Investigating prosodic ability in Williams syndrome

This article was downloaded by: [University of Sheffield] On: 7 April 2009 Access details: Access Details: [subscription number 731709316] Publisher Informa Healthcare Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Clinical Linguistics & Phonetics Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t713693308 Investigating prosodic ability in Williams syndrome Catherine Catterall a; Sara Howard b; Vesna Stojanovik c; Marcin Szczerbinski b; Bill Wells b a Dorset Speech & Language Therapy Service, UK b Department of Human Communication Sciences, University of Sheffield, UK c Department of Linguistic Sciences, University of Reading, UK Online Publication Date: 01 September 2006 To cite this Article Catterall, Catherine, Howard, Sara, Stojanovik, Vesna, Szczerbinski, Marcin and Wells, Bill(2006)'Investigating prosodic ability in Williams syndrome',Clinical Linguistics & Phonetics,20:7,531 — 538 To link to this Article: DOI: 10.1080/02699200500266380 URL: http://dx.doi.org/10.1080/02699200500266380 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. Clinical Linguistics & Phonetics, Sept–Oct 2006; 20(7–8): 531–538 Investigating prosodic ability in Williams syndrome CATHERINE CATTERALL1, SARA HOWARD2, VESNA STOJANOVIK3, MARCIN SZCZERBINSKI2, & BILL WELLS2 Dorset Speech & Language Therapy Service, UK, 2Department of Human Communication Sciences, University of Sheffield, UK, and 3Department of Linguistic Sciences, University of Reading, UK Downloaded By: [University of Sheffield] At: 00:09 7 April 2009 1 (Received 12 July 2004; accepted 24 October 2004) Abstract This paper investigates whether people with Williams syndrome (WS) have prosodic impairments affecting their expression and comprehension of four main uses of intonation. Two adolescent males with WS were assessed using the PEPS-C battery, which considers prosodic abilities within a psycholinguistic framework, assessing prosodic form and function in both the input and output domains. The performances of the subjects with WS were compared with control data for age and language-comprehension matched children. The results revealed significant prosodic impairment affecting all areas of the profile. Crucially, however, different profiles of strengths and weaknesses were revealed for the two subjects. The results support the growing view that WS is a heterogeneous population in terms of linguistic abilities. Keywords: Williams syndrome, prosodic impairment, intonation, psycholinguistic framework, PEPS-C Introduction Williams Syndrome is a rare genetic disorder which results from a gene deletion on chromosome 7. It has attracted significant research interest because of claims that individuals with WS present with an uneven profile, where it is claimed that linguistic abilities are relatively strong, compared to general cognitive functioning and abilities in the non-verbal domain (Bellugi, Marks, Bihrle, & Sabo, 1988; Clahsen & Almazan, 1998). Much of the research into linguistic abilities in WS has focused on the areas of grammar and semantics, where recent research has begun to question the claim that linguistic abilities are well-preserved in WS (Stojanovik, Perkins, & Howard, 2001; in press). In comparison to this grammatical and semantic research, relatively little attention has been paid to phonological abilities in WS, including prosody, although abnormalities with the use of affective prosody in story-telling have been reported by Reilly, Klima and Bellugi (1990). The present study investigates the prosodic processing and production of two adolescent males with WS. The aim is to contribute to the theoretical discussions described Correspondence: Sara Howard, Department of Human Communication Sciences, The University of Sheffield, 31 Claremont Crescent, Sheffield, S10 2TA, UK. E-mail: s.howard@sheffield.ac.uk. ISSN 0269-9206 print/ISSN 1464-5076 online # 2006 Informa UK Ltd. DOI: 10.1080/02699200500266380 532 C. Catterall et al. above, and also to provide data that may inform the clinical and educational management of individuals with WS. The following questions are addressed: 1. Are the prosodic abilities of the two WS subjects at an age-appropriate level? 2. Are the prosodic abilities of the two WS subjects in line with their other linguistic abilities? 