AUTHOR: Kruspe, Nicole TITLE: A Grammar of Semelai SERIES: Cambridge Grammatical Descriptions PUBLISHER: Cambridge University Press YEAR: 2004 Announced at http://linguistlist.org./issues/14/14-2149.html
Pogibenko Tamara, Institute of Oriental Studies, Moscow
This monograph is a comprehensive description of Semelai, a language that belongs to Aslian group of Mon-Khmer (MK) branch of Austroasiatic (AA) family. Semelai is spoken by some 4,000 people in Peninsular Malaysia. The monograph is a revision of the doctoral dissertation submitted to the University of Melbourne in 1999.The study is based on a corpus of data collected during the period of fieldwork from July 1990 to June 1991 as well as the data collected during visits in 2000-1, while Kruspe was a postdoctoral student at the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany.
This book is a pioneer full description of a language belonging to the least explored group of AA language family, an area much underrepresented in cross-linguistic studies. Located at the periphery of AA, Semelai and other Aslian languages are key for our understanding of the paths of historical development and typological evolution of a most challenging language family, which includes such typological extremes as polysyllabic Munda languages with complex verb and noun morphology on the one hand, and monosyllabic isolating Vietnamese, on the other hand. The description is carried out in terms of Basic Linguistic Theory; an abundant number of clearly presented examples, and an impressive body of texts representing different oral genres are included, therefore this work would be equally useful to researchers of the AA languages, typologists and descriptivists. The method for data collection was based on the following principles: monolingual work with a cross-section of the community; collecting narratives in a range of genres, the narrator performing in a setting of an interactive audience, direct elicitation of language material kept to a minimum; texts' transcription and annotation being carried out with one principle speaker in group settings which included people of all ages from different settlements, spontaneous and lively debate increasing the researcher's insight into the intricacies and subtleties of the language.
The book consists of fourteen chapters (the last chapter is the collection of texts), the list of Tables and figures, Preface, Acknowledgements, list of Abbreviations and conventions, three maps, Vocabulary (Semelai-English), References, and Index.
Chapter 1 ''Semelai'' (1-31) provides general information about the typological characteristics of Semelai, the subgrouping of the Aslian languages and their geographical distribution, affiliation of Aslian with MK, previous research, the prehistory of the Aslian languages, ethnographic background. Semelai exhibits many of the typological features of a mainland AA language, as well as incorporates aspects of Austronesian due to sustained contact with Malay. It is an isolating language with agglutinating features and complex morphology. The inventory of affixes includes those of AA origin, as well as those borrowed from Malay. These two different sets of affixes have different forms of attachment: a) non-concatenative type for prefixes and infixes of MK origin b) concatenative type for borrowed prefixes, suffixes and a circumfix. A most intriguing morphophonological process is coda/onset copying. Roots can feed multiple affixation, combining affixes from both systems of attachment. Borrowed Malay lexemes can feed affixation of indigenous affixes. Word order in NPs is essentially MK: the attribute, possessor, and relative clause follow the head. Word order at the clause level drastically differs from that in most non-Aslian MK languages: it is not used to express grammatical relations, it is fluid, with variations driven by pragmatic factors. Semelai combines core role marking on the predicator and NPs: subject marking in transitive clauses on the predicator and subject NP and object marking on the object NP. Semelai speakers practice avoidance speech styles which depend on the environment, including particular household activity locations.
Chapter 2 ''Phonology and Phonotactics'' (32-60) is a full description of the phonological system. Semelai has a rich phonemic inventory: thirty-two consonants and twenty vowels (ten oral vowels, each with a nasal counterpart). The structure of the syllable is essential to the organization of the phonology and morphological processes. Semelai, as all Aslian languages, has a strong monosyllabic tendency, although disyllabic and polysyllabic words are attested. Word can consist of one main syllable, or main syllable with one or two presyllables.
Chapter 3 ''Morphology'' (61-93) describes morphemic composition of words in terms of structures and structural processes. Two subsystems are described separately, they are the indigenous system and the borrowed system. Structural units such as lexeme, grammatical word, clitic, root, affix and base are distinguished, defined and their phonological and morphophonological features described. A phonologically specific affix type involving coda and onset copy is described in terms of templatic morphology.
