|Date: Mon, 16 Jan 2006 15:04:58 +1000
From: Ruth Singer and Sebastian Fedden
AUTHOR: Vajda, Edward J.
SERIES: Languages of the World/Materials 204
PUBLISHER: Lincom GmbH
Ruth Singer and Sebastian Fedden, Department of Linguistics and Applied
Linguistics, University of Melbourne.
The book summarises Vajda's work on tone and verb morphology in Ket
which has been published elsewhere in various forms. It is a grammar
based on Vajda's own field work together with research by other scholars
that was largely published in Russian and German. Vajda states that the
main purpose of the book is to make research on Ket more widely
accessible. This hundred-page book is bound to be useful to typologists or
anyone else with an interest in Ket.
THE KET LANGUAGE
Ket is the last remaining language of the Yeniseic language family. A few
hundred speakers of three different dialects live in north-central Siberia.
Ket is a clear isolate, as there are no strong similarities between the
Yeniseic lexicon, phonology or syntax and those of surrounding Eurasian
languages. Throughout the book, Vajda points out correspondences
between Ket and Athabaskan-Eyak-Tlingit language family but he notes
that he is still gathering the evidence to present a strong case for a genetic
relationship. Some notable typological features of Ket are the three-way
noun class distinction between masculine animate, feminine animate and
inanimate classes. These classes are only partially semantically transparent.
There are five tones in Ket but interestingly tones are not found in other
Siberian languages. The verb is polysynthetic with ten position slots. There
are five different strategies used to encode subject/object agreement and
the strategy used is lexically specified for each verb.
The book is divided into three main sections: phonology, morphology and
syntax. The chapter on Ket phonology comprises only nine pages and Vajda
himself calls it a phonological sketch of the language. The section on verb
morphology accounts for one third of the book. There is also an
introduction with background information on Ket and Vajda's research. The
book concludes with a short account of the lexicon and a text. This review
discusses the sections on phonology and the section on verb morphology
in detail and does not cover the sections on simple and complex clauses,
nominal morphology or noun phrases. The discussion of Ket phonology
was written by Sebastian Fedden and the discussion of Ket verb
morphology was written by Ruth Singer.
The phonology chapter falls into two main parts of roughly equal
proportions. In the first half of the chapter, Vajda deals with the segmental
phonology of Ket. He gives a succinct account of the set of segmental
phonemes and the phonological processes they are involved in for the
three dialects Southern Ket (SK), Northern Ket (NK), and Central Ket (CK).
The section also provides information on phonotactic restrictions and
syllabifcation of cluster-final sonorants and liquids. The second half is
dedicated to suprasegmental features of Ket, namely word-tone and other
prosodic phenomena at and beyond the phrase level. In this section, Vajda
analyses Ket as a word-tone language, i.e. a language in which the word
rather than the syllable is the relevant domain for tone assignment. On this
analysis, Ket has five distinctive tonemes: glottalized rising, falling, high-
even, rising/falling, and rising/high falling, which extend over the first or —
in the case of the latter two — the first two syllables of the phonological
word. Diachronically, most tonemes can be traced back to lost or reduced
consonantal elements in syllable-final clusters. All five contrastive pitch
contours are illustrated by diagrams which give the traces of fundamental
frequencies for relevant tokens. A short list of examples of minimal tone
pairs support the tonal analysis proposed by Vajda. The phonology chapter
is rounded off by a short account of how Russian loan words are treated
phonologically in Ket.
Ket verb morphology is very complex which has led to the development of
many different analyses. The account of Ket verb morphology in this book
is very similar to that in Vajda (2003), although there are more examples in
this book. Two aspects of Ket verb morphology are particularly unusual.
