TY - JOUR
AU - Podobryaev, Alexander
AB - Abstract We offer a solution to a puzzle in the number interpretation of nominals in Buriat. Buriat has a two-way number opposition in morphology (unmarked vs. plural), but semantically, both forms may be number neutral. We show that even though the number neutrality of unmarked nominals is heavily restricted, it does not boil down to (pseudo-)incorporation. Our proposal is that unmarked nominals can be either singular (projecting a NumP) or numberless (lacking a NumP). In case they are singular, they are semantically strictly atomic, but when there are numberless they are truly number neutral, just like the plurals. The plurality inferences of plurals and the consistent number neutrality of numberless nouns are accounted for in a Katzirian system with structurally defined alternatives. 1 INTRODUCTION In a variety of languages, plural DPs like books systematically show non-singularity inferences in some contexts (1-a), but not in others (1-b): (1) Mary sent me Buriat books. (False in a situation where Mary sent me one Buriat book) Mary didn’t send me Buriat books. (False in a situation where Mary sent me one Buriat book) Competition-based theories of plurality derive non-singularity inferences from reasoning about the meaning of the plural form against its singular alternative (Sauerland, 2003; Spector, 2007; Zweig, 2009). A parallel claim has been made about the meaning of singular DPs – in particular, that the non-plurality inferences that singular forms come with (= strictly atomic interpretations) are the result of competition with a plural alternative. This has been suggested for languages with a landscape of nominal number quite different from English (Western Armenian, Bale & Khanjian 2014) – and for English as well (Farkas & Swart, 2010). Competition theories of number face a number of criticisms. First, it’s been noticed that plurality inferences can arise even in the absence of a relevant singular alternative in the language (Ivlieva & Sudo, 2015; Magri, 2011; Sudo, 2017). Second, the mechanism constraining the alternatives that play a role in competition-based meaning enrichments in general (including number) is also currently under debate. In particular, the implementation of a general theory of competition-based enrichments which refers to the structural complexity of alternatives as a constraining mechanism (Fox & Katzir, 2011; Katzir, 2007) has recently been challenged (Romoli, 2013; Swanson, 2010; Trinh & Haida, 2015). We will not try to provide an overview of this debate (see Breheny et al. 2018) or argue for any particular theory. Our goal is more modest. We describe nominal number in a language with two kinds of semantically number-neutral DPs, only one of which can ever be strengthened. As it turns out, the size of the projection of those DPs which cannot be strengthened is smaller than the size of those that can. We take the very existence of such a language – here, Buriat – to be an argument in favor of a competition-based analysis of number inferences, where structural complexity is a factor in how alternatives are constrained. 2 DATA Nouns in the Barguzin dialect of Buriat as spoken in the village of Baragkhan in the Republic of Buryatia of the Russian Federation (henceforth referred to simply as Buriat) show a morphological distinction between two forms: one traditionally referred to as ‘singular’ (morphologically unmarked) – and the other ‘plural’ (hosting an overt plural suffix):1 (2) nom ‘book’ vs. nom-uːd ‘books’ xbː(n) ‘boy’ vs. xbː-d ‘boys’ We focus on the range of number interpretations of morphologically unmarked and morphologically plural forms in different contexts. Let’s start with the unmarked (‘singular’) form. In Buriat, inanimate nouns unmarked for plurality systematically get number-neutral interpretations in various contexts, illustrated here for the direct object position: (3) b|$^{\textrm{j}}$|i nom un ʃ-aː-b I book read-PST-1SG ‘I read a book / books’ The word nom ‘book’ in (3) doesn’t have number marking (or any other marking, for that matter) and, in this sentence, it can refer to one book or to more than one book. This number interpretation of morphologically unmarked inanimate nouns is not restricted to direct object positions – the possibilities include objects of postpositions (4) and nominative subjects (5): (4) b|$^{\textrm{j}}$|i nom-iːn t|$ {u\!\!\!\!-}$|l|$ {o\!\!\!\!