Abstract

The theoretical aim of this article is to integrate the singulative into the theory of division proposed by Borer (2005) and other theoretical linguists (e.g., Krifka 1995, Doetjes 1996, 1997, Chierchia 1998, Cheng and Sybesma 1999). To illustrate my claim, I offer a brief case study of Ojibwe, an Algonquian language, which I argue uses gender shift (from inanimate to animate) to mark singulativization. Singulatives, as morphological markers, are primarily known from Celtic, Afro-Asiatic, and Nilo-Saharan languages, but are not a known feature of Algonquian languages. Further support for my claim that the grammar of Algonquian languages embeds a singulative system comes from Fox (Mesquakie).

Borer (2005) proposes that all nouns in all languages are in need of being portioned out before they can interact with the count system. In Chinese, this is achieved by (count) classifiers for count nouns and (mass) classifiers for mass nouns (Cheng and Sybesma 1999).1 In English, plural marking takes the role of classifiers for count nouns, and measure phrases are used in lieu of mass classifiers (Sanches and Slobin 1973, Doetjes 1996, 1997).2 For singular individual nouns in English such as cat in one cat, the numeral one functions both as a divider and as a counter. In Hungarian, Tagalog, and Turkish, this is generalized to all numerals, in which case it is possible for the noun to be free of plural marking (Borer 2005).

I will show that atomization3 for count nouns in Ojibwe (an Algonquian language)4 can proceed exactly as in English, namely, by way of the plural. However, in the absence of many measure phrases or productive mass classifiers, Ojibwe uses a singulative system (or the remnants thereof) to perform division that yields unit-of-measure readings (from mass nouns). As is the case in other languages with the singulative, atomization via the singulative in Ojibwe is also used to create individuals from collectives.

The idea that the singulative performs the same function as a classifier goes back to Greenberg (1972),5 who states that ‘‘[t]he classifier is an individualizer which performs the same function as a singulative affix in languages with the collective/singulative opposition’’ ( p. 26).6 Reversing this claim (a singulative affix performs the same function as a classifier) and adding the singulative system to the theory of division is only a logical step that makes the present proposal a very natural one.

The degree to which the content of Div0 (the locus of individuation according to Borer (2005)) varies is consequently higher than previously thought (see Ritter and Wiltschko 2009 for the idea that the content of functional categories can vary crosslinguistically). The plural, numeral classifiers, and atomizing numerals are all different flavors Div0 can take, but we must now add the singulative. I will show that the content of the singulative itself varies: in Ojibwe, it comes in the form of gender shift or in the form of the diminutive. Crosslinguistically, gender shift comes in two subflavors: shift from masculine to feminine in Breton, Welsh, Somali, and Arabic, but shift from inanimate to animate in Algonquian.7 (1) summarizes the different flavors that Div0 can take.

(1)

  • a.

    graphic

  • b.

    Possible content/flavors of Div0

    A = plural (Number system 1)

    B = numeral classifier (Number system 2)

    C = numerals (Number system 3)

    D = singulative (Number system 4)

    a = gender shift

    b = diminutive

Since there is evidence that singulative forms can be pluralized, my final claim will be that division is not the sole function of the plural (as in Borer 2005); it can also simply be used as a counter, as taken for granted by traditional grammars and common wisdom. The plural thus has a dual role in the grammar.

In tackling issues related to the count/mass distinction (e.g., the role of number in aboriginal languages or the derived versus default nature of the plural), this article contributes to many current issues in linguistics.

Section 1 incorporates the singulative into Borer’s (2005) system of division and discusses known languages that have the singulative as part of their grammars. Section 2 introduces the basic facts about Ojibwe. It also introduces a puzzle: why Ojibwe mass nouns can systematically be pluralized, a problem whose solution will be connected to the singulative. Section 3 gives a full analysis of the Ojibwe singulative, focusing on gender shift from inanimate to animate. Section 4 introduces another means by which the singulative in Ojibwe can be realized: the diminutive. Section 5 concludes.

1 The Singulative

The singulative is a process by which a collective or a mass noun (of a certain kind, depending on the language) is turned into a unit. It is common crosslinguistically for gender shift to mark the singulative. For example, languages like Russian that have remnants of a singulative system (which was once fully productive; Greenberg 1972) use a shift from masculine to feminine by way of a suffix on a mass noun to create a unit-of-measure reading (lyod (MASC) ‘ice’ ~ l’dina (FEM) ‘block of ice’) or a simple unit reading (grad (MASC) ‘hail’ ~ gradina (FEM) ‘hailstone’). Hebrew also has remnants of a singulative system whereby a mass noun that is masculine is turned into an individual marked as feminine: se’ar (MASC) ‘hair’ ~ sa’ar-a (FEM) ‘hair’ ~ sa’arot ‘hairs’ (Doron and Müller 2011). In that language, it is also possible to shift from feminine to masculine: alv-a (FEM) ‘foliage’ ~ ale (MASC) ‘leaf ’ ~ al-im (MASC) ‘leaves’.

In Breton (a Celtic language, still spoken in some western parts of France), nouns that refer to collections or masses can be portioned out with the help of a feminine suffix, -enn, yielding singulative nominals.8 The feminine is marked and used to distinguish count from mass/collective. Collective nouns, as in (2), are semantically plural, but morphosyntactically singular (examples from Stump 2005:62). The collective is typically used to denote a kind (in contrast, the singulative term is never associated with kinds).

(2)

graphic

The resulting singulative form can in turn often be pluralized, as in (3). This shows that the singulative first performs the atomizing operation and that the plural is added only afterward. The plural does not itself perform the dividing function. This, a priori, poses a challenge for Borer’s (2005) theory of division; but as we will see, there is a natural solution to this problem. Once we take seriously the idea that the singulative can be a divider, like classifiers and the plural, then we must come to terms with the idea that the role of the plural in this instance is simply that of counting rather than dividing.

(3)

graphic

The singulative by way of the feminine morpheme -enn also targets mass nouns; see (4). These examples (from Trépos 1980:67) show that the function of the singulative consists not only in turning abstract object types (collectives) into identifiable objects, but also in picking discrete entities out of an undifferentiated mass. The mass nouns in (4) are all masculine; the singulative nouns are all feminine.

(4)

graphic

Division by the singulative can also be performed on substances and some liquids, as shown in (5) and (6), respectively ((5a–c) and (6a–b) are from Trépos 1980:67; (5d) is from Ternes 1992: 416).9

(5)

graphic

(6)

graphic

Although -enn joins very freely with collective nouns (see (2)), it is apparently much more sporadic in its combinations with mass nouns (however, note that in other singulative languages, such as Classical Arabic, not all collectives are the target of the singulative either; Greenberg 1972). In addition, as pointed out by Acquaviva (2008:245), when singulatives are derived from mass nouns, the meaning/value of the massifier varies with the word (piece of ice, but blade of grass and drop of rain). This state of affairs is widespread in languages where the singulative is available. In Syrian Arabic, for example, singulatives are formed by suffixing -e/-a, both feminine forms.10 The translated measure word varies according to the noun that is used (here, ‘grain’ or ‘blade’). Examples are from Cowell 2005:298.

(7)

graphic

The singulative is not productive with all mass nouns. In Syrian Arabic, a few mass nouns designating plants, for example, either have no unit derivative at all, or have one that appears very rarely. In such cases, a periphrastic phrase is used instead. An example from Cowell 2005: 298 is tūm ‘garlic’ ~ rās tūm ‘a garlic bulb’ (rās means ‘head’).11

The incomplete productivity and the shifting meaning of the singulative are sometimes taken as a sign that although clearly an inflectional process (see footnote 13), it nevertheless has many derivational properties (see, e.g., Acquaviva 2008). However, I wish to abstract away from the inflectional/derivational distinction (following Distributed Morphology; inflectional vs. derivational processes share many properties) and I follow Booij (1993, 1995), who shows that inflectional and derivational morphology are not two different kinds of morphology, but two different processes: inflectional morphology, in particular, can be used contextually—purely syntactically—or inherently. I therefore assume that (a) the absence of certain singularized mass nouns is simply a reflection of accidental gaps, and (b) the exact meaning of singularized nouns is obtained through context (contextual determination of meaning/compositionality being a property of syntax).

That the context is important for the interpretation of singulativized nouns is clear from the case of nouns that denote substances (see (5)). Sometimes the meaning is the measure reading (‘a piece of’, ‘a blade of’, etc.); sometimes it is the simple unit meaning (‘a lighter’, ‘a glass’). Sometimes the translation of the word varies according to the source: gwer ‘steel’ in (5c) is sometimes translated as ‘piece of steel’ in online dictionaries.

