Different types of animacy effects can be found in the languages of the world. 1 On the one hand, animacy properties of arguments may incur morphosyntactic marking, such as differential case-marking, i.e., when the absence or presence of a certain type of case is triggered by the animacy properties of arguments (see Malchukov 2008 for a cross-linguistic overview of these effects). As Malchukov (2008: 210) observes, “it is redundant to mark animacy per se (as it is clear from the lexical content of the nominal in question)”. This clearly distinguishes animacy from definiteness, because a distinction of definiteness can be the result rather than the trigger of differential case-marking (cf. de Swart and de Hoop 2007). On the other hand, languages may highlight certain degrees of animacy to make a distinction in prominence, with animate referents being conceptually more prominent than inanimate ones. In that sense, animacy is comparable to definiteness (cf. de Hoop 2009), since an argument can be prominent due to its inherent properties such as animacy, or because of its status in the discourse (de Swart 2007; van Bergen 2011). Although animacy is often thought of as an inherent property of arguments (cf. de Swart and de Helen 2007), this property can change as well, either by the discourse (e.g., in the context of a fairy-tale) or by linguistic means, such as case-marking or the use of a specific construction. Take for example Malayalam, which is considered a one-dimensional differential object marking language, in which all and only animate objects are assigned accusative case (Aissen 2003). Egger (2016) found that speakers of Malayalam in fact use accusative case-marking on inanimate objects when these are characters in a fairy-tale. We hypothesize that in the absence of such a fairy-tale context, the use of an accusative case-marker on an inanimate object could itself elicit such an interpretation in the sense that an inanimate entity is conceptualized as animate. A clear example of this is discussed by de Swart (2014). Dutch exhibits a pattern of differential object marking with a restricted class of contact verbs, such as hit, bite, and kick. These verbs take animate arguments as direct objects but inanimate ones as prepositional objects, due to an implication of sentience (hence, animacy) on behalf of the undergoer argument of these verbs. The pattern is illustrated in (1) and (2) (de Swart 2014: 445–446):
‘The dog bit the man.’
‘The dog bit the bread.’
De Swart (2014) argues that the use of the preposition with inanimates can be interpreted as a means to overtly signal a type mismatch, as inanimates are not sentient. The preposition signals a shift in the selectional restrictions of the verb, making it compatible with inanimate arguments. Given these two types of structures for the object of verbs of contact in Dutch, one can also use the prepositional structure for an animate argument or the nominal structure for an inanimate argument. The resulting pattern is not ungrammatical, but pragmatically odd. Hearers arrive at an interpretation in which the animate argument is interpreted as inanimate, and the other way around. Examples of this conceptual shift in animacy are given in (3) (de Swart 2014: 461) and (4) (Hoekstra 1992: 163; de Swart 2014: 458).
In (3) the prepositional object obtains an inanimate interpretation of the man, for example that the man refers to a statue or a dead body. Hoekstra (1992) notes that (4) receives a fairy-tale interpretation in which the inanimate cup receives an animate interpretation. Hence, due to the difference in linguistic marking, the hearer may feel sorry for the cup in (4) but not for the man in (3), while in the prototypical constructions, such as in (1) and (2), this would be the other way around. We thus observe a difference between overt shifts, in which the type incompatibility is overtly marked and the result is a change in the selectional restrictions of the predicate, and covert shifts, in which the argument is conceptually shifted without any overt signalling.
The goal of this paper is to arrive at a better understanding of these animacy shifts, and hence of the notion of animacy. We do this by exploring the similarity between the shifts presented above and the phenomenon of type shifting known from the semantic literature (see de Hoop 2012). To make this maximally explicit, we pursue the idea (following Dahl and Fraurud 1996; Dahl 2008; a.o.) that animacy represents an ontological category with human, animate and inanimate entities representing different types. In Section 2, we first discuss the more traditional (functionalistic) view of treating animacy effects in terms of a hierarchy. We will argue that such a hierarchy, although crucial for some languages, is not a necessary property of each language in which animacy plays a role. Instead, we can use a binary feature or type opposition. Section 3 then provides some linguistic evidence for establishing such types. Section 4 focuses on type mismatches as the ones illustrated above. Following earlier work (Aristar 1997; Nunberg 1979) we will show that these mismatches can be analysed in terms of a type shifting operation, which can be either overt or covert. In accordance with Malchukov’s (2008) claim that animacy status does not need to be marked in language, we will argue that an overt type shifter does not change the conceptual animacy of the noun phrase, whereas covert type shifting does. The proposed analysis relies on discrete semantic types and this seems at odds with the generally accepted conceptualization of animacy as a gradient category (e.g. Yamamoto 1999). In Section 5 we will argue that this incompatibility is only apparent. We maintain that even though conceptually animacy may be a gradient notion, linguistically it manifests itself in a discrete (binary) way due to the fact that the phenomena in which animacy is involved are binary (e.g. absence or presence of case). Thus, by focusing on overt semantic shifts as in (1) and (2), and covert shifts as in (3) and (4) above, we arrive at a more nuanced view on animacy and its effects on language.
