Prosodic Domains Across Time and

Document Sample
Prosodic Domains Across Time and Powered By Docstoc
					                      Prosodic Domains Across Time and Space

                                    Juliette Blevins
                  Max Planck Institute for Evolutionary Anthropology

Prosodic domains are central to synchronic descriptions of sound patterns. Whether one
is determining the inventory of contrastive features, identifying phonotactics, or
analysing alternations, reference must be made to a specific prosodic domain, most
commonly, the phonological word. In contrast, historical descriptions of sound patterns
make little reference to prosodic domains. Though words are central units of comparison,
and regular sound change is assumed to apply within such units, there has been relatively
little study of the role of prosodic domains in determining the nature and direction of
sound change. This talk presents an overview of two ways prosodic domains can
influence sound change, working primarily with historical data from Austronesian and
neighboring languages. In the first case, an inherited prosodic domain is self-reinforcing
over time. In the second case, a prosodic domain is superimposed on a language via
contact, leading to a range of sound changes that are unattested in non-contact situations,
but which serve to satisfy the imposed prosodic domain. A range of hypotheses are
suggested by this study, from restrictive contact origins of sesquisyllabic domains, to
accounts of prosodically determined allomorphy.
                             The Optimal Word in Hindi
                                                                             Gyanam Mahajan

This paper analyzes the phonological word in some Indo Aryan languages. While it is
true that it is difficult to find evidence for a “phonological word” in Indo Aryan
languages or South Asian languages in general, it is striking that there is an “optimal
word” in these languages. The basic claim is that the “optimal” Hindi word is a
predictable, preferred form of bi-syllabic, with each syllable being heavy. While this is
not an expected unit since it is not a definable foot, this paper will show several word
formation strategies that are prosodically motivated to yield the bi-syllabic, each syllable
bi-moraic form. First, the processes that are considered are related to “naming” in Indo-
Aryan languages. Specifically, this paper looks at the following naming processes: (i)
assigned nicknames, kinship terms, other common names for objects (which are all
overwhelmingly bisyllabic, bimoraic) (ii) hypocoristics or nickname formation (e.g. long
names are reduced to two syllables, each syllable heavy, while one syllable names are
added onto to yield bisyllabic, bimoraic syllables). Additonal data is provided from
assimilation of borrowed words into Indo Aryan languages (lengthening in borrowed
Arabic words or gemination in borrowed English words) . Secondly, the paper analyzes
verb forms and provides an analysis with schwa epenthesis, which is again
overwhelmingly constrained by a preference for bisyllabic, each syllable heavy shape of
a word (e.g. verb stems are overwhelmingly monosyllabic and affixes attach to yield the
“optimal” word shape). Additional data is discussed from heritage language learners.

While it may be difficult to characterize this preference in terms of a phonological unit,
binarity is of itself an expected occurrence in language. Kayne (1984), provides several
arguments for binary branching, including the constrained structures that it yields and its
simplicity for acquisition. Although binary branching structures are assumed as a given in
Syntax, in Phonology, Foot Binarity is the main accepted constraint within Optimality
Theory. This paper argues for binarity not for the foot but the prosodic word in Hindi,
which skips the foot level to require binary syllables where each syllable is further binary
to require bi-moraic syllable. The foot is shown to be an irrelevant level in Indo Aryan
languages. At the same time, the prosodic word level itself has to be argued for. The
prosodic word cannot be defined in terms of any identifiable form of foot in Hindi. And
yet, it is not that the phonological shape of the prosodic word cannot be defined at all – in
fact, it seems overwhelmingly to prefer a very specific form. That is, the prosodic word
cannot be defined in terms of its next layer, the foot, but is distinct in terms of its syllable
structure and each syllable also shows a distinct preference with respect to the moraic
structure within it. The paper proposes a high ranking BIN constraint both at the word
level and the syllable level that interacts with other faithfulness constraints to yield the
“optimal” word.

While most of the data is drawn from Hindi, the paper considers other closely related
languages to look at the similarities and differences with Hindi. We observe similar data
for Gujarati and Punjabi. We also pursue the idea of India as a Linguistic Area to
consider data from Tamil, which is a Dravidian language.

