Jump to ContentJump to Main Navigation
Show Summary Details
More options …

The Linguistic Review

Editor-in-Chief: van der Hulst, Harry

4 Issues per year

IMPACT FACTOR 2017: 0.558
5-year IMPACT FACTOR: 0.813

CiteScore 2017: 0.56

SCImago Journal Rank (SJR) 2017: 0.403
Source Normalized Impact per Paper (SNIP) 2017: 0.876

See all formats and pricing
More options …
Volume 32, Issue 4


Animating the narrow syntax

Martina Wiltschko / Elizabeth Ritter
  • Ben Gurion University of the Negev, Be’er Sheva, 8410501, Israel
  • University of Calgary, Calgary, AB T2N 1N4, Canada
  • Other articles by this author:
  • De Gruyter OnlineGoogle Scholar
Published Online: 2015-11-24 | DOI: https://doi.org/10.1515/tlr-2015-0011


In this paper we demonstrate that there are two related, but distinct types of animacy in the Plains Algonquian language, Blackfoot – morphological animacy (m-animacy) and high animacy (H-animacy). We argue that the two types of animacy are constructed in different ways: M-animacy is a head feature that determines noun class and plays a role in syntactic agreement operations, whereas H-animacy is a selectable feature of arguments. The two kinds of animacy also have different distributions: Only languages that have animacy-based form classes have m-animate nominals, but H-animate nominals are universal because all languages have predicates that select for high animates. We discuss empirical differences between m-animate and H-animate nominals that are intended to serve as diagnostics for the exploration of animacy in other languages.

Keywords: animacy; humanness; noun classification; selection; Blackfoot

1 Introduction

Algonquian languages are famous for their animacy-based noun class system. In this paper we demonstrate that there are two related, but distinct, types of animacy in the Plains Algonquian language, Blackfoot. The first is a morphosyntactic element that determines agreement. The second is a semantic element that serves as a restriction on arguments. We argue that the two types of animacy are constructed in different ways: Morphological animacy (henceforth m-animacy) is a formal feature of a classificatory functional category; it determines noun class and plays a role in syntactic AGREE operations. High animacy (henceforth H-animacy), which is semantically conditioned, is also embodied in the syntactic representation as a formal property. In particular, we take H-animacy to be an index which associates with specifiers of functional categories and places a restriction on arguments that occupy these positions.

We offer two kinds of evidence for the distinction between m-animacy and H-animacy: Focusing first on Blackfoot, we show that H-animate nominals are a proper subset of m-animate nominals, notably because some argument positions require an H-animate nominal. As a result, H-animate nominals have a broader distribution than nominals that are m-animate only. Looking beyond Blackfoot, we propose that while m-animacy is a language specific property, H-animacy is a universal property, which is also present in languages that lack m-animacy. That is, all languages have a more complex representation for nominals that refer to humans and other ‘high animate’ beings and even in languages where m-animacy is absent, H-animate arguments may display distinctive properties. We review evidence from English and Spanish to this effect, and we argue that these distinctive properties are due to the presence of an abstract H-animate index in the syntactic representation.

The term animacy is used with a certain amount of ambiguity in the literature. Sometimes animacy is used to refer to a fundamental distinction between animate beings and inanimate objects. Elsewhere, animacy is used to refer to a distinction between human (and humanoid) beings and other animals and objects, as in Figure 1. 1

Two notions of animacy.
Figure 1:

Two notions of animacy.

We show that the conceptual category human(oid) is special in that it is visible to the narrow syntax; other conceptual categories, such as animal or biological organism are not visible to the narrow syntax, though they may be the basis for morphological classification. We speculate that the reason for this is that H-animates are the only type of arguments that may bear event roles (agent, undergoer) as well as speech act roles (speaker, addressee); other types of arguments lack this capacity.

The paper is organized as follows: in Section 2, we discuss the empirical differences between m-animates and H-animates in Blackfoot. In Section 3, we analyze m-animacy as a head feature associated with nominal inner Aspect. In Section 4, we argue that H-animacy is a property of arguments. In Section 5, we discuss evidence for our proposal that goes beyond Blackfoot, and in Section 6, we conclude.

2 Empirical differences between m-animates and H-animates

In this section, we show that the grammar of Blackfoot formally represents both animacy and humanness. However, we also show that these two notions serve very different purposes. Animacy is a formal morpho-syntactic feature that plays a role in agreement; humanness is a semantic restriction on arguments.

2.1 Animacy as a morphological form class system

Like all Algonquian languages, Blackfoot makes a morphological distinction between animate and inanimate nouns. 2 Animate nouns and their modifiers take the plural suffix -iksi (1a), whereas inanimate nouns and their modifiers take the suffix -istsi (1b). 3, 4


‘Those boys are tall.’


‘Those houses are tall.’

Bliss 2013: 31 (1c, 2b)

As shown in Table 1, the inflectional paradigms for animate and inanimate nouns differ in both content and complexity: singular animate nouns are either proximate or non-proximate (i.e., obviative), whereas singular inanimate nouns are unspecified for this feature. See Section 3 for a formal analysis.

Table 1:

Inflectional paradigm for animate and inanimate nouns.

This asymmetry between the two paradigms is reminiscent of other morphological splits that are based on the animacy hierarchy in (2). 5


human > non-human animate > inanimate

For example, in the typological literature, the animacy hierarchy is intended to capture the fact that plural marking in different languages is only possible, or only obligatory, for animate nouns.

The animacy hierarchy in (2) consists of semantic categories. Consequently, if the facts in Table 1 were a reflex of this hierarchy, we might expect the availability of proximate -wa to be restricted to nouns that denote humans and other animate beings. However, a survey of the lexical semantic classes of nouns in each of the morphological classes fails to bear out this prediction, as illustrated in Table 2.

Table 2:

Lexical semantic classes of animate and inanimate nouns.

As expected, the animate class includes all nouns that denote humans and animals, as well as spirits and deities. Together, these nouns comprise a lexical semantic class that denotes entities that are variously referred to as mental state holders, rational or sentient beings (e.g. Reinhart 2002; Corbett 1991; Tenny 2006; a.o.). Significantly, though, the animate class also includes all nouns that denote objects that roll, as well as metal objects and a seemingly random assortment of other nouns that denote inanimate objects. These are obviously not mental state holders, so why are they classified as animate?

Psychological research shows that “infants distinguish between inanimate objects and animates, namely humans, in important ways. For example, they recognize that humans are self-propelled while inanimate objects move only after contact with another object,” (Kuhlmeier et al. 2004: 9). Arguably, then, there might be a cognitive motivation for the animate classification of objects that roll since, like humans and other animate beings, they are potentially animated, i.e. able to move. Metal artifacts might be classified as animate because they are man-made. However, this kind of rationale is unavailable for nouns that denote berries, body parts and items of clothing. Any attempt at rationalizing their animacy is doomed in light of the fact that these classes of nouns are sometimes animate and sometimes inanimate. In this respect, Blackfoot is a typical Algonquian language, as is evident from Bloomfield’s (1946: 94) description of the facts:

Nouns are in two […] classes, inanimate and animate; the latter includes all persons, animals, spirits, and large trees, and some other objects such as tobacco, maize, apple, raspberry (but not strawberry), calf of leg (but not thigh), stomach, spittle, feather, bird’s tail, horn, kettle, pipe for smoking, snowshoe.

Thus, while there is semantic motivation for the classification of some nouns as animate or inanimate, it is impossible to predict the form class of all nouns in Blackfoot, or any other Algonquian language, based solely on their meanings (Goddard 2002). 6 What this means is that the full range of facts cannot be adequately captured as an animacy hierarchy effect. This must be treated as a morpho-syntactic property, rather than a semantic one. Regardless of their lexical semantic classification, all animate nouns belong to one inflectional paradigm; all inanimate nouns belong to another. Moreover, whether a noun is animate or inanimate is reflected in the form of modifying demonstratives, numerals and relative clauses. We conclude, then, that animacy in Blackfoot is a formal morpho-syntactic feature of nouns and their modifiers.

2.2 Only H-animate DPs are selectable

In the previous subsection we established that Blackfoot animate nouns belong to the same morphological paradigm, regardless of whether they denote animate beings or inanimate objects. We now present evidence that Blackfoot distinguishes DP arguments that denote humans, animals and spirits from those that denote inanimate objects. Henceforth, we will use the term H-animate for this subclass of human and humanoid DPs.

Previous research on Blackfoot has determined that only H-animate DPs can (a) be subjects of transitive verbs (Frantz 2009); (b) be added to the structure as non-core objects (Bliss 2010); (c) trigger inverse marking on the verb (Bliss 2005); and (d) co-occur with PP goals of directed motion verbs (Kim 2015). We propose that in each of these contexts, the H-animate DP is an argument of a functional head. We interpret this cluster of facts as evidence that H-animate DPs have some property that is visible for argument selection. In the remainder of this section we briefly review the facts and discuss the functional categories responsible for this set of constraints.

