Who is Fu? – Perception of L2 sounds that are partially neutralized in L1

Abstract

This study investigates the interference of L1 phonology with L2 speech perception, focusing on the identification of the English fricatives /f/ and /h/ by Japanese listeners. In Japanese, the contrast of /f/ vs /h/ is lost only in the /_u/ environment by neutralization (/fu/ and /hu/ > [ɸu]). This makes food vs who’d extremely difficult for Japanese to discern while fee vs he poses no problem. Perceptual experiments demonstrated that while /f/ and /h/ were accurately identifiable when isolated from non-/u/ syllables [e.g. f(a) vs h(a)], those isolated from /_u/ were rather hard [f(u) vs h(u)] and became even harder when presented with the vowel [fu vs hu]. Such Cs alone [f(u)/h(u)] were easier than the whole CVs [fu/hu] from which those Cs were excised, contrary to our general expectations. A cross-splicing experiment further revealed it was not the acoustics of coarticulated fricatives but the presentation of /u/ that made the identification difficult. The basic phonetic process of f/h was debilitated in fu/hu, where the L1 neutralization applied, cancelling the contrast and the need for attunement, which was then transferred to L2. It was argued that the L1-L2 sound correspondence can be affected by the knowledge of L1 phonological processes.

Introduction

This study examines the effect of L1 phonology on the perception of L2 sounds. Specifically, it addresses the difficulty for Japanese listeners to identify the English fricatives /f/ and /h/ in the /u/ environment as in /fu/ and /hu/; e.g., feet vs heat is easy, but food vs who’d can be extremely difficult to discern. The two English fricatives are generally contrastive in Japanese, but they are neutralized in the /u/ environment and become non-contrastive (Vance, 2008). Therefore, the said difficulty in identifying the fricatives in the similar environment in English may be attributed to the negative transfer of the L1 phonology, namely the partial, vowel-context-dependent neutralization (hereafter “partial” neutralization), onto L2 speech perception.

Two L2 phonemes are assumed to be easy to identify when there are also two corresponding phonemes in L1 (Best & Tyler, 2007). The English fricatives /f/ and /h/ are assimilated to /ɸ/ and /h/ respectively in Japanese (Guion et al., 2000, Vance, 1987). The bilabial /ɸ/ is a phoneme which was once-extinct in classic times but was revived to represent the foreign /f/ sound in loanwords. The use of /ɸ/ is indispensable and well-established in modern Japanese (See 2.1 for details).¹ Therefore, the contrast between /f/ and /h/ in English is generally preserved in Japanese as /ɸ/ and /h/. For example, English /fɑ/ vs /hɑ/ translates into Japanese /ɸa/ vs /ha/, which are contrastive and readily identifiable. However, /h/ in its allophonic variation becomes [ɸ] when followed by /u/ in Japanese (/hu/ > [ɸu]), and this is neutralized with another [ɸu] (loan /fu/ > /ɸu/ > [ɸu]) (e.g. Takebayashi, 1997, Vance, 2008), as illustrated in Fig. 1. Due to this neutralization, there is no contrast between [ɸu] of /ɸu/ representing foreign /fu/ on one hand and [ɸu] derived from /hu/ on the other. As a result, this can cause tremendous confusion to Japanese listeners in identifying the English correspondents in similar environments as in /fu/ ‘Fu’ vs /hu/ ‘who.’

Such difficulty has been experienced by EFL instructors (Nogita, 2010) but described only in the limited literature (e.g. Kondo et al., 2015, Takebayashi, 1997, Walker, 2010). Moreover, there is not much empirical research on the perception of those fricatives by Japanese listeners. In one of only a few studies on this issue, Lambacher, Martens, Nelson, and Berman (2001) tested Japanese listeners on English fricative identification in nonce syllables and demonstrated low identifiability of /f/ and /h/ in /u/ conditions. Their data and analyses are an objective presentation of the phenomenon, but they did not go into the detailed causes of it.

We recognize the vowel context is certainly a factor. However, it is yet to be known (i) whether Japanese listeners can identify the fricatives in real words; (ii) whether they are able at all to identify the isolated fricatives alone; and (iii) whether they are less sensitive to the fricatives excised from the /u/ context or if it is the presentation of the following vowel /u/ that causes perceptual confusion since /f/ and /h/ (assimilated into /ɸ/ and /h/) do not need to be differentiated in that environment in L1.

The current study employed a series of experiments to answer the above questions with respect to Japanese perception of English /f/ and /h/. The experiments will help elucidate what causes the said difficulty: whether it is (i) the acoustics of fricatives /f/ and /h/; (ii) the coarticulatory effect from the following vowel; (iii) or the L1 phonological influence arising from the /fu, hu/ sequences. These acoustic, phonetic and phonological levels are known to be fundamental factors involved in L2 speech perception. This study also provides data of linguistic interest on how L2 sounds are processed when they are generally contrastive (phonemes) but partially neutralized to be non-contrastive (allophones within those phonemes) in L1. Furthermore, we will discuss our results in wider perspectives in the light of several L2 speech perception and learning models, including the Perceptual Assimilation Model (PAM, Best, 1995, Best and Tyler, 2007), PAM-L2 (Best and Tyler, 2007, Tyler, 2019), Speech Learning Model (SLM; Flege, 1995), and the phonological inference model by Gaskell and Marslen-Wilson (1998). This may allow us to posit that the knowledge and experience of L1 phonological processes can override basic comparisons of L1/L2 sounds on the acoustic level as well as formation of L1-L2 sound correspondences on the whole-phoneme level.