3. What are the similarities and differences in the profiles of the two WS subjects? Method Downloaded By: [University of Sheffield] At: 00:09 7 April 2009 Subjects Two adolescent males with Williams syndrome, referred to as B and C, were selected for this study. At the time of data collection, B was aged 12;5 and in his first year of secondary education. He had been diagnosed with Williams syndrome at the age of 7;6. C was aged 13;3 and was in his second year of secondary education at the time of the study, having been diagnosed with Williams syndrome during very early childhood. Scores for both subjects were obtained for a number of standardized verbal and nonverbal measures including TROG (The Test for Reception of Grammar, Bishop, 1989), BPVS (The British Picture Vocabulary Scales, Dunn, Dunn, Whetton, & Pintilie, 1982), CELF (Clinical Evaluation of Language Fundamentals–Revised, Semel, Wiig, & Secord, 1987), and four non-verbal measures from the WISC–R (Wechsler Intelligence Scale for Children–Revised, Wechsler, 1981): Block Design, Picture Completion, Picture Arrangement and Object Assembly. Further information about both subjects is reported in Stojanovik (2002), from which the information in Table I is drawn. Table I shows that both subjects scored below the age-appropriate range on all tests, linguistic as well as non-verbal, with the exception of B’s score on the BPVS. Control data for this study was taken from research by Wells, Peppé, and Goulandris (2004), who used a battery of prosodic tasks to compile a picture of the development of intonation in school aged children. Data had been collected from 120 normally developing children divided into four equal sized groups with mean chronological ages of 5;5, 8;6, 10;8, and 13;8. In order to provide a chronological age (CA) control group for the present study, 15 children from the earlier study were matched with B and C for sex, age and ethnicity. The chronological age (CA) control group thus consisted of fifteen males aged between 11 and 13.9 years. The control data for the four focus sub-tests (see below) comes from different children, as initial difficulties in the original study led to the redesigning of these subtests and subsequent testing took place with different children (Wells, Peppé, & Goulandris, Table I. Verbal and non-verbal performance from standardized tests. Test TROG BPVS CELF-E Picture Completion Picture Arrangement Block Design Object Assembly a B: Raw Score B: Standard Score 12 22 73 4 6 4 11 69a 97a 64a 1b 2b 1b 4b Standard Wechsler scores (M5100; SD515). b C: Raw Score C: Standard Score 10 10 90 9b 4b 5b 18b Standard scaled scores (M510; SD53). 72a 57a 70a 3b 1b 1b 7b Investigating prosodic ability in Williams syndrome 533 2004). Therefore, the CA control group for this section consists of eight males aged between 11 and 13.9 years. A further language matched control (LC) group was identified, using the Test of Reception of Grammar (TROG: Bishop, 1989). B and C had raw scores of 12 and 10 respectively. These correspond to control children with a median age of 5;5–5;8. For the control group, boys were selected who had a TROG score of between 7 and 15. This group consisted of nine boys for the chunking, affect and interaction sub-tests, and six boys for the focus sub-tests. Once again, only a small group could be found for the focus tests because of the smaller group used in the original study. Downloaded By: [University of Sheffield] At: 00:09 7 April 2009 Data collection Data was collected during three sessions with B and four sessions with C. The sessions lasted up to 1 hour depending on the attention span of the subject, with the shortest session lasting 45 minutes. The sessions consisted of a combination of formal assessment using the PEPS-C battery (Wells & Peppé, 2001) and informal conversational interaction between the subject and an adult. All data collection sessions were held in a quiet room in a university speech and language therapy clinic. Materials Both subjects were assessed using the PEPS-C battery, (Wells & Peppé, 2003). PEPS-C incorporates the following dimensions: input (perception/comprehension) vs. output (generation/production); and form (referring to lower level phonetic processing, where meaning is not involved) vs. function (involving higher level processing, drawing on stored knowledge, relating phonetic form to meaning). PEPS-C covers four communicative areas, where intonation is generally agreed to have an important role: grammar, affect, interaction and pragmatics. The specific