The important point is that the singulative performs division: whether it yields a measure reading or a simple unit reading, we obtain an individual. I propose that regardless of the obtained interpretation, the singulative is encoded in Div0 (the number head proposed by Borer (2005)). There is no need to postulate an additional projection for the measure reading (see footnote 19). This idea is consistent with the facts.

The singulative is an alternative way to create division in the grammar. While classifiers and the plural are well-studied mechanisms for creating individuation (Krifka 1995, Doetjes 1996, 1997, Chierchia 1998, Cheng and Sybesma 1999, Borer 2005), the singulative has not been previously considered a divider in generative grammar. I propose that like classifiers and the plural, a singulative morpheme originates in the head Div0 and that it is in that position that it creates an individual. By way of illustration, (8) gives the derivations for Breton geot ‘grass’, from a mass noun (8a) to a singulative noun (8b).12

graphic

(8)

graphic

The process is productive and syntactic in the sense that it occurs in exactly the same syntactic contexts as the choice between an ordinary singular noun and its plural counterpart (Stump 2005).13

Additional evidence for the idea that the singulative resides in a unique head Div0 and that it is in complementary distribution with the plural comes from Dagaare, a Gur language spoken in Ghana and Burkina Faso. This language has a type of number marking known as ‘‘inverse’’ or ‘‘polarity’’ number marking (Baerman 2007), whereby the same element marks singularity or plurality depending on the syntactic context. Grimm (2009) shows that the marker -ri marks the plural for nouns denoting individual objects (child, dog), while for collective nouns (seeds, insects), -ri marks the singular. I assume that in the first case, -ri is the regular plural acting as a divider (under Div0), but that in the second case, -ri is used as a singulative (also under Div0). This shows clearly that the plural is not alone in performing division and that, importantly, in some languages the plural and the singulative are one and the same morpheme (examples from Grimm 2010: 169).14

(9)

graphic

Interestingly, Dagaare also possesses a singulative marker -ruu, which is primarily restricted to granular mass terms (‘pepper’, ‘straw’, ‘grass’), designating ‘a piece of’. The use of -ri as a singulative is reserved for collective aggregates, such as vegetation, insects, or inherently plural body parts. In this language, it is the form of the singulative, rather than two different positions, that distinguishes the simple unit from the unit-of-measure reading. Examples are from Grimm 2010:176.

(10)

graphic

Finally, let me briefly discuss the case of liquids and substances. Grimm (2009) argues that substance and liquid terms in Dagaare are not the target of any individuating operations. He argues that Dagaare is like Welsh in this respect (like Breton, Welsh has a singulative system) and that both languages pattern with English in this regard. Table 1 (adapted from Clausen et al. 2010) summarizes the crosslinguistic variation, adding Breton.

Table 1.

Kinds of nouns targeted by the singulative crosslinguistically

LanguageLiquids/SubstancesGranular aggregatesCollective aggregatesIndividual objects
English plural (-splural (-s
Breton singulative (-ennsingulative (-ennplural (-
Welsh singulative (-ynsingulative (-ynplural (-od
Dagaare singulative (-ruusingular (-riplural (-ri
LanguageLiquids/SubstancesGranular aggregatesCollective aggregatesIndividual objects
English plural (-splural (-s
Breton singulative (-ennsingulative (-ennplural (-
Welsh singulative (-ynsingulative (-ynplural (-od
Dagaare singulative (-ruusingular (-riplural (-ri

This cannot be correct, however. This is because as the examples in (5) and (6) show, it is possible in Breton for substance and liquid terms to be targets of the singulative operation. In fact, in other singulative languages, it is possible for substance and liquid terms to be inputs to singulativization. For example, in Classical Arabic, ‘wood’ and other nongranular aggregates such as ‘mud’ and ‘butter’, as well as liquids such as ‘buttermilk’ and ‘wine’, can be targets of the singulative.15 ((11a) is from Greenberg 1972:21; (11b–e) are from Saleh AlQahtani, pers. comm.)

(11)

graphic

Table 2 is a revised version of table 1. It now includes both Breton and Arabic. What is clear is that not all singulative languages are the same: some allow substances and liquids to be singulativized, some either substances or liquids, others neither. In section 3, we will see that Ojibwe tolerates the singulativization of substances, but not of liquids.16

Table 2.

Revised version of table 1. (sing = singulative; sgl = singular)

LanguageLiquidsSubstancesGranular aggregatesCollective aggregatesIndividual objects
English plural (-splural (-s
Breton sing (-ennsing (-ynsing (-ennsing (-ennplural (-
Welsh sing (-ynsing (-ynsing (-ynsing (-ynplural (-od
Dagaare sing (-ruusgl (-riplural (-ri
Arabic sing (-ahsing (-ahsing (-ahsing (-ahplural (-t
LanguageLiquidsSubstancesGranular aggregatesCollective aggregatesIndividual objects
English plural (-splural (-s
Breton sing (-ennsing (-ynsing (-ennsing (-ennplural (-
Welsh sing (-ynsing (-ynsing (-ynsing (-ynplural (-od
Dagaare sing (-ruusgl (-riplural (-ri
Arabic sing (-ahsing (-ahsing (-ahsing (-ahplural (-t

A natural consequence of the idea that the singulative is a divider is that the plural of a singulativized noun is not a divider, but something else. The question is, what is it?

I want to propose that it is in this instance simply a counter. The plural is thus not exclusively tied to division. This follows naturally from the idea that the singulative is used for division and thus is generated under Div0. We know independently that the plural has other uses: the plurals of modesty and abundance, and the exaggerative, hyperbolic, approximative, and antiassociative plurals (Corbett 2000). The counting plural is different from these, though, since it is completely systematic and not tied to the meaning of the noun (see footnote 29).

That the plural does not have a unique denotation has been proposed by Wiltschko (2008): on her account, in some languages number is a functional head (#, as head of #P) while in others it is a modifier. Another way to express this idea is to claim that the plural is sometimes inflectional and sometimes derivational. In fact, the interpretation we get for a pluralized singulative is exactly that of counting. If in cases such as Breton (12) the plural had a dividing function, then we would expect the derivation geot-où-enn, but this is completely ill-formed. The counting interpretation is exactly the one we obtain: geotennoù in (12) means ‘more than one blade of grass’.

(12)

graphic

It also follows from this proposal that numerals like ‘one’ in Breton and other singulative languages are not always dividers. For Borer (2005), the numeral ‘one’ is special in that it acts both as a divider and as a counter in English and similar languages. However, if division is already performed by the singulative, the counter element ‘one’ must simply be a counter when used in combination with a singulative. However, when used with an ordinary singular noun (in the absence of the singulative), it does act both as a counter and as a divider.

In summary, in singulative languages, such as Breton, the singulative can target collective nouns and mass nouns with the same kind of effect (individualization of mass and collective terms). The singulative was incorporated into the theory of division and was argued to be a particular flavor of Div0.

2 The Basic Ojibwe Facts and a Puzzle

On the basis of the fact that in Ojibwe, words such as mkwam ‘ice’ can be pluralized (mkwamiig ‘ice-PL’) and that mkwam can equally mean ‘ice’ (mass), ‘a piece of ice’ (measure unit), or even ‘icicle’ (individual), it has been argued in the literature on Algonquian that there is no grammaticized count/mass distinction in Ojibwe (Rhodes 1990:153; see also Corbett 2000:87).17

While it is admittedly common in Native North American languages for plural marking to target all sorts of nouns (count or mass) and/or for plural marking and agreement in general to be completely optional (Whorf 1941 for Hopi, Davis and Matthewson 1999 for Lillooet Salish, Wiltschko 2008 for Halkomelem Salish), and while it might be tempting to conclude that the absence of a grammaticized count/mass distinction is a pervasive feature of aboriginal languages, it is not the case that all Native North American languages lack such a distinction. For example, as Mithun (1988) points out, all nouns in Taos, Kiowa, Zuni, and the Algonquian languages are inflected for number. In addition, the Ojibwe examples in (13) show that subject and object nominals have to agree with the verb.18

(13)

  • graphic

    1SG-PAST-see.3PL men.3PL

    ‘I saw men.’

  • b.

    graphic

    men.3PL 1SG-PAST-see-3PL

    ‘Men saw me.’

As seen in (14), the opposition between singular and plural cuts across the paradigms of nouns, pronouns, and verbs. Agreement is obligatory with all these categories. (Here and later, IN = inanimate, AN = animate.)

(14)

graphic

Number is therefore clearly inflectional in Ojibwe (see Mathieu 2012 for details): it is not a modifier (to a nominal root; i.e., derivational) but a functional head (see Wiltschko 2008).

