first sought to investigate whether the childrensanswers reflect an inability to integrate relevant information from context. Within groupcomparisons of incorrect answer types for each group were first examined by task followedby between group comparisons on each task.

    Within group paired incorrect answer comparisons for each task were examined usingWilcoxons signed ranks test with a Bonferroni correction (p = 0.0127). Comparisons ofincorrect answer types revealed that the children with SLI produced significantly more worldknowledge and irrelevant answers on the verbal task compared to dont know and scenarioanswers (World knowledge and Dont know Z = 4.40, p = 0.0001, Irrelevant and Dont know

    123

  • 52 J Psycholinguist Res (2014) 43:4558

    Table 3 Descriptive statistics for verbal and story tasks

    Group Verbal task Story task

    WK Irrel DK/NR Scenario WK Irrel DK/NR Story

    SLIM 20 21 2 1 12 13 4 8SD 13.8 20.0 4.6 2.3 14.9 14.5 9.3 12.256 yearsM 21 9 3 2 6 12 4 10SD 13.2 9.7 7.4 4.6 10.2 13.1 8.9 12.8711 yearsM 4 3 2 1 2 7 2 0SD 7.4 5.4 4.1 2.1 5.5 10.6 4.6 0.00

    Means represent the average percent incorrect. WK world knowledge, Irrel irrelevant, DK/NR dont know orno response

    Fig. 1 Incorrect answers of children with SLI with pragmatic language difficulties (PLD) and without prag-matic language difficulties (NPLD) by answer type and task

    Z = 3.95, p = 0.0001, World knowledge and Scenario Z = 4.41, p = 0.001 and Irrelevantand Scenario Z = 4.24, p = 0.001). See Fig. 1. On the storybook task, the children withSLI produced significantly more irrelevant answers than dont know answers (irrelevant anddont know Z = 2.59, p = 0.007) all other comparisons were insignificant (world knowledgeand dont know Z =2.12, p = 0.034, irrelevant and story Z = 1.22, p = 0.26 and worldknowledge and story Z = .734, p = 0.46).

    The 56 years old, on the Verbal task, produced significantly more world knowledgeanswers than the other answer types (world knowledge and irrelevant Z = 3.95, p = 0.0001,world knowledge and dont know Z = 4.27, p = 0.0001, world knowledge and scenario

    123

  • J Psycholinguist Res (2014) 43:4558 53

    Z = 4.48, p = 0.0001, irrelevant and dont know Z = 2.69, p = 0.007 and irrelevant andscenario Z = 3.02, p = 0.0003). On the story book task they produced less world knowledgeanswers but an increased number of answers using the story context. There were no significantdifferences between the incorrect answer types (world knowledge and irrelevant Z = 2.73,p = 0.08, world knowledge and dont know Z = 0.78, p = 0.43, world knowledge and storyZ = 1.66, p = 0.09, irrelevant and dont know Z = 2.22, p = 0.03, irrelevant and story Z =0.31, p = 0.75).

    The 711 year old group, produced significantly more answers based on world knowledgethan dont know answers (Z = 2.77, p = 0.006) on the verbal task, significantly more worldknowledge than scenario answers (Z = 2.94, p = 0.003) and more irrelevant than scenarioanswers (Z = 2.64, p = 0.008). On the storybook task the only significant finding wasmore irrelevant answers than dont know answers (Z = 2.94, p = 0.003) with a Bonferronicorrected p value of 0.016. There were no incorrect answers based on story context.