2.2.1 Only H-animate DPs can be subjects of transitive v

In Blackfoot, only H-animate nominals can function as external arguments of transitive verbs. Consequently, morpho-syntactically inanimate nominals and animate nominals that denote inanimate objects, such as instrumental subjects, as in (3), are impossible. In order to express the proposition in (3), speakers of Blackfoot must use an impersonal construction, as illustrated in (4) (Frantz 2009). Observe that the prefix iiht- has been added to this example to license the instrumental adjunct. There are also differences in the form of the theme sign and the third person agreement on the verb in the two examples. 7 In (3), the theme sign –m signals that a third person animate subject is acting on an inanimate object, and the suffix -wa on the verb constitutes number agreement with the singular subject. In (4), on the other hand, the direct/inverse marker is -‘p, indicating that a non-specific subject is acting on an inanimate object, and in this context, the number suffix -yi signals agreement with the plural object.


‘That knife cut off those branches.’

Frantz 2009: 46 (f)


‘By means of the knife [somebody] cut off those branches.’

Frantz 2009: 46 (g)

A second pair of examples that illustrates this same point is provided in (5). These examples show that the transitive verb iksikiin ‘wake (somebody) up’ must also have an H-animate DP as its external argument. In (5a), the H-animate DP anna pookaa ‘the child’ is the external argument, as indicated by the presence of the theme sign and the absence of any adjunct licensing prefixes. In (5b), on the other hand, the non H-animate nominal ootsistsapis ‘(the) noise’ is an adjunct licensed by the prefix oht-, a variant of the prefix iiht- in (4).


‘The child woke me up.’

1-inst- wake.up.ta-3:1noise

‘The noise woke me up.’

We interpret these facts as evidence that transitive v, the functional category that selects the external argument of the transitive verb, requires an H-animate DP, as in (6).


2.2.2 Only H-animate DPs can license direct/inverse marking

Direct/inverse markers only occur on transitive verbs with an m-animate object in Blackfoot. A direct marker indicates that the person prefix on the verb is specified for features of the subject; an inverse marker indicates that this person prefix is specified for features of the object, as schematized below.


The choice between direct and inverse marking in Algonquian languages has traditionally been described in terms of the animacy hierarchy, augmented to distinguish among third person arguments that bear proximate or obviative marking. The former indicates that the referent is discourse salient while the latter indicates that it is not.


Direct/Inverse Hierarchy – original version

1/2 > 3proximate > 3obviative > 3inanimate

However, on the basis of examples like (9), Bliss (2005) demonstrates that sentient arguments outrank non-sentient ones, regardless of their obviation status.

dem-proxball-proxdem -obv3-dur-throw-dir/*invbe.red

‘The ball that he is bouncing is red.’

Bliss 2005: 34 (30)

In this example, the embedded clause (‘he is bouncing it’) has an obviative argument (‘he’) acting on a proximate argument (‘the ball’). According to the hierarchy in (8), we should find an inverse marker on the transitive verb but, in fact, we find a direct marker. The reason, for this, according to Bliss, is that the obviative agent ‘he’ is both animate and sentient (H-animate in our terms) and thus outranks the proximate theme ‘the ball’, which is animate, but not sentient. In other words, what this example demonstrates is that direct/inverse morphology is governed primarily by semantic sentience, rather that morphological obviation. In light of these facts, the animacy hierarchy that governs direct/inverse marking must be revised as in (10).


Direct/Inverse Hierarchy – modified version

1/2 > 3sentient.prox > 3sentient.obv >3non-sentient 8

In Section 2.2.1, we established that external arguments (subjects) of transitive verbs must be H-animate. In this subsection, we have seen that both direct and inverse marking can only be licensed by H-animate arguments.

An anonymous reviewer correctly points out that the fact that direct marking is only licensed by an H-animate transitive subject follows from the H-animacy restriction on subjects of transitive verbs. However, there is no such restriction on objects of transitive verbs. Thus, the fact that inverse marking only occurs when there is an H-animate object provides strong evidence that the argument selected by the functional category that hosts the direct/inverse marker is necessarily H-animate, just like the argument selected by transitive v. Following Ritter and Wiltschko (2009), we assume that direct/inverse markers instantiate Viewpoint Aspect in Blackfoot – the category that Bliss labels Point of View. Like v, Viewpoint Aspect requires an H-animate argument in its specifier position.


This restriction is perhaps unsurprising, given that only sentient beings can hold a point of view. 9 However, not all environments that require H-animate arguments are semantically motivated in this way. We turn to such an environment in the next subsection.

2.2.3 ONLY H-animate nouns can be non-core objects

As observed by Bliss (2010), non-core arguments constitute a third type of argument that has to be H-animate. These are arguments, such as benefactives, that function as the grammatical object of a transitive verb, and are licensed by the addition of a benefactive or other applicative morpheme to the verb stem. This is illustrated by the minimal pair in (12) and (13). Observe that when the benefactive is H-animate, as in (12), it functions as the grammatical object of the verb, as is evident from the fact that the 3rd person agreement suffix agrees with the singular benefactive object (ana issitsimann ‘the baby’), rather than the plural-marked logical object (amiksi si’káániksi ‘the blankets’). Conversely, when the benefactive is a not H-animate, such as ani ákssin ‘the bed’ in (13), it cannot function as the grammatical object. Rather, this type of goal is realized as an adjunct, licensed by the addition of a purposive prefix (iht-) on the verb. In this context, the logical object is the grammatical object, as indicated by the plural agreement on the transitive verb.


‘I sewed those blankets for the baby.’


‘I sewed those blankets for the bed.’ Bliss 2010: 63 (12)

Pylkkänen (2008) argues that benefactives are licensed in the specifier position of the functional head Appl. Based on the examples in (12) and (13) we can conclude that in Blackfoot the specifier position of ApplP is restricted to H-animate benefactives.


2.2.4 ONLY H-animate nouns can be themes of motion Vs with PP+path/goal

Kim (2014) identifies a fourth context restricted to H-animate DPs, namely the themes of directed motion verbs with a prepositional object. Unsurprisingly, directed motion verbs without a path or goal object are acceptable with both H-animate and non-H animate theme subjects, as shown in (15). However, if a path or goal object is added to the predicate, then the subject must be H-animate, as illustrated in (16).

anna akiikoan sainnisoo

‘That girl went downward.’

anna ainaka’si itskoo

‘That wagon passed by.’

Kim 2014: 132 (16a,b)
anna akiikoan iihtoo anni niitahtaayi

‘That girl will go along the river.’

*anna ainaka’si iihtoo anni niitahtaayi

‘That wagon will go along the river.’

Kim 2014: 131 (15a,b)

Blackfoot verbs of this class always contain a prefix that specifies the direction or path of motion. Frantz (2009) calls these prefixes linkers when they occur with a path/goal object, as in (16), and non-linkers when there is no such object, as in (15). Kim (2014) argues that linkers belong to the functional category p (‘little p’.), whereas non-linkers are lexical Ps. She develops an analysis in which the path/goal object is the complement of the linker element, and together they form a vP adjunct. Suppose that these pP adjuncts also include a theme argument (pro), in Specp. The contrast in (16), suggests that this theme argument must be H-animate. Thus, p, like other functional heads (i.e. v, Asp, Appl), requires an H-animate external argument.


Note that just as with non-core arguments, the requirement for H-animates in this context is not straightforwardly understood as a semantic restriction. Rather it appears to be a configurational property of Blackfoot that specifiers of these functional categories require an H-animate argument. 10

2.3 Summary

We have now seen that all nouns that form their plural with –iksi belong to the [+animate] form class. However, not all [+animate] DPs have the same grammatical distribution. As summarized in Table 3, only H-animate DPs, i.e. those that denote human or humanoid individuals, can occupy the specifier of a functional head. The residue of the [+animate] form class, like [-animate] DPs, cannot serve as arguments of such heads.

Table 3:

Blackfoot Noun Classes – morphology and selection.

Animacy hierarchy approaches implicitly assume that humanness and animacy are different points on a single scale (Dixon, 1979). Others have suggested that what we are calling H-animacy is encyclopedic knowledge, that is, it is outside the narrow syntax (Folli and Harley 2008; Ramchand 2008). The evidence discussed in this section suggests that neither of these claims is quite right – the two kinds of animacy play different roles in the grammar, and both must be formally represented in the narrow syntax of Blackfoot.

An anonymous reviewer points out that the difference between m-animacy and H-animacy mirrors the traditional distinction between grammatical and semantic animacy much discussed in the Algonquianist tradition and beyond (e.g. Brousseau 2009; Leonard 2007; Piriyawiboon 2007; Quinn 2006; Rhodes 2010 on Algonquian and Corbett 1991; Dahl and Fraurud 1996 for a broader discussion.) However, this is not so. In particular, the classical distinction between grammatical and semantic animacy is meant to capture the fact that the morphological form class [±animate] is not fully predictable based on the ontological status of the referent. As we have seen in Section 2.1 some nouns denoting inanimate objects are classified as [+animate]. In contrast the difference between m-animacy and H-animacy is meant to capture the fact that a certain class of arguments is selectable by functional heads that impose an animacy restriction on their arguments, namely those that refer to animate beings that are capable of holding mental states. Hence, while m-animacy corresponds to grammatical animacy, H-animacy is a notion that is distinct from semantic animacy.

According to our proposal, m-animacy is a matter of individual grammars, subject to cross-linguistic variation. Some languages use m-animacy as a nominal classification device, while others do not. In contrast, H-animacy is a universal property and plays a role in all languages. However, languages differ as to the range of categories that are sensitive to H-animacy. For example, while in Blackfoot v, Asp, Appl and p all select for H-animate arguments, in English only Appl seems to do so.