The puzzle is thus as follows: why is it possible in some cases for mass nouns in Ojibwe to be pluralized (e.g., mkwamiig ‘ice-PL’) or for singular mass terms to denote individuals (e.g., mkwam ‘ice/piece of ice/icicle’) and even appear with numerals (e.g., bezhig mkwam lit. ‘one ice’)?19 The answer, I will argue, is related to the fact that Ojibwe lacks productive massifiers and certain measure phrases, but also more generally to the fact that the grammar embeds a singulative system, albeit not always a visible one.

Although Ojibwe admittedly has some massifiers of the Chinese kind in its grammar, their use has diminished in recent years (Valentine 2001:502). They attach to numerals to indicate measure units: -aatig is used for wooden, pole-like elements; -eg for cloth-sheet-like elements; -aabik for metal, glass, plastic, or stone; and -aabiig for string-like elements. Some examples appear in (15).

(15)

  • a.

    niizh-waatig mishi

    two-CL firewood

    ‘two sticks of firewood’

  • b.

    niizh-weg zenibaa

    two-CL silk

    ‘two pieces/sheets of silk’

Other classifiers include -naagans ‘cupful’, -ooshkin ‘bagful’, -oonag ‘boatload’, -baneninj ‘handful’, -sag ‘barrelful’, but some of these are listed as rare in dictionaries (e.g., ‘boatful’ in Rhodes 1985) and they appear to be no longer productive.20 Two examples appear in (16).

(16)

  • a.

    niizh-naagans ziisbaakwad

    two-CL sugar

    ‘two cupfuls of sugar’

  • b.

    niizh-ooshkin mnoomin

    two-CL rice

    ‘two bagfuls of rice’

Ojibwe not only lacks productive mass classifiers of the Chinese kind, but also appears to lack measure phrases of the English kind (e.g., ‘piece of’, ‘bit of’, ‘portion of’).21 As Greenberg (1972:16) points out, ‘‘[T]here are a considerable number of Amerind languages . . . which do not have measure constructions [including Ojibwe; Greenberg 1974]. Numerals occur directly both with nouns designating mass as well as countable objects.’’

The main empirical fact introduced in this article is that in the absence of measure phrases such as ‘piece of’, ‘bit of’, ‘portion of’, Ojibwe (and Fox) uses a singulative system to denote units of measure. The same mechanism is used to create individuals out of collectives. While the Ojibwe (and Fox) singulative marker is the animate form -a, the collective and mass nouns that are the input to the singulative are inanimate forms ending in -i.

Ojibwe, like other Algonquian languages, has a productive gender system based on animacy. Some nouns are animate; others are inanimate. The animate/inanimate contrast is grammaticized (like the masculine/feminine gender system of, say, Romance languages): for example, some types of berries are inanimate while others are animate, some body parts are inanimate while others are animate.

A reviewer asks whether animacy and gender are really the same in Ojibwe, citing Wiltschko (2012), who claims on the basis of Blackfoot that animacy is not gender, but nominal aspect. For Ojibwe, the answer is definitely yes: animacy in Ojibwe is unquestionably gender and not nominal aspect. Following Rijkhoff (1991), Wiltschko makes an analogy between the verbal and the nominal domain in terms of boundedness. Bounded events have an inherent endpoint/culmination, beyond which the same event cannot continue (e.g., walk to the store), while unbounded events have no inherent endpoint/culmination: the same event can continue over an indefinite period of time (e.g., play cards). For the nominal domain, the idea is that mass nouns are unbounded (they are not collections of individuals), while count nouns are bounded (they form collections of individuals). Wiltschko’s (2012) idea is that English has a clear grammaticized count/mass distinction; therefore, nominal aspect is instantiated by a bounded versus unbounded distinction. On the other hand, according to Wiltschko, Blackfoot does not have a grammaticized count/mass distinction (Wiltschko cannot find examples of mass terms that cannot be pluralized in that language). This means that the language lacks a nominal aspect instantiated via a bounded versus unbounded contrast. Instead, in Blackfoot nominal aspect is instantiated via animacy. Compare (17a) with (17b).

(17)

graphic

To support her claim, Wiltschko uses the fact that in Blackfoot, verbs have different forms depending on whether the participants are animate or inanimate (see, e.g., (23) in Ojibwe where waabam ‘see’ = AN and waabnd ‘see’ = IN). She argues that in Blackfoot, Aktionsart is thus not based on [+/−bounded]/telicity; rather, it is based on [+/−animate]. She also argues specifically that animacy in Blackfoot is not like gender. She shows that in German nominalizers are specified for gender but that in Blackfoot they are not: German die Grammat-ik ‘the grammar’ is feminine, whereas der Lehr-er ‘the teacher’ is masculine; the Blackfoot nominalizer a’tsis is used for both animate nouns (saa’kssoya-a’tsis ‘poison ivy’) and inanimate nouns (isoohkama-a’tsis ‘container’).

In Ojibwe, the count/mass distinction is clearly grammaticized, as I have shown. For example, not all mass nouns can be pluralized: liquid terms resist pluralization. Also, while it is true that verb stems in Ojibwe end in either animate or inanimate forms, this is completely independent of boundedness. For example, the Ojibwe verb ‘be’ has two forms: one for animate nouns, -wi (18a), and another for inanimate nouns, -wan (18b) (see Valentine 2001). As Valentine (2001) points out, verbs such as these can be interpreted not only as ‘be’ but also as ‘become’, where a change of state is clearly asserted.

(18)

  • a.

    Ninii-wi.

    man-be

    ‘He is a man.’

  • b.

    Oodenaa-wan.

    town-be

    ‘It is a town.’

More generally, Ojibwe verbal expressions such as ‘walk to the store’ are bounded while expressions such as ‘play cards’ are not. As in English, telicity is therefore a concept very much anchored in the grammar of Ojibwe.

Finally, while it is true that Ojibwe nominalizers are indifferent to whether the noun they merge with is animate or inanimate (e.g., the nominalizer -gan can appear with either), it is not clear how this is relevant to distinguishing languages with a grammaticized count/mass distinction from languages without one. In conclusion, there is no reason to treat animacy in Ojibwe as (nominal) aspect instead of the more traditional notion of gender. Everything points to the view that animacy is gender in Ojibwe.

Let us now turn to gender shift in Ojibwe. While it is well-known that it is possible in Algonquian languages to shift gender from inanimate to animate as a mark of elevated status, especially in storytelling (a case of semantic shift (Black-Rogers 1982, Valentine 2001:118) or perspectival marking (Muehlbauer 2008)),22 it is less well-known that it is also possible to shift gender from inanimate to animate as a way to divide what would otherwise be a mass or collective noun. In this case, a unit is created, either of a simple kind as in (19) (a simple unit reading) or of a measure kind as in (20) (a unit-of-measure reading). The examples are from Fox (Mesquakie) (see Goddard 2002); -i is the inanimate ending on nouns while -a is the animate ending. Goddard (2002) does not correlate these facts with the notion of the singulative, but I believe this is exactly what is at work here.23

(19)

graphic

(20)

graphic

Once singularized, nouns such as zhooniyaaha (AN) ‘silver, money’ can be pluralized (with an animate plural form), giving zhooniyaaha-ki (AN) ‘coins, bills, money’. The same is true for miichipehi (IN) ‘game (collective)’ → miichipeha (AN) → miichipeha-ki (AN) ‘game animals’.

It is evidently difficult to notice the gender shift that is the spell-out of the singulative in Ojibwe. This is because most singular inanimate nouns in Ojibwe have lost their final -i and most animates have lost their final -a. There are, nevertheless, residual visible effects of this process in modern Ojibwe, as I will show.

First, let me point out that there is evidence, as Piggott (2007) argues, that number is present in the derivation of every Ojibwe noun. Each of the singular forms in (21a–c) ends in a vowel that is demonstrably not part of the exponent of the root morpheme. For example, the root allomorphy in (21c) [mi∫ → mis] results from a palatalization process (s → ∫) that only applies in a derived environment (Kaye and Piggott 1973). This means that there is a singular suffix -i that attaches to inanimate nouns and a counterpart -a that attaches to animates.

(21)

graphic

Since singular number and gender are fused morphologically, it is reasonable to assume that gender is also present in the derivation of every Ojibwe noun. In words where no final -i or -a surfaces, it is assumed that the vowel has been truncated. However, the vowel can only be truncated if the word meets minimality requirements. If the word is too small (e.g., if it is bisyllabic, as in (21)), the vowel cannot be truncated. As Piggott (2007) argues, the process is therefore systematic and predictable.