-}$|ːxoʒomd-o:-b I book-GEN because.of late-PST-1SG ‘I was late because of the book / books’ (5) nom hon|$^{\textrm{j}}$|in baig-aː book interesting be-PST ‘The book(s) was/were interesting’ Morphologically unmarked animate nouns don’t give rise to number-neutral interpretations – (6), for example, is only compatible with the speaker having seen one boy: (6) b|$^{\textrm{j}}$|i xbːxar-aː-b I boy see-PST-1SG ‘I saw a boy / #boys’ Morphologically plural nouns (both inanimate and animate) in Buriat give rise to non-singularity inferences in upward-entailing (UE) contexts – (7) requires the speaker to be late because of more than one book; in (8) there was more than one book that was interesting: (7) b|$^{\textrm{j}}$|i nom-uːd-iːn tlː xoʒomd-oː-b I book-PL-GEN because.of late-PST-1SG ‘I was late because of the books / #book’ (8) nom-uːd hon|$^{\textrm{j}}$|in baig-aː book-PL interesting be-PST ‘The books were / #book was interesting’ In environments that typically suppress scalar implicatures (such as questions, if-clauses and downward-entailing (DE) contexts), however, the non-singularity inferences of morphologically plural nouns disappear – (9) is false if the speaker has one Buriat book; the question in (10) can get a true positive answer if the addressee has only read one book in Buriat; in (11), one book satisfies the condition: (9) namda bur|$^{\textrm{j}}$|aːd nom-uːd giː I.DAT Buriat book-PL COP.NEG ‘I don’t have Buriat books’ (10) ʃi xezeːʃtaː bur|$^{\textrm{j}}$|aːd-aːr nom-uːd-iːje unʃ-aː-nʃ? you ever Buriat-INST book-PL-ACC read-PST-2SG ‘Have you ever read books in Buriat?’ (11) ʃamda bur|$^{\textrm{j}}$|aːd nom-uːd b|$^{\textrm{j}}$|i: xada-n|$^{\textrm{j}}$|⁠, tede:n|$^{\textrm{j}}$|-iːje asar-a:-raj you.DAT Buriat book-PL COP if-3.POSS they-ACC bring-PST-PRSCR ‘If you have Buriat books (even if you only have one), bring them.’ For the rest of this paper, we will focus on Buriat inanimate nouns and the following generalizations about them: 1) nouns morphologically unmarked for number are semantically number-neutral (their denotation includes both singularities and pluralities); 2) morphologically plural nouns are semantically number-neutral as well (as revealed by DE contexts). To make our reasoning easier, let’s formulate these generalizations against a formal background in which the domain of individuals has the structure of a complete join semilattice (Landman 1991; Link 1983 a.o.). The distinction between atomic (john, bill,...) and sum (john|$\sqcup $|bill,...) subdomains will be crucial. We notate predicates ranging exclusively over atoms as |$P$|⁠, predicates with the whole semilattice as their domain will be |$^* P$| (⁠|$^{\ast}$| being the closure of |$P$| under ‘join’), and predicates ranging over the non-atomic part of the semilattice will be |$^* P$|∖|$P$|⁠. Reformulating our generalization in these terms, both morphologically unmarked and morphologically plural Buriat nouns are |$^* P$| predicates. The puzzle is thus twofold: Why does Buriat have two forms with different number marking have the same number interpretation – |$ P$|? And under what mechanism does one of these forms come to bear non-singularity inferences? Before moving on to approach this puzzle, we will show that the situation in Buriat is different from those described for other languages where morphologically unmarked nouns are semantically number-neutral. 3 TYPES OF LANGUAGES WITH NUMBER-NEUTRALITY Buriat is not the only known language in which nouns unmarked for number exhibit semantic number-neutrality. Similar observations have been made for Turkish (Bale et al., 2010; Öztürk, 2005), Hungarian (Farkas & Swart, 2010), Western Armenian (Bale & Khanjian, 2014) and Hindi (Dayal, 2011). However, Buriat differs from each of these languages either in the properties of the unmarked forms, or in the properties of the plural ones. In most of these languages the distribution of number neutral unmarked nominal forms is very limited. In Hindi (Dayal, 2011), Hungarian (Farkas & Swart, 2003) and Turkish (Öztürk, 2005), they can only be found in the (pseudo-)incorporation construction. It has been argued that such constructions don’t involve full-fledged DPs, and that these forms don’t have an argumental semantic type at all. This DP deficiency has been linked to number-neutrality (Farkas & Swart 2003 a.o.). In Buriat, it would hardly be possible to treat all instances of number neutrality of unmarked forms as cases of pseudo-incorporation. The range of syntactic positions where number neutrality arises in Buriat is greater than what (pseudo-)incorporation is usually assumed to be able to target. Even for direct objects (DOs), which would