182 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 Martine Robbeets and Hubert Cuyckens (eds.). 2013. Shared Grammaticalization: With Special Focus on the Transeurasian Languages. Amsterdam: John Benjamins (Studies in Language Companion Series Vol. 132), XIV, 360 pp. No practitioner of the language sciences will need to be reminded that the study of grammaticalization developed into one of the most prominent fields of comparative linguistics over the past two decades. Attitudes toward grammaticalization vary, with some hailing it as nothing less than the cornerstone of a new theory of language change and others taking more cautious positions.The latter, while acknowledging the phenomenon and its importance for the understanding of language variation and change as such, tend to warn against overemphasizing it at the expense of tried and proven methods of describing and explaining language change and its outcomes.1 This fine collection brings together leading practitioners of historical linguistics, many of whom are well known for major and often seminal contributions to all aspects of grammaticalization (whether within a “grammaticalization theory” framework or not), under the general scheme of “shared grammaticalization”— defined in the opening paragraph of the editors’ introduction (M. Robbeets and H. Cuyckens, Towards a typology of shared grammaticalization, pp. 1-20) in a very clear and straightforward way as “(...) a state whereby two or more languages have the source and the target of a grammaticalization process in common”.2 The subtitle of the volume “With special focus on the Transeurasian languages” prepares the reader to find here (mostly, not exclusively) papers dealing with one or several of the “Altaic” languages (Turkic, Mongolic, Tungusic, Korean, and Japanese).3 Most of the papers do in fact deal with these languages, and some of 1) For a very clear and convincing—and much less polemical than the title might lead one to expect—argumentation in favour of a sober and non-iconoclastic use of the concept of grammaticalization see Joseph (2004). 2) This formulation, clear as it is, may attract the attention of some observers, since it raises the question, whether grammaticalization can be regarded as a “process” in the first place. While it may seem to be, at least prima facie, counterintuitive to deprive anything which can be defined in terms of a “source” and an “outcome/target” of the label “process”, Joseph (2001) argues precisely in this direction. The important nuance here is, of course, whether “grammaticalization-as-process” means that the phenomena described by this term constitute (a) unique kind(s) of “processes”, which bypass (or are not describable in terms of) well-known and well-understood mechanisms of language change. Joseph’s answer to this question is negative (and for very good reasons). If a particular discussion does not focus on this subtle (but important) distinction between the two uses of “process” (“a grammaticalization process is a process like no other” and “grammaticalization is the outcome of diachronic processes, which are understandable and describable in terms of well-understood mechanisms of language change”), there is of course no need to avoid the “process”-metaphor at all costs. 3) The term “Transeurasian” is obviously meant to replace the time-honoured name “Altaic” (which is reserved by some authors in this volume for the grouping—whether viewed as © Koninklijke Brill NV, Leiden, 2013 DOI: 10.1163/19606028-00422P05 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 183 them provide very valuable and welcome insights into interesting—and sometimes intriguing—phenomena, whether in the narrower context of grammaticalization or not. It goes without saying that it will not be possible to do full justice to all the fine and highly instructive papers in this volume here, so it must suffice to give short characterizations of each—however, in one case some more words are in order, and, I’m afraid, these will have to be rather critical. After the editors’ introduction, A.Y. Aikhenvald opens the series of papers with “Areal diffusion and parallelism in drift: Shared grammaticalization patterns” (pp. 23-41). On the basis of two quite different linguistic areas, for both of which Aikhenvald is a leading expert and author of widely acclaimed grammars (Northern Arawakan Tariana in contact with unrelated Tucanoan languages in Northern Amazonia, and the results of areal influence on Manambu (Ndu-Family, Middle Sepik Area, Papua-New Guinea) from (unrelated) Kwoma and related, but “long disconnected” Ambulas). The preliminary hypotheses (p. 25) are that, in a contact situation between unrelated languages, the recipient language can be expected to acquire traits that are rather unusual for the family it belongs to (“change against the grain”), whereas the second, more complicated (and, needless to say, more interesting) situation features genealogically related languages, which, after “long separation”, may show “parallelism in drift”, by which their “typological unity and similarity” will get reinforced. One example of such a “drift” process is the grammaticalization of a lexical verb meaning “put” to a (bound) directional marker, which has seemingly happened in Manambu and Ambulas independently. In the Amazonian area, an example is the (again, parallel, since the languages are related but “long separated”) development of a verb “to stay” into a durative marker in Tariana and Piapoco, which (p. 37) “maintain(s) a certain uniformity within the subgroup”. Had the author left it at that, these examples could well be taken as convincing examples that “areal diffusion” will indeed more likely than not produce “change against the grain”, whereas “parallelism in drift” will tend to level out the typological profile of a language family (and, to take this one step further, that the very fact of the genealogical relationship of the languages involved will, in genetic or areal in nature—without Korean and/or Japanese). I find this new name unnecessary and even misleading, and would not advocate that it be used. Linguists not particularly familiar with the language map of Asia will be directed away from the—still ongoing, and still undecided—debate on the validity of Altaic (whether in its “Micro”- or “Macro”-form), and may easily come under the impression that everything which was not fine with “Altaic”, is now perfectly in order in “Transeurasian”—which it certainly is not. Second, the term bears an uncanny (and certainly unwanted) resemblance to (i.a.) J. Greenberg’s “Eurasiatic” (a giant “lumping” endeavour, going way beyond even the various versions of “Nostratic”). And, last but not least, I see little reason to insert the syllable “eur-” into a term, when its purpose is to draw attention to the fact that a language grouping (traditional “Altaic”—Turkic, Mongolic, Tungusic), which has (if only marginal and recent) extensions to Europe, is now augmented by two languages/small families in East Asia. 184 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 some enigmatic way, have something to do with it). This is in essence what the editors of this volume seem to read in Aikhenvald’s paper, saying (p. 13f), that Aikhenvald “proposes a criterion to distinguish contact-induced grammaticalization from genealogically motivated grammaticalization”. Such a “criterion” could, in the hands of, say, macrocomparativists constantly on the lookout for such “shortcut” criteria for determining languages as related when other methods fail, certainly do more harm than good, but fortunately this is not what Aikhenvald proposes. While the phenomena discussed are of course valid and do confirm the initially formulated expectations, the explanation of “parallel drift” is by no means easy, and Aikhenvald is at pains to point out the biggest potential pitfalls of jumping to conclusions here, neatly