2 Animacy as a hierarchy?
human > animate > inanimate
This hierarchy distinguishes three categories of animacy and prioritizes them such that human entities are considered the most prominent and inanimate ones least prominent with animates sitting in between. The animacy hierarchy in itself may reflect conceptual prominence, but in the linguistic literature it is often employed in relation to some grammatical phenomenon such as case marking. In such situations it is used in an implicational way: if a phenomenon applies to a category C on the hierarchy it will also apply to categories higher up in the hierarchy but not to the ones lower down in the hierarchy. Thus, we may find a language which restricts case marking to the human category leaving animates and inanimates unmarked. A language which implies a certain marking to both humans and inanimates to the exclusion of animates would go against the predictions of the hierarchy. In this sense, the animacy hierarchy works like a semantic map obeying the contiguity hypothesis (Corbett 2010).
The status of prominence hierarchies, such as the animacy hierarchy, is a fiercely debated issue in the linguistic literature and proponents and opponents seem to be split largely on the functional-formal divide in linguistic theory (see e.g. Aissen 2003; Newmeyer 2002; de Swart 2007; de Swart et al. 2008; Bornkessel-Schlesewsky et al. 2015; for discussion). Functionally oriented researchers generally rely on hierarchies to describe and explain (cross)linguistic patterns (e.g. Croft 2003; Corbett 2010). Linguists with a more formal inclination, by contrast, generally shun the use of hierarchies (see Carnie 2006; Newmeyer 2002; for discussion). We see merit in both positions.
On the one hand, evidence for the (language-specific) grammatical reality of the animacy hierarchy can be found in languages that make explicit reference to it. For instance, Mam-Maya, a language from Guatemala, requires the transitive subject to be higher in the animacy hierarchy than the object (Minkoff 2000). As shown in the b-examples of (6) and (7) the reverse situation is considered ungrammatical. In order to express such events a passive-like construction has to be used.
‘The woman sees the dog.’
‘The dog sees the woman.’
‘The dog killed the plant.’
‘The plant killed the dog.’
A similar situation is found in the Papuan language Fore (Scott 1978; see also below) and hierarchy effects on word order are documented in Navajo and Sesotho (cf. Siewierska 1988). These languages make explicit reference to an animacy hierarchy in the formulation of language-particular grammatical constraints.
At the same time, we find phenomena in which reference to a hierarchy is not necessary. This may, in fact, be the dominant pattern. Take, for instance, the phenomenon of differential object marking in Spanish. The object marker a in this language is obligatory with animate objects, but not with inanimates (in addition definiteness and/or specificity play a role as well), cf. the following examples (from García García 2007: 63).
‘I know this film.’
‘I know this actor.’
This case marking pattern can be described through reference to the opposition [±animate] (see e.g. López 2012) without reference to a hierarchy, although such a hierarchy may come into play through cross-linguistic comparison (e.g. Aissen 2003; Haspelmath 2015). Ortmann (1998) observes that when it comes to animacy effects, situations such as that in Spanish DOM are manifold. The hierarchy in (5) above divides the domain of entities into three different types, namely human, animate and inanimate. Nevertheless, it seems to be rather uncommon for languages to adopt such a three-way split, what we may call a “tripartite” system, differentiating a given phenomenon for human, animate and inanimate entities. Instead, we find that the “middle” category of animate is sometimes combined with human and sometimes with inanimate. The latter type of system, which we may call “homonist” following Ortmann (1998), can be best described in terms of a feature opposition [±human] (see Table 1). The first type of system, which Ortmann calls “vitalist”, can be described in terms of a feature opposition [±animate] (or, in fact, as the opposition [±inanimate]).
