THE PRODUCTION OF RHOTIC SOUNDS BY BRAZILIAN SPEAKERS OF ENGLISH

Abstract

This study investigates how three Brazilians surface the rhotic sounds in free speech when speaking English, taking into consideration the phonological environment, the frequency,  and the occurrence of patterns. Attempts are made to relate the differences in pronunciation to possible theoretical explanations (the Markedness Differential Hypothesis (MDH) [Eckman, 1977], the Structural Conformity Hypothesis (SCH) [Eckman, 1991], and the Speech Learning Model (SLM) [Flege, 1995]). The data shows that the SLM appears to explain much of the participants’ difficulty, since L2 learners are not able to separate their L1 and L2 phonetic subsystems. In two-member onsets, for example, participants recognize the English rhotic, but fail to surface it phonetically accurately (e.g., central [ˈs ɛ nt ɾ aw]). The MDH might explain the great number of two-member onsets/codas and the very few three-member onsets/codas, since the former are less marked. It might also partially account for the variability of retroflex liquids in syllable-final (e.g., York [ ˈi ɔ Xki]), since the retroflex is the most marked sound among the rhotic sounds analyzed (Maddieson, 1984). However, the MDH cannot encompass all the complexities found in the participants’ interlanguage phonological processes. The SCH also appears to fail to provide an explanation of the participants’ interlanguage processes, since generalizations found in children’s acquisition of the retroflex diverge from the processes identified in the participants’ interlanguage.