For most singular nouns, however, especially those that end with a consonant, it is impossible to tell whether they are animate or inanimate from the endings, creating a situation where the singulative is not morphologically visible. In addition, because some words have lost their final -i together with final consonants, inanimate nouns can end in -a, which is the mark for animates rather than inanimates (and vice versa).

With this in mind, take the case of the collective noun zhoonya ‘money’ in (22a). It is listed in dictionaries as an inanimate noun (although it ends in -a; in Fox, as shown in (19a), it clearly ends in -i); but interestingly, in the plural it is listed as animate. The same is true for mitig in (22b): it is listed in dictionaries as inanimate when it is interpreted as ‘wood/forest’ but as animate when it means ‘tree’.24 Although there is no morphological surface difference between mitig ‘wood’ (IN) and mitig ‘tree’ (AN), the change from inanimate to animate is a clear residual effect of the gender shift correlated with the singulative use: from a collective we obtain a simple unit reading via gender shift.

(22)

graphic

To see that nouns such as zhoonya ‘money’ and mitig ‘tree’ are animate on their individual reading, we can consider them in a sentence. Since Ojibwe has two kinds of verb stems, transitive animate (TA) and transitive inanimate (TI), the verb stem selected will depend on the gender of the noun. Moreover, Ojibwe distinguishes between two third persons in a sentence or a narrative by means of obviation. In the sentence ‘John saw Fred’, for example, there are two third persons, ‘John’ and ‘Fred’. In such a case, one of the third persons is seen as primary and is called proximate (as if it were somehow closer to the interest of the speaker); the other one is seen as secondary and is called obviative. In Ojibwe, the obviative is marked only on animate nouns. In (23a), makwa ‘bear’ is animate; it takes the obviative form -n and the verb stem is TA. In (23b), jiimaan ‘boat’ is inanimate; it takes no obviative form and the verb stem is TI.

(23)

graphic

Turning back to zhoonya ‘money’ and mitig ‘tree’, we see in (24a) that zhoonya ‘money’ is obviative when interpreted as ‘coin’, and that in this case the verb stem is TA. The same is true for mitig when interpreted as ‘tree’; see (24b).

(24)

graphic

Turning now to mass nouns that can be turned into units of measure, consider the following examples. The plural versions of the animate mass nouns in (25) are all interpreted as unit-ofmeasure nominals ( pieces of x, portions of x, blades of x—depending on the noun).25

(25)

graphic

We can now see why mass terms can be systematically pluralized in Ojibwe. It is not because the language lacks a grammaticized count/mass distinction, but because in these cases, a gender shift process has operated on the singular to give a plural that is of a different gender from the original mass noun (this is typical of singulative systems). We can also understand why mikwam can mean ‘ice’ or ‘piece of ice’. There is a gender shift from mikwam as mass noun to mikwam as unit of measure, although it is no longer spelled out morphologically.

This process is different from so-called coercion in English. As is well-known, although mass nouns cannot normally be pluralized in English, exceptions to that generalization are possible provided that the interpretation of a mass noun is coerced to that of a kind (26a) (via the Universal Sorter) or a standard serving (26b) (via the Universal Packer; see Bunt 1985).

(26)

  • a.

    There are only three waters available (still, sparkling, and flavored).

  • b.

    Bill ordered three waters (i.e., glasses, bottles, etc.).

Count nouns can also be made mass (see David Lewis’s Universal Grinder)—for instance, There was dog all over the road (example from Pelletier 1979). It has been argued that the pervasiveness of these effects suggests that count/mass status is not tied to a lexical item but is instead rather free (Pelletier 1979).

In Ojibwe, pluralization of mass nouns is not free; rather, it is dependent on syntactic structure. First, pluralized mass nouns are never interpreted as kinds (Rhodes 1990:153–154). Speakers reject the direct translations of ‘What sugars do you have?’ (in the context of, say, entering a grocery store: castor sugar, brown sugar, Turbinado sugar, etc.).26 This follows from the fact that in a singulative system, it is the nonindividuated term (the inanimate in Ojibwe) that denotes a kind. This is the situation we find in Breton and other singulative languages where collectives (and mass terms) are used to refer to kinds (recall section 1). Moreover, in most cases the unitof- measure reading is completely independent of conventions: although one can think of portions of semaag ‘tobacco-PL’ as conventionalized, it is impossible to refer to ‘pieces of ice/clay/grass’ as conventionalized units.

I should also point out that some mass nouns in Ojibwe resist pluralization completely (despite what is claimed in the literature; see, e.g., Rhodes 1990:153, Corbett 2000:87). The speakers I consulted rejected pluralization of the nominals in (27), both on the kind reading and on the serving/measure reading, even though it is not difficult to think of standard servings/ conventionalized units (or different kinds, for that matter) for oil (27a), milk (27c), or water (27d). Here, Ojibwe clearly differs from Breton and Arabic, where it is possible for liquid terms to be targets of the singulative.

(27)

graphic

Nouns referring to abstract things such as qualities and states cannot occur in the plural either. Examples are from Valentine 2001:182.27

(28)

graphic

Finally, contrary to their Ojibwe counterparts zhoonyag ‘coins’ and mitig ‘trees’, which have a simple unit reading (see (22)), in English no coercion can achieve this result: monies does not mean ‘coins’ and woods or forests cannot mean ‘trees’.

To summarize, we have seen that (a) the pluralized version of an Ojibwe or Fox mass or collective term is accompanied by a change in the gender class of the noun; (b) gender shift also targets singulars; (c) not all mass nouns can be pluralized in Ojibwe (liquid terms resist pluralization); and (d) the function of gender shift in Ojibwe is to perform individualization. These properties are exactly those found in languages with a singulative system (see section 1). In the next section, I give a full account of the singulative in Ojibwe, with complete derivations.

3 An Analysis of the Ojibwe Singulative

In the simplest cases, pluralization in Ojibwe creates a series of discrete individuals. Just like English nuts, Ojibwe bagaanag ‘nuts’ in (29a) means ‘more than one nut’. (29a–c) are animate nominals while (30a–c) are inanimates.

(29)

graphic

(30)

graphic

Since everything points to the idea that there is a count/mass distinction in Ojibwe (see section 2 and Mathieu 2007, 2012), the category Number (by way of Div0) is projected in the language. Therefore, the structure of niizh gaazhagensag ‘two cats’ must be (31). Division is performed by the plural (as in Borer 2005 for other languages). The noun raises to Div0 via head movement,28 and the numeral niizh ‘two’ is added as a counter in Spec,#P.

(31)

graphic

In the absence of a DivP, nouns are always interpreted as mass. So far, this is exactly like English. The meanings of (31) and (32) are read off from the structure (neoconstructionist view; Borer 2005 and many others).

(32)

graphic

Where Ojibwe differs from English is in the use of measure phrases. Since Ojibwe has no measure phrases of the type pieces of, bits of, portions of in English, and since Ojibwe also makes less and less use of its mass classifier system, I want to argue that the language uses another system to obtain measure readings: the singulative. In fact, this system is probably very old and must have been in competition with the classifier system for a long time. The singulative system is still fairly productive, but often not visible for independent reasons (loss of final vowels like -i and -a; see section 2).

In Fox, the gender shift in question is clear: -i becomes -a. In the collective version of ‘money’, the noun is inanimate (33a) (and is undivided; i.e., no DivP is present), while in the individual version of ‘money’, the noun is animate (33b) (DivP is projected and gender shift operates in Div0).

(33)

graphic

(34) illustrates the case of a mass noun turned into a unit of measure in Fox. The mass noun ‘meat’ starts undivided and is inflected inanimate (34a). In (34b), the addition of DivP and gender shift under Div0 creates a unit of measure.

(34)

graphic

In Ojibwe, exactly the same derivations take place, (35) and (36). The difference between Fox and Ojibwe is that in Ojibwe, gender shift is not morphologically visible for most nouns.

(35)

graphic

(36)

graphic

When the numeral ‘one’ is added to the structure in (35b), it is directly inserted in Spec,#P, as shown in (37). This differs from nonsingulative contexts, where in Ojibwe (as in English) the numeral ‘one’ first performs division under Div0 and then moves to #0 (as in Borer 2005), as shown in (38).

(37)

graphic

(38)

graphic

Let us now turn to the case of pluralized singulatives. Although it may have appeared earlier (when I first introduced mass plurals) that the plural in Ojibwe plays the role, not only of a count classifier, but also of a mass classifier, it should be clear by now that this is only an illusion. The idea is that it is the gender shift prior to plural insertion that performs division. The question thus arises, what is the status and role of the plural in singulative contexts? I want to argue that the plural is in this case not a divider but a simple counter. Its function has changed and it appears under #0, the locus of the counting function.