Chapter 4 ''Word classes'' (94-104) deals with general principles of division into word classes; a short description of each class is given in terms of subclasses, word formation, inflexion and function. The following word classes are distinguished for Semelai: nominal, verb, expressive, preposition, adverb, auxiliary, existential and ascriptive predicators, negator, connective, interjection. Generally, assignment to a particular word class raises no problems (including noun and verb distinction) except for some specific cases such as Malay loans, which can function both as verbs and nouns, a few cases of verbs which are used as modifiers without undergoing nominalisation, locative nominals, which may function as prepositions. Each word class comprises prototypical, less prototypical and boundary members. In contrast to other AA languages, deverbal nouns in Semelai do not generally function as arguments except subjects; they can also function as predicates in non-verbal clauses and in interrogative sentences. Indigenous classifiers are derived from nouns. Prepositions include deictic dimensions: three degrees of height and speaker orientation; they also feed nominalization.
Derivational morphology is most productive for verb to verb derivation, more constrained for verb to noun derivation, restricted for noun to noun derivation and highly restricted for noun to verb derivation. Affixes of the same phonological form found in verb to noun and noun to noun derivation are treated as homophones.
In chapter 5 ''The verb'' (105-169), word structure, morphological categories and semantic classification of verbs are described. A characteristic feature of verb morphology in Semelai is ''peaceful'' and even ''cooperative'' coexistence of forms with similar functions derived by indigenous and Malay borrowed affixes. Morphology of intransitive verbs is simpler than that of transitive verbs. Verb form can be root form or can include a number of affixes and/or clitics. Affixes are mostly prefixes or infixes, but there is also a suffix and a circumfix. Trivalent verbs do not feed derivational processes, even valency decreasing ones. Verb roots can host combination of up to two affixes. The particular function of the derivational morphemes is dependent on the root. Thus, coda copy can derive the following: imperfective (from transitive verbs), new lexical item (from active intransitive verbs), noun (from adjectives), intransitive verbs (from bound roots), and stative intransitive verbs (from nouns). Section 5.3 describes derivational morphology, i.e. verbal forms, derived from verbal or nominal roots with prefixes, infixes and suffixes including those borrowed from Malay, the latter include: the imperfective prefix mN-, found in those Malay loans where coda copy is inapplicable; the middle voice prefix b(r)-; the collective circumfix br-...an; the causative prefix p-; the applicative suffix –i; the happenstance prefix t(r)-. The forms derived with AA inventory include: the imperfective derived with coda copy; the causative with par-, tar-, -r-; the comparative with raʔ-/-raʔ-; and the intensive derived by means of light syllable reduplication. Functions of verb forms in different environments are described in detail and several subsections are supplied with comparative notes concerning corresponding affixes in other Aslian languages and Malay. Section 5.5 describes verbal clitics: pronominal and irrealis proclitics (which are mutually exclusive and primarily associated with transitive roots or stems and a limited number of intransitive roots), imminent aspect proclitic. Section 5.6 deals with means of expressing modal meanings.
In Chapter 6 ''Pronouns: personal, ignorative, and demonstrative'' (170-201), three types of pronouns are described. Personal pronouns include familiar forms for 1SG and 2SG, inclusive/exclusive distinction for 1SG and 1PL, and plural forms both simple and derived with the clitic =ʔen (first and second person free pronouns). Third person singular pronominal clitic can be used with a distributive meaning. Ignoratives in Semelai are numerous and they are described in detail. Two different constructions are used to express interrogative ''whose'': one with ''who'' in postposition to the noun head and the other with ''who'' in preposition to the noun head with the possessive preposition. Semelai has three basic demonstratives of distance orientation -- proximal, distal, and non-specific proximal, it also has an elaborated paradigm of spatial demonstrative forms, which are derived from deictically specified directionals with the help of the prefix naʔ- ~ na-. Personal pronouns can be modified by proximal/distal demonstratives and by the temporal deictic. Kruspe also observes that the demonstrative may precede the head in imperative and directive constructions.
