See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/274532717

A grammar of Kolyma Yukaghir

Article  in  Linguistic Typology · January 2005

DOI: 10.1515/lity.2005.9.2.299

CITATIONS READS
4 1,080

1 author:

Irina Nikolaeva
SOAS, University of London
58 PUBLICATIONS   456 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Prominent Possessors View project

Endangered languages and cultures of Siberia View project

All content following this page was uploaded by Irina Nikolaeva on 08 December 2016.

The user has requested enhancement of the downloaded file.

[Review of:] Elena Maslova, A grammar of Kolyma Yukaghir. (Mouton grammar library 27.) Berlin;
New York: Mouton de Gruyter, 2003, xviii + 609 pages, ISBN 3-11-017527 4.
Linguistic Typology, 9 (2). 299 -325

1. General

Kolyma Yukaghir and Tundra Yukaghir are closely related languages spoken in the north-eastern part
of Russia. Kolyma Yukaghir at present has about 40 speakers. The language is not entirely
undescribed: Jochelson (1905) and especially Krejnovič (1979, 1982) provide useful accounts of
morphology, as well as basic phonological and syntactic information, while Nyikolajeva (2000) is a
short overview of Yukaghir grammar in historical perspective. There are a number of articles by
various authors focusing on particular grammatical points. Several folklore text collections have been
published (Jochelson 1898, 1900; Nikolaeva 1989, 1997; Maslova 2002). These works present
essential facts about the language. However, most of them are written in Russian or Hungarian and
based on non-standard terminology and representational conventions, while the presentation of
examples uses a format that is opaque in terms of modern scholarly practice (i.e. without a
morpheme-by-morpheme analysis). Most importantly, the previous descriptions have many gaps by
modern standards and for the most part ignore syntax. Since they are incomplete, idiosyncratically
presented and not universally accessible, Yukaghir grammar has played little or no role in the
considerations of modern theoretical linguists and typologists.
Elena Maslova’s (henceforth abbreviated as M) new grammar will certainly remedy this situation
as it is the only lengthy book on Yukaghir easily accessible for the international linguistic audience. It
is written in English and provides a thoroughly organized and conventionally represented description
reflecting the standards of modern linguistic scholarship. The presentation of the data is transparent
and the discussion is clear. The description is intended to be comprehensive, i.e. to cover all major
aspects of the language. The grammar offers a thoughtful, new treatment of grammatical topics that
had not been adequately addressed before. Moreover, it considers many that the earlier works did not
discuss at all. The book is richly exemplified and therefore is valuable as a timeless piece of
documentation of this highly endangered and understudied language. A useful feature of the book is
that examples are clearly marked either as elicited or as obtained from naturally occurring discourse,
with preference given to the latter. The description is based on materials from M’s own fieldwork, as
well as on data from a published text collection.1
 The book contains a grammatical description, a 15 page word list of core vocabulary and words
occurring in the cited examples, a 6 page list of non-productive suffixes of verbal derivation, 19 pages
of analyzed texts, a list of references, and a subject index. The first two chapters offer the basic facts
about the language and its speakers, including population figures, geographical location, the
sociolinguistic situation, language contacts, and the relationship between Kolyma and Tundra
Yukaghir, together with relevant references.2 As is emphasized on p. 28, the grammar deviates from
the previous tradition in that it treats the Kolyma and Tundra varieties of Yukaghir as two closely
related languages rather than dialects of a single language, a choice which is certainly justifiable but
ultimately dependent on the criteria chosen. The first chapter also presents a short overview of
Kolyma Yukaghir grammar and describes the purpose and the structure of the book, emphasizing that
the grammatical description combines the “form-to-function” and “function-to-form” approaches. The
descriptive chapters are organized around general semantic categories such as “Aspect” or “Tense and
Mood”, while the internal structure of each section is determined by grammatical distinctions
expressed in Yukaghir. This certainly ensures a useful balance between widely assumed descriptive
concepts and language-particular phenomena.
The grammatical description starts with Chapter 3, which briefly deals with phonology and covers
such topics as phonemic inventory, syllable structure, vowel harmony, some phonological and
morphophonemic alternations, and word stress. Chapter 4 introduces the Yukaghir parts of speech and
the inflectional paradigms associated with each of them. Two major parts of speech, nouns and verbs,
are fairly easily distinguishable by inflectional criteria. Nominal inflection comprises Plural,
Possessive (in the 3rd person) and case suffixes (in that order). In the finite verbal inflection the
suffixes ordered as follows: mood – number – tense – person. Importantly, property words constitute a
subclass of verbs having the full finite paradigm and most of the non-finite forms. Thus, “the most
typologically important feature of the Yukaghir part-of-speech system is the absence of adjectives” (p.
61; but see section 4 below).
Chapter 5 outlines nominal morphology. An interesting feature of the Plural marking is that in the
context of the Possessive affix it can signify either the plurality of the entity denoted by the host noun,
or of the entity cross-referenced by the possessive marker, or both. Thus, numø-pe-gi (house-PL-POSS)
can mean ‘his/her houses’, ‘their house’, or ‘their houses’ (p. 86).3 It should be noted that the same is
observed in the neighboring language Yakut (Stachowski & Menz 1998: 422), so apparently we are
here dealing with an areal feature. The focus of Chapter 6 is verbal categories. The finite verbal forms
are organized around two parameters: the illocutionary force and the expression of the grammatical
Focus. Non-assertive force is expressed by means of the Imperative, Prohibitive and Interrogative.
2
The assertive (Indicative) paradigm has four sub-paradigms: intransitive Main forms, transitive Main
forms, intransitive Subject-Focus forms, and transitive Object-Focus forms. The finite verb agrees in
person/number with the subject, but in Subject-Focus and Object-Focus sub-paradigms agreement is
morphologically meager. The only obligatorily expressed tense distinction is Future vs. non-Future.
The distinction between past and non-past is achieved by aspectual affixes in combination with the
lexical meaning of the verb.
Chapter 7 deals with the morphology of closed classes. The syntactic part of the grammar opens
with Chapter 8 concentrating on the noun phrase and the postpositional phrase. It describes possessive
constructions, modification and NP coordination. Yukaghir has three strategies of NP conjunction,
comitative, asyndeton and conjunctional, but the latter is infrequent. As in many languages where the
conjunction ‘and’ is absent or marginal, the primary strategy employs the Comitative case, either on
one conjunct or on both.
Chapter 9 provides information on key aspects of clausal syntax. Yukaghir is known to be
typologically interesting in that it exhibits a close association between the case marking of core
participants and their information structure status (Comrie 1992; Fortescue 1996; Maslova 1997). The
primary participant in the finite intransitive clause and the secondary participant in the finite transitive
clause receive alternative grammatical encodings depending on their Focus/non-Focus status. In
intransitive clauses Yukaghir shows a special “Focus-oriented” case of split-intransitivity: the
non-Focus primary participant takes the Nominative (1a), while the Focus primary participant takes
the Predicative (1b).

(1) a. čomo:-d’e alme mon-i

big-ATTR shaman say-INTR.3SG
‘The great shaman said …’ (p. 89)
b. ta:t touke-lek jedej-l
then dog-PRED appear-SF
‘Then a/the dog appeared.’ (p. 9)

In transitive clauses the Focus object is encoded by the Predicative (2a). The marking of the
non-Focus object depends on the deictic status of the primary participant: it takes the Nominative if
the primary participant is the 1st or 2nd person (2b), or the Accusative/Instrumental if it is the 3rd
person (2c). In all three cases the primary participant stands in the Nominative.

3
(2) a. omo-s’e jaqada:s’e-lek kes’i:-l’el-me
good-ATTR horse-PRED bring-INFR-OF.2SG
‘You have brought a good horse.’ (p. 461)
b. met me:me: iŋi:
I bear be.afraid.TR.1SG
‘I am afraid of the bear.’ (p. 89)
c. titte ču:l-gele miŋ-ŋa:
their meat-ACC take-TR.3PL
‘They took their meat.’ (p. 93)

