Chapter 7
Prosody in language documentation
Nikolaus P. Himmelmann
Introduction
Prosodic aspects of a linguistic message such as intonation and lexical ac-
cent are essential elements of its formal make-up. To date, the basics of
analyzing prosodic features have not yet become an integral part of linguis-
tic fieldwork training, and, accordingly, a reasonably detailed and compre-
hensive documentation and description of prosodic features is not yet part
of standard linguistic fieldwork practices. This chapter is specifically con-
cerned with the documentation of prosodic features, i.e. with the question
of what kind of data a language documentation has to contain so that a
thorough analysis of prosodic features is possible. In order to be able to
productively apply the suggestions discussed in this chapter, a basic under-
standing of the core units and procedures of prosodic analysis is necessary.
For a more comprehensive introduction to basic prosodic fieldwork focus-
ing on issues of analysis and description, see Himmelmann and Ladd
(forthcoming).
Given that a language documentation includes a large corpus of record-
ings of communicative events of different types, it may well be questioned
whether there is any need to pay special attention to prosody when compil-
ing it. Provided that the recordings are of a reasonable quality,
1
there can be
no doubt that such a corpus can be used for prosodic analyses even when no
particular attention was paid to prosodic features at the time of compiling
the corpus.
2
However, there are essentially three reasons why some special
attention for prosodic features is necessary when compiling a corpus of
primary data so that it becomes really useful for prosodic purposes:
1.
Prosodic phenomena are highly variable and susceptible to contextual
influences. This makes it difficult to recognize basic distinctive patterns.
Prosodic pattern recognition is much facilitated by having the same ut-
terance produced by a number of different speakers (or at least to have
multiple versions of the same utterance). See further Section 2.
164
Nikolaus P. Himmelmann
2.
Words produced in isolation are minimal utterances showing both lexi-
cal and utterance-level (post-lexical) features. Hence, the widespread
practice of recording words in isolation when recording a wordlist is of
limited use for prosodic purposes. See further Section 3.
3.
Acoustic and auditory data (i.e. recordings of spontaneous and elicited
utterances) do not provide direct evidence with regard to the perception
of native speakers, i.e. what native speakers actually perceive as rele-
vant prosodic contrasts (conversational material may provide indirect
evidence, though; see further below). The most straightforward way to
obtain perception data is to run perception experiments, as further dis-
cussed in Section 5.
Before these points are further elaborated, Section 1 provides a bit more
detail on what exactly the term prosody is intended to refer to here. Further-
more, when discussing points (1) and (2), it will be repeatedly suggested
that elicitation may provide useful materials to complement the data found
in recordings of spontaneous speech. However, eliciting prosodic data is
not an easy task, as discussed in Section 4.
1. Prosodic phenomena
Table 1 lists the major prosodic phenomena according to the different do-
mains in which they are manifest, i.e. the recordable sound wave (acoustic),
the perceptual impression (auditory), and as a component of the language
system (phonological category). The rightmost column lists the most
widely attested functions which may be conveyed by prosodic features (but
of course can also be conveyed by non-prosodic means).
Chapter 7 – Prosody in language documentation
165
Table 1. Prosodic phenomena according to domain
Acoustic Auditory
Phonological
category
Function/meaning
–
fundamental
frequency
–
duration
–
intensity
–
spectral
characteristics
3
–
pauses/silence
–
pitch
–
length
–
loudness
–
stress/
prominence
–
rhythm/tempo
–
grouping
–
voice quality
(creaky, etc.)
–
tone
–
quantity
–
(lexical) accent
–
intonation
–
levels in pro-
sodic hierarchy
(syllable, foot,
etc.)
–
delimiting units
–
distinguishing
lexical units
–
grammatical
categories
–
speaker attitude
–
sentence modality
–
information
structure
–
interactional tasks
In discussing prosody, it is important to keep the different domains distinct
and to be aware of the fact that there is no unambiguous mapping relation
between features in different domains. To take just pitch as an example,
regular correspondences exist between changes of fundamental frequency
(F0) observed in the acoustic signal, changes in pitch perceived by the hu-
man ear, and tonal or intonational distinctions. But these correspondences
do not consist of simple and direct mapping relations between the domains.
Thus, there are changes in fundamental frequency which are generally not
perceived as such by the human ear. These are known as “microprosodic
perturbations” and include phenomena such as the lowering of F0 regularly
induced by voiced consonants.
4
Furthermore, while it is true that tonal and
intonational categories are primarily marked by changes in pitch, other
auditory parameters such as length, loudness, and voice quality often also
play a role in the marking of these categories.
In the present chapter, the above distinctions and the corresponding ter-
minology will be observed rather strictly. Many of the terms are widely used
in the literature in the sense they are used here, but it may be worth pointing
out that the strict distinction also applies to the terms (lexical) accent and
stress, which
are
used
in
many
different
and
often
somewhat
confusing
ways
in the literature. Both terms refer to the phenomenon that a given syllable is
in some sense more prominent than neighboring ones, but lexical accent
here designates this property with reference to the phonological structure of
lexical items (i.e. as a phonological category), while stress refers to an
Dostları ilə paylaş: |