Intrinsic advantage for canonical forms in spoken word recognition: myth or reality?

ABSTRACT

In connected speech, many words are produced with a pronunciation that differs from the canonical form. How the speech recognition system deals with this variation is a fundamental issue in the language processing literature. The present study examines the roles of variant type, variant frequency, and context in the processing of French words with a canonical (schwa variant, e.g. semaine “week”) and a non-canonical pronunciation (no-schwa variant, s’maine). It asks whether the processing of canonical pronunciations is faster than the processing of non-canonical ones. Results of three lexical decision experiments reveal that more frequent variants are recognised more quickly, and that there is no advantage for canonical forms once variant frequency is accounted for. Two of these experiments further failed to find evidence that the context in which the words are presented modulate the effect of variant type. These findings are discussed in the light of spoken word recognition models.

KEYWORDS:

• Spoken word recognition
• phonological variation
• exposure frequency
• French schwa
• context

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. The authors perform an analysis in which they predict the difference between the two variants in acoustic duration by the difference in response times between the two variants and report a non-significant outcome. This is however not sufficient to show that the effect of canonical form in other analyses is not due to differences in duration. This is a non-significant result, obtained in an analysis which is independent from the other analyses.

Additional information

Funding

The authors acknowledge funding from the Swiss National Science Foundation [grant number 100014_159374] to A. Bürki.