Skip to main content icon/video/no-internet

Language Production, Incremental Processing in

Language production involves the generation of several successive levels of representation. These representations map between the thought to be expressed and the motor commands that articulate it. Current models of language production are in broad agreement about the nature of these representations. Speaking begins with the construction of a conceptual representation of the meaning to be expressed. This representation triggers grammatical encoding processes, which select the appropriate words from the mental lexicon and generate a syntactic structure to fix their linear order. Phonological encoding processes then generate the abstract sound structure of the utterance prior to articulation.

Models of language production agree that speakers need not generate all levels of representation for an entire utterance before beginning to speak. They propose that fluent speech output is accomplished by incremental processing, so that the articulation of early parts of an utterance occurs in parallel with the planning of upcoming segments. However, exactly how processing at different levels is coordinated remains a matter of dispute. In particular, there is disagreement about how much of an utterance must be generated at a particular level of representation before processing at the next level can begin.

This entry will introduce different proposals about the degree of incrementality operating during speech production. It will discuss the relevant findings for processes involved in the generation of grammatical and phonological structure.

Generating Grammatical Structure

There is conflicting evidence about how much of an utterance must be grammatically encoded prior to articulation. Some theories claim that the verb of a sentence must be retrieved before the generation of syntactic structure can commence. Verbs control the structure of clauses and there is evidence that the clause operates as a planning unit during speech production. Planning pauses in speech have been shown to occur more frequently between clauses than clause internally. Speech error data have been used to claim that the words for a whole clause are retrieved prior to speech onset. Word exchange errors such as “put the drawer in the cloth” suggest that the exchanging words drawer and cloth are retrieved in parallel. However, it is also proposed that these effects could occur during conceptual, rather than grammatical, processing.

Experimental evidence exists for tightly incremental planning. In recent years, eye-tracking studies have recorded the gaze patterns of speakers while they describe pictured scenes. These experiments demonstrate that speakers almost always fixate on pictured objects in their order of mention and rarely look ahead at objects to be named later (although some peripheral processing of immediately adjacent objects can occur). The time spent looking at an object is a function of the ease with which it can be identified, as well as of the ease with which the phonological form of its name can be retrieved. These findings suggest that we plan speech word-byword with planning progressing only slightly ahead of articulation. However, these experiments usually involved the production of one sentence structure to one fixed pattern of pictures. It is possible, therefore, that the observed visual and linguistic processing patterns are strategic rather than typical.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading