LINGUIST List 10.1193

Thu Aug 12 1999

Review: Revithiadou: Headmost Accent Wins

Editor for this issue: Andrew Carnie <carnielinguistlist.org>

Directory

1. Paul Hopkins, Review of Revithiadou 1999: Headmost Accent Wins

Revithiadou, Anthi (1998) Headmost Accent Wins: Head Dominance and Ideal
Prosodic Form in Lexical Accent Systems. Holland Academic Graphics, The
Hague. 327 pages.

Reviewed by Paul Hopkins, University of Victoria

SUMMARY

This dissertation, the result of four years of research at Leiden
University, shows that the lexical marking of accent is used productively
by some languages to provide cues to morphological structure and is
furthermore constrained by prosodic structure so that what superficially
seems an impediment to language learning may actually facilitate it. The
dissertation includes in-depth analyses in the Optimality Theoretic
framework of the stress systems of Greek, Russian and the (Interior) Salish
languages and so should be of interest to those working on these languages
as well as linguists interested in Optimality Theory, metrical phonology,
and the prosody-morphology interface.

After a brief introduction in which R. sketches out her proposal and gives
an outline of her thesis, she presents in Chapter 1 a typology of stress
systems. A primary distinction is made between fixed-stress systems, where
stress is primarily determined on the basis of prosodic criteria, and
interface systems, where morphology plays a role in determining the
position of stress. The latter are divided into morphology-dependent
systems, head-dependent systems and head-stress systems. In the first of
these, stress is largely determined by structural constraints but some
grammatical markers are lexically prespecified to prevail. In the latter
two, the morphological/prosodic head of each word is given priority in
determining the position of primary stress: they differ in that whereas in
head-dependent systems the head determines stress position only if it is
lexically marked, in head-stress systems the head determines stress
position whether or not it is marked. Chapter 1 also includes an overview
of Optimality Theory and an appendix, in which R. shows that different
ranking of four archetypical constraints (Structural, Faith, HeadFaith and
HeadStress) can account for the attested stress patterns in fixed-stress
systems and all types of interface systems.

Chapter 2 presents R's theory of lexical accents and contrasts it with that
of other researchers. In this theory, a lexical accent is an autosegment
sponsored by a morpheme. Its appearance in surface forms is controlled by
the contraints used to ensure faithful parsing of other autosegments. Its
position, however, is determined by two constrasting constraints, *Flop and
*Domain. *Flop requires each lexical accent to manifest itself within the
morpheme it is lexically associated to. *Domain, however, requires each
lexical accent to "extend beyond the restricted domain of a morpheme and
become a property of the word" (p. 51) by appearing outside of its sponsor.
Interaction between the constraints shows that morphemes differ in their
relationship to lexically sponsored accents in that some are lexically
associated with the accent and some are not: in the former case, the accent
cannot appear outside the sponsoring morpheme without violating *Flop, in
the latter case, the accent can (and usually does). A further distinction
is made between strong and weak lexical accents, where strong accents
attract prominence and weak accents avoid prominence but retain duration.
After then showing that her conception of the lexical accent helps avoid
problems encountered by other researchers, R. concludes the chapter with an
appendix giving evidence for unaccentability in tone languages.

Chapter 3 is the first of two chapters in which R. develops and exemplifies
her theory through an in-depth analysis of two languages, Greek and
Russian. In this chapter the focus is on forms without derivational
morphemes. For each language, R. presents the accentual facts in great
detail, focussing primarily on nouns but also discussing verbs and
adjectives. She shows that in spite of certain differences relating
primarily to the default stress pattern, the two languages have many
similarities which R. characterizes as typical properties of head-dependent
systems. First, both languages restrict the prosodic forms available to
lexically marked (accented and unaccentable) words: although the position
of stress in marked words is not predictable, these words have fixed
paradigmatic stress and a predictable prosodic shape, advantages which
unmarked words lack. Second, both languages deal with conflict between an
accented root and an accented inflectional suffix by giving priority to the
root. Third, in both languages loan words are prespecified with diacritic
stress, but as they are assimilated they conform to the requirements
imposed on marked native words. The chapter concludes with an appendix
giving phonetic evidence supporting the claim that Russian words are
exhaustively parsed into trochaic feet and an analysis of these facts.

