BUFFER
Speech needs to be pre-assembled in chunks of several words before we produce it; only in this way could we impose intonation patterns upon an utterance. It is therefore assumed that the speaker has some kind of mental buffer in which to store a blueprint of each upcoming utterance prior to articulation. The term is borrowed from the buffer in which a PC stores information ahead of printing it. Without such a resource, we would be incapable of producing connected speech as fluently as we do.
What goes into the buffer at this final stage of planning must be extremely detailed– or there may be several buffers operating in parallel. Levelt (1989) describes a phonetic plan which assembles the entire set of articulatory gestures which are necessary in order to produce the target words. In addition, intonation, lexical stress, word order etc. must be fully specified prior to articulation.
A further function of a speech buffer is to enable the pre-assembled sequence to be reviewed at the last possible moment before it triggers articulation. Evidence from self-correction data indicates that speakers not only monitor their utterances at the time they are producing them, but also subject them to a final check immediately before speaking.
Writers appear to need a similar storage process to that of speakers. While Dickens was penning the line It was the best of times, he must have had the rhetorical structure of the whole sentence already marked out, with the reprieve it was the worst of times already available as ‘pre text’. There is uncertainty as to the form that written language takes when it is stored in this way. Both reading and writing appear to have a strong phonological component; it may well be that words in the writing buffer are stored in some kind of quasi-phonological form as ‘inner speech’. In this case, there may be a separate buffer dedicated to the low-level process of assembling graphemes.
See also: Articulation, Self-monitoring, Speech production, Writing
Further reading: Levelt (1989)