Chapter 7 ''The noun phrase'' (202-226) describes the structure of the noun phrase and its constituents: numerals, classifiers, measure nouns, quantifiers; different types of noun phrases are also described: the attributive phrase and the associative phrase, the latter covers a range of relations holding between two nouns: filiation, whole-part relation, possession, the second component can also denote performer, creator, user, beneficiary, purpose, source, time. Of all these relations only the possessive one can be (optionally) marked by the possessive preposition and the enclitic =hn. Bordering phenomena such as compounds and criteria for distinguishing them from the associative phrase construction are also dealt with. This chapter includes sub-section ''Nominal derivation'' which is comprised of two paragraphs: ''Noun to noun derivation'' and ''Deverbal nouns''. Last but not least, it includes small paragraphs dealing with syntactic peculiarities of derived nouns.
Chapter 8 ''Prepositions and the prepositional phrase'' (227-246) deals with prepositions, the structure of the prepositional phrase and its functions. Much of the space is taken up by deictic locative prepositions and directionals. Morphosyntactic characteristics of individual prepositions are described in detail.
Chapter 9 (pp. 247-268) focuses on grammatical relations, constituent order and coding strategies. Grammatical relation coding and constituent order are the features that drastically distinguish Semelai from the majority of MK languages. Word order in Semelai does not encode grammatical relations and allows a range of variations. The options are determined by sentence structure, e.g. transitive vs. intransitive vs. non-verbal clause; such pragmatic factors as type of reference, e.g. general/background statements vs. perfective clauses (describing a specific instance of an event); and communication structure of discourse. General statements in Semelai have invariable SVO order. The unmarked order in the transitive perfective clause is VSO ~ VOS, V-IO-O. Three argument sequences are not attested in spontaneous discourse. Fronting is a strategy to place an NP into focus position. Fronting does not affect verbal morphology but fronted core NPs lose their role-marking. Interrogative formations also require fronting of the ignorative and relativisation of the clause.
Two basic coding strategies are discussed in detail: agent cross-referencing and NP role coding. Post-verbal agents are marked with la=, which has other functions: introduces an NP or a clause providing information about the source of causation or reason; only post-verbal agents can be cross-referenced on the verb with pronominal proclitics. A small subset of intransitive verbs (human activity and emotion) allow their sole NP to be marked as agent. Agent marker can be omitted when it is recoverable from the context or universal knowledge as well as in absence of post-verbal O. Post-verbal direct objects can be marked by hn=, which has other functions as well. Pronominal proclitics have an anaphoric function, allowing high frequency of zero representation of agent NPs. Zero anaphora is a prominent feature of the Semelai clause.
Chapter 10 ''Basic clauses'' (269-338) deals with different types of clauses: non-verbal clauses, irrealis constructions, negative, interrogative, and imperative clauses; it also includes a subsection which covers all types of adverbial modifiers. Semelai has three types of non-verbal clauses: identificational, ascriptive and locational. There are no copulas of identity or predication in Semelai but there is a copular of existence and a copular of transformation, the latter is used to describe some natural processes of development, i.e. changes of state. In questions with interrogative pronouns, different strategies are used depending on the type of the argument that the fronted ignorative refers to.
A considerable space is allocated to the irrealis mood (marker ma=), which covers a variety of modalities. It is only used with verbs which can host pronominal proclitics. The irrealis marker is used in place of pronominal proclitics so that the two markers are mutually exclusive. The irrealis in Semelai is used to express a range of modalities, which, as Kruspe maintains, fall into two categories: epistemic modalities (inferential, experiential, mirative, and hypothetical) and deontic modalities (negative imperatives, denial of permission, prohibition, and lack of obligation). The irrealis marker is only obligatory in case of deontic modality. The agent in an irrealis clause has zero representation; it can be only expressed externally as an unattached NP.
Chapter 11 'Complex clauses' (338-395) deals with relative and complement clauses, serial verb constructions, concatenated and connective clauses. In Semelai relativisation strategy depends on the head NP status: those representing core arguments are relativised with the marker of relativisation, while non-core ones are relativised without the marker. The relativisation strategy also operates in questions to O and IO, in which case the question structure is an equative sentence where the first component is an ignorative and the second component is a headless relative clause.