be the most likely to undergo (pseudo-)incorporation, it is easy to demonstrate that they lack the hallmark properties of the construction, which suggests that they are full-fledged DPs: they are separable from their predicate, can serve as antecedents of discourse anaphora, don’t interact with aspect in a way typical for pseudo-incorporation and can have wide scope w.r.t. other quantificational elements in the sentence (tests following Farkas & Swart 2003; Mithun 2010 a.o.). Here we illustrate the compatibility of unmarked forms with discourse anaphora and with individuating distributive adjuncts, and their interaction with aspect. Discourse anaphora can target unmarked DPs even in DO position, with a number-neutral reading, compatible with a plural referent: (12) b|$^{\textrm{j}}$|i nom unʃ-aː-b. teden|$^{\textrm{j}}$|-iːje b|$^{\textrm{j}}$|bliot|$^{\textrm{j}} $| eka-da tuʃaː-gaː-b I book read-PST-1SG they-ACC library-DAT return-PST-1SG ‘I read the books. I returned them to the library’ Unmarked nominal forms (in DO position and elsewhere) are compatible with distributive adjuncts that individuate each atomic element of the plurality: (13) xojno-xojnohoːn|$^{\textrm{j}}$| b|$^{\textrm{j}}$|i nom unʃ-aː-b one.by.one I book read-PST-1SG ‘I read books one by one’ Dayal (2011) argues that NPs undergoing pseudo-incorporation in Hindi and Hungarian are always specified for number (singular), and the apparent number neutrality is a result of embedding under certain aspectual operators. Crucially for her argument, pseudo-incorporated DOs in Hindi are incompatible with a telic interpetation. In Buriat this generalization does not seem to hold. In (14), the unmarked nom ‘book’ denotes a plurality in a telic clause. (14) glː-g:r xbːn nom unʃ-aː-d bai-gaː morning-INST boy book read-CONV be-PST ‘By the morning, the boy has read the books’ Beyond that, there is extensive evidence that the number neutrality of unmarked nouns in Buriat cannot be reduced to atomicity under aspectual operators. Consider (1), which is ambiguous between an atomic and a non-atomic interpretation of the unmarked noun, with the non-atomic interpretation (hiding between houses) clearly not arising from quantifying over events of hiding in the middle of a house, even if the clause is atelic. (15) badma ger dunda xorgod-oː Badma house middle hide-PST ‘Badma hid in the middle of a house’ ‘Badma hid between houses’ We conclude that unmarked number-neutral nominal forms in Buriat are not (pseudo-)incorporated, and can constitute regular full-fledged argumental DPs. Semantic number-neutrality is thus a property of unmarked argumental DPs in Buriat. Western Armenian may be the only language that has been claimed to have number-neutral NPs unmarked for number in argument positions without (pseudo-)incorporation (Bale & Khanjian, 2014). In this respect, Western Armenian patterns with Buriat; however, there is a crucial difference. As Bale & Khanjian (2014) argue, plural forms in Western Armenian are not semantically number-neutral, but rather range over only the non-atomic part of the domain of entities (⁠|$^* P\backslash P$|⁠).2 In Buriat, on the other hand, there is evidence that the plural forms are number neutral. The number neutrality of plurals becomes evident in non-upward-entailing environments, as in examples (9)-(11) above. It seems that the situation in Buriat is unique: it has a two-way number distinction in morphology (unmarked and plural) and both forms show semantic number-neutrality in argument positions, where unmarked number-neutral forms can’t be brushed off as pseudo-incorporation and plural forms can’t be argued to denote strictly non-atoms. The next section briefly summarizes two big classes of theories of number as candidates for an analysis for this situation. 4 TWO COMPETITION-BASED THEORIES OF NUMBER Competition-based semantic theories of number can be divided into two broad groups – after Farkas & Swart (2010), we call them STRONG SG / WEAK PL theories and WEAK SG / STRONG PL theories. STRONG SG / WEAK PL theories of number offer an analysis for the situation in which morphologically plural forms get a number-neutral (⁠|$^* P$|⁠) interpretation in DE contexts and a strictly plural (⁠|$^* P\backslash P$|⁠) interpretation in UE contexts (the situation in Buriat). According to these theories, morphologically plural forms have both atoms and sums as their domain (=|$^\ast P$|⁠) (Sauerland, 2003; Sauerland et al., 2005; Spector, 2007; Zweig, 2009). To account for non-singular inferences in UE contexts, these