The role of linguistic context in deriving word meanings in individuals with Down Syndrome

Research in Developmental Disabilities 34 (2013) 605–615

Contents lists available at SciVerse ScienceDirect

Research in Developmental Disabilities

The role of linguistic context in deriving word meanings in individuals with Down Syndrome Maja Roch *, Elena Florit, Chiara Levorato University of Padova, Department of Developmental Psychology, via venezia 8, 35131 Padova, Italy

A R T I C L E I N F O

A B S T R A C T

Article history: Received 27 March 2012 Received in revised form 17 September 2012 Accepted 20 September 2012 Available online 1 November 2012

Deriving the meaning of unknown words from context and its relationship to text comprehension was investigated in 24 individuals with Down syndrome and in 24 typically developing children matched for the Peabody Picture Vocabulary Test (PPVT) score. The study consisted of three phases. Unknown words were identified during the first phase (PPVT). Those words were presented embedded in brief linguistic contexts during the second phase. Recognition (maintenance) of word meanings was verified in the third and final phase. Both groups of participants recognized the meanings of a noteworthy number of words in contexts and a high percentage of these was maintained when they were presented at a later date without the support of context. Over and above group differences and basic linguistic skills, text comprehension seems to predict the ability to use context. Context provides the semantic information necessary to extract word meaning by activating relevant world knowledge. ß 2012 Elsevier Ltd. All rights reserved.

Keywords: Use of context Vocabulary acquisition Text comprehension Down syndrome

1. Introduction This study focuses on the ability of individuals with Down Syndrome (DS) to deduce the meaning of new words from linguistic contexts and the relationship between that ability and language comprehension. Its background is linked to the theory concerning the ability to use context clues to deduce meanings (Cain, Oakhill, & Elbro, 2003; Cain, Lemmon, & Oakhill, 2004; Cain, 2007; Cunningham, 2005; Levorato & Cacciari, 1992; Nagy & Scott, 2000). When a listener or reader encounters unfamiliar expressions (e.g. words, idioms) he/she analyzes the surrounding context for clues to its meaning (Fukkink, 2005). It is well documented that linguistic expressions can be acquired after as little as a single exposure in a supportive linguistic context (Goodman, McDonough, & Brown, 1998; Nagy, Herman, & Anderson, 1985). Proposed to explain how children interpret figurative language (Cacciari & Levorato, 1989, 1998; Levorato, 1993; Levorato & Cacciari, 1999; Levorato, Nesi, & Cacciari, 2004), the Global Elaboration Model (GEM Levorato & Cacciari, 1995) illustrates how they use linguistic context to derive the meaning of unfamiliar expressions and how these are discerned. Deducing the meaning of unknown words involves exploiting knowledge about the meaning of the surrounding context and the world (Hannon & Daneman, 2004; Levorato, Roch, & Nesi, 2007). Consistent with the GEM, the meaning assigned to an unfamiliar expression depends on the context in which it appears. Indeed, according to the construction–integration model, ‘‘The word meaning is always situation specific and context dependent’’ (Kintsch, 1998, p. 165). Only when a child is able to construct a coherent semantic representation can the meaning of an unknown expression be gathered from a text in which it is embedded.

* Corresponding author. E-mail address: [email protected] (M. Roch). 0891-4222/$ – see front matter ß 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ridd.2012.09.014

606

M. Roch et al. / Research in Developmental Disabilities 34 (2013) 605–615

GEM’s implication that there is a relation between text comprehension and the ability to use context (Cain, Towse, & Knight, 2009; Levorato et al., 2004, 2007; Nesi, Levorato, Roch, & Cacciari, 2006) has been supported by studies on typically developing (TD) school-age children at different levels of text comprehension (Cain & Towse, 2008; Levorato et al., 2004, 2007; Nesi et al., 2006) and, more importantly for the current work, on individuals with DS (Levorato, Roch, & Beltrame, 2009; Roch & Levorato, 2010). Roch and Levorato (2010) studied a group of 20 individuals with DS and a group of 20 TD children matched for their level of text comprehension in order to investigate their ability to interpret ambiguous idiomatic expressions (e.g. to break the ice). The participants with DS (mean age 14.9 years; SD = 2.2) and the TD children (mean age 6.7 years; SD = 0.4) were presented with unfamiliar idioms embedded in short linguistic contexts such as stories. The correct meanings of idioms were presented within multiple choice tasks together with a paraphrase of the literal meaning and other alternatives linked to the situation narrated in the context. The participants with DS were found to be as proficient as the TD children in identifying the meaning of idioms presented in context. Since the two groups were matched for their level of text comprehension, the investigators concluded that comprehension of an idiom is a function of text comprehension. In fact, the probability that the participants would choose the appropriate meaning of an idiom was associated to good text comprehension skills while that of choosing one of the two incorrect answers was associated to poor ones. Levorato et al. (2009) studied a group of 16 adolescents with DS and a group of 16 TD six-year-olds matched for text comprehension to investigate their ability to use context to facilitate the comprehension of morphosyntactically complex sentences. All the study and control participants took the standardized version of the Test of the Reception of Grammar (TROG – Bishop, 1989) which assesses children’s grammatical comprehension and tests understanding of sentence comprehension (the assessment phase). The TROG has a multiple-choice format; children taking the test are asked to listen to a sentence and to select the picture that matches its meaning. All the items use a simple vocabulary with grammatical complexity increasing as the test proceeds. Four 4-choice items are used to test understanding of each of 20 sentence types. Two weeks later, the participants taking part in that study were administered a modified version of the test during which the same sentences were once again presented in a spoken context three or four sentences long. The participants were again asked to match each sentence to the correct picture. The results showed that the individuals with DS, who were not as proficient as the TD children at recognizing the meaning of sentences when presented in isolation, did better in identifying the correct pictures when the sentences were introduced after a short context. This result suggests that individuals with DS are able to use contextual clues to infer the meaning of sentences they were otherwise be unable to understand. Over and above the ability to understand sentences per se, the ability to comprehend sentences in a wider context seems to be related to text comprehension which specifically contributes to comprehension of sentences embedded within a context. In accordance with the GEM, the two studies cited above indicate that context facilitates identifying the meaning of unknown expressions and that there are common processes linking the use of context and text comprehension. In the light of these findings the hypothesis has been advanced that individuals with DS could also use context to identify the meaning of unknown words. Chapman and other investigators demonstrated that individuals with DS are able to fast map (determine the meaning of a new word from seeing/hearing it used in a familiar context) the meaning of new words from a story context (Chapman, Sindberg, Bridge, Gigstead, & Hesketh, 2006; Kay-Raining Bird, Chapman, and Schwartz (2004); McDuffie, Sindeberg, Hesketh, & Chapman, 2007). In those studies, adolescents with DS were matched with TD children for either non-verbal mental age or for syntax comprehension to evaluate and compare their ability to define new words presented in the context of a spoken story. The adolescents with DS had more difficulty in defining words than the TD children matched for their nonverbal ability, but their performance was similar to that of the children matched for syntax comprehension. These results suggest that individuals with DS do indeed use information provided by a story’s context to infer the meanings of novel words at the level predicted by their language capacity scoring. The present investigation intends to take the studies described a step further and to extend the findings reported in the literature. The participants studied were thus provided with meaningful contexts simulating everyday life situations. Since individuals with DS seem to have poor memory and text comprehension abilities (Levorato, Roch, Florit, 2011), the contexts were only two or three sentences long and selected with the intent of providing the world knowledge necessary to identify the meaning of target words. The procedure differed from the one utilized by Kay-Raining Bird et al. (2004) as far as two details were concerned: (a) the words were not novel, as they were real Italian words from the Peabody Picture Vocabulary Test (PPVT, 1991) and (b) in view of the fact the persons with DS have difficulty with oral and written language, comprehension of the meaning of words embedded in a context was assessed utilizing a recognition rather than a definition task. 1.1. The current study: its aims and research questions The current study consisted of three phases. During the first, a baseline assessment of receptive vocabulary was made to identify words that were unknown to the participants. During the second, carried out ten days later, each participant was presented with the words he/she was unfamiliar with at the baseline assessment but this time embedded in short linguistic contexts. Finally, during the third and final stage carried out two weeks later the unknown words were presented out of context.