                          Paroma Sanyal



       Central Institute of English and Foreign Languages,
                  THE CASE OF BANGLA, TAMIL AND AO

        In this paper we look at three languages from different language families of South
Asia to argue in favour of positing the prosodic category “phonological word”, between the
foot/prosodic word and the phonological phrase. We show that certain phonological
generalisations are best stated at this level of prosodic organization.
        In Bangla, which belongs to the Indo-Aryan family, iterative feet are built from the
left edge of the stem in the domain of the phonological word. Importantly, excluding the
prefix, derivational and inflectional suffixes including the genitive marker (clitic?) is parsed
and re-syllabified along with the stem to form a phonological word.
    1. (a). [(akor)( n)] pr.wd + er =>         [(akor)(ner)]ph wd
              attraction          GEN
       (b). [(pori)(brton)]pr.wd + er =>      [(pori)(brto)(ner)]ph wd
              change              GEN
In the above examples, the suffix/clitic gets incorporated into the prosodic domain of the
stem. This is reiterated by the licensing of the vowel // and vowel length observations in the
language. However, this domain is smaller than a phonological phrase which can include
more than one phonological word.

         In the Dravidian language Tamil, the vowels /u/ and // (unrounded, high, back vowel)
are in complementary distribution. While the former occurs in morpheme-initial position, the
latter is rounded elsewhere in the environment of labial consonants. The domain for this rule
of labial harmony is the phonological word (stem + suffix + clitic) as the data in 2 shows.
     2. (a). perkk           ‘sweep’
       (b). perkku-m         ‘sweep. F.neut’

        Chungli, one of the major dialects of Ao, a Tibeto-Burman tone language with three
contrastive tones, attested on every syllable also provides evidence for the phonological word
between the foot and the phrase. The tonal sequence M-L does not occur within words but is
attested across word boundaries.
    3. [[(an) (tukln)-i]ph.wd [asm-a]ph.wd ]PP ‘run to the east’
          sun rise -LOC           run-imp
In the above, the M in /an/ changes to H when followed by a L tone. However, the M in /-i /
when followed by L of the second phonological word, undergoes no change.

       We attempt to look at evidence like foot structure, licensing of vowels, vowel
harmony, resyllabification, tonal spreading and constraints on tonal sequences in the above
languages to strengthen the argument that the phonological word is a necessary intermediate
prosodic domain between the foot/prosodic word and the phrase.
                  Word prosodies in Tamang and some related languages

                                      Martine Mazaudon
                                      Lacito-CNRS, Paris

    The languages of the Tamang branch of Bodish, or TGTM languages, are at a stage in
their evolution where tone is still in the process of creation by transphonologisation from
segmental material, while tonal oppositions are now threatened by reduction or loss, at least in
the bilingual speakers of some of these languages, under the influence of increased contact
with non tonal languages.
     In these languages, the domain of the lexical assignment of tone is the morpheme,
whether monosyllabic or polysyllabic, and the domain of its phonetic implementation is the
word, comprising a lexical root and a (possible) string of affixes. Word-tone is a characteristic
of this group of languages and of (?most) languages of its sister branch, the Tibetan branch.
But some Qiangic languages are also reported with word-tone and it may be more an areal
character than a character of a branch of Tibeto-Burman. We will look at the different
structures in Risiangku Tamang, Standard Tibetan and Dzongkha.
    Secondly, tones in Tamang are defined by a bundle of features or cues, not all based on
F0, like voicing, open quotient -- breathiness -- which we have studied instrumentally, and
others which we have not.
    Finally we will look at other ‘prosodies’ (in the Firthian sense), like restrictions on vowel
length, and intervocalic voicing which help define the word-domain or smaller domains in
                                      The Prosodic Structure of Prinmi

