A Linear Grammar of Speech 107
in a tone unit. He states that latter two types, types 2 and 3, represent
instances of hesitation phenomena (ibid. 31–2). In the fi rst case the speaker
indicates a momentary processing diffi culty in locating the correct word
while the latter case is a holding operation which gains the speaker time to
plan the rest of the utterance. As both types may result in incomplete tone
units or tone units with level tone, it will not always be possible to state with
certainty whether the incomplete tone unit or the realization of level tone
is an example of Cruttenden’s type 2 or 3 pause. Some examples from
Brazil (1997: 147–8) may clarify.^24
(46) // p he GAMbled // and LOST... // p LOST a FORtune //
Type 2
N V & Ø V V d N
(47) // he WAITed... // p he THOUGHT he’d better WAIT //
Type 3
N V... N V Ø N V V'
(48) // and... // p he THOUGHT he’d better WAIT //
Type 3
& N V Ø N V V’Example (46) appears to be best interpreted as an instance of a type 2
pause with the speaker struggling to locate fortune, though Biber et al.
(1999: 1055) state that speakers repeat the same piece of speech in order to
gain planning time. Examples (47) and (48) appear to be instances of type
3 pauses. The speaker in (47) produces a false start, hesitates, gains plan-
ning time, and assembles his message. In (48) the speaker hesitates in order
to gain planning time to allow for the assembly of the remainder of the
utterance. It is possible, however, to construct examples such as (49) which
are not readily interpretable as either a type 2 or 3 pause.^25
(49) // – and the ANswer is // \ TWENty pints of BEER //
& d N V d° e N P NIt is not clear whether the speaker pauses at the tone unit boundary because
he/she is searching for the lexical item twenty or marshalling the remainder
of the utterance.
Brazil (1995: 211–13) discusses four possible types of ‘on-line amendments’
which he explains as speakers in mid-increment realizing that they are