M. Roch et al. / Research in Developmental Disabilities 34 (2013) 605–615

607

The individuals with DS were compared to a group of TD children who were matched for their baseline assessment, namely their score on the PPVT and the number of words they did not know. The ability being investigated was used as the matching criterion (see also Roch & Levorato, 2009; Roch & Levorato, 2010) to verify whether, at the same level of receptive vocabulary, the participants with DS showed the same ability to identify word meanings from context. Participants were also assessed with regard to other language skills: sentence comprehension was assessed in order to control for basic linguistic skills; listening text comprehension was assessed in order to test the prediction based on the GEM that the ability to identify the meaning of unknown words presented in a context is linked to the ability to process the context itself. In view of the literature data demonstrating that context can facilitate comprehension of literal and idiomatic sentences (Roch & Levorato, 2010; Levorato et al., 2009) and studies showing that individuals with DS can fast map new words (Chapman and colleagues), the focus of the current work was twofold: (1) To investigate the ability of persons with DS to use linguistic contexts to deduce the meaning of unknown words. The issue was addressed by asking two research questions: (A) To what extent does linguistic context facilitate deducing the meaning of unknown words? (B) Are those words recognized later on when they are presented without the support of a context? (2) To investigate the role of the ability to understand texts in deducing the meaning of new words from a context and in recalling those meanings at a later date when they are presented without the support of a context. The research question that was specifically addressed in this study was then: Above and beyond receptive vocabulary and sentence comprehension, does text comprehension account for the ability to use context clues to deduce word meanings and to maintain knowledge of newly acquired word meanings? 2. Methods 2.1. Participants Twenty-four individuals with DS (10 males and 14 females, between 10 years + 6 months and 20 years + 8 months; mean age 15.9 years, SD = 2; 6) took part in this investigation. Assisted by special education teachers, they had been attending mainstream public school classes together with TD children of approximately the same chronological age for a period between 5 and 15 years (M = 9 years, SD = 3 years). From middle and lower-middle class socio-economic backgrounds, all the participants were Caucasian and spoke Italian as their native language. According to the files of the Italian National Health Care Service, which basically provides for all citizens’ medical and mental health services, none of the participants had any medical, psychological, or neurological impairments at the time this study was conducted. In order to assess their non-verbal abilities at the time this study was carried out and to verify if these were linked to performance on linguistic tests, the participants were administered the Coloured Progressive Matrices (CPM, Raven, 1996) on which they attained a mean score of 17 (SD = 2.8, range 13–24) out of 36 items. The participants also took the revised PPVT (Dunn & Dunn, 1981) – a version adapted for Italian speakers by Stella, Pizzoli, & Tressoldi (2000) – (see the materials section for a detailed description of the test). In order to identify and recruit children with an age-appropriate receptive vocabulary that was similar to that of the study participants, 38 TD children attending kindergarten and elementary schools also took that test. Each of the study participants was matched to one of the TD children who had obtained a similar score and had made a similar number of errors at the baseline assessment. Twenty-four TD children (12 males and 12 females, between 4 years + 6 months and 6 years + 10 months) were thus selected and made up the control population of this investigation. From middle and lower-middle class socio-economic backgrounds, all the TD children were Caucasian, spoke Italian as their native language, and attended public schools. According to their teachers, none had learning difficulties nor had been diagnosed with developmental, psychiatric, psychological, or neurocognitive disorders. It can be assumed then that their cognitive and linguistic abilities fell within typical parameters. The raw score on the PPVT for the individuals with DS and the TD children was 69.42 (SD = 15.28) and 70.88 (SD = 16.13), respectively. The scores of the two groups almost overlapped [t (46) = .322, p = .749], and the p level calculating group differences was such that these were considered matched as far as that variable was concerned (Mervis & Klein-Tasman, 2004). The study and the control groups also made a similar number of errors: mean = 18.29 (SD = 7.59) and mean = 19.21 (SD = 7.85), respectively [t (46) = .411, p = .683]. Again, the p value was sufficiently high that the two groups were considered matched for this variable. 2.2. Materials The PPVT Revised (Dunn & Dunn, 1981), adapted for Italian speakers by Stella et al. (2000) is a standardized test which measures the receptive vocabulary of children between 3 and 12. It consists of a list of 175 words that are presented verbally to participants who are asked to indicate which out of four pictures is the right one. The items are presented in order of

