Rob Goedemans, (1998) Weightless Segments, Holland Academic Graphics,
The Hague. 262 pages, NLG 61.50
Reviewed by Andreas Mengel, IMS Stuttgart.
This book addresses the apparently universal phenomenon in stress rules
for the languages of the world that syllable onsets (consonants in front
of the first vowel of a syllable) are irrelevant for stress assignment
whereas the weight, i.e. segmental complexity, of the rhyme (vowels and
final consonants) of a syllable are good predictors for the location of
stress in words. In his book Goedemans looks for phonetic and
phonological explanations for this fact.
>From results of a series of speech production experiments Goedemans
concludes that the duration variation in onsets of syllables caused by
more complex consonant clusters is similar to duration changes of
equivalent rhyme structures. Thus, one possible explanation for the
weightlessness of onsets - phonological unexploitability of onsets due
to duration stability - must be dropped.
Since acoustical measurements of speech data need not necessarily
reflect their perception, psycho-acoustical experiments might shed
different light on the issue. The next experiments show that (Dutch)
listeners are less sensitive to duration variation in onsets than to
those in changes applied to the rest of the syllable. This finding can
by explained by a look at the sonority of the segments of these
structures. Sonority enhances duration perception sensitivity.
Another explanation of the relative weightlessness of onsets that is
tested is the existence of a trigger in the syllable that activates the
duration estimation mechanism, which - in case the trigger is located
after the onset - has the result that onsets before the trigger position
cannot be estimated properly. Trigger candidates are intensity maximum,
the P-centre (Pompino-Marschall 1989), and the CV-transition or PIVOT.
None of these candidates is proven to be a trigger, although for the
PIVOT this cannot be excluded completely.
In the last experiment it is tested whether the difference in duration
perception of onset, nucleus and coda is specific to speech by using
non-speech syllable signals whose spectral shape is syllable like. For
the onset, nucleus and coda of these stimuli, no difference in duration
perception sensitivity is found.
In the phonological part of the book, Goedemans evaluates languages
which are described in the literature as having onset sensitive stress
rules since the general claim that syllable onsets are weightless is
doubted by some authors, e.g. Davis (1985). Reviewing the data of these
languages, alternative descriptions to account for these exceptions are
provided. Goedemans concludes that it is doubtful whether these
languages can still correctly be labeled as onset sensitive.
In the conclusion Goedemans outlines an interpretation of his findings
and perspectives of further research. Under the assumption that the
experiments were conducted properly he supposes that physical duration
cannot be seen as the phonetic correlate of phonological weight as such
but only duration perception is correlated with phonological weight.
The fact that the perception of properties of non-speech syllable
stimuli is different from the perception of speech syllables might show
that onset weightlessness and the asymmetry of duration perception is an
abstract phenomenon which has its source in a linguistic centre in our
brain where the abstract syllabic model contains no weight unit for
onsets. For the case that part of his experiments have conceptual gaps
or are based upon inappropriate stimuli, Goedemans develops further
experiments to be run. Possible results and consequences for
phonological models are outlined.
Goedemans gives very concise descriptions of the rationale behind the
questions in focus and possible hypotheses. The phonetic experiments
are well documented and evaluated critically. This book is well
structured and the reader is provided with all necessary details to
follow the experiments and the theoretical considerations.
However, the theoretical assumptions and the starting point of some of
the experiments need further thought. First of all weight is assumed to
be an absolute property of syllables. This is the reason why it is
appropriate to investigate the structure of isolated syllables. Since
word stress is a phenomenon that involves at least two syllables, this
assumption must be questioned. Yet, it has been shown for many
languages that duration is the dominant signal property that correlates
with stress location (among others Jessen et al. 1995, Potisuk et al.
1996, Sluijter et al. 1997, Mengel 1997) so the assumption that relative
syllable weight, i.e. segmental complexity, duration and word stress
location correlate can be taken for granted. Considering words with
more than one syllable as a starting point, the question still remains
why mostly only rhymes are considered in quantitative stress rules. The
psycho-acoustic explanation for this fact is that the beginning of
acoustic events caused by speech utterances start at the location of the
P-centre, i.e. at a point in the signal where spectral change is most
prominent because of a sudden energy rise (Zwicker & Fastl 1990).
Phonologically speaking, within a syllable this is the transition
between the onset and the rhyme. Thus, the onset of a syllable belongs
to another - previous - psycho-acoustic event. As this is so, the
prominence relation of the two syllables in <houses> and <blouses> is
equivalent, the difference in the onsets of the first syllable does not
count because the relevant psycho-acoustic entities are <ous> and <es>
in these words and they are compared with respect to their duration.
The longer entity is perceived as bearing stress.
Goedemans is right in stating that onsets are weightless, yet one should
add, they are weightless in as much as they do not contribute to the
weight of the syllable that includes the following vowel, because
psycho-acoustic syllables are marked by the beginning of vowels for the
most part, and in this respect they obviously differ from articulatory
Davis, S. (1985): Topics in syllable geometry. PhD Dissertation,
Jessen, M.; Marasek, K.; Schneider, K.; Clah\223en, K. (1995): Acoustic
Correlates of Word Stress. International Congress of Phonetic
13, 4: 428-431.
Mengel, A. (1997): Domains and Properties of Lexical Stress in German.
Conference on the Word as a Phonetic Unit, Berlin.
Sluijter, A.M.C.; van Heuven, V.J.; Pacilly, J.J.A. (1997): Spectral
Balance as a Cue in the Perception of Linguistic Stress. Journal of
Acoustical Society of America 101, 1: 503-513.
Pompino-Marschall, B. (1989): On the Psychoacoustic Nature of the
Phenomenon. Journal of Phonetics 17: 175-192.
Potisuk, P.; Gandour, J; Harper, M.P. (1996): Acoustic Correlates of
in Thai. Phonetica 53: 200-220.
Zwicker, E.; Fastl, H. (1990): Psychoacoustics. Facts and Models.
Biography of reviewer
Andreas Mengel (http://www.ims.uni-stuttgart.de/~mengel) studied
linguistics and communications at Technical University Berlin. He
compiled the German section of a pronunciation dictionary of German
names in EU funded project ONOMASTICA. He wrote his doctoral thesis
about German word accent (Deutscher Wortakzent. Symbole und Signale)
including phonological, phonetical and morphological analysis. At the
moment he is developing standards for the uniform representation of
linguistic annotation data and a query machinery in EU project MATE
(http://mate.nis.sdu.dk) at IMS Stuttgart.