Prinmi is a Tibeto-Burman language unique to southwestern China. It falls within the linguistic
area of South-East Asia (Matisoff 2001). Like many languages of this area, Prinmi morphemes
are predominantly monosyllabic and the language has developed a tonal system, where all
lexemes are assigned with a particular tonal category. Unlike the typical tonal language of
South-East Asia, the domain for lexical tones in Prinmi is not based on the syllable; it also
differs from word-tone languages such as Tibetan.
Tonal patterns in Prinmi are generated with two parameters: specification of the locus of H (the
high tone) in an underlying domain independent of the grammatical word, and the potential
spreading of H to the next syllable. Such characteristics of the tonal system of Prinmi are akin
to those found in the pitch-accent system of Japanese. Consequently, Prinmi has been analyzed
as a pitch-accent language in Ding (2001; 2006). To avoid unnecessary confusion from the
rubric ‘pitch-accent’ (cf. Hyman 2006), the kind of tonal system found in Prinmi and Japanese
will be referred to as ‘melody-tone’ in this paper. Accordingly, the underlying domain for
lexical tones will be labeled as the ‘melody’.
The prosodic structure of Prinmi involves three types of domain at various level, viz. the
melody, the phonological word, and the grammatical word. Minimal triplets that contrast solely
in terms of tonal categories occur in many monosyllabic words, e.g. (1). For the sake of
convenience, all syllables in Prinmi data are marked for tone (with H for ‘high’, F for ‘falling’
R for ‘rising’, and L for ‘low’.
    (1)       Tone A [H.L.L]melody              Tone B [H.H.L]melody                Tone C [L.H.L]melody
       a       iF louse                          iH hundred                          iR new
       b      bj F urine                        bj H busy                           bj R to fly
The grammatical words in (1), being monosyllabic, has a great impact on the realization of
these tonal categories: it neutralizes the contrast between Tone A and B. However, when the
words occur in a larger domain of phonological word after an enclitic is attached, the different
patterns become transparent:
       (2)           Tone A [H.L.L]melody                             Tone B [H.H.L]melody
       a              iH gjeL louse                                    iH gjeH hundred
                       ~                                                ~
       b             m H gjeL hair                                    m H gjeH name
       c             bi H gjeL honey
                      -                                               bi H gjeH sun
In compounding a complex grammatical word often extends into a larger phonological word,
where the second formative loses its original tone, e.g.
                       ~                               ~
      (3)a             oH sheep +       F
                                         meat      → oH         H
         b           tsh-F goat +
                        i              F
                                         meat      → tsh-H
                                                        i        L
                                                                    goat meat
However, it is possible for a compound word longer than two syllables to have more than one
melody. Similarly, the phonological word may also contain more than one melody.
Ding, P. S. 2001. The pitch-accent system of Niuwozi Prinmi. Linguistics of the Tibeto-Burman Area 24: 57-83.
     . 2006. A typological study of tonal systems of Japanese and Prinmi: Towards a definition of pitch-accent
     languages. Journal of Universal Language 7: 1–35.
Hyman, L. M. 2006. Word-prosodic typology. Phonology 23: 225–257.
Matisoff , J. A. 2001. Genetic versus contact relationship: prosodic diffusibility in South-East Asian languages,
     291–327 of Dixon & Aikhenvald (eds), Areal Diffusion and Genetic Inheritance. Oxford: OUP.
  Observations on the Production and Perception of Tone in Manang-
 Kristine Hildebrandt (University of Manchester) and David Hünlich
                       (University of Leipzig)

Like most Tibeto-Burman languages, Gurung is described as possessing
a contrastive word-tone system (specifically, four tones aligning
with vowel F0 and segment phonation properties). To date, the best
existing description of Gurung tone properties may be found in an
account of the Kaski variety of Gurung, spoken near Pokhara, Nepal
(Glover 1974), however, Gurung tone has been included in other
synchronic and diachronic accounts of tone in Bodish and Tamangic
languages (cf. Burton-Page 1955; Sprigg 1997; Mazaudon 1977, 1978,
1988). Despite this literature, there are still a number of open
questions about Gurung tone, including (but not limited to):

     What are the (acoustic) phonetic details behind tone in this
     language, and what is the functional load of tone in lexical
     There are many Gurung dialects spread out across Nepal, and
     different evidence suggests that tone variation is one
     dialectal diagnostic (cf. Glover and Landon 1980). Given this
     observation, how do tone properties in the dialect of Gurung
     spoken in Manang compare with properties of Kaski Gurung tone?
     Is the Tone Bearing Unit in Gurung the phonological word, as
     it has been demonstrated for other closely related languages
     like Tamang and Manange?