(39)

graphic

Of course, it is always possible to add a numeral in this case, since Spec,#P is available. This is shown in (40).

(40)

graphic

Morphologically, the counting plural is identical to the dividing plural. It simply behaves differently depending on where it appears in the structure. When in Div0, the plural behaves as a divider; when in Num0, the plural behaves as a counter. We know independently that the same morpheme/grammatical element can have two very different functions: for example, -ri in Dagaare encodes both singular and plural (see section 1), and ‘one’ in Ojibwe sometimes behaves as a divider, sometimes as a counter ((37) vs. (38)).29

Finally, table 3 adds Ojibwe to the picture. Ojibwe is similar to yet different from other singulative languages in that it allows substance terms to be singulativized, but not liquid terms.

Table 3.

Revised version of table 1. (sing = singulative; sgl = singular)

LanguageLiquidsSubstancesGranular aggregatesCollective aggregatesIndividual objects
English plural (-splural (-s
Breton sing (-ennsing (-ennsing (-ennsing (-ennplural (-
Welsh sing (-ynsing (-ynsing (-ynsing (-ynplural (-od
Dagaare sing (-ruusgl (-riplural (-ri
Arabic sing (-ahsing (-ahsing (-ahsing (-ahplural (-t
Ojibwe sing (iasing (iasing (iaplural (-g/-n
LanguageLiquidsSubstancesGranular aggregatesCollective aggregatesIndividual objects
English plural (-splural (-s
Breton sing (-ennsing (-ennsing (-ennsing (-ennplural (-
Welsh sing (-ynsing (-ynsing (-ynsing (-ynplural (-od
Dagaare sing (-ruusgl (-riplural (-ri
Arabic sing (-ahsing (-ahsing (-ahsing (-ahplural (-t
Ojibwe sing (iasing (iasing (iaplural (-g/-n

In summary, although gender shift has been known to be used in Algonquian languages to signal change of perspective or expression of power (see references in Goddard 2002), what is less well-known is that gender shift in these languages can target mass and collective nouns. This is discussed at length in Goddard 2002 for Fox. What has never been proposed before is that this type of gender shift is a feature of an underlying singulative system.

4 The Diminutive as Singulative

In this section, I argue that in addition to gender shift, Ojibwe uses the diminutive as a means to singulativize collective and mass nouns. While the individuating property of the diminutive is well-known (it is widespread crosslinguistically; see Dressler and Barbaresi 1994, Jurafsky 1996), the idea that it is a form of the singulative is, to the best of my knowledge, not often (if ever) entertained. There is a good reason for this: in many cases, the diminutive behaves differently from gender shift in its individualizing capacity. However, I want to show that in Ojibwe there is evidence to treat both types on a par.

On its most basic use, the diminutive process in Ojibwe turns nouns such as mkisin ‘shoe’ into mkiznens ‘little shoe’. However, as Rhodes (1990:152) points out, ‘‘The highly lexicalized English diminutives frequently used to gloss these forms belie this productivity; any concrete Ojibwa noun can form a diminutive up to the limits of semantic compatibility. This extends even to borrowings.’’ Some examples are given in (41).30

(41)

graphic

When the diminutive ending is attached to the name of an animal, the meaning is the young of that animal (42a–c). The early stages of human life are also marked with the help of the diminutive (42d–e). Examples are from Rhodes 1990:152–153.

(42)

graphic

Interestingly for the present discussion, the diminutive can also help individuate mass or collective nouns as shown in (43). These in turn can be pluralized, as the third column shows (examples from Rhodes 1990:153–154 and Pedchenko, Lippert, and Gambill 2003).

(43)

graphic

In these examples, we see that the diminutive achieves the same result as gender shift would. Nouns denoting substances, granular aggregates, and collective aggregates are targeted by the diminutive to form an individual. As with the case of gender shift, the diminutive cannot apply to liquid terms, (44a–b).

(44)

  • a.

    *nbiishins intended: ‘puddle of water’

  • b.

    *bimidens intended: ‘bit of oil’

Despite the similarities, there is one main difference between gender shift and the diminutive as markers of the singulative: the diminutive tends to yield a simple unit rather than a unit-ofmeasure reading. For example, in (43d) the diminutivized noun does not mean ‘a piece of sugar’; it means ‘a candy’. In (43g), the diminutivized noun does not mean ‘a piece of bread’; it means ‘a cookie’. However, it must be noted that the diminutive targets exactly the same kind of collective and mass nouns as the singulative. In addition, although it is common for mass nouns in singulative languages with gender shift to be ambiguous between the simple unit reading and the measure reading, it appears that in Ojibwe gender shift correlates with unit-of-measure readings while the diminutive correlates with simple unit readings. However, this is not a strict dichotomy. Instead, I speculate that in order to avoid the ambiguity that a term faces when singulativized, the diminutive as divider is used to mark simple units, some of which are specialized (‘bread’ becomes ‘cookie’, etc.—although bkwezhgaans ‘bread-DIM’ is in fact ambiguous among ‘cookie’, ‘biscuit’, ‘roll’, ‘doughnut’, etc.). The difference in meaning is reflected in the context or across dialects.

Since I am not making a difference between the measure reading and the simple unit reading (both are exemplars of Div0), I propose that the diminutive is another flavor that Div0 can take. (45) shows the derivation for a mass noun such as mkwam ‘snow’ that becomes singulativized via the diminutive.

(45)

graphic

Ojibwe is far from being the sole language to use the diminutive as an individualizer. In many languages, an element that also expresses smallness must be used to achieve individuation (see Rijkhoff 1991 for discussion). Stroomer (1987:87) reports that the singulative affix -ittii in Wellegga Oromo (Cushitic) also has a diminutive reading.

In summary, the diminutive is a full individualizer on a par with gender shift. My account applies to Ojibwe and is not meant to apply to other languages where diminutives have been shown to encode atomization independently of a singulative system (absent from the language). For example, in some dialects of German and Dutch, it is possible to individuate a mass noun with the diminutive (Wiltschko 2006, De Belder 2008, 2011, Ott 2011). Borer (2005:92n6) mentions similar examples in passing, attributing the observation to Henk van Riemsdijk. These are sometimes said to pose a problem for her view that classifiers and number are in complementary distribution (De Belder 2008, 2011). It appears that in these languages the diminutive works differently from the diminutive in Ojibwe. In the dialects mentioned, it is not possible to diminutivize granular aggregates.31

5 Conclusion

In this article, I argued for integrating the singulative into Borer’s (2005) system of division. Division or atomization comes in different flavors cross- and intralinguistically: via the use of classifiers, number, and atomizing numerals, but also, as I have shown, via the use of the singulative. The singulative itself introduces different flavors: gender shift (from masculine to feminine, but also, as I have argued, from inanimate to animate) as well as the diminutive. With a comprehensive view of the singulative, I was able to give an account of why mass nouns in Ojibwe can often be pluralized. One major consequence of my proposal is that the plural has a dual function in the grammar: it can be used as an atomizer (as Borer (2005) argues) but also as a regular counter (as in traditional grammars).

Notes

First, and foremost, thank you to Philomene Chegahno, Berdina Johnston, Donald Keeshig, Joanne Keeshig, Isabel Millette, Juanita Pheasant, Ernestine Proulx, and Ella Waukey for sharing their beautiful language with me. Special thanks to Joanne Keeshig for introducing me to the members of the community and to Sheila Keeshig for introducing me to the teachers of the Chippewas of Nawash Elementary School. Many thanks also to Shirley Williams from Trent University. Miigwech! Second, for their comments and chats about the topic of this article, thanks to Michael Barrie, Hagit Borer, Rose-Marie Déchaine, Bethany Lochbihler, Richard Rhodes, Elizabeth Ritter, and Martina Wiltschko. Finally, many thanks to the reviewers. Funding by SSHRC is gratefully acknowledged: grants 230424-120699-2001 and 230611-120699-2001.

1 Following Cheng and Sybesma (1999), classifiers can roughly be divided into two groups: classifiers that create a unit of measure, and those that simply name the unit in which the entity denoted by the noun naturally occurs. Cheng and Sybesma refer to the classifiers that create a unit of measure as massifiers (short for mass-classifiers) and to the ones that simply name the unit of natural semantic partitioning as classifiers (short for count-classifiers).

2 The idea that number is not about counting but about dividing undivided stuff is most apparent in English in the case of fractions (as originally described by Krifka (1989)). The noun apple is systematically pluralized in (i) although there is only one apple involved in each case. This is an effect found in many languages, not just English.

(i)

  • a.