As stated, it is not necessary to stipulate the presence of an animacy hierarchy in the grammars of such languages. These languages can be adequately described using the feature oppositions [±human] or [±animate]. This does not mean that the animacy hierarchy may not be available to these languages. Evidence that such a hierarchy is available may come from the fact that we do not seem to find systems in which human and inanimate are grouped together to the exclusion of animate. As stated above, the absence of such a system can be explained when we assume that the possible feature (or type) space is constrained by the animacy hierarchy. When we interpret this hierarchy as a semantic map, the contiguity hypothesis indeed excludes the system (Corbett 2010). De Schepper and Zwarts (2009) show that the meanings on a semantic map can be decomposed into sets of primitive and binary features, since these meanings have properties that relate them to each other. The features can be used to define only certain sets of meanings. Two meanings are related to each other if they differ in just one feature. That is, human and animate entities share the feature [+animate] while they differ in the feature [±human], while animate and inanimate entities share the feature [-human] but differ in the feature [±animate]. Human and inanimate entities differ in both features and thus are not related to each other, which defines the structure of a semantic map in such a way that the two meanings lack contiguity.
Thus, we certainly do not reject the assumption that an animacy hierarchy plays a role in language, but we do think that many phenomena are more parsimoniously analysed in terms of binary types or features. This may even hold for phenomena that have been linked to an animacy hierarchy. For instance, since the work of Silverstein (1976) it has been argued that the marking of core arguments is affected by the position of a referent in the animacy hierarchy. Aristar (1996, 1997) argues that similar hierarchical patterns can be used to describe the distribution of non-core cases in many languages. Aristar studies the distribution of what he labels “non-grammatical cases”, such as locative and instrumental case, in comparison to the “grammatical” dative case. He reports that dative case is cross-linguistically oriented towards animacy whereas locative and instrumental cases are oriented towards inanimacy in ways similar to the hierarchical patterns observed by Silverstein (1976). Although Aristar shows how these case-marking patterns can be described in terms of hierarchies, he actually claims that this hierarchical patterning is an epiphenomenon of selectional restrictions on case-arguments. He refers to this preference of cases to combine with certain referents as typing, a notion that can be directly linked to the kind of type-theoretical system used in the formal semantic tradition (cf. De Hoop 2012). We take up this idea in this paper. In the next section, following work by Dahl (2008), we will show that animacy can indeed be established in terms of ontological types.
3 Animacy as ontological types
The two basic semantic types in semantic type theory are e for “entity” and t for “truth value”. Partee (1987) argues that noun phrases get interpretations in three types, the referential type e, the predicative type <e,t>, and the generalized quantifier type <<e,t>,t> . Type shifts from one type to the other can be either overt or covert. Example of overt type shifters are indefinite articles and copular verbs. An indefinite determiner such as a(n) in English can be used to shift the predicative type <e,t> of a bare noun to the referential type e or the existential quantifier type <<e,t>,t>, whereas the copular verb be takes a noun phrase of type e or <<e,t>,t> and turns it into a predicate of type <e,t>. Likewise, De Hoop and Kemperman (2015) argue that certain adjectives may function as overt type shifters that turn a proper noun of type e into a predicate of type <e,t> which can in turn combine with an indefinite article, as in a relieved Obama. Obviously, languages may differ in whether they have an overt type shifter or not. Chierchia (1998: 360) assumes that only if a language does not have a particular lexicalized type shifter, a covert type shift may occur, a principle he calls “Type Shifting as Last Resort”.
Focusing on the domain of entities, Partee (1987) also discusses several sort-shifting operations within the single type e, for instance converting count to mass reference and vice versa, and proposes to allow for events and plural individuals to denote in e. Additional structure has been proposed within the domain of entities. De Swart et al. (2007), for instance, introduce subtypes eC for capacities (e.g., teacher or president) and eK for kinds. These two subtypes are considered similar but sortally distinct. Similarly, we take animacy properties to lead to different ontological subtypes in the domain of entities, i.e., ehuman for human beings, eanimate for animate entities and einanimate for inanimate ones. We will see below that these types can be shifted both overtly and covertly.
Animacy is thus considered an ontological category, a fundamental dimension along which we categorize entities around us, in this case into categories of human beings, animate entities and inanimate entities. Dahl and Fraurud (1996) propose that these categories can be aligned with the notion of type as advanced in semantic type theory. Dahl (2008) provides several arguments to support the claim that these categories function as ontological types. We reiterate and exemplify two of them here.