summarized as (p. 38) “[b]ut can we exclude some prehistoric or ancient contacts between the populations speaking genetically related languages? We cannot.” And another caveat is, of course, that the features under discussion “may in fact have been present in the proto-languages and may well be interpretable as parallel genetic inheritance” (ibid.).4 The theme of “Sapirian drift” is of course already present here (and duly mentioned in the paper), which elegantly opens the stage for B. Joseph’s brilliant study “Demystifying drift. A variationist account” (pp. 43-65). Using examples from Germanic and Indo-Iranian, he goes a long way towards removing the ominous “invisible hand” (p. 43) from the discussion of Edward Sapir’s famous “drift”, or “independent but parallel developments”. The rationale he proposes—and illustrates with a rich set of phonological and morphological examples—is largely sociolinguistic in nature, and summarized on p. 63 as (omissions mine): “There was a variation in a proto-language (...) and the variation can be inherited into later stages as continued variations, but it can be modified, e.g. generalized (...), or it can be “submerged”, e.g. sociolinguistically restricted, for a while before re-surfacing, ‘bubbling up to the surface’ (...)”, or, shorter, as “parallel resolutions of proto-language variation” (ibid.). If we add, as Aikhenvald implies, intra-branch contacts (which, in some cases, may have escaped historical records), we may be gradually filling up a toolbox that allows us to deal with many, if not most, cases of “Sapirian drift”—without mystical “inherited forces”.5 B. Heine and M. Nomachi’s “Contact-induced replication: Some diagnostics” (pp. 67-100) are concerned with the question of how language changes may be identified as contact-induced, even when no “form-meaning units and no phonetic substance are involved” (p. 68). They propose a series of “diagnostics” (all illustrated by ample, and mostly well-chosen, examples), none of which, they stress, 4) The problem can only be to (try to) decide between post-proto-language contact or the “going underground” of inherited features, which may later “bubble up” in separated languages in the sense of Joseph (this volume, pp. 43, 63)—using such phenomena for phylum recognition, i.e. for an attempt to justify a genealogical grouping in the first place, would be putting the horse before the cart. 5) As invoked by Robbeets (this volume, p. 170), thus “remystifying” drift in spite of this. Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 185 can be “in itself (...) entirely sufficient to ‘prove’ language contact, but the more critical criteria apply, the stronger the case for contact-induced grammatical change” (p. 71). These are: Intertranslatability, Genetic patterning (meaning that some feature, common to languages A and B, is common or widespread in dialects and/or family cousins of A, but isolated in B), the involvement of Rare grammatical categories (admitted to be a rather “relative notion”), Paired structural similarity (i. e. shared polysemy patterns), Demographic variables (the restriction of certain changes to speaker groups, for which contact situations are more frequent, as opposed to a lack (or weaker representation) of such features in more isolated groups), Differences in grammaticalization (“the more intensive the contact is, the more grammaticalized the contact-induced elements will be” (paraphrase mine)), Rare grammaticalization. The last criterion is said to help in determining the direction of the transfer, and is labelled Degree of grammaticalization, postulating that “If two languages have undergone the same process of grammaticalization as a result of language contact but one of them exhibits a high and the other a low degree of grammaticalization, then the former is more likely to have provided a model of replication than the other way round” (p. 89f.). This catalogue of criteria is very useful (and summarizes many of the criteria that have been and are routinely—if often tacitly—used by many practitioners of historical and comparative linguistics), and the authors back them up with theoretical and factual observations and data. I do, however, have some problems with some of the examples presented. Thus, on p. 75 Fn. 6, the grammaticalization of the verb for “to come” into a modal auxiliary for the expression of necessity in Estonian is identified as the model for the same phenomenon in Latvian. In Fn. 6, however, it is mentioned as an alternative that Russian may be the culprit, which may have passed this grammaticalization to all of the languages involved (i.e. Finnish, Estonian, and Latvian, with which Russian is in contact).6 But which Russian model construction is meant here? Moreover, structure-changing language-contact between Russian as donor and Finnish as recipient language is marginal at best, in spite of Finland having been a Russian possession for some centuries— there was never any Russian settlement in Finland to speak of, and the official language of the territory was Swedish, never Russian. The ultimate donor language for Latvian was, in all likelihood, not Estonian, but Liv—by a substratum relation6) And not with Lithuanian, which should explain the absence of this feature in this language. But this is hardly correct; it may look like this, if one looks at a modern map of the region, which will show the Belorussian language as the “neighbour” of Lithuanian—overlooking the fact that, nevertheless, it was (“Great-, East”-) Russian (the language of the Muscovite state and its successors) which exerted much more linguistic influence on Lithuanian than any variant of Belorussian ever had the chance to—and, what is more, Belorussian is, historically, nothing more than a variant of the East Slavic dialect continuum, like Russian itself—given that the time-depth of the contact situation involved here is neither mentioned nor discussed, the contact may as well date from a period, in which the notions “Russian” and “Belorussian” were indeed still meaningless. 186 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 ship (the historical Liv(onian) territory is largely coextensive with that of the East Baltic dialect, which developed into Latvian)—the polysemy of “to come” and “necessity auxiliary” is, indeed, present in Liv as well (cf. Sjögren 1861: 118). Another example on which I have to express some disagreement is an instance of shared polysemy between Azeri and Persian (p. 79); here, the polysemy of the Azeri verb goy- (“to put”, but also “to let, allow”) is explained as due to influence from Persian, where the verb gozāštan shows the same semantic range: “Since this particular kind of ‘polysemy’ is hard to find in other languages, but occurs in these neighbouring languages known to share a long history of contact, there is reason to assume that it is due to language contact.” But this is not rare. It is widespread in the Turkic language family and found, e.g., in East Turki (“Modern Uyghur”), Uzbek, Kumyk, Kirgiz, Karakalpak, and, while some of these languages show a certain degree of direct (but most of them rather indirect and literary) influence from Persian, this is certainly excluded for Karaim, for which cf. Baskakov et al. 1974: 329).7 So it needs to be added that, while using a set of diagnostic criteria as the one presented here by Heine and Nomachi will certainly be useful, and we have all reasons to thank them for presenting it here in an exceptionally clear and stimulating way, it can never replace the one criterion (which may, it is true, not always be available) that should precede every purely linguistic (typological, grammaticalization-theoretical, or other) reasoning: the thorough knowledge of all available (and retrievable) data on the languages studied and the phenomena under scrutiny. Opinions offered without such an inevitable backing, and exclusively based on