Complement clauses exhibit three types of syntactic compression depending on the main verb: full, reduced, and mixed. Thus, verbs of obligation and permission take reduced complement clauses, the latter undergo obligatory deletion of the argument co-referential with an argument in the main clause, their verb losing the ability to be marked independently for negation, aspect and mood. Interestingly, in the reduced-type complements with the subject co-referential to the subject of the main clause dependent verbs describing unrealized activity can be embedded under the verb ''desire''.
The subsection on serial verb constructions includes the defining criteria; accurate and full account of different semantic types with peculiarities of their internal structure; full lists of supplementary verbs; as well as a detailed description of the most specific serial constructions.
In the last two subsections concatenated and connective clauses are examined. Sentences consisting of juxtaposed independent clauses are one of the most prominent features of Semelai as well as other MK languages. They function parallel to multi-clause constructions with overtly coded semantic relationships between the clauses. Concatenated clauses in Semelai include consecutive, purposive, conditional and simultaneous semantic types. The syntax of concatenated clauses is described in terms of representation of core arguments and peripheral NPs and their positioning within the clause. Indigenous connectives in Semelai are only those of cause and condition, the majority of connectives is borrowed from Malay. The following types of connective clauses are described: causal, conditional and temporal. There are two points of interest: the clitic =hn, which is optionally used with some connectives, elsewhere, it has a range of role-making functions; causal conjunction la=, whose other functions are introducing NPs of cause/reason and encoding post-verbal subjects of transitive clauses.
Chapter 12 (396-402) gives a brief account of expressives -- a distinctive shared feature of the AA language family.
Chapter 13 (403-419) deals with miscellaneous phenomena: quotative marker, interjections and discourse clitics. In Semelai quotative marker is used to mark reported speech; utterance verbs are not used in this function. Quotation marker is also used to frame expressives, interjections, exclamations and other indexical expressions of human interaction.
Chapter 14 (420-447) contains a collection of samples of Semelai oral literature: riddles, autobiographical narratives, description of traditional activities, and traditional narratives.
_A Grammar of Semelai_ is an important breakthrough in one of the most obscure areas of Austroasiatic. This work sets a model of grammatical description for future researchers in this field. Kruspe is being meticulous in every respect: from choosing the approach to data collecting to carefully selecting theoretical background for presenting particular segments of the language structure or choosing solutions for the most tricky topics. Kruspe's work proves that sticking to method and theory is a good recipe for surviving in this troublesome but rewarding language area. Another positive point in Kruspe's approach is that she is very much concerned about choosing the most comprehensible way of presenting the material. To this end the author selects a rational and efficient structure of the book and her presentation of examples is clear and exhaustive both in terms of grammar and cultural and conversational contexts of the utterances. When describing a language like Semelai it is also essential that the author is capable of providing reliable comparative data from the relevant areas: the genetically unrelated Malay language, the main source of borrowings into Semelai, the closely related Aslian languages, as well as other AA languages. It is only too natural that a research like that, full of detail and theoretical insight, should be thought provoking, and encourage alternative opinion and attitudes.
Although the templatic morphology approach adopted in Chapter 3 makes the description impeccably formal and consistent, as well as compatible with data from non-AA languages, it is rather unusual for a MK language and leaves little chance for a comparative linguist who would rather have morphology described in terms of conventional affixes and syllabification processes they involve to make it more compatible with historical and comparative data. What is more, this theoretical background overshadows the unique phenomena, most relevant in the context of Austroasiatic: Semelai has morphophonological strategies designed to stop or slow down the destruction of the indigenous affixation. They are: an absolute veto on main syllable infixation and expanding infixes to a heavy syllable by coda copy. These strategies provide and support morphological transparency of affixed forms, stop the process of resyllabification, which eventually results in the loss of derived forms, as it was the case with many MK languages.
Verb morphology in Semelai appears to be much elaborated considering its MK affinity. The accurate, consistent and detailed description that Kruspe provides, supported by a great number of examples carefully presented and thoroughly explained in terms of contexts they are used in, enables the reader to understand all the intricacies and subtleties of form functions. It is both very informative for typological and AA comparative studies, if we are to understand how a MK language could have developed a morphology of that kind.
