|Title:||Ranking Transfer and Constraint Emergence in the Interlanguage|
|Abstract:||Evident throughout the SLA literature is the assertion that phonological properties of the L1 are often transferred into the L2, and these properties are particularly apparent during the learning stages (Gass & Selinker 1994). Previous accounts of this transference using Optimality Theory (Prince & Smolensky 1993) include Broselow et al. (1998), who argued that unmarked structures that deviate from both L1 and L2 could occur in the interlanguage phonology. The aim of this paper is to determine the ranking of interlanguage English phonology of native speakers of Brazilian Portuguese (henceforth BP), adopting an OT framework. I show that there is transfer of the L1 ranking and structures that deviate from both the L1 and L2 appear in the interlanguage English.
First, the syllable structures of BP and English are contrasted. It was found that BP is considerably more restrictive in terms of permissible consonants and consonant clusters in the coda position than English. BP allows only [s] and [r] to surface in the coda position. The lateral /l/ is glided to [w] and syllabified in the nucleus, and /n/ regressively assimilates its [nasal] feature onto the vowel and is deleted. Thus, BP has a high ranking CODACOND that ranks low in English. An OT account of BP syllable structure determined the constraints and their ranking at work in BP phonology.
Five native speakers of BP were recorded reading English sentences, which contained target words whose output constructions deviated from the typical output constructions of their L1. The data was analyzed acoustically on spectrograms using WinSurfer 095. It was evident throughout the data that the same processes that occurred in the L1 were transferred into the interlanguage English. For example, when /n/ occurred in the coda position of the target word, the feature [nasal] was regressively assimilated onto the vowel and /n/ was deleted, and when /l/ occurred in the coda position of the target word, it was glided to [w].
However, when a complex coda like [nt] occurred in the target word, [t] did remain in the coda in the IL, a construction not allowed in the L1. This follows Broselow et al. (1998), where, because of the richness of the base in the L2, it is possible for structures not native-like to the L1 to surface in the interlanguage.
The interlanguage constraint ranking thus deviates from both the L1 and L2 rankings. Although CODACOND plays a role in the IL, in order to retain [t] in the IL output, MAX-IO (OBS) ranks high, even though its ranking in BP is uncertain because BP does not provide obstruents (with the exception of /s/) in the input. Otherwise, if only CODACOND was ranked high in the interlanguage then the expected output form of /plant/ would be [pla~](a~ denotes a nasalized vowel). Also, constraints militating against the complete featural loss of the segments must be ranked high in the interlanguage, as they are in BP. As the language learner continues to learn and acquire the target language they begin to rerank the constraints from a ranking most like their native language to one that more resembles the target language. It is this continuous reranking of the constraints during the acquisition process that allows OT to account for the acquisition of the phonological properties of the target language. This paper is the first to apply OT to BP syllable structure and the interlanguage coda constructions of BP speakers learning English.
|Area/Keywords:||Phonology, Language Acquisition, Phonetics|