    0.2 apples/*apple

  • b.

    0.1 apples/*apple

  • c.

    1.5 apples/*apple

  • d.

    1.0 apples/*apple

    (Borer 2005:115)

3 I use division, individualization, and atomization interchangeably.

4Ojibwe encompasses varieties of the language called by different names in English, including Odawa, Ottawa, Chippewa, or Ojibway. While the language is spoken over a vast region of central Canada and in US border states from Michigan to Montana, the varieties of the language used in this study are those described in Valentine 2001: that is, dialects spoken in southern Ontario between the shores of Lake Huron to the east roughly as far as the Ottawa River. Many of the original data used in this article come from fieldwork undertaken with members of The Chippewas of Nawash Unceded First Nation at Cape Croker (Neyaashiinigmiing) on the Saugeen (Bruce) Peninsula. Ojibwe is spoken there; a few individuals speak Odawa. Other data come from published material such as grammars, dictionaries, and articles on the syntax and morphology of the language.

5 Greenberg’s work is, of course, independently a source of inspiration for Borer 2005.

7 I take it that while Div0 is universal, its content is not. The list in (1b) is not meant to be exhaustive: other exponents of Div0 and of the singulative itself might be discovered. This is why I want to avoid viewing the different flavors of Div0 as traditional parameters (see, e.g., Newmeyer 2005 and Boeckx 2010 for the idea that parameters are best relegated to outside systems of the grammar).

8 Old Breton had two singulative endings, -in and -en, one for each gender value (Hemon 1975:39). Middle and Modern Breton retain only the feminine -enn, so that singulative forms are invariably feminine (Acquaviva 2008:243).

9 ‘‘Il y a d’autres collectifs qui sont en réalité des noms de matière, ou d’espèce’’ (Trépos 1980:67). ‘There are other collectives that are in fact names of matter, or kind’ (my translation).

10 Three points are worth noting here. First, in Classical Arabic, traditional grammars distinguish two kinds of collectives: nouns of collections (applies to sentient beings), like ‘herd’ and ‘company’, and nouns of collective kinds. The latter applies to animals, plants, and inanimate objects, and comprises many nouns denoting fruits, vegetables, flowers, grains, insects, and birds. These forms serve as the basis for the derivation of singulatives (unit nouns), which in turn can be pluralized.

Second, another way to express semantic packaging in Arabic is through classifiers. Examples from Egyptian Arabic (Cairo dialect) are hittit lahm ‘a chunk of meat’, lit. ‘piece meat’, and hiitit haddiid ‘a piece of iron’, lit. ‘piece iron’ (Holes 2004:171, cited in Acquaviva 2008:223). This illustrates again different forms for Div0.

Third, collectives are also the target of the singulative in Syrian Arabic, as shown in (ia). In turn, the singulative can be pluralized, as shown in (ib). Examples are from Cowell 2005:297.

(i)

graphic

11 Breton also expresses the analogue of a singulative derivation by the use of classifier-like elements, penn ‘head’ or pez ‘piece’: moc’h ‘pigs’ ~ pennmoc’h ‘a pig’, dilhad ‘clothes’ ~ pez-dilhad ‘a garment’ (Trépos 1956:236, cited in Acquaviva 2008:244). These classifiers are in complementary distribution with singulative -enn.

12 It is admittedly more difficult to see gender shift in the case of Breton collectives, since collectives are not listed as masculine or feminine in dictionaries. This must be due to the fact that such nouns are used as kinds without articles, numerals, or quantifiers that might signal masculine or feminine gender; and since the nouns in question refer to kinds, it must be difficult or impossible to refer to them with pronouns, which are marked for masculine and feminine.

13 For example, Stump (2005) shows that the syntactic contexts that determine the choice of the ordinary singular noun potr ‘boy’ (lenited form botr) and its plural counterpart potred ‘boys’ in the following table (from Stump 2005: 63) likewise determine the choice between the singulative noun sivienn ‘strawberry’ (lenited form zivienn) and its collective counterpart sivi ‘strawberries’.

The singulative in Breton is also clearly inflectional in that it appears after inflectional morphemes: for instance, after the plural in examples such as peskedenn ‘fish-SINGULATIVE’, whose derivation is pesk ‘fish’ → pesk + ed ‘fish’ + ‘PL’ → pesk + ed + enn ‘fish’ + ‘PL’ + ‘SINGULATIVE’.

graphic

14 Dagaare is not unique in its ability to use the same morpheme in two very different contexts. In Tangga, a language of Papua New Guinea, the same element an/am denotes not only the plural for count nouns, but also the singulative for noncount nouns. The situation is not unlike that of the Kiowa inverse system, as described in Harbour 2008 (though here only singular and plural values are involved; see Corbett 2000:163). Examples are from Ross 1988:295, 298.

(i)

graphic

15 In fact, it appears that even in Dagaare, it is not completely impossible for material terms to be targets of the singulative. For example, Grimm himself shows that fanfama ‘soap’ (general term) can become fanfanuu ‘a piece of soap’ (Grimm 2011).

16 Relatively closely related languages might also vary in their ability to pair the singulative with substances and liquids. Bari and Turkana (Nilo-Saharan languages) both use the singulative, but whereas in Bari it is possible to use the singulative -tat as a unit of individuation with the word le ‘milk’ (litat ‘a drop of milk’), this is not possible in Turkana, where instead a special word for ‘drop’ must be used (Dimmendaal 2000).

17 As Corbett and Rhodes put it:

Or consider the Algonquian language Ojibway (Richard Rhodes 1990:153–4, and personal communications). Nouns which might be expected not to have a plural do in fact form plurals freely, interestingly with the unit reading and not with the sort reading. Thus mkwam ‘ice’ or ‘piece of ice’, mkwamiig ( plural) ‘pieces of ice’. Rhodes is unable to find a noun that cannot be pluralized in Ojibway. (Corbett 2000:87)

In Ojibwa there is no grammatical distinction like the mass/count distinction of Indo-European. Thus mkwam can equally mean ‘ice’ or ‘a piece of ice’. Nbiish can mean ‘water’ or ‘an amount of water’. (Rhodes 1990:153)

18 Verbal agreement in Ojibwe follows a person hierarchy: 2nd > 1st > animate 3rd > obviative (animate) 3rd > inanimate. This is why the proclitic in (13) is always first person singular. Third person marking is often not pronounced.

19 Ojibwe is not the only Native North American language with plural mass nouns triggering a unit-of-measure reading. Mithun (1999:80) mentions the case of Yup’ik: ‘‘Yup’ik mass nouns such as uquq ‘oil’ and meq ‘water’ often appear with number suffixes to indicate units of substance, such as ‘containers of ’: uqu-k ‘oil-DUAL’ = ‘two sealpokes of oil’; mer’-et ‘water-PL’ = ‘buckets of water’.’’ Yup’ik number appears to have all the properties associated with number in Ojibwe: it is obligatory, disjunctive, transparent, and word-final, yet it allows the pluralization of mass nouns.

Ojibwe has all the properties of a language with inflectional/grammatical number: plural marking is obligatory, it triggers agreement, it is not possible inside compounds, it is not possible inside derivational morphology, and pluralia tantum as well as bare plurals are available (see Wiltschko 2008 for a discussion of this cluster of properties in relation to Halkomelem Salish, a language without, it seems, inflectional number).

20 Since Ojibwe number is in complementary distribution with mass classifiers (e.g., the noun naagans ‘cup’ in (16a) cannot be pluralized in this context, *niizh-naagansanziisbaakwad ‘two cupfuls of sugar’; of course, in other nonclassifier contexts, it can be pluralized) and since independently Chinese count classifiers and mass classifiers are in complementary distribution (Fassi Fehri and Vinet 2007), I take it that number and mass classifiers target the same head, namely, Div0. In English, on the other hand, it is possible for measure phrases to be pluralized: six piecesof sugar. Therefore, it may be the case that in English two different heads are needed for this type of construction (see Borer 2005).

21 A reviewer asks how ‘‘real’’ measure words such as ‘pound’ and ‘kilo’ work. I have not been able to find a word for ‘kilo’, but the language uses the word dibaabiishkoojigan for ‘pound’. Like classifiers, it attaches to numeral roots, as in ingo-dibaabiishkoojigan ‘one pound’ and ingodwaaso-dibaabiishkoojigan ‘six pounds’.

22 This is a separate phenomenon and, in my view, not the realization of the singulative. Moreover, as Goddard (2002) points out, the use of gender shift in narratives to express ‘power’ should not be exaggerated. For example, in Fox inanimates seem to be freely assigned the powers of speech, comprehension, and thought without shifting gender (Goddard 2002:208).