One of the most thought provoking verbal forms in Semelai is imperfective derived from transitive roots by coda copy or coda and onset copy (coda copy is also involved in verb to noun derivation.) Although this derivation does imply concomitant changes in aspectual meanings, i.e. is used to describe an ongoing event, a persistent event, a permanent state, a habitual action or a non started action, one can notice that imperfective forms exhibit some marginal features. Firstly, imperfective derivation is actant decreasing. Secondly, imperfective forms cannot be used with agentive pronominal proclitics. At the same time, imperfective forms exhibit some features of non-finite ones. Thus, they are used in multi-verb constructions such as serial constructions (with verbs ''go'' and ''take'') and complement clauses (as arguments of verbs ''desire'', ''want'', ''know'', ''be difficult''). Imperfective forms are also used in subjectless sentences of identity with the underlying copular of identity (38, p. 117).
Semelai causative forms exhibit a variety of specific features distinguishing them from causative formation in other MK languages. Of the four causative affixes, i.e. par-, tar-, -r-, p-, the last one is a Malay borrowing notwithstanding its AA appearance. Causative bases (except those which are derived from noun roots), can feed happenstance t(r)-, which Kruspe also considers to be a Malay borrowing though we can find traces of an involuntary t- in non-Aslian MK languages too (Katu). Semelai has two semantic types of causation. Causatives with par- describe a situation where the causer only assists or facilitates the situation in which the causee is willing to participate (''feed''). Where causative formation is inapplicable, causation is expressed syntactically with the connective la= introducing an NP or a clause providing information about the source of causation. In three-participant causative sentences both objects can be marked with the object clitic hn=, the indirect object having one of the locative prepositions. Causatives with tar-, which Kruspe labels as manipulative causatives, describe events which involve direct physical manipulation on the part of the causer (''close someone's eyes'') or ''hands on'' manipulation (''get someone to weave'', ''put food in someone's mouth''). The opposition of the two semantic types of causation is only possible with one-syllable roots and only several roots are found with both prefixes (all of them being bound ones). Causatives derived from disyllabic roots, i.e. those with p- and -r- affixes, can have cooperative auto-causative meaning. Overtly, such derivation is not valency increasing; however, such causatives imply that the action is performed with some sort of consideration and ''being-cooperative'' attitude towards someone else. The use of this type of causative in imperative sentences, which Kruspe points out too, might qualify as an indirect speech act designed to make the imperative less direct. Kruspe's remark to ex.107 (p. 133) that the auto-causative is only used when addressing a woman supports that suggestion. Interestingly, in Semelai we find semantic types of causative similar to those found in other MK languages; however, their distribution across the inventory of presumably related affixes is different. Thus, in MK languages where different semantic types of the causative are distinguished, deliberate causative is generally expressed by forms with a labial prefix, non-deliberate causative by forms with a dental prefix or by an analytical form, and in Katu we even find a form, which is semantically comparable to cooperative causative in Semelai: it is a combination of deliberate p- causative and non-deliberate analytical causative with the verb ''make''; the form is used as an apology for causing inconvenience to someone else.
Pronominal proclitics in Semelai which encode person, number and degree of respect, mark the agent in transitive clause, the agent can also be represented in a post-verbal NP with/without the agent marking proclitic la=. Actually, examples found in the texts suggest that post-verbal position of the subject may have to do with rheme shift (4, p. 432; 14, p. 430). As Kruspe points out, pronominal proclitics are also found with intransitive verbs of motion and emotion describing an atypical situation induced by some indirect external cause (230, p. 160) or an action instigated by an external force (231, p. 160). Actually, the examples suggest that a pronominal proclitic on an intransitive verb may mark a rheme shift too.
Possessive constructions in Semelai exhibit idiosyncratic features, which might be of some relevance for comparative studies. Core argument third person possessor can be marked on the possessee NP by the enclitic hn= homonymous to object proclitic. Coincidence of object and possessor marking is found elsewhere in MK (Old Khmer and Nicobarese) (see Long Seam 2000, Man 1889).