So Focus marking works on an ergative basis, cf. (1b) and (2a). This mechanism can be viewed as a
result of the close association of the Focus function with the primary participant of the intransitive
verb, on the one hand, and the secondary participant of the transitive verb, on the other hand, as
attested in the grammar of other languages (e.g. Du Bois 1987). What makes Yukaghir typologically
remarkable is that this association is highly grammaticalized.
The question that arises in connection with this system is that of grammatical relations, namely,
which grammatical units (if any) can be associated with the syntactic functions of the subject and the
direct object. The chapter thoroughly argues that although the non-Focus primary participant of
intransitive clauses demonstrates split case marking, in both cases it behaves as a grammatical subject.
Thus, Kolyma Yukaghir can be called a “subject-prominent” language. Subjecthood is defined on
participant roles, without regard for their surface expression (case marking). Similarly, direct object
properties are associated with the secondary participant of transitive clauses, independently on its
case.
Chapter 10 outlines the phenomenon of clause chaining, while Chapter 11 covers in turn three
major types of non-finite subordinating constructions: complement clauses, relative clauses, and
adverbial clauses. Sentential complementation is mostly expressed by clausal nominalizations.
Various types of sentential subjects and sentential objects are fairly typical from the cross-linguistic
point of view. Relative clauses are built by means of attributive forms of verbs and deverbal nominals
(the Action Nominal and Result Nominal). The following grammatical functions can be relatitivized:
the subject, the direct object, the spatial or temporal locative, and the possessor of the intransitive
subject. This distribution is typologically interesting, as it involves the highest and the lowest ranks in
the Noun Accessibility Hierarchy (Keenan & Comrie 1977). Finally, adverbal clauses are headed by
deverbal nominals with various case and postpositional markings.
4
 Chapter 12 addresses nominal predicates. It also deals more thoroughly with Focus constructions,
since the morphology employed in Focus marking shows a similarity to cleft and cleft-like structures.
As can be seen from (1) and (2) above, if a clause is marked for a grammatical Focus by means of the
Predicative case, the verb form is selected from the Subject-Focus or Object-Focus sub-paradigms.
Otherwise the verb form is selected from the Main sub-paradigm. The Subject-Focus sub-paradigm is
morphologically based on the Action Nominal, whereas the Object-Focus paradigm is based on the
Passive attributive form. So the Focus NP takes the Predicative marking which is also employed for
nominal predicates, while the verbal forms are identical to those employed for relativization. In
contrast to her own previous analysis of Focus constructions as sort of biclausal clefts (Maslova 1997:
465-466), M comes to a well-argued conclusion that the Focus NP is a clause-internal argument rather
than a clause-external main predicate. This is not to deny that Focus structures in modern Yukaghir
could have emerged as a result of a historical reanalysis of constructions where the Focus element
functioned as a predicate, as has been suggested (in somewhat opaque terminology) in Nikolaeva &
Xelimskij (1997: 162).
Chapter 13 describes the formation of imperatives, questions, negation, quotations, exclamatives
and optatives. The main focus of Chapter 14 is coreference and discourse coherence. It discusses such
phenomena as reflexivization, switch-reference, reference tracking by means of zero anaphora,
agreement markers, pronominalization and the like, as well as how the links between two finite
clauses or two clause chains are established at the paragraph level.
As much as I admire this informative grammar, I have a few reservations about three aspects of the
book, to which I turn in the following three sections of this review article.

2. Presentation of the data

The transcription in the book is intended to be “essentially phonemic” (p. 32), but there are many
inconsistencies in how the language data are presented. This is quite unfortunate, since the grammar is
likely to become a standard reference and its examples will certainly be cited in many works. Most
often inconsistencies concern the transcription of the vowel ø. For example, the vocabulary contains
the word aduøn ‘this one, that one’ (p. 542), which in other places is written as aduon (p. 225, 500).
The same concerns the words numø ‘house’ (p. 550) sometimes written as nume (p. 66, 87, 106), køj
‘boy’ (p. 548) sometimes written as koj (p. 277, 536), and others. In principle these inconsistencies
could reflect subphonemic variations. However, such variations are not addressed in the grammar.
Moreover, in my experience with the same speakers there is no variation in the pronunciation of these
5
words: all speakers consistently pronounce ø. Therefore the transcription of ø as e or o is likely to
represent auditory errors.
Similarly, M renders the negative copula as ojl’e (p. 145, 160, 299, 308, 332, 476, 477, 525, 579),
but according to my data it should be transcribed as øjl’e. This is confirmed by the lengthening of the
final vowel. As described on pp. 55-56, the final vowel is lengthened before certain affixes. The
quality of the resulting long vowel depends on the harmonic quality of the stem. In the word in
question the final vowel is lengthened into e:, e.g. øjle:-nuj (IMPF). If it contained the back vowel o,
the respective form would have been *ojla:-nuj, but this form does not exist.
There are numerous examples of the inconsistent use of vowel length, e.g. n’ied’i- ‘speak, tell’ (p.
229, 386, 406, 419, 473, 518, 564, 580) vs. n’ied’i:- (p. 550) and ažu ‘word’ (p. 230) instead of ažu:
(p. 543). The word mi:d’i: ‘sledge’ is sometimes written as mid’i: (e.g. p. 549), sometimes as mi:d’i
(p. 438) and sometimes as mied’i: (p. 576). Such inconsistencies can be quite annoying when they
occur on the same or adjacent pages and/or pertain to grammatical morphemes and functional words.
Thus, the pronouns ša:r ‘something’ and i:le ‘some’ (p. 253 ff.) are cited as šar and ile, respectively,
on p. 71, 172, 282, 387, 428, 535, and 580. The connective particle aji: (p. 542) is sometimes written
as a:ji (p. 173, 430, 440, 456) and the postposition arqa: ‘near’ (pp. 268-269) is sometimes written as
arqa (p. 380, 393, 429). The question word noŋo:n ‘what for’ is cited twice in this form on p. 488, but
on the very same page it is written as noŋon. The Prospective suffix is sometimes transcribed as
moži:- (p. 6, 26, 175, 176) and sometimes as moži- (p. 7, 139, 166, 167, 175, 440). The 3rd person
Plural affix of transitive verbs -ŋa: (p. 140) is written as -ŋa on p. 171, 201, 311, and 509. The
iterative affixes -ji:-, -d’i- ~ -n-d’i-, -či- ~ -s’i- and -uji- are transcribed either with a short vowel (e.g.
on p. 131, 195, 196, 197, 221, 236, 399, 539) or with a long vowel (e.g. on p. 193, 195, 197, 237, 287,
545) throughout the grammar.
The book does not explain why these words and affixes are written in several different forms. If
the alternations in vowel length were free variations, this would have been worth mentioning
explicitly, given how numerous they are. In fact, just as in the case of the phoneme ø, I have seen no
evidence that there are variations in vowel length: speakers always pronounce long vowels in the
words and suffixes in question.
In other instances alternative transcriptions do reflect subphonemic variations. For instance, back
vowels are optionally fronted after palatal consonants. This is reflected in transcriptional variations,
e.g. joŋži- ‘fall asleep’ (p. 547) vs. jeŋžo:d’e ‘dream’ (p. 505). However, these variations are not
addressed in the text. Other free variations are mentioned, but there is no indication that they are to be
reflected in the transcription of the book. P. 30 says that mid long vowels are sometimes pronounced
6
as diphthongs, usually in stressed syllables. They are written differently throughout the book,
sometimes in the same phonological environment, e.g. je:n ‘other’ (p. 113, 250, 255, 281, 283, 571)
vs. jien (p. 251, 546, also jen on p. 391) and jo:bi: ‘in the forest’ (p. 65, 546) vs. juobi: (p. 259, 293,
520).
The second problem concerns segmentation into morphemes. In many cases epenthetic u and i are
not indicated in the glosses as such. For instance, the attributive and predicative forms of numerals
form the following pairs: ataq-u-n and ataq-lo:- ‘two’, ilek-u-n and ilek-lo:- ‘four’, but ja:-n and
ja:-lo:- ‘three’ (segmentation into morphemes is mine). This data demonstrates that the Attributive
marker is -n, while the preceding u is epenthetic. This is especially obvious since -n is the regular
Attributive marker. Nevertheless, M writes the Attributive forms of numerals as ataq-un and ilek-un,
respectively, suggesting in her glosses that the vowel u belongs to the Attributive morpheme. The
same concerns the Dative forms of personal pronouns divided into morphemes as met-in (I-DAT),
although the Dative marker is in fact -n. Although not mentioned in the grammar, it occurs sometimes
with proper nouns, e.g. Jakutskej-n ‘to Jakutsk (DAT)’.
Furthermore, p. 57 states that vocalic epenthesis only occurs in word-final syllables. But it also
serves to prevent ill-formed word-internal clusters. For instance, the iterative suffix is cited as -(u)žu-
(p. 562), but the first u is epenthetic, as it appears only after consonantal stems, cf. eg-u-žu- ‘walk
(ITER)’ and qodo-žu- ‘lie (ITER)’. The idea that some word-internal occurrences of i and u are
epenthetic explains seemingly unmotivated variations in verbal stems. For example, the verb ‘come’ is
cited as kel- in the vocabulary, but on p. 317, 330 and 383 the same stem is represented as kelu-. The
stem ‘eat’ is cited as leg-, but in its derivative legite- ‘feed’ the morphemic division is legi-te-
(eat-CAUS), while -te- is a frequent Causative affix. The same concerns the verb šoh-e:- ‘get lost’ and
its Causative counterpart šohu-še-. There is no account of the alleged stem alternations kel-/kelu-,
leg-/legi- and šoh-/šohu-, although they may appear on the same page (e.g. p. 219). Yet, the
comparison with other verbs reveals that u and i here are epenthetic, and consequently the morpheme
division must be kel-u-, leg-i- and šoh-u-. As a matter of fact, most occurrences of short i and u in
non-first syllables are epenthetic. Although the phonological environment of word-internal epenthesis
is not always clear, most cases are straightforward. However, the grammar does not contain any
discussion of this matter. Epenthetic vowels are glossed as parts of either stems or affixes and, as a
result, some morphemes occur in several shapes.
There are other inconsistencies in representing verbal stems. For instance, the frequent stem ‘go’ is
cited in at least five variants: kew- (p. 106, 107, 408, 428, 462, 467, 506, 521), kebe- (p. 165, 170,
429), keb- (p. 459, 486, 569), køu- (p. 320) and kewe- (p. 164, 181, 259, 375, 395, 442, 520, 558, 566,
7
567, 568). This word is given in the vocabulary in the latter variant, although the same entry contains
its derivatives køw-de- ‘drive out’ and kep-či:- (the Iterative Causative). But the vowel e in the second
syllable is likely to belong to the perfective suffix, which often follows the stem. This suffix is cited as
-j- on p. 559, but for a number of verbs it takes the shape -ej- ~ -aj-, cf. moro-l-aj- (dress-0-PFV) on p.
559 and jed-ej- (be.visible-PFV) on p. 392.4
There is no definition of the sign “plus” (+) which is sometimes employed to separate morphemes.
I presume the idea here is that it indicates a somewhat looser connection between the respective
morpheme and the rest of the word. However, the morphological status of such morphemes is not
addressed. The plus is used for what M refers to as prefixes (more precisely they can be termed
proclitics, see Endo (1997)). They show properties of both bound morphemes and free words. On the
one hand, they are separable from the host word and do not participate in stress assignment.
Vowel-final proclitics can precede a vowel-initial stem, thus creating a sequence of two vowels,
which is normally prohibited word-internally. On the other hand, the vowel optionally assimilates in
rounding to the root. Next, the plus is employed before the Prospective marker -možu:- and the copula
ŋo:-. These morphemes are bound to the stem, but do not participate in vowel harmony or, rather,
create their own harmonic domain and bear independent stress. Finally, the plus sign is used to
separate the Attributive nominal from the head. The Attributive is fully productive and can be derived
from virtually any noun. It is phonologically independent from the following word except that the
Attributive marker typically shows the alternation n ~ d depending on the quality of the next segment.
In sum, there are at least three different cases here, but there is no discussion of this matter.
Lastly, although in the majority of cases M has provided satisfactory and well-formulated
translations, some corrections can be made. For example, amun’d’a: (p. 138) is not any kind of bony
fish, as its literal translation may suggest, but a specific sort of fish, namely, Amur ide (Leuciscus
waleckii).5 The stem tami- (p. 152) does not mean ‘help’ but ‘put on (clothes)’ (cf. тaмi- in Jochelson
1900), while emes’ke (p. 385) is not ‘noiselessly’ but ‘suddenly’ (cf. Yakut emiske ‘suddenly’). The
word šøtkuri: (p. 290, 490) is translated as ‘ski’, but it means ‘boots with soles made of the fur from
the inside of reindeer hooves’ (cf. шoткурi in Jochelson 1900).