23 Although rarer, there appear to be cases where the gender shift is from animate to inanimate. Goddard (2002: 211) mentions the following cases: Fox aamoowa (AN) ‘bee, member of any species of the non-formicid aculeate hymenoptera’ versus aamoowi (IN) ‘honey’ and Ojibwe ishkodekaan (AN) ‘fire-steel’ versus ishkodekaan (IN) ‘lighter.’ To quote Goddard (2002:211): ‘‘Usable honey did not exist in North America until the arrival of the European honey-bees, but F[ox] aamoowa ‘bee’ has widespread cognates and is clearly an old word. Therefore, ‘honey’ must have been named after ‘bee’ rather than the other way around’’ and ishkodekaan (AN) ‘fire-steel’ ‘‘names the older fire-making device (Baraga 1850 . . . ), while the inanimate noun names the later one’’ (i.e., ishkodekaan (IN) ‘lighter’; Nichols and Nyholm 1995:69). These cases are clearly not the product of the singulative, since no unit is created. In other languages (e.g., Hebrew; see section 1), however, it appears to be possible to use reverse gender shift to express the singulative.

24 According to Valentine (2001:116–117), while in most Ojibwe dialects all terms for money are animate, in Nishnaabemwin (Ojibwe, Odawa) they are consistently inanimate in the singular, but not in the plural, the latter being listed in dictionaries as animate. The term ‘money’ also clearly takes the animate plural -ag (the inanimate plural being -an). The singular term mitig ‘wood’ is also listed as inanimate in dictionaries, but its pluralized counterpart is listed as animate (and it surfaces with the animate plural, -oog being an allomorph of -ag).

25 As shown in (i), it is possible for certain mass nouns to be pluralized with a unit-of-measure reading, even though they are inanimate. These examples show that the singulative system is no longer fully productive morphologically. Some mass terms tend to remain inanimate when pluralized. It must be noted, however, that there is extreme dialectal variation. For example, in some dialects the plural of ‘rice’ is animate while in others it is inanimate.

(i)

graphic

26 Ojibwe has a special word that it uses for kinds (call it a kindifier), namely, dnawa or dowa (see Valentine 2001: 593; my consultants used daawa).

27 My generalization differs slightly from Valentine’s (2001:182), since he argues that mass terms such as ‘bread’, ‘snow’, and ‘sand’ cannot appear in the plural, contrary to fact: bkwezhgan-ag ‘pieces of bread’, mkwam-ig ‘pieces of snow’, and negaw-an ‘grains of sand’ are all possible.

28 The head movement operation is triggered by the fact that -agPL.AN’ is an affix. Whether head movement is carried out in the narrow syntax (Travis 1984, Roberts 2010) or at PF (Chomsky 2001) makes no difference for my proposal.

29 There also appear to exist plurals that are neither dividing nor counting (plural of modesty; exaggerative, hyperbolic, and approximative plurals; antiassociative plurals; and the plural of abundance; see Corbett 2000). These can often take special forms. For example, in Banyun, a language of the West Atlantic branch of Niger-Kordofanian, spoken in Senegal and Guinea Bissau, the plural of abundance (sometimes called the greater plural) is used when the exact number is impossible to pinpoint or when it is irrelevant, and it surfaces as a different form from that of the dividing plural. In this language, nouns typically have singular and plural forms, distinguished by prefixes, as in (ia) and (ib) (Corbett 2000:31, from Sauvageot 1967:227–228). The greater plural is shown in (ic).

(i)

graphic

In the case of the greater plural, I assume that no Div0 head is present, perhaps not even #P. The plural in this case is probably some kind of modifier (as in Wiltschko 2008).

