Loanword formation: a neural network approach

Loanword phonology seeks to model the process by which foreign words are ‘nativised’ or incorporated into the phonological system of the ‘borrowing’ language. We can conceive of this as a parsing of the phonetic input provided by the foreign word forms, in accordance with phonological output constraints of the borrowing language. Following Silverman (1992) we conceive loanword formation as fundamentally a twostage process: the first of which yields a parsing of the phonetic input into segmentally organised phonetic feature bundles, interpretable as segmental targets in the borrowing language. In the second stage of processing, these segmental targets are parsed into phonological structures (syllables, mora, feet...etc) compatible with the word-prosody of the borrowing language. Japanese borrowings from English provide a good test-bed for models of loanword formation, because examples are abundant and, while the segmental mapping from English to Japanese is relatively straightforward, their respective word-level prosodies are strikingly different, providing ample opportunity to observe prosodic restructuring in loanword formation. The theoretical aim of this study is to investigate the learning mechanisms required for phonological parsing in loanword formation. Phonologists have typically sought to construct ‘symbolic’ or ‘analytical’ parsing algorithms for this task, either with or without reference to a framework of learnability theory. This approach came under strong challenge from ‘empirical’ Connectionist models of language processing in the 1980's. The debate between these competing paradigms, or the search for some suitable ‘hybrid’, continues. It may be argued that loanword formation provides a more restrictive, and hence better controlled, environment for studying parsing mechanisms than other natural language processing tasks, which involve a host of lexical, morphological, syntactic or pragmatic influences. In this paper, we report the results obtained from a feed-forward neural network, trained on an 1100 word corpus of American English loanwords `borrowed’ into Japanese, postWWII. We make preliminary comparisons with an analytical constraint-based approach to modeling loanword formation. An intended application for the NN parser, was to devise an English-Japanese proper or place-name translator, which would map an English phoneme sequence into Katakana (e.g.: Brisbane /brwzbcn/ => L^X=^M^] ). Introduction The Japanese language has borrowed thousands of words from English, particularly since World War II, under the overwhelming economic and cultural influence of the United structure of the English source word is States. This massive borrowing over a converted to Japanese moraic timing. English comparatively short period of time provides a tense (long) vowels and diphthongs are unique window on processes of loanword usually treated as two-mora units, whereas lax formation. Borrowed words may reflect (short) vowels are assigned to a single mora. varying degrees of nativisation to the (Stress pattern in the English source word phonological patterns of the borrowing plays a moderating role. Tense vowels or language. diphthongs in unstressed syllables may While the segments and phonetic features of [clwmpwk].) English words tend to be remarkably well preserved by the process of loanword English voiceless obstruents following a lax formation, the resulting Japanese word forms vowel are almost always treated as geminate are so completely transformed in terms of (two mora) stops. But voiced obstruents their prosodic structure that English listeners usually do not geminate in this environment in almost invariably fail to recognize their loanword formation. Also, if the preceding English sources, when loanwords are vowel is tense (long) in the English source presented to them as isolated words carefully word, gemination of voiceless obstruents does spoken by a native speaker of Japanese not occur in the loanword. The reasons for (Ingram, 1998). The main factor underlying these timing changes need not concern us poor recognition of the English source words here. appears to lie in the extensive resyllabification, involving vowel epenthesis, The segmental mapping from the English which is required to parse the segmental input source word to segments in the Japanese loans into Japanese prosodic frames. Some is basically one-for-one, without any feature examples are given below: exchange between adjacent segments. There orthographic phonemic Japanese Olympic climpik oriNpikku