The syntactic alignment of Ket verbs is lexically specified - different verbs
mark their core arguments in different ways. Argument agreement may also
be lexicalised in which case it no longer gives a true reflection of the
Ket is of interest to typology in that it lacks a single type of syntactic
alignment. Vajda divides Ket verbs into five distinct conjugation classes
which are defined according to agreement strategy and the form of the
TAM morphemes. Each of the agreement patterns instantiates a different
type of syntactic alignment. For example, one conjugation class uses a
classic split-subject system; some intransitive subjects are cross-referenced
using the same strategy as for as transitive subjects while others are cross-
referenced by the same strategy as transitive objects. Another conjugation
class uses a straightforward ergative/absolutive system. The third
conjugation class uses a nominative/accusative system. The other two
conjugation classes use more unusual systems. Unlike most systems of verb
conjugation classes, the Ket system is not exhaustive. Some verbs cannot
be classified as having a conjugation type. These verbs include weather
verbs that have an inherent weather subject such as 'rain' or 'snow' and lack
agreement altogether. These cannot be categorised as belonging to any
particular conjugation class.
The verbal agreement strategy is lexically specified for Ket verbs. In
addition, the exact form that agreement takes is partly lexicalised for some
verbs. The morphemes in slots P1, P3 and P8 which typically function as
third person agreement morphemes are lexically specified for some verbs.
These morphemes no longer show agreement that corresponds to any
syntactic argument. Vajda calls these morphemes 'pseudo-actant markers'.
For example, in the absolutive (ergative/absolutive) conjugation the
morpheme da in P8 position usually indicates that the verb has a third
person transitive subject of feminine or inanimate noun class. However, a
small group of verbs have no argument that corresponds to this agreement
morpheme. These include verbs with meanings such as 'turn
yellow', 'broken' and 'slip'. These verbs have a single core participant which
is cross-referenced as an absolutive. The transitive subject marker has
become lexicalised in these verbs and is effectively a dummy argument.
The lexicalisation of verbal argument-agreement morphology is found in
various unrelated languages (see for example Evans 2004; Frantz 1995;
Harvey 2002; Singer (forthcoming). Lexicalised agreement is yet another
link between Ket and the Athabaskan-Eyak-Tlingit languages as lexicalised
agreement is also found in Athabaskan languages (Givón and Bommelyn
2000; Young et al. 1992). There are strong parallels between the meanings
of verbs with lexicalised agreement in Ket and those in other languages. In
Ket as elsewhere lexicalisation of agreement is restricted to third person
agreement. There a set of verbs with pseudo-actant markers which Vajda
describes separately as 'inversional verbs'. Most have intransitive stems and
a few such as 'I got sleepy (lit: sleep took me inside)' can have an external
nominal that corresponds to the dummy argument (in this case the noun
for 'sleep'). The inversional verbs encode their sentient experiencer as non-
subject (absolutive or inactive) and usually have a dummy subject. The
inversional verbs are similar to those described as experiencer-object verbs
in Iwaidjan languages (Evans 2004).
It is the strength of Vajda's analysis that some phonological processes for
consonants, such as sonorant devoicing, and the segmental and
suprasegmental features of vowels, i.e. vowel allophony, differences in
vowel quantity and voice quality (glottal or neutral), are entirely predictable
from the tonal contour of a syllable. This economical analysis gives a very
neat and consistent picture of Ket phonology. Furthermore, the description
of segmental allophony and tonemes is not confined to SK but includes
information on NK and CK as well. However, for the reader's orientation, it
would have been helpful to include a chart which gives an overview of all
segmental phonemes, their allophones, and the segmental and
suprasegmental contexts which determine allophony. Also, the posited
segmental phonemes are not substantiated by minimal pairs.
The analysis of Ket as having prosodic word-tone is plausible because in
mono- and disyllabic words tonal melodies contrast at the level of the word
and not the syllable (examples are almost exclusively nouns). It is not the
case that each syllable is allowed to bear a distinctive tone independent of
the other syllables in the word. The glottalized and the falling toneme occur
on monosyllables, the high-even toneme occurs on monosyllables but
allows some spreading to excrescent vowels in following syllables. The
rising/falling and rising/high falling tonemes, however, normally occupy
two consecutive syllables. Both of these tonemes can also appear on words
which surfaces as a monosyllable with a geminate vowel due to intervocalic
consonant deletion in a (underlyingly) disyllabic word, which is mainly a
feature of fast speech.