30 The diminutive in Ojibwe has several allomorphs.

31 Thanks to a reviewer for pointing this out to me.

References

References
Acquaviva
,
Paolo
.
2008
.
Lexical plurals
.
Oxford
:
Oxford University Press
.
Baerman
,
Matthew
.
2007
.
Morphological reversals
.
Journal of Linguistics
43
:
33
51
.
Baraga
,
Friderik
.
1850
.
A theoretical and practical grammar of the Otchipwe language
.
Detroit, MI
:
Jabez Fox
.
Black-Rogers
,
Mary B
.
1982
.
Algonquian gender revisited: Animate nouns and Ojibwa ‘power’—an impasse?
Papers in Linguistics
15
:
59
76
.
Boeckx
,
Cedric
.
2010
.
What principles & parameters got wrong
.
Ms., Universitat Autònoma de Barcelona
.
Booij
,
Geert
.
1993
.
Against split morphology
. In
Yearbook of morphology 1993
, ed. by
Geert
Booij
and
Jaap
van Marle
,
27
50
.
Dordrecht
:
Kluwer
.
Booij
,
Geert
.
1995
.
Inherent versus contextual inflection and the split morphology hypothesis
. In
Yearbook of morphology 1995
, ed. by
Geert
Booij
and
Jaap
van Marle
,
1
16
.
Dordrecht
:
Kluwer
.
Borer
,
Hagit
.
2005
.
In name only
.
New York
:
Oxford University Press
.
Bunt
,
Harry
.
1985
.
The formal representation of (quasi-)continuous concepts
. In
Formal theories of the commonsense world
, ed. by
Jerry
Hobbs
and
Robert
Moore
,
37
70
.
Norwood, NJ
:
Ablex
.
Cheng
,
Lisa Lai-Shen
, and
Rint
Sybesma
.
1999
.
Bare and not-so-bare nouns and the structure of NP
.
Linguistic Inquiry
30
:
509
542
.
Chierchia
,
Gennaro
.
1998
.
Reference to kinds across languages
.
Natural Language Semantics
6
:
339
405
.
Chomsky
,
Noam
.
2001
.
Derivation by phase
. In
Ken Hale: A life in language
, ed. by
Michael
Kenstowicz
,
1
52
.
Cambridge, MA
:
MIT Press
.
Clausen
,
David
,
Alex
Djalali
,
Scott
Grimm
,
Sven
Lauer
,
Tania
Rojas-Esponda
, and
Beth
Levin
.
2010
.
Extension, ontological type, and morphosyntactic class: Three ingredients of countability
.
Ms., Stanford University, Stanford, CA. Available at http://www.stanford.edu/~bclevin/bochum10abst.pdf
.
Corbett
,
Greville
.
2000
.
Number
.
Cambridge
:
Cambridge University Press
.
Cowell
,
Mark W
.
2005
.
A reference grammar of Syrian Arabic (based on the dialect of Damascus)
.
Washington, DC
:
Georgetown University Press
.
Davis
,
Henry
, and
Lisa
Matthewson
.
1999
.
On the functional determination of lexical categories
.
Revue québécoise de linguistique
27
:
30
69
.
De Belder
,
Marijke
.
2008
.
Size matters: Towards a syntactic decomposition of countability
. In
WCCFL 27
, ed. by
Natasha
Abner
and
Jason
Bishop
,
116
122
.
Somerville, MA
:
Cascadilla Press
.
De Belder
,
Marijke
.
2011
.
A morphosyntactic decomposition of countability in Germanic
.
Journal of Germanic Linguistics
14
:
173
202
.
Dimmendaal
,
Gerrit J
.
2000
.
Number marking and noun categorization in Nilo-Saharan languages
.
Anthropological Linguistics
42
:
214
261
.
Doetjes
,
Jenny
.
1996
.
Mass and count: Syntax or semantics?
In
Proceedings of Meaning on the HIL
,
34
52
.
Leiden
:
HIL/Leiden University
.
Doetjes
,
Jenny
.
1997
.
Quantifiers and selection: On the distribution of quantifying expressions in French, Dutch and English
.
Doctoral dissertation, Leiden University
.
Doron
,
Edit
, and
Ana
Müller
.
2010
.
The cognitive basis of the mass-count distinction: Evidence from bare nouns. Abstract
.
The Hebrew University of Jerusalem and University of São Paulo
.
Dressler
,
Wolfgang
, and
Lavinia Merlini
Barbaresi
.
1994
.
Morphopragmatics: Diminutives and intensifiers in Italian, German, and other languages
.
Berlin
:
Mouton de Gruyter
.
Fassi Fehri
,
Abdelkader
, and
Marie-Thérèse
Vinet
.
2007
.
Number and classifier distributions in Arabic and Chinese and their complementarity
. In
Architectures, rules, and preferences: Variations on themes by Joan Bresnan
, ed. by
Annie
Zaenen
,
Jane
Simpson
,
Tracy Holloway
King
,
Jane
Grimshaw
,
Joan
Maling
, and
Chris
Manning
,
167
193
.
Stanford, CA
:
CSLI Publications
.
Goddard
,
Ives
.
2002
.
Grammatical gender in Algonquian
. In
Papers of the Thirty-Third Algonquian Conference
, ed. by
H. C.
Wolfart
,
195
231
.
Winnipeg
:
University of Manitoba
.
Greenberg
,
Joseph H
.
1972
.
Numeral classifiers and substantival number: Problems in the genesis of a linguistic type
.
Working Papers on Language Universals
9
:
2
39
.
Greenberg
,
Joseph H
.
1974
.
Studies in numeral systems, I: Double numeral systems
.
Working Papers on Language Universals
14
:
75
89
.
Grimm
,
Scott
.
2009
.
Number marking and individuation: A view from Dagaare. Abstract
.
Stanford University, Stanford, CA
.
Grimm
,
Scott
.
2010
.
Number and markedness: A view from Dagaare
. In
Proceedings of Sinn und Bedeutung 14
, ed. by
Martin
Prinzhorn
,
Viola
Schmitt
, and
Sarah
Zobel
,
168
184
.
Vienna
:
University of Vienna, Department of Linguistics
.
Grimm
,
Scott
.
2011
.
Countability in Dagaare
.
Paper presented at the Department of African Studies, Humboldt University, Berlin, December
.
Harbour
,
Daniel
.
2008
.
Morphosemantic number: From Kiowa noun classes to UG number features
.
Dordrecht
:
Springer
.
Hemon
,
Roparz
.
1975
.
A historical morphology and syntax of Breton
.
Dublin
:
Institute for Advanced Studies
.
Holes
,
Clive
.
2004
.
Modern Arabic: Structures, functions, and varieties
. 2nd ed.
Washington, DC
:
Georgetown University Press
.
Jurafsky
,
Daniel
.
1996
.
Universal tendencies in the semantics of the diminutive
.
Language
72
:
533
578
.
Kaye
,
Jonathan
, and
Glyne
Piggott
.
1973
.
On the cyclical nature of Ojibwa t-palatalization
.
Linguistic Inquiry
4
:
345
362
.
Krifka
,
Manfred
.
1989
.
Nominal reference, temporal constitution and quantification in event semantics
. In
Semantics and contextual expression
, ed. by
Renate
Bartsch
,
Johan
van Benthem
, and
Peter van Emde
Boas
,
75
115
.
Dordrecht
:
Foris
.
Krifka
,
Manfred
.
1995
.
Common nouns: A contrastive analysis of English and Chinese
. In
The generic book
, ed. by
Gregory N.
Carlson
and
Francis Jeffry
Pelletier
,
398
411
.
Chicago
:
University of Chicago Press
.
Mathieu
,
Éric
.
2007
.
The count/mass distinction and the role of number in Ojibwe
.
Paper presented at the Workshop on Nominal and Verbal Plurality, Paris, November 9–10
.
Mathieu
,
Éric
.
2012
.
On the mass/count distinction in Ojibwe
. In
Count and mass across languages
, ed. by
Diane
Massam
,
172
198
.
Oxford
:
Oxford University Press
.
Mithun
,
Marianne
.
1988
.
Lexical categories and the evolution of number marking
. In
Theoretical morphology: Approaches in modern linguistics
, ed. by
Michael
Hammond
and
Michael
Noonan
,
211
234
.
San Diego, CA
:
Academic Press
.
Mithun
,
Marianne
.
1999
.
The languages of Native North America
.
Cambridge
:
Cambridge University Press
.
Muehlbauer
,
Jeffrey
.
2008
.
kâ-yôskâtahk ôma nêhiyawêwin: The representation of intentionality in Plains Cree
.
Doctoral dissertation, University of British Columbia
.
Newmeyer
,
Frederick
.
2005
.
Possible and probable languages: A generative perspective on linguistic typology
.
Oxford
:
Oxford University Press
.
Nichols
,
John
, and
Earl
Nyholm
.
1995
.
A concise dictionary of Minnesota Ojibwe
.
Minneapolis
:
University of Minnesota Press
.
Ott
,
Dennis
.
2011
.
Diminutive formation in German: Spelling out the classifier analysis
.
Journal of Comparative Germanic Linguistics
.
Online
.
Pedchenko
,
Svetlana ‘‘Weshki-ayaad,’’ Charlie Lippert
, and
Guy
Gambill
.
2003
.
Freelang Ojibwe-English dictionary and English-Ojibwe dictionary
. .
Pelletier
,
Francis Jeffry
.
1979
.
Non-singular reference: Some preliminaries
. In
Mass terms: Some philosophical problems
, ed. by
Francis Jeffry
Pelletier
,
1
14
.
Dordrecht
:
Reidel
.
Piggott
,
Glyne
.
2007
.
Deriving word minimality by phase
.
Ms., McGill University, Montreal
.
Rhodes
,
Richard
.
1985
.
Eastern Ojibwa-Chippewa-Ottawa dictionary
.
Berlin
:
Mouton de Gruyter
.
Rhodes
,
Richard
.
1990
.
Lexical hierarchies and Ojibwa noun derivation
. In
Meanings and prototypes: Studies in linguistic categorization
, ed. by
Savas L.
Tsohatzidis
,
151
158
.
London
:
Routledge
.
Rijkhoff
,
Jan
.
1991
.
Nominal aspect
.
Journal of Linguistics
8
:
291
309
.
Ritter
,
Elizabeth
, and
Martina
Wiltschko
.
2009
.
Varieties of Infl: Tense, location, and person
. In
Alternatives to cartography
, ed. by
Jeroen
van Craenenbroeck
,
153
202
.
Berlin
:
Mouton de Gruyter
.
Roberts
,
Ian
.
2010
.
Agreement and head movement: Clitics, incorporation, and defective goals
.
Cambridge, MA
:
MIT Press
.
Ross
,
Malcolm D
.
1988
.
Proto Oceanic and the Austronesian languages of Western Melanesia
.
Canberra
:
Pacific Linguistics
.
Sanches
,
Mary
, and
Linda
Slobin
.
1973
.
Numeral classifiers and plural marking: An implicational universal
.
Working Papers on Language Universals
11
:
1
22
.
Sauvageot
,
Serge
.
1967
.
Sur la classification nominal en baïnouk
. In
La classification nominale dans les langues négro-africaines
,
225
235
.
Paris
:
Centre national de la recherche scientifique
.
Stroomer
,
Harry
.
1987
.
A comparative study in three southern Oromo dialects in Kenya (phonology, morphology, and vocabulary)
.
Doctoral dissertation, Leiden University
.
Stump
,
Gregory
.
2005
.
Word-formation and inflectional morphology
. In
Handbook of word-formation
, ed. by
Pavol
Štekauer
and
Rochelle
Lieber
,
48
71
.
Dordrecht
:
Springer
.
Ternes
,
Elmar
.
1992
.
The Breton language
. In
The Celtic languages
, ed. by
Donald
MacCaulay
,
371
452
.
Cambridge
:
Cambridge University Press
.
Travis
,
Lisa
.
1984
.
Parameters and effects of word order variation
.
Doctoral dissertation, MIT, Cambridge, MA
.
Trépos
,
Pierre
.
1956
.
Le pluriel Breton
.
Rennes
:
Imprimerie Réunies
.
Trépos
,
Pierre
.
1980
.
Grammaire Bretonne
.
Rennes
:
Ouest-France
.
Valentine
,
Randy
.
2001
.
Nishnaabemwin reference grammar
.
Toronto
:
University of Toronto Press
.
Whorf
,
Benjamin
.
1941
.
The relation of habitual thought and behavior to language
. In
Language, culture and personality: Essays in memory of Edward Sapir
, ed. by
Leslie
Spier
,
A. Irving
Hallowell
, and
Stanley
Newman
,
75
93
.
Menasha, WI
:
Sapir Memorial Publication Fund 1941
.
Wiltschko
,
Martina
.
2006
.
Why should diminutives count?
In
Organizing grammar: Studies in honor of Henk van Riemsdijk
, ed. by
Hans
Broekhuis
,
Norbert
Corver
,
Riny
Huybregts
,
Ursula
Kleinhenz
, and
Jan
Koster
,
669
679
.
Berlin
:
Mouton de Gruyter
.
Wiltschko
,
Martina
.
2008
.
The syntax of non-inflectional plural marking
.
Natural Language and Linguistic Theory
26
:
639
694
.
Wiltschko
,
Martina
.
2012
.
Decomposing the mass/count distinction: Towards a formal typology
. In
Count and mass across languages
, ed. by
Diane
Massam
,
146
171
.
Oxford
:
Oxford University Press
.