3. Comprehensiveness

My second reservation is related to the coverage of the Yukaghir material. Obviously, no grammatical
account can be fully comprehensive and M is well aware of this fact. However she states that the
purpose of the grammar is “to cover all attested forms and constructions of contemporary Kolyma
8
Yukaghir” (p. 13). Although the book provides a very rich account of the language, I would not say
that this promise is entirely fulfilled. In this section I will cite a number of grammatical facts which
are not discussed in the book but appear to belong to the core grammar. Some of them can be found in
the published works on Yukaghir including the source analyzed by M. A few others are attested in the
course of my own fieldwork. For the most part I will not provide references and examples for reasons
of space, but they are available upon request.
The description of nominal derivational patterns is not quite comprehensive. Firstly, some
additions could have been made to the list of nominal derivational suffixes. The list of non-productive
suffixes on p. 134 can be augmented by -me, -le ~ -l’e, -r, and several others (see Krejnovič 1982).
The nominalizer -ben is not only used to substantivize deverbal nouns, but can also substantivize
adverbs and the Temporal form of nouns, e.g. pude-ben ‘top’ from pude ‘over, upward’. Action
Nominals derived from qualitative verbs can be nominalized by means of the pronominal element
-taŋ. What M describes as the suffix -(i)l’ (p. 134) is -l’, while i is epenthetic (see section 2). This
suffix seems to represent a variant of the frequent nominal suffix -l. The latter is attested, for example,
in such words as ad-i-l ‘youth’ (cf. ad-uø ‘son’ lit. ‘youth child’) and lud-u-l ‘iron’ (cf. lun-bugø
‘caldron’ lit. ‘iron box’). The grammar does not list the suffix -l among nominal derivational suffixes,
although it was mentioned by Krejnovič (1982: 85-86). Instead M states on p. 136 that many nouns in
-l result from the lexicalization of deverbal Action Nominals. However, for a number of nouns in -l,
such as those cited above, no corresponding verb exists in modern Yukaghir. Although they are not
synchronically derived, the affix -l can still be identified based on other related forms.
Second, p. 134 says that the only regular type of nominal compounding is the pattern where the
first component of the compound takes the Attributive marker -d ~ -n. No examples of another,
perhaps less regular, type where both components occur in the basic form are cited. In fact, there are
many such, e.g. o:ži:-nodo ‘duck’ lit. ‘water bird’ and melut-pod’erqo ‘round metal decoration women
wore on the chest’ lit. ‘chest sun’. In some compounds the first component is reduced, as in the words
cited in the previous paragraph, or relates to a qualitative verbal stem, e.g. kej-lebejdi: ‘cowberry’ lit.
‘red berry’ and čam-ani: ‘sheefish (Stenodus leucichthys)’ lit. ‘large fish’.
Derivational patterns of smaller grammatical classes could have been described in more detail. The
section on interjections on p. 489-490 does not mention that interjections have some morphology. A
subclass of interjections is derived from verbs by means of the suffix -ge ~ -he. Two examples are
cited in the grammar, but there are many more, e.g. mejlu-ge ‘what a sorrow!’ and pažilu-he ‘it is
disgusting!’. A few items are missing from the list of postpositions, which is intended to comprise “all
attested forms” (p. 267), e.g. mi:de ‘along, according to’, s’is’kin ‘along’ (see p. 566 of the grammar),
9
me:kl’e: (with derivatives) ‘near’, titel’o: ‘like, as’ (see p. 569 of the grammar), and n’a:čin ‘against,
in front of’. Information about particles, conjunctions and focus items is scattered throughout the
book, but it would be useful to see a more or less representative list of them in one place. Some are
not discussed, e.g. jaqaj ‘too, also’,6 ere ‘only’, as well as a number of discourse particles.
The grammar does not mention words derived by mean of the suffix -ke ~ -qo, such as pøm-ke
‘round’ (cf. pøm-ne- ‘round’), čič-ke ‘long, tall’ (cf. čit-ne- ‘long’), pon’-qo ‘white’ (cf. pojne-
‘white’), and jolo-qo ‘last’ (cf. jolo:- ‘last’). This group is remarkable because it shows neither
nominal nor verbal properties. The words in question have no inflectional morphology and, unlike
nouns and verbs, are only used in the attributive function. In fact, they can be termed adjectives. As
mentioned above, M concludes that Yukaghir has no adjectives, but this is not entirely accurate. She
herself points out that there are two pairs of attributive words that cannot be classified as nouns or
verbs, namely juku ‘small’ vs. čomo ‘big’, and čø:l’e ‘old, ancient’ vs. il’l’e ‘new, fresh, (an)other’.
The first pair is mostly used in frozen collocations but the second freely combines with any noun, and
so do the attributive words in -ke ~ -qo. To this list I can also add the rare word n’an’če ‘big, large’.
So Yukaghir appears to have a small closed class of basic adjectives which includes at least five
non-derived attributive words and several derived words showing totally idiosyncratic grammatical
behavior. This conclusion seems to argue against Wetzer’s (1996) point that even in languages where
adjectives are a distinct class they tend to associate with one of the two major word classes, noun or
verb.
Some additions could have been made to the inflectional patterns. It is not mentioned that the
Transformative form of nouns is compatible with the oblique Possessive affix -de- and so perhaps can
be analyzed as a case. P. 234 says that Free Possessive forms are typical of personal pronouns, but
absent from the nominal paradigm. However nouns are able to attach the same Free Possessive marker
-l’e, and so does the interrogative pronoun kin ‘who’. Although the Free Possessive forms primarily
form headless NPs, sometimes they serve as attributes, e.g. pugø-l’e čugø ‘summer road
(summer-POSS road)’. When describing the morphology of indefinite pronouns in Chapter 7, the
author surprisingly omits the primary type, namely indefinite pronouns formed from interrogative
pronouns by means of the clitic ere ~ ure. This pattern is quite regular and can derive indefinite
pronouns from virtually any interrogative. The clitic follows the case inflection, cf. kin-ere ‘somebody
(who-INDF)’ and kin-ŋin-ere ‘to somebody (who-DAT-INDF)’. As in many languages, it also functions
as a Focus item meaning ‘only’. There are other lacunae in the description of the pronominal system.
As for verbal morphology, Kolyma Yukaghir has an additional modal form not mentioned in the
grammar. This form is derived with the suffix -mi:bi:- ~ -bi:bi:- and typically denotes a situation
10
which the speaker judges as certain to take place, e.g. nu:-mi:bi:-m ‘(s)he will definitely find it
(find-MODAL-TR:3SG)’, but can also express various deontic meanings. As stated on p. 158, the
Different-Subject converb takes the Possessive marker -de- if the subject is in the 3rd person.
However, there is no Possessive marking in constructions with the expletive subject pøn, e.g. pøn
emidej-ge ‘when it got dark (darken-DS)’. Sometimes the Possessive marker is absent even with the
referential subject. It is not mentioned that the Periphrastic Past is compatible with the Periphrastic
Prospective. The resulting form expresses intention in the past and has three components: the Supine
of the content verb, the attributive form of the verb l’e- ‘be’ (optionally with the nominalizer -ben)
and (with the 1st or 2nd person subject) the copula o:- that takes subject agreement: ejre-din l’e-j(-ben)
o:-d’e (walk-SUP be-ATTR(-RELNR) COP-INTR.1SG) ‘I was going to walk’.
The functions of the verbal categories could have been described in more detail. The pro-verb
monohot- is not exclusively used for requests to repeat the previous utterance (p. 480), but as noticed
by Krejnovič, has a whole paradigm and serves in other functions, namely, in embedded questions and
with the negative marker for negation: el-monohod-i (NEG-say.what-INTR.3SG) ‘(s)he did not say
anything’. The grammar says that the Interrogative forms are only used with questions words (p. 143).
But the 1st person Singular Interrogative can have a hortative meaning (cf. Krejnovič 1982: 149-151).
A question word in not required in this case.