First, membership of an ontological type determines what predicates can be applied to it, as predicates often apply to a certain type of argument. In other words, some predicates select one of the types as their argument. Indeed, different things can be said about animate and inanimate types. The verb scare, for instance, requires its object to be animate, whereas it puts no such restriction on its subject. In some cases we even find verbs that come in two flavours, one for animates and one for inanimates. A well-known example is the distinction between the two existential verbs in Japanese iru for animates (10) and aru for inanimates (11), which can be related to a more general cross-linguistic pattern of classificatory verbs based on animacy (Croft 1994).
‘There is a tiger.’
‘There is a fence.’
(from Nicol 1998: 118)
The linguistic relevance of such selectional restrictions has been debated. McCawley (1971), for instance, arguing against selectional restrictions as grammatical constraints, states that a person who utters My toothbrush is alive and is trying to kill me “should be referred to a psychiatric clinic, not to a remedial English course” (McCawley 1971: 219). It is indeed not clear that this sentence violates any selectional restrictions. An alternative analysis is that the toothbrush, although lexically inanimate and normally referring to an inanimate entity, is used here to refer to an animate entity. Thus, the sentence presents another example of the animacy shifts with which we started this paper. This kind of meaning transfer is very common particularly in those contexts, i.e., belief reports and cartoons, which McCawley considers prime examples of contexts in which selectional restrictions are violated. The fact that animacy shifts occur frequently can be considered evidence for the existence of selectional restrictions, which can only be met in these cases after a shift in animacy has taken place. In the Spanish examples (8) and (9) above, the verb know does not select for either an animate or inanimate object. As a consequence, a conceptual shift in the object’s animacy may be more difficult to obtain. Unlike in a sentence such as My toothbrush is trying to kill me, where selectional restrictions immediately trigger the animacy shift, the result in (8) and (9) is ungrammaticality when the case-marker is used with an inanimate object or when no case-marker is used with an animate object. In an appropriate context, however, the case marker could elicit an animacy shift and hence lift the ungrammaticality (see (20) below).
We take selectional restrictions to be linguistically relevant. In fact, we take them to be the driving force behind the shifts exemplified in the introduction. Additional evidence for the linguistic relevance of selection restrictions can be gleaned from the fact that language users rely on them to make inferences about categories (see Dik 1997: 91–97 for further arguments). For instance, Yoshida and Smith (2003) report that Japanese children make use of the linguistic cues in the verbs aru/iru in their conceptualization of objects that are perceptually ambiguous as to their animacy status.
Resistance of combination either within a single lexical item or by means of conjunction of items, is a second property of ontological types argued by Dahl (2008) also to apply to animacy. He notes for lexical items that “there are few if any nouns whose denotation includes both animates and inanimates” (Dahl 2008: 145). Some languages do not allow the conjunction of animate and inanimate nouns (see e.g. Bril 2011 for discussion of such facts in Austronesian languages). Given that conjunction applies to elements of like type, the exclusion of conjunction for animate and inanimate entities is expected under the view that animacy properties lead to different ontological subtypes in the domain of entities.
In addition to linguistic evidence, the fundamental status of animacy can also be observed from cognitive facts. There is some evidence from research on language impairment that animate and inanimate concepts may be stored in separate brain regions. Some patients have been reported to have lost the ability to name animate entities while maintaining that of inanimate ones, and vice versa (see Radanović et al. 2016 for references). Studies from concept development suggest that children already learn the distinction between the two categories at a very young age (Gelman and Opfer 2002). Also, language users are reported to be better at remembering animate entities than inanimate ones, although this ability may decrease with age (Bugaiska et al. 2016).
What is the best place to look for linguistic reflexes of ontological types? According to Dahl (2008), they “work behind the scene, channeling the ways we speak about entities in the world” (p. 146). Phenomena involving sortal partitions such as gender, possessives, pronouns, question words, and quantification are thus a good starting point. Indeed do we find effects of animacy in these places. The opposition between animate and inanimate gender discussed for Algonquian languages is probably one of the best-known illustrations of an animacy dichotomy influencing the grammar of a language (e.g., Mithun 2001). As for possessive marking, Malchukov (2008: 204) notes that Russian has a special possessive form -in that can only be used on proper nouns and kin terms (e.g., mam-in ‘mother’s’) but not on inanimates. Personal pronouns are used to refer to entities in the (non)-linguistic context and many languages make a distinction between animate and inanimate referents for third person (singular) pronouns. Several languages make a further subdivision within the animate category based on gender of the referent, as in English he vs. she, but animacy is the primary factor (Croft 1994; Ortmann 1998).