A preliminary study of the acoustic properties and perception of
tone in the Manang dialect of Gurung conducted by Hünlich (2006) was
inconclusive, due to unforeseen methodological issues. However,
despite a number of unanswered questions, there were tantalizing
indications that:

     Manang-Gurung “tone” (as it exists in the phonology of this
     dialect) is acoustically both unlike Kaski Gurung and also
     unlike Manange (a language with which Manang-Gurung speakers
     have had regular contact over an extended period of time)
     Despite its acoustic uniqueness, there is strong evidence that
     Manang-Gurung speakers perceive of and use distinctive pitch
     (among other cues) to disambiguate (segmental) homophones

A more recent and methodologically revised investigation of tone
production and perception among Manang-Gurung speakers confirms
these previous suspicions, and also reveals new patterns about the
phonetic properties and domain of tonal contrasts in Manang-Gurung.
As such, the goals of this presentation are to detail the
methodological challenges confronted in a detailed documentation of
the tone system of Manang-Gurung, along with the findings on the
production, perception and domain of tonal contrasts in a dialectal
and cross-linguistic perspective.
Workshop on Phonological Word, University of Leipzig, Sept. 19-20 2007, Germany

                                   Minimal Word in Mandarin Chinese

                                                   Shengli Feng
                                                 Harvard University

        Minimal Word (MinWd) is of unique significance in characterizing certain Prosodic-
Morphological phenomena. This paper explores the minimal word effect in Mandarin
Chinese. It is shown that the minimal word phenomena in Chinese provide strong evidence
supporting the minimality theorem (McCarthy & Prince 1990) and most importantly, as
argued in this paper, the minimal word constraint can also be extended to much broader
applications such as compound formation and its interaction with syntactic constructions in
languages like Chinese.

        The notion of MinWd is derived from the interactions of Prosodic Hierarchy and Foot
Binarity (M & P 1990, 1993, 1998). The Prosodic Hierarchy impinges on every prosodic
word to contain at least one foot, while the Foot Binarity demands that every foot be
bimoraic or disyllabic. As a result, a prosodic word must contain at least two moras or
syllables according to the transitivity of the Prosodic Hierarchy. The Minimal Word is
therefore a single PrWd in the system. As we will see in this paper, the Minimal Word is of
singular importance in characterizing a wide range of prosodic-morphological phenomena
not only in languages known in previous studies, but also in Mandarin Chinese. It is
demonstrated, first, that among all VO forms in the language, only the ones that meet the
minimal word requirement exhibit word properties while longer forms are all on a par with
phrases. Secondly, it is shown that a process of category changing from a [Aux V] verbal
expression to a [Aux-V]adjective compound is conditioned strictly on whether or not the [Aux
V] is a minimal word. Finally, there is a clear distinction, in Mandarin Chinese, between
MinWd and non-MinWd [A+N] forms both morphologically and syntactically.

         It is well known that in Prosodic Morphology, the core area of previous
investigations has focused mainly on reduplication and infixation. The present study,
however, extends the notion of PrWd into the area of compounding and its interaction with
syntax. For all three types of forms discussed in this paper, we will see that the minimality
constraint indeed controls the word formation of compounding in Chinese. That is, a
compound (or more specifically a lexical compound) must first be a PrWd, even if a PrWd is
not, by necessity, a compound. This conclusion has several implications in Prosodic
Morphology. First, the Templatic Constraint not only determines morphological operations
like reduplication and infixation, but also controls the word formation of compounding in
languages like Chinese. Furthermore, it is indubitable that the minimal word requirement
functions in Chinese. Yet, when it does, it not only affects word formation, but also syntactic
structures in a way that the MinWd may also circumscribe certain syntactic phrases; as a
result, the native speaker’s intuition about what is considered a “word”, as shown in this
paper, is unquestionably affected by the prosodic notion of MinWd.

        This paper raises interesting questions as to why and how the Templatic Constraint
could control not only morphology but also syntax, questions that are extremely important for
theoretical as well as empirical inquiries in future research.

Shengli Feng

Professor of The Practice of Chinese Language
Director of Chinese Language Program and,
Harvard-Beijing Academy
Department of East Asian Languages and Civilizations
Harvard University
Cambridge, MA 02138

 Gta bo, Khmu tmlùuy, Khasi kdor, Ksingmul pta:, Santali dalet’metahkanae:
             an overview of Munda and Austroasiatic word structure
                               Gregory D. S. Anderson
                         Living Tongues Institute for Endangered Languages

