Machine translation - a view from the lexicon

Books describing novel approaches to machine translation (MT) are always welcome. This is all the more so when the approach is one not covered by general MT surveys such as those in Hutchins and Somers (1992) or Arnold et al. (1994). Bonnie Jean Dorr's Machine Translation: A View from the Lexicon is a book with a novel approach. It describes the interlingual MT system UNITRAN rooted in two Massachusetts-based frameworks of theoretical linguistics: Chomskyan principles-and-parameters government-binding (GB) theory for the syntactic component and Jackendovian lexical conceptual structure (LCS) for the lexical-semantic component, which also serves as the interlingua. The main claim is that cross-linguistic lexical-semantic divergences between source and target languages (at least across basic English, Spanish, and German) are of roughly only seven types, thus leading to a simple systematic translation mapping (relating the interlingua to the corresponding syntactic structures) parameterized by switches, with no language-specific rules. Besides introductions, conclusions, and appendices, the book is organized into three parts encompassing UNITRAN's syntactic component, its lexical-semantic component, and application of the model. Chapter 1 is an introduction to the book. It briefly describes the basics of MT, including alternative approaches, and attempts to justify Dorr's parameterized interlingual principle-based design. It also begins a preliminary discussion of translation divergences, such as the lexical-semantic categorial type as in the English I am hungry, in which the predicate hungry is adjectival, compared with the German Ich habe Hunger ('I have hunger'), in which the corresponding Hunger is nominal. Chapters 2 and 3 form the part dealing with the syntactic component. The former discusses the implementation of GB modules coupled with the parameters particular for each of English, Spanish, and German. The latter deals with the two-level morphological processor used in UNITRAN for analysis and generation. Chapter 4, the first in the part dealing with the lexical-semantic component, is to a large extent a variant of Dorr (1993a). It describes the interlingual representation of UNITRAN. The chosen interlingua is an extended version of LCS, used also as a representation of lexical entries. Dorr justifies this choice as follows: