Lexical activation in late bilinguals: effects of phonological neighbourhood on spoken word production

ABSTRACT

This research explores patterns of lexical activation of both languages of a bilingual when producing spoken words. Specifically, it investigates the influence of within- and cross-language measures of phonological neighbourhood, on the English picture naming performance of a diverse group of French (L1) – English (L2) late bilinguals. A novel phonological neighbourhood density measure was used that took into account the phonological and lexical constraints of this sample of late bilinguals. Results showed that both within- and cross-language neighbours influenced picture naming performance and that the length of exposure to L2 and familiarity of the target modulated these effects, with both facilitatory and inhibitory effects observed. Additionally, the previously well-attested cognate facilitation effect decreased with L2 language exposure. These results support interactive models of word production, within a language non-specific view of lexical access, and show the dynamic nature of a bilingual’s lexical connections.

KEYWORDS:

• Late bilingual
• lexical retrieval
• phonological neighbourhood density
• cross-linguistic influence
• L2 exposure
• cognate

Acknowledgements

We thank Dr Anthony Shook for helpful communication regarding the CLEARPOND database. During the preparation of this paper, the first author was funded by an Australian Research Council Discovery Project grant (190101490), and by an International Macquarie University Research Excellence Scholarship (iMQRES) in a previous version of this manuscript. In addition, we thank two anonymous reviewers for their useful suggestions aimed at improving the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 For the sake of simplicity, we refer to the population under investigation as late bilinguals or just bilinguals (other ways to refer to our population would be second language -or L2- learners). We are acutely aware of the debate around what constitutes a bilingual, and variability within this population and we do not intend to imply that our findings will necessarily generalise to all (late) bilingual speakers.

2 Forty English monolingual students from Macquarie University (Hameau, 2016; Hameau et al., 2018).

3 In CLEARPOND, French neighbours are drawn from the Lexique 2 database (New et al., 2004), using a frequency cut-off that ensures that the French neighbours are likely to be words from an educated French monolingual adult’s vocabulary (i.e. excluding rare words that potentially very few participants would know).

4 Note that in a previous version of this manuscript (as part of the first author’s PhD thesis: Hameau, 2016), we reported analyses run both with and without word length in phonemes, despite the absence of an effect on either response time or accuracy of this variable. The presence or absence of length as a predictor did not change the pattern of significance (interactions) of the experimental predictors.

5 Since Log frequency significantly predicted accuracy (as opposed to response time), we tested the interaction between frequency and PND in line with Hameau et al. (2018): this interaction did not improve the PND model: X²(1) = 0.0176, p = .894.

Additional information

Funding

This work was supported by Australian Research Council [grant number DP 190101490] and an International Macquarie University Research Excellence Fellowship (iMQRES) to Solène Hameau.