Acta Linguistica Lithuanica

LVI (2007), 117 – 144

Reviews

Reviews

Vytautas Ambrazas

The Historical Syntax of the Lithuanian Language

Vilnius: Lietuvių kalbos institutas, 2006, 610 pp., ISBN 978-9955-704-18-8

Vytautas Ambrazas’s The Historical Syntax of the Lithuanian Language is a synthesis of the author’s many years of research, the results of which have been published in numerous separate publications. It is unsurprising that particular attention is devoted to those matters the author had previously investigated in greater detail. These are questions of major importance in Lithuanian historical syntax, such as word order and the development of participial constructions, infinitival constructions, and the passive. The book also fills some gaps, that is, it discusses matters to which the author had not previously devoted separate monographs or articles. Of course, completeness is an ideal that can only be approached, and even in major synthetic works we can always discern traces of an author’s individual investigations and research preferences. With these introductory remarks, I wish to make clear that I do not intend to reproach the author for having been able to discuss this or that aspect of the book more fully, or for having left this or that matter unmentioned. When the time comes for a new synthesis, another author will write another work, likewise reflecting that author’s individual preferences.

After an introductory chapter devoted to the principles of research in historical syntax, the book has a chapter on word order and its changes. This is followed by three chapters discussing, in turn, three sentence types, whose distinction, in the author’s view, reflects the most important criterion for classifying sentences from a historical perspective. First, two basic types are distinguished—nominal and verbal sentences—while the third type consists of what are called “mixed sentences,” that is, sentences with nominal (primary and secondary) predicates. The mixed type would include, for example, the sentence The son returned safe and sound, which combines features of the separate sentences The son returned and The son is safe and sound. Whether these sentences “arose in the distant antiquity of the Indo-European languages as a result of the interaction of verbal and nominal sentences” (p. 491) is something no one really knows, and it would probably be impossible to find out. But, of course, this point does not need to be proved in order to serve as a criterion for organizing the material. The following chapters discuss, in turn, the case marking of nouns, prepositional constructions, infinitival and participial constructions, and complex sentences with subordinate clauses introduced by conjunctions—

117

Recenzijos

—with conjunctional subordinate clauses. A separate chapter presents conclusions about the historical development of Lithuanian syntax, a German summary of the book, the bibliography, and indexes.

The organization of material is always a matter of convention, often dictated by practical considerations or simply by tradition, and should not be an object of principled criticism. But the structure of the contents often also reflects the conceptual framework used in the work, which should indeed be evaluated critically. The organization of the material in the book under review has many unusual features that make it difficult to read for an unprepared reader—especially someone who has not attended a Lithuanian school or studied Lithuanian school grammar (in much the same way, reading academic Grammars of the Lithuanian Language or Grammars of Contemporary Lithuanian can also be challenging). What should such a reader do when encountering, say, the notion of “double cases”? The notion itself seems fairly straightforward: two nominatives, two accusatives, two datives, and so on, are used in a sentence, for example, varnų ben porelę sav pietums nusišauju (Donelaitis; p. 267). So are sentences such as Ąžuolas kaitrus medis (p. 133) presumably instances of a double nominative? Nothing of the kind: this sentence contains two nominatives, but for some reason it does not meet the conditions for inclusion in the category of double cases. In fact, the chapter on double cases (apart from constructions of a wholly different kind mentioned in the introduction, such as prašau tavęs menkniekio) discusses mostly constructions with secondary predicates (secondary predicates), for example, Asch tawa Dukterį neradau Mergą BrB V Moz 22.17. Having discovered a familiar category beneath the mysterious label “double case forms,” the reader feels relieved and goes on to expect more detailed information about the marking of secondary predicates. We know (in Slavic studies, a vast literature is devoted to this phenomenon and the factors that condition it) that secondary predicates can be marked in various ways: by agreement, the instrumental case, or a prepositional construction (cf. The court found him guilty / as guilty; At the farmer’s, he worked as a farmhand / as a (through) farmhand). How did the relationship among these three ways of marking develop in the history of Lithuanian? We find no direct answer to this question in the book, since, for instance, the predicative instrumental is discussed in an entirely different chapter, where it is hidden under the no less enigmatic title “sentences with the instrumental” (pp. 158–167). On opening this chapter, the reader may find it difficult to understand why Bread turns into flesh is a “sentence with the instrumental,” whereas Jonas rides a bicycle is not. This curious point becomes clear only after consulting a school grammar of Lithuanian, in which the predicate complement of a copula (including a predicate used with so-called “impure copulas” such as become, turn into, and the like) is not a separate sentence constituent, but only part of a compound predicate, linked to the subject by agreement (more precisely, a “mutual relation”). It is possible to speak of agreement in the construction mes esme Liudininkais io (BrB Apd 5.32)

118

Recenzijos

difficult, so this predicative instrumental ends up in no man’s land and, to avoid using a term that implies a strict taboo, it is safer to hide behind the name of a case or another morphological category. Thus we get “sentences with the instrumental” or “nominal sentences with a neuter form.” How, then, can one speak about something that one is not allowed to speak about?

