Production of Language

Producing a linguistic expression involves retrieving a set of words, arranged in a grammatical hierarchically organized sequence so as to convey speakers’ intended thoughts. This is accomplished through a series of processing steps. A speaker begins an act [Page 616]of language production by deciding what to say: message formulation. For example, a cat owner might want to tell you his or her pet happened to eat an unspecified arachnid. Then a speaker must decide which words to use to express that message, lexical selection, and then retrieve those words, lexical retrieval. The cat owner might choose the words my cat, ate, and a spider. A speaker must also assign these words to roles that convey who did what to whom, function assignment. The cat owner might assign “my cat” to the subject function and “a spider” to the object function. At least in spoken languages such as English, words can only be produced one after another, so a speaker must then use the principles of the grammar of the speaker's language to order the role-assigned words, constituent assembly. In English, the subject goes before the verb and the object after, so the cat owner's sentence will use the sequence “my cat ate a spider.” With words and their order (at least partly) determined, a speaker can send a plan off to phonological encoding so that the sound of an utterance can be formulated, followed by articulation so that a signal is actually generated for an audience. Additionally, most approaches to language production allow monitoring—that is, assessing formulated speech (before or after articulation) for adequacy and accuracy.

To be precise, though it is convenient to present these processing components as operating in a strictly sequenced or staged fashion, it is not clear that the language system works this way. Certainly, it is not the case that each component must finish its tasks completely for an entire sentence before the next stage can start. Rather, production is incremental in the sense that once some initial part of an utterance has been formulated at one level of processing, that part can be sent for processing at the next level of processing as upcoming parts of the utterance are formulated at the first level. For example, once a speaker has retrieved the words my and cat, they can be sent off for phonological encoding as the word ate is selected. Such incrementality allows speakers to start their utterances sooner and, more generally, permits some parallel processing so as to make production more efficient. A more controversial aspect of production planning, whether processing is strictly staged or is more free-flowing, is discussed below where lexical selection and retrieval are detailed further.

Each of these processing components is discussed in turn.