Date: Thu, 17 Mar 2005 09:07:52 -0800
From: Carolina Gonzalez <sauce@euskalnet.net>
Subject: The Phonological Spectrum: Volume II: Suprasegmental structure
EDITORS: van de Weijer, Jeroen; van Heuven, Vincent J.; van der Hulst,
Harry
TITLE: The Phonological Spectrum
SUBTITLE: Volume II: Suprasegmental structure
SERIES: Current Issues in Linguistic Theory 234
PUBLISHER: John Benjamins
YEAR: 2003
Carolina González, Department of Linguistics, University of California,
Los Angeles
BRIEF SUMMARY
This is a collection of ten papers on suprasegmental phonology. The volume
opens with a short, very general preface on the aims and contents of the
two volumes of The Phonological Spectrum. After this, a note by the
editors titled 'Syllables, feet and higher up', briefly discusses the
organization and contents of the second volume. The remainder of the book
is organized into three sections: (1) syllabic structure, (2) metrical
structure, and (3) prosodic structure.
COMMENTS ON THE PAPERS
The first section of the book comprises four papers on syllable structure.
In 'Phonetic Evidence for phonological structure in syllabification',
Heather Goad and Kathleen Branner consider the representation of word
final consonants in the earlier stages of language acquisition. The
authors propose that word-final consonants (henceforth C#) are first
syllabified as onsets regardless of whether they are onsets or codas in
the adult stage. Departing from standard Government Phonology, the authors
assume, following Itô (1986) and Piggott (1991, 1999), that C# can be
syllabified as codas or as onsets of a syllable with an empty nucleus.
This choice depends on various properties of the language, including
segmental profiling (if C# are restricted in respect to word internal
codas, they are onsets; otherwise, they are codas), rhyme shape (languages
with word final codas allow CVC structures only; languages where C# are
onsets allow CVXC structures), and release properties (C# with onset- like
release properties will be onsets; final codas will usually be
neutralized).
According to Goad and Branner, early acquisition of C# in English proceeds
in two stages. In the first stage the syllable structure is CV; epenthesis
after C# and word truncation are common. In the second stage C# are
pronounced, but they are onsets, since they emerge before word-internal
codas and show onset-like release properties (aspiration, nasal release,
and lengthening; the preceding vowel may also be lengthened or have a post-
vocalic pause).
Goad and Branner propose that in early stages of acquisition C# syllabify
as onsets regardless of the constraints of the target grammar. But this
conclusion is based only on English, where the syllabification of C# in
the adult stage is ambiguous; minimal words have a CVX shape, which
suggest that C# have weight and are codas. On the other hand, CVXC are
allowed at the right edge, which argues for a syllabification of C# as the
onset of an empty headed syllable. This is acknowledged by the authors
(footnote 6, p. 26) but the implications for their analysis are not
discussed. Additionally, Goad and Branner propose that C# forms a rhyme
unit with the following empty nucleus, this structure being most
consistent with the release and timing properties found in this position.
It is unclear that this structure is unmarked enough to be posited at this
stage. Finally, the reference title for Hildegard's data, Leopold (1939),
mentions that Hildegard is a bilingual child; this is not discussed in the
paper either.
'The phonology-phonetics interface and Syllabic Theory', by Shosuke
Haraguchi expands and revises Haraguchi 1998 Set Theory of the Syllable
(STS). This theory assumes that syllables consist underlyingly of
unordered sets of features, with ordering determined by principles of
linearization and phonetic implementation. Syllable structure is
parametrized, with the unmarked syllable composed of the 'core' (onset and
nucleus) and the 'coda', as in Japanese, and the marked case formed
of 'onset' and 'rhyme' (nucleus and coda),as in English. The unmarked
syllable structure explains the universal preference for CV syllables.
Various universal constraints determine the principles of linearization,
including that the onset precedes the nucleus, and the nucleus precedes
the coda.
Haraguchi considers a number of properties of linearization and the
phonology-phonetics interface in Tokyo Japanese and English. In Tokyo
Japanese the syllable template is (C)V (V)(X), where X is either a moraic
nasal or the first half of a geminate. Restrictions in what constitutes a
possible syllable in this language make underlying unordering possible.
Additionally, the morphological structure of words is responsible for
determining the relative ordering of verbal roots and suffixes, as
exemplified by the alternations found in inflectional verbal forms in
Japanese.
Haraguchi defends that STS captures various facts in English without the
need for additional machinery. For example, since /h/ and /ng/ are
restricted to onset and coda positions respectively, any word with these
segments will consist on a sequence of unordered segments with only one
possible interpretation. However, linearization has to apply at some point
in the phonology to allow for cases where segments need to be adjacent, as
in voicing assimilation in the plural or in the past tense. Haraguchi
concludes that STS accounts for a number of previously unnoticed or
unexplained facts. A number of issues remain to be investigated, as
pointed out in p. 55; among them, to make the system of linearization more
explicit and to examine languages with more complicated syllable
structures.