608

M. Roch et al. / Research in Developmental Disabilities 34 (2013) 605–615

increasing difficulty. A basal level is defined as the ability to give at least eight consecutive correct answers; the test is continued until the participant gets six answers wrong out of the last eight words presented (ceiling level). Each participant is thus presented with a different number of words depending on his/her performance. Raw scores correspond to the number of correct answers minus the number of errors. 2.2.1. PPVT in context A total of 104 words were unknown to the study participants: 68 nouns, 17 verbs and 19 adjectives. The mean number of unknown words was 18.5 (SD = 8, range 7 to 32 words): these were presented during phase 2 embedded in a context. In addition, 12 words chosen from among the most difficult words on the PPVT, not utilized during Phase 1 as they were above the vocabulary level of all the participants, were also presented at this time embedded in short linguistic contexts two or three sentences long (mean length 15 words, SD = 1.1). The reliability of this task (Crombach’s a) calculated on the 12 difficult words administered to all participants was .59. The contexts. Examples of contexts created for each of the word classes can be found in the Appendix. The contexts describe simple situations aiming to activate world knowledge stored in the participant’s long-term memory concerning some quality linked to the word’s meaning. The procedure for constructing contexts was as follows: - The contexts were discussed by the authors who chose sentences that provided the simplest and most typical descriptions of the word meanings - Examples of possible uses of words were taken from the Italian dictionary, excluding (a) figurative meanings and (b) examples giving indirect clues (e.g. in the context describing the word ‘birdcage’ any reference to bars was avoided since only the picture of the birdcage had bars); - Twelve adults were asked to rate how appropriate the contexts were in aiding children deduce a target word on a scale of 1–5 (1 = not appropriate; 5 = very appropriate). The adults were presented with the list of 104 target words, followed by short contexts. The mean score obtained was 4.17 (SD = 0.44), which is different from the neutral score 3 [t (106) = 26.9, p < .001, d = 5.2]. Contexts attaining a score below 3 were then modified. A full list of words and contexts can be obtained by contacting the first author.

2.2.2. The TROG (Bishop, 1989); reduced version, adapted for Italian (edited by Bisiacchi, Cendron, Gugliotta, Tressoldi, & Vio, 2005) which was used in this study consisted of 18 sentences containing morphosyntactic clues. The procedure for administering the test calls for the examiner to read one sentence at a time to the child who is asked to indicate which picture out of four corresponds to the sentence. Each correct answer is given a point, with the total score ranging from 0 to 18. 2.2.3. The test for the evaluation of oral text comprehension (TOR 3-8). (Levorato & Roch, 2007) The TOR 3-8 test was standardized utilizing 1700 Italian children between 3 and 8. During testing the study participants were asked to listen to two stories appropriate for 6- to 8-year-old Italian children. The tester read the stories out loud and, in order to minimize the cognitive and memory load, interrupted reading at two predetermined points and then asked multiple-choice questions testing for comprehension. The tester verbally presented four alternative answers/pictures. Participants are asked to point to the correct picture; in this way listening comprehension was tested without involving expressive language. Comprehension was assessed for each story using 10 questions, half concerning information explicitly stated in the story and half requiring inferences to be made. The score consists of the sum of correct answers, 10 for each story, with a maximum score of 20. 2.2.4. Procedure 2.2.4.1. Phase 1. PPVT standard. The baseline. In order to have a baseline assessment of vocabulary knowledge for each participant, the PPVT was administered in its standard form. 2.2.4.2. Phase 2. PPVT in context. The test. Ten days after Phase 1, the PPVT in Context was administered to test the comprehension of unfamiliar words embedded in linguistic contexts. Each participant was presented with only the words whose meaning he/she had not correctly identified in Phase 1. The participants were asked to choose which picture illustrated the word’s meaning. As we were investigating ‘‘incidental learning’’ (Rice, Oetting, Marquis, Bole, & Pae, 1994), no training was given. The participants were not told that the context served to provide clues to facilitate recognition of word meanings, nor were they told that they had already heard the words during a previous testing session. The only feedback given was satisfaction and appreciation for their collaboration. 2.2.4.3. Phase 3. PPVT out of context: maintenance. Two weeks after phase 2, the words previously presented in context were administered without the support of context. The participants listened to each unknown and difficult word and were asked to point to which of four pictures was appropriate. As in the previous phases, the same recognition task was used in order to render performance across the three phases of the study comparable.