Chapter 4 examines the influence of derivational suffixes on stress in
Greek and Russian. In contrast to what is found in chapter 4, the accentual
properties of roots do not (necessarily) prevail here. Rather, it is the
accentual properties of the derivational suffix which determines the stress
pattern of the word. The explanation for this is found in the notion of the
morphological/prosodic head. In a derived word, it is generally the
derivational suffix which determines the categorial/class properties of the
word as the morphological head. Given an appropriate means of directly
mapping prosody onto morphology, which R. does with a principle of prosodic
compositionality borrowed from formal semantics, the morphological head
becomes a prosodic head, and R. argues that prosodic heads are accentually
prominent in head-dependent and head-stress systems. The former highly rank
the constraint HeadFaith, which requires lexical accents sponsored by a
morphological head to be given priority, and the latter highly rank the
constraint HeadStress, which requires morphological heads to be stressed.
High ranking of HeadFaith is sufficient to account for the Greek stress
pattern, since the accentual properties of the derivational suffix are
always respected in this language, but Russian throws a few additional
curves. First, Russian has some evaluative suffixes which, although
apparently marked, do not draw stress away from a marked root. Although
derivational suffixes, they do not determine the class properties of the
words they are in, and so are not morphological heads: this is reflected in
the prosody by their lack of accentual prominence. Second, in a certain
number of Russian words, stress is retracted from an inflectional suffix
onto a root in some forms. Whereas other researchers have had to describe
this by the application of a arbitrary rule, R. can appeal to the broader
notion of head attraction: apparently, the forms in question give top
ranking to HeadStress forcing stress to be associated with the
morphological head even at the cost of violating HeadFaith (stressing an
unaccentable root) and Faith (preferring an unmarked root over a marked
suffix).

Chapter 5 extends R's analysis beyond the fusional languages to
polysynthetic ones, where word structure is determined not only by
morphology but also by syntax. R. examines four Salishan languages: two
Northern Interior languages (Lilloet and Thompson) and two Southern
Interior languages (Moses-Columbian and Spokane). Building on seminal work
on Moses-Columbian by Czaykowska-Higgins (see CzH 1996, revised as CzH
1997), R. recognizes two morphological units in the Interior Salish word:
the morphological stem, containing the root, reduplicative affixes and
lexical suffixes, and the morphological word, which also includes
transitive and aspectual markers. Examining each of the languages in turn,
most extensively Thompson, R. shows that at the level of the morphological
word these Salish languages can be seen as exhibiting the characteristics
of head-dependent languages if the lexical suffixes are assumed to be
non-heads when they act as incorporated arguments and heads when they form
participate in compounds. When transitive, aspectual or modal suffixes are
included to form a morphological word, however, these Interior Salish
languages exhibit the characteristics of head-stress languages: the
morphemes representing the higher functional categories attract stress, and
when the morpheme representing the highest of these functional categories
lacks a vowel, stress is associated with the morpheme representing the next
highest category it is able to, be that another functional category, the
root or a lexical suffix (these languages are quite rich in vowelless
morphemes). This chapter shows that prosody is sensitive not only to
morphological but also to morphosyntactic structure. Giving prominence to
prosodic heads in a structure-sharing relationship with morphological heads
provides learners with cues to morphological information, and thus it is
not surprising that languages with complex morphology employ lexical
stress.

EVALUATION

This dissertation provides many insights into the prosody of languages with
lexical stress systems and certainly merits a wide distribution. Of
particular value are R's demonstration that prosodic structure can act as a
parsing cue for morphological structure and that in interface systems words
with unpredictable stress have predictable prosodic shape. The specifics of
her proposals will, of course, have to be tested to determine whether they
are truly applicable to all languages and what consequences they might have
for other parts of the grammar: the dissertation successfully shows, for
example, that head dominance voids the need, with respect to the
determination of stress, for both the derivational machinery of cyclicity
and the general metaconstraint RootFaith >> SuffixFaith (McCarthy & Prince
1995), but both have other applications which head domination may not
address. The notion of weak stress, used to explain the pre-accenting
properties of some Greek suffixes and the non-reduction of certain vowels
in Thompson, is another innovation which bears further examination.