In this talk I offer an overview of some of the issues in Munda and Austroasiatic [AA] word
structure. The internal classification of the AA language phylum remains an open question.
It is clear that Munda is divergent from other AA languages. However, it has not yet been
demonstrated how exactly Proto-Munda differs from other putative proto-languages within
AA nor what exactly the relationship[s] is[/are] between the different identifiable subgroups
of AA (with notable exceptions, e.g. Katuic and Bahnaric, which seem to form some higher-
level group together). The term Mon-Khmer [MK] for example has been used for numerous
and as yet inadequately justified subgroupings such as ‘core MK’: Monic, Khmeric,
Bahnaric, Katuic, Pearic, Mangic, Palaung-Waic, Khmuic, Khasic, and Viet-Muong; a
“mainland” MK that includes all of the above and Aslian (and the as yet unclassified
Chinese AA languages as well as Palyu languages) and a macro-MK which further includes
         Austroasiatic is an old phylum and the subgroups are not closely connected; most
likely, there were additional branches that have died out unrecorded. In addition, AA
languages fall into several different areal linguistic spheres (Sino-, Indo-, and Malayo-
sphere) which have altered the nature of the languages from their original structure. But a
detailed comparative analysis can begin to reveal what the original kernel of “AA-ness”
was; despite their obvious differences, all languages have a basic AA component.
         In this presentation I offer an overview of the nature of AA languages from the
perspective of word structure. This addresses in brief both the nature of the phonemic
systems found in these languages and the broader word-based phonotactics that, for MK
at least (but arguably elsewhere in AA as well), are highly characteristic and indicative of
the family as a whole. Said in other terms, there is a core AA word structure that is found
more or less across the languages of the family. Once the basics of the phonemic systems,
and their combinatorial possibilities are introduced, I give a short overview of
suprasegmentals as well as basic word structure. In all discussions, I move from Munda
to other AA languages in a roughly West to East trajectory.
         After an overview of the synchronic nature of AA languages from the perspective
of the phonological structure of the word is offered, I then mention some of the major
overarching issues in AA diachronic linguistics. A study of AA word structure is of
course in part morphological and due to the peculiar history of the family and probable
original proto-AA structure, comparative AA phonology and morphology are
inextricable. Preliminary reconstructions of word forms in AA subgroups are offered and
the issues they elucidate discussed in brief. Whether or how categorization into parts of
speech is relevant to the phonology (or morphophonology) of the word in Munda and
other AA languages is also examined in brief. In addition to the possible differences (or
lack thereof) between nominals and verbals, the nature of ‘expressives’ and the role of
reduplication in AA word structure are briefly discussed and exemplified. The talk is
summed up by offering thoughts on fruitful areas for future research in comparative
Munda and AA word structure.
                                                         John Peterson,

            "Words" in Kharia: Phonological, syntactic and graphematic aspects
In the South Munda language Kharia, spoken in eastern India, there is no native term
corresponding to the English term word and in fact, this turns out to be a highly complex issue
in this language which must be approached from at least three different perspectives:
phonological, syntactic or "grammatical", and graphematic.
     Kharia has no distinctive dynamic accent. Phonologically, a "word" in Kharia begins with
a low tone which then rises throughout the remainder of the "word". With bisyllabic lexical
morphemes, the first syllable has a low tone while the second syllable has a high tone, with
monosyllabic morphemes, a rising contour is found, ròchó b 'side', la 'tongue'. This fits in
well with the fact that almost all lexical morphemes in Kharia are bisyllabic, with an L-H
pattern. When this is not the case, this morpheme is obligatorily reduplicated in certain
syntactic environments to create a bisyllabic root, with very few (highly common) exceptions.
Recent research suggests that this bisyllabic constraint has resulted from the "re-analysis" of
an earlier, proto-Munda bimoraic constraint as a bisyllabic constraint.
     The issue, however, immediately becomes more complex once grammatical markers such
as TAM, case, person marking, etc., come into play. These markers are all enclitic in Kharia:
Phonologically, they do not have the L-H pattern of phonological words and may be low,
high, rising or falling, depending on their environment. Thus, phonologically speaking, they
are not integrated into the "word". However, preliminary evidence suggests that a sequence of
three or more clitics can combine to form their own "clitic word", with the typical L-H
pattern. With respect to their distribution, these same units may also be considered
"grammatical" or "syntactic" words, as they are "phrasal affixes", i.e., they attach to their
hosts at the syntactic level and are not suffixes. Thus, the relationship between phonological
and grammatical words in Kharia is highly complex, as many grammatical words do not form
part of any phonological word, at least not with respect to tone, whereas others can apparently
combine with other grammatical words to form a single phonological word. There even
appear to be cases where two syntactic words combine to form two phonological words, with
the phonological boundary occurring within the second syntactic word.
     Finally, there is the issue of what corresponds to a written word. Note that, although
literacy among the Kharia has been estimated to be as high as 90% and phoneme-grapheme
relations have largely been standardized, the issue of "orthographic words" has to my
knowledge never been dealt with and it is left to each writer's intuition where a boundary is to
be made. The general tendency here is to represent phonological words as single written units,
which, however, is complicated by the ambivalent status of the enclitics referred to above.
This often results in highly divergent writings even in shorter texts: While some writers tend
to consider enclitics part of the same "word" as their host, others prefer to write them
separately, although no writers are entirely consistent. Some writers also prefer to combine
some or all of the clitics to form written words. Consider the following example, showing the
three different writings for a "single word" by different writers:

