||This article needs additional citations for verification. (February 2008)|
A speech disfluency, also spelled speech dysfluency, is any of various breaks, irregularities, or non-lexical vocables that occurs within the flow of otherwise fluent speech. These include false starts, i.e. words and sentences that are cut off mid-utterance, phrases that are restarted or repeated and repeated syllables, fillers i.e. grunts or non-lexical utterances such as "uh", "erm" and "well", and repaired utterances, i.e. instances of speakers correcting their own slips of the tongue or mispronunciations (before anyone else gets a chance to).
Fillers are parts of speech which are not generally recognized as purposeful or containing formal meaning, usually expressed as pauses such as uh, like and er, but also extending to repairs ("He was wearing a black—uh, I mean a blue, a blue shirt"), and articulation problems such as stuttering. Use is normally frowned upon in mass media such as news reports or films, but they occur regularly in everyday conversation, sometimes representing upwards of 20% of "words" in conversation. Fillers can also be used as a pause for thought ("I arrived at, um—3 o'clock").
Research in computational linguistics has revealed a correlation between native language and patterns of disfluencies in spontaneously uttered speech. In addition to this research, there are other subjective accounts reported by individuals. According to one commentator, Americans use pauses such as "um" or "uh," the British say "er" or "erm", the French use "euh", the Germans say "äh" (pronounced eh or er), Japanese use "ā", "anō", or "ēto", and Spanish speakers say "ehhh" (also used in Hebrew), "como" (normally meaning 'like'), and, in Latin America but not Spain, "este" (normally meaning 'this'). Besides "er" and "uh", the Portuguese use "hã or é". In Mandarin "nà gè" and "zhè ge" ('that' and 'this') are used. In Bosnian, Serbian and Croatian, speakers vocalize an "ovaj", "onaj" or "to jest" ('this', 'that' and 'that is'.) Arabic speakers say "يعني", the pronunciation of which is close to "yaa'ni", [jæʕni] or [jaʕni], (literally, "he means", there being no grammatically gender-neutral third person) and in Turkish, they say "şey" in addition to "yani" (without the [ʕ] found in Arabic) and "ııı". A more complete list can be found on the fillers page.
Recent linguistic research has suggested that non-pathological disfluencies may contain a variety of meaning; the frequency of "uh" and "um" in English is often reflective of a speaker's alertness or emotional state. Some have hypothesized that the time of an "uh" or "um" is used for the planning of future words; other researchers have suggested that they are actually to be understood as full-fledged function words rather than accidents, indicating a delay of variable time in which the speaker wishes to pause without voluntarily yielding control of the dialogue. There is some debate as to whether to consider them a form of white noise or as a meaning-filled part of language.
See also 
- Robert Eklund (PhD thesis, 2004): "Disfluency in Swedish human-human and human-machine travel booking dialogues". Chapter 2 is an account of disfluency research from a wide variety of perspectives. Download (4 MB) http://liu.diva-portal.org/smash/record.jsf?pid=diva2:20923 (Linköping University)
- Herb Clark & Jean E. Fox Tree