Patterns of speech sounds after surgery: investigating infants' phonological development following full cleft palate restorative surgery.

在英国，每700名婴儿中就有一名出生时患有唇裂和/或腭裂(NHS，2020)。大多数患有腭裂(CP)的婴儿都接受了全面的修复手术，但不一定会继续完全获得目标类语音(Hardin-Jones，2003，Kummer等人，2012)。Wren(2017)显示，这些婴儿中有多达50%的人从未达到完全的语言熟练程度，即使在完全重建腭部后也是如此。这项研究旨在通过对早期发声的分析，更好地理解为什么成功接受腭裂修复手术的婴儿可能会持续存在语音困难。该方法整合了Vihman(2014：121)提出的已建立的理论--发音过滤假说(AFH)，该假说认为婴儿注意到输入语音中与他们的发音相匹配的单词，这反过来又鼓励了音素-喋喋不休相似的匹配行为(反馈环)，从而鼓励进一步的发声。AFH突出了早期言语产生的意义，语音反馈回路为后来的语言进步提供了关键的基础。关于婴儿发声和他们的知觉之间关系的研究还没有探索这种关键的联系是如何在腭裂修复后建立的。对AFH的深入了解可以表明与照顾者输入的声学/语音特性相匹配的重要性，以进行类似目标的生产。研究问题：在13-14个月大的婴儿和腭裂修复后的婴儿中，典型发育的婴儿和婴儿之间，BABBLE的声学-语音特性(即，BUBLE辅音的发音和语音特性)有什么不同？脑瘫婴儿在多大程度上对照顾者在他们自己的胡言乱语中说话的声音做出反应(遵循莱恩和伯格尔森，2020年的框架)？喋喋不休和照顾者输入匹配的模式在多大程度上预测了24个月和36个月后的言语和语言结果？其中一个是比另一个更好的预测者吗？数据将来自裂隙集体言语和语言语料库(CC-SL)，其中包括接受完全CP恢复的讲英语的婴儿(Wren等人)。2017年)。CC-SL包含了从出生到5岁的600多名儿童的数据。数据是通过语言环境分析或Lena-Device(Lena Research Foundation，2018)收集的，它捕获了照顾者在婴儿环境中的言语。关键变量包括人口统计学、心理学、外科手术、听力和语音/语言数据。来自DePaolis等人的48名TD婴儿的语料库。(2010)将作为有价值的对照样本；它包括一组14个月时母子互动的自然记录，这些记录将与CP数据进行对比，以确定群体差异。CC-SL的语言结果数据(24个月和36个月)将测试不同CP婴儿的个体差异，以评估他们是否可以预测语言结果。该方法将利用40名CP婴儿(手术后)和40名TD婴儿(对照组)的样本来分析语音发育，并确定叽叽喳喳中的发音特性是否可以预测CP婴儿未来的语音结果。描述性统计和语料库分析方法将用混合方法探索和分析CP和TD婴儿的两个主要变量：早期发声时婴儿发声的声学-语音特性，以及照顾者输入和婴儿发声之间的偶然性。分析将包括以下阶段：-声学语音分析--使用PRAAT(Boersma等人)。2019年)-辅音库存13-14个月。-语音转录--使用Elan(2020)--调查婴儿的生产。-婴儿发声之前一致/不一致的照顾者输入的转录，改编自莱恩和贝格尔森(2020年)。-比较13-14个月的辅音库存与24个月和36个月的语音和语言结果

In the UK, 1 in every 700 babies are born with cleft lip and/or palate (NHS, 2020). Most infants with cleft palate (CP) receive full restorative surgery, but nevertheless will not necessarily go on to fully acquire target-like speech (Hardin-Jones, 2003, Kummer et al., 2012). Wren (2017) shows that as many as 50% of these infants never reach full speech proficiency, even after complete palate reconstruction. This research will aim to better understand why speech difficulties might persist in infants who have had successful cleft palate repair surgery through an analysis of early vocal production. The methodology integrates an established theory-the articulatory filter hypothesis (AFH)-presented by Vihman (2014:121) which proposes that infants notice words in input speech that match their production, and this, in turn, encourages a matching behaviour (feedback-loop) of phoneme-babble resemblances that encourage further vocal production. The AFH foregrounds the significance of production in early speech and that the phonological feedback-loop provides a key foundation for later linguistic advances. Research into the relationship between an infant's vocal production and their perception of has not yet explored how this crucial connection is established after cleft-palate repair. Insight into the AFH could indicate the importance of matching to the acoustic/phonetic properties of caregiver input for target-like production. Research Questions:How do the acoustic-phonetic properties of babble at 13-14 months (i.e., the articulatory and phonetic properties of babbled consonants) differ between typically developing infants and infants after cleft palate repair? To what extent do CP infants respond to the sounds of caregiver speech in their own babble (following the framework set out in Laing and Bergelson, 2020)? To what extent do patterns of babbling and caregiver input matching predict speech and language outcomes at 24 and 36 months? Is one a better predictor than the other? The data will be from the Cleft Collective Speech and Language corpus (CC-SL), which includes English-speaking infants who have undergone full CP restoration (Wren et al. 2017). CC-SL contains data from over 600 children who were studied from birth to 5 years. Data was collected with a language environment analysis-or LENA-device (LENA Research Foundation, 2018), which captures caregiver speech in the infant's environment. The key variables include demographic, psychological, surgical, hearing, and speech/language data. A corpus of 48 TD infants from DePaolis et al. (2010) will serve as a valuable control sample; it includes a set of naturalistic recordings of mother-child interactions at 14 months, which will be tested against CP data for group differences. The CC-SL data on language outcomes (at 24 and 36 months) will be tested for individual differences across CP infants to assess whether they predict language outcomes.The method will draw on a sample of 40 CP infants (post-surgery) and 40 TD infants (control group) to analyse phonological development and to determine whether the articulatory properties in babble can predict later speech outcomes in the CP infants. Descriptive statistics and corpus analysis approaches will serve to explore and analyse two main variables in CP vs. TD infants with a mixed methods approach: the acoustic-phonetic properties of infant babble produced in early vocalisations and the contingency between caregiver input and infant vocalisations. Analyses will include the following stages: -Acoustic phonetic analysis-using PRAAT (Boersma et al. 2019)-of consonant inventories at 13-14 months. -Phonetic transcriptions-using ELAN (2020)-to investigate infant productions. -Transcription of congruent/incongruent caregiver input preceding infant vocalisations, adapted from Laing and Bergelson (2020). -Comparisons of consonant inventories at 13-14 months with speech and language outcomes at 24 and 36 months across CP