In the remainder of this paper, we propose a formal representation of the two kinds of animacy. In particular, we argue in Section 3 that m-animacy is best analyzed as a bivalent morpho-syntactic head feature, which gives rise to two form classes of nouns in Blackfoot. In contrast H-animacy, is a formalized as a phrasal index that determines semantic selection of arguments. We argue in Section 4 that it is best analyzed as a restriction on the semantic role associated with an abstract argument in the specifier of nominal functional heads.

3 Analysis Part I: morphological animacy is a head-feature

Following Wiltschko (2012), we assume that the morpho-syntactic feature [±animate] associates with inner Aspect, the nominal functional projection that is responsible for the classification of nouns. 11 In order to account for the differences in proximate/obviative marking described in Section 2.1, we propose that animate noun phrases are structurally more complex than inanimate ones. More specifically, only animate noun phrases have a DP layer, whose head associates with the morpho-syntactic feature [±proximate]. Thus, we assume that nominal inflection is distributed across different positions in the syntactic tree. We develop our proposal as follows: In Section 3.1 we introduce our proposal and the background assumptions regarding the functional architecture of nominal arguments. In this subsection we also discuss in detail the distribution of the pieces of morphology associated with nominal arguments in Blackfoot. In Section 3.2 we discuss the association of animacy (in the form of plural marking) with inner Aspect. In Section 3.3 we discuss the association of proximate -wa with D. In Section 3.4 we discuss the association of the singular marker -yi with Num, and in Section 3.5 we summarize this discussion.

3.1 The distribution of nominal inflection along the spine

Recall from Section 2.1 that there exists an asymmetry in the inflectional paradigms of animate and inanimate nouns. As shown in Table 1 repeated below, inflectional suffixes that combine with animate nouns are either specified for animacy and plural number, or for singular number and proximate/non-proximate status. In contrast, inflectional suffixes that combine with inanimate nouns are specified for animacy and plural number or for singular number, but not for proximate/non-proximate status.

Thus, both animate and inanimate nouns are specified for animacy and singular/plural number. However, animate nouns differ from inanimate ones in that they are additionally specified for proximate/non-proximate. Our goal in this section is to explore the significance of this asymmetry. We develop a model that is designed to account for the manner in which these contrasts and their exponents are deployed in the nominal syntax. In particular, we hypothesize that each of the affixes is located in a different functional head, which gives rise to a difference in structural complexity for the two classes of noun phrases, as shown in (18):

Table 1:

Inflectional paradigm for animate and inanimate nouns.


The structures in (18) allow us to understand the different patterns of contrast: Consider first the animate noun phrase in (18a). The suffix -wa is a proximate marker and it contrasts with a zero non-proximate marker. We do not analyse -yi as a non-proximate marker, but rather as a marker of singular number. When it is part of an animate DP, it appears to be non-proximate because it can only be spelled out in the context of the zero non-proximate marker in D. Turning to the inanimate noun phrase in (18b) in this context, -yi is not interpreted as non-proximate marking because there is no DP layer and hence there is no zero non-proximate to trigger that interpretation.

Note that the structures in (18) are also conceptually motivated. That is, observed cross-linguistic uniformity of functions points to the existence of a universal syntactic spine consisting of a set of hierarchically organized functional layers, each with a distinct interpretive function. Thus, we assume following Ritter and Wiltschko (2014), and Wiltschko (2014) that Universal Grammar provides a set of hierarchically-organized, abstract functional categories (κ) that are defined by their interpretive function, as schematized in (19). For the present purpose we focus on the following three categories: Inner Aspect (iAsp), which is responsible for classification, Number (Num), which is responsible for individuation, and Determiner (D), the position where discourse roles are established including, speech act roles, point-of-view roles, and topicality.


Cross-linguistic variation in syntactic categories arises because the heads of these abstract categories associate with some language specific material, such as a morpheme and/or a formal feature.

Finally, we assume that the mapping of syntax to morphology is many to one. In Bliss et al. (2014) we argue that different verbal inflectional morphemes associate with clausal functional categories but compete for a single slot in the verbal morphological template. This competition is resolved by prominence hierarchies. 12 The evidence considered in the present paper indicates that this is also the case for Blackfoot nominal inflection. In other words, although the various nominal inflectional affixes associate with different functional categories, they all compete for a single slot in the noun’s morphological template, as schematized in (20).


In the remainder of this section, we provide independent evidence for this proposal.

3.2 Animacy (and plural marking) associates with inner Aspect

Unlike -wa and -yi, the suffixes -iksi and -istsi are specified for two kinds of features – animacy and number. We begin by summarizing Bliss’s (2013) argument that these suffixes are associated with a lower functional category than either -wa or -yi, and Kim and Ritter’s (2014) evidence that animacy, rather than number, is the head feature of this functional category. We then argue that this functional category is inner Aspect, based on evidence first discussed by Wiltschko (2012) and Ritter (2014b).

Some morphologically intransitive verbs in Blackfoot optionally take direct objects (Taylor 1969; Frantz 1970, 2009). However, as exemplified by the contrast between (21a,b) and (21c,d) below, there are strict constraints on the form of these objects: they may be either bare nouns or plural nouns, but not -wa or -yi marked nouns. 13


Vintrans + N


‘I am hunting elk/an elk.’

Bliss (2013) : 119 (90a)

Vintrans + N{-iksi/-istsi}


‘S/he baked breads.’

Bliss (2013) : 125 (102)

*Vintrans + N-wa


intended: ‘I am hunting an/the elk (proximate)’

Bliss (2013) : 119 (90b)

*Vintrans + N-yi


intended: ‘I am hunting an/the elk (obviative)’

Bliss (2013): 119 (90c)

Bliss (2013) interprets these facts as evidence that formally intransitive verbs select a nominal constituent that is smaller than a full DP. The reason that only plural markers are available in small nominals is that they associate with one of the categories present in a small nominal; -wa and -yi, on the other hand, associate with higher functional categories. What this means is that plural suffixes associate with the spine in a position lower than either the proximate or singular suffixes.

However, the contrast in (21) only allows us to draw conclusions about the relative position of -istsi/-iksi with respect to the other nominal suffixes. In order to determine the absolute position of these suffixes we need to consider other kinds of evidence. We cannot develop arguments based on morpheme order due to the fact that all nominal inflectional markers map onto the same position in the morphological template. Consequently, we must appeal to their interpretive function.

Significantly -istsi/-iksi are specified for both number and animacy, unlike the singular suffix -yi (and, as we will argue below, also unlike the proximate suffix -wa). Following Wiltschko (2009), we assume that only one of the features of these plural markers is in fact a head feature; the other is a modifying feature. The two types of features can be distinguished on the basis of a difference in valency: head features are bivalent whereas modifying features are monovalent. This difference has interpretive consequences. The absence of an overt morpheme with a head feature [+F] is interpreted as the presence of a null morpheme with the head feature [-F]. In contrast, the absence of an overt morpheme with a modifying feature [F] is interpreted as the lack of specification for [F].

Applying Wiltschko’s (2009) diagnostic to the Blackfoot plural markers, Kim and Ritter (2014) demonstrate that [plural] behaves like a modifying feature, whereas animacy behaves like a head feature. 14 Evidence for this comes from the following considerations: small nominal objects that contain a plural-marked noun always refer to plural individuals, whereas small nominal objects with a bare unmarked noun can refer to singular or plural individuals (or to kinds), as in (21a,b). This interpretive contrast indicates that bare nouns lack a number specification altogether, whereas plural-marking contains a monovalent feature [plural]. 15

The fact that -iksi only appears with animate nouns and -istsi with inanimate ones suggests that these markers are additionally specified for a bivalent feature [± animate]. The fact that all nouns, including bare nouns are specified either as [+animate] or [-animate], indicates that animacy is not only bivalent, but also obligatory. In other words, it is a head feature associated with a functional category.

Having established that [±animate] and [plural] are associated with the same functional category, and that this category appears in a relatively low position in the nominal spine, we turn to the question of the identity of the functional category. Following Wiltschko (2012), we assume that the interpretive function of [±animate] in Blackfoot is lexical aspectual classification, and that it associates with the functional category Inner Aspect (iAsp) (in the sense of Travis 2010).

Note that this analysis contrasts with the standard view of Algonquian scholarship according to which animacy is a type of gender (e.g. Corbett 1991; Dahlstrom 1995; Darnell and Vanek 1976; Goddard 2002; Greenberg 1954; Hockett 1966; Joseph 1979). Here we offer two arguments against this view. First, if the animacy distinctions were indeed a type of gender we would expect both animate and inanimate nouns to show the same inflectional contrasts, but in fact they do not. As we have seen in Section 2, only animate nouns are inflected for obviation. This is shown in Table 1 repeated from above.

Table 1:

Inflectional paradigm for animate and inanimate nouns.

Second, the view that animacy is a form of gender, fails to capture the fact that [±animate] plays a comparable role in the lexical classification of verbs, as well as nouns in all Algonquian languages. For example, Bloomfield (1946) develops a classification system that divides verbs into four lexical classes, which differ in transitivity and animacy of a designated argument (Table 4). Languages do not typically classify their verbs based on nominal gender distinctions. However, if animacy instantiates lexical aspect, it is not surprising that verbs are also sensitive to this distinction. Significantly, at least for Blackfoot, these verb classes appear to constitute the verbal Aktionsarten (cf. Louie 2008; Ritter 2014a, 2014b). The hypothesis that both nominal and verbal inner Aspect associate with [± animate] elegantly captures the fact that this feature has the same interpretive function in the two domains.