theories invoke pragmatic mechanisms which rely on the singular form as an alternative to the plural form (⁠|$P$| vs. |$\ast P$| semantically). Implementations range from Maximize Presupposition (Sauerland, 2003; Sauerland et al., 2005) to scalar implicatures based on the comparative logical strength of |$P$| and |$^\ast P$| alternatives in context (Spector 2007 a.o.). Importantly, all these theories build on singular alternatives being atom-denoting (⁠|$P$|⁠). This is where the Buriat data becomes problematic – in the examples we’ve seen so far, morphologically unmarked forms don’t have exclusively atomic reference. Similar concerns have been raised for the non-singularity inferences of Japanese plurals in Sudo (2017). Under one version of the WEAK SG / STRONG PL approach (Bale & Khanjian, 2014), the domain of singular nouns includes both atoms and sums (⁠|$^* P$|⁠) and plural forms are strictly plural (⁠|$^* P \backslash P$|⁠). Singular forms sometimes – but not always – give rise to non-plurality inferences. They do so when in proper competition with morphologically plural forms. The conditions for such competition are structural. Bale & Khanjian (2014) argue that in Western Armenian, unmarked indefinites are not embedded in a DP, while indefinites marked for plural are. This makes the unmarked indefinites and the plural-marked indefinites too different structurally for competition, so the non-plurality inferences of unmarked forms don’t arise. Unmarked and plural definites, to the contrary, both form DPs and thus compete, which gives rise to non-plurality inferences of unmarked forms. Although we believe that structural properties are crucial for number inferences (see below), two considerations preclude application of this particular theory to our data: 1) Unmarked number-neutral forms do form DPs in Buriat, unlike what Bale & Khanjian (2014) argue for Western Armenian; 2) Buriat plural forms are not strictly non-atomic, but number-neutral. A more complicated version of the WEAK SG / STRONG PL approach (Farkas & Swart, 2010) suggests that the domain of singular nouns (in Hungarian, and potentially more generally) includes both atoms and sums (=|$^\ast P$|⁠), while plural forms are ambiguous between the same and exclusively non-atomic reading (⁠|$^\ast P$|∖|$P$|⁠). This seems promising as it in principle allows for both unmarked and plural forms to satisfy the diagnostics for semantic neutrality (⁠|$^* P$|⁠), as is the case in Buriat. According to Farkas & Swart (2010), plurals are subject to an additional requirement of having sum witnesses in their denotation, precluding them from having exclusively atomic reference. The choice between the |$^\ast P$| and |$^\ast P$|∖|$P$| readings of the plural is regulated by a pragmatic principle (Strongest Meaning Hypothesis), giving rise to non-singularity inferences in UE contexts. Singular DPs under this view are structurally strictly less complex than plural DPs – they lack a layer hosting the privative [Pl] feature. Still, in argument positions singulars and plurals form alternatives, and via this competition a strictly singular reading of non-plurals should arise. This is compatible with the Hungarian data – non-pseudo-incorporated argumental unmarked forms are semantically strictly atomic. However, this is not the situation in Buriat, as shown above – argumental unmarked inanimate DPs are still semantically number-neutral. Apart from this empirical problem, this account has a theoretically dangerous property – the derivation of the non-plurality inference for the unmarked form invokes an alternative that is structurally more complex than the original item. Summing up, existing theories do not account for the data. Either they rely on basic meanings which cannot be maintained for Buriat, or they make wrong empirical predictions, while being potentially problematic in the light of ideas about structural constraints on alternatives. The next section turns to such constraints: first, we formulate them explicitly, relying on Fox & Katzir (2011); Katzir (2007), and then we introduce Buriat data which strengthens the point that structural considerations are relevant for the number interpretation in Buriat. After that, we can formulate our analysis. 