However, there is a major difference between Ket as a word-tone language
and typical word-tone languages, such as Shanghai, Mende (Sierra Leone)
and many Papuan languages (e.g. Kairi, Golin). In these languages tone is
lexically specified for roots. Tonal melodies are taken out of their respective
lexical entries as unpredictable information pertaining to a lexeme. In Ket,
on the other hand, one of the five tonemes is assigned to the first or the
first two syllables of a phonological word. Some toneless subject agreement
prefixes obviously fall outside the domain of tone assignment. As the Ket
verb is heavily prefixal, a verbal root is often quite far from the point where
tone is assigned to the phonological verbal word. If tone is purely prosodic
in Ket and is associated after the whole word has been assembled by the
morphology, how does the language know what tone it has to use? If tone
is lexical, why is it that roots in polysyllabic words normally do not bear any
tone? Answers to these questions would have been welcome.
In polysyllabic words, i.e. in the verbal vocabulary, which unfortunately is
hardly considered in the phonology chapter (but the reader can find many
examples in the chapter on verb morphology), the tone situation is
different from the nominal vocabulary in other respects as well. The tones
which appear on verbs are almost exclusively the rising/falling and the
rising/high falling ones. It would have been interesting to know in how far
verbal and nominal tone differ. Also, it does not become quite clear what
the tonal specification for those syllables are which do not carry any of the
five tonemes because they happen to lie outside the first or the first two
syllables of a word, a very common situation in polysyllabic verbs.
Moreover, sometimes a phonological verbal word seems to bear more than
one toneme (maybe because an incorporated element retains its tone?)
though one would probably not want to treat the verb as consisting of two
distinct phonological words. I am aware that detailed information on tone is
probably not the first priority for most readers of the book nor would it be
feasible to include such details in a phonological sketch, but I would have
liked more on tonal behavior in complex words, that is nominal compounds
and especially the verb.
Vajda describes the categorisation of Ket verbs into five conjugations as his
major contribution to Ket studies. He argues that each verb is lexically
specified for conjugation class, which is based on verbal agreement
strategy and TAM morphology. Vajda's analysis seems to have been well
received by others working on Ket. The main problem with the conjugation-
class analysis is that it separates what we would want to categorise as a
single verb on the basis of semantics and lexical morpheme distribution
into separate lexical items. The lexical morphemes of Ket verbs are found
in the P0 and P7 slots. However many verbs which have the same lexical
morphemes and the same basic meaning are classified as separate lexical
items in Vajda's analysis because they differ in conjugation class. These
include verbs related as causatives, inceptives, resultatives, reflexives,
reciprocals and infinitives. An example of three forms classified as separate
verbs because of their different conjugation classes is given below.
(Examples are from Vajda (2005); see Vajda's publications for abbreviations
and descriptions of the position slots marked by suffixed numerals.) Note
that all three verbs have the same lexical morphemes in their P0 and P7
Coreferential absolutive conjugation: quick trip
'I go to the river for a few hours and come back.'
Active conjugation: medium trip
'I go to the river for a few days and return.'
Coreferential inactive conjugation: long trip
'I go to the riverbank and remain there.'
Vajda's analysis creates a split between verb meaning and lexical
morphemes on the one hand and agreement strategy on the other. No
doubt this is a necessary step along the path to understanding Ket verb
morphology. However, it is hard to see how this system of verb
classification alone will provide much insight into the semantic and
syntactic properties of Ket verb roots. It is not actually clear whether Vajda
believes that the concept of 'verb root' is useful for describing Ket verbs.