secondary literature, all too often turn out to be ill founded after some checking of data is done. Also, the authors’ quite adamant insistence that the Ukrainian synthetic future is definitely based on the grammaticalization of the verb ‘to take’ (which would make it a rare grammaticalization and the fitting model for a similar development in some Romani dialects in Russia), against B. Joseph’s objections (cited as “p.c.”) that this may just as well represent a (by no means rare) ‘have’-future, invites opposition. The authors’ comment that the synthetic Ukrainian future cannot be based on the “have”-verb maty is correct, but anachronistic. It is, of course, based on the diachronically precedent (and now obsolete) verb iměti 8 (attributable to at least East Slavic—i.e. Pre-Russian/Ukrainian/Belorussian, and further to Proto-Slavic), with a range of meanings between “bringing oneself into the state of possessing” (i.e. ‘to take’) and “being in this state” (i.e. ‘to have’). Since even Old Church Slavic (= Old Bulgarian) bases a (periphrastic) future on this verb (where it definitely departs from ‘have’-semantics), there is little justification to justify the cross-linguistically rarer ‘take’-interpretation over the (more widespread, Romance-like) ‘have’-scenario for the Ukrainian synthetic future. 7) To this should be added that attestations of this verb with both meanings abound in Middle Turkic (14th century). 8) Cf. Žovtobrjuch et al. (1980: 214). Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 187 L. Johanson’s “Isomorphic processes: Grammaticalization and copying of grammatical elements contribution”, (pp. 101-109) undertakes to clarify the author’s thinking on grammaticalization and especially on the mechanisms by which grammaticalization phenomena can come to be shared between different languages. He underlines most emphatically that grammaticalization processes cannot be copied (his terminology for borrowed), and that, in such cases, everything that is copied is the output of the process (p. 105).9 The same holds for the “inheritance” of grammaticalizations.10 Here, too, there can be no talk of “processes” being copied; only “elements”, not “processes”, can ever be the undergoers of copying. Johanson’s paper is not accompanied by much language data, and the example he does mention I find somewhat surprising. This is the author’s dictum that Chuvash and Yakut, two “non-mainstream” Turkic languages, do not show the grammaticalization of the numeral *bir “one” to an indefinite article, whereas all other Turkic languages do (with implications for the history of the whole family). But this is simply not correct—it is, on the contrary, not at all difficult to find instances of the respective Chuvash (pĕr) and Yakut (biir) numeral in exactly this function, cf.: (1) Chuvash: ǝlǝk pǝr pujanśi̮n pulnǝ̂ earlier one/INDEF rich person was ‘Once upon a time, there lived a rich man.’ (Paasonen 1949: 175).11 (2) Yakut: Bılır manna biir ėmėėxsin olorbuta ühü earlier here one/INDEF old.woman lived (hearsay) ‘In former times, an old woman lived here, it is said.’ (Ubrjatova 1982: 177).12 9) Recall the discussion of “grammaticalization as a process” mentioned above; for Johanson’s purposes, it is of course quite impossible to avoid the term “process” here, in the most general sense of the “X”, which combines two temporally distinct, but diachronically identical, states of affairs (this phrasing is mine, and may or may not represent Johanson’s thoughts correctly). 10) On a more abstract level—with all due caution, and certainly without wishing to downplay important and salient differences—it could be said that (lateral) “borrowing” and (vertical) “inheritance” are but two variants of the same phenomenon—in both scenarios, linguistic elements (to use the broadest possible term), are “copied”—both have a “Model Code”—the “donor” language in a borrowing scenario, the “ancestral” language in an inheritance scenario; while direct/vertical inheritance is, then, characterized by the absence of a pre-existing “Basic Code”—this code is produced by the process—individuals (in this case first language learners) certainly “copy” linguistic behaviour/elements from a “Model Code” (that of the linguistic community they grow up in). 11) These folklore texts, collected in 1900, literally abound in clear and unambiguous instances of pĕr in indefinite article function, in narratives, riddles, and songs—they are present on every single page of the book. 12) It may be true that, in Yakut (certainly not in Chuvash), this function is “less obligatory/ frequent/widespread” or, then, less “grammaticalized” than in other Turkic languages, but 188 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 (3) Biir küölü one/INDEF lake-ACC ‘We passed a lake.’ aastıbıt we passed (ibid.) V. Gast and J. van der Auwera’s contribution “Scalar additive operators in Transeurasian languages: A comparison with Europe” (pp. 113-145) deals with the semantics and the distributional properties of adverbs with meanings like German sogar or English even, which (p. 114) “indicate that a given focus value (...) ranges higher on a scale of ‘pragmatic strength’ than any alternative value under discussion”. The paper builds on and expands the authors’ findings presented in Gast and van der Auwera (2010), where European languages were in focus, by taking a closer look at some Altaic languages, namely Turkish, Gagauz, Azeri, Kumyk, Tatar, Modern Uyghur, Uzbek, Yakut, Ewenki, Udihe, Khalkha Mongolian, Japanese, and Korean. Their contribution is a welcome addition to an area, which is certainly not well studied, and the typology around the dimensions scalar/nonscalar, upward/ downward entailing, negative/non-negative they develop is highly interesting. Needless to say, even this long paper can only scratch the surface of this matter, and it is certainly not the authors’ fault that they could obtain quite a number of examples only from their colleagues (“p.c.”), rather than from available and good studies of these operators in grammars or linguistic literature.13 I have several problems with the contribution of one of the editors of this volume, M. Robbeets’s chapter “Genealogically motivated grammaticalization” (pp. 147-175). The purpose of this paper is clearly (and chiefly) to convince its readers— once more—that the “Altaic” languages (Turkic, Mongolic, Tungusic, Korean, and Japanese, here renamed as “Transeurasian”) form a valid genealogical family, and not, as critics of this hypothesis maintain, a grouping of languages, whose (undoubted) commonalities are due to areal consolidation of originally unrelated whether this is an archaic or innovative trait of Yakut is less than clear. I found a further surprising statement on bir as indefinite article in Turkic in Fn. 2 (p. 11) of the editors’ introduction: “The presence of the grammaticalization pattern in Tatar, in spite of being under Uralic influence, suggests that the absence of the indefinite article in Yakut and Chuvash is due to genealogical rather than areal factors”. The non sequitur of this statement (and the error on Chuvash and Yakut) apart, Tatar is not “under Uralic influence”. It is spoken in the vicinity of some Finno-Ugric languages (which are definitely more influenced by Tatar than the other way round), Cheremis and Mordvin, and both of these (especially the former) do show the numeral “one” in indefinite article function, a host of examples can be found in any text collection, as e.g. in Genetz 1895, passim. The editors’ quote is, thus, at the very least, confused and raises the question, whether they really feel comfortable in the realm of Altaic linguistics. 13) The Azeri, Kumyk and Uzbek example sentences Nos. 49, 50 and 54 (“p.c.”) come directly from standard Russian-based dictionaries of these languages, and are all findable there under the lemma entry даже (for future studies along these lines, and with these languages, it may be hoped that primary sources will play a greater role than isolated and normative dictionary examples). Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 189 languages. M. Robbeets has devoted a long series of books and papers to this question before, some of which, as this one, concentrate on what she claims to be morphological evidence in favour of the genealogical interpretation. The core of the argumentation here is a list of 13 morphological comparanda, adapted from earlier lists with varying membership (yet around a somewhat stable core), presented by the author on numerous occasions. This is now accompanied by a set of criteria, all centered around the concept of grammaticalization, which allegedly allow one to determine whether the comparanda are more likely to be due to common heritage from a proto-language (“genealogically motivated”) or rather motivated by chance, general tendencies or principles of language change (“universal factors”), or borrowing. It is not surprising that she finds, in most cases, the criteria to be fulfilled and, consequently, yet another case for the genealogical scenario to be successfully made. Robbeets’ seven criteria for shared grammaticalization phenomena, which are said to strengthen the case in favour of an ultimate relationship, are the following: (1) The shared grammaticalization is global (fulfilled, when source and target of the grammaticalization are shared by the languages involved both in terms of semantics and form, the opposite being a “selective” grammaticalization). An immediate reaction to this line of argumentation is that this criterion is rather artificial—why should the source of a grammaticalization be present in the first place? All too often, elements are lost in the course of language change (or alter their meanings), so why should such a loss or alteration play a game-changing role for linguistic assessments of past processes?14 But there are more problems with this, of a more fundamental nature. Thus, looking at Robbeets’ list of “shared global grammaticalizations”, it quickly becomes clear that the very claim that we are dealing with cases of grammaticalization here in the first place is built on shaky grounds at best. Instead of using the core definition of grammaticalization, which is used to describe the development of erstwhile lexical elements to bound grammatical markers (with semantic bleaching, loss of phonetic substance etc.), the term is expanded here to include the development of elements from “less grammatical to more grammatical status” (p. 156). This is certainly defendable (and widespread practice in grammaticalization theory), but it has to be mentioned that the examples presented here15 are almost exclusively of this kind and do raise serious doubts whether they really fit into the framework of grammaticalization. To wit: what, then, is “more grammatical” in a “loan verb marker”, as opposed to its function as a general verbalizer?16 The same question can be asked for the causative-passive 14) Apart from, of course, making the linguist’s life a bit more difficult. 15) With the exception of No. (4): “do, make” → causative marker, and possibly (12) and (13): negative verb →negation marker. Both of these are not discussed with examples by Robbeets in this paper, and can, consequently, not be discussed here, since it would entail too much guesswork to know what she really has in mind. 16) The ubiquitous verbalizer *-la- (in Turkic, but also in Mongolian), can be used to verbalize virtually everything, which is semantically suited to make a verbalization useful for the 190 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 polysemy, which is indeed found in some Altaic languages (but also elsewhere: Johanson, this volume (p. 108), mentions, a bit vaguely, Caucasian languages),17 but again, there is no need to invoke the concept of grammaticalization here— Johanson explains Altaic causatives/passives as instances of the category of transcendence,18 without any allusion to grammaticalization (or, then, any category-changing “process”) at all.19 And, finally, the “pathway” from non-finite to finite use of some affixes invites similar comments. Neither here, nor in other publications of hers on these markers,20 is a convincing argumentation given for why any grammaticalization should be involved in the fact that (most) verbal nouns in (most) Altaic languages can be used for finite and non-finite purposes. Once a language (say, Mongolian) has reached (whether through areal consolidation or not) the stage where all (or most) verbal nouns can invariably be used in finite and non-finite (adnominal, relativizing etc.) function(s), this language can accommodate such verbal nouns from a donor language (say, Turkic) with exactly this set of functions (whether these are the results of grammaticalization in the source language or not) into its system, without anything in the way of grammaticalization. This possibility is, needless to say, systematically overlooked in Robbeets’ approach (see further on this below on Malchukov’s paper). Criterion (2) concerns the rarity of some grammaticalization patterns—this is certainly a quite subjective parameter (and greatly dependent on the observer’s— or then the discipline’s—current and available knowledge of what is rare, and what is not). But, again, I find a more fundamental flaw in this way of thinking, since a geographical clustering of phenomena (globally rare or not) will always lead to a suspicion of areal interaction, if the languages involved are not in an inspectionally obvious way related or otherwise (by traditional, non-”shortcut” methods) confirmed as members of a true family.21 To make them look a bit “rarer”, loan-verb users of the language, including, of course, the subset of “everything”, which is made up by borrowed verbal roots. Promoting this use to the status of a special case of grammaticalization is tantamount to hypostasizing what is at best a co-function of a functionally already quite broad morphological element in order to obtain just another “grammaticalization process” (and, importantly, these are counted in the end to make the case appear more convincing). This, and the other cases named here, are classic examples of reification by inflating nomenclature. 17) Some Uralic languages could be mentioned here, too. 18) See Johanson (1990:211) with an exceptionally clear formulation, also Johanson (1974). 19) Johanson carefully avoids the “grammaticalization” metaphor for this in this volume (108) as well, and not without reason. 20) As, e.g., in Robbeets (2009), where the allegedly different functions of the morphological markers are demonstrated either from the same chronological layer of the respective language or, sometimes, even with an example for a putative source of a grammaticalization in later, and the target of this in an earlier layer (e.g. p. 70, with two misinterpretations of Mongolian morphology, see Georg forthcoming). 