The question of marking predicatives by agreement or by the instrumental is discussed very briefly (pp. 165–167), but the author seems to decline the possibility of saying anything new about it on the basis of the material he has collected, since even in Jablonskis’s grammar a predicative agreeing in case falls into the category of “double cases,” while there would not even be a way to raise the question there of the relationship between an argument of the sentence verb and the instrumental. Thus, in traditional Lithuanian grammar, the two constructions lead separate lives and, within its framework, cannot in principle meet. Yet it is interesting that, for example, depictive secondary predicates (depictive secondary predicates), which in present-day Lithuanian consistently agree with the subject or object, are sometimes marked by the instrumental in Old Lithuanian, e.g., kad nevaykščcotu nogu ChNT Apr 16.15 (now: kad nevaikščiotų nuogas); gime aktu ChNT Jn 9.20 (now: gimė aklas). Perhaps this is merely the influence of Slavic languages¹, or perhaps the principles governing case use were simply different from those of today? For Chylinskis also uses agreement in another construction that should probably be recognized as complementative (complementative: neradau nes darba tawo pitno podiewu ChNT Apr 3.2. In the absence of specialized studies, the author may not have been able to answer this question immediately, but at the very least, noting the fact itself would have been useful as a pointer to what other researchers might investigate in the future. But to open up new research perspectives, one would need to abandon the categories and concepts of school grammar and adopt some of the conceptual resources of modern theoretical linguistics. There is no doubt that the author has read foreign-language scholarship, as a look at the book’s bibliography makes clear: it cites, for example, Johanna Nichols’s book on predicative marking (Nichols 1981), but the author keeps the insights gained from it to himself, offering readers instead the old, familiar wisdom of Jablonskis’s grammar. This is a pity, since such an approach effectively blocks the way to new research.

The concept of “double cases” is flawed also because it lumps together fundamentally different syntactic phenomena, leaving no possibility of distinguishing constructions with a (depictive or complementative) secondary predicate (O Ioab palikka giwus atlikusius Mieste BrB I Kron 12.8; ką wadini mane gera BrB Mt 19.17) from constructions such as these

¹There could have been no direct influence from Polish here, since the corresponding passages in the Gdańsk Bible use adverbs (nago, ślepo).

119

Recenzijos

constructions such as neger yra žmogui vienam būti VE 50₁₄, in which the predicative of the infinitival clause agrees, at best, only with a coreferential complement of the main clause. For such a sentence could syntactically be interpreted either as neger(a) yra [žmogui būti vienam], or as neger(a) yra žmogui. [PRO₁, būti vienam], depending on whether we treat the dative as the subject of the infinitival clause or as a complement of the main clause. The meaning of the sentence requires choosing the first interpretation; compare non est bonum esse hominem solum, οὐ καλὸν εἶναι τὸν ἄνθρωπον μόνον. Therefore, to pretend that sennam girtam jūkis B 73 and neger yra žmogui vienam būti have the same syntactic structure is possible in school grammar, but not in scholarly grammar. This is, moreover, not merely a matter of interpretation, since an erroneous theoretical position also leads to factual errors that can be difficult to correct later. Suppose we find “double-accusative” constructions in old writings in which a secondary predicate agrees with the object: Ghys... bagotiusius tusczus apleida VEE 159; yet throughout the entire corpus of Old Lithuanian writing, we find no constructions in which the predicative of an infinitival clause agrees with the accusative object of the main clause—in all such cases there is an instrumental, e.g., sakens jemus nessibijoti / liepe ios linksmais būti BP I 83.15. Why, then, is the “double accusative” not used here (*liepe ios linksmus būti)? It is true that standard Lithuanian now requires the accusative instead of the instrumental (privertė juos atsargius būti), but the construction prescribed for current usage is entirely artificial, created (most likely) by Jablonskis and based on the erroneous assumption that the construction here is the same as, for example, padarė juos atsargius, namely, a “double accusative”². The “correct” construction privertė juos atsargius būti (that is, unsupported by historical tradition but presumably consistent with the “spirit of the language” as recognized by speakers) was introduced together with its “correct” interpretation as a “double accusative.” It would therefore probably be tactless to ask why Bretkūnas does not use the construction *liepe ios links-mus būti, even though he uses agreement in the construction Diewas prisake mums mielaschirdingiems būti (BP II 253,4). Ambrazas does not raise this question either, contenting himself with citing examples of “double datives,” such as neger yra žmogui vienam būti. There would simply have been no place to cite and discuss the construction liepe ios linksmais būti, because it is obvious to everyone that it lacks the double accusative required by present-day grammar; nor can it be acknowledged as a predicative instrumental with the infinitive būti, since that would mean acknowledging that this construction in fact differs from

²Leonardas Dambriūnas challenged this artificially created construction, revising the sentence Pavojų, verčiantį mus būti akylesnius, ir čia netrūksta to... būti akylesni aus (Ur-nėžiutė 2006: 7). From a historical perspective, he was right, since the “double accusative” in this construction is based on a misunderstanding.