    To further investigate the strategies used by the groups and whether the strategies ofthe children with SLI were similar to younger typically developing children, comparisonsbetween groups on each task were made. Comparisons on the Verbal task revealed significantdifferences in the number of world knowledge (Kruksal Wallis 2 32.03(2), p = 0.0001) andirrelevant answers (Kruksal Wallis 2(2), p = 0.001). The performance of the children withSLI and the 711 age group (age matched peers) on the Verbal tasks were compared usingMann-Whitney U -tests with the critical p value set at 0.016 (Bonferroni adjustment). Thechildren with SLI produced significantly more answers based on world knowledge than the711 age group (Man U 182.0, p = 0.0001) as well as irrelevant answers (Man U 240.50,p = 0.01). Similarly, the 56 year old produced significantly more answers based on worldknowledge (Man U 167.00, p = 0.001) and irrelevant answers (Man U 346.00, p = 0.003)than the 711 year old. The pattern of responses of the children with SLI and the 56 yearold was similar except the children with SLI produced significantly more irrelevant answers(Man U 316.00, p = 0.016). There were no other significant differences in incorrect answertypes between these two groups.

    On the Storybook task, Kruksal Wallis comparisons revealed significant differencesbetween groups on answers based on world knowledge (Kruksal Wallis2(2), p = 0.005) andanswers utilising storybook information (Kruksal Wallis 2(2), p = 0.001). Man U compar-isons (with Bonferroni correction, p = 0.016) revealed that the children with SLI producesignificantly more answers based on world knowledge (Man U 395.50, p = 0.038) andanswers based on storybook information than the 711 year old (Man U 315.00, p = 0.001).Again the children with SLI and the 56 year olds performed similarly and no significant dif-ferences in the type of incorrect answers were found though there were more world knowledgeanswers produced by the children with SLI (Man U 3.777, p = 0.063).

    The third research question examined whether the incorrect answers given by the subgroup of SLI children with pragmatic language difficulties (PLD) would reveal differentstrategies than the SLI children without pragmatic difficulties. Twelve of the 30 childrenwere diagnosed as primarily pragmatically impaired by a speech and language therapist whotreated these children on a daily basis (and administered the language assessments), and therelevant descriptive statistics for SLI children with pragmatic language difficulties (PLD)and without pragmatic language difficulties (NPLD) are given in Table 2. Overall, on bothtasks, the children with PLD answered more questions incorrectly (see Fig. 1).

    Because of distributional anomalies and the small group sizes a Kruksal Wallis test wasused to compare incorrect answers on the two tasks. On the Story task, the PLD groupproduced more answers based on world knowledge (2(1), p = 0.007) and on the Ver-bal task they produced more irrelevant answers (2(1), p = 0.029). There were no other

    123

  • 54 J Psycholinguist Res (2014) 43:4558

    significant differences in answer types on the two tasks (Verbal task: world knowledge (2(1),p = 0.537; dont know (2(1), p = 0.674; scenario (2(1), p = 0.208); Story task: irrel-evant (2(1), p = 0.125), dont know (2(1), p = 0.208), story (2(1) p = 0.670). Thechildren with PLD relied on world knowledge preferentially rather than the use of storycontext. Where there was no supporting context the incorrect answers of these children wereirrelevant in the context of the verbal scenario on which the questions were based.

    In the analysis of the children with SLI compared to the 56 years old above, these groupsperformed similarly except the SLI group produced more irrelevant answers. Given that thechildren with PLD produced more of these answers than the NPLD children, we removedthe PLD children from the SLI group to examine whether the difference in the use ofirrelevant answers would remain. The irrelevant answers on verbal and story tasks of theNPLD group and 56 year old group were compared and revealed no significant differences(Mann-Whitney U Verbal task p = 0.911; Story task p = 0.263).Language Abilities and Performance of Children with SLI

    The data in this study are part of a data set used in a previous study focusing on childrenspragmatic language comprehension. The extent to which structural language abilities andauditory memory of the children with SLI were correlated with their performance on impli-cature questions was examined and it was found that though receptive grammar (TROG) wasmoderately related to performance, an analysis of covariance controlling for TROG scoresrevealed the lower means of the PLD children on the tasks were not due to their weakerperformance on the TROG (Ryder et al. 2008).