M. Roch et al. / Research in Developmental Disabilities 34 (2013) 605–615

609

The time intervals between phases 1 and 2 (10 days) and between phases 2 and 3 (two weeks) were the same for all the participants. These intervals were brief enough that it was unlikely that words were learned in everyday situations during that period and long enough that any memory effect across the three phases was unlikely. All the study and control participants individually took part in all three sessions: The standard PPVT, the TROG, plus the CPM test which was administered only to the individuals with DS during the first session. Ten days later the PPVT in Context was administered during the second session to all participants. During the third session held two weeks later, the participants were administered the PPVT out of context and the TOR 3-8. Each session lasted between 30 and 40 min including short breaks which were scheduled as needed. School authorities and parents provided written consent for the TD children to be assessed at school, and community authorities and parents or guardians gave written consent authorizing the individuals with DS to be assessed in community centers. 3. Preliminary survey In order to test the context facilitation hypothesis, target words were shown out of context to 5 individuals with DS the same age and with similar language skills as those taking part in the main study. Like the main part of the study, this one consisted of three phases. The first phase (baseline assessment) was the same as in the main part and likewise served to identify words unknown on the PPVT by each participant. During the second phase, the experimenter explicitly attempted to teach unknown words by saying the word and pointing to the right picture out of 4 being shown. This was done twice (e.g. ‘‘This is the dromedary, look at the dromedary’’). The participant was then asked to point to the right (e.g. ‘‘Show me which one is the dromedary’’). All the participants correctly indicated the appropriate picture for all the words, showing that they did indeed understand the association between the picture and the word. The third phase (maintenance), which was carried out two weeks later, was the same as that in the main study: the experimenter said the target word and asked the participant which of the four pictures was the right one. The mean percentage of correct answers at this phase was 23.3% (SD = 4.6%). None of the participants performed above chance levels in recognizing correct pictures. In 53% of the cases, participants tended to choose the same picture they had during Phase 1 (the incorrect one). The fact that the participants did not identify the meanings of those words suggests that the attempt to explicitly teach the word was unsuccessful and that using the simple word-picture association technique to teach new words is ineffective. Presumably, any percentage higher than 23.1 found during the maintenance phase of the main study can be attributed to the facilitating effect of context. 4. Results The possible effect of individual characteristics such as score on the CPM test, age, years of schooling of the participants with DS was analysed in relation to their linguistic tasks: the PPVT at the three sessions, the TROG and the TOR 3-8. Since no significant correlation emerged (r ranges between .008 and .34), individual characteristics were not included in subsequent analyses. The performance of the two groups with regard to the 12 difficult words and the set of unknown words which differed across participants in function of each person’s vocabulary was compared using a t test. Since the comparison did not show any difference, the two types of words were merged into a single sample (from now on called ‘‘unknown words’’) in order to improve the statistical power of the analyses. A further set of analyses were performed in order to verify if the number of items presented to each participant had any effect on the results. A zero order correlation between the number of items presented and the percentage of correct answers in phase 2 was performed. The correlation was close to 0, specifically, r = .032 and r = .049, respectively, for the individuals with DS and for the TD children showing that their performance was not influenced by any variation in the number of items. 4.1. Descriptive analyses Table 1 illustrates the performance of the two groups on the two linguistic tasks during the three testing phases. Performance on each task by the two groups was compared using a series of t tests. As the tasks have different units of measurement, Cohen’s d were calculated and compared across the tasks to estimate the effect sizes. In this way the group differences in the level reached on each task acquired significance. A Bonferroni correction for multiple comparisons was adopted (.05/5 = .01). Only the p values lower than .01 were considered statistically significant. 4.2. Performance in language tasks: TROG and TOR 3-8 Both raw and standardized mean scores on the TROG and the TOR 3-8 are reported here since it is well recognised that standard and age equivalent scores are only partially useful when the performance of individuals with developmental disorders is being interpreted (see Maloney & Larrivee, 2007, for details). Standardized scores on the two tests are likewise reported in order to assign practical relevance to the performance of the two groups. All the TD children performed appropriately for their age since no one obtained a score below two standard deviations from the mean performance appropriate for their age. The individuals with DS were less proficient with respect to the TD children on both the TROG and the TOR 3-8 tests: the effect sizes for the differences are high for the analyses computed on both raw and standard scores.

610

M. Roch et al. / Research in Developmental Disabilities 34 (2013) 605–615

Table 1 Descriptive statistics for the two groups.

PPVT standard (Phase 1)

PPVT in context (Phase 2) PPVT out of context (Phase 3) TROG

TOR

Raw score Range Standard scorea Raw score Range Raw score Range Raw score (Max 18) Range Standard score^ Range Raw score (Max 20) Range Standard scoreo Range

Down syndrome

Typical development

t test, d of Cohen

69 (15) 48–102 83 (11) 69–107 80% (8%) 55–95 64% (15%) 25–95 10.25 (1.9) 7–15 4;7 (11 months) 4;1–8;2 7.79 (2.3) 3–12 8.1 (1.2) 5–12

71 (16) 45–103 87 (8) 70–101 71% (8%) 57–83 43% (12%) 26–79 12.92 (2.5) 8 – 18 6;4 (1;7) 4;1–9;3 12.33 (3.1) 5–16 10.7(1.1) 8–12

t (46) = .322, d = .09 t (46) = .1.3, d = .36 t(46) = 3.68* d = 1.09 t(46) = 5.31* d = 1.57 t(46) = 4.15*, d = 1.22 t(46) = 4.4*, d = 1.32 t(46) = 5.76*, d = 1.70 t(46) = 7.7*, d = 2.31

o

M = 10, SD = 2. M = 100, SD = 15. ^ Linguistic age. * p < .05. a

4.3. Performance on PPVT administered in the three phases of the study Raw and standardized scores are reported for the PPVT administered during the baseline phase. Scores are listed as percentages with regard to the second and third phases of the study as every participant was presented a different number of words (see Table 1). Performance on the PPVT during the second and third phases yielded a large number of correct answers by both groups. Despite the fact that the two groups were matched for their performance during the baseline assessment of receptive vocabulary, the individuals with DS were more proficient in deducing the meaning of words both in context (test phase) and out of context (maintenance phase) than the TD children. Similar effect sizes, respectively in phase 2 (1.09) and in phase 3 (1.57), suggest that the group differences remained relatively stable across the two sessions in which the PPVT was administered with and without the context. 4.4. The ability to deduce the meaning of words by using context clues How context facilitates deducing the meaning of unknown words was first assessed. The individuals with DS correctly recognized 80% and the TD children 71% of the words that they did not known at a previous session (see Table 1). Both of these percentages exceed the chance levels of 25% [t (23) = 33.1, p < .001 d = 13.8 and t (23) = 27.5, p < .001 d = 11.5 respectively for the two groups]. Both of these percentages are, moreover, different from zero [t (23) = 48.28 d = 20.1, and t (23) = 42.46 d = 17.6 respectively for the two groups] and the effect sizes are high. It can be concluded from these findings that a short context greatly facilitates the recognition of word meanings. The second thing to be assessed was if unknown words were recognized even without the support of a context during the third phase of the study. The percentage of correct answers was calculated considering (a) the total number of words learned with respect to the unknown words presented (i.e. comparing phases 3 and 1; (b) the total number of learned with respect to the words presented (i.e. comparing phases 3 and 2). With respect to the first, the individuals with DS learned the meanings of 64% of the previously unknown words and the TD children learned 43%. Both of these percentages differ from chance levels [t (23) = 27.5, p < .001 d = 11.5 and t (23) = 7.7, p < .001 d = 3.2, respectively for these two groups], suggesting that both provided correct answers systematically rather than casually. Nonetheless, a comparison of these percentages with 0, which was the baseline knowledge condition, yielded the following statistics: [t (23) = 21.06, p < .001 d = 8.8] and [t (23) = 18.09, p < .001 d = 7.9] for the two groups, respectively. When the percentage of correct choices in the third phase was compared with that in the second phase, it was found that the individuals with DS retained an average of 80% of the meanings recognized in the context while the TD children retained 60%. The meaning of a relatively high number of words was thus retained. 4.5. The role of text comprehension in identifying and maintaining new word meanings A zero-order correlation between the tasks was performed for each of the two groups. Table 2 illustrates the results. A Bonferroni’s correction for multiple comparisons was applied (.05/10) and only the p values lower than .005 were considered as statistically significant. The pattern in the findings was different in the two groups studied. Significant correlations emerged between the TOR 38, on the one hand, and the baseline TROG and the PPVT, on the other, in the TD children. Their text understanding seems

M. Roch et al. / Research in Developmental Disabilities 34 (2013) 605–615

611

Table 2 Correlations between the tasks (Down syndrome above the diagonal, typical development below the diagonal).