Table 4:

Algonquian Verb classes.

Further evidence that animacy is a distinction in lexical aspect and hence associates with inner Aspect in Blackfoot comes from the classic argument of complementarity. That is, if it is indeed [±animate] which associates with iAsp, then this functional head cannot simultaneously associate with other head features, such as [± bounded]. According to Wiltschko, [± bounded] is the feature that associates with inner Aspect in languages, such as English, that distinguish between telic and atelic verbs, on the one hand, and count and mass nouns, on the other. See Ritter and Rosen (2010) for arguments that Blackfoot lacks a telic/atelic distinction and Wiltschko (2012) for arguments that it lacks a count/mass distinction.

Finally, it is a genetically and typologically rare property of Blackfoot that plural marking is more highly specified than singular marking. Our analysis of -istsi/-iksi as [± animate] markers of the category inner Aspect with a modifying [plural] feature provides some insight into this property.

3.3 Proximate -wa associates with D

Next, we consider the Blackfoot proximate marker -wa. Although, like plural marking, -wa is sensitive to animacy, it does not associate with iAsp. This can be deduced from its distribution and interpretation. Regarding distribution, -wa does not occur in small nominal complements of formally intransitive verbs. As discussed in the Section 3.2, this constitutes evidence of the relative position of -wa compared to the plural suffixes. Since small nominals contain iAspP, we are led to conclude that -wa associates with a category that is higher than iAspP.

Regarding its interpretive function, -wa marks a noun phrase as discourse salient, though the precise discourse role is subject to debate. For example, Genee (2009) proposes that -wa marking signals topicality, while Bliss (2013) argues that it serves as a reference tracking device. Frantz (2009:13) uses the term major third person, and describes the function of -wa marking as follows: “When two or more nouns of animate gender occur in the same sentence, only one of them can be what we will call major third person; the others, if particular in reference, must be demoted to minor third person.”

Thus, -wa identifies a particular referent, based on its role in the discourse, and this is a function we associate with D. The constraint that Frantz describes implies that there may be at most one -wa marked 3rd person DP in a sentence, and that this DP is both morphologically animate and pragmatically salient. The fact that all animate DPs are either proximate or non-proximate indicates that D is associated with a bivalent head feature: [+proximate] D is realized as -wa, and [-proximate] D is realized as a null morpheme. 16

Significantly, morphologically inanimate nouns are never proximate, and hence are never inflected with -wa. Why should this be the case? We cannot appeal to a semantic difference because all animate nouns can be -wa marked, including those that denote inanimate objects, as shown in (9), repeated below as (22). Here, the morphologically animate noun ‘ball’ denotes an inanimate object. Nevertheless, it is suffixed with the proximate marker -wa.


‘The ball that he is bouncing is red.’

Bliss 2005: 34 (30)

This suggests that the explanation must be morpho-syntactic in nature. We propose that morphologically inanimate nouns cannot be -wa marked because they lack a DP layer. In other words, if -wa always associates with D, then -wa will not be available simply because inanimate noun phrases are never DPs. 17

3.4 Singular -yi associates with Num

Finally, we consider the remaining nominal suffix, -yi. According to Frantz (2009), there are two homophonous -yi suffixes: (i) an obviative suffix that only appears on singular, non-proximate, animate nouns and (ii) a singular suffix that only appears on non-plural inanimate nouns. However, we propose that, in fact, there is only one suffix -yi, and that it is a singular suffix that associates with the functional category Num. 18 Our reasoning is as follows: First there are good reasons to rule out the possibility that -yi associates with other functional heads in the noun phrase. In particular, -yi cannot associate with D because it appears on inanimate nouns, and by hypothesis, inanimates do not have a DP layer. Similarly, -yi cannot associate with iAsp because it cannot appear in small nominals, which are iAspPs.

Second, since -yi is an overt marker of singular number, we assume that its formal feature content is [+singular], which is compatible with the interpretive function of the head Num. 19 This leads to the prediction that when -yi is absent, Num associates with [-singular], expressing a contrast between singular and non-singular nominals. This prediction is borne out. Earlier, we saw that, at the level of iAsp, overtly marked plural nouns contrast with number neutral bare nouns. Now observe that at the level of NumP, overtly marked plural nouns contrast with overtly marked singular ones. Kim and Ritter (2014) attribute these different contrasts to the fact that both [-singular] in Num and [plural] in iAsp give rise to a plural interpretation. As a result, plural marked nouns are in fact structurally ambiguous – in the context of small nominals they are of the category iAspP, but elsewhere they are DPs if the head noun is animate and NumPs if the head noun is inanimate. This is schematized in (23).


It remains to be explained why proximate, singular, animate nouns always bear the suffix -wa, and never -yi. We attribute this to the assumption that -wa and -yi compete for a single slot in the noun’s morphological template, and thus these exponents are subject to post-syntactic spell-out restrictions (see Section 3.1). These spell-out restrictions will also account for the fact that plural animate nouns always bear the plural animate marker -iksi, regardless of their proximate specification. This suggests that the spell-out restrictions rank -iksi above -wa, and -wa above -yi. These rankings have the following results: -wa is spelled out in the context of -yi, but not -iksi; and -yi is only spelled out if it is not in competition with another overt marker.

3.5 Summary

We have argued that there are (at least) three nominal functional categories – D, Num and iAsp, and that each of the nominal inflectional suffixes associates with one of these categories. The animate (plural) suffix -iksi and the inanimate (plural) suffix -istsi both associate with iAsp; the singular suffix -yi associates with Num; and the proximate suffix -wa associates with D. Their distribution is based in part on differences in their availability in small nominals, and in part on the interpretive function of the different suffixes. This is summarized in Table 5 .

Table 5:

The distribution of features across the spine.

4 Analysis Part II: H-animacy is a property of arguments

Recall that the class of animate nouns can be subdivided into those that denote humans or humanoid beings on the one hand, and those that denote inanimate objects, on the other. In particular, the former class, which we refer to as H-animates are a proper subset of the class of morphologically specified animate nouns, which we refer to as m-animates. This is illustrated in Figure 2.

H-animate nouns are a subset of m-animate nouns.
Figure 2:

H-animate nouns are a subset of m-animate nouns.

In this section we turn to the analysis of H-animacy. Particularly, in order to account for the distinctive properties of H-animate DPs, we propose that they are the only class of nominals that contain an abstract argument whose referent is restricted to human(oid)s. In other words, the unique properties of H-animate DPs is not attributable to a difference in morpho-syntactic head categories or head features, but rather to the content of an argument of one of those heads.

We begin by laying out our assumptions about universals of phrase structure, and in particular the structure of noun phrases (Section 4.1). We then introduce a formal analysis of H-animacy, in which we postulate an H-index that restricts referents to human(oid)s (Section 4.2). Finally, we move on to discuss the relation between H-animacy and m-animacy (Section 4.3).

4.1 Background: the universal structure of categories

We assume, following Wiltschko (2014), that, universally, functional categories are transitive predicates which establish a relation between two abstract arguments as schematized in (24) (see also Speas 2010).


This idea has its roots in the work of Demirdache and Uribe-Extebarria (1997, 2000), who propose that the clausal functional heads, Tense and Aspect, are transitive predicates that relate arguments denoting times by a relation of coincidence (in the sense of Hale, 1986). Based on this work, Ritter and Wiltschko (2009, 2014) (henceforth R&W) analyze present tense as a two-place predicate that asserts that the reference time (an abstract argument associated with SpecAspP) coincides with the utterance time (an abstract argument associated with SpecTP). In contrast, past tense relates the same two abstract arguments, but asserts that they do not coincide. According to R&W’s formal implementation, the functional category (in this case INFL) comes with an unvalued coincidence feature [ucoin]. This feature is valued by the substantive content of the morphology that associates with INFL (i.e. tense morphology in English).


To accommodate languages in which INFL has spatial or participant-based content, R&W further argue that the abstract arguments ordered by INFL are not intrinsically temporal. Instead they suggest that these arguments are situation arguments, which contain times, places, and individuals. Depending on the substantive content a language uses to value the coincidence feature, a particular aspect of the abstract argument is highlighted. Accordingly, all clausal functional heads are analyzed as predicates that relate abstract situation arguments. 20

Wiltschko (2014) extends this analysis to nominal functional categories. There are two differences between nominal and clausal functional categories: i) the unvalued feature in the head of the category is [coincidence] for clausal categories, but [identity] for nominal categories (cf. Baker 2005); ii) the abstract arguments are situation arguments for clausal categories and individual arguments for nominal categories. This is summarized in Table 6.

Table 6:

The difference between clausal and nominal categories.

The morphemes that associate with the nominal categories value the unvalued [identity] feature, thereby establishing a relation between the two abstract individual arguments. Moreover, just as there are different types of situation arguments in the various categories of the clausal spine, there are different types of individual arguments in the various categories of the nominal spine. Of particular relevance for the current discussion are the arguments associated with the nominal category D. Following Wiltschko (2014), we assume that the abstract argument associated with the specifier of DP is an utterance individual, defined as in (26). 21


Uttind=def an individual associated with a file-card at the utterance situation.

Wiltschko, 2014: 222 (58)

This argument is ordered relative to the reference individual associated with NumP, as illustrated in (27).


With these assumptions in place we now turn to a formal analysis of H-animacy.