(3) met id’e:-de met ti: modo-m!

I alone-CP I here live-ITR.1SG
‘Let me live here alone!’

This form of the Interrogative often combines with the hortative particle ošto: ‘let’.
Turning now to syntax, a few case functions and valence patterns are not accounted for. The Dative
often expresses the purpose of the movement in combination with the frozen Imperfective converb of
the copula ‘be’ l’et. In addition to the metaphorical meanings of the Ablative listed on p. 111-113, the
Ablative denotes material and a partitive object, mainly with imperative verbs, see example (910b) in
the grammar and Nikolaeva & Xelimskij (1997: 160). The Instrumental denotes the measure of
comparison. The Transformative functions as an argument of significantly more verbs than the two
verbs listed on p. 449, e.g. qarte- ‘divide into parts’ and o:ži:- ‘consider as’. As described on p.
352-353 and 356-357, two-place intransitive verbs of speech take the Dative, while intransitive verbs
of movement open a directional slot, i.e. their second argument can be expressed by the Dative,
Locative or a postpositional phrase. But there are other patterns involving these cases. Some verbs of
11
emotions and perception take the Dative or a postpositional phrase, while the Locative is impossible.
Thus, the verb jø:de- ‘look at’ takes the object expressed by the Dative or a postpositional phrase with
the postposition laŋi(de) ‘at’. On the contrary, there are verbs that only take the Locative, e.g. čaŋde-
‘rub oneself against’ or tan’n’e- ‘owe’.
P. 213 says that the causee of transitive causatives is encoded as the indirect object, i.e. stands in
the Dative. It is not mentioned that the transitive causee exhibits alternative grammatical encodings: in
addition to the Dative, it may be encoded as the direct object and take the Accusative or the
Predicative. This fact was discussed in M’s own earlier paper on Yukaghir causatives (Maslova 1993).
The causative construction with the direct object causee is formally identical to other double object
patterns: ditransitive double object constructions (p. 354-355) and the constructions where some kind
of part-whole relationship obtains between two objects (p. 93-94). Note that geographically close
Tungus languages exhibit the same three types of double object construction, while the distinction
between the direct object and the indirect object causee correlates with an interpretational difference:
the former is available when the construction has a factitive meaning, while the latter allows both
factitive and permissive interpretations. It would be interesting to see whether a similar distinction
obtains in Yukaghir.
Secondary predicates are not addressed in the grammar. Yet Yukaghir possesses quite a distinct
strategy for resultative secondary predicates. The verbal form coincides with the verbal stem
augmented by the Resultative marker -o:.

(4) a. ø:rpe-p-ki amd-o: qodo:-l’el-ŋi

children-PL-3 die-RES lie-INFR-INTR.3PL
‘(It turns out that) his children lie dead.’
b. Noj aŋsi:-t lebe: arimel-ge an-i-pe-n’e n’aha: ejr-o:
Noah search-SS.IMPF earth bottom-LOC fish-0-PL-COM together walk-RES
nuŋ-ŋa:
find-TR.3PL
‘They looked for Noah and found him walking at the bottom of the earth together with the
fishes.’

As can be seen from these examples, the resultative secondary predicate is controlled either by the
subject (4a) or by the direct object (4b).

12
 The grammar does not describe many patterns of complex sentence formation, such as temporal
clauses with the Temporal form of Action Nominals, causal clauses formed with the Action Nominal
in the Instrumental, purpose clauses headed by the Transformative form of the Action Nominal,
“instead”-clauses formed by means of the Action Nominal with the postposition ta:her ‘instead’,
relative clauses formed with Different-Subject Converbs, and comparative finite clauses with the
conjunction n’eno:de ‘as if’. Only three types of sentential objects are listed: the object complement
clause, the Instrumental complement clause, and the directional (Locative or Dative) complement
clause. However, some sentential objects are expressed with the Action Nominal in the Prolative: they
involve such verbs as šørile- ‘write about’ and pundu- ‘tell about’, as well as the expression pas’i:be
‘thanks for’. The lists of complement-taking verbs could have been expanded. For instance, the verb
l’orqaj- ‘be unable’ introduces the Instrumental complement, and there are significantly more
predicates that take the sentential subject, such as čejluke:j ‘late, long ago’ and the like.
Finite complementation is not addressed. Many descriptions of Siberian languages suggest that the
native subordination strategies involve non-finite forms and no complementizers, while alternative
strategies, if present, have developed under Russian influence. Yet a number of languages exhibit
native types of finite complementation. A frequent historical source of such complementizers is verbs
that have undergone some kind of semantic bleaching (cf. Vincent 1993: 159). Languages areally
close to Yukaghir use grammaticalized forms of the verb of saying. For example, in the Tungus
languages Evenki (Brodskaja 1988: 72-73) and Udihe (Nikolaeva & Tolskaya 2001: 662-664) the
converb of the verb of saying serves as a complementizer in the complement clauses which do not
necessarily express the content of speech, as well as in purpose clauses. A similar construction exists
in Yakut (Čeremisina 1995: 225-226) and Buriat (Koptjevskaja-Tamm 1993: 29). Yukaghir shares
this property. It frequently uses the complementizer monut grammaticalized from the Same-Subject
Imperfective converb of the verb mon- ‘say’ which has fully lost its original semantics. It may
introduce complement clauses with verbs other than verbs of saying (5a) or purpose clauses (see
example (7) below)). The finite complement clause has the verb in the Indicative or the Interrogative.

(5) a. met naha: iŋlu:-bed-ek [kin’d’e bojs’e amde-t-i monut]

I very afraid-RELNR-PRED moon totally die-FUT-INTR.3SG that
‘I was very afraid that the Moon would completely die.’
b. mend’e-lek mødi:-l’el-u-l [kereke qaŋis’e-k kel-te-j monut]
news- PRED hear-INFR-0-OF Koriak hunter-PRED come-FUT-INTR.3SG that
‘He heard news that a Koriak hunter would come.’
13
Example (5b) additionally demonstrates that monut cannot be analyzed as a converb because there is
no same-subject relationship between the main clause subject and the presumed subject of monut.
Interestingly, the grammaticalization process in Yukaghir goes even further. The element monut
can serve as a kind of evidential marker in the main clause.

(6) jø:-ŋi-de-ge … taŋ čereuro: ložka-p-ki d’e:tehe ta:

see-PL-POSS-DS that silver spoon-PL-POSS still there
pon’o:-l’el du, čugø-ge løudu:-l’el du monut
remain-INFR.INTR.3SG either road-LOC fall-INFR.INTR.3SG either INFR

‘They saw that either his silver spoons had remained there or they had fallen out on the road.’

In (6) the element monut is located in the main clause and does not function as a complementizer. Its
role is apparently to strengthen the meaning of the Inferential marker: it indicates that the respective
event has not been witnessed directly but was inferred on the basis of secondary evidence. What we
find here is the subordinate clause which functions as the main clause. This sentence presents an
example of insubordination defined by Evans (forthc.) as “conventionalized independent
(main-clause) use of what, on prima facie evidence, appears to be a formally subordinate clause”.
Evans argues that epistemic and evidential meanings are commonly associated with the
insubordinated structure, and this is exactly what is observed in Kolyma Yukaghir.
I would like to have seen more information concentrated in one place in the grammar on
disjunction, subjectless (impersonal) constructions and the order of attributive modifiers within the NP
(relative clauses, nominal and verbal attributive forms). There is no summarized discussion on the
clausal position of adverbials and Focus elements. The grammar says that the Focus normally
precedes the verb (p. 341), but it remains unclear whether Focus is associated with the immediately
preverbal structural position as in many SOV languages (Kim 1988; É. Kiss 1995). A few examples
from the grammar suggest that lexical material may intervene between the Focus constituent and the
verb, but exactly what may be located there and under which conditions is not discussed. Related to
this, on p. 480-481 M says that the wh-question words are either sentence-initial or occur in the
position most typical of the respective constituent type. However, only very short examples with no
more than two or three constituents are provided to illustrate this point, so it is not obvious from the
surface order whether the non-initial wh-question is separable from the verb.