grammatical function of intonation tested in PEPS-C is ‘‘chunking’’. This refers to prosodic delimitation of the utterance into units (or intonation phrases) for grammatical, semantic or pragmatic purposes, e.g.,/COFFEE-CAKE/AND HONEY/vs. /COFFEE /CAKE/AND HONEY/. In the second utterance, there are three intonation phrases, each with its own accent, constituting the utterance as a list of three food items. In the first utterance the absence of a separate intonation phrase for CAKE constitutes ‘‘coffee-cake’’ as a compound noun. As an instantiation of the use of prosody to convey affective or attitudinal meaning, PEPS-C uses the distinction between expressing strong liking as opposed to reservation. One way in which this distinction can be expressed is by using rise-fall vs. fall-rise pitch movement respectively, for instance with the syllable [m:]. In order to assess the role of prosody in interaction, PEPS-C tests the prosodic opposition between a low fall, meaning ‘‘yes I understand’’ as opposed to a high rise, meaning ‘‘no I didn’t understand, please repeat’’. Focus refers to the speaker’s use of phonetic prominence to indicate which item is most important in an utterance, e.g.,/CHOCOLATE AND HONEY/vs./CHOCOLATE AND HONEY/. Each of the four communicative areas is tested for both input and output, with different tasks for form and for function. This gives a total of 16 tasks. Each input task has 16 items, and each Output task has 12 items. Their content is summarized in Table II. For a full description of the tasks, see Wells and Peppé (2003). The input form task for each of the four communicative areas comprises a same-different procedure. The child is presented with stimuli in a form where the lexical and grammatical information is not audible. The stimuli in fact consist of the laryngograph signal only, 534 C. Catterall et al. Table II. Brief description of PEPS-C tasks. Task name Chunking input form Chunking input function Chunking output function Chunking output form Affect input form Affect input function Downloaded By: [University of Sheffield] At: 00:09 7 April 2009 Affect output function Affect output form Interaction input form Interaction input function Interaction output function Interaction output form Focus input form Focus input function Focus output function Focus output form Description Same or different: pairs of 4–7 syllable stimuli differ in location of prosodic boundaries; no segmental information. Identification: recorded voice names two foods (e.g., FRUIT-SALAD AND MILK) or three foods (e.g., FRUIT, SALAD AND MILK). Naming: picture-strip shows either two foods (e.g., FRUIT-SALAD, MILK) or three foods (e.g. FRUIT, SALAD, MILK) Imitation: two numbers (e.g., FORTY-TWO, ONE) or three numbers (e.g., FORTY, TWO, ONE). Same or different: pairs of one syllable stimuli differ in pitch pattern (rise-fall vs. fall-rise); no segmental information. Identification. Single food item on picture. Recorded voice likes it ([m] with rise-fall) or is not keen ([m] with fall-rise). Child hears food-item (e.g., BANANAS) and, with [m] only, expresses liking or not keen. Imitation task: monosyllabic words, (e.g., ONE) with rise-fall or fall-rise. Same or different: pairs of one syllable stimuli differ in pitch direction (fall vs. rise) and pitch height at onset (low vs. high); no segmental information. Identification. Child names picture (e.g., CUP) which tester repeats either fall with low onset (affirming, i.e., ‘‘go on’’) or rise with high onset (questioning, i.e., ‘‘repeat’’) Child decides whether the tester wants child to go on to the next item or to repeat. Recorded voice speaks a non-word (e.g., PARGLE) or a real word (e.g., CARROT). Child repeats word, to sound as if checking understanding (non-word) or confirming understanding (real word). Imitation task: monosyllabic words with fall or rise, with low or high onset. Same or different: pairs of 4–6 syllable stimuli differ in location of main prosodic prominence; no segmental information. Identification. Recorded stimuli, e.g., ‘‘I wanted CHOCOLATE AND HONEY’’/‘‘I wanted CHOCOLATE AND HONEY’’. Child decides which food the speaker had not received. Tester offers child a picture saying e.g., ‘‘How about a green bike?’’ Child has to respond so as to get the picture s/he actually needs e.g., ‘‘I WANT A WHITE BIKE’’. Imitation of items consisting of three numbers, varying in accent location (e.g., THREE TWO ONE; THREE TWO ONE.) derived from spoken pairs such as the ones illustrating the four input function tasks in Table II. Pitch, loudness and length variations are preserved; the result is a ‘‘buzz’’—not dissimilar to listening to a speaker in an adjacent room, where the intonation is audible, but the content of the utterance is not. The input function tasks are designed as identification tasks: the child hears a spoken stimulus and has to assign it to one of two meaning categories, by pointing at the correct picture of two. The output function tasks involve a range of elicitation procedures, including picture naming (see Table II for details). The output form tasks involve repetition of short phrases. Administration of the PEPS-C battery was preceded by a phase in which the child’s knowledge of the vocabulary to be used was evaluated. This took the form of a picture naming procedure, in which coloured pictures of each of the vocabulary items used in the PEPS-C were presented in turn to the child. Stimuli for the input tasks had been prerecorded on digital audiotape (DAT) in a recording studio and were presented to participants via tape recorder in free field. All responses were recorded on DAT. Investigating prosodic ability in Williams syndrome 535 Analysis The children’s performance was scored by the first author from live observation and listening to the recording. On the PEPS-C input tasks, each of which comprises 16 items, the child has only two choices for each item—the response is either right or wrong. On the output tasks, each of which comprises 12 items, the scorer rates the child’s production of each item as right (2 points), wrong (0 points) or ambiguous (1 point), giving a possible maximum of 24 points. Downloaded By: [University of Sheffield] At: 00:09 7 April 2009 Results In order to address the first research question, ‘‘Are the prosodic abilities of the two WS subjects at an age-appropriate level?’’, the scores of B and C are compared to those of the CA control group. Means and range of scores for the CA control group for the 16 PEPS-C sub-tests are presented in Table III, along with raw scores and z scores for B and C. Inspection of the raw scores of the two subjects shows that in comparison to the CA control group, B scored below the mean score on 14/16 sub-tests. He scored below the range of control subjects’ scores on 5/16 sub-tests. C scored below the mean on 14/16 subtests and below the range of the control subjects’ scores on 2 out of 16 sub-tests. In order to address the second research question, ‘‘Are the prosodic abilities of the two WS subjects in line with their other linguistic abilities?’’, the scores of B and C are compared to those of the LC control group. Means and range of scores for the LC control group for the 16 PEPS-C sub-tests are presented in Table IV, along with raw scores and z scores for B and C. Inspection of the raw scores of the two subjects shows that in comparison to the LC control group, B scored below the mean on 13/16 sub-tests and below the range of LC subjects’ scores on 3/16 sub-tests. C scored below the mean on 10/ 16 sub-tests and below the range of LC subjects’ scores on 4/16 sub-tests. Thus there are a large number of subtests where both children are performing below the mean level for their CA peers. On several sub-tests, B and C are also performing below the level of their younger LC matched controls. In order to specify the distribution and extent of these deficits, a cut off point of -1.5 standard deviations (SD) from the norm, is used to show a potentially clinically significant deficit (marked * on Tables III and IV), while -2.0 SD (**) is used to indicate a particularly severe impairment (Wells & Peppé, 2001, 2003). The distribution of scores across PEPS-C sub-tests in Tables III and IV shows that B and C have different profiles. This was confirmed statistically: coefficient of correlation (Spearman’s rho) between 16 PEPS-C z-scores of B and C was –.071. This value is not significantly different from 0, where 0 indicates a lack of any similarity whatever between the two profiles. Discussion PEPS-C identified prosodic difficulties in both subjects with Williams syndrome. It is unlikely that their poor performance is due to general difficulties with PEPS-C arising from their low intelligence, since (a) each performed within normal range on some sub-tests and (b) they differed on the tasks which they did well on. For each subject, a wide