4.2. H-animacy as a restriction on abstract individual arguments

We assume that the structure in (27) is universal, and propose that restrictions may be added to the abstract arguments within this structure. In particular, we propose that the utterance individual, i.e. the abstract argument in SpecDP, may be restricted to humans and humanoids. We implement this by associating an index H to this argument, as shown in (28). Thus, when a functional category selects an H-animate argument it selects a DP with an H-indexed utterance individual in SpecDP. 22


This amounts to saying that there is a restriction on the utterance individual associated with DP: it must denote a human or humanoid individual. (In Blackfoot this includes animals and spirits). It is the utterance individual in SpecDP (along with its restriction) that serves as a selectable property. In particular, we have seen that in Blackfoot only H-animate DPs can be external arguments of certain clausal functional heads (see Section 2). 23 Essentially then, our proposal is that these functors select H-animate DPs as illustrated in (29).


Selecting H-animates in Blackfoot

This accounts for the distribution of H-animates in Blackfoot. Unlike m-animacy, which is a property of nominal heads, H-animacy is a property of nominal phrases. In our analysis, this is implemented by means of an H-index, which is in turn associated with the abstract argument in SpecDP: the utterance individual. Since these abstract arguments are, by hypothesis, associated with specifier positions, it follows that any restriction placed on them will necessarily be a restriction on the entire phrase rather than a head. Hence, H-animacy does not display any of the distinctive properties of head-features, such as triggering agreement or concord. We thus predict that there is no direct syntactic relation between m-animate nouns and H-animate arguments. Nevertheless, the two notions interact with each other in predictable ways, as we now show.

4.3 The relation between H-animacy and m-animacy

As discussed in Section 3.3, we assume that in Blackfoot, the overt proximate marker -wa associates with the category D. Here we argue that -wa serves to value the unvalued identity feature as [+ident], thereby asserting that the referent is identical to an utterance individual (i.e., an individual with a file-card at the time of utterance). It contrasts with a zero morpheme, which serves to value the identity feature as [-ident] thus deriving non-proximate (i.e., obviative) arguments. This is illustrated in (30).


Note that the source of the H-index on the utterance individual (Uttind) is not in any way tied to morphological animacy. The animacy feature associated with iAsp (see Section 4.2) does not enter into a syntactic relation with the abstract argument in SpecDP. Instead, we assume that the presence of the H-index is purely a matter of semantic interpretation: nouns that denote human(oid) referents are compatible with an H-index in SpecDP. Thus, if the noun denotes a human (e.g, ‘girl’), then the utterance individual associated with the functional architecture above that noun will denote a human referent as well.

This analysis allows us to understand why human(oid) nouns are a proper subset of m-animate nouns. Recall that m-animacy is represented as a formal feature associated with inner Aspect (Section 3.2). Since all nouns that denote human(oid) beings are classified as [+animate] it follows that all H-animates are also [+animate]. 24 However, it is not the case that all nouns that denote inanimate objects are classified as [-animate]. As discussed in Section 2.1, some nouns that denote inanimate objects are (arbitrarily) classified as [+animate]. While such nouns still contain a DP-layer, the abstract utterance individual argument in SpecDP does not receive the H-index because it does not denote a human(oid) being. In other words, the presence of the H-index depends on the real world referent of the noun rather than its morphological classification. Hence, it follows that H-animate nouns are a proper subset of [+animate] nouns.

Finally, unsurprisingly nouns that are classified as [-animate] are not associated with an H-index. This is because the referent introduced by such a noun denotes an inanimate individual. We have seen evidence that [-animate] nouns do not have proximate/obviative marking, and hence lack the DP-layer, which is the locus of H-indexation. For these reasons, m-inanimate nouns, just like m-animate nouns that do not denote human(oid) referents, cannot appear in the specifier positions of v, Asp, Appl or p, as these positions are restricted to DPs with an H-index on their utterance individual arguments.

There is, however, one exception to the generalization that H-animates are a proper subset of m-animates. In particular, in descriptions of fictional worlds where inanimate objects can think, feel and act, we find that even nouns that are classified as [-animate] can have the distribution of H-animate arguments. In this context, consider the example in (31), which describes some imaginary smiling flowers:


‘These flowers are happy.’

Here, the verb ‘be happy’, which is normally predicated of an H-animate subject, takes as its subject an argument containing an inanimate noun. Note crucially that the noun remains classified as [-animate] as evidenced by the fact that the noun is suffixed by the inanimate plural marker -istsi. We treat this as a case of discourse determined H-indexation, i.e., an instance of coercion. That is, while H-animacy is typically determined by the denotation of the noun, it may also be exceptionally determined by contextual reference. This can be understood if we assume that this form of coercion is structurally represented. Whenever a normally inanimate nominal is treated as an H-animate argument, a DP layer containing an H-indexed utterance individual is added.

As long as the referent of the DP is considered to be a mental state holder the abstract argument will be associated with the H-index, even if this is not compatible with the typical referent of the noun contained in the phrase. This is further evidence for the assumption that m-animacy and H-animacy are independent of each other. H-animacy depends on real world knowledge, m-animacy is a formal classificatory feature.

On this view, it is a coincidence that Blackfoot has both m-animacy as well as H-animacy. These two types of animacy play different roles in the grammar: m-animacy as a head feature participates in head-syntax (agreement and concord), while H-animacy is a property of arguments. In Blackfoot, H-animacy plays a crucial role in the licensing of arguments in the specifier of functional categories. 25 If this analysis is on the right track, we predict that even in languages where m-animacy plays no role, H-animacy may play a role. We show in the following section that this prediction is indeed borne out.

5 Beyond Blackfoot

The role of animacy in natural language is a matter of debate and its status within grammar is still far from settled. Consider Dahl and Fraurud’s (1996: 47) opening statement: “Animacy, or the distinction between animate and inanimate entities, is so pervasive in the grammars of human languages that it tends to be taken for granted and become invisible.” What is it that gives us the impression that it is simultaneously pervasive and invisible? We suspect that it has to do with its dual nature. On the one hand, animacy may serve as a morpho-syntactic feature that serves to classify nouns. Based on languages such as Blackfoot, in which this is the case, we are lead to conclude that animacy is indeed a property of grammar. And in such languages it is a pervasive property in the sense that it does pervade the entire grammar. As we have seen throughout this paper, animacy plays not only a role in nominal classification, but also affects other areas of the grammar: plural marking, agreement within the nominal phrase, classification of predicates, etc. Thus, animacy marking in Blackfoot is far from invisible. What may turn it into an invisible property is the fact that, as a feature of the classifying functional category, it cannot be reduced to semantic or ontological properties. Given that there are mismatches between real-world animacy and grammatical animacy, one may conclude that the ontological notion animacy does not, in fact, play a role here. However, at the same time, we have seen evidence that ontological animacy does plays a role elsewhere in the grammar of Blackfoot. It is pervasive in that it affects nominals that occupy the specifier position of a variety of clausal functional categories. But, at the same time, it appears to be invisible in that it has not figured prominently in the way the Blackfoot grammatical system is traditionally described. This may be due to the fact that animacy as a feature that determines nominal classification does not align with animacy as a restriction on arguments. Nevertheless, we have argued that both notions of animacy (m-animacy and H-animacy) play a role in the grammar of Blackfoot, albeit in different ways. Thus, the ‘invisibility’ of animacy in the grammar of natural languages may have two sources. When it serves as a morpho-syntactic head feature it need not reflect ontological animacy, and when it serves as a restriction on arguments it is not formalized as a morpho-syntactic feature.

Consider in this context the role animacy plays in English. In a recent paper, Folli and Harley (2008) argue that animacy (more precisely humanness) is a property of referents, not Ns or DPs. Hence they conclude that it must be outside of grammar (i.e., outside of narrow syntax; cf. also Ramchand 2008). Let us look at some effects of animacy in English. With some exceptions, agent and experiencer roles must be instantiated by noun phrases denoting animate individuals. Thus, we find the contrast in (32) where the subject of the psych predicate scared must be animate.

  • a.

    The boy was scared.

  • b.

    #The tree was scared.

We also find a similar contrast with one type of possessor role. While inalienable possessors may be animate, as in (33a), or inanimate, as in (33b), this is not so for alienable possessors. Rather with alienable possession, the possessor has to be animate as shown in (34). With animate possessors, the sentence is well-formed (34a); with inanimate possessors, the sentence is degraded (34b). To rescue this structure, a locative phrase (in it) has to be added (34c).

  • a.

    John has a broken arm.

  • b.

    The oak tree has many branches.

  • a.

    John has a bird.

  • b.

    #The oak tree has a family of birds.

  • c.

    The oak tree has a family of birds in it.

What is striking about this pattern in English is that it is completely based on real world knowledge. That is, the judgments reported in (33) and (34) hold for the actual world only. In fictional worlds where trees are anthropomorphized, (33b) and (35b) are well-formed. This means that the well-formedness of these examples depends on the context, not on their grammatical expression. Hence, it comes as no surprise that animacy in English has been treated as a property outside of narrow syntax.

