Comparative L2 English [r]Cluster Production by Cantonese and Mandarin Speakers

English initial C[r] clusters are often mispronounced by Cantonese and Mandarinspeakers. Cantonese speakers tend to delete the [r] (e.g., [pɪnt] for print) or substitute it to [w](e.g., [kwaɪ] for cry). Mandarin speakers, on the other hand, tend to insert sounds as theypronounce print as [pu’rɪnt]. Studies on L2 Cantonese and Mandarin have been mostlydescriptive and pedagogical studies addressing their English pronunciation (Chan & Li, 2000;Hung, 2002; Au, 2002; Broselow et al., 2002; Lee, 2003; and Chan, 2006a). Among themChan (2006a) and Lee (2003) particularly investigated Cantonese and Mandarin productionerrors of English clusters: the former observed deletion and substitution in Cantonese English;and the latter found deletion and epenthesis in Mandarin English, respectively. However, nofurther distinction on the distribution regarding articulation has been made.Best’s Perceptual Assimilation Model (1995, 2001) suggested that greater similarity of aL2 sound to a L1 sound would increase mapping of that sound onto L2. Brown (2000) andWang (2008) further developed Best’s model in the perspective of articulation and prosody. InBrown’s words, ‘proximity of constriction location’ defines similarity, which enables onesound in different articulatory environments to have different mappings. Wang, on the otherhand, focuses on prosody and states that perceptual difference of prosody of Mandarin andEnglish, such as tones, would cause more variations in the mapping. Thus, this study is toinvestigate the distribution of errors of Cantonese and Mandarin productions of Englishclusters by constriction location and prosody.The participants of the study include ten Cantonese speakers and ten Mandarin speakers intheir twenties. Stimuli differ in place of articulation of the initial consonant, the vowel quality,and the syllable structure (e.g., trip, free, greet). Words were put into short paragraphs forparticipants to read, and then recorded and digitized for the spectrogram analyses with Praat.First, the formant transition between the initial consonant and the vowel were examined: F2and F3 were measured to identify the [w] substitution; and change of formants was used toidentify deletion/epenthesis. Furthermore, pitch, intensity and duration patterns of each tokenwere studied against the sound data in Ladefoged (1993) for any significant differenceamongst the three prosodic factors.Preliminary results show that most Mandarin speakers were adapted to epenthesiswhereas Cantonese speakers were more adapted to deletion and substitution. However,mispronunciations were mostly in the labial and velar clusters but seldom occurred in thealveolar clusters. Such distinction could be explained by Brown (2000)’s articulatoryproximity under the Articulatory Phonology. In the C[r] clusters, it is suggested that gesturesor [r] are reduced to a [lip] gesture only, forming a [w] in labial clusters as [r] has a [lip]gesture similar to that of [w]. Velar consonants, which share a common gesture [dorsal] with[w], may also trigger substitution. Findings on the prosodic effects show that Mandarinspeakers’ English involves more contour tones in overall pitch pattern whereas Cantonesespeakers mainly use level tones. Given the fact that contour tones require more tone-bearingunits (mostly vowels) than level tones, and that a greater demand of vowels may triggerepenthesis, deletion by Cantonese speakers and epenthesis by Mandarin speakers can beexplained as different speech strategies influenced by the prosody of the learners’ nativelanguages.