Native Language Cognate Effects on Second Language Lexical Choice

Native Language Cognate Effects on Second Language Lexical Choice Ella Rabinovich⋆▲ Yulia Tsvetkov† Shuly Wintner⋆ ⋆Department of Computer Science, University of Haifa ▲IBM Research †Language Technologies Institute, Carnegie Mellon University ellarabi@gmail.com, ytsvetko@cs.cmu.edu shuly@cs.haifa.ac.il Abstract We present a computational analysis of cog- nate effects on the spontaneous linguistic pro- ductions of advanced non-native speakers. In- troducing a large corpus of highly competent non-native English speakers, and using a set of carefully selected lexical items, we show that the lexical choices of non-natives are af- fected by cognates in their native language. This effect is so powerful that we are able to reconstruct the phylogenetic language tree of the Indo-European language family solely from the frequencies of specific lexical items in the English of authors with various native languages. We quantitatively analyze non- native lexical choice, highlighting cognate fa- cilitation as one of the important phenomena shaping the language of non-native speakers. 1 Introduction Acquisition of vocabulary and semantic knowledge of a second language, including appropriate word choice and awareness of subtle word meaning con- tours, are recognized as a notoriously hard task, even for advanced non-native speakers. When non- native authors produce utterances in a foreign lan- guage (L2), these utterances are marked by traces of their native language (L1). Such traces are known as transfer effects, and they can be phonological (a foreign accent), morphological, lexical, or syn- tactic. Specifically, psycholinguistic research has shown that the choice of lexical items is influenced by the author’s L1, and that non-native speakers tend to choose words that happen to have cognates in their native language. Cognates are words in two languages that share both a similar meaning and a similar phonetic (and, sometimes, also orthographic) form, due to a com- mon ancestor in some protolanguage. The definition is sometimes also extended to words that have sim- ilar forms and meanings due to borrowing. Most studies on cognate facilitation have been conducted with few human subjects, focusing on few words, and the experimental setup was such that partici- pants were asked to produce lexical choices in an artificial setting. We demonstrate that cognates af- fect lexical choice in L2 spontaneous production on a much larger scale. Using a new and unique large corpus of non- native English that we introduce as part of this work, we identify a focus set of over 1000 words, and show that they are distributed very differently across the “Englishes” of authors with various L1s. Impor- tantly, we go to great lengths to guarantee that these words do not reflect specific properties of the var- ious native languages, the cultures associated with them, or the topics that may be relevant for particu- lar geographic regions. Rather, these are “ordinary” words, with very little culture-specific weight, that happen to have synonyms in English that may re- flect cognates in some L1s, but not all of them. Con- sequently, they are used differently by authors with different linguistic backgrounds, to the extent that the authors’ L1s can be identified through …

Content truncated to 500 words