1.PTTV standard 2. PTTV (in context) 3. PTTV (out of context) 4. TROG 5. TOR

1.

2.

1 .228 .277 .344 .640*

.311 1 .428 .323 .465

3. .196 .465 1 .094 .040

4.

5.

.094 .192 .034 1 .662*

.142 .382 .502* .400 1

* p < .005 with Bonferroni’s correction

then to be correlated to vocabulary and sentence comprehension. A significant correlation between the TOR 3-8 and the PPVT in the maintenance phase was, instead, found in the individuals with DS. In fact, the better their text understanding, the higher was the probability of improving their vocabulary skills. In order to test the hypothesis that text comprehension is a predictor of the ability to use context to infer and maintain word meanings, two multivariate hierarchical regressions were carried out, one using the performance on the PPVT in Context and the other using the PPVT out of context as dependent variables. For both analyses, the predictors were established a priori. The group was inserted in the first step in order to verify the contribution of individual characteristics related to membership. In the second step, the scores on the PPVT during the baseline assessment were inserted, while in the third one the TROG was added as a control variable. These latter two steps of the analysis aimed to verify, over and above group differences, the contribution of basic language skills to the ability of processing a short context. In the fourth step, the TOR 3-8 was inserted in order to verify if, over and above group differences and basic linguistic skills tested by the PPVT and the TROG, this higher order skill predicted the ability to use context to understand and maintain the meaning of new words. Finally, in order to test if the contribution of TOR 3-8 and of the two basic linguistic skills was similar in the two groups, the interactions between the group and performance variables on the three linguistic tasks was tested. Since the product of the two variables included in the interaction (Group x TOR 3-8, TROG and PPVT) is correlated to each of the main effects (i.e. Group, TOR 3-8, TROG and PPVT) the standard error increases and the power of the model decreases. In order to avoid this effect, we used centralized variables instead of raw scores (Tabachnick & Fidell, 2001see Tabachnick and Fidell’s indications, 2001). The interactions did not contribute with significant variance to the models suggesting that the way in which these skills predicted performance on the PPVT in context and out of context is similar in the two groups. More parsimonious regression models excluding the interactions were therefore run. Table 3 shows the model on the ability to use context clues: a total of 38.8% of variance is explained. The group variable accounted for 22.8% of variance: the individuals with DS outperformed the TD children. The variance uniquely explained by the PPVT and the TROG was 5.5% and 3.3%, respectively, a contribution which did not yield significance. Performance on the TOR 3-8 added a unique and significant contribution of 6.8% of variance to the model over and above group differences and language skills. Table 4 shows the model on the maintenance of word meanings, explaining a total of 49.7% of variance, distributed as follows. The group factor explained 38.1% of the total variance while performance on the baseline PPVT added 3.3% of the variance, which did not yield significance. Performance on TROG did not add further significant variance to the model. The TOR 3-8, inserted in the fourth step added a significant and unique value of 7.9% over and above group differences, Table 3 Summary of hierarchical regression analysis for variables predicting PPVT in context (N = 48): R2 = .388 [F(4,47) = 5.3, p < .001]. Variable

R2 change

Step 1

Group

.228

Step 2

PPVT (standard)

.055oo

Step 3

Step 4

o

TROG

TOR

F change (1,46) = 13.6, p < .001. F change (1,45) = 3.5, p = .069. ooo F change (1,44) = 2.1, p = .152. oooo F change (1,43) = 4.7, p < .05. * p < .05. ** p < .001. oo

B

SE B

B

o

Group

.087

.023

.478**

Group PPVT

.109 .001

.027 .001

.601** .191

Group PPVT TROG

.139 .001 .001

.029 .001 .006

.769** .131 .021

Group PPVT TROG TOR

.139 .001 .001 .011

.029 .001 .006 .005

.769** .131 .021 .424*

.033ooo

.068oooo

M. Roch et al. / Research in Developmental Disabilities 34 (2013) 605–615

612

Table 4 Summary of hierarchical regression analysis for variables predicting PPVT out of context (N = 48): R2 = .496[F(4,47) = 10.59, p < .001]. Variable

R2 change

Step 1

Group

.381

Step 2

PPVT (Standard)

.033oo

Step 3

Step 4

TROG

TOR

B

SE B

B

o

Group

.206

.039

.617**

Group PPVT

.203 .002

.038 .001

.608** .182

Group PPVT TROG

.218 .002 .006

.045 .001 .009

.652** .199 .084

Group PPVT TROG TOR

.278 .003 .008 .022

.049 .001 .010 .008

.832** .263* .129 .455*

.005ooo

.078oooo

o

F change (1,46) = 28.3, p < .001. F change (1,45) = 2.5, p = .119. ooo F change (1,44) = .37, p = .546. oooo F change (1,43) = 6.6, p < .05. * p < .05. ** p < .001. oo