120

Recenzijos

The “dual accusative” constructions made them cautious, and then one would have to admit that the interpretation of the construction Dievas prisakė mums mielaširdingiems būti as a dual case is not adequate either. What, then, is to be done? Best to say nothing. The article by the author of these lines (Holvoet 2004), which draws attention to the troubling fact that the old Lithuanian construction *liepe jos linksmus būti is absent and proposes an interpretation, is rightly not cited by Ambrazas. Indeed, a discussion with the author of that article would not have been possible simply because the problem he raised cannot even be formulated within the framework of Jablonskis’s grammar, so raising it at all was presumably unwelcome.

A few words must also be said about the principles of syntactic description applied in the book, since they determine many of the formulations characterizing both the development of individual constructions and the general trends in the development of Lithuanian syntax.

This historical grammar abandons the three-way classification of syntactic relations adopted in the Lithuanian grammatical tradition—government, agreement, and adjunction—and instead proposes a simpler two-way distinction between government and adjunction (p. 20). By definition, this would correspond to the distinction between government and modification adopted in Lietuvių kalbos gramatikos darbuose (e.g., Holvoet & Judžentis 2005). As there, the author considers agreement to be a formal expression of syntactic dependency (p. 20). It is odd, however, that only agreement is considered a form of such dependency: in linguistic literature, agreement as a type of morphosyntactic dependency is usually mentioned alongside morphosyntactic government (English government, French rection, German Rektion). Ambrazas’s book, however, does not traditionally distinguish morphosyntactic government (rection) from syntactic government (complementation). This, incidentally, is not the only feature transferred unchanged from the old system of syntactic relations. The relation of “mutual connection” between subject and predicate is retained (p. 21). This concept is inherited from the DLKG conception, according to which the verb is, in a certain sense, syntactically dependent on the subject because it agrees with it. Since the subject, in turn, depends on the verb (after all, the latter opens a position for a nominative subject), we get mutual dependency. But once agreement is rejected as a third type of dependency, there is no longer any “mutual connection” either, and the verb’s agreement with the subject is merely a classic example of head marking. Since Johann Nichols’s 1986 article is cited in the bibliography, the concept of “head marking” should be familiar to the author, although he ignores it. The old inventory of concepts also retains “double dependency,” which is supposedly what links predicative modifiers simultaneously to both the predicate and the subject or object. Traditionally, this concept is justified by the predicative’s agreement with the subject or object, e.g., paži-

121

Recenzijos

nauji jauną: since agreement is traditionally considered a type of syntactic relation, in this construction the predicative adjective jauną must depend on the object pronoun jį (because of agreement), but at the same time it must also depend on the predicate verb (otherwise the difference between the constructions mačiau brolį linksmą and mačiau linksmą brolį could not be described syntactically). But if agreement is removed from the list of types of syntactic relations, there is no basis for speaking of “double dependency” either. Dependency grammar does not recognize double syntactic dependency at all, and it is not entirely clear how the author himself understands it; for instance, on p. 491 we read: “The second nominal in sentences of this type [i.e., sentences with predicative modifiers—A. H.] has double syntactic relations: it is adjoined to the first [nominal] as a kind of apposition and at the same time is semantically related to the verb.” But if, as the author claims, the nominal is connected to the verb by a semantic relation, then he presumably no longer postulates a syntactic relation between them, so it is unclear why the syntactic relation should be double. Let us suppose, however, that the author is simply imprecise in his wording, and that, alongside semantic dependency, syntactic dependency is in fact also postulated. But then we have two different criteria for establishing syntactic dependency: (i) in one case (jį jauną), syntactic dependency is inferred from morphosyntactic dependency between words (agreement), whereas (ii) in the other case (mačiau jauną), the inference is made despite the absence of agreement and of morphosyntactic dependency between the words altogether—the aforementioned semantic relation between them apparently being considered sufficient grounds for establishing a syntactic relation. It should be clear that this inconsistency in the procedures for establishing dependency is theoretically undesirable and in practice leads to irresolvable contradictions. For it is hard to deny that, for example, in the sentence nudažė namą žaliai, the words namą and žaliai are semantically related, since the house is made green, and many languages also emphasize this relation through agreement (Latvian nokrāsoja māju zaļu). If agreement is a sufficient, though not necessary, condition for establishing a syntactic relation, then on the basis of the semantic relation one can freely claim that the construction nudažė namą žaliai also has double dependency. But then the whole system, which posits a syntactic distinction between a construction with an adverbial (laiko butelį stačiai) and a construction with a predicative modifier (laiko butelį stačią) collapses.