    Discussion

    This study examined whether childrens incorrect answers reflected an inability to integraterelevant information when answering pragmatically demanding questions. Results suggestthat children gradually develop the ability to use relevant context in interpretation. The num-ber of irrelevant answers reduced as age increased for the typically developing children.All the children attempted to provide an answer and there were few dont know answers.The younger children and the children with SLI tended to rely on world knowledge whenanswering questions. In Sperber and Wilsons view (2002) these children were consider-ing the first relevant interpretation that came to mind, without considering or integrating thegiven contextual information to reach a conclusion/answer. These findings along with studiesof younger children (Paul 1990), suggest that in answering pragmatically demanding ques-tions children utilise their world knowledge or experience, either inferring and enriching thesemantic meaning of a word in the question or recalling something they have knowledge ofon the basis of semantic meaning, which is familiar to the scenario or question being asked.The childrens answers reflect a developing ability to utilise the given context even if theyhave not quite understood the intended focus of the question.

    Irrelevant answers reflected an inability to integrate relevant contextual information.Examination of these answers revealed that the children appeared to make erroneous infer-ences, for example inferring the noise the ghost made to be the noise of an owl. Though thistype of answer did fulfil the discourse role in providing an answer, it reflects an inferencemade on the basis of a word or phrase alone rather than the recovery of an implicature onthe basis of the context. In the RT model, the hearer attends to the information and makesinferences on the basis of that information at the time of interpretation. A system geared

    123

  • J Psycholinguist Res (2014) 43:4558 55

    towards relevance ensures that the number of inferences which are potentially possible areconstrained by consideration of the relevant context. The owl interpretation implies that thechild relies on semantic inference without combining any contextual information. If the childdoes not make inferences on the basis of the context then as the story progresses the childwill not have previously generated relevant inferences available to integrate when attemptingto answer the question. This study supports the view of Sperber and Wilson that childrendevelop the ability to integrate information (and possibly to understand the communicativeintention of the speaker). Increased experience of language situations where implicaturesare recovered is in their view likely to result in more procedural or automatic processing.In providing an irrelevant answer, the child makes inferences based on world knowledge(something which is familiar in their experience) but which is irrelevant in the context ofthe story and question. Sperber and Wilson (2002) suggests that the child stops at the firstavailable interpretation rather than positing that the child has a difficulty in understanding(or attending to) the focus of the question. It is not clear from the model how the child moveson from this phase but the child is said to develop the ability to consider the intention of thespeaker as they develop.

    An emergentist view may provide an explanation for this developmental change. Anemergentist perspective would propose that childrens ability to combine different sourcesof information to process language is constrained by their experience of language use. Inearly development, changes in external contextual demands are said to result in shifts of thechilds attention from one attractor (i.e. content word) to another. This experience of shiftingattention is said to result in self-organised patterns and the stability of these vary over time. Forchildren with SLI, atypical language experience may incur instability of attention preventingcomprehension of the focus of the question. Irrelevant answers might therefore be seen as aresult of instability which may also lead to atypical organisation (Evans 2001).

    Lastly we explored whether within the SLI group the incorrect answer strategies of childrenwith pragmatic language difficulties reflect more specific difficulties in utilising relevantcontext compared to children without pragmatic difficulties. The most significant differencewas the increased number of irrelevant answers of the children with pragmatic languagedifficulties (PLD) compared to the children without pragmatic difficulties (NPLD) group.This finding on the Verbal task suggests that where these children have no other contextualcues except verbal information, this increased the level of difficulty. In the story task, irrelevantanswers were not so increased and the children relied more on their world knowledge (and toa much lesser extent story context) as was the pattern with typically developing children. Thepattern for all groups showed that in the verbal task the scenario information was not evidentin their incorrect answers. This may be because the scenarios were short (two sentences) andeach scenario was not related to the next. Whereas, in the story context, previously generatedinferences and literal interpretation provided knowledge which could be used in interpretation(albeit not always relevant).