vocabulary, and sentence comprehension. This result supports the hypothesis that learning and maintaining new word meanings is related to higher order linguistic skills, namely text comprehension, rather than to more basic linguistic skills. 5. Discussion This study examined the ability of individuals with DS to use context clues to identify the meanings of unfamiliar words and to recognize those meanings at a later date without the support of a context. Three main results emerged. First, both individuals with DS and TD children are able to recognize the meanings of numerous unknown words when these are embedded in short linguistic contexts and a relatively high percentage of those words are also recognized out of context at a later date. Second, over and above vocabulary size and sentence comprehension, in both groups the ability to recognize meanings is predicted by the capacity to comprehend a text. In accordance with implications of the GEM, this result suggests that common processes link text comprehension and the ability to use context (Levorato et al., 2007). Third, even though the individuals with DS had less proficient language skills, they benefited from context more than the TD children matched for receptive vocabulary. The theoretical and practical implications of these findings are discussed below. 5.1. Contextual facilitation The theory that context has a facilitating effect on recognizing the meaning of unknown words is based on findings demonstrating that individuals with DS possess fast mapping skills (Kay-Raining Bird et al., 2004) and are able to process contexts to comprehend literal and idiomatic sentences (Levorato et al., 2009; Roch & Levorato, 2010). To our knowledge, this is the first study to examine comprehension of words from everyday life in short contexts in individuals with DS. The facilitating effects of context on both the study and control groups were striking as more than 70% of the answers to the questions asked were correct. This singular finding is important in view of the fact that the contexts were specifically constructed to make the meanings of the words easily accessible. The authors, in fact, did not aim to create an artificial condition but to stimulate one that is typical of the process of learning new meanings. It can be argued that the linguistic contexts we presented to the participants were so effective because they activated basic world knowledge linked to characteristics or functions of the word’s meaning: What is known is the basis for acquisition of what is unknown. In accordance with the GEM (Levorato & Cacciari, 1995), the facilitating effect of context is due to activation of semantic and episodic memory to integrate existing knowledge with a linguistic input. A striking number of words recognized in context by both groups of participants were also identified subsequently during the maintenance phase when these were presented out of context. It should be underlined that the unknown words were presented in context only once during testing. Moreover, the participants were not provided with any explanation about the purpose of the contexts, nor were they advised that they would be expected to utilize them to deduce meanings of unknown words. Finally, no reinforcement was given when the participants made a correct association between a word in context and the picture. The participants discovered appropriate associations by themselves and recalled many of these. Learning word meanings was thus incidental rather than deliberate and can be considered similar to the process of acquisition of new words in everyday life when words are encountered in various communicative contexts that are not specifically dedicated to voluntary or strategic learning. This was confirmed by the results obtained during the preliminary survey during which an

M. Roch et al. / Research in Developmental Disabilities 34 (2013) 605–615

613

attempt was made to explicitly teach some individuals with DS the association between a word and a picture illustrating the word without any contextual support. In fact, the probability that they would choose the correct picture at a later date at a maintenance phase did not differ from chance. This explicit teaching was presumably ineffective because it involved only memory of the association between the word referent to the picture. Ricketts, Bishop, Pimperton, & Nation (2011), in fact, demonstrated that what makes a learning episode successful is the presence of specific information about the meaning of the word. We believe that the contexts used in the current study activated this kind of world knowledge and therefore facilitated the recognition of the word meaning. This finding highlights the importance of incidental learning to increase vocabularies both for individuals with DS and for TD children. The facilitating effect of context should be further analyzed to test the permanence of knowledge of meanings of new words over a longer period of time. In particular, what was neglected in the current work but is essential for the construction and maintenance of new vocabulary words is the role of mental resources involved in those tasks. Future studies should investigate the contribution of both visual and verbal memory which have been shown to be linked to the ability to deduce and to maintain the meaning of unknown words in TD children (Cain, 2007; Cain et al., 2003, 2004). The facilitating effect of context reported in the current study is higher than that described in studies of Chapman and colleagues (Chapman et al., 2006; Kay-Raining Bird et al., 2004; McDuffie et al., 2007), but the methodological differences between the studies are relevant. Chapman and colleagues used novel words, while in the current study the items were real words from daily living. More importantly, while in Chapman et al.’s works, maintenance was verified through a definition task, in the current study a recognition task was adopted. To define a word implies some sort of awareness of its relevant features and it is certainly a good measure of knowledge of its meaning. But the ability to define words also relies heavily on expressive language which is impaired in individuals with DS. The difficulty involved in producing an appropriate definition might have obscured the real word knowledge detected from the contexts. On the other hand, using a multiple choice task to test acquisition, as in this study, indicates that the participant possesses some knowledge about the word’s meaning but little is known about underlying semantic comprehension. Another important difference between studies is that we used very short contexts, and not stories, which did not involve a high processing load. Our aim was to activate previous knowledge without overloading memory. Future works should extend the investigation by adapting the trials in order to obtain data comparable to those reported by Chapman and colleagues (Chapman et al., 2006; Kay-Raining Bird et al., 2004; McDuffie et al., 2007). 5.2. The ability to use context is related to text comprehension skill The current work’s main contribution concerns the link between text comprehension and the ability to use contexts which confirms the GEM theory according to which text comprehension and using context clues to deduce the meaning of new words share common integrative processes (Cain et al., 2009; Cain & Towse, 2008; Levorato et al., 2004; Nesi et al., 2006). In fact, over and above the ability to process words and sentences, text processing involves integrative and constructive processes (cf. Kintsch, 1998). The ability to use context is a cognitive skill mainly involving top-down processes aiming to integrate new linguistic information with previous world knowledge. The context anchors new information to existing world knowledge thanks to integrative processes. A link between text comprehension and the ability to use context is consistent with previous studies focusing on TD children’s capacity to use contextual information to identify the meanings of new words by (Cain, 2007; Cain et al., 2003, 2004) and on the association between text comprehension and the ability to understand different types of sentences in context found in individuals with DS (Levorato et al., 2009; Roch & Levorato, 2010). The current study takes these results a step further concentrating on the process involved in deducing the meanings of unfamiliar words from linguistic contexts. The link between the use of context and text comprehension should be analyzed in greater depth in future investigations. Attention should be dedicated to the role of other underlying component skills that are involved in the abilities to use context and to comprehend a text and in particular inferential processes that are necessary to integrate existing knowledge and new meanings and verbal memory (see Cain, 2007). The relevance of these processes for successful text comprehension have been demonstrated in TD preschoolers (Florit, Roch, Altoe`, & Levorato, 2009) and in individuals with DS (Levorato et al., 2011; Roch, Florit, & Levorato, 2012). Future studies should investigate the role these components play in the acquisition of new meanings and knowledge from the context. 5.3. 5.3.Group differences in contextual facilitation The individuals with DS were able to benefit to a greater extent from contexts than the TD children. In fact, they deduced more new meanings and maintained more words after a two week interval. Although we do not have empirical data to explain this finding, some inferences can be made. Group differences that emerged in the analyses are due to a series of individual characteristics which are peculiar to the two groups of participants. In fact, comparison of the two groups must take into consideration both genetic differences as well as experiential factors. As different authors have argued, the performance of individuals characterized by developmental disorders such as Down syndrome is the outcome of a complex interaction between genetic predisposition and experience (Karmiloff-Smith, 2009, 2006; Thomas & Karmiloff-Smith, 2003). In the current study, the individuals with DS being studied were older, had more years of schooling and more life experience than the TD children. With respect to the controls, it can be said that they had experienced greater opportunities to become