The use of lexical morphemes to construct meaning is briefly discussed on
page 61 but this is a very interesting topic that deserves further research. It
would be interesting to see how productively lexical morphemes can
combine and how Ket verbs with two lexical morphemes compare with
complex verbs in other languages. Ultimately it seems that it will be
necessary to set up a number of cross-cutting verb categorisations, one of
which is based on verbal agreement strategy. It is likely that
correspondences between the classes set up by these categorisations will
emerge. Vajda greatly emphasises the independence of conjugation class
and verb meaning. He describes the five conjugation classes as simply
formal groupings with no semantic basis. However, the fifth conjugation
group, the 'possessive conjugation' clearly has a semantic basis. The verbs
in this group all refer to events portrayed ''as a spontaneous occurrence
(which it may not actually be)'' (2004:58) and most refer to the production
of sounds such as whistling or clapping. In addition, those verbs which fall
outside of the conjugation class system as all their arguments are internal
also form semantic groupings, such as weather verbs.
The discussion of pseudo-actant markers is very brief. Given that this is a
relatively unknown phenomenon it would have been interesting to know
more about it. In particular, the extent to which those uses described
as 'productive' occur and examples of all uses described as 'nonproductive'.
Vajda gives a concise account of both the segmental and the
suprasegmental phonology of Ket. His analysis is economical and
convincing. Tonal analysis is plausible and supported by phonetic data and
minimal tone pairs. More information on tone in complex words would
have been desirable to fully appreciate the complexity of the language's
tonal phonology. The summary of Ket verb morphology in this book will be
a great resource for typologists because of the many interesting questions
Ket raises. Vajda's analysis of Ket verb morphology demonstrates that not
all languages can be said to have a single syntactic alignment. The question
of what constitutes a single verb or verb root in Ket is also of importance to
typology. If verbs are categorised according to conjugation class then each
verb has its own syntactic alignment. If verbs are categorised according
their lexical morphemes, a single verb can have different syntactic
alignments. The discussion of pseudo-actant markers is tantalising and
there are comments throughout the section on verb morphology showing
that it has become lexicalised in different ways. Lexicalisation of agreement
and other aspects of verb morphology is yet another parallel between Ket
and Athabaskan-Eyak-Tlingit languages.
Evans, Nicholas. 2004. Experiencer objects in Iwaidjan languages
(Australia). Non-nominative subjects, ed. by Peri Bhaskararao and Karumuri
Venkata, 77-100. Amsterdam: John Benjamins.
Frantz, Donald G. 1995. Southern Tiwa argument structure. Grammatical
relations: Theoretical approaches to empirical questions, ed. by Clifford S.
Burgess, Katarzyna Dziwirek and Donna Gerdts, 75-98. Stanford: CSLI.
Givón, T. and Bommelyn, Loren. 2000. The evolution of de-transitive voice
in Tolowa Athabaskan. Studies in Language, 24.41-76.
Harvey, Mark. 2002. A Grammar of Gaagudju. Berlin: Mouton de Gruyter.
Singer, Ruth. (forthcoming). Agreement in Mawng, PhD thesis, Linguistics
and Applied Linguistics, University of Melbourne.
Vajda, Edward J. 2003. Ket verb structure in typological perspective.
Sprachtypologie und Universalienforschung (Language Typology and
Vajda, Edward J. 2005. Distinguishing referential from grammatical function
in morphological typology. Linguistic diversity and language theories, ed.
by Zygmunt Frajzyngier, David S. Rood and Adam Hodges, 397-420.
Amsterdam: John Benjamins.
Young, Robert W., Morgan, William and Midgette, Sally. 1992. Analytical
lexicon of Navajo. Albuquerque: University of New Mexico Press.
| ABOUT THE REVIEWER:
ABOUT THE REVIEWERS
Ruth Singer is currently completing a thesis which is a description of the
Australian language Mawng, with a focus on the different functions of
verbal agreement morphology. She is also interested in the typology of
lexicalised agreement, inclusory constructions and complex predicates.
Sebastian Fedden is working on a descriptive grammar of the Papuan
language Mian as a thesis. His main interests are in word tone and nominal