    The children with PLD did however, use the same strategies as the other children with SLIand the youngest typically developing children, but they produced more irrelevant answersif the task was purely verbal. This may implicate a greater developmental delay in utilisingrelevant context in interpretation or recovering implicatures. Studies of task difficulty in falsebelief tasks for example, have shown that 34 year old children were more likely to succeed ifthe verbal scenario and question included indexical referents (those that could be seen in thepictures) than if the referents were symbolic (referred to verbally but not visible in pictures)(see the meta-analysis of Abu-Akel and Bailey 2001). Wilson and Sperber (2002) suggestdevelopment includes an ability to see the relevance of the context and the questioner as acooperative communicator with communicative intent. Children learn that the given context

    123

  • 56 J Psycholinguist Res (2014) 43:4558

    (or background) renders the question relevant. That is, questions often lead the hearer toassume that there is some reason to believe that the information in the question is relevantand relates to the background information (Schwarz 1996; Wegner et al. 1981). In this studythe scenarios and story are the given context and were relevant to the question. The increasednumber of irrelevant answers by the children with PLD suggests that they are not utilising thegiven context perhaps because they have not yet developed the ability to understand that this isthe relevant information rather than utilising other knowledge. This may be due to an inabilityto consider what is meant by the questioner (understanding the intended meaning) as mightbe suggested by a more immature comprehender who stops at the first interpretation withoutconsidering the intention of the speaker (Wilson and Sperber 2002). The comparativelybetter performance of children with PLD on the story task suggests impairment is greaterwhen there is little additional context. In the clinical setting when considering appropriatetherapy, additional contextual support will make the task of interpretation more achievablewhere meaning of the language expression is implied or non-literal. Increased contextualsupport should be provided such as pictures or storybooks with pictures and conversationalexchanges about the characters perspective or questions targeting the characters point ofview are useful in assessing the stability of the childs language system. It is also importantto draw the attention of the child with SLI or PLD to the relevant context to make them awareof the alternatives in interpretation and to discourage stopping at the first interpretation (orrelying on a key word as is often reported in responses of children with SLI). The differencesin the clinical group highlight the importance of assessing pragmatic language understandingalongside structural language assessments (i.e. grammar, vocabulary or semantic knowledge)and the need for a theoretically based framework to analyse instances of pragmatic languagebreakdown. Relevance theory is a useful tool with which to investigate the nature pragmaticlanguage comprehension and analysing childrens answers in terms of the pragmatic languagedemands and contextual complexity (inferring a referent, enriching semantic meaning andrecovering implicatures). The difficulties of children with PLD can be addressed in theclinical setting by encouraging repetition and understanding of how the context is relevant tothe interpretation. Within this framework it is possible for clinicians to explore the difficultieswith pragmatic language interpretation of language impaired children, and to consider whatkinds of contributions may be most facilitative (i.e. role play to increase different languageexperiences). Further well-grounded research is necessary to provide a reliable and validassessment.

    The study did not investigate childrens intention reading ability which Sperber and Wilson(1995) and others (Tomasello 2002) have proposed to explain how children develop the abilityto understand meaning in context. The emergentist view (Evans 2001) however, argues thatthere is no reason to assume an intention reading ability and suggests atypical experienceof language will result in a reduced ability to attend to relevant contextual information.However, given that both impaired groups have atypical experience and the children withSLI without pragmatic difficulties produced less irrelevant answers than the PLD group thisexplanation is problematic. Further research would be needed but it appears that a delayin language development includes a delay in combining different sources of information toreach an implicature in children with PLD.

    Overall, the study suggests that children gradually develop the ability to integrate giveninformation and to utilise relevant context and that verbal information without supportingreferents/context increases the pragmatic load. RT provides a framework in which to interpretand explain the strategies. Childrens experience of being asked questions where the answerdepends on recovering implicatures may vary when they are very young. It is common inthe classroom to use open questions and imply meaning rather than explicit instruction.

