Theories of speech production are based, to a large extent, on the Indo-European languages (e.g. Bock & Levelt, 1994; Dell, 1986; Levelt, Roelofs & Meyer, 1999). Although crosslinguistic checks of the theories have been attempted, they were carried out primarily within that language family. The purpose of this chapter is to describe some attempts to extend production research outside this family, to the Chinese language, in particular, Mandarin Chinese or Putonghua.
To produce a word, a speaker must convert a meaning to a phonetic code that can be executed by the articulatory organs. The conversion is not a one-step job, but consists of several processing levels, each involving the construction of a linguistic representation that is appropriate for (characteristic of) that level. It is typically assumed that the representations consist of two parts, a constructed hierarchical frame, and a retrieved set of discrete units which are inserted into slots in the frame. The following levels of processing have been postulated (Levelt, Roelofs & Meyer, 1999). Initially at the syntactic level, a sentence frame is constructed which is composed of a number of categorically labeled slots. The syntactic code of the word, called its “lemma,” is also retrieved. The lemma is a memory representation that uniquely identifies the word and stores its syntactic properties such as gender, number, person, tense, etc. When the lemma is retrieved, these syntactic properties determine the word's slot in the sentence frame.