range of difficulties was evident in comparison to age-matched peers. The extent and severity of these deficits are no less than that evident in the children with diagnosed specific speech and language impairments studied by Wells and Peppé (2003). Thus, based on the data Input form CA mean CA SD CA min–max B Bz C Cz Input function Output function Output form Ch Aff Inter Foc Ch Aff Inter Foc Ch Aff Inter Foc Ch 14.34 2.01 5–16 10** -2.16 11** -1.66 13.46 2.68 6–16 9* -1.66 12 -.54 13.54 2.74 6–16 12 -.56 11 -.93 14.13 1.46 11–16 8** -4.20 13 -.77 13.76 1.99 9–16 14 .12 10* -1.89 15.02 2.02 6–16 16 .48 9** -2.98 14.7 2.38 6–16 9** -2.39 8** -2.81 14.63 1.41 12–16 6** -6.12 9**# -3.99 19.96 3.36 11–24 10** -2.96 18 -.58 21.6 3.16 12–24 12* -3.04 10**# -3.67 20.42 3.88 7–24 14 -1.65 14* -1.65 21.38 2.20 18–24 17* -1.99 22 .28 21.94 3.44 11–24 15** -2.02 13** -2.60 Aff Inter 20.08 20.98 4.25 3.21 8–24 11–24 15 17 -1.20 -1.24 6**# 12** -3.31 -2.79 Foc 23.13 0.83 22–24 19** -4.98 23 -.16 Note. Ch5Chunking, Aff5Affect, Int5Interaction, Foc5Focus. *5performed at least 1.5 SD below mean of control group. **5performed at least 2 SD below mean of control group. #5Raw score below range of scores obtained by control group. Table IV. Raw and z scores for both subjects with data for language comprehension (LC) matched control group, (n59 ( n56, focus)). Input Form LC mean LC SD LC min–max B Bz C Cz Input Function Output Function Output Form Ch Aff Int Foc Ch Aff Int Foc Ch Aff Int Foc Ch Aff Int Foc 14.11 .93 12–15 10**# -4.42 11**# -3.34 11.33 3.12 8–16 9 -.75 12 .21 10.78 3.23 7–16 12 .38 11 .07 10.83 2.32 8–15 8* -1.22 13 .94 12.89 1.90 9–15 14 .58 10* -1.52 14.22 1.56 11–16 16 1.14 9**# -3.35 12.78 3.96 6–16 9 -.95 8 -1.21 7.83 .75 7–9 6**# -2.44 9 1.56 18.89 4.28 11–23 10**# -2.08 18 -.21 19.89 4.46 12–24 12* -1.77 10**# -2.22 18.89 6.01 7–24 14 -.81 14 -.81 19.33 2.94 15–24 17 -.79 22 .91 17.33 4.61 12–24 15 -.51 13 -.94 16.89 6.43 8–24 15 -.29 6*# -1.69 18.78 5.93 10–24 17 -.30 12 -1.14 21.83 2.14 19–24 19 -1.32 23 .55 Note. Ch5Chunking, Aff5Affect, Int5Interaction, Foc5Focus. *5Performed at least 1.5 SD below mean of control group. **5performed at least 2 SD below mean of control group. #5Raw score below range of scores obtained by control. C. Catterall et al. Downloaded By: [University of Sheffield] At: 00:09 7 April 2009 536 Table III. Raw and z scores for both subjects with data for age matched (CA) control group, (n 515 (n58, focus)). Downloaded By: [University of Sheffield] At: 00:09 7 April 2009 Investigating prosodic ability in Williams syndrome 537 examined here, there are no grounds for assuming that prosodic aspects of speech and language processing are age-appropriate in Williams syndrome. In comparison to the younger, language comprehension matched controls, both subjects again showed a pattern of deficit: each scored below 1.5 SDs on 5/16 sub-tests. There were two sub-tests (chunking input form, affect output function) on which both B and C performed well below the language comprehension matched controls. This suggests that aspects of prosodic comprehension and/or expression may be particularly challenging for people with WS—more so than aspects of language that have typically been assessed, such as grammatical comprehension. Although both subjects displayed a prosodic deficit as measured by PEPS-C, the pattern of deficits is rather different between the two. For example, while B scored at ceiling on affect input function, on the same task C scored around 3 SD below both sets of controls. On the other hand, B struggled with chunking output function, whereas C performed relatively well compared with both control groups. Subject B’s difficulties compared to CA controls were pervasive, but when compared to LC controls B was found to have particular difficulty within the area of output function. This means that he is predicted to have difficulty using prosodic resources in a meaningful context. In terms of his ability to communicate, this is an important finding as the functional domain is the most noticeable to the listener and concerns the speaker’s ability to express the meaning conveyed by the four functions of intonation. This type of difficulty could be described by the term prosodic disability (Brewster, 1989, p. 179), the inability to ‘‘deploy the prosodic resources of … language appropriately’’. While subject C, like B, demonstrates pervasive difficulties across the