5 STRUCTURAL CONSTRAINTS As discussed above, accounts of number inferences of DPs often make use of some mechanism which crucially refers to the set of alternatives of a nominal form. This section discusses a constraint on the set of alternatives, a constraint which has been argued to be active no matter the particular alternative-using mechanism the set is input to (scalar implicature, focus, etc.). Fox & Katzir (2011); Katzir (2007) argue that the ability of a structure to enter the set of alternatives of another structure depends on the relative complexity of these structures. Here is how structural complexity is defined: (16) STRUCTURAL COMPLEXITY (somewhat simplified)Let |$\phi $|⁠, |$\psi $| be parse trees. If we can transform |$\phi $| into |$\psi $| by a finite series of deletions, contractions, and replacements of constituents in |$\phi $| with constituents of the same category taken from the lexicon, then either |$\psi $| is structurally less complex than |$\phi $|⁠, or |$\psi $| and |$\phi $| are of the same structural complexity; we will write |$\psi \lesssim \phi $|⁠. For a structure |$\phi $|⁠, the alternatives will be all those structures that are at most as complex as |$\phi $|⁠: (17) STRUCTURALLY DEFINED ALTERNATIVES Let |$\phi $| be a parse tree. The set of structural alternatives for |$\phi $|⁠, written as |$A_{str}$|⁠, is defined as |$A_{str}(\phi ):= \{ \phi ^{\prime } \ | \ \phi ^{\prime } \lesssim \phi \}$| Motivating examples for this kind of constraint are along the lines of (18) (from Katzir 2007): (18) If any tall man comes to the party, he will be disappointed. If any man comes to the party, he will be disappointed. (18-b) |$\lesssim $| (18-a). The DE context makes sure that the less complex pair is the more informative one. Empirically, (18-a) bears the implicature that (18-b) couldn’t have been used instead. This means that (18-b) |$\in $||$A_{str}$|(18-a), although it is strictly less complex than (18-a) – we can transform (18-a) into (18-b) by deletion. If the context is changed from DE to UE, the more complex structure will be the more informative one. However, empirically, (19-b) does not have the implicature that (19-b) was not assertable – in fact, (19-a) suggests nothing about its (19-b) counterpart, so (19-b) |$\notin $||$A_{str}$|(19-a): (19) A man came to every party. A tall man came to every party. Back to the semantics of number in Buriat – can unmarked and plural forms in Buriat end up in each other’s |$A_{str}$|? Would non-singularity inferences of plural forms then be derivable? What is the structure of the DPs these forms are part of? To answer this, we introduce further data showing that number readings of non-plural forms in Buriat are conditioned morphosyntactically. Configurations that induce strictly atomic reference include DOs with overt ACC case marking, 1&2-person possessive morphology and adjectival modification.3 Data concerning positions requiring oblique cases are less straightforward and we omit them. (20) b|$^{\textrm{j}}$|i {nom-iːje} / {hon|$^{\textrm{j}}$|in nom} un ʃ-aː-b I book-ACC / interesting book read-PST-1SG ‘I read a(n interesting) book’ (#books) (21) {nom-ʃn|$^{\textrm{j}}$|i} / {ulaːn nom} hon|$^{\textrm{j}}$|in baigaː book-2SG / red book interesting was‘Your book / The red book was interesting’ (#books) We were not able to find semantic factors that would give an independent explanation to this generalization. (Lack of) definiteness or genericity doesn’t seem to be one of them. Adjectives like ‘interesting’ or ‘red’ tend to be able to form subkinds when attached to generic nouns, and thus we would expect a DP to retain its number-neutrality in combination with these adjectives, if the said number-neutrality were due to genericity – but, as (20) and (21) show, number-neutrality is lost. Definiteness is a more complicated issue in Buriat, which lacks grammatical marking of definiteness. One could, in principle, suggest that number-neutrality is linked to indefiniteness – possessive morphology and overt accusative marking enforcing a definite environment, in which number-neutrality is lost. However, it’s not clear why 3-person possessivity would be different from 1- and 2-person possessivity, which have a different effect on number interpretation in Buriat. Although we don’t have much to say about (in)definiteness of DOs with overt accusative marking, we have data suggesting that indefiniteness and number-neutrality do not correlate with each other. Speakers report a contrast between (1) with and without the adjective ‘interesting’: (22) b|$^{\textrm{j}}$|i hon|$^{\textrm{j}}$|in nom xudalda-ʒa aba-xa johotoj-b I interesting book sell-CONV take-POT need-1SG ‘I have to buy an interesting book (#books)’ First, (22) can be used with an indefinite interpretation of ‘interesting book’ – any interesting book will do. Second, unlike