    123

  • J Psycholinguist Res (2014) 43:4558 57

    In view of the developmental nature of pragmatic interpretation, this may mean that somechildren do not understand what is being asked of them, particularly children who maybe language delayed. Results of this study suggest the pragmatic language difficulties ofchildren with SLI and PLD should be considered in language therapy and in situations wherethe combining of information is required, the relevant information should be identified for thechild.

    Acknowledgments We gratefully acknowledge the support of the Hertfordshire Primary Care ResearchNetwork (Hertnet) for funding this project. Our thanks also to Margaret Ellis and Claire Hammond (Speechand Language Therapists) for their valuable advice on the design of materials. We would also like to thankthe staff and children for their participation in this study.

    Appendix: Examples of Incorrect Answers for each Category (Five Categories)

    Examples of Answers to Storybook Questions

    - - - - - Insert illustration here - -The Storybook had colour pictures.Text: It was midnight. The room was dark except for the torch which Mark shone on his

    face. He told a scary story and spoke in a whisper. The other boys were silent as they listened.Question: Why did Mark speak in a whisper?Story answer: To hear the monsters/ghosts (previous text explicitly states that the chil-

    dren are at a ghosts and monsters sleepover). Mark was making the story more scary bywhispering.

    World knowledge answer: So that others cant hear. (World knowledge of situationswhen one might whisper).

    Irrelevant answer: Because all the other boys havent heard the owl and they wanted tohear it. (The childs answer is irrelevant in the context of the story, there is no owl).

    Examples of Incorrect Answers to Verbal Questions

    Freddie helped his dad paint the fence. Freddie had to put on old clothes.Question: Why did Freddie put on old clothes?World knowledge answer: Because he doesnt like his new clothes.Scenario answer: His dad mightve told him to.Irrelevant answer: Because he found them

    References

    Abu-Akel, Ahmad, & Bailey, Alison L. (2001). Indexical and symbolic referencing: What role do they playin childrens success on theory of mind tasks? Cognition, 80, 263281.

    Bishop, D. V. M. (2000). Pragmatic language impairment: A correlate of SLI, a distinct subgroup, or part ofthe autistic continuum? In D. V. M. Bishop & L. Leonard (Eds.), Speech and language impairments inchildren: Causes, characteristics, intervention and outcome (pp. 99113). Hove: Psychology Press.

    Bishop, D. V. M. (2003a). Test for reception of grammar 2. London: Psychological Corporation.Bishop, D. V. M. (2003b). The childrens communication checklist version 2. London: Psychological Corpo-

    ration.Bishop, D. V. M., & Adams, C. (1992). Comprehension problems in children with specific language impair-

    ment. Literal and inferential meaning. Journal of Speech and Hearing Research, 35, 119129.

    123

  • 58 J Psycholinguist Res (2014) 43:4558

    Bishop, D. V. M., Chan, J., Adams, C., Hartley, J., & Weir, F. (2000). Conversational responsiveness inspecific language impairment: Evidence of disproportionate pragmatic difficulties in a subset of children.Development and Psychopathology, 12, 177199.

    Botting, N., & Adams, C. (2005). Semantic and inferencing abilities in children with communication disorders.International Journal of Language and Communication Disorders, 40, 4966.

    Evans, J. L. (2001). An emergent account of language impairments in children with SLI: Implications forassessment and intervention. Journal of Communication Disorders, 34, 3954.

    Kirk, S. A., McCarthy, J. J., & Kirk, W. D. (1968). Illinois test of psycholinguistic abilities. Illinois: UniversityOf Illinois.

    Kirk, E., Pine, K. J., & Ryder, N. (2011). I hear what you say but I see what you mean: The role of gesturesin childrens pragmatic comprehension. Language and Cognitive Processes, 26(2), 149170.

    Leinonen, E., Ryder, N., Ellis, M., & Hammond, C. (2003). The use of context in pragmatic comprehensionby specifically language-impaired and control children. Linguistics, 41, 407423.