However, based on Blackfoot, we have shown that even the kind of animacy that is sensitive to context has a reflex in the grammar (in the form of an H-index, which serves to restrict the argument fulfilling a given role). The evidence suggests that several functional categories in Blackfoot select arguments with an H-index. This finding leads us to conclude that H-indices are part of universal grammar. This conclusion is further supported by the observation that the set of functional categories that require arguments with an H-index differs from language to language. For example, in Blackfoot functors that select agents and experiencers, point of view holders, benefactives, themes of goal-directed motion and possessors all select H-animate arguments. In contrast, in English only those functors that select experiencers and possessors do so. If H-animacy were indeed outside the grammar but instead rooted in the way we perceive the world, we would not expect this type of cross-linguistic variation.

A similar point can be made based on Spanish. In this language, H-animacy underlies differential object marking. Direct objects that have a human referent are marked with the dative preposition as in (35a); otherwise they are unmarked as in (35b).


‘Juan saw a girl.’


‘Juan saw a book.’

But as in English and Blackfoot, what counts as H-animate, is contextually determined. This is illustrated by the minimal pair in (36). If death is conceptualized as personified, it is preceded by the case-marker a (36a); in contrast, if death is conceptualized as an abstract concept, then it is unmarked as in (36b). 26


‘S/he called out to death.’


‘S/he called out to death.’

Hanssen (1945): 296

Thus, there is no doubt that whether a DP is H-animate depends on the referent. In other words, this determination is extra-grammatical. At the same time however, the Spanish facts indicate that H-animacy must be visible to the grammar because it determines case-marking, an unambiguously grammatical property. Extending the treatment of H-animacy in Blackfoot to Spanish, allows us to explain the Spanish pattern. In particular, we suggest that a-marked arguments occupy a different position than unmarked arguments. And in addition, the specifier position that a-marked arguments occupy appears to be restricted to H-animates.

Thus, there are two differences between Blackfoot and Spanish. On the one hand, Spanish lacks m-animacy (i.e., Spanish nouns are not classified based on animacy). On the other hand, in Spanish, only the specifier position hosting a-marked arguments is restricted to H-animates, whereas in Blackfoot specifiers of several functional heads are restricted to H-animates. Whether or not this analysis of animacy in Spanish can be independently supported remains to be determined. However, we contend that the logic of our analysis may shed new light on other familiar animacy effects, most notably those that have been described in terms of the animacy hierarchy (see Section 2.1).

6 Conclusion

Based on data from Blackfoot, we have argued that there are two qualitatively different notions of animacy that may play a role in the languages of the world. M(orphological)-animacy is a head feature which is responsible for agreement. It serves as a nominal classification device. Given the assumption that the substantive content that associates with functional categories is not universally fixed, it follows that not all languages make use of m-animacy. In particular, we have assumed, following Wiltschko (2012), that animacy is the Blackfoot counterpart of boundedness, the feature that serves to classify nouns as mass or count in languages like English.

The second type of animacy (H-animacy) we have discussed here is a property of arguments. We contend that H-animacy is a universal property. Its effects are not dependent on the presence of m-animacy. As a result, there are languages without m-animacy, but which still display evidence for H-animacy. Moreover, even in a language where m-animacy plays a role (such as Blackfoot), we observe that m-animate nouns do not always form H-animate arguments.

What our exploration of Blackfoot has revealed is that the conceptual category human is special in that it is visible to the narrow syntax. In contrast, other conceptual categories, such as self-propelled, are not visible to the narrow syntax. They may however be the basis for morphological classification. We speculate that the reason for the special status of humans in grammar is the fact that humans are the only individuals that can bear both event roles and speech act roles.


  • Adger, David. 2003. Core syntax: A minimalist approach. Oxford: Oxford University Press. Google Scholar

  • Barrie, Michael. 2009. The locality of wh-feature checking in Blackfoot. Paper presented at Minimalist approaches to syntactic locality, Hungarian Academy of Sciences, Budapest, Hungary, 26–28 August. 

  • Baker, Mark. C. 2005. Lexical Categories. Cambridge: Cambridge University Press. Google Scholar

  • Béjar, Susana 2003. Phi-syntax: A theory of agreement. Toronto: University of Toronto Dissertation.Google Scholar

  • Bliss, Heather. 2005. Topic, focus and point-of-view in Blackfoot. West Coast Conference on Formal Linguistics (WCCFL) 24. 61–69. Google Scholar

  • Bliss, Heather. 2010. Argument structure, applicatives and animacy in Blackfoot. In Heather Bliss & Raphael Girard (eds.), Proceedings of WSCLA 13 & 14: Workshop on the structure and constituency of languages of the Americas. UBC Working Papers in Linguistics 26, 58–69. Vancouver: UBC Department of Linguistics. Google Scholar

  • Bliss, Heather. 2013. The Blackfoot configurationality conspiracy: Parallels and differences in clausal and nominal structures. Vancouver: UBC dissertation. Google Scholar

  • Bliss, Heather, Elizabeth Ritter & Martina Wiltschko. 2014. A comparative analysis of theme marking in Blackfoot and Nishnabemwin. In Monica Macaulay & Rand Valentine (eds.), Papers of the forty-second Algonquian conference, 10–33. Albany: SUNY Press. Google Scholar

  • Bloomfield, Leonard. 1946. Algonquian. In Harry Hoijer & Cornelius Osgood (eds.), Linguistic Structures of Native America [Viking Fund Publications in Anthropology 6], 85–129. New York: Viking Fund Publications. Google Scholar

  • Borer, Hagit. 2005. In name only. New York: Oxford University Press.Google Scholar

  • Brousseau, Kevin. 2009. Medials in the historical Cree-Montagnais-Naskapi dialect of Nehirawewin. Montreal: Université du Québec à Montréal MA thesis. Google Scholar

  • Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Roger Martin, David Michaels & Juan Uriagereka (eds.), Step by step: Minimalist essays in honor of Howard Lasnik, 89–155. Cambridge, MA: MIT Press. Google Scholar

  • Corbett, Greville. 1991. Gender. Cambridge, UK: Cambridge University Press. Google Scholar

  • Croft, William H. 2003. Typology and universals, 2nd edn. Cambridge: Cambridge University Press. Google Scholar

  • Dahl, Östen & Kari Fraurud, 1996. Animacy in grammar and discourse. In Thorstein Fretheim, & Jeanette K. Gundel, (eds.), Reference and referent accessibility, 47–64. Amsterdam: John Benjamins. Google Scholar

  • Dahlstrom, Amy. 1995. Motivation vs. predictability in Algonquian gender. In David H. Pentland (ed.), Papers of the twenty-sixth Algonquian conference, 52–66. Winnipeg: University of Manitoba. Google Scholar

  • Darnell, Regna & Anthony L. Vanek. 1976. The semantic basis of the animate/inanimate distinction in Cree. Papers in Linguistics 9(3–4). 159–180.Google Scholar

  • Demirdache, Hamida & Myriam Uribe-Extebarria. 1997. The syntax of temporal relations: A uniform approach to tense and aspect. In Emily Curtis, James Lyle & Gabriel Webster, (eds.), Proceedings of WCCFL 16, 145–159. Stanford, CA: CSLI Publications.

  • Demirdache, Hamida & Myriam Uribe-Extebarria. 2000. The primitives of temporal relations. In Roger Martin, David Michaels & Juan Uriagereka (eds.), Step by step: Essays on minimalist syntax in honour of Howard Lasnik, 157–86. Cambridge, MA: MIT Press. Google Scholar

  • Dixon, Robert M. W. 1979. Ergativity. Language 55(1). 59–138.Google Scholar

  • Folli, Raffaela & Heidi Harley. 2008. Teleology and animacy in external arguments. Lingua 118(2). 190–202. Google Scholar

  • Frantz, Donald G. 1970. Toward a generative syntax of Blackfoot. Edmonton: The University of Alberta dissertation. Google Scholar

  • Frantz, Donald G. 2009. Blackfoot Grammar, 2nd edn. Toronto: University of Toronto Press. Google Scholar

  • Frantz, Donald G. & Norma Jean Russell. 1995. Blackfoot dictionary of stems, roots, and affixes. Toronto: University of Toronto Press. Google Scholar

  • Genee, Inge 2009. What’s in a morpheme? Obviation morphology in Blackfoot. Linguistics 47(4). 913–944. Google Scholar

  • Goddard, Ives. 2002. Grammatical gender in Algonquian. In H. C. Wolfart (ed.), Papers of the thirty-third Algonquian conference, 195–231. Winnipeg: University of Manitoba. Google Scholar

  • Greenberg, Joseph. H. 1954. Concerning inferences from linguistic to nonlinguistic data. In Harry Hoijer (ed.), Language in culture: Proceedings of a conference on the interrelations of language and other aspects of culture, 3–19. Chicago: University of Chicago Press. Google Scholar

  • Hanssen, Frederico. 1945. Grammática histórica de la lengua castellana. Buenos Aires: El Ateneo. Google Scholar

  • Heim, Irene. 1982. The semantics of definite and indefinite noun phrases. Amherst, MA: University of Massachusetts Amherst dissertation. Google Scholar

  • Hockett, Charles F. 1966. What Algonquian is really like. International Journal of American Linguistics 32(1). 59–73.Google Scholar

  • Johansson, Sara. 2013. A participle account of Blackfoot relative clauses. Canadian Journal of Linguistics 58(2). 217–238. 

  • Joseph, Brian D. 1979. On the animate-inanimate distinction in Cree. Anthropological Linguistics 21(7). 351–354. Google Scholar

  • Kim, Kyumin. 2014. Refining the syntax of non-core arguments: P, applicative, and functional p. In Jyoti Iyer & Leland Kusmer (eds.) North east linguistic society (NELS) 44. 205–219. 