Numeral phrase in Semelai exhibits both regular MK and specific features. The former are: optional character of classifiers and alternative positioning of numeral-classifier cluster before or after the noun head. As it is often the case in descriptions, Kruspe, when making observations about the position of numeral-classifier cluster, does not distinguish between predicative and non-predicative use of numerals (p. 204). It also should be mentioned that the communicative structure can influence the positioning as well (e.g. 5, p.421). All components of Semelai numeral phrase exhibit idiosyncratic features, which would be of interest to a MK comparativist. Firstly, in contrast to other MK languages, where we only find occasional instances of classifiers derived from common nouns, in Semelai all indigenous classifiers are derived from common nouns by the nominaliser n-/-n- (with/without coda copying). Secondly, noun head forms derived by means of the same morphological process and numeral forms derived by means of coda copy can be used instead of classifiers. Derivation of numeral forms exhibits irregularities: only numerals 1-7, i.e. indigenous ones, can have such forms, derivation is only applied to main syllables, all pre-syllables being dropped, in derivation from numerals 6 and 7 the prefix t-(or ta-) is additionally involved.
Nominal derivation in Semelai makes an extensive use of coda copy both in combination with other affixes (n-/-n, p-/-p, -m-) and alone. One can hardly agree with Kruspe who tends to qualify coda copy as a process independent of and prior to affixation, introducing such notion as coda-copy base (p. 222). It looks like coda copy used alone in formation of the imperfective is a morphological device, while coda copy found in noun derivation is a morphophonological device generating allomorphs of one morpheme.
In Semelai we find two nominalising affixes: pn-/-pn- (with irregular allomorphs pN- and p-), and pC-/-pC- (''C'' stands for coda copy), which Kruspe describes as two different morphemes, the first one deriving object nouns and the second one process nouns (p. 223-24). This seems to be disputable (compare s-pn-wak from swak 'go' in ex. 93, p 223). It might well be one affix (at least by origin) with a range of allomorphs. It would be interesting to find out in what relation does it stand to labial nominalising affix found elsewhere in AA.
In Semelai we find the infix –m(C)- which is the reflex of the rare MK agent prefix/infix. In Semelai it is both found in agent and instrument nouns, however, the examples are too few to conclude, whether the latter are cases of assimilation of the affix n-/-n-. Kruspe refers to it as ''agent nominalisation'' (p. 80) and ''agent/attributive nominalisation'' (p. 224), and suggests that it corresponds to the Proto-Monic *-m-, which derived agent nouns from transitive verbs and attributive adjectives from stative verbs. It remains unclear what exactly enables the author to label it as attributive in Semelai, and, actually, Kruspe seems to have failed to spot the attributive usage of the infixed form in 96a (p. 225), interpreting it as a noun.
To compare with other MK languages, it should be noted that while deverbal nouns in Semelai can function as modifiers in noun phrase, verbs can only function as modifiers in capacity of a relative clause predicator. Another specific feature of Semelai deverbal nouns is that they can maintain aspect marking adverbials (104, p. 226).
Prepositions are an interesting as well as highly language specific domain. They also tend to be the least described. Kruspe's work is an exception to the rule. Her account is scrupulous and methodical as well as full of detail, the use of a handful of prepositions is illustrated with 130 examples. What strikes most is that in Semelai, unlike in most MK languages, there are no prepositions homonymous to verbs. Some prepositions, though, bear resemblances to verbs in other MK languages. Yet, there are specific functions of the PP in Semelai, not found in other MK languages, which suggest that prepositions do have verbal origin. Thus, PPs in Semelai, in addition to their normal functions (core arguments, non-core arguments, adjuncts, attributes) can function as the main predicator and in that capacity host aspectual markers (ga=); in non-core PPs pronominal complements can be marked with the proclitic hn=, which otherwise can only mark core arguments; several prepositions can take verbs as complements; locative PPs can host the marker of relativisation, the construction, as Kruspe points out, is used to generate avoidance lexemes (e.g. ''the one who is under the house''). On top of that, locative and deictic prepositions derive demonstrative pronominals with the prefix naʔ- ~ na-, the process is similar to the derivation of deverbal nouns, the prefix could well be allomorph of n(C)- used in deverbal derivation, since the majority of locative prepositions are monosyllables with glottal stop coda.
