8th Speech in Noise Workshop, 7-8 January 2016, Groningen

Effects of exposure to noise during second language consonant acquisition

Martin Cooke(a)
University of the Basque Country, ES

Maria Luisa Garcia Lecumberri
University of the Basque Country, ES

(a) Presenting
(b) Attending

The process of acquiring the sounds of a first language takes place in everyday settings that are potentially noisy and uncontrolled, yet most formal training in a second language is based on exposure to exemplars heard in quiet conditions in the language laboratory, leading to the question: could sound acquisition in noise be beneficial? Exposure to noisy tokens might help in one of two ways: by forcing a listener to focus on robust cues that on average survive masking, or through the formation of noisy exemplars. On the other hand, it could be harmful, by masking relevant information on some trials, or increasing attentional load and fatigue.

The current study addressed these issues in the context of English consonant acquisition by Spanish learners. A cohort of 86 learners undertook a pre-test involving 24-alternative forced choice consonant identification in quiet and in two masked conditions (SSN: speech shaped noise; BAB: 8-talker babble). The cohort was then divided into 4 groups. One (CON-NOISE) underwent consonant in noise training; another (CONS-QUIET) was exposed to the same consonants in quiet; a third (VOW-NOISE) heard vowels in noise; a fourth (VOW-QUIET) heard vowels in quiet. The two noise-trained groups heard exemplars in SSN at a range of SNRs. Each group underwent 10 training sessions over 5 weeks, after which they performed a post-test identical to the pre-test. The vowel-trained groups acted as a control to measure any benefits from simultaneous curriculum activities on consonant acquisition.

All groups showed gains from pre- to post-test. However, gains for vowel-trained groups were 2-4 p.p., while those for the consonant groups ranged from 10 to 14 p.p. suggesting only a small effect of other activities. The CONS-QUIET group produced larger gains when tested in quiet than the CONS-NOISE group. Conversely, the CONS-NOISE group displayed higher gains than the CONS-QUIET when tested in SSN. The noise advantage did not transfer to the BAB condition, where the CONS-NOISE and CONS-QUIET groups showed near-identical gains. No clear evidence was found of noise exposure during vowel training transferring to the consonants in noise test. These outcomes suggest that exposure to noise during training helps in identifying consonants both in noise and in quiet, with a benefit for matched noise exposure and test conditions and little evidence of a transfer to other maskers (SSN/babble) or sound type (vowels/consonants).

Last modified 2016-05-12 14:22:09