21) Both criteria do, of course, not hold for the Altaic languages. If the typological rarity of features could play an independent role for lowering the scales in favour of genealogical Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 191 markers (on which see above) are said to be non-existent in Uralic languages, but they do exist in some of them.22 The 3rd criterion, which stipulates that “globally shared grammaticalization (which) concerns two or more instances” strengthens a case for relationship, is of a quantitative nature and stands and falls with the item-by-item assessment of the validity of the elements discussed (see below). Number (4) says that a good criterion is fulfilled, when the phenomena are “not restricted to contact zones”. This is true as far as it goes (and also holds for lexical and other commonalities), but, again, some caveats here: since the languages involved are Proto-Languages in the first place, what are the contact zones between Proto-Turkic and Proto-Mongolic, then? Can we locate them in space? While some would say that we can, others (the present writer included) maintain that we cannot—at best we can make more or less educated guesses on this. When Robbeets repeatedly mentions “low contact” between the members of our grouping (pp. 156, 168) this obviously refers to the location of all these languages on a modern map—a grave error in this context, since, which is often overlooked, both hypotheses—that of the genealogical unity of the Altaic languages, and the areal consolidation scenario—operate with the assumption that all of them were once spoken in much closer geographical proximity than in modern (or documented) times (the genealogical hypothesis, by virtue of being itself, even has to assume historical identity, i.e. locate the five protos at one point in time and space, with a distance of zero between them, gradually moving apart from there).23 The 5th criterion ascribes a higher “genealogical value” to grammaticalization patterns, which move already bound elements from higher to lower grammatical status (i.e. not involving lexical sources), because bound morphemes are “more inheritance as opposed to areal consolidation, most “Balkanisms” could be taken as immediate evidence for the relationship of the languages involved—which they cannot (notwithstanding that Balkan languages are ultimately related, of course; they are, but for different reasons). 22) E. g. in Ostyak and Vogul (cf. Kálmán 1961: 109-111), Ganschow (1965: 107-109); Robbeets’ characterization of Uralic languages as “exceptional in this respect” (p. 169) is an artefact of Wohlgemuth’s (2009, her source) sampling, which did not include these languages for Uralic—are generalizations of this kind really made so quickly these days? 23) Modern language geography can (and should, of course) play a role in our discussion, but in a different perspective: elements only present in historically attested or directly observable contact zones (cf. Pakendorf, this volume), and absent from other parts of the territory a given family occupies, may strengthen the suspicion that contact is the culprit, but this cannot be simply reversed, as it seems to happen here, much less so in the case of reconstructed proto-languages with no known localization in space or time, let alone any robust information about the possible details of their putative geographical subdivisions (“Western Proto-Mongolian stood in more intensive contact relations with Eastern Proto Turkic than with Southern PT”). Note that the absence of any information of this kind sets boundaries for the speculations of both camps—the differentiation of “low” vs. “high” contact zones is meaningless for any reconstructed proto-language. 192 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 resistant to code-copying” than independent lexemes. The value of this criterion is again limited, since it can be (and has been) repeatedly shown that bound morphemes may indeed get borrowed somewhat less frequently than words, but that their “copy-proofness” is an illusion at best—if the contact situation is intensive enough, bound morphemes can and will move from language to language (cf. Georg 2011 for more on this). The 6th criterion says that a strong point in favour of relatedness is scored, when “the globally shared grammaticalization spreads over more than two (proto-) languages”. Interesting as such situations may be, I can see no reason why they should be of any special value for the argument at hand. Robbeets mentions, by way of an example, the Old Chinese verb 得 (*tək, Mandarin dé) “to obtain”, borrowed into Vietnamese as được, with the same meaning, both also grammaticalized as a passive marker. On p. 156 we read, quite apodictically: “Vietnamese has (not) globally copied its verb ‘to receive’ and subsequently transferred the grammaticalization of ‘to receive’ into a passive marker to any third language”. But similar verbs, with similar grammaticalizations, can be found elsewhere in (un- or distantly related) languages in South-East Asia, too (cf. Matisoff 1991: 420-425), to wit: Thai thùuk, Yao (Hmong-Mien) tùˀ, and possibly Khmer trǝw.24 And, what is more, this criterion can exert its convincing power only, when the comparanda adduced to underpin it have been vindicated as valid—which is far from achieved, despite repeated claims to the contrary (this holds, of course, also for the seventh and last criterion, which involves that a “specific pathway of grammaticalization is recurrent in more than one cognate set”); to this may be added that a given grammaticalization, say, from deverbal noun to converb25 (and which may or may not be contact-induced) may spawn further similar grammaticalizations in the language by pattern extension, without any further input from any donor language, which further reduces the diagnostic value of “shared grammaticalizations” for discussions of genealogical descent. 24) The Khmer verb may as well not belong here for phonological reasons, and the Thai verb does not show the expected lexical (source) meaning (“hit, touch, be right”). Add here Laha (Kra, in Northern Vietnam) hôp1 “obtain, passive marker” for a “non-global” grammaticalization along the same lines as in Chinese and Vietnamese (Solncev 1986, 69). Whatever the ultimate solution to the questions around these SE-Asian words may be, there is no need to assume that the elements (and their grammaticalized functions) were necessarily passed successively from one language to the next, as Robbeets’ formulation implies. The early contact situation(s) in Altaic may (and will) have been just as complex and non-linear as in this case. The (expectable) objection that this example misses the criterion of “globality” is of less importance for this comparison, since, contrary to Robbeets’ claims, the Altaic examples in her paper fall short of this criterion as well (and, as mentioned, most of them do not even involve grammaticalizations in the first place). 25) Examples (10) and (11) on Robbeets’ list; I see problems with both of them, which will be discussed in Georg (forthcoming). Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 193 Finally, Robbeets’ list of comparanda itself cannot be passed over in complete silence here. Impressive as it may look at first sight, there are serious problems with it, some of which have been mentioned in the literature before (obviously without any consequences for the authors’ confidence in them—one would be interested to learn why). Of the thirteen comparisons, only four are presented in some detail,26 which allows (especially non-specialist) readers to form some opinion (*-lA-, *-dA-, *-i, *-rA); for all the others, the reader is referred to previous publications of the author (without detailed references). All these comparisons should, however, be scrutinized from different angles, to see how many (and which) of them really hold water. Readers familiar with the discussion on Altaic will certainly not be surprised to hear that I insist that this list be rigorously sifted, and that many of the elements present on it will have to be eliminated, because known and demonstrable facts of the history of the languages involved do not allow for many (I dare say the majority) of the reconstructions and functional descriptions given by Robbeets. I do not expect any reader of these lines to simply take my word for this. Knowing that underpinning such a claim with concrete data is an absolute necessity, I may use the opportunity to announce a book I have currently in preparation, in which I intend to discuss Robbeets’ (and other Pro-Altaicists’) morphological and other comparanda in all due completeness (and I feel justified to claim serious reasons of space for not doing it here).27 I have added some comments on a similar list by Robbeets in Georg (2011). On this, Robbeets comments here (p. 148) that the present writer “(has) attempted to explain these correspondences in terms of borrowing”. This is wrong. I have done nothing of the kind there, I expressly said, in what I hoped to be comprehensible English, that I do not think what she says I do, and that my approach to these comparisons—which I indeed view as not indicative of any genealogical relationship between these languages—is at least a trifle more complex. I will of course welcome each and every objection or rejoinder to anything I ever say or imply (and I certainly do not claim infallibility), but a misrepresentation of expressed views by 180 degrees falls, I may be allowed to say, somewhat short of proper academic conduct.28 26) “For reasons of space” (p. 159)—were ten more pages such a big problem in a self-edited volume? At least, clear references to previous publications of the author, where more details could be found, should have been expected as a minimum. I was able to identify most, but not all of such loci in Robbeets’ earlier publications. 