'Hungarian as a strict CV language' by Krisztina Polgárdi, argues for an
abstract representation of syllabic structure in Hungarian as CV,
following Lowenstamm (1996). This is based on two pieces of evidence.
First, there is an almost complete lack of phonotactic restrictions
between word-internal consonant clusters; clusters where the second member
is more sonorous than the first are very common. This is analyzed in
standard Government Theory as two single onsets separated by an empty
nucleus. Secondly, long vowels cannot precede consonant clusters. This is
explained if an empty nucleus inhibits the preceding vowel from being
long, since long vowels are represented as a vowel expanding to the next
nucleus position. However, long /a/ and /e/ are exceptional in that they
can precede clusters. Polgárdi shows that these vowels differ both
qualitatively and quantitatively from their short counterparts; they also
are unique in that they can undergo low vowel lengthening and alternate in
shortening stems. Polgárdi analyzes long /a/ and /e/ as sequences of short
vowels separated by an empty onset. This explains their occurrence before
clusters. Polgárdi concludes that more surface-based accounts cannot
explain important generalizations in Hungarian.
'Syllable structure at different levels in the speech production process:
Evidence from aphasia', by Dirk-Bart den Ouden and Roelien Bastiaanse,
reports an experimental study on syllable simplifications in aphasic
patients. This study attempts to establish the extent to which syllable
structure has a phonetic basis. Two groups of aphasics were tested: non-
fluent patients, which are commonly assumed to have a phonetic deficit,
and fluent aphasics, which are supposed to have a more abstract
phonological deficit. Their performance on a repetition task concerning
words with different onset/coda combinations was contrasted; the results
show that both groups of aphasics behave similarly, producing more errors
in codas than in onsets, and more in marginal syllabic elements. The
findings suggest that both groups of aphasics have similar syllable
representations, and that syllable structure plays a role at both the
(cognitive) phonetic level and the phonological level.
The section on metrical structure opens with 'Quantity- sensitivity of
syllabic trochees revisited: The case of dialectal gemination in Finnish',
by Heli Harrikari. In various eastern and northern Finnish dialects,
gemination targets the medial consonant of a (CV.CVV) sequence, giving
rise to the structure (CVC.CVV). Harrikari proposes that this is foot-
sensitive gemination that occurs to repair the worst type of trochee in
Finnish (CV.CVV is an uneven trochee with a heavy syllable in non-head
position) in order to create a more balanced trochee. Harrikari analyzes
this phenomenon in Optimality Theory, with the crucial ranking of IDENT-
Nuc-syllable stress (a positional faithfulness constraint which penalizes
changes in the moraic structure of the stressed vowel) over trochaic-
oriented constraints. This interaction explains other cases where
gemination is preferred over vowel lengthening, even if is more marked.
According to Harrikari, syllable weight is relevant in syllabic trochees
in Finnish, which calls for the reevaluation of the general notion that
syllabic trochees are quantity insensitive.
'Ternarity is Prosodic binarity' by Jay I. Rifkin, analyzes ternary
structures as prosodic words, building on Itô and Mester (1992) and Hewitt
(1992). Rifkin proposes that prosodic words are binary: they can be
composed of two feet or a foot and an unparsed syllable. This is encoded
as the constraint Bin (P).Rifkin exemplifies his proposal with the
analysis of ternary structures in Cayuvava and Pirahã. In Cayuvava, a
three-syllable iteration continues across the stress contour domain, and
various segmental processes that interact with the prosody evidence the
emergence of optimal prosodic words. Vowel deletion is blocked when it
would result in a less optimal prosodic word. Additionally, semivowels j,
w and vowels i, u are allophonic; j and w occur if there are already three
vowels in the word, and i, u occur if there are only two vowels. This
process reduces the number of unfooted syllables in the word. In Pirahã,
word boundaries are left aside to create a better prosodic word, which
provides evidence for the status of Bin (P). Rifkin suggest to extend this
model to binary and quaternary systems (as Palestinian Arabic), and
proposes that Bin (P) can substitute the Lapse constraint, and that feet
can only be maximally binary.
