Neural Network Model of Lexical Organisation

Dr Michael Fortescue is at Department of Linguistics, University of Copenhagen, Denmark.

• Part 1: The Basics; Part 2: Applications; Part 3: Cognitive Justification of the Model; Appendix 1: The relationship to Burnod's neurological model; Appendix 2: Paradigmatic features of English words; Appendix 3: Sample derivations; List of templates; Graphic conventions as first introduced; References; Index.

"Michael Fortescue sets forth "to provide a bridge between neurology and theoretical linguistics" and a patient reader can only conclude that he accomplishes this rather daunting, synthetic task exceptionally well. Neuroscientifically, the author is very well-informed, building on the work of established authorities such as Deacon and Pulvermüller, but mostly on the relatively less-known model of Yves Burnod. As an established cognitive-functional linguist, however, he far surpasses such work in linguistic coverage, rigor and theoretical sophistication. In the first part of the book, Fortescue develops a neurologically plausible model of the mental lexicon, distinguishing systematically between sensori(motor) affordances, and what he calls "mirco-functional affordances" (of key importance for grammar) and "macro-functional affordances" (related to pragmatics and context), and at the same time shows how they are integrated in speech production and comprehension. An important contribution is a three-dimensional graphic representational format, which may at first appear complicated, but is clearly worth the effort, since it possesses considerably more structure than alternative (connectionist) models, and therefore possibly for the first time shows how language could be neurally realized. In part 2, the model is applied to phenomena that have concerned linguists for quite some time: semantic fields, compositionality, grammatical constructions, polysemy (including metaphor and metonymy) and cross-linguistic variation (typology), both casting new light on them theoretically, and showing how they could be treated neuroscientifically. The third and final part is reserved for the most difficult issues: the relations between lexicon, grammar and context in language use and acquisition, and concludes by specifying predictions that can be experimentally tested. All this makes this book truly deserve the rather overused epithet "ground-breaking"". - Associate Professor Jordan Zlatev, Centre for Cognitive Semiotics, Lund University, Sweden