27) The points to be illustrated there include that the Turkic verbalizer *-la- can be shown to be composite (and borrowed into other, Altaic and also numerous non-Altaic, languages); some substantial objections against the Japanese comparandum may be found in Vovin (forthcoming); Turkic *-(A)d- “passive” is to be eliminated, also the Mongolian “passive “ *-da-, and the Mongolian “converb *-i etc. Not each and every single element will have to leave this list, but the proper sifting will leave it considerably reduced and way less useful for the purpose it was compiled for. 28) What I do there instead (Georg 2011: 29-33) is to take a 17-item (that is four items more than on the list presented here) list of M. Robbeets’ Altaic morphological comparanda and 194 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 A. Malchukov’s contribution “Verbalization and insubordination in Siberian languages” (pp. 177-208) argues in favour of an areal propensity of Siberian languages for the so-called insubordination scenario, by which finite verb forms are renewed through forms of non-finite origin (p. 177). He illustrates such patterns (and verbalizations, i.e. the reanalysis of nominal predicates as verbal predicates (p. 201) in Tungusic (no other Altaic languages are discussed), and then in Chukchi, Yukaghir, Ket, Eskimo, and Nivkh. For Ket (which is cited in three different transcription/parsing systems), he mentions a rather marginal and very confined pattern from the system of verbal agreement (the so-called Vth conjugation, cf. Georg 2007: 201f). Actually, Ket lacks clear instances of “Altaic type verbal nouns”, and it may be worthwhile to mention that the areal pressure from neighbouring Turkic languages has led to a quite interesting process in this language, by which a clearly finite verb form (with the hallmark of its finiteness, morphological subject reference, intact) can get case marked (“Turkic style”) to allow for a verbal-noun-like subordination pattern, as in (parsing simplified, cf. Georg 2008: 163f.): (4) ǝ̄t dǝ́ŋon-diŋal qō sɨ́kŋ ukon we we.went-ABL ten year.PL they.went ‘After we went away, ten years have passed.’ This is completely parallel in structure to the following Khakas (South-Siberian Turkic), with the difference that the predicate of the subordinate clause is a “normal” finite verb form in Ket, while a verbal noun in Khakas: (5) pis par-γan-naŋ on žıl irt-ip we go-VN-ABL ten year pass-CONV ‘After we went away, ten years have passed.’ par-dı go-PST Thus, while Ket certainly participates in the areal mechanism mentioned by Malchukov, it also has its fair share of opposing trends, which are also operative in the area—here, Ket «imitates» a canonical Turkic structure, but not quite, because it simply does not have the morphological means for a perfect emulation of it. to show that for all morphemes on it, functionally and formally equivalent comparisons can be found in Volga-Finnic languages as well (fifteen in Cheremis alone, two more in Mordvin). Many of these suffixes are clearly borrowed from neighbouring Turkic languages, but many clearly are not (and many of these can, further, be shown not to be of Finno-Ugric or Uralic age, thus the possible escape route to an expansion of Altaic to the long-abandoned UralAltaic, let alone Nostratic, remains closed). I described my Cheremis/Mordvin-control list as the result of the “conspiracy of inter-language borrowings and chance similarities” (p. 34). Similar reasons are responsible for the Altaic comparanda as well, not more, not less. Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 195 On a general note, I have to say that I find this phenomenon here (and also in Robbeets 2009) somewhat «overterminologized».29 The use of this (“-ation”) term implies that the observed phenomena are to be viewed of necessity as outcomes of a (grammaticalization) process—and one with a known input at that. This is not necessarily the only way to see the phenomena. It has always been a hallmark of (the description of) “Altaic” languages that their verbal system can be subdivided into the three categorial realms of (exclusively) finite markers, converbs, and verbal nouns.30 The latter are mostly (if not in every single case—Mongolian, with its lack of person agreement, is a prototypical representative of this type) also capable of being the sole finite verbal element in a sentence. Now, “Altaic” languages are certainly not as uniform in this respect as traditional scholarship might have wanted to make us believe, and thus it is absolutely correct and important, as Malchukov does, to explore and point out varying degrees of “finiteness” (which seems to be especially fruitful in Tungusic, his section on “The noun-verb continuum in Tungusic”, pp. 183-189, is particularly interesting in this context), and to try to give the findings of such research a diachronic dimension (in search of a “grammaticalization history”). On the other hand, the underlying assumption that the finite use of what seem to be mostly infinite verbal markers (“verbal nouns”) has to be (always?) regarded as the endpoint of a grammaticalization process is more problematic—in most of such cases the starting point of such a “grammaticalization process” cannot be observed, let alone a gradual movement to the target state, but has to be postulated at best. Instead, the canonical Altaic model, in which verbal nouns are polyfunctional and may fulfil non-finite and finite roles simultaneously, should, at least provisionally, be taken at face value and treated as typologically justified (notwithstanding that fine studies as this one will, then, shed more light on subtle differences, which may indeed show that Tungusic is “less canonically Altaic” than many observers still might think). J. Janhunen’s contribution “Personal Pronouns in Core Altaic” (pp. 211-226) does not deal with grammaticalization phenomena at all, but takes up the time-honoured problem of explaining the undoubted isomorphisms and material commonalities of the 1st and 2nd person pronouns in Turkic, Mongolic, and Tungusic. His conclusion is that, in spite of the striking (and well-known) similarities between these elements, they cannot be used as a decisive point for a genetic scenario involving these languages, and that borrowing, but also “shared drift”, are the mechanisms to be reckoned with here. When he (220) reconstructs the primary shapes of the Turkic pronouns as *bi and *si (“confirmed by Chuvash”, p. 220), it might be useful to compare a recently published study on the same problem by A. Vovin (2011), where it is argued that the Chuvash /ĕ/’s in these pronouns should be seen as reflexes of Proto-Bulghar *i, itself an innovation from Proto-Turkic *e, which, if accepted, adds another—crucial—element to the roster of Bulghar loans in early 29) With the imminent danger of reification. 