'The status of word stress in Indonesian', by Ellen van Zanten, Rob
Goedemans and Jos Pacilly, reports a number of production and perception
experiments designed to determine the location of word stress in
Indonesian. The production study was conducted on two Indonesian speakers,
one with a Javanese substratum, and the other with a Toba Batak
Substratum. These speakers produced four and five syllable words in
sentence final focus and non-final non-focus position. The perception
experiments also tested Indonesian speakers with different substrata (Toba
Batak, Javanese, or Jakartan Indonesian/Malay). The results show that
stress can be final or penultimate in Indonesian, suggesting that word
stress is free or that it does not apply. Additionally, it was found that
the substrate makes a difference in stress realization and perception
(speakers with a Toba Batak substratum have strictly penultimate stress).
The authors speculate that prominence in Javanese is phrase based, with
one phrasal accent located near the end of the phrase.
The third section consists on three experimental papers on different
aspects of the prosody of Dutch. 'Perceived prominence and the metrical-
prosodic structure', by Karijn Helsloot and Barbertje M. Streefker,
reports the results of a perceptual experiment on word prominence in Dutch
and proposes a model that incorporates these results to improve text to
speech systems. Ten subjects listened to a number of sentences in Dutch
and indicated which words were realized with emphasis. The results were
mapped with theories of prosodic phonology, specifically with the
relational-based metrical grid representation of Liberman and Prince
(1977). Four levels of prosodic structure were found: (i) syllable, (ii)
weakly stressed syllables(function words/non-head long context words, head
syllables of verbs), (iii) lexically stressed head syllables of nouns, and
(iv) head syllables of adjectives, adverbs, negative particles and
deictically used demonstrative pronouns. These four levels were captured
through a series of prosodic input constraints. The experiment also
provided evidence for an increase of prominence in the first prominent
word in the sentence, deletion of the last highest prominence in the
sentence, and avoidance of clashes and lapses. These facts were captured
through prosodic output constraints, which are higher ranked than prosodic
input constraints; among these, avoidance of lapses and clashes are ranked
highest.
'Phonetic variation or phonological difference? The case of the early
versus the late accent-lending fall in Dutch', by Johanneke Caspers,
reports a perceptual experiment that tested whether different melodic
differences in Dutch are relevant phonologically. The study involved
pairwise and rating tasks aiming to see if subjects distinguish different
meanings and attitudes in the use of two types of falling pitch accents:
early vs. late accent lending falls. The results support the hypothesis
that a difference in timing relates to a difference in meaning; the early
accent is more prone to project information , while late accent stresses
new information. Additionally, there is a difference in perceived
attitude, with the early accent-lending fall perceived as 'less
acceptable, more detached, more final and more irritated' than the late
accent-lending fall (p. 217). These results suggest that both pitch
accents are separate phonological categories.
'On the categorical nature of intonational contrasts- an experiment on
boundary tones in Dutch', by Bert Remijsen and Vincent J. van Heuven,
closes the volume. This paper reports an experiment that tested whether
the low and the high boundary tone, which in Dutch signal the end of a
statement and a question marker respectively, are distinguished
continuously or categorically. The results from an identification and
discrimination task show that there is a categorical distinction between
the two boundary tones. This provides evidence for the phonological status
of these tones and constitutes a clear example of categorical perception
of intonational contrasts.
GENERAL COMMENTS AND SUMMARY
The papers represent diverse theoretical and experimental approaches to
suprasegmental structure. The range of languages investigated is quite
varied, and includes English, Japanese, Hungarian, Finnish, Indonesian,
Cayuvava, Pirahã and Dutch. There is a good balance between experimental
and theoretical contributions. However, the choice of experimental papers
could have been more diverse; four out of the five contributions focus on
the Dutch language.
REFERENCES
Haraguchi, Shosuke. 1998. A Theory of the Syllable. Paper presented at
LP'98, Columbus, Ohio, September 15, 1998.
Hewitt, M. 1992. Vertical maximization and metrical theory. Ms., Brandeis
University.
Itô, Junko. 1986. Syllable theory in prosodic phonology. Ph.D.
Dissertation, University of Massachusetts, Amherst. Published 1988 by New
York: Garland.
Itô, Junko and Armin Mester. 1992. Weak layering and word binarity. Ms.,
UC Santa Cruz.
Leopold, W. F. 1939. Speech development of a bilingual child; a linguist's
record. New York: AMS Press.
Liberman, Mark and Alan Prince. 1977. On stress and linguistic rhythm.
Linguistic Inquiry 8. 249-336.
Lowenstamm, J. 1996. CV as the only syllable type. In J. Durand & B. Laks
(eds.), Current Trends in Phonology: Models and Methods . CNRS, ESRI,
Paris X. 419-441.
Piggott, G. L. 1991. Apocope and the licensing of empty-headed syllables.
The Linguistic Review 8, 287-318.
Piggott, G. L. 1999. At the right edge of words. The Linguistic Review 16,
143-185.
|