Of the specific features of individual prepositions one should mention the following. The locative, directional, comitative, and possessive prepositions can occur in a ''sandwich'' construction (Pi NP Pi). The preposition 'SOURCE' is used with epistemic predicates to denote the subject of epistemic judgment (e.g. ''That was so far for us''). The preposition 'WITH' besides instrumental and comitative meanings is also used to encode ''actual path of the event or locations encountered on the path'' (p. 244), to compare, the same is true for the instrumental/comitative preposition in Khmer.
The scope of functions of the core role-coding marker hn=/=hn might seem somewhat puzzling (3.3.4, 9.3.1 and 11.5.2). The proclitic is used optionally to mark direct objects in post verbal position. At the same time, O marking is used in single argument constructions: middle voice and irrealis. It is used to mark pronominal indirect objects in addition to their role marking prepositions. It is also attested in pronominal non-core prepositional phrases. The enclitic is used anaphorically on the verb when O has zero representation as well as on the possessee NP to mark third person possessor. As Kruspe assumes, it is also found as a fused marker of third person pronominal subject in intransitive clause. Finally, the enclitic is optionally used in temporal and conditional conjunctions. In the last case Kruspe suggests that a possible function of the enclitic could be marking the clause as backgrounded information (p. 382). All this would indeed be puzzling if it were not for the fact that functions of this Semelai clitic largely overlap with the functions of the Old Khmer ta-marker introducing dependent predications, appositive NPs, possessors, and non-core NPs, (see Sak-Humphry, 1993; Pogibenko, 2005); to a less degree, its functions overlap with ta in Nancowry (Man, 1889); reflexes of this function word (supposedly, of pronominal origin) are found elsewhere in MK and, possibly, in Munda. In addition, the grammatical status of the clitic in Semelai is somewhat unclear; it is optional in all its usages and, as Kruspe points out, in case of direct object marking, direct elicitation never yielded examples with the proclitic, whereas spontaneous clauses when rechecked proved to be correct (p. 262). As a matter of fact, Man recognized only two functions of ta in Nancowry: in passive constructions and with modifiers, all other occurrences he considered as purely euphonic (Man, 1889: xlvii.)
Kruspe's account of the irrealis provides a good deal of detail (pp. 278- 296). Her description of the many facets of this modality in Semelai is accurate and precise. Apparently, this work is going to be much quoted by linguists studying mood and modality. It is evident that Kruspe's insight into the intricacies of this highly elusive domain was acquired through extensive analysis and discussion with informants. However, there remain some ambiguous points which deserve further consideration. A strong (and successful too) advocate of systematic rather than pragmatic approach to description, Kruspe is trying to place all usages of the irrealis within the scope of deontic and epistemic modalities. Yet, judging by the examples, the use of the irrealis in Semelai goes beyond that scope. Firstly, it is very likely that it is used for pure future reference (58. p. 281), the more so since the only other means to refer to future in Semelai is the imminent aspect. Secondly, it looks like the irrealis is used to express ability (72, 73, p. 284; 77, 78, p.285, 16, p. 423), as well as possibility (238, p. 161, 59, p. 281). Finally, the usage which Kruspe labels as ''retrospective experiential'' does not seem to belong to epistemic usage, at least the examples given do not suggest that a degree of commitment by the speaker to the truth of what is being said is involved. On typological considerations, there are no reasons why the only oblique mood in Semelai should not be asymmetrical, multifunctional, with vague boundaries between different usages.
Of the different strategies used in questions with interrogative pronouns two are of particular interest. One is relativisation of the verb group in questions to core arguments O or IO, the result of applying this strategy being an equative sentence. Interestingly, this structural type is occasionally found in Khmer. The other strategy, in Kruspe's wording, involves nominalisation of verbal predicate, resulting in a nominalised clause; it is used in questions about direction and, as Kruspe puts it, about possession. However, the latter usage actually might be the result of reanalysis of questions about the agent, induced by the form of the verb (compare ex. 302, p. 330). Bearing in mind that the nominalised verb form in Semelai has other marginal functions, such as modifier in nominal compounds (101, p.226), main verb in subordinate clauses (108, p. 226), its status requires further analysis.