30) Often referred to as participles in the literature. 196 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 Mongolian (and would, thus, add an “indication as to what the direction(s) of borrowing could have been” , p. 222). H. Nugteren’s “Postposed indefinite articles in Mongolic and Turkic languages of the Qinghai-Gansu Sprachbund” (pp. 227-250), is a magisterial treatment of a grammaticalization phenomenon (the development of the numeral “one” into a postposed indefinite article in Salar, Baoan, Kangjia, Mongghul, and Mangghuer, clearly replicating an Amdo-Tibetan model), with ample discussion of details of usage in these languages—very commendably, a Salar text sample is appended (pp. 248-250), which illustrates some of the findings for this fascinating Turkic language. É.Á. Csató’s “Growing apart in shared grammaticalization” (pp. 251-258) discusses the historical development of “indirectives” in Turkic languages along the lines of retention, loss and renewal of this category. The discussion is very concise, mentions morphological elements, but without giving illustrative examples, and will, it has to be said, fully comprehensible only for linguists who are intimately acquainted with the life-long work of Lars Johanson and his school on aspectotemporal categories in Turkic and beyond (and its terminology). This apart, Csató argues convincingly that the renewal of this category (after semantic erosion) with new (or “redeployed”) morphological material is best attributed to intra-family contacts in the central zone of the Turkic speaking world. It would, however, be quite wrong to attribute the ease with which such contacts could and can “stabilize” (or even “repair”) categories to the very fact of the relatedness of the languages involved.31 Turkic languages are related, but they are also (sometimes very) similar,32 which of course presents numerous opportunities (and pathways) to renew eroding categories by imitating external models (“endo-hybridization”). Would the contact-induced patterns be observable in languages, which are (demonstrably, but not intuitively) related, like, e.g., Albanian and Serbo-Croat, Hausa and Arabic, or Vietnamese and Khmer? Hardly. While contact between these languages may produce all kinds of interesting phenomena, the very fact of their ultimate relationship—long forgotten and irretrievably beyond the reach of their speakers—cannot play any role for them.33 The scenario Csató discusses 31) Such a misunderstanding might be present in Csató’s conclusion that “intensive contact between genealogically related varieties can result in shared selective grammaticalization” (p. 258), and certainly is in Robbeets’ assessment (p. 170) that “related languages exhibit a strong tendency to maintain pre-existing categories in spite of formal renewal”, which is even elevated to the status of a “language-internal force”. 32) With well-known exceptions, most of them still maintain a considerable degree of mutual intelligibility—certainly, most speakers of a Turkic language/dialect will recognize any neighbouring variant (even if linguists put them in different subbranches of the family) as “something like their own language”. 33) And any observation made in the contact situations of such languages can, of course, not be used for any typology of language contact between “related” vs. contact between “unrelated” languages, with the ulterior motive of developing criteria for phylum recognition “through the backdoor”. Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 197 here, is, moreover, to be sharply separated from “shared drift” (p. 257, also misunderstood this way by Robbeets and Cuyckens, p. 16, “parallel drift”), since the “drift” metaphor describes cases, where prima facie intra-family contacts are excluded as immediate instruments for the explanation of parallel developments—whereas in this Turkic case the opposite is the true.34 B. Pakendorf’s “Incipient grammaticalization of a redundant purpose clause marker in Lamunxin Ėven: Contact-induced change or independent innovation?” (pp. 259-283) is a masterly study of an intricate problem in the contact history of (Tungusic) Ewen/35 Lamut and (Turkic) Yakut. The question of whether the Lamunxin purposive construction (using a purposive converb, accompanied by a redundant converb form of the verb “to say”) is to be viewed as a structural borrowing from Yakut is, in the end, answered positively, but before allowing herself this conclusion, Pakendorf discusses a host of possible objections to and problems for such a statement from all imaginable angles (extent of use in the donor and recipient languages, general typology of purpose expressions, language geography, intra-Turkic and intra-Tungusic observations). This is a model of scholarship from a true expert in the field, which shows that fruitful results are only obtainable, if any jumping to conclusions is carefully avoided and the discussion of possible problems for one’s own hypothesis is given due and unbiased attention from the outset. The long and data-rich paper by H. Narrog and S. Rhee “Grammaticalization of space in Korean and Japanese”36 (pp. 287-315) deals with case particles, relational nouns, postpositional verbs and demonstratives in the two languages, which are used, in one way or another, to encode concepts of space, including terms of absolute and relative orientation, giving due attention to the pervasive influence of Chinese on Japanese and Korean. The authors remain somewhat undecided (which is not to be criticized), whether the striking similarities they find in the grammaticalizationpatterns involved are to be explained areally or genealogically (but they do mention that “the majority of lexical sources is not shared” (p. 311), which should direct further thinking on this matter). A. Antonov’s “Grammaticalization of allocutivity markers in Japanese and Korean in a crosslinguistic perspective” (pp. 317-339) examines the history of alloc34) In other words: “Sapirian drift”of the kind B. Joseph demystifies in this volume would not be detectable in this situation in the first place. If anything, “drift” is not a particular and unique “process” at all, but an observational situation, which calls for explanations not immediately available. In spite of the clear exposition of Joseph, this misunderstanding is still present in some contributions to this volume (cf. Robbeets, p. 170—a “language-internal force”). 35) This rendering of the name of the language (and that of closely related Ewenki) is, in my view, to be preferred over the direct transliteration from /w/-less Russian; in this volume, it is constantly used only by Janhunen. 36) For the three following contributions, I insert here the common disclaimer that I must and will leave the judgement of all language facts presented and discussed for Korean and Japanese to competent specialists. 