lebu=ki=te=ga 'the men' (plural, object, focused)

4 words: lebu, ki, te, ga    2 words: lebu, kitega         1 word: lebukitega

In my proposed talk, I will deal with these and other issues in more detail, highlighting the
difficulties involved in any attempt to find a single definition of the "word" in Kharia.
               The Structural Identity of the Phonological Word in Vietnamese:
                                   constituency as domain
                                      Shanti Úlfsbjörninn
                          The School of Oriental and African Studies

Vietnamese has been described as lacking a phonological word, a minor syllable and minimal word
constraints (Schiering et al. 2007). It will be shown in this essay that Vietnamese phonology viewed
from a Government Phonology or Minimalist Phonology perspective does have minimal word
constraints and a differentiation between phonological words and clitics (Pham 2003) and prefixes
(Thompson 1959) which in Government Phonology (contra Optimality Theory) are also lexically
endowed with syllabic structure. What this paper will also illustrate is the correlation between
minimal word sizes and viability of holding stress.

In part one, it will be shown that in a good number of languages (Mandarin, Kammu, Khmer,
Italian, English, Turkish, Thai, Mongolian) we can see how, in order to gain primary stress and
therefore ‘wordship’, the syllabic structures in question must conform to a particular configuration,
namely: the Charette foot (Charette in press.). The Charette foot is a binary relationship between
two nuclei. One, the head, is obligatorily filled, and another, the dependent, obligatorily silent. In
this model, the head of the foot is head of the domain and therefore able to (A-)license its fringes
(Kaye et al. 1990; Harris 1997). All the while, being an obligatory binary relationship between two
nuclei automatically creates what have been called ‘bi-moraic’ or ‘ONON’ templates. The paper
will show evidence that if the minor syllable of Austro-Asiatic is large enough ‘ONON’ it can hold
stress/tone (Kammu) while if it cannot be fitted with a Charette foot, ie. ‘ON’ it may not bear stress
(Svantesson and Karlsson 2004; Schiering et al. ms.).

Part two will then show how one can motivate a templatic analysis for the phonological word of
Vietnamese; its minimal size, and most importantly its behaviour, contrasts with that of the prefixes
and clitics which attach themselves to these phonological words for stress. This is contra Schiering
et al. (2007), although, some of their other observations do make the Vietnamese phonological
domain a strange one.

Part three will show that, even within GP, Vietnamese has a strange phonological domain, the
templatic analysis of part two which shows Vietnamese having some sort of phonological word is
argued to be arbitrary and flawed in terms of licensing. More importantly it misses certain
phonological observations made in Vietnamese grammars and studies (particularly Pham 2003).
What is salvaged is the following: there is a structural correlation between the minimal word and
the ability to hold stress.

Part four will conclude the paper by illustrating that Pham’s (2003) observations that the nucleus
and the rhyme form a constituent is the key to understanding the true extend of the correlation
between minimal word size and stress. The observation that the Vietnamese rhyme manifests trade
offs of length between the vowel and the rhyme [ta:m] vs. [tam:] (Pham 2003) is exactly what
Minimalist Phonology (Pöchtrager 2006) would expect and mirrors the better known Italian stress
domain: [fat:o] vs. [fa:to] (Lowenstamm 1996). It will be argued that Vietnamese and other South
East Asian languages limit projecting their phonological words above this stress domain while
European languages’ phonological words may project further, although, all the while, maintaining
this stress domain as a constituent which behaves identically to SE Asian phonological words. This
is revealed in European languages’ when study is turned to their minimal words.
On the notion of a Cerberean node
Nozomi Kodama
Kumamoto University, Japan