This rather lengthy digression was necessary in order to understand and assess one of the main theses of the work under review—the claim about “strengthening of government.” The author draws on a quotation from the introduction to Antoine Meillet’s comparative grammar of Indo-European (Meillet 1934):

„Il n’y avait pas en indo-européen de „rection“ d’un mot par un autre,

122

Recenzijos

comme il y en a en latin par exemple; l’autonomie du mot est le principe qui commande la structure de la phrase indo-européenne“ (p. 196, išnaša).

It is not entirely clear whether, in writing this, Meillet had syntactic or morphosyntactic dependency in mind, although the term “rection” traditionally denotes case (i.e., morphosyntactic) government. Judging from some formulations, the author of the book under review has precisely morphosyntactic dependency in mind: compare, for instance, the formulation “When they occur with inflected words, prepositions govern them, that is, determine their grammatical form” (p. 283). An indeclinable word apparently cannot be governed, so in constructions such as iš čia, per kur and so on, we presumably have adjunction. But what, then, are we to make of the definitions given on p. 20? There, after all, the issue is whether there is a syntactic requirement or not, whereas on p. 283 it turns out that this is merely a matter of morphology: if the dependent word is inflected, we have government; if not, adjunction. Let us suppose, however, that the definitions lead a life of their own somewhere in a separate corner and have no bearing on the author’s reasoning. Even with this qualification, it is difficult to understand fully what the author means when he claims that “in the development of the Lithuanian language, an increase in the SYNTACTIC dependence of cases on verbs can be observed” (p. 197; emphasis mine—A. H.). After all, case dependence is not equivalent to word dependence, so the author is in fact talking about rection, not complementation.

In places, reading the text gives the impression that the author himself has not fully decided what exactly is strengthening in the development of Lithuanian. For in specific cases it is complementation that is emphasized: “In general, constructions with temporal and locative accusatives show the same developmental direction: from an adjoined adverbial to a direct object governed by the verb” (p. 206). This claim is supported by examples such as pra- vaikščiojau naktį, ištarnavau metus and so on. However, if we set aside the special derivational types in which duration is incorporated into the verb’s argument structure (these derivational types may be new, but their emergence is nevertheless a matter of word formation and only indirectly of syntax), it is not clear what has changed. In the construction miegojo visą naktį, does the word naktį presumably remain “adjoined,” just as before? In any case, becoming a direct object is undoubtedly a syntactic process, and postulating it would indicate that the emergence of a complementation relation is at issue. Elsewhere, however, what is clearly meant is once again case selection: “As the contrast between partitive genitive and accusative weakens, there is a growing tendency to generalize either one or the other of these cases depending on the meaning of the verb. In this way, semantic case agreement with the verb is replaced by increasingly strong syntactic government” (p. 224). Thus it is no longer the word that is governed, but the case. If the partitive genitive weakens

123

Recenzijos

and the opposition with the accusative (atsiminti brolio: atsiminti brolį) is generalized to the accusative (atsiminti brolį), then only the marking of the complement changes; it is not possible, however, to speak of the emergence of a complementation relation. So what actually changes? Perhaps “government” merely means the consistent use of a particular case to mark the complement of a particular lexical item? Or is it that case selection is no longer determined semantically? It would be good to know.

Amicus Plato, sed... No one doubts Antoine Meillet’s enormous significance for the development of historical linguistics, but his syntactic claims should no longer be cited at the beginning of the twenty-first century. However one understands his claim quoted above, whether in connection with syntactic or morphosyntactic dependency, it is probably wrong in either case.

First, if we understand “government” as syntactic dependency—complementation—then, following Meillet, we attribute to the reconstructed proto-language structural principles different from those of living languages. Supposedly, the words of the proto-language were syntactically independent and were not linked by relations of dependency such as those found in present-day languages. Such a claim can of course be made, but then we return to the era of the early comparativists (Bopp and Schleicher), when the view was still widespread that languages could have different structural features and laws of change at different stages of development (the “childhood” and “old age” of languages). Since the time of the Neogrammarians, the “uniformitarian principle” has been followed implicitly, and since the second half of the twentieth century, explicitly (uniformitarian principle, cf. Harris & Campbell 1994: 24–32), according to which linguistic explanations must be uniform, regardless of whether one is studying a present-day language or a hypothetical ancient proto-language. Nothing that is impossible now should be postulated when investigating the development of languages in the past. The idea of primitive, grammatically unformed languages may never have been strictly and definitively rejected, but at least at present it is not seriously considered by linguists. A stage in the development of a proto-language at which the concepts of complementation and modification could not be applied to it is merely a matter of semi-scientific speculation for us.

