Planning in Language Production

Scope

Speakers appear capable of planning and storing in memory (i.e., buffering) a representation containing the sounds of an entire utterance prior to articulating it, but for multiword utterances they do not do so spontaneously. Just as pauses in playback and slow connection warnings indicate that downloading has not stayed ahead of playback, the prevalence of disfluencies (e.g., silent pauses and delay signals such as um and uh) suggests a limited scope of planning and buffering. Moreover, buffering material is resource demanding and feedback from listeners may often make buffered material irrelevant, similar to downloading a whole movie that a viewer cancels after a few minutes.

Message

The preverbal message for a descriptive utterance contains a topic (what an utterance is about) and information that the speaker wishes to express about the topic. As the topic is usually the first thing mentioned, its message representation is probably fully specified and the words and sounds used to express it retrieved before any part of the utterance is articulated. Suggesting that speakers outline a message prior to utterance onset, initial pitch varies with sentence length, while sentence length is largely a function of message content and complexity. Also, having outlined preverbal messages allows speakers to order simple content before complicated content in languages structured like English (e.g., “The philanthropist donated to a charity a wide array of products and services that would prove to be very useful”) and complex content before simple in languages such as Japanese. Patterns of spontaneous word anticipations, exchanges, and subject-verb agreement errors suggest that speakers rarely plan or simultaneously consider messages beyond the clause that they are currently articulating.

Speakers need to simultaneously represent and maintain message representations that correspond to entire noun phrases to fluently produce prenominal modifiers (e.g., the long, brown hair) and conjoined noun phrases (dog and cup). Speakers can modify the content of a noun phrase while articulating its initial parts, but this leads to disfluencies when there is insufficient time to encode the new content. For example, a speaker may fluently utter, “the alien with the small spots,” and only decide to mention the size of the spots around the time he or she starts articulating alien. However, if the speaker wants to say “the small alien” fluently, he or she needs to consider size at least a half second before the onset of the.

...