the corresponding sentence without the adjective, (22) can’t be used in a situation where the requirement is to buy more than one book. Thus, at least it’s not the case that indefiniteness is sufficient for number-neutral interpretations of unmarked forms. We were not yet able to find a reliable example of the opposite – it not being the case that all unmarked definites are strictly atom-denoting. However, even if all unmarked definites turn out to be strictly atom-denoting, this will leave the number-variability of unmarked indefinites a mystery. In any case, the only reliable factor predicting the number-interpretation for unmarked forms in Buriat so far seems to be the presence or absence of elements in particular structural positions, regardless of their semantics. We take these facts to mean that the interpretation of number is sensitive to the size of the DP structure the unmarked noun is part of – extended structure requires atomic semantics. We build our analysis on this suggestion. 6 ANALYSIS We propose that unmarked DPs in Buriat are structurally ambiguous. They can either lack the projection hosting number morphology or have a silent singular morpheme in it (ignoring linearisation): [|$_{DP}$|⁠...[|$_{\sqrt{}}$| nom]] ‘book’ vs. [|$_{DP}$|⁠...[|$_{NumP}$||$\varnothing $| [|$_{\sqrt{}}$| nom]]] ‘book-SG’ (we locate NumP below DP following Farkas & Swart 2010, a.o.). Thus we conjecture that in Buriat, the lack of a NumP layer does not preclude the formation of a DP (unlike, maybe, in some other languages). All DPs with overt plural morphology contain a NumP layer. The interpretations of the three relevant substructures are the following (note that both the form without the number projection and the plural form have number-neutral interpretation): (23) |$[\hskip-1.5pt[ [_{\textrm{NP}} {\it \sqrt{nom}}] ]\hskip-1.5pt] = \lambda x. ^* $|BOOK(x) A corresponding numberless DP: (24) |$[\hskip-1.5pt[ [_{\textrm{NumP}} \varnothing _{\textrm{SG}} {\it \sqrt{nom}}] ]\hskip-1.5pt] = \lambda x. $|BOOK(x) A corresponding singular DP: (25) |$[\hskip-1.5pt[ [_{\textrm{NumP}} {\it{u:d}}_{\textrm{PL}} {\it \sqrt{nom}}] ]\hskip-1.5pt] = \lambda x. ^* $|BOOK(x) A corresponding plural DP: The argument for such a solution comes from the data in (20)-(21) that suggest that extended syntactic structure correlates with the strictly atomic reading of unmarked forms. We think that this has to do with the syntactic requirements of certain elements of DP structures. Namely, we suggest that adjectives, possessive morphology and certain case markers can’t merge in the absence of NumP. Similarly, animate nouns require NumP (while inanimate nouns don’t). We treat such facts as strictly syntactic. The marked/unmarked direct object contrast buttresses this argument. Although in general the DO position can remain unmarked for ACC case (the conditions under which it happens are orthogonal to our point), whenever ACC is present, nouns with non-atomic reference have to host a plural morpheme and nouns not overtly marked for number denote strictly within atoms. This suggests that ACC selects for NumP. Similarly, (20)-(21) suggest that adjectival modification and possessive morphology generally require NumP to be projected. Given the structures and meanings in (23), (24), and (25), we can ask which of them serve as alternatives to each other. [|$_{\textrm{NP}}$||$\sqrt \it{nom}$| ] is strictly the least complex of these structures (the other two can be transformed into it by Num head deletion) – therefore, |$A_{str}$|([|$_{\textrm{NP}}$||$\sqrt \it{nom}$| ]) will be empty. |$A_{str}$|([|$_{\textrm{NumP}}$||$\varnothing _{\textrm{SG}}$| [|$_{\textrm{NP}}$||$\sqrt \it{nom}$|]]) and |$A_{str}$|([|$_{\textrm{NumP}}$||$\it{u:d}_{\textrm{PL}}$| [|$_{\textrm{NP}}$||$\sqrt \it{nom}$|]]) can in principle contain the other two forms, as they are at most as complex – either of the same complexity, or, in the case of [|$_{\textrm{NP}}$||$\sqrt \it{nom}$|], strictly less complex. However, regardless of the entailment properties of the environment, [|$_{\textrm{NP}}$||$\sqrt \it{nom}$| ] cannot be kept as an alternative to [|$_{\textrm{NumP}}$||$\it{u:d}_{\textrm{PL}}$| [|$_{\textrm{NP}}$||$\sqrt \it{nom}$|]]: since they are synonymous, the negation of the sentence containing [|$_{\textrm{NP}}$||$\sqrt \it{nom}$| ] would contradict the original sentence. There is no such dependence in (1-b): (26) |$A_{str}$|([|$_{\textrm{NumP}}$||$\it{u:d}_{\textrm{PL}}$| [|$_{\textrm{NP}}$||$\sqrt \it{nom}$|]]) = {[|$_{\textrm{NumP}}$||$\varnothing _{\textrm{SG}}$| [|$_{\textrm{NP}}$||$\sqrt \it{nom}$|]]} |$A_{str}$|([|$_{\textrm{NumP}}$||$\varnothing _{\textrm{SG}}$| [|$_{\textrm{NP}}$||$\sqrt \it{nom}$|]]) = {[|$_{\textrm{NP}}$||$\sqrt \it{nom}$|], [|$_{\textrm{NumP}}$||$ \it{u:d}_{\textrm{PL}}$| [|$_{\textrm{NP}}$||$\sqrt \it{nom}$|]]} In sum, the plural (⁠|$^*P$|⁠) form invokes the singular (⁠|$P$|⁠) form as an alternative; the singular (⁠|$P$|⁠) form invokes |$^* P$| forms as alternatives. In this way, the problem of non-singular inferences of plural DPs via competition of two |$^\ast P$|-denoting forms does not arise – these forms are not in competition. This system is not very different from English, except for the existence of one more |$^* P$| form as an alternative to the singular one (0-b). This unmarked alternative is not always active: sometimes, using the unmarked form instead of the singular one will result in ill-formedness due to requirements of other elements in the DP (case or possessive morphology or the adjective), but sometimes not. Even in the latter case, the semantic relationship between the source and its alternatives is never |$^* P$| vs. |$^* P$|⁠. We do not argue here for any particular flavour of a STRONG SG / WEAK PL theory deriving non-singularity inferences by some pragmatic mechanism – be it scalar implicature (Spector, 2007; Zweig, 2009) or Maximize Presupposition (Sauerland, 2003; Sauerland et al., 2005). Rather, we point out that underlyingly, the number system in Buriat, although it looks quite exotic, turns out to be surprisingly similar to that of English, and is reducible to it with the help of structurally filtered alternatives.4 This is an example of a syntactic cross-linguistic difference that manifests itself semantically in the available range of interpretations, in this case, in the number domain. It’s a matter of further work to see if a similar analysis could cover further variation in nominal number (see, for example, Sağ 2018 on Turkish). Acknowledgements The data discussed in this paper was collected during a field trip to Baragkhan village, Kurumkansky District, Republic of Buryatia, Russian Federation, in the summer of 2017. The authors thank our language consultants, as well as the Department of Theoretical and Applied Linguistics of Moscow State University for organizing this trip and letting us participate in it. The first author acknowledges support from the Netherlands Organisation for Scientific Research / VENI Grant no. 275-70-045. Thanks go to the audience at the Amsterdam Colloquium 2017, as well as Natalia Ivlieva, Benjamin Spector, Yasutada Sudo, Henriëtte de Swart and Sergei Tatevosov for discussion and comments. Footnotes 1 Buriat has several plural morphemes, each of which comes with non-trivial morphophonological properties. For our current purposes, we treat them as variants of one plural suffix due to the lack of semantic differences between them. 2 Bale & Khanjian (2014) add Turkish and Korean to the class of languages with strictly non-atomic plurals. At least for Turkish, it has since been argued that plural forms are in fact number-neutral (see Renans et al. 2017 for experimental evidence). 3 There is a certain amount of inter-speaker variation in whether adjectival modification precludes number-neutrality. Within our system, it may signal different attachment sites of adjectives in individual grammars. 4 An alternative analysis of our data involving the optionality of a feature rather than the optionality of a projection is, in principle, possible. Under such an analysis, truly singular DPs in Buriat would have a [Sg] feature, plural ones would have [Pl], and number-neutral unmarked ones would lack a number feature altogether. The difference between Buriat and English would be in the optionality of the number feature in the former vs. its obligatoriness in the latter. We do not pursue this analysis here, as spelling out its syntactic and semantic consequences is not trivial. Its immediate semantic disadvantage is that it requires postulating a Horn scale for number features, while our current analysis doesn’t. However, we admit that this alternative analysis, once spelled out, might in principle be a viable alternative. References Bale , A. & Khanjian , H. ( 2014 ), ‘ Syntactic complexity and competition: the singular-plural distinction in Western Armenian ’. Linguistic Inquiry 45 : 1 – 26 . Google Scholar Crossref Search ADS WorldCat Bale , A. , Khanjian , H. & Gagnon , M. ( 2010 ), ‘ Cross-linguistic representations of numerals and number marking ’. In