Thus, only the second, morphosyntactic, conception of “government” remains. But here too Meillet’s claim seems highly questionable. At various stages in the development of particular languages, we find cases in which semantic oppositions encoded by case are eliminated and structural cases (e.g., the accusative of the object) become established, but we also find cases of the opposite tendency. For example, partitive marking of the object (from which the genitive of negation also developed) became considerably stronger in the Baltic and Slavic languages, most likely encouraged by the syncretism of the ablative and genitive (ablatives are a well-attested source of partitive constructions): this is clearly evident when Baltic and Slavic languages are compared with texts in Ancient Greek and Lo-

124

Recenzijos

tin. In the predicate position alongside the nominative (accusative), the Slavic and Baltic languages introduced the instrumental, which was initially associated with marking additional semantic content (today this predicative instrumental function has almost completely faded in both language families). In short, in the Baltic and Slavic languages, some semantic oppositions encoded by case disappeared, while others arose. Of course, we also see a reduction in semantic oppositions encoded by case in languages where cases have disappeared, such as the modern Romance languages, but here it is difficult to discern any further spread of the idea of morphosyntactic “government.”

Returning specifically to the thesis of the book in question, its evaluation is made more difficult by the fact that the author does not define “case government” more precisely—as already noted, his conception of case government is intertwined with a purely syntactic phenomenon, complementation. The examples provided are contradictory and show that the author does not clearly distinguish syntactic government (complementation) from morphosyntactic government (government in the case-marking sense). If the book’s thesis mentioned above is to be understood as claiming that, in the development of Lithuanian, the number of semantic oppositions encoded by case in the expression of subjects, objects, and the like decreased, then such a claim is theoretically meaningful, but probably factually incorrect. For example, the opposition between the accusative and the partitive genitive is still alive in Lithuanian. In what sense would it have weakened and given way to “government”? Of course, isolated cases can be found in which the alternation between the genitive and the accusative was eliminated. According to the author, the genitive used with the verbs atmintį, užmiršti is of partitive origin (atmink manęs BrB Moz. I 40,14), and has been displaced from modern usage by the accusative. But competition between the genitive and accusative is already observable with Latin memini and Greek μιμνήσκω. This phenomenon is so old that, most likely, even in the language of the old Lithuanian writings this genitive was already a semantically unmotivated syntactic relic (its partitive origin can only be conjectured), and from the perspective of Lithuanian its elimination was merely the replacement of one semantically unmotivated type of “government” with another (the principle of economy probably came into play here: one type of government remained instead of two). In this respect, genitives with an ablative meaning differ somewhat. Of verbs of the type bijoti, vengti, the author writes: “The genitive, which formerly had a separative meaning, is gradually drawn into the verb’s field of semantic and syntactic attraction, loses its independent meaning, and becomes the case of its governed complement” (p. 245). It is hard to believe that the genitive vilko in the construction bijoti vilko was not a complement before “government” emerged; such a view is merely part of the same myth of a primitive proto-language devoid of syntactic relations. Most likely, this genitive in the construction bijoti vilko has always marked a complement. The only question is whether it (that is, this particular case) is governed now and/or in the old writings. In this

125

Recenzijos

case the case alternation was not eliminated, as it was with the predicate-type atminti: now, as before, the genitive is used. Thus, in this case, according to the author, the shift occurred when the ablative meaning of the genitive faded: the genitive came to be used not so much because its ablative meaning matched the type of situation being conveyed, but because the verb itself began to require it. It would be difficult, however, to prove such a thing, since the cases of complements or prepositions often redundantly duplicate what is already inherent in the meaning of the verb itself. Suppose that the ablative meaning is undoubtedly an integral part of the lexical meaning of the verbs bijoti, vengti, but is redundantly expressed in the corresponding Latvian constructions baidīties no vilka ‘fear the wolf’, vairīties no vilka ‘avoid the wolf’ (these constructions replaced the older constructions baidīties, vairīties vilka). One could probably say that the prepositional phrase no vilka is “governed” by the corresponding verbs, just as the genitive is with the Lithuanian verbs bijoti, vengti (since Latvians have no other construction). But the fact cannot be denied that the preposition no has an ablative meaning. Thus, for “government” to become established, it is by no means necessary for the meaning of the case to “fade.” Finally, it would also be difficult to prove that the genitive in Lithuanian has lost its ablative meaning. After all, the genitive with an ablative meaning is still in active use with prepositions (iš/nuo miesto). Similarly, the instrumental is used in colloquial speech with the preposition su (rašyti su pieštuku, važiuok su autobusu), but we would not claim that the instrumental in a prepositional construction has lost its “instrumental,” i.e. instrumental, meaning. Incidentally, this “instrumental” meaning can also be discerned in cases where, according to Ambrazas, this case has already become “governed” (e.g., naudotis peiliu), and if someone wanted to argue that the genitive in the construction bijoti, vengti vilko still has an ablative meaning, it would be difficult to refute that.

I do not intend to discuss more specific examples here. The claim that “government strengthens” in the development of Lithuanian is problematic not only, or even primarily, because it is wrong, but because it is probably not entirely clear even to the author himself what it actually means. Neither the definition of “government” given at the beginning of the book nor the explanations found elsewhere in it help the reader fully understand what, in the author’s terms, is strengthening in the development of Lithuanian.

Vytautas Ambrazas is the author of many studies important to the history of the Lithuanian language. Their findings are reflected, synthesized, and presented in a broader context in the book under discussion. One may mention here the author’s research on Lithuanian infinitival constructions (Ambrazas 1995). These studies convincingly reveal the mechanisms of syntactic reanalysis that led to the emergence of a nominative object in constructions such as šienas reikia pjauti (→ reikia šienas pjauti); the nominative object is fir-

126

Recenzijos

the result of the reanalysis of the former subject’s nominative, while the influence of Finnish languages emphasized by some scholars probably only contributed to the preservation of this nominative and was by no means a necessary condition for its emergence. Ambrazas also convincingly explains the use of the dative to mark the subject and object of purpose and intended-use infinitives in adverbial and adnominal positions (a place for the hen to roost, an axe for chopping wood) on the assumption that in this case reanalysis took place in a construction with a double dative of purpose, and the first dative was reinterpreted as the subject or object of the infinitive (the author rightly rejects the case-attraction hypothesis). These studies by Ambrazas are a lasting contribution to the historical syntax of Lithuanian and the Baltic languages.

Some of the author’s other explanations of the historical development of syntactic phenomena, presented here and also found in his earlier works, have partly lost their relevance in light of new typological research. One might mention, for example, the hypothesis concerning the emergence of the indirect mood.

As in his earlier publications, in the book under review Ambrazas derives indirect-mood constructions from constructions in which a participle is used with verbs of perception, experience, or saying, for example, he felt / thought / said that he knew the way. “The meaning of a reported or inferred action could have been transferred here to the participle, which then began to be used independently” (p. 393). This explanation, continuing the tradition of Delbrück’s and Endzelīns’s accounts, is unsatisfactory, since it covers only the reported meaning of the constructions in question. It is understandable why this meaning was most important to Endzelīns: Latvian evidential constructions do, after all, have only a reportative function. Lithuanian evidential constructions, however, also serve an inferential function (a conclusion based on traces of an event, e.g., there has been a thief here) and a mirative function (expressing surprise, e.g., he turns around—and there stands the devil). In typological literature, the parallels cited from other languages convincingly explain how an inferential meaning first arises from the perfect meaning, followed later by other evidential meanings, including the reportative meaning. The reverse development—from reportative to inferential meaning—is unattested and difficult to explain. The explanation proposed by Litvinov in 1989, which regards the perfect as the source of Baltic evidential constructions, therefore seems much more attractive. According to this author, alongside evidential constructions of perfect origin (they have left), constructions with present participles (they live far away) and even with future participles (they are about to leave) emerged later. Ambrazas briefly discusses Litvinov’s explanation but rejects it on the grounds that the precedence of past-tense forms has not been proven. Yet it would be difficult to prove this, given that the written history of the Baltic languages is too short. On the basis of direct evidence from the sources

127

Recenzijos

it is impossible to “prove” either Ambrazas’s or Litvinov’s hypothesis, but the latter is superior in that it does not contradict the facts of language typology.

The author frequently mentions language typology and cites a considerable amount of typological literature. In some cases, it supplements the author’s discussion; in others, it serves merely a decorative role. Evidentiality—or the previously mentioned marking of the head of a syntactic phrase—is a fairly striking example of this. Johanna Nichols’s article on head and dependent marking certainly adds distinction to any bibliography. But that does not mean that it is useful to acquaint Lithuanian readers with the ideas it presents, since, as has been observed, these pernicious ideas bring down the entire edifice of “Lithuanian” syntax.

Incidentally, any typological claim may be contradicted by an example from some language that has not previously been studied with respect to the matter in question—even Lithuanian—but that language must be thoroughly investigated. Thus, for example, on p. 186 the author claims that the data he presents refute Shibatani’s (1998) assertion that the implicit agent of the impersonal passive, in Lithuanian as in other languages, can refer only to a human being. In fact, Frajzyngier (1982) proposed this as a universal feature of the impersonal passive. I argued in an article from 2004 (2004b), not mentioned by the author, and Holvoet (2007) discussed at greater length, that Lithuanian evidential constructions with passive participles are only an apparent exception to Frajzyngier’s generalization. In fact, Lithuanian evidential constructions with passive participles violate several well-attested typological generalizations and, moreover, some of them are difficult to explain using the usual rules for deriving passive constructions (see Holvoet 2007, 96–104, for further discussion); it is therefore worth considering whether they should be regarded as passive constructions at all.

