Influence of phonology and individual differences in adults’ statistical word learning

Ge, Yuxin and Rebuschat, Patrick and Monaghan, Padraic (2024) Influence of phonology and individual differences in adults’ statistical word learning. PhD thesis, Lancaster University.

Text (2024GePhD)
2024GePhD.pdf - Published Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (4MB)

Abstract

Language learners often acquire vocabulary rapidly from their environment, usually without explicit instruction. One explanation for this process is that learners track input statistics to map words to meanings, a mechanism known as cross-situational statistical word learning (e.g., Escudero et al., 2022; Monaghan et al., 2019; Rebuschat et al., 2021). Despite its efficacy, limited research has explored how the phonological properties of words interact with the statistical learning process, especially in second language acquisition. This thesis presents a series of studies investigating the effects of phonological overlap and learners' native languages on the statistical learning of novel, non-native words. Additionally, it takes into account the factors that predict individual differences in word learning performance. In Study 1, English-native and Mandarin-native participants were trained with Mandarin tonal pseudowords via a cross-situational, statistical word learning (CSWL) task (Yu & Smith, 2007). The CSWL task contained ambiguous word-referent mappings: participants were presented with two referent pictures and one pseudoword in each trial, and they had to decide which picture the spoken word referred to by tracking the word-picture co-occurrences across trials, with no feedback provided. It was found that similar-sounding words (i.e., minimal pairs) were harder to acquire, and words that contrast in a non-native phonological feature (i.e., tonal minimal pairs for English-native speakers) were even harder. Study 2 extended the CSWL task by doubling the number of trials to test whether extended training facilitated learning. Results suggested that doubled exposure did not significantly improve English-native speakers’ word learning performance. Study 3 targeted the heritage speaker population and explored whether early exposure to Mandarin tones promotes tonal minimal pair learning later in life. It was observed that heritage speakers of Mandarin, who were exposed to Mandarin at an early age but then acquired and became dominant in English, also showed difficulty with tonal minimal pair learning. Study 4 examined whether the difficulty with non-native minimal pairs may be modulated by individual differences in lower-order, domain-general auditory processing ability (encoding and reproducing fundamental acoustic features; Mueller et al., 2012). Results indicated that more precise auditory processing (pitch discrimination and melody reproduction) was associated with better learning of non-native tonal words. Overall, the findings demonstrated the significant influence of phonology in implicit, statistical word learning. Additionally, variations in learning outcomes can be partially explained by individual differences in auditory processing ability.