Proceedings of SALT 20 : 1 – 15 . Google Scholar Crossref Search ADS WorldCat Breheny , R. , Klinedinst , N., Romoli , J. & Sudo , Y. ( 2018 ), ‘ The symmetry problem: current theories and prospects ’. Natural Language Semantics 26 : 85 – 110 . Google Scholar Crossref Search ADS WorldCat Dayal , V. ( 2011 ), ‘ Hindi pseudo-incorporation ’. Natural Language and Linguistic Theory 29 : 123 – 67 . Google Scholar Crossref Search ADS WorldCat Farkas , D. & de Swart , H. ( 2003 ), ‘ The semantics of incorporation: from argument structure to discourse transparency ’. CSLI . Stanford, CA . Google Scholar Farkas , D. & de Swart, H. ( 2010 ). ‘The semantics and pragmatics of plurals’. Semantics and Pragmatics3:1–54. Fox , D. & Katzir , R. ( 2011 ), ‘ On the characterization of alternatives ’. Natural Language Semantics 19 : 87 – 107 . Google Scholar Crossref Search ADS WorldCat Ivlieva , N. & Sudo , Y. ( 2015 ), ‘ Another problem for alternative-based theories of plurality inferences: the case of mass plurals ’. Snippets 29 : 7 – 8 . Google Scholar Crossref Search ADS WorldCat Katzir , R. ( 2007 ), ‘ Structurally-defined alternatives ’. Linguistics and Philosophy 30 : 669 – 90 . Google Scholar Crossref Search ADS WorldCat Landman , F. ( 1991 ), Structures for Semantics . Kluwer . Berlin . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Link , G. ( 1983 ), ‘ The logical analysis of plurals and mass terms: a lattice-theoretical approach ’. In R. Bauerle, C. Schwarze and A. von Stechov (eds.), Meaning, Use and Interpretation of Language . De Gruyter . Berlin . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Magri , G. ( 2011 ), ‘ The plurality inference of object mass nouns ’. Snippets 24 : 9 – 10 . Google Scholar OpenURL Placeholder Text WorldCat Mithun , M. ( 2010 ), Constraints on Compounding and Incorporation. 37–56 . John Benjamins . Amsterdam . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Öztürk , B. ( 2005 ), Case, Referentiality and Phrase Structure . John Benjamins . Amsterdam . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Renans , A. , Tsoulas , G., Folli , R., Ketrez , N., Tieu , L., de Vries , H. & Romoli , J. ( 2017 ), ‘ Turkish plural nouns are number-neutral: experimental data ’. In A. Cremers (ed.), Proceedings of the 21st Amsterdam Colloquium . 365 – 74 . Romoli , J. ( 2013 ), ‘ A problem for the structural characterization of alternatives ’. Snippets 27 : 14 – 5 . Google Scholar OpenURL Placeholder Text WorldCat Sağ , Y. ( 2018 ), ‘ The semantics of Turkish numeral constructions ’. In U. Sauerland and S. Solt (eds.), Proceedings of Sinn und Bedeutung 22, vol. 2, ZASPiL 61, ZAS, Berlin. 307 – 324 . OpenURL Placeholder Text WorldCat Sauerland , U. ( 2003 ), ‘ A new semantics for number ’. In R. B. Yound and Y. Zhou (eds.), Proceedings of SALT 13 . Cornell Linguistics Club. Ithaca , NY . 258 – 75 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Sauerland , U. , Andersen , J. & Yatsushiro , K. ( 2005 ), ‘ The plural is semantically unmarked ’. In S. Kepser and M. Reis (eds.), Linguistic Evidence . Mouton de Gruyter . 413 – 34 . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Spector , B. ( 2007 ), ‘ Aspects of the pragmatics of plural morphology: on higher-order implicatures ’. In U. Sauerland and P. Stateva (eds.), Presuppositions and Implicatures in Compositional Semantics . Palgrave-Macmillan . New York . 243 – 81 . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Sudo , Y. ( 2017 ), ‘ Another problem for alternative-based theories of plurality inferences: the case of reduplicated plural nouns in japanese ’. Snippets 31 : 26 – 8 . Google Scholar OpenURL Placeholder Text WorldCat Swanson , E. ( 2010 ), ‘ Structurally defined alternatives and lexicalizations of xor ’. Linguistics and Philosophy 33 : 31 – 6 . Google Scholar Crossref Search ADS WorldCat Trinh , T. & Haida , A. ( 2015 ), ‘ Constraining the derivation of alternatives ’. Natural Language Semantics 25 : 249 – 70 . Google Scholar Crossref Search ADS WorldCat Zweig , E. ( 2009 ), ‘ Number-neutral bare plurals and the multiplicity implicature ’. Linguistics and Philosophy 32 : 353 – 407 . Google Scholar Crossref Search ADS WorldCat © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com
TI - Plurality in Buriat and Structurally Constrained Alternatives
JF - Journal of Semantics
DO - 10.1093/jos/ffz017
DA - 2020-02-14
UR - https://www.deepdyve.com/lp/oxford-university-press/plurality-in-buriat-and-structurally-constrained-alternatives-cQU1UzsKKc
SP - 117
EP - 128
VL - 37
IS - 1
DP - DeepDyve
ER -