14
4. Interpretations

To begin this section, I would like to stress that most of M’s interpretations of the material are
convincing and well-argued. However, there are a few cases where I believe a different analysis can
be provided.
In the phonology chapter consonantal alternations could have been addressed in more detail. On p.
39-42 M describes syllable-final alternations of the voiced consonants such as devoicing, assimilation
to a nasal, word-final nasalization, and d ~ n alternation in nominalizers and Attributive nominal
forms. These processes are treated as independent and it is not clear if they interact and whether they
involve the same stems. In fact, some words show syllable-final devoicing whereas others show
nasalization in the same position, cf. pad-um ‘cook (TR.3SG)’ but pat (IMP) vs. jad-um ‘send (TR.3SG)’
but jan (IMP).
As a matter of fact, the range of alternations of syllable-final voiced consonants is broader than
suggested in the grammar, and they crucially fall into two types. One type of alternations involves the
consonants d, d’, g and h. They alternate with the (near-)homorganic nasals before a consonant or a
pause. This alternation does not depend on the quality of the next segment, cf. jad-um ‘send (TR.3SG)’)
but jan (IMP), jan-te-m (FUR-TR.3SG), jan-delle (SS.PFV), jan-mele (OF.3SG), jan-ŋa: (TR.3PL), etc. The
alternation d ~ n is especially frequent and is by no means limited to nominalizers, Attributive forms
and some verbs, as suggested in the grammar. The second type of alternations involves the consonants
d, g, h and b. They have different realizations in the syllable-final position depending on the quality of
the next segment. Many of them were described in Krejnovič (1982). For example, d is realized as t
before an occlusive or a pause, as l before l, and as n before a nasal, while g is realized as k before a
voiceless occlusive or a pause, as ŋ before a voiced occlusive or ŋ, as m before m and as u ~ w before
l. These alternations reflect the individual assimilative properties of each consonant and do not
conform to the general rule “occlusive + vowel ~ homorganic nasal + consonant”, as in the first type.7
Thus, the consonants d, g and h show two types of alternations. This situation can be lexically
marked, for example, by introducing different diacritic signs for different types of voiced consonants
or representing the first type as an underlying cluster. In any case it is worth mentioning in a
comprehensive grammar that there are two classes of stems with different assimilative properties.
Some objections can be made to the analysis of the consonantal inventory. According to M, b is
absent as a separate phoneme in Kolyma Yukaghir. She takes b to be a variant of w. However, b
shows the range of assimilative alternations parallel to the alternations of other voiced consonants in
the syllable-final position: p before a pause, u ~ w before an occlusive, m or l, and ŋ before ŋ. So we
15
are likely to be dealing here with the voiced labial stop phoneme. In this instance the system of
occlusives is symmetrical. All voiceless occlusives have a voiced counterpart: p ~ b, t ~ d, k ~ g, č ~
d’, š ~ ž, and q ~ h (in the latter there is also a difference in the mode of articulation).
On the other hand, w does not seem to have a phonemic status, but is rather optional
pronunciational variant of either b or u. When u is syllable-final, it is part of a diphthong, e.g. what M
transcribes as køwde- ‘drive out’ is phonologically køude-. In a similar manner, some occurrences of j
are better analyzed as glide components of raising diphthongs than consonants. This analysis is
supported by the phonotactic properties of the language. On p. 34 M says that consonantal clusters are
disallowed in the syllable-final position, except for j-initial clusters, e.g. qojl ‘God’ and qojl-get (ABL).
The diphthong analysis strengthens the generalization: if j and w are analyzed as vocalic, such words
do not present exceptions. Moreover, the diphthongs behave as long vowels with respect to
consonantal epenthesis, stress assignment and other phonological processes conditioned by syllable
weight.
Next, the account of vocalism and vowel harmony appears to miss some generalizations. On p.
52-55 M correctly says that all stems fall into two morphonological classes, {+E}-stems and
{-E}-stems. The “morphophoneme” {E} is postulated on the basis of its idiosyncratic
morphonological behavior. In particular, a number of suffixes show the alternation CE ~ E. When
they follow each other, the allophones are chosen in an alternating manner so that they create
sequences CE – C – CE or C – CE – C.8 What is not mentioned is that the distribution of {E} is not
random. Two-syllabic stems ending in a short vowel can only be of the types CV:CE, CVCCE and
CVCV, where V is not equal to {E}. The stems *CV:CV, *CVCCV or *CVCE are ruled out. In
subsequent syllables all short non-high vowels are represented as {E}, as evidenced by their
alternations and assimilative properties. This leads to the following generalization: the only non-high
short vowel that occurs after the initial sequence CVC, CVCV or CV: in native words is {E}. In other
words, short non-high vowels are underspecified with respect to place of articulation after the first
bimoraic foot (for detail see Nikolaeva 1998).9
This bears on the treatment of vowel harmony. M’s basic insights can be formulated in the
following way: (i) vowel harmony is based on the frontness feature, while the only vowel that can
optionally undergo labial harmony is {E}; (ii) the domain of the harmony is the word; (iii) the
harmony applies to long and short vowels, and (iv) apart from optional rounding, {E} does not
participate in the harmony but is realized as e. However, statement (iv) is only partially true. {E}
shows optional assimilation to o, e and sometimes a, but is mostly represented as the neutral vowel
schwa (rather than e). Generalizations (i) are (ii) do not seem to hold. Since the only non-high vowel
16
that occurs after the first foot is the non-harmonizing {E}, the harmony is restricted to the first foot,
i.e. the structures #CVCV. In contrast to what M says, they show rounding harmony in addition to
frontness harmony, but the rounding harmony operates only on a subsystem of vowels, namely, on
non-front vowels. These must share the rounding feature within the first foot, therefore the structures
*CaCo or *CoCa are forbidden.
As for long vowels, they participate in harmony only when they correspond to {E} which has
undergone secondary lengthening. In this case we find front-back harmonic pairs, for example, the
Ingressive suffix -a:- ~ -e:-. However, the grammar does not make it explicit that there is a number of
non-harmonizing suffixes with long vowels (e.g. the Diminutive -de: and the Causative-Resultative
-e:-), which apparently do not result from the lengthening of {E}. Long round vowels do not
harmonize either. So with the exception of secondary long vowels and the optional assimilation of {E}
the domain of the harmony is the first foot. However, synharmonism, i.e. alternations of velar and
uvular consonants depending of the frontness of the stem, applies to the whole word, except for a
number of inflectional morphemes. Although on p. 37 M promises to give a list of such morphemes, it
is not provided in the grammar.
Turning now to verbal morphology, on p. 140 the following paradigm for the Imperative is
described: 2SG -k (Simple form), -gek (Complex form), 3SG -gen, 1PL -ge, 2PL -ŋik (Simple form),
-ŋigek (Complex form), and 3PL -ŋigen. This analysis seems to be influenced by the Russian tradition
most richly represented in the typological work Xrakovskij (2001). The authors of this volume argue
that the core meaning of the imperative is speech causation irrespective of who is the performer of the
prescribed action. Under this broad definition the imperative subject’s person is irrelevant.
However, there are good formal and semantic grounds to draw a distinction between imperatives
and what can be referred to as hortatives. Imperatives are prototypically used to express directive
force. Roughly speaking, this means that the speaker intends the addressee to take the utterance as a
reason to perform action X and believes that the addressee can do X (Allan 1994). The cooperative
response from the addressee would be the doing of X. This description reflects the interaction between
speech act participants and has grammatical consequences for the imperative clause type: the
imperative is the 2nd person (Palmer 1986; Bybee, Perkins & Pagliuca 1994; Potsdam 1998, and
others). On the other hand, hortatives are not associated with the directive force but are closer to
expressives and are often loaded with additional modal meanings. If, in Sadock’s (1994) terminology,
imperatives emphasize the effective, social aspect of speech, hortatives rather express an affective,
emotional aspect primarily used to display the speaker’s feelings. The respective speech act can
roughly be defined as follows. The speaker believes it appropriate to express a wish that situation X
17
takes place and intends the utterance to be taken as expressing his/her wish that it takes place. Unlike
the imperatives, hortatives do not necessarily presuppose a non-verbal response. Therefore they do not
require the speaker’s assumption that the addressee can carry out event X and do not even require the
presence of the addressee for a successful speech act. Since hortatives are not primarily directive,
there are no restrictions on the hortative subject: they can express exhortations to any person (Sadock
& Zwicky 1985).
Hortatives tend formally to differ from 2nd person imperatives. For examples, in a number of
African languages there are special forms usually referred to as the subjunctive, injunctive, hortative
or optative (see Carlson 1994 for Supyire, Dimmendaal 1983 for Turkana, and Bergelson 2001 for
Bambara). They serve as complements of modality or manipulative verbs and may form a dependent
purpose clause. But when used independently, these forms either fulfill a sequential narrative function
or serve as hortatives. Unlike imperatives, which are limited to the 2nd person and express direct
prescriptions, hortatives occur in all three persons and convey the illocutionary meaning of
inducement or wish. When they are in the 2nd person, they express an indirect causation and function
as more polite or remote future imperatives. Thus, although imperatives and hortatives have common
components in their meaning (the reference to the future event intended by the speaker), languages
often grammaticalize the clause type distinction between them.
The situation in Yukaghir is basically similar. M’s paradigm appears to conflate two different
moods, the Imperative and the Hortative. The Imperative is only employed in the directive function
and is limited to the 2nd person. It is marked with the suffixes -k in the Singular and -ŋik in the Plural.
The Hortative has the following affixes: 2SG -gek, 3SG -gen, 1PL -ge, 2PL -ŋigek, and 3PL -ŋigen.
There is no distinct form for the 1st person Singular Hortative, probably because the respective
meaning is expressed by the 1st person Singular Interrogative (see example (3) above). This analysis
makes both paradigms morphologically homogenous.
The Hortative is not intended as a directive but rather emphasizes the speaker’s emotional state.
When used in the 2nd person it is perceived as the remote imperative. Indeed, according to M, the
difference between the Simple and the Complex forms of the 2nd person Imperative (i.e. between the
Imperative and the 2nd person Hortative, in my terminology) is that the latter expresses commands or
requests that have to be fulfilled later. Most importantly, although this is not mentioned in the
grammar, the Hortative can be used in dependent clauses, normally in purpose clauses with the
complementizer monut ‘for, in order to, so that’.

(7) n’e:-ŋa: [[pundu-ge-n monut [qamun ulege-k køude-ŋi]]

18
 call-TR.3PL tell-HORT-3 for how.much grass-PRED beat-TR.3PL
‘Theyi called themj so that theyj could tell how much grass they had mowed.’

As noticed in Sadock & Zwicky (1985: 174), true imperatives do not occur in dependent clauses, but
examples such as (7) are predicted by the Hortative analysis.
Next, Kolyma Yukaghir has two possessive constructions with lexical possessors: with and without
a 3rd person cross-reference marker on the head. Under M’s analysis, in both cases the possessor is
NP-internal. The agreeing possessor is not analyzed as a syntactic topic, because it does not have to be
clause-initial and may not be separated from the possessum by other clausal constituents. What is
more, there is no evidence for the prominence of syntactic topics other than possessors in Yukaghir
syntax.
However, topic analysis cannot be altogether excluded. First, the generalization that the agreeing
possessor must be NP-internal appears to be too strong. In some instances it can be separated from the
rest of the NP, as in (8) where it is separated by a free-standing adverb ta:t ‘so, then’.

(8) pulun-de: ta:t šaši-gi øjl’e

old.man-DIM so trap-POSS NEG.3SG

‘So the old man had no trap.’

M also cites one example (p. 345) where the agreeing possessor is right-detached, i.e. postposed after
the clause as some kind of afterthought. This shows that the agreeing possessor can be adjoined to the
clause either from the right or from the left. In the latter case it is reminiscent of double-subject
constructions in languages like Japanese and Chinese. Second, other languages (Ob-Ugric and
Samoyed) employ a similar detached construction to topicalize the possessor in the absence of
non-possessor clause-external topics. In these languages the topical status of arguments is
systematically expressed by clause-internal means. It is conceivable that the Yukaghir system of
Focus marking provides sufficient formal distinction between topical and non-topical participants.
Since the possessor is not involved in this system, topicalization of the possessor is encoded via
syntactic dislocation.
On the other hand, as correctly argued in the grammar (p. 303-304), when the agreeing internal
possessor is NP-internal, it is somewhat structurally more autonomous from the head than the non-
agreeing possessor. The demonstrative preceding the agreeing possessor modifies the latter (9a), while
in the absence of possessive agreement two interpretations are possible (9b).
19
(9) a. tiŋ pulut terike-gi
this old.man wife-POSS
‘the wife of this old man’ / *this wife of the old man’s’
b. tiŋ pulut terike
this old.man wife
‘the wife of this old man / this wife of the old man’s’

M generalizes on the basis of similar data that an NP where the possessor triggers agreement cannot
form a constituent of a larger NP. Apparently, what she means by a “larger NP” is an NP where the
head noun takes another dependent in addition to the possessor. Yet, the head-marked possessum can
take any dependents, including demonstratives. What is important is their position: they must follow
the possessor. If the demonstrative referring to the possessum is located between the possessor and the
latter, possessive agreement is required, e.g.:

(10) pulut tiŋ terike-gi / *terike

old.man this wife-POSS / wife
‘this wife or the old man’s’.

The contrast between (9b) and (10) appears to indicate that the phrase-internal positions of the
agreeing and non-agreeing possessor differ: the former precedes the determiner and the latter follows
it. To put it differently, the agreeing possessor is located at the left periphery of the NP.
Thus, the constructions with the agreeing possessor seem to fall into two types. In the first type the
possessor and the possessum do not form a single NP, but the possessor corresponds to an external
topic. In the second type the possessor remains NP-internal, but its position in the NP is such that it
precedes all other NP-internal constituents. This situation is not typologically unique. As well as the
Ob-Ugric and Samoyed mentioned above, it exists in another Uralic language, Hungarian (Szabolcsi
1994).
From a functional perspective, M argues that the agreeing possessor is construed either as the
clausal topic or as a participant involved in the clausal situation. This creates a sort of form-function
mismatch. Although under her analysis the possessor remains NP-internal, it plays an independent
role in the overall information structure of the clause (p. 296ff.). However, sometimes neither topic
nor “involvement” interpretations seem to be appropriate for the NP-internal possessor, e.g.:
20
(11) tabun kin zajezdka-gi?
that who fish.trap-POSS
‘Whose fish trap is that?’

In (11) the possessor kin cannot be topical, as it corresponds to a wh-question and so is associated
with the Focus function. It is equally unlikely that its referent (if any) is conceived as a participant of
the situation. So the NP-internal agreeing possessor does not always exhibit clause-level prominence.
It may be worth considering that it could be prominent at the level of the NP, as supported by the fact
that in most cases it tends to be inalienable and therefore indispensable for the conceptualization of
the possessum. If this is on the right track, the structure is essentially iconic. Both the external and the
internal agreeing possessors are assigned a special level of prominence in the interpretation of their
respective syntactic domain, i.e. the clause and the NP.
Related to that is the analysis of relative clauses. They are construed as possessive constructions:
the relative clause subject is encoded as grammatical possessor, i.e. it takes the same form as the
possessor in regular possessive constructions. According to the grammar, relative clauses fall into two
classes. In the Nominal strategy the subject is construed as the possessor of the dependent verbal
form. In the Attributive strategy the subject can be formally construed as the possessor of the
relativized NP. “These constructions are formally distinguished by the placement of the possessive
marker cross-referencing the subject of the relative clause: in the Attributive relative construction, it is
placed on the head noun, in the Nominal relative construction, on the verb form itself” (p. 416-417).
This classification may be worth rethinking, unless the two types are shown to exhibit further
syntactic distinctions. First, the grammar does not provide an account of what determines possessive
agreement in relative clauses and whether the conditions applying to it are similar to those applying to
regular possessive constructions. In the Attributive strategy the possessive marker is optional, cf.
examples (743a) and (743d) from the grammar, repeated here in a modified version.

(12) a. odu-pe modo-l jalhil-pe-gi

[Yukaghir-PL live-ANR] lake-PL-POSS
‘the lake where the Yukaghirs lived’
b. tiŋ lebe:-ge odu-pe titt-id’e: modo-l para:-ge
[this earth-LOC Yukaghir-PL they-INTS live-ANR] time-LOC
‘at the time when only the Yukaghirs inhabited this land’
21
In (12a) possessive agreement on the head cross-references the relative clause subject, but in (12b)
the agreement morpheme is absent. Still (12b) represents the Attributive strategy. In contrast, the
Nominal strategy is said not to allow optionality: the possessive marking must be present on the
verbal form. Consider (13), which repeats example (748b) from the book.

