Word Etymology as Native Language Interference

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2702–2707 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Word Etymology as Native Language Interference Vivi Nastase University of Heidelberg Heidelberg, Germany nastase@cl.uni-heidelberg.de Carlo Strapparava Fondazione Bruno Kessler Trento, Italy strappa@fbk.eu Abstract We present experiments that show the influence of native language on lexical choice when producing text in another lan- guage – in this particular case English. We start from the premise that non-native En- glish speakers will choose lexical items that are close to words in their native lan- guage. This leads us to an etymology- based representation of documents written by people whose mother tongue is an Indo- European language. Based on this repre- sentation we grow a language family tree, that matches closely the Indo-European language tree. 1 Introduction In second-language writing, vestiges of the native language such as pronunciation, vocabulary and grammar are well-attested, and the phenomenon is called native language interference (Odlin, 1989). At the lexical level, the choice as well as the spelling can be indicative of the native language, through the choice of cognates, true or false friends – e.g. a writer with native language Ger- man may choose bloom cognate with blume, while a French one may choose flower, cognate with fleur. Misspellings – cuestion instead of question are also indicative, as the writer will tend to spell words close to the form from her original language (Nicolai et al., 2013). In this paper we also look at native language in- terference starting from the lexical level, but ab- stract away from the actual word forms, and focus instead on the language of the etymological an- cestors. The hypothesis we investigate is that the collective evidence of etymological ancestor lan- guages are indicative of the language of the native speaker, and that this effect is sufficiently strong Eng.: flower Eng.: bloom Eng.: blossom ↓ ↓ ↓ Middle Eng.: flourMiddle Eng.: blome Middle Eng.: blosme ↓ ↓ ↓ Anglo Norm.: flur Old Norse: bl¯om Old Eng.: blostm ↓ ↓ ↓ Latin: florem Proto Ger.: *bl¯omˆo Proto Ger.: *bl¯ostama ↓ ↓ ↓ Proto IE: *bhleh3 Proto IE: *bleh3 Proto IE: *bhleh3 − s− Figure 1: Examples of etymological ancestry from Wiktionary to allow us to rebuild an Indo-European language family tree. We use a corpus of essays written by English language learners, whose native language cover the languages from the Indo-European fam- ily. Etymological information is provided by an et- ymological dictionary extracted from Wiktionary. The fact that word etymologies are included in the native-language interference phenomena could be used in various ways, i.a.: (i) to influence the selection of material for language learning, by fo- cusing on vocabulary closer etymologically to the native language of the student, thus facilitating lexical retention; (ii) to reveal lexical choice er- rors caused by “etymological interference”; (iii) together with other interference phenomena, for the automatic corrections of language errors. 2 Related Work English is a widespread common language for communication in …

Content truncated to 500 words