614

M. Roch et al. / Research in Developmental Disabilities 34 (2013) 605–615

familiar with the use of context in gathering new information both at school and through everyday experience. It is presumable that those opportunities enhanced their incidental learning. Individuals with DS, moreover, often have greater exposure to formal instruction which may have increased their ability to consolidate knowledge (see Perfetti & Hart, 2001). It is possible that reading and literacy in general are important sources for vocabulary acquisition and lead to a more ingrained knowledge than oral communication (for an extensive review of literacy skills in individuals with DS, see Roch & Jarrold, 2008, 1012; Roch & Levorato, 2009; Roch, Florit, & Levorato, 2011). Future studies can address this issue by analysing individual differences in literacy levels of individuals with DS. Consistent with recent findings in TD children (Cunningham, Perry, & Stanovich, 2001), it is possible that reading contributes more to individual differences in vocabulary than oral language and this, of course, must be verified by further studies. The fact that individuals with DS are able to use context to deduce meanings of unknown words and to acquire new vocabulary through incidental learning might partially explain why they seem to have such good vocabulary skills. Many studies have shown that the dimension of lexicon seems relatively preserved with respect to syntax, phonological awareness, and other aspects of language development which are severely impaired in DS. Also, recent findings confirm that individuals with DS continue to widen their vocabulary throughout their adolescence/adulthood (Chapman & Hesketh, 2000). Incidental learning of new words from contexts may in part explain this finding. Individuals with DS are thus good candidates for continued intervention in language and related skills as well as in acquisition of knowledge through incidental learning which allows them to integrate their previously acquired world knowledge with situations described in a text or sentence. As Chapman and Hesketh (2001) stated, ‘‘goals for receptive vocabulary learning should be dictated by environmental demands, social and cognitive skills, and the individual’s interests and comprehension’’ (p.4). These practices would undoubtedly lead to a wider and presumably richer, more complex conceptual and word knowledge which, in turn, would have a positive impact on other linguistic and cognitive skills. Appendix Examples of contexts used in the study for different classes of words: noun, verb and adjective Narici/Nostrils Non vado a scuola perche´ ho il raffreddore. Le mie narici prudono e starnutisco. [I’m not going to school because I have a cold. My nostrils itch and I am sneezing.] Tuffarsi/To Dive I bambini vanno in piscina. L’insegnante dice loro di non tuffarsi perche´ potrebbe essere pericoloso. [The children are going to the swimming pool. Their teacher tells them not to dive because it might be dangerous.] Inclemente/Inclement Oggi c’e` un vento inclemente. Il vento soffia forte sugli alberi e le case. [Today there is an inclement wind. The wind is blowing hard through the trees and against the house.] References Bishop, D. V. (1989). Test for reception grammar (2nd ed.). University of Manchester: Manchester UK Age and Cognitive Performance Research Centre. Bisiacchi, P. S., Cendron, M., Gugliotta, M., Tressoldi, P. E., & Vio, C. (2005). BVN 5-11: Batteria di valutazione neuropsicologica per l’eta evolutiva. Trento: Erickson. Cacciari, C., & Levorato, M. C. (1989). How children understand idioms in discourse. Journal of Child Language, 16, 387–405. Cacciari, C., & Levorato, M. C. (1998). The effect of semantic analysability of idioms in metalinguistic tasks. Metaphor and Symbol, 13(3), 159–177. Cain, K. (2007). Deriving word meanings from context: does explanation facilitate contextual analysis? Journal of Research in Reading, 30(4), 347–359. Cain, K., & Towse, A. S. (2008). To get hold of the wrong end of the stick: Reasons for poor idiom understanding in children with reading comprehension difficulties. Journal of Speech, Language, and Hearing Research, 51, 1538–1549. Cain, K., Lemmon, K., & Oakhill, J. (2004). Individual differences in the inference of word meanings from context: The influence of reading comprehension, vocabulary knowledge and memory capacity. Journal of Educational Psychology, 96(4), 671–681. Cain, K., Oakhill, J. V., & Elbro, C. (2003). The ability to learn new word meanings from context by school-age children with and without language comprehension difficulties. Journal of Child Language, 30, 681–694. Cain, K., Towse, A. S., & Knight, R. (2009). The development of idiom comprehension: An investigation of semantic and contextual processing skills. Journal of Experimental Child Psychology, 102, 98–108. Chapman, R. S., Sindberg, H., Bridge, C., Gigstead, K., & Hesketh, L. (2006). Effect of memory support and elicited production on fast mapping of new words by adolescents with Down syndrome. Journal of Speech, Language and Hearing Research, 49, 3–15. Chapman, R. S., & Hesketh, L. J. (2000). Behavioural phenotype of individuals with Down syndrome. Mental Retardation and Developmental Disability Research Reviews, 6, 84–95. Chapman, R. S., & Hesketh, L. J. (2001). Language, cognition, and short-term memory in individuals with Down syndrome. Down Syndrome Research and Practice, 7, 1–7. Cunningham, A. E. (2005). Vocabulary growth through independent reading and reading aloud to children. In E. H. Hiebert & M. L. Kamil (Eds.), Teaching and learning vocabulary: Bringing research to practice (pp. 45–68). Mahwah, NJ: Lawrence Erlbaum. Cunningham, A. E., Perry, K. E., & Stanovich, K. E. (2001). Converging evidence for the concept of orthographic processing. Reading and Writing: An Interdisciplinary Journal, 14, 549–568. Dunn, L., & Dunn, L. (1981). Peabody picture vocabulary test-revised. Circle Pines, MN: American Guidance Service. Florit, E., Roch, M., Altoe`, G., & Levorato, C. (2009). Listening comprehension in preschoolers: The role of memory. British Journal of Developmental Psychology, 27(4), 935–951. Fukkink, R. G. (2005). Deriving word meanings from context: A process analysis. Learning and Instruction, 15, 23–43. Goodman, J. C., McDonough, L., & Brown, N. B. (1998). The role of semantic context and memory in the acquisition of novel nouns. Child Development, 69(5), 1330– 1344. Hannon, B., & Daneman, M. (2004). Shallow semantic processing of text: An individual-differences account. Discourse Processes, 37(3), 187–204. Karmiloff-Smith, A. (2006). The tortuous route from genes to behaviour: A neuroconstructivist approach. Cognitive, Affective and Behavioural Neuroscience, 6(1), 9– 17. Karmiloff-Smith, A. (2009). Nativism versus neuroconstructivism: Rethinking the study of developmental disorders. Developmental Psychology, 45(1), 56–63.

