How speech works - questions and preliminary answers.
Björn E. F. Lindblom · 1999
ABSTRACT In this presentation I attempt to discuss researchfrom a wide spectrum of topics all relevant to the studyof speech communication. It is my hope that placingmore narrowly focused work in a common, broadercontext will produce a kind of symbiosis effect that willhelp us see the individual subtopics in perhaps new andinsightful ways. Admittedly, such an undertaking mightseem ambitious and even ill-advised. However, as ourfield is put under increasing pressures to deliver clinicaland technological applications of greater real-liferelevance, it does appear useful to try to assess what weknow and do not know about speech processes. Also, asis sometimes the case, the whole tends to be more thanthe sum of its parts. FROM LAB SPEECH TO UNSCRIPTEDCONVERSATION For many years phoneticians have tried to understandspeech by investigating carefully designed speechsamples, so-called ’laboratory speech’. Recently, severalstudies of spontaneous, unscripted speech have beenreported (SPOSS 98). It is seems fair to say that the newfindings have clearly indicated that laboratory speechtends to drastically undersample the signal variabilityexhibited by natural speech patterns. Patterns ofextensive reduction and elaboration are typicallyobserved and are likely to challenge many of our mostcherished theoretical ideas about how the production andperception of speech work.