Are higher-level processes delayed in second language word production? Evidence from picture naming and phoneme monitoring

ABSTRACT

There are clear disadvantages in the speed of word production and recognition in a second language (L2), relative to the first language (L1). Some accounts claim that these disadvantages occur because of a slow-down in lexical retrieval and phonological encoding. But it is also possible that the slow-down originates from a later part of the production process, namely articulatory planning or articulation. We used a phoneme monitoring task to study the time course of conceptualisation, lexical retrieval, and phonological encoding during language production in the absence of articulation. First, we demonstrated that there was indeed an L2 disadvantage of 102 ms in a picture-word interference (PWI) task with phonologically related and unrelated distractor words. Next, participants from the same population performed a combined phoneme monitoring task / PWI task with the same stimuli: they monitored for the occurrence of a phoneme in a picture name while ignoring a distractor word. In both the PWI task and the combined phoneme monitoring/PWI task, there was phonological facilitation, suggesting that both tasks are similar up to the level of phonological encoding. Importantly, L2 speakers were not slower in phoneme monitoring than L1 speakers. These findings suggest that the slow-down typically observed in L2 speech production may not be situated at phonological or pre-phonological stages of speech production, but rather in a later stage of speech production.

KEYWORDS:

• Self-monitoring
• picture word interference task
• phonemic overlap
• second language processing
• bilingualism

Acknowledgements

We would like to thank Michael Stevens and Toru Hitomi for their advice on the statistical analyses, specifically on creating and fitting the linear mixed effects models. This paper received funding from the special research fund of Ghent University (GOA – Concerted Research Action BOF13/GOA/032). We would also like to thank the members of the GOA-project for their input and suggestions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Only half of these pictures were analyzed because of the experimental design of Experiment 2. In that experiment, a phoneme monitoring task had to be performed. The phoneme was present in the picture name in half of the trials and absent in the other half. Since we wanted to keep the set-up of Experiment 1 as similar as possible to that of Experiment 2 (Experiment 2 was conducted first) we only analyzed the trials where the phoneme was present. Therefore, only half of the pictures were analyzed in the end, leading to a total of 7200 target trials (25*12*48/2 = 7200).

2. If more than 20 out of 25 trials were answered incorrectly per category (e.g. double overlap, yes answer), then the participant was excluded from the data set. Four participants answered at least 24 out of 25 trials incorrectly, indicating that they clearly misunderstood the task and were therefore excluded. Other participants showed a range from 0 to 8 incorrect trials out of 25 (although the majority of the participants only answered 1 or 2 trials incorrectly per category).

Additional information

Funding

This study received funding from the special research fund of Ghent University (GOA – Concerted Research Action BOF13/GOA/032).