(13) čø:l’e-d omni:-pe en-pe-de para:-ge

[ancient-ATTR people-PL live-PL-POSS] time-LOC
‘at the time when ancient families were alive’

As was mentioned above, possessive agreement is not obligatory and is conditioned by the functional
prominence of the lexical possessor. If the suggested parallelism between possessive constructions
and relative clauses is to be maintained, we can expect that relative clauses demonstrate similar effects
independently of the location of the suffix. That is, the possessive suffix is likely to be optional both
in Nominal and Attributive relatives.
Second, in the absence of overt possessive marking it is virtually impossible to decide which of the
two strategies we are dealing with. This concerns the relative clauses where the subject is the 1st or 2nd
person and possessive agreement is altogether impossible, as well as the clauses with the non-agreeing
3rd person subject. For instance, (12b) can be classified together with (13) rather than with (12a).
Moreover, one example in the grammar (747a) demonstrates that the possessive affix expressing
subject agreement can be located both on the head noun and the Action Nominal. This example does
not fit neatly into the classification of relative clauses into Attributive and Nominal.
On p. 369 clause chaining is defined as “a structure in which one (typically final) clause contains a
finite verb form, and each non-final clause is marked to indicate whether or not it has the same subject
as the controlling clause”. Clause chaining is employed both to conjoin clauses and to modify the
proposition rendered by the finite clause. What is more, M in principle does not want to draw a strict
distinction between coordination and subordination in Yukaghir. She states that there are two
structural types of chains: linear structure, where a switch-reference clause is followed by the
controlling clause, and the “nesting” structure, where a switch-reference clause is located within the
controlling clause. Obviously, the availability of the nesting structure is the primary argument for
syntactic embedding. However, as stated on p. 379-380, chains are often structurally ambiguous in the
sense that it is impossible to decide based on their surface shape whether we are dealing with the

22
linear or nested structure. M concludes, much in line with Longacre (1985), that Yukaghir has no
strict formal opposition between coordination and subordination.
This analysis a priori excludes another possibility: that expressions identical on the surface obscure
underlying structural distinctions and therefore the ambiguity of chains may only be apparent. There
are languages where the structural differences in intra-clausal syntax are not accompanied by any
overt morphological or word order distinctions. For example, although traditional Altaic studies treat
all converbial clauses as subordination (e.g. Ubrjatova & Litvin 1986), Rudnickaja (1997) argues that
Korean converbs in -ko show properties either of subordination or of coordination, depending on how
their semantic relation with the finite clause is construed.
The situation in Yukaghir may be similar. The verbal forms involved in chaining are typical
instances of what Nedjalkov (1995) refers to as “contextual converbs”: the type of semantic relation
between the dependent and the main situation is not expressed unambiguously by the verbal form, but
determined contextually. Converbial clauses convey a wide range of semantic relations such as causal,
temporal, conditional, concessive, purpose or manner relations, and are even involved in
complementation (p. 413-414). At least in some of these cases we may well be dealing with
subordination. For instance, simultaneous same-subject chains normally have a purpose/manner
interpretation and are associated with the embedding nesting structure.
According to p. 380, the structural ambiguity is “an essential property of this strategy of clause
linking (rather than an artifact of inadequate tools of syntactic analysis)”. However, there is no
analysis of the syntactic properties of the chains as such. Coordinated and subordinated structures are
known to exhibit different properties with respect to anaphoric binding, gapping, extractions, and
scope of quantifiers (van Oirsouw 1987; Goodall 1987; Haspelmath 1995; Johannessen 1998, and
others). These tests are not discussed.10 In the absence of such a discussion the conclusion about
ambiguity between coordination and subordination seems premature.
I have a few minor terminological disagreements. I am not sure that the so-called Attributive
nominal form in -n ~ -d does not have the status of the Genitive case, given that it is not restricted to
attributive use, but may serve as the object of a postposition. There are also a few examples where it
functions as the object of a non-finite clause. The treatment of case marking of the non-Focus object
can be questioned. According to M, with the 3rd person subject non-Focus objects take either the
Accusative in -gele ~ -jle or the Instrumental in -le. The Instrumental mostly applies to unmodified
nouns, while the Accusative applies to modified nouns and inherently definite objects, such as proper
and possessive nouns, as well as personal and demonstrative pronouns. In other words, case marking
of the object depends on its deictic status. An alternative solution would be to state that there is no
23
difference in grammatical case but the deictic status of the NP influences the actual form of the case
marker. The Accusative has two variants, the “definite” variant -gele ~ -jle and the “indefinite” variant
-le, while the latter happens to be homonymous to the Instrumental and may ultimately go back to it
historically. What we gain with this analysis is a more unified account of case marking and a better
conformity to the typological data. Note that a similar solution has been employed elsewhere in the
grammar. On M’s own account, the Predicative, has two forms, -lek and -k, and their distribution
essentially depends on whether or not the host noun takes a modifier.
Finally, I do not think that the so-called Active verbal attributive form in -j(e) and the Passive
verbal attributive form in -me are successfully termed; at least some additional argumentation is
needed to show that the difference between them correlates to the active/passive distinction. On the
one hand, the form in -j(e) mostly serves to relativize the subject. But it is also employed for
relativization of the direct object (examples (281b) and (740) in the grammar). This function is
impossible for active participles. On the other hand, the form in -me relativizes direct objects and
locatives. However, there are no apparent reasons to analyze these constructions as passive, especially
in the absence of the syntactic passive outside relative clauses. Normally passive participles relativize
the passive subject, while the primary participant is encoded as an oblique. In contrast, in relative
constructions with me-forms the primary participant is encoded as the subject, as follows from the fact
that it triggers agreement. Perhaps it would be more appropriate to term the Passive and the Active
attributive forms the Perfective and Imperfective attributive forms, respectively. The tendency is such
that the form in -me mostly refers to an event that precedes in time the main clause event, while the
event described by the form in -j(e) either follows the main clause event, or coincides with it in time,
or has a universal timeless character.

5. Final remarks

To conclude, the author of the grammar under review should be given full credit for writing a very
thought provoking description of this intriguing language. Overall, I consider A grammar of Kolyma
Yukaghir an extremely useful piece of work. The book is both informed in modern linguistic
scholarship and rich in neatly organized empirical material. It brings to light many interesting new
facts about Kolyma Yukaghir and offers clear but sophisticated analysis of its intricate grammatical
phenomena. Since any aspect of grammar in a language as little researched as Yukaghir may prove a
theoretical challenge, the book is likely to be of great value for theoretical linguists and can be highly
recommended to typologists.
24
 Notes

1
Most of the data comes from Vasilij Shalugin who was the principal language consultant and the narrator of most of the
texts analyzed in the grammar (from Nikolaeva 1989). Sadly, this talented man passed away in 2002.
2
However, some relevant references are missing. For example, in the discussion of the genetic affiliation of Yukaghir on
p. 1 the most recent and comprehensive book on the topic (Fortescue 1998) is not mentioned. There is no mention of
publications employing the Yukaghir writing system, in particular the school primer (Spiridonov & Nikolaeva 1993).
3
In this review I have followed M’s transcriptional conventions. In particular, the letter h stands for a voiced uvular
fricative and d’ for a voiced palatal affricate. Yet there are two important deviations: I indicate long vowels with a colon
rather than a macron and do not transcribe long mid vowels as diphthongs, as M sometimes does. I also use M’s
conventions for glosses, although some minor deviations are possible.
4
Th vowel-initial variant of the affix may contract with the stem-final vowel, cf. čolho- ‘hollow’ vs. čolh-aj- (PFV). On p.
191 this word is written as čolha-j-, but this transcription leaves the stem-final alternation o ~ a unexplained.
5
Incidentally, this word is said to be formed via the lexicalization of the Active attributive form, presumably of the
Proprietive verb amun-n’e- ‘have bones’. However, the attributive form of this verb is amun-n’e-je rather than amun’d’a:
since, first, it never ends in a: (contrary to M’s assertion on p. 137) and, second, the Proprietive suffix here is -n’e- rather
than -n’-. The word amun’d’a: is likely to be formed from amun ‘bone’ with the suffix -ča: ‘abundant in something’
through the following regular phonological development: amun + ča: > amund’a: > amun’d’a:.
6
M transcribes it as ieqa:j.
7
This results from their historical origin: voiced consonants that participate in alternations of the first type go back to
clusters, while voiced consonants that participate in alternations of the second type go back to single consonants (see
Nikolaeva 1988).
8
On p. 49 M refers to Krejnovič (1982) for this generalization. In fact, Krejnovič only discussed the suffixes which
immediately follow the root and concluded that their phonological shape was determined by the root-final vowel. He made
no explicit claims about the shape of the following suffix in the sequence.
9
This most recent paper on Yukaghir phonology is not in M’s list of references. The paper demonstrates among other
things that the alternations of {E} are prosodically motivated.
10
A possible reason for this is that such data are difficult to obtain from texts. In her analysis M relied primarily on corpus
data, having stressed “the methodological drawbacks of interviews” (p. 16). But no corpus, however large, contains
information about all the areas of grammar a linguist might want to explore (cf. Fillmore 1992), and the analyzed Yukaghir
corpus is not even particularly large. With respect to the point in question, corpus data could have been usefully
augmented by elicited materials.

Abbreviations

ABL Ablative, ACC Accusative, ANR Action Nominalizer, ATTR Attributive, CAUS Causative, COM

Comitative, COP Copula, CP Connective Particle, DAT Dative, DETR Detransitive, DS Different Subject,

25
DIM Diminutive, FUT Future, HORT Hortative, IMP Imperative, IMPF Imperfective, INDF Indefinite, INFR

Inferential, INTR Intransitive, INTS Intensive, ITR Interrogative, ITER Iterative, LOC Locative, NEG

Negation, 0 Epenthetic element, OF Object Focus, PFV Perfective, PL Plural, POSS Possessive, PRED

Predicative, RES Resultative, RELNR Relative Nominalizer, SF Subject Focus, SG Singular, SS Same
Subject, SUP Supine, TR Transitive.