    Marinac, J. V., & Ozanne, A. E. (1999). Comprehension strategies: The bridge between literal and discourseunderstanding. Child Language Teaching & Therapy, 15(3), 233246.

    Nelson, K. (2006). Advances in pragmatic developmental theory: The case of Language acquisition. HumanDevelopment, 49, 184186.

    Paul, R. (1990). Comprehension strategies: Interactions between world knowledge and the development ofsentence comprehension. Topics in Language Disorders, 10, 6375.

    Perrusi, C., Brandao, A., & Oakhill, J. (2005). How do you know this answer?Childrens use of text dataand general knowledge in story comprehension. Reading and Writing, 1, 687713.

    Renfrew, C. E. (1988). The renfrew action picture test. Bicester: Winslow Press Ltd.Robinson, E., & Whittaker, S. (1987). Childrens conceptions of relations between messages meanings and

    reality. British Journal of Developmental Psychology, 5, 8190.Ryder, R. J., & Graves, M. F. (1998). Reading and learning in content areas (2nd ed.). New York: Wiley.Ryder, N., & Leinonen, E. (2003). Use of context in question answering of 3, 4 and 5 year old children. Journal

    of Psycholinguistic Research, 32(4), 397415.Ryder, N., Leinonen, E., & Schulz, J. (2008). A cognitive approach to assessing pragmatic language com-

    prehension in children with specific language. International Journal of Language and CommunicationDisorders, 43(4), 427447.

    Sperber, D., & Wilson, D. (1995). Relevance communication and cognition (2nd ed.). Oxford: Blackwell.Sperber, D., & Wilson, D. (2002). Pragmatics, modularity and mind-reading. Mind and Language, 17(12),

    323.Strohner, H., & Nelson, K. (1974). The young childs development of sentence comprehension: Influence of

    event probability, nonverbal context, syntactic form, and strategies. Child Development, 45(3), 567576.Schwarz, N. (1996). Cognition and communication: Judgmental biases, research methods, and the logic of

    conversation. Mahwah, NJ: Lawrence Erlbaum.Tomasello, M. (1999). The cultural origins of human cognition. Cambridge: Harvard University Press.Tomasello, M. (2002). The new psychology of language (Vol. 2). Mahwah: Erlbaum.Waddell, M., & Granstrm, B. (1997). Bens bring your bear party. London: Walker Books Ltd.Wegner, D. M., Wenzlaff, R. M., Kerker, R. M., & Beattie, A. E. (1981). Incrimination through innuendo: Can

    media questions become public answers? Journal of Personality and Social Psychology, 4, 822832.Wiig, E., & Secord, W. (1992). Test of word knowledge. London: The Psychological Corporation.Wilson, D., & Sperber, D. (2002). Relevance theory. UCL Working Papers in Linguistics, 14, 249287.Winer, G. A., Cottrell, J. E., Mott, T., Cohen, M., & Fournier, J. (2001). Are children more accurate than

    adults? Spontaneous use of metaphor by children and adults. Journal of Psycholinguistic Research, 30(5),485496.

    Yazdi, A. A., German, T. P., Defeyter, M. A., & Siegal, M. (2006). Competence and performance in belief-desire reasoning across two cultures: The truth, the whole truth and nothing but the truth about false belief?Cognition, 100(2), 343368.

    123

  • Copyright of Journal of Psycholinguistic Research is the property of Springer Science &Business Media B.V. and its content may not be copied or emailed to multiple sites or postedto a listserv without the copyright holder's express written permission. However, users mayprint, download, or email articles for individual use.

    Pragmatic Language Development in Language Impaired and Typically Developing Children: Incorrect Answers in ContextAbstractIntroductionMethodParticipantsProcedure and Task Design

    ResultsLanguage Abilities and Performance of Children with SLI

    DiscussionAcknowledgmentsAppendix: Examples of Incorrect Answers for each Category (Five Categories)Examples of Answers to Storybook QuestionsExamples of Incorrect Answers to Verbal Questions

    References