  • Kim, Kyumin. 2015. Spatial PPs and the structure of motion verbs in Blackfoot. In Natalie Weber and Sihwei Chen (eds.) Proceedings of the 18th and 19th Workshop on Structure and Constituency in the Languages of the Americas – Papers for WSCLA 18 & 19, UBC Working Papers in Linguistics 39: 125–139.

  • Kim, Kyumin & Elizabeth Ritter. 2014. The categories of number in Blackfoot. Paper presented at the 46th Algonquian Conference, Uncasville, CT, 23–26 October. 

  • Kuhlmeier, Valerie A., Paul Bloom & Karen Wynn. 2004. Do 5-month-old infants see humans as material objects? Cognition 94(1). 95–103. Google Scholar

  • Leonard, Wesley Y. 2007. Miami language reclamation in the home: A case study. Berkeley: University of California Berkeley dissertation.Google Scholar

  • Lomashvili, Leila. 2011. Complex predicates: The syntax-morphology interface. Amsterdam: John Benjamins. Google Scholar

  • Louie, Meagan. 2008. Atemporal anchoring of individuals, events and sub-events in Blackfoot. Toronto: University of Toronto MA thesis. Google Scholar

  • McGinnis, Martha. 2008. Phi-feature competition in morphology and syntax. In David Harbour, David Adger & Susanna Béjar (eds.), Phi theory: Phi-features across modules and interfaces, 155–184. Oxford: Oxford University Press. Google Scholar

  • Piriyawiboon, Nattaya. 2007. Algonquian obviation reanalysis. 2007 Canadian Linguistics Association (CLA) Proceedings. http://homes.chass.utoronto.ca/~cla-acl/actes2007/Piriyawiboon.pdf (accessed 8 August 2015).

  • Piriyawiboon, Nattaya. 2008. Reconsidering the obviative. In Seok Koon Chin & Fusheini Hudu (eds.) Proceedings of the 12th Workshop on Structure and Constituency in the Languages of the Americas (WSCLA XII), UBC Working Papers in Linguistics 21, 100–113. 

  • Pylkkänen, Lina. 2008. Introducing arguments. Cambridge: MIT Press. Google Scholar

  • Quinn, Connor McDonough. 2006. Referential-access dependency in penobscot. Cambridge, MA: Harvard University dissertation. Google Scholar

  • Ramchand, Gillian C. 2008. Verb meaning and the lexicon: A first phase syntax. Cambridge: Cambridge University Press. Google Scholar

  • Reinhart, Tanya. 2002. The theta system–an overview. Theoretical linguistics, 28(3), 229–290. Google Scholar

  • Rhodes, Richard A. 2010. Relative root complement: a unique grammatical relation in Algonquian syntax. In Michael Cysouw, Orin Gensler, and Jan Wohlgemuth (eds.), Rara & rarissima: Documenting the fringes of linguistic diversity (Empirical Approaches to Language Typology), 305–324. Berlin: de Gruyter. Google Scholar

  • Rijkhoff, Jan N. M. 1990. Nominal Aspect. Journal of Semantics 8(4) 291–309. Google Scholar

  • Ritter, Elizabeth. 2014a. Argument structure inheritance and the structure of nominalized DPs. In Ileana Paul (ed.), Cross-linguistic investigations of nominalization patterns, 25–50. Amsterdam: John Benjamins. Google Scholar

  • Ritter, Elizabeth. 2014b. Featuring animacy. Nordlyd 41(1). 103–124.Google Scholar

  • Ritter, Elizabeth & Sara Thomas Rosen. 2010. Animacy in Blackfoot: Implications for event structure and clause structure. In Malka Rappaport-Hovav, Edit Doron & Ivy Sichel, (eds.), Lexical semantics, syntax, and event structure, 124–152. New York: Oxford University Press.Google Scholar

  • Ritter, Elizabeth & Martina Wiltschko. 2009. Varieties of INFL: TENSE, LOCATION, and PERSON. In Jeroen van Craenenbroeck (ed.), Alternatives to cartography, 153–202. Berlin: Mouton de Gruyter. Google Scholar

  • Ritter, Elizabeth & Martina Wiltschko. 2014. The composition of INFL: An exploration of tense, tenseless languages, and tenseless constructions. Natural Language & Linguistic Theory 32(4). 1331–1386.Google Scholar

  • Ritter, Elizabeth & Martina Wiltschko (in prep.) Case, humanness and argument-licensing. 

  • Rivero, Maria Luisa. 2005. Topics in Bulgarian morphology and syntax: a minimalist perspective. Lingua 115(8). 1083–1128. Google Scholar

  • Silverstein, Michael, 1976. Hierarchy of features and ergativity. In R. M. W. Dixon (ed.), Grammatical categories in Australian languages, 112–171. Canberra: Australian Institute of Aboriginal Studies. Google Scholar

  • Speas, Peggy. 2010. Evidentials as generalized functional heads. In Anna-Maria Di Sciullo & Virginia Hill (eds.), Edges, heads and projections: Interface properties, 127–150. Amsterdam: John Benjamins. Google Scholar

  • Taylor, Allan R. 1969. A grammar of Blackfoot. Berkeley: University of California dissertation. Google Scholar

  • Tenny, Carol L. 2006. Evidentiality, experiencers and the syntax of sentience in Japanese. Journal of East Asian Linguistics 15(3). 245–288. Google Scholar

  • Travis, Lisa. 2010. Inner aspect: The articulation of VP. Dordrecht: Springer.Google Scholar

  • Trommer, Jochen 2006. Plural insertion is constructed plural. In Gereon Müller & Jochen Trommer (eds.), Subanalysis of argument encoding in distributed morphology, 197–228. (Linguistische Arbeits Berichte 84). Leipzig: Universität Leipzig. Google Scholar

  • Wiltschko, Martina & Elizabeth Ritter. 2014. Animating the narrow syntax. Paper presented at GLOW 37. Brussels, Belgium. 

  • Wiltschko, Martina. 2009. What’s in a determiner and how did it get there? In Jila Ghomeshi, Ileana Paul & Martina Wiltschko (eds.), Determiners: universals and variation, 25–66. Amsterdam: John Benjains. Google Scholar

  • Wiltschko, Martina. 2012. Decomposing the count-mass distinction: Evidence from languages that lack it. In Diane Massam (ed.), Count and mass across languages, 146–171. Oxford: Oxford University Press. Google Scholar

  • Wiltschko, Martina. 2014. The universal structure of categories: Towards a formal typology. Cambridge: Cambridge University Press.Google Scholar


  • 1

    Humanoid beings include spirits, gods, pets and anthropomorphized objects that are ascribed the capacity to hold mental states. 

  • 2

    Blackfoot is a typical Algonquian language in that the most reliable diagnostic of noun classification is the form of nominal inflection that appears on the noun itself, and on any DP internal modifiers. See Section 3.2 below for discussion. 

  • 3

    The set of modifiers includes demonstratives, numerals and relative clauses (cf. Frantz 2009, Johansson 2013, Bliss 2013 and Kim and Ritter 2014). Plural marking does not appear on adjectives because Blackfoot lacks this category. Rather, adjectival notions are either expressed as verbs (ia) or as noun stem internal modifiers (ib).



    ‘be.tall (predicated of inanimate/animate subject)’

    Frantz and Russell 1995: 227




    ‘little shields’

    Frantz and Russell 1995: 16


  • 4

    Unless otherwise indicated, all data are from original fieldwork by the authors. Abbreviations are as follows: 1 = 1st person; 3 = 3rd person; 3:1 = 3rd person acting on 1st person; ai = animate intransitive verb; an = animate; ben = benefactive; dat = dative; dem = demonstrative; dir = direct; dur = durative; evid = evidential; fut = future; in = inanimate; inst = instrumental; intns = intensifier; inv = invers; ipfv = imperfective; obv = obviative; pl = plural; prn = pronoun; prox = proximate; pst = past; purp = purposive; sg=singular; ta = transitive animate verb; ti = transitive inanimate verb; x = impersonal. 

  • 5

    Focusing specifically on common nouns, the animacy hierarchy is generally assumed to distinguish at least between the three lexical semantic classes listed in (2). However, as originally formulated in Silverstein (1976), the animacy hierarchy is normally extended to include other types of nominals, notably personal pronouns and proper names, as shown in (i).(i) 1st/2nd person pronouns > 3rd person pronouns > proper names > human common Ns > non-human animate common Ns > inanimate common Ns(Dixon 1979: 85, as cited in Croft 2003: 130)Thus, it is unsurprising that in Blackfoot, both pronouns and proper names pattern with human and animate common nouns (Bliss 2005, Frantz 2009). 

  • 6

    See Wiltschko and Ritter (2014) for a quantitative analysis of noun classification in Frantz and Russell’s (1995) Blackfoot Dictionary of Stems, Roots, and Affixes that is consistent with the generalizations discussed here. 

  • 7

    Theme sign is a traditional Algonquianist term for the morpheme that appears immediately to the right of the transitive verb stem. When the object is morphologically inanimate, the form of the theme sign is determined by the person feature of the subject. When the object is morphologically animate, the form of the theme sign depends on features of both the subject and the object. In the latter case, the term direct/inverse marking is also used. See Section 2.2.2 for further discussion. We follow Frantz (2009: 44) in analysing ‘p as a theme sign. An anonymous reviewer points out that cognate morphemes in other Algonquian languages are not theme signs, and that it is likely that ‘p has undergone some kind of diachronic shift. 

