Computational Speech Production Models On Serial Language Processing

1390 Words Mar 27th, 2016 6 Pages
There are many computational speech production models regarding serial language processing. In these models the stages are independent and information flow is unidirectional/top-down – lemma level processes feed lexeme level processes, but not the other way around. One of the most widely known and discussed speech production models was proposed by Bock & Levelt (1994). This model presented four distinct stages of processing. The process begins at the conceptual stage, a message-level representation, which captures the idea the speaker wishes to convey. Then two stages of syntactic (grammatical) processing follow: functional processing and positional processing. Grammatical encoding begins with functional-level processing, which is subdivided into a lexical selection, where lemmas that contain grammatical information are retrieved and associated with lexical concepts. Another substage is function assignment, which determines syntactical functions and grammatical roles (subject/object), and those are assigned to words. The other component of grammatical processing is positional processing. During this stage, linear word order and constituent structure are decided (constituent assembly), and inflections and inflectional affixes are built. In addition, other ordering decisions are made at the positional level: order of adjectives, conjunctions, and function words are also inserted. Thus, positional-level processing determines both: the serial order of phrases and the order of…

Related Documents