The relativisation marker in Semelai has specific features that are common with relativisation markers in Old Khmer (ta/ti, tel) and Mod. Khmer (dael): in Old Khmer the markers ta/ti are also role-oriented; in Mod. Khmer relativisation strategy is also found in questions; both in Old and Mod. Khmer relativisation strategy is involved in derivation of sentences with rheme shift (see Pogibenko, 2005). It seems that the latter usage of the relativisation marker in Semelai has escaped Kruspe's attention (see 23, p. 343). What is more, while in Khmer in sentences with rheme shift the markers of relativisation can appear in combination with the copular of identity (which is often perceived as a sort of emphatic particle), in Semelai it appears in combination with a discourse clitic =pa (which is rather vaguely dealt with in the last subsection of Chapter 13). Lastly, as Kruspe mentions, in Semelai a headless relative clause is often associated with a contrastive function; a similar function of Old Khmer ta is preserved in Mod. Khmer.
Serial verb constructions (SVC) in Semelai are mostly of the types found in other MK languages, where the supplementary verb expresses manner, various aspectual characteristics, purpose, result, or is quasi-synonymous with the main verb. The relative ordering of the components for the most part follows the MK patterns. Although Kruspe correctly observes that in Semelai participant-oriented supplementary verbs (see Himmelmann and Schultze-Berndt, 2005) precede main verbs, she seems to miss that so do aspectual supplementary verbs (except those expressing result). Of special interest is the fact that besides resultive SVC proper, Semelai has a construction which expresses semi-result or suspended result (with pay 'put aside', p. 364).
The fact that the majority of Semelai conjunctions were borrowed from Malay on no account can be puzzling (see, however, p. 381). The need for explicit codification of the relationship holding between conjoined clauses is more urgent in a written language rather than a spoken one. Vernacular forms of speech can surprisingly do without any conjunctions, the context providing the necessary unambiguity (Spoken Russian is a good example in that respect). In a vernacular language nothing ever triggers the indigenous mechanisms of grammaticalisation coining the means of that sort of codification, but they are readily borrowed when available. That is why Kruspe's suggestion that Semelai knaʔ 'when' is borrowed directly from the Malay verb kena 'to affect' can hardly be justified. Actually, it could as well be a borrowing from the Khmer kaal naa 'when'.
To conclude, the main point of the discussion is that this comprehensive, scrupulous and detailed description of Semelai, accomplished by a talented and dedicated scholar, based on the data collected in a demanding and strenuous fieldwork environment, is an outstanding contribution to the field of Austroasiatic studies as well as general typology. It fills a considerable gap in our knowledge and sets an example for future research. For the years to come it will be a valuable source of information about an idiom that is unique, though, as many minority languages on our planet, regretfully, threatened with extinction. Hopefully, Kruspe's description of Semelai will encourage further research.
Himmelmann, Nikolaus P. and Schultze-Berndt, Eva (2005). Issues in the syntax and semantics of participant-oriented adjuncts: an introduction. In Himmelmann, N. P., Schultze-Berndt, E. (eds.). Secondary predication and adverbal modification: the typology of depictives. Oxford University Press, 1-67.
Long Seam (2000). Dictionnaire du Khmer Ancien (D'après les inscriptions du Cambodge du YIe. – YIIIe. siècles). Phnom Penh.
Man, E. H. (1889). Dictionary of the Central Nicobarese Language (English-Nicobarese and Nicobarese-English). London.
Pogibenko, T. (2005). Drevnekhmersky i palaungicheskije yazyki: zavisimyje predicativnyje konstructsii [Old Khmer and Palaungic languages: dependent predicative constructions]. In Alpatov, V., Shalyapina, Z. (eds.). Actualnyje voprosy yaponskogo i obtschego yazykoznanija. [Issues of the Japanese and General Linguistics.] Moscow: Vostochnaja Literatura RAN, 359-388.
Sak-Humphry, Chany (1993). The syntax of nouns and noun phrases in dated pre-Angkorian inscriptions. Mon-Khmer Studies 22, 1-126.
ABOUT THE REVIEWER:
Tamara Pogibenko is a research fellow in the Institute of Oriental Studies, Moscow. She is a graduate of the Department of theoretical and applied linguistics, Moscow State University. Her current research is in Khmer (grammar, semantics, pragmatics), Old Khmer, and Austroasiatic comparative studies. Since 1978 she has taken part in several joint linguistic expeditions in Vietnam, collecting data on a number of minority languages.