198 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 utive37 markers in Korean (-(su)pni-) and Japanese ((i)mas-), and finds that their grammaticalization went along strikingly parallel lines (departing from objectexalting verbs). His study is the only real historical contribution to this volume, since Antonov traces the development of these markers through the whole documented history of both languages, and he finds little reason to adduce this parallelism to “Sapirian drift” (since he is quite positive that a common ancestor for Korean and Japanese cannot be reconstructed) or (for chronological reasons) to straightforward language contact—but I think this dilemma can be resolved. “Drift”, as the term is commonly used, refers to parallel developments of related languages, which, after separation, lost contact, true enough. Yet the lines of thought described by Joseph in his enlightening demystifying paper in this volume can easily be used to explain similar situations from early (and, just like ProtoLanguages, not immediately observable) contact scenarios (cf. also Aikhenvald’s paper). What, in at least one of the languages involved, “goes underground”, only to “bubble up” later, can, just as well as “inherited material”, be the result of an early contact scenario, thus reflecting, instead of “proto-language variation”, variation in the original recipient (and possibly also in the donor) language; the tertium comparationis between both situations is mainly the (observational) absence of contact between the languages involved at the time the parallels can be observed for the first time. This documentational gap is what the notion of “drift” is, in essence, all about, the question, whether relatedness or early intensive contact is the primary reason behind the phenomena is, then, of secondary importance. In the last paper of the collection, J.M. Unger’s “A possible grammaticalization in Old Japanese and its implications for the comparison of Korean and Japanese” (pp. 341-353) discusses, inter alia, the history of the Japanese adjectival auxiliary be—(necessity), which he views as a grammaticalization similar to that which led to modern Korean ya ha—“must” (p. 347).38 He finds parallels to this structure in Lamut and Manchu, where collocations like “VERB-conv, (it is) good/bad/it fits/is OK” are used as equivalents of (positive or negative) necessitative constructions (“one should (not) VERB”). I must admit that I fail to see much structural similarity here, let alone a Tungusic equivalent of the “provisional” element—not to speak of the fact that the Lamut and Manchu constructions certainly do not give the impression of being very old and deeply entrenched features of the languages (on which I may be wrong).39 Consequently, I see little reasons to view any “new perspective” (p. 348) on the debate on the genealogical relationship of Korean and Japanese to each other and/or to Tungusic (and “Altaic”) emerging from this. 37) Allocutivity is defined as a situation, in which an addressee who is not an argument of the verb is systematically encoded in all declarative main clause conjugated verb forms (p. 317); in the general literature, such elements are often described as markers of “politeness styles”. 38) Where ya is glossed as a “provisional” marker. 39) On a different way of expressing “necessitive” semantics in Ewen/Lamut, cf. Malchukov in this volume, pp. 183f. Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 199 Speaking of a “common innovation” (clearly meant in the classical Leskien’ian sense), shared by K and J with two Tungusic languages, is certainly a gross overinterpretation of the facts, the more so, since no kind of material commonality whatsoever is involved in this. I cannot avoid mentioning that Unger’s argumentation brings us back to a stage, in which typological parallelisms/isomorphisms were seen as legitimate reasons for the postulation of common (genealogical) origin. But, luckily, these times have passed, and I can see no argument in favour of any genealogical relationship between any subset of the Macro-Altaic group of languages here. To sum up, this is a very interesting volume, presenting its readers with mostly highly competent studies, which show how the concept of grammaticalization (whether viewed as a “process” or not) can yield very worthwhile and intriguing insights into the history of the Altaic (and other) languages—if, however, rash conclusions are avoided and all possibly available sources of knowledge on the history of the languages involved are given due credit. On the other hand, B. Joseph’s warning (as embodied in the very title of his 2004 paper), that grammaticalization theory (which is a different thing from the study of grammaticalization phenomena) cannot replace “traditional” historical linguistics in any way, should always be borne in mind—the danger of doing “a-historical historical linguistics” in this way is a real one, as some of my critical remarks above hopefully show. And, finally, it should be clear enough that the study of grammaticalization (even if “shared” by several languages) may shed light on a great deal of their history, but certainly not on the question of their possible common origin or lack thereof. References Baskakov, N.A. et al. 1974. Karaimsko-russko-pol’skij slovar’. Moskva: Nauka. Gast, V. and J. van der Auwera. 2010. Scalar additive operators in the languages of Europe. Language 87.1: 2-54. Ganschow, G. 1965. Die Verbalbildung im Ostjakischen. Wiesbaden: Harrassowitz. Genetz, A. 1895. Volmari Porkka’s Tscheremissische Texte mit Übersetzung. Helsingfors: Finnische Literatur-Gesellschaft. Georg, Stefan. 2007. A Descriptive Grammar of Ket (Yenisei-Ostyak), Part 1: Introduction, Phonology, Morphology. Folkestone/Kent: Global Oriental. Georg, Stefan. 2008. Yeniseic languages and the Siberian linguistic area. In: A. Lubotsky, J. Schaeken, J. Wiedenhof (eds.). Evidence and Counter-Evidence. Essays in Honour of Frederik Kortlandt, Vol. 2: General Linguistics, pp. 151-168. Amsterdam and New York: Rodopi. Georg, Stefan. 2011. The Poverty of Altaicism. Paper presented at the symposium “Altaique ou pas?”, Paris/INALCO, 10. December 2011. Pre-publication version available at: http://www.academia.edu/1638942/The_Poverty_of_Altaicism. 200 Reviews / Comptes rendus / Cahiers de Linguistique Asie Orientale 42 (2013) 178-200 Georg, Stefan. (forthc.) Haeretica Altaica. Critical Essays on Altaic Linguistics, with a special focus on the alleged relationship of Turkic, Mongolic, Tungusic, Korean, and Japanese. Johanson, L. 1974. Zur Syntax der alttürkischen Kausativa, ZDMG, Suppl. II, 529-540. Reprinted in: L. Johanson. 1991. Linguistische Beiträge zur Gesamtturkologie, 198209. Budapest: Akademiai Kiadó. Johanson, L. 1990. Subjektlose Sätze im Türkischen. In: B. Brendemoen (ed.). Altaica Osloensia. Proceedings from the 32nd Meeting of the Permanent International Altaistic Conference, Oslo, June 1-16, 1989, 193-218. Oslo: Universitetsforlaget. Joseph, Brian. 2001. Is there such a thing as “Grammaticalization”?, Language Sciences 2001, 163-186. Joseph, Brian. 2004. Rescuing traditional (historical) linguistics from grammaticalizationtheory. In: Fischer, O., Norde, M., Perridon, H. (eds.). Up and Down the Cline— The Nature of Grammaticalization, 45-71. Amsterdam: Benjamins. Kálmán, B. 1961. Die russischen Lehnwörter im Wogulischen. Budapest: Akademiai Kiadó. Matisoff, James A. 1991. Areal and universal dimensions of grammatization in Lahu. In: E. Traugott, and B. Heine (eds.). Approaches to Grammaticalization, Vol. II, 383-454. (Typological Studies in Language 19). Amsterdam: Benjamins. Paasonen, H. 1949. Gebräuche und Volksdichtung der Tschuwassen (E. Karahka and M. Räsänen, ed s.). Helsinki: Suomalais-Ugrilainen Seura. Robbeets, M. 2009. Insubordination in Altaic, Voprosy filologii. Serija “Uralo-altajskieissledovanija” 1.1, 61-79. Sjögren, A.1861. Gesammelte Schriften, Band II, Theil II: Livisch-deutsches und deutschlivisches Wörterbuch. St. Petersburg: Kaiserliche Akademie der Wissenschaften. Solncev, V.M. et al. (eds.) 1986. Materialy sovetsko-v’etnamskoj lingvističeskoj ėkspedicii 1979 goda. Jazyk Lacha. Moskva: Nauka. Ubrjatova, E.I. (ed.). 1982. Grammatika sovremennogo jakutskogo literaturnogo jazyka. Fonetika i morfologija. Moskva: Nauka. Vovin, Alexander. 2011. First and second person singular pronouns: a pillar or a pillory of the ‘Altaic hypothesis’?, Türk Dilleri Araştırmaları 21.2, 251-278. Vovin, Alexander. (forthc.) Why Japonic is not demonstrably related to ‘Altaic’ or Korean. Prepublication available on: http://www.academia.edu/4208284/why_japonic_is_ not_demonstrably_related_to_altaic_or_korean. Wohlgemuth, J. 2009. A Typology of Verbal Borrowings. (Trends in Linguistics. Studies and Monographs 211.) Berlin: Mouton de Gruyter. Žovtobrjuch, M.A. et al. 1980. Istoryčna gramatyka ukraïns’koï movy. Kiïv: Višča Škola. Stefan Georg University of Bonn [email protected]