232 Notes
(^17) Chun (2002: 61) interprets Sag and Liberman’s notation in a different way. She
argues that it is key and not tone which distinguishes a literal question from a
suggestion: high key realizes a literal question while low key realizes a suggestion.
Nonetheless, the point that this is not a generally applicable rule remains.
Examples (13) and (14), regardless of key, are unlikely to be interpreted as
genuine enquires!
(^18) Their fi ndings have not, as of yet, been replicated for other dialects of English.
(^19) See Ladd (1996: 82) for a list of correspondences between Pierrehumbert’s
notation and that of the ‘British style nuclear tones’.
(^20) As discussed previously, mutual assumptions do not appear to be psychologically
feasible or necessary for the description of how speakers estimate the extent of
shared speaker/hearer convergence.
(^21) Another scholar who concurs that the fall-rise labels information as part of the
background while the fall labels information as updating the background
is Steedman (1991, 2000: 656) who argues that the ‘theme’ of an utterance
(information already shared by the speaker and the hearer) is either de-accented
or receives a fall-rise contour while the ‘rheme’ (information not previously
shared by the speaker and the hearer) receives a falling contour.
(^22) In his most recent work, Gussenhoven labels the communicative value realized
by rising tone as testing and claims that ‘testing leaves it up to the listener to
decide whether the message is to be understood as belonging to the background’
(2004: 299).
(^23) Brazil proposes an identical relationship between the fall (p) tone and rise-fall
(p+) tone. Both tones proclaim but only p+ realizes the extra communicative
value of dominance. Gussenhoven (1983) does not include rise-fall tone among
his three primary tones. However, he speaks of nine secondary tones which
are produced by the application of a number of phonetically specifi able modifi -
cations which are also assigned morphemic status (ibid. 193). One of the four
phonetic modifi cations is timing realized as delay (ibid. 216). Gussenhoven states
that a delayed fall is a rise-fall (1983: 217 his example 28a). The modifi cation
delay adds the extra communicative value that the manipulation (in this case
V-addition) of the variable is very signifi cant or non-routine (see also O’Connor
and Arnold 1973: 78–82 and Cruttenden 2001: 269). As rise-falls are rare in
discourse, I will discuss dominance only in respect to rising tones.
(^24) A discussion of planning diffi culties in assembling speech on the fl y is deferred
until Chapter 4 pp. 106–110.
(^25) An alternate explanation for the preponderance of level tones found in public
scripted prayer may be that it is diffi cult to speak in unison if speakers employ a
tone other than level tone (Martin Hewings, personal communication). How-
ever, such an explanation does not explain Crystal’s fi nding that the level tone is
also the most frequent tone found in individual liturgical prayer.
(^26) Neither Ladd nor Gussenhoven recognize the existence of an independent level
tone. They argue that what is realized phonetically as level tone is phonologically
either a stylized rise or a stylized fall. Gunter (1982) criticizes the labelling of a
surface level tone as a realization of an underlying tone as an artefact arising out
of Ladd’s theory.