Tonal features demarcate three tiers in Japanese prosodic structure. An intonational phrase, with an
edge tone, and optionally, a tone bearing segment at the rightmost end, consists of one or more
phonological phrases. A phonological phrase, demarcated by a tone rise from the beginning, has
other various prosodic as well as segmental features such as a mora timing domain, compulsory
devoicing of high vowels between two voiceless consonants and the domain of catathesis of
‘accented’ words as claimed by Pierrehumbert and Beckman(1988), which may be better explained
simply by the absence of the distinctive tone rise within. (See Kawakami 1956). Phonological words
that make up the phonological phrase have their respective lexically specified tonal feature that
realizes as one pitch fall (accent) at most within or at the rightmost edge. Phonological words and
phrases generally correspond to syntactic phrases and words (or clitic groups in Southwestern
dialects where postpositions and clitics are not lexically specified for tonal features), although a
single syntactic phrase or word may be broken into two or more phonological phrases. Examination
of such cases suggest that syntactic structural nodes at various (morpho-)syntactic levels are divided
into two types: one that phonologically unifies (either as a phonological word or phrase) and the
other that phonologically realize as divided by the phonological phrase boundary, the latter of which
I call a ‘Cerberean’ node, a non-endocentric node that does not have a single ‘head’ constituent.
 Examples: {a’zia no bu’Nka} “Asian cultures” {a’zia to}{ahurika no bu’Nka} “Asian and African
cultures”; aziasyo’koku   //Asian country// {a’zia}{ahurikasyo’koku} “Asian and African countries”
{a’zia ahurikasyo’koku}{rekihoL} “a visit of Asian and African countries”
Applied to other prosodically contrastive pairs such as {beLtoLbeN no}{uNmeL} “Fate, the
symphony, by Beethoven” vs. {beLtoLbeN no uNmeL} “Beethoven’s fate” or {ha’yaku}{ki’te}
“Come, hurry” vs. {ha’yaku ki’te} “Come early”, this departure from the unified treatment of
syntactic nodes may have further implications.

Note. I understand that Japanese is not included in South or Southeast Asian languages. So I am
currently working on similar Telugu constructions. At least morphologically similar constructions
abound in Telugu (such as Dvandva compounds that bear plural marking only on the latter
constituent). But as for prododic features, I could make only tentative analyses by September. Rather
I expect to gain insights from other presentations.
                            Wordhood in Riau Indonesian
                                          David Gil

      Isolating languages pose a particular problem for the notion of word. In languages
with a substantial amount of morphology it is generally possible to identify a set of
criterial features, some universal, others language-specific, distinguishing word-external
syntactic structure resulting from the concatenation of words and phrases, from word-
internal morphological structure resulting from processes such as affixation,
compounding and the like. Such criteria form the basis for the existence, within
linguistic theory, of autonomous disciplines of syntax and morphology. However, in
isolating languages, characterized by a paucity of morphological structure, there may not
be enough morphology to support a robust and systematic distinction between
morphological and syntactic structure. Accordingly, in isolating languages, there may be
relatively little evidence for the existence of words as a viable unit of linguistic structure,
as distinct from morphemes. Moreover, what little evidence is available tends to come
from the domain of phonology.
      This paper examines the notion of word in one extreme exemplar of the isolating
language type: Riau Indonesian. In general, it is argued that, compared to other, non-
isolating languages, the word plays a much smaller role in the grammar of Riau
Indonesian. Nevertheless, it is still possible to support a distinction between
morphological and syntactic structure in Riau Indonesian.
      This paper argues for the following morphological structure underlying words in
Riau Indonesian:

                                 outer word

                                 inner word

                     pre-core                 core foot

                                        syllable          syllable

                                    onset rhyme onset rhyme

Evidence for the above structure is derived from a number of different sources of
evidence, as summarized in the following table:
                        CORE FOOT        INNER WORD     INNER WORD OUTER WORD
                                         (terminal)     (non-terminal)
focus intonation        X
no reduction            X
epenthesis              X
loanword expansion      X
obligatory si-          X
N- realized as nge-     X
Warasa ludling          X                                               X
final k realized as ?                    X
Sabaha ludling                           X
Bahasisa ludling                         X
Pantun rhythm                            X
reduplication                                           X
spelling                                 X              X               X
Evidence for Word Structure Categories

In total, the above phenomena provide convincing evidence for the posited underlying
word structure, but hardly overwhelming evidence: the extent to which the grammar of
Riau Indonesian makes reference to the various categories constituting word structure is
still much less than in most other languages.

Shared By: