Growing up around different accents: the effect of speech variability on infant word recognition

父母的言语在婴儿的语言习得中起着至关重要的作用(Bruner，1983)。在许多家庭中，父母说话的口音不一样。不同的口音(地区口音或外国口音)会导致语音差异，即同一个单词的发音有两种不同的方式，这意味着婴儿收到的输入是不一致的。之前的研究对于这种不一致对婴儿语言学习的影响提出了相互矛盾的发现。婴儿早在五个月大的时候就能够区分熟悉和不熟悉的地区口音(Butler等人，2010年)。然而，这种敏感性对投入可变性的影响尚不清楚。一方面，接触不同的口音可能有助于理解其他不熟悉的口音。这一点已经在许多基于实验室的研究中得到了证明，在这些研究中，来自单一口音背景的孩子短暂地熟悉了一种外国口音，然后又接触到了另一种不熟悉的口音。例如，来自美国英语背景的18个月大的婴儿在接触过澳大利亚英语、南美英语和印度英语后，识别英国英语单词的能力更强(Potter&Saffran，2017)。此外，短暂接触特定的陌生口音有助于这种口音的单词学习(Schmer等人，2012)和单词识别(Paquette-Smith等人，2020)。在实验室之外，当出生时就接触到两种不同的口音时，Kartushina等人说。(2021)显示，在双重口音家庭长大的婴儿仍然更善于学习以陌生口音呈现的新词。然而，对于熟悉的单词识别，还没有得出任何结论。另一方面，接触两种不同的口音可能会延迟单词识别。例如，在家中同时接触加拿大口音和非母语英语口音的两岁儿童在识别加拿大英语单词方面比来自单一口音背景的婴儿更差(Buckler等人，2017年)。与来自单一口音环境的婴儿相比，出生时接触两种口音的婴儿对单词发音错误的敏感度也较低(Durant等人，2015)。因此，在实验室中操纵口音暴露的研究与那些从出生起就观察多重口音暴露的研究在单词识别方面出现了差异。这可能是由于长时间和短暂的多口音暴露之间的真正差异，也可能是由于方法上的差异。到目前为止，关于这一主题的所有研究都是基于行为衡量标准。然而，使用事件相关电位(ERP)数据来测量婴儿的单词识别是可能的(Korpiahti等人，2001年)。在现有的研究中加入事件相关电位的测量方法，可以帮助调和由于行为测量方法的不同而产生的任何差异。本项目旨在通过结合行为测量和脑电测量，系统地研究多重口音暴露对婴儿单词识别的影响。这项研究将包括三个阶段，一个阶段是检查来自单一和多种口音背景的婴儿的单词识别，下一个阶段是在实验室中介绍对口音暴露的操纵，最后一个阶段是调查暴露在不同地区和外国口音变化中的婴儿的单词识别技能是否存在不同的调节。这将清楚地区分短期和长期接触一种以上口音的影响。

Parents' speech plays a crucial role in infants' language acquisition (Bruner, 1983). In many families, parents do not speak with the same accent. Different accents (regional or foreign) introduce speech variability, where the same word is pronounced in two different ways, meaning that the input received by the infant is inconsistent. Previous research presents conflicting findings as to the effect of this inconsistency on infants' language learning. Infants are able to distinguish familiar and unfamiliar regional accents as early as at five months of age (Butler et al., 2010). The effect of this sensitivity to input variability, however, is unclear. On one hand, exposure to different accents may facilitate understanding of other, unfamiliar accents. This has been shown in a number of lab-based studies where children from mono-accent backgrounds are briefly familiarised with a foreign accent, and then exposed to yet another unfamiliar accent. For example, 18-months-olds from an American English language background were better at recognising British English words after they had been exposed to Australian, Southern American and Indian English (Potter & Saffran, 2017). Moreover, brief exposure to a particular unfamiliar accent facilitates both word learning (Schmale et al., 2012) and word recognition (Paquette-Smith et al., 2020) in this accent. Outside of the laboratory, when the exposure to two distinct accents happens from birth, Kartushina et al. (2021) showed that infants raised in bi-accent households are still better at learning novel words presented in unfamiliar accents. No conclusion, however, has been reached about familiar word recognition. On the other hand, exposure to two different accents may delay word recognition. For example, two-year-olds exposed to both Canadian and a non-native English accent at home are worse at recognising words in Canadian English than infants from mono-accent backgrounds (Buckler et al., 2017). Infants exposed to two accents from birth are also less sensitive to word mispronunciations than infants from mono-accent environments (Durrant et al., 2015).A discrepancy, therefore, emerges between studies manipulating accent exposure in the laboratory and those looking at multi-accent exposure from birth, regarding word recognition. This could either be due to a genuine difference between prolonged and brief multi-accent exposure or due to methodological differences. All research on the topic to date has been based on behavioural measures. It is possible, however, to measure word recognition in infants using event-related potentials (ERP) data (Korpilahti et al., 2001). Adding ERP measures to the existing body of research could help reconcile any discrepancies that may arise from the differences in behavioural measures.This project aims to systematically examine the effect of multi-accent exposure on infant word recognition by using a combination of behavioural and EEG measures. The study will consist of three phases, one examining word recognition in infants from mono- and multi-accent backgrounds, the next one introducing a manipulation of accent exposure in the laboratory, and the final one investigating whether exposure to regional and foreign accents variations modulate differently infants' word recognition skills. This will allow for a clear distinction between the effects of brief and prolonged exposure to more than one accent.