  • 8

    Bliss (2005: 67) also provides arguments that 1st person outranks 2nd person. Since this is not relevant to the discussion in this section we omit this detail. Bliss proposes the hierarchy in (i) to account for both the sentience and person facts.


    1 > 2 > 3sentient > 3non-sentient

    The hierarchy in (10) departs from (i) in order to additionally account for the fact that in clauses with both a 3rd person sentient (H-animate) subject and a 3rd person sentient (H-animate) object, direct/inverse marking is also sensitive to obviation (cf. Frantz 2009: 52–63 for discussion.) 

  • 9

    An anonymous reviewer points out that the restriction to H-animate arguments in SpecAspP is essentially different from those in the other functional categories. In particular, s/he points out that while the other categories are all associated with their specifiers via external merge, SpecAsp is filled via internal merge (i.e., movement). We note that it is not unusual for functional categories to impose restrictions on their specifiers even if they are filled via internal merge (i.e., movement). Wh-movement is an example where SpecCP has to be filled by a particular type of phrase, i.e., a wh-phrase.Moreover, under minimalist assumptions, restrictions on a specifier that is filled by movement are in fact expected. In particular, it is assumed that both selectional restrictions and movement may be derived by means of an uninterpretable feature (see Adger 2003 for the former and Chomsky 2000 for the latter). Everything else being equal, there is no reason why the semantic roles associated with externally merged arguments in SpecvP or SpecApplP should differ from the grammatical (aspectual) roles associated with internally merged arguments in SpecAspP. 

  • 10

    Given that v, Asp, Appl and p all require an H-animate specifier the question arises as to whether all functional heads in the language impose this selectional requirement on their specifier. This appears not to be the case as non-H-animate DPs may undergo wh-movement, to SpecCP and object movement to the specifier of a vP internal functional head (cf. Barrie 2009, Ritter 2014a). 

  • 11

    The term nominal aspect (or Seinsart) is due to Rijkhoff (1990), who proposes that this is the nominal counterpart of verbal aspect, and that it underlies the mass/count distinction. Wiltschko (2012) formalizes Rijkhoff’s insight, by introducing a category, inner Aspect into the nominal spine. She hypothesizes that inner Aspect serves to classify nouns as mass or count when it has the interpretable feature [±bounded], but that it may alternatively be specified with the feature [±animate], in which case it classifies nouns as animate or inanimate. Both in terms of its interpretive content and its position in the structure, inner Aspect specified for [±bounded] is similar to Borer’s (2005) DivP. Here, we adopt the label inner Aspect to emphasize the parallel with verbal inner Aspect, and to reflect our assumption that the interpretable feature content of this category is subject to cross-linguistic variation. 

  • 12

    Similar proposals have been made for other related and unrelated languages. For example, Bejar (2003) and McGinnis (2008) both propose that in Algonquian languages multiple agreement controllers compete for the same agreement slot. Similarly, both Rivero’s (2005) analysis of Bulgarian clitics and Lomashvili’s (2011) analysis of Georgian complex predicates require that different syntactic constituents compete for spell-out in a single morphological slot. Trommer (2006) also assumes that the mapping from syntax to morphology is many to one in his exploration of a range of genetically unrelated languages in which a single vocabulary item combines the content of two syntactic heads. 

  • 13

    Descriptive grammars of Blackfoot refer to this class of verbs as pseudo-intransitive (Taylor 1969) or paratransitive (Frantz 2009). Their objects must be non-specific, and consequently they are incompatible with demonstrative determiners. Bliss (2013) shows that the semantic and syntactic restrictions on these arguments are due to the fact that they are pseudo-incorporated into the verb. 

  • 14

    This constitutes a departure from the view that plural marking in Blackfoot associates with Num (e.g. Bliss, 2013, Ritter 2014b, Wiltschko, 2012). Bliss (2013) argues that what we are calling small nominal objects are pseudo-incorporated into the AI verb, and that such objects are bare NumPs if plural-marked, and bare NPs otherwise. We note that the empirical generalizations reported in Bliss differ from those in Kim and Ritter (2014). It may be the case that this difference reflects a dialectal difference or else is a sign of language change (the speakers consulted for the latter paper are both speakers of new Blackfoot). 

  • 15

    In other words, we assume that iAsp without a modifying plural feature refers to a set of atoms plus the sums/sets that can be created from those atoms (*P), while [plural] iAsp refers to the subset of *P consisting only of sums/non-singleton sets (*P-A). Thanks to an anonymous reviewer for this clarification. 

  • 16

    We use the term non-proximate, rather than Frantz’s term, obviative to describe [-proximate] animate nominals. Frantz further assumes the traditional Algonquianist distinction between obviative and further obviative nominals. This distinction applies in transitive clauses with two non-proximate DPs. In such clauses, if the verb has direct marking, then the subject is called obviative and the object further obviative, but if the verb has inverse marking it is the object which is obviative and the subject further obviative. Given that so-called further obviative DPs are formally identical to obviative ones, we assume that this distinction is not formalized with different features or feature values in D or any other nominal functional head. Further obviatives, like obviatives are simply non-proximate DPs, formally represented with the feature [-proximate] in D. 

  • 17

    In Section 4.3, we discuss an exception to this generalization. In fictional contexts, inanimate noun phrases can be used as subjects of experiencer and agentive predicates. It appears that in such contexts the inanimate noun phrase refers to an animated object or a humanoid being. This is structurally represented with an added DP layer. 

  • 18

    See Piriyawiboon (2007, 2008) for a different re-analysis of the so-called obviative suffix in Nishnaabemwin, a variety of the Algonquian language Ojibwe. As in Blackfoot, the Nishnaabemwin obviative suffix is homophonous with a suffix that functions as a number marker on inanimate nouns, and Piriyawiboon argues that they are in fact the same suffix. However, Nishnaabemwin differs from Blackfoot in that the suffix in question marks both singular and plural animate nouns as obviative, and it marks inanimate nouns as plural, rather than singular. Piriyawiboon claims that obviation in Nishnaabemwin is the result of a feature deletion operation that turns animate nouns into inanimate ones. While her analysis may account for the facts of Nishnaabemwin it cannot be extended to the Blackfoot facts considered here because Blackfoot -yi marked animates do not undergo a class shift. In other words, they consistently pattern with -wa marked animates and not yi- marked inanimates. 

  • 19

    On this analysis, [-singular] is the unmarked value for the head feature of Num. We assume that it is spelled-out as a phonetically null morpheme, and that it co-occurs with the plural markers in iAsp. In terms of their semantic interpretation, we assume that [+singular] denotes a set of atoms (A), while [-singular] denotes its complement, i.e. sums/non-singleton sets (*P-A). What this means is that [-singular] has the same denotation as the monovalent feature [plural], as defined in footnote 15. Thus, the difference between [-singular] and [plural] lies in the contrast set. In the first case, if the head is not specified as [-singular] it must be [+singular] and refer to atomic entities; in the second case, if the head is not specified as [plural] it has no number feature, and refers to both atomic and non-atomic entities. 

  • 20

    See Bliss, et al. and Wiltschko (2014) for a discussion of the functional category ASPECT. 

  • 21

    The notion of file-card here is in the sense of Heim (1982). The definition in (26) does not imply that inanimate referents cannot be associated with a file card. Rather it asserts only that utterance individuals must have a file card. (Referent individuals may also do so.) 

  • 22

    Note that the restriction that the utterance individual be H-indexed is independent of the value assigned to [uident]. In Blackfoot, for example, proximate marking values [uident] as [+ident], hence asserting that the reference individual is identical to the utterance individual. The latter may or may not be H-indexed. Similarly, the [-ident] value derived by non-proximate marking is compatible with an H-index on its abstract utterance argument, but does not require one. 

  • 23

    In the present paper we do not address the restriction on H-animates in SpecIP. See Ritter and Wiltschko 2014 for detailed discussion of this effect. 

  • 24

    An anonymous reviewer asks why nouns that denote humanoid entities are always [+animate] in Blackfoot. It appears that for this class of nouns the ontological properties of their nominal roots always determine the aspectual classification, cf. also Wiltschko (2012) for discussion. 

  • 25

    In Ritter and Wiltschko (in prep.) we develop the idea that H-animacy is Blackfoot’s version of case-licensing. 

  • 26

    We are grateful to Karen Zagona for pointing us towards this example. 

About the article

Published Online: 2015-11-24

Published in Print: 2015-12-01

Citation Information: The Linguistic Review, Volume 32, Issue 4, Pages 869–908, ISSN (Online) 1613-3676, ISSN (Print) 0167-6318, DOI: https://doi.org/10.1515/tlr-2015-0011.

Export Citation

©2015 Wiltschko and Ritter, published by De Gruyter. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License. BY-NC-ND 3.0

Citing Articles

Here you can find all Crossref-listed publications in which this article is cited. If you would like to receive automatic email messages as soon as this article is cited in other publications, simply activate the “Citation Alert” on the top of this page.

Kyumin Kim, Elizabeth Ritter, Martina Wiltschko, and Hotze Rullmann
Glossa: a journal of general linguistics, 2017, Volume 2, Number 1

Comments (0)

Please log in or register to comment.
Log in