References

Allan, Keith (1994). Speech act classification and definition. In Asher, Ronald E. (ed.) The encyclopedia
of language and linguistics. Vol. 8, 4124-4127. Oxford: Pergamon Press.
Bergelson, Mira B. (2001). Imperative constructions in Bamana. In Xrakovskij (ed.), 485-498.
Brodskaja, Larisa. (1988). Složnopodčinennye predloženija v èvenkijskom jazyke. [Subordinate clauses
in Evenki.] Novosibirsk: Nauka.
Bybee, Joan L., Revere D. Perkins & William Pagliuca. (1994). The evolution of grammar: tense,
aspect and modality in the languages of the world. Chicago: University of Chicago Press.
Carlson, Robert. (1994). A grammar of Supyire. (Mouton grammar library 14). Berlin: Mouton de
Gruyter.
Čeremisina, Maja (ed.) (1995). Grammatika sovremennogo jakutskogo literaturnogo jazyka. Sintaksis.
[A grammar of modern literary Yakut. Syntax.] Novosibirsk: Nauka.
Comrie, Bernard. (1992). Focus in Yukagir (Tundra dialect). In Howard I. Aronson (ed.), The Non Slavic
languages of the USSR. Linguistic studies, 55-70. Chicago: Chicago Linguistic Society, University
of Chicago.
Dimmendaal, Gerrit J. (1983). The Turkana language. (Publications in African languages and linguistics
2). Dordrecht: Foris.
Du Bois, John W. (1987). The discourse basis of ergativity. Language 63: 805–55.
É. Kiss, Katalin (ed.) (1995). Discourse configurational languages. New York; Oxford: Oxford University
Press.
Endo, Fubito. (1997). Does Yukaghir really have prefixes? Keizai Riron (The Wakayama Economic
Review) 276: 138-149.
Evans, Nicholas (forthcoming). Insubordination and its uses. In Irina Nikolaeva & Frans Plank (eds.),
All over the clause: Finiteness. Oxford: Oxford University Press.
Fillmore, Charles J. (1992). "Corpus linguistics" or "Computer-aided armchair linguistics". In Svartvik,

26
 Jan (ed.), Directions in Corpus Linguistics. Proceedings of Nobel Symposium 82, Stockholm, 4-8
August 1991. (Trends in linguistics. Studies and monographs 65), 35-60. Berlin; New York: Mouton
de Gruyter.
Fortescue, Michael (1996). Grammaticalized focus in Yukaghir. Is it really grammaticalized and is it
really focus? In Elisabeth Engberg-Pedersen et al. (eds.), Content, expression and structure. Studies
in Danish Functional Grammar, 17-38. Amsterdam; Philadelphia: John Benjamins.
_____. (1998). Language relations across Bering Strait: Reappraising the archaeological and linguistic
evidence. London; Washington: Cassell.
Goodall, Grant T. (1987). Parallel structures in syntax: coordination, causatives, and restructuring.
(Cambridge studies in linguistics 46). Cambridge: Cambridge University Press.
Johannessen, Janne Bondi (1998). Coordination. (Oxford studies in comparative syntax). New York:
Oxford University Press.
Haspelmath, Martin (1995). The converb as a cross-linguistically valid category. In Haspelmath &
Ekkehard König (eds.), 1-55.
Haspelmath, Martin & Ekkehard König (eds.) (1995). Converbs in cross-linguistic perspective.
(Empirical approaches to language typology 13). Berlin; New York: Mouton de Gruyter. 1-55.
Kim, Alan (1988). Preverbal focus position in type XIII languages. In Michael Hammond, Edith Moravcsik,
& Jessica Wirth (eds.), Studies in syntactic typology, 148–171. Amsterdam: John Benjamins.
Jochelson, Woldemar (1898). Obrazcy materialov po izučeniju jukagirskogo jazyka i fol'klora,
sobrannye v Jakutskoj èkspedicii [Samples of materials on Yukaghir language and folklore collected
during the Yakut expedition.] Izvestija Imperatorskoj Akademii Nauk 8: 151-177.
______. (1900). Materialy po izučeniju jukagirskogo jazyka i fol'klora. [Materials on Yukaghir language
and folklore.] Saint Petersburg: Imperatorskaja Akademija Nauk.
_______. (1905). Essay on the Grammar of the Yukaghir Language. Annals of the New York Academy of
Sciences 16, 2.
Keenan, Edward L. & Bernard Comrie (1977). NP accessibility and universal grammar. Linguistic
Inquiry 8: 63-100.
Koptjevskaja-Tamm, Maria (1993). Nominalizations. Theoretical linguistics series. London: Routledge.
Krejnovič, Eruxim (1979). Jukagirskij jazyk [The Yukaghir language.] In Jazyki Azii i Afriki. Vol. 3.
Jazyki drevnej Perednej Azii: nesemitskie, iberijsko-kavkazskie jazyki, paleoaziatskie jazyki, 348–
368. Moscow: Nauka.
_____. (1982). Issledovanija i materially po jukagirskomu jazyku [Studies and materials on the Yukaghir
language.] Leningrad: Nauka.

27
Longacre, Robert E. (1985). Sentences as combination of clauses. In Shopen, Timothy (ed.), Language
typology and syntactic description. Complex constructions, 235-286. Cambridge: Cambridge
University Press.
Maslova, Elena (1993). The causative in Yukaghir. In Comrie, Bernard & Maria Polinsky (eds.)
Causatives and transitivity. (Studies in language Companion series 23), 271-285. Amsterdam:
Benjamins.
____. (1997). Yukaghir focus in a typological perspective. Journal of Pragmatics 27: 457-475.
Maslova, Elena (ed.) (2002). Yukaghir Texts. (Tunguso-Sibirica, 7). Wiesbaden: Harrassowitz.
Nedjalkov, Vladimir (1995). Some typological parameters of converbs. In Haspelmath & König (eds.),
97-136.
Nikolaeva, Irina (ed.) (1989). Fol’klor jukagirov verxnej Kolymy [Folklore of the upper Kolyma
Yukaghirs.] Vol. 1-2. Yakutsk: JaGU.
_____. (1997). Yukaghir texts. (Specimina Sibirica, 13). Szombathely: Savariae.
Nikolaeva, Irina. (1998). Optimal syllables are not always optimal: a prosodic structure of Yukaghir.
Lingua 105: 201-229.
Nikolaeva, Irina & Evgenij Xelimskij (1997). Jukagirskij jazyk [The Yukaghir language.] In Volodin,
Aleksandr (ed.) Jazyki mira. Paleoaziatskie jazyki, 155-168. Moscow: Indrik.
Nikolaeva, Irina & Maria Tolskaya (2001). A grammar of Udihe. (Mouton Grammar Library 21). Berlin:
Mouton de Gruyter.
Nyikolajeva, Irina (2000). Chrestomathia jukagirica (Uralisztikai Tanulmányok 10). Budapest: ELTE
BTK.
Palmer, Frank Robert (1986). Mood and modality. (Cambridge textbooks in linguistics). Cambridge:
Cambridge University Press.
Potsdam, Eric (1998). Syntactic issues in the English imperative. (Outstanding dissertations in
linguistics.) New York-London: Garland
Rudnickaja, Elena (1997). Problema altajskogo sočinenija v korejskom jazyke [The problem of “Altaic
coordination” in Korean.] Voprosy jazykoznanija 6: 89-99.
Sadock, Jerrold M. (1994). Toward a grammatically realistic typology of speech acts. In Savas L.
Tsohatzidis (ed.), Foundations of speech act theory: philosophical and linguistic perspectives,
393-406. London: Routledge.
Sadock, Jerrold M. & Arnold M. Zwicky (1985). Speech act distinctions in syntax. In Shopen, Timothy
(ed.), Language typology and syntactic description. Vol. 1. Clause structure, 155-197. London; New
York; New Rochelle; Melbourne; Sydney: Cambridge University Press.

28
 Spiridonov Vasilij & Irina Nikolaeva (1993). Bukvar’ dlja 1 klassa jukarirskix škol (Verxnekolymskij
dialect). [An alphabet primer for grade 1 of Yukaghir schools (Upper Kolyma dialect).] St.
Petersburg: Prosveščenie.
Stachowski, Marek & Astrid Menz (1998). Yakut. In Johanson, Lars & Éva Á. Csató (eds.), The Turkic
languages, 417-433. London; New York: Routledge.
Szabolcsi, Anna. 1994. The noun phrase. In: Kiefer, Ferenc & Katalin É. Kiss (eds.), The syntactic
structure of Hungarian. (Syntax and Semantics 27), 179-274. San Diego; New York; Boston;
London; Sydney; Tokyo; Toronto: Academic Press.
Ubrjatova, Elizaveta & Fedor Litvin (eds.) (1986). Strukturnye tipy sintetičeskix polipredikativnyx
konstrukcij v jazykax raznyx sistem. [Structural types of polipredicative sentences with synthetic
verbal forms in languages of different types.] Novosibirsk: Nauka.
Van Oirsouw, Robert R. (1987). The syntax of coordination. (Croom Helm linguistics series). London:
Croom Helm.
Vincent, Nigel (1993). Head- versus dependent-marking: the case of the clause. In Corbett, Grevill G.,
Norman M. Fraser & Scott McGlashan (eds.), Heads in grammatical theory, 140-163. Cambridge:
Cambridge University Press.
Wetzer, Harrie (1996). The typology of adjectival predication. (Empirical approaches to language
typology 17). Berlin: Mouton de Gruyter.
Xrakovskij, Victor (2001). Typology of Imperative Constructions. München: Lincom Europa.

29

View publication stats