M. Roch et al. / Research in Developmental Disabilities 34 (2013) 605–615

615

Kay-Raining Bird, E., Chapman, R. S., & Schwartz, S. E. (2004). Fast mapping words and story recall by individuals with Down syndrome. Journal of Speech, Language and Hearing Research, 47, 1286–1300. Kintsch, W. (1998). Comprehension: A paradigm for cognition. Cambridge: Cambridge University. Levorato, M. C., & Roch, M. (2007). TOR 3-8: Test per la valutazione della comprensione del testo orale [TOR 3-8. Test for the evaluation of oral text comprehension]. Florence, Italy: OS Organizzazioni Speciali. Levorato, M. C., Roch, M., & Beltrame, R. (2009). Text comprehension in Down syndrome: The role of lower and higher level abilities. Clinical Linguistics & Phonetics, 23(4), 285–300. Levorato, M. C., & Cacciari, C. (1992). Children’s comprehension and production of idioms: the role of context and familiarity. Journal of Child Language, 19, 415– 433. Levorato, M. C., & Cacciari, C. (1995). The effects of different tasks on the comprehension and production of idioms in children. Journal of Experimental Child Psychology, 60, 261–283. Levorato, M. C., & Cacciari, C. (1999). Idiom comprehension in children: Are the effects of semantic analysability context separable? European Journal of Cognitive Psychology, 11(1,), 51–66. Levorato, M. C. (1993). The acquisition of idioms and the development of figurative competence. In C. Cacciari & P. Tabossi (Eds.), Idioms. Processing, structure and interpretation (pp. 105–128). Hillsdale, N.Y Laurence Erlbaum Associates. Levorato, M. C., Nesi, B., & Cacciari, C. (2004). Reading comprehension and understanding idioms: A developmental study. Brain and Language, 202, 4–16. Levorato, M. C., Roch, M., & Nesi, B. (2007). A longitudinal study of idiom and text comprehension. Journal of Child Language, 34, 1–22. Levorato, Roch, & Florit, (2011). The role of verbal memory in reading text comprehension of individuals with Down syndrome. American Journal of Intellectual and Developmental Disabilities, 16(2), 99–110. Maloney, E., & Larrivee, L. (2007). Limitations of age-equivalent scores in reporting the results of norm-referenced tests. Contemporary Issues in Communication Sciences and Disorder, 34, 86–93. McDuffie, A. S., Sindeberg, H. A., Hesketh, L. J., & Chapman, R. S. (2007). Use of speaker intent and grammatical cues in fast-mapping by adolescents with Down syndrome. Journal of Speech, Language and Hearing Research, 50, 1546–1561. Mervis, C. B., & Klein-Tasman, B. P. (2004). Methodological issues in group matching designs: Alpha levels for control variable comparisons and measurement characteristics of control and target variables. Journal of Autism ad Developmental disorders, 34(1), 7–17. Nagy, W. E., & Scott, J. A. (2000). Vocabulary processes. In Kamil, M. L., Mosenthal, P. B., Pearson, P. D., & Barr, R. (Eds.), Handbook of reading research (vol. 3, pp. 269– 284). Mahwah, NJ: Lawrence Erlbaum Nagy, W. E., Herman, P. A., & Anderson, R. C. (1985). Learning words from context. Reading Research Quarterly, 20, 233–253. Nesi, B., Levorato, M. C., Roch, M., & Cacciari, C. (2006). To break the embarrassment: Text comprehension skills and figurative competence in skilled and lessskilled text comprehenders. European Psychologist, 11(2), 128–136. Perfetti, C. A., & Hart, L. (2001). The lexical quality hypothesis. In L. Verhoeven, C. Elbro, & P. Reitsma (Eds.), Precursors of functional literacy (pp. 189–214). Amsterdam, The Netherlands: John Benjamins. Raven, J. C. (1996). CPM: coloured progressive matrices: Series A, AB, B. Board and Book Form. London: H.K. Lewis. (Progressive matrici colorate, It. Trans., Florence, Italy: Organizzazioni Speciali). Rice, M. L., Oetting, J. B., Marquis, J., Bole, J., & Pae, S. Y. (1994). Frequency of input effects on word comprehension of children with specific language impairment. Journal of Speech and Hearing Research, 37(1), 106–122. Ricketts, J., Bishop, D. V. M., Pimperton, H., & Nation, K. (2011). The role of self-teaching in learning orthographic and semantic aspects of new words. Scientific Studies of Reading, 15(1), 47–70. Roch, M., & Jarrold, C. (2008). A comparison between word and nonword reading in Down syndrome: The role of phonological awareness. Journal of Communication Disorders, 41, 305–318. Roch, M., & Jarrold, C. (2012). A follow-up study on word and non-word reading skills in Down syndrome. Journal of Communication Disorders, 45(2), 121–128. Roch, M., & Levorato, M. C. (2010). Idiom understanding in children and adolescents with Down syndrome: The role of text comprehension skills. Applied Psycholonguistics, 31, 531–550. Roch, M., & Levorato, M. C. (2009). Simple view of reading in Down Syndrome: The role of listening comprehension and reading skills. International Journal of Communication Disorders, 2, 206–223. Roch, M., Florit, E., & Levorato, M. C. (2012). The advantage of reading over listening text comprehension in Down syndrome: What is the role of verbal memory? Research in Developmental Disabilities, 33, 890–899. Roch, M., Florit, E., & Levorato, M. C. (2011). A follow-up study on reading comprehension in Down syndrome: The role of reading skills and listening comprehension. International Journal of Language and Communication Disorders, 46(2), 231–242. Stella, G., Pizzoli, C., & Tressoldi, P. (2000). Il Peabody Test. Test di vocabolario recettivo [Peabody: Test of receptive vocabulary]. Omega: Turin. Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). New York: Allyn & Bacon. Thomas, M., & Karmiloff-Smith, A. (2003). Modelling language acquisition in atypical phenotypes. Psychological Review, 110(4), 647–682.

.