In connection with constructions containing passive participles, I would also like to draw attention to another instance in which the work under review fails to take account of the distinctiveness of evidential constructions. Discussing the origin of the agent genitive, the author rejects my explanation (Holvoet 2001), according to which genetivus auctoris was transferred from constructions with transitive verbs to constructions with intransitive verbs. According to this hypothesis, the agent genitive, being possessive in origin (as shown by the use of possessive genitive forms my, your, one’s own in passive constructions), must have arisen in an adnominal position ([the house [built by the father]]); following reanalysis, the genitive was understood as agentive ([[built by the father] house]), and together with the participle it could be shifted to predicate position (the house is [built by the father]); when the participle was extracted from the participial phrase [built by the father], a full passive with an agent genitive, the house was built by the father, emerged in Lithuanian (but not Latvian). From constructions of this kind, the genitive was transferred to constructions with intransitive verbs

128

Recenzijos

(the father has lived here). In rejecting this explanation, the author stresses that “[constructions with neuter forms] are used in the area of Lithuanian dialects where periphrastic passive constructions are generally uncommon” (p. 189). Of course, it is not the presence or absence of agreement that is crucial here, because, as in present-day Lithuanian dialects, constructions with participle agreement in adnominal position (the house built by the father) could also have coexisted earlier with corresponding non-agreeing predicates (the house (is) built). Thus, what the author probably has in mind is that constructions of the type the father has lived here are used in dialects in which the standard passive construction the house was father-built is generally uncommon. But this essentially only shows that constructions of the type the father has lived here, as evidentials, became functionally detached from passive constructions and began to exist independently, while indeed inheriting many of their features. One such feature inherited from the passive construction was, undoubtedly, the agent genitive. In rejecting my explanation, the author points to the nominal use of neuter participles: the primary meaning of the sentence the father’s work is supposedly ‘the father’s work’, and that of the sentence my birthplace there is ‘my birthplace there’ (p. 191). However, although it is easy to imagine that nominal constructions such as the father’s work might give rise to constructions such as *I like the father’s work, *I respect the father’s work, it is much harder to imagine how they could have given rise to sentence structures in which a passive participle takes over the functions of a finite predicative verb. Moreover, such a substantivized participle as what has been could mean only ‘das Gewesene’, i.e. ‘that which was’ (cf. things that are not). The so-called passive participles of Indo-European languages, when formed from intransitive verbs, always have an intransitive active meaning, not a passive one: cf. Latin impransus ‘having not eaten breakfast’; German der spät gekommene Gast ‘the guest who arrived late’ (that is, in Edith Moravcsik’s terms, an “ergative pattern”; ergative pattern, cf. Moravcsik 1978). The German construction mein Gesagtes ‘what I said, that which I said’ can be interpreted meaningfully, but *mein Gekommenes is much harder to interpret. If such intransitive participles are nevertheless used in passive constructions, this is a secondary development: a construction based on transitive verbs is extended to include some intransitive verbs as well (the so-called “unaccusatives”).

While acknowledging the role of Lithuanian, like any other language, as a “corrective authority” in testing the validity of typological generalizations, I would nevertheless be inclined to view more cautiously attempts to project into the proto-language phenomena of Lithuanian unattested in other Indo-European languages. Discussing the absence of a number opposition in the third-person forms of Baltic verbs (the child plays: the children play), the author suggests that the Baltic languages supposedly did not lose the third-person plural

129

Recenzijos

forms, as is traditionally claimed, but rather forms that they never had. He describes the traditional hypothesis as an “irrational” attempt “to derive the Baltic (like the Germanic and Anatolian) verbal system [...] from the model characteristic of Indo-Iranian” (p. 115). But separate third-person endings for singular and plural are found not only in Indo-Iranian, but also in Greek, Latin, Germanic, Hittite, and, finally, Slavic; thus, the failure to distinguish the third-person number forms in the Baltic languages is an isolated phenomenon. One may agree that the verbal model created by the Neogrammarians on the basis of Greek and Indo-Iranian needs to be revised in many respects in light of the discovery of Anatolian and other Indo-European languages. However, the special status traditionally assigned to Indo-Iranian and Greek and, in our time, for example, to Anatolian as well, reflects not geographical bias but, above all, differences in the dates of attestation. However remarkable the phonetic archaism of Lithuanian may be (which, for example, makes the Lithuanian nominative form vyras closer to the proto-language form than Latin vir or Sanskrit viráḥ), one cannot uncritically contrast facts about Lithuanian, historically attested only since the sixteenth century, with those of languages whose attested histories go back to the first or, in some cases, even the second millennium BCE. In view of the incongruence (lack of agreement) in passive participle forms in Lithuanian constructions with nominal predicates, which the author discusses at greater length (pp. 139–144) (pradalgiai suvėlta, šimušta; keliu karvės genama and so on.), cases of verbal incongruence in the Baltic languages are unsurprising as well. These matters should therefore rather be examined in a broader context, which suggests the conclusion that the absence of third-person verb forms may be due to the incongruence of nominal predicates and verbs, widespread in Proto-Baltic (although, given the τὰ ζῷα τρέχει rule, rooted in Proto-Indo-European). The third-person plural form, displaced from its original use, probably survives in participial forms such as dirbą, as Cowgill (1970) already argued.