profile when compared to CA controls, the comparison with LC controls revealed a specific difficulty with affective prosody, with low scores on the input function, output function and output form tasks. This is consistent with informal observation of his prosodic output in conversation, which appeared to lack emotional expression. It contrasts, however, with the little existing research in this area which reports that although ‘‘prosody has not been systematically studied … it appears to be preserved, though possibly over rich in affect tone’’ (Trevarthen, Aitkin, Papoudi, & Robarts, 1998, p. 35). Conclusions Our results indicate that the prosodic abilities of the two subjects with WS are not at an age appropriate level. These findings furnish further evidence, this time from prosody, against the received view that the linguistic skills of individuals with WS are age-appropriate, or at least superior to non-verbal performance. Moreover, there is some indication that aspects of prosodic processing may be relatively more impaired than some other linguistic abilities, as revealed by comparison with controls matched for grammatical comprehension. It may therefore be important to take particular note of prosodic aspects when assessing the communicative abilities of individuals with WS, and when considering management strategies. Finally, although based on a study of just two people with WS, our research shows that a marked divergence in patterns of prosodic deficit can be found across individuals, suggesting that there may be no single, specific underlying prosodic deficit associated with WS. This provides further evidence of the heterogeneity of this population with regard to their profile of linguistic strengths and weaknesses. This heterogeneity needs to be taken into account both when drawing on data from WS to address theoretical issues (e.g., regarding the modularity of language); and also when planning individual intervention and management programmes. 538 Investigating prosodic ability in Williams syndrome Downloaded By: [University of Sheffield] At: 00:09 7 April 2009 References Bellugi, U., Marks, S., Bihrle, A. M., & Sabo, H. (1988). Disassociations between language and cognitive functions in Williams Syndrome. In D. Bishop, & K. Mogford (Eds.), Language Development in Exceptional Circumstances. London: Churchill Livingstone. Bishop, D. (1989). Test for Reception of Grammar (second edition). Manchester: University of Manchester. Brewster, K. (1989). Assessment of prosody. In K. Grundy (Ed.), Linguistics in Clinical Practice. London: Whurr. Clahsen, H., & Almazan, M. (1998). Syntax and morphology in Williams syndrome. Cognition, 68, 197–198. Dunn, L., Dunn, L., Whetton, C., & Pintilie, D. (1982). British Picture Vocabulary Scales. Windsor: NFER-Nelson. Reilly, J., Klima, E., & Bellugi, U. (1990). Once more with feeling: affect and language in atypical populations. Development and Psychopathology, 2, 367–391. Semel, E., Wiig, E., & Secord, W. (1987). Clinical Evaluation of Language Fundamentals–Revised. London: The Psychological Corporation. Stojanovik, V. (2002). Williams Syndrome, Specific Language Impairment and Modularity, Unpublished PhD Thesis, University of Sheffield. Stojanovik, V., Perkins, M., & Howard, S. (2001). Language and conversational abilities in individuals with Williams Syndrome: How good is ‘‘good’’? International Journal of Language and Communication Disorders, 36, 234–239. Stojanovik, V., Perkins, M., & Howard, S. (in press). Williams Syndrome and Specific Language Impairment do not support claims for developmental double dissociations and innate modularity. Journal of Neurolinguistics. Trevarthen, C., Aitkin, K., Papoudi, D., & Robarts, J. (1998). Children with Autism: Diagnosis and intervention to meet their needs (second edition). London: Jessica Kingsley. Wechsler, D. (1981). Wechsler Intelligence Scale for Children–Revised. London: The Psychological Corporation. Wells, B., & Peppé, S. (2001). Intonation within a psycholinguistic framework. In J. Stackhouse, & B. Wells (Eds.), Children’s Speech and Literacy Difficulties 2: Identification and Intervention. London: Whurr. Wells, B., & Peppé, S. (2003). Intonation abilities of children with speech and language impairments. Journal of Speech Language and Hearing Research, 46, 5–20. Wells, B., Peppé, S., & Goulandris, N. (2004). Intonation development from 5 to 13. Journal of Child Language, 31, 749–778.