In a dissertation of such wide scope, written in the author's second
language and under the pressure of needing to be published before a
doctorate could be awarded, it is perhaps inevitable that certain
inaccuracies have crept in. The ennumeration of typological errors,
mismatches between example numbers and their textual references and similar
peccadillos will be dispensed with here - those interested may request a
list of errata from the reviewer - however I would like to make some
general comments.

First, the description of data varies in reliability depending on the
language in question. R's description of Greek is highly nuanced and, as
far as I can determine, completely reliable. Her description of Russian is
very detailed, but it is clear that she has mostly relied on secondary
sources and there are minor inaccuracies. The following are some examples:
(1) it is not true that monosyllabic words always constitute a closed
syllable in Russian (p.122); certain roots traditionally claimed to contain
a yer vowel such as _t'ma_ 'darkness (Nom.Sg.)' and _l'va_ 'lion (Gen.Sg.)'
are examples of monosyllabic words consisting of an open syllable; (2) it
is not true that Russian assigns inflectional endings to all loan words
(p.139): loan words ending in a stressed vowel such as _pal'to_ 'overcoat'
and _kenguru_ 'kangaroo' are indeclinable (they are also missing from the
table of loan word accentual patterns, p.139); and (3) R's arguments for
exhausitive footing in Russian based on examples such as _s_advokatam_
(p.156, should be _s_advokatom_) are questionable because the first _a_ is
only partially reduced because it is word-initial, not because it appears
in the pre-pre-stressed syllable, and because the syllabification of this
word is sa.dvo.ka.tom, not sad.vo.ka.tom as R. claims. The interested
reader is referred to Zubritskaya (1995), which as the most comprehensive
Optimality Theoretic examination of Russian syllable structure to date
surely deserved a mention somewhere in R's 60 pages of discussing Russian
data. In presenting Salish data, R. generally stays close to her sources
although those tempted to quote are advised to consult the original
sources, as there are some discrepancies in the presentation of forms, the
references and, perhaps most importantly, the glosses.

In conclusion, "Headmost Accent Wins" presents a thought-provoking vision
of phonology-morphology interaction which goes a long way towards
explaining the nature of lexical accent systems. Given the scope of this
thesis, an index might have been useful but this lack is well compensated
for by excellent summaries which extract the essence of each chapter and
provide the reader with ample opportunity for review.

Bibliography:

Czaykowska-Higgins, E. (1996) "What's in a Word? Word Structure in Moses-
Columbia Salish (Nxa'amxcin)". Voices of Rupert's Land, University of
Manitoba, Winnipeg.

Czaykowska-Higgins, E. (1997) "The Morphological and Phonological
Constituent Structure of Words in Moses-Columbia Salish (Nxa'amxcin)". In
E. Czaykowska- Higgens & M. D. Kinkade, eds., Salish Languages and
Linguistics, Mouton de Gruyter, 153-195.

McCarthy, J. & A. Prince (1995) "Faithfulness and Reduplicative Identity".
In J. Beckman, L. Walsh Dickey & S. Urbanczyk, eds., University of
Massachusetts Occasional Papers in Linguistics 18: Papers on Optimality
Theory, 249-384.

Zubritskaya, K. (1995) "The Categorial and Variable Phonology of Russian".
Doctoral dissertation, University of Pennsylvania.

The Reviewer:

Paul Hopkins (e-mail address: phopkinsuvic.ca) is a doctoral candidate in
the Linguistics Program at the University of Victoria (BC, Canada). His
dissertation (in progress) is an Optimality Theoretic account of syllable
structure and stress system in the Slavic language Kashubian. His research
interests include the linguistics of the Germanic, Slavic and Salishan
languages and he has taught Esperanto, German, Polish and Russian as well
as introductory linguistics.