Let us turn to the conclusions. Without a doubt, this synthesis of the research of Vytautas Ambrazas, the foremost expert on the historical syntax of the Lithuanian language, is an important work for Baltic studies. One may agree or disagree with many of the explanations he offers, but they are always interesting. The author’s main theses concerning trends in the development of Lithuanian syntax have been mentioned above. The thesis most frequently emphasized in the book, concerning the “strengthening of government,” is of questionable value, primarily because the very concept of “government” used in the study is unclear. Also surprising is the author’s attitude toward contemporary theoretical grammar, never openly expressed but becoming increasingly apparent as one reads the text: it is viewed as an intoxicating drink that is stylish and prestigious, yet too potent, from which the younger generation of Lithuanian linguists should be protected as much as possible. A linguist who has read as much foreign

130

Recenzijos

literature as Vytautas Ambrazas has is hardly likely to take seriously a syntactic model rife with internal contradictions, which is presented in The Grammar of Contemporary Lithuanian and, in essence (despite some definitions having been partly modified), is retained in the work discussed here—especially since such a model is applied nowhere in the world (except Lithuania, Latvia, and a few other countries that experienced the “occupation” of Russian Soviet grammar). But the criteria of domestic and foreign policy apparently differ. Ideas developed in foreign linguistics may exist on their own and adorn bibliographies, but in principle they are unsuitable for Lithuanian readers—unless, once passed through the proper filter, they no longer threaten the quasi-school, quasi-scientific doctrine of LKG and DLKG, whose validity is expected to continue indefinitely. The Historical Syntax of the Lithuanian Language is a valuable work, but its value would have been much greater had the author abandoned his dual role—as researcher and guardian of the national heritage—and undertaken to show young Lithuanian linguists paths for further research.

BIBLIOGRAFIJA

AMBRAZAS, V. 1970: Dėl lietuvių kalbos netiesioginės nuosakos (modus relativus). In: V. RŪKE-DRAVIŅA, ed., Donum Balticum. To professor Christian S. Stang on the occasion of his seventieth birthday, 15 March 1970. Stockholm: Almqvist & Wiksell, 6–13.

AMBRAZAS, V. 1995: Lietuvių kalbos bendraties konstrukcijų raida. Lietuvių kalbotyros klausimai 33, 74–109.

COWGILL, W. 1970. The nominative plural and preterite singular of the active participles in Baltic. In: Th. F. MAGNER & W. R. SCHMALSTIEG, eds., Baltic Linguistics. University Park-London: Pennsylvania State University Press, 23–37.

DLKG = V. Ambrazas, red., 2006: Dabartinės lietuvių kalbos gramatika. 4-oji papildyta laida. Vilnius: Mokslo ir enciklopedijų leidybos institutas.

FRAJZYNGIER, Z. 1982: Indefinite agent, passive and impersonal passive. A functional study. Lingua 58, 267–290.

HARRIS, A. C. & CAMPBELL, L. 1995: Historical Syntax in Cross-Linguistic Perspective. Cambridge: University Press. (Cambridge Studies in Linguistics, 74.)

HOLVOET, A. 2001. Once more the Baltic genitive of agent. Baltistica 35 (2), 45–58.

HOLVOET, A. 2004a: On the marking of predicate nominals in Baltic. In: BALDI, Ph. & DINI, P. U.

131

Recenzijos

(eds.), Studies in Baltic and Indo-European linguistics in honor of William R. Schmalstieg. Amsterdam & Philadelphia, 75–90.

Holvoet, A. 2004b: Evidencialumo kategorija. In: A. Holvoet & L. Semėnienė, red., Gramatinių kategorijų tyrimai. Vilnius: LKI, 105–120.

Holvoet, A. 2007: Mood and Modality in Baltic. Kraków: Wydawnictwo Uniwersytetu Jagiellońskiego.

Holvoet, A. & Judžentis, A. 2005: Sintaksinės priklausomybės tipai: papildymai ir patikslinimai. In: A. Holvoet & R. Mikulskas, red., Gramatinių funkcijų tyrimai, Vilnius: LKI, 11–38.

Litwinow, W. P. 1989: Der Modus relativus baltischer Sprachen aus typologischer Sicht. Baltistica 25 (2), 146–155.

LKG = K. Ulvydas, red., 1965–1976: Lietuvių kalbos gramatika. Vilnius: Mintis.

Meillet, A. 1934. Introduction à l’étude comparative des langues indo-européennes. 7^e éd. Paris: Hachette.

Moravcsik, E. 1978. On the distribution of ergative and accusative patterns. Lingua 45, 233–279.

Nichols, J. 1981: Predicate Nominals: A Partial Surface Syntax of Russian. Berkeley: University of California Press.

Nichols, J. 1981: Head-marking and Dependent-marking Grammar. Language 62, 1, 56–119.

Shibatani, M. 1998. Voice parameters. In: L. Kulikov & H. Vater, eds., Typology of Verbal Categories. Papers Presented to Vladimir Nedjalkov. Tübingen: Niemeyer, 117–138.

Urnežiūtė, R. 2006: Leonardui Dambriūnui – 100 metų. Gimtoji kalba 2006 vasaris Nr. 2 (464), 3–7.

Axel Holvoet

Lietuvių kalbos institutas

P. Vileišio g. 5, LT-10308 Vilnius, Lietuva

132