Question words seem prime candidates for linguistic elements reflecting ontological distinctions. Thus, in English we find separate question words for established ontological types such as location (where) and times (when) but also for animate (human) and inanimate entities in the opposition between who and what. Whaley (1997: 242) states that although languages differ in the number of question words they have, “most languages at least make a distinction between human and nonhuman question words (e.g. who or what)”.
A similar pattern can be observed for quantifiers. We find that languages use different quantifying words depending on ontological type. Looking at English again we find separate universal quantifiers for humans (everyone or everybody) and nonhumans (everything) in addition to words dedicated to times (everytime) and places (everywhere). The same holds for existential quantification and negative quantifiers. Some languages take the distinction even further and apply it also to other types of quantification. The Uto-Aztecan language Ute, for instance, has separate animate and inanimate forms for quantifiers expressing “many”, “much”, “few” and “other” in addition to “all” and “some” (Givón 2011: 53–54). The quantifier “other” seems to behave similarly in English, at least in plural, with the opposition between others and other things. Likewise, a normative rule of written Dutch requires substantivized quantifiers to be written differently when used for humans (andere-n ‘other-hum vs. andere ‘other’), although in practice not many language users adhere to this (van der Horst 2007).
4 Overt and covert type shifting
We have argued in the previous section for a type representation of animacy. In this section we discuss a few examples that illustrate the different ways in which languages may resolve type clashes. Predicates and functional markers (such as case) may impose typing restrictions on their arguments. When an argument is not of the type required by the predicate a type clash occurs. This may result in ungrammaticality but generally there are also ways to resolve the clash, notably by type shifting. This shifting may be overtly marked or not.
Aristar (1996, 1997) identifies three ways in which languages deal with a mismatch between the type restriction of a case-marker and the animacy feature of its complement in the locative case domain. The first is the exclusion strategy. For instance, as Aristar (1997: 346) reports, in Old Hittite dative case appeared only with animate nouns, while the locative case was reserved for inanimate nouns. In Even, animates take the comitative case-marker nún (e.g., gia-nún ‘with a friend), while inanimates take the form -lkan (e.g., turki-lkan ‘with a slade’) (Malchukov 2008: 204). Ditransitive constructions in the Algonquian language Fox (Anderson 1997) are restricted to animate indirect objects. Likewise, in Basque benefactive case occurs exclusively with animates and instrumental case with inanimates. A type mismatch in these languages results in ungrammaticality.
A second strategy is reinterpretation, dubbed meaning extension by Aristar. In this case, the composition of a mismatch can proceed through the reinterpretation of the semantic role to be associated with the noun phrase. A clear case is when an instrumental marker is combined with an animate argument resulting in a comitative interpretation (an animate instrument). Note that in these cases it is not the ontological type of the noun phrase itself that shifts (unlike in cases of conceptual shifts as advanced in semantic literature, cf. Pustejovksy 1991; Nunberg 1995; discussed below), but rather the interpretation of a linguistic marker, i.e. its associated semantic role.
This second strategy is sometimes accompanied by a third one, in which a type mismatch is resolved by the use of some additional morphology (‘bridges’ in Aristar’s terminology). Some languages have a special morphological marker which has to be added to the argument noun before the appropriate case ending can be attached. This is illustrated in (12) and (13) for the ablative case in the Australian language Yidiny (Aristar 1997: 317):
‘from the hand’
‘because of the woman’
The inanimate noun mandi “hand” in (12) takes the ablative case ending directly. The animate noun buña: “woman” in (13), by contrast, must first be marked with a special suffix before the ablative ending can be added. Moreover, an extension to a causal meaning can be observed.
The morphological bridges identified by Aristar (1997) can be interpreted as overt markers of a type shifting procedure. In Yidiny, the ablative marker requires an inanimate argument, i.e., an argument of type einanimate. However, woman is of type eanimate and can as such not be the argument of the ablative marker. A type shift takes place by means of the bridge morpheme. As a result, composition can proceed. However, there is no conceptual shift in the animacy status of the argument. That is to say, the woman is not interpreted as conceptually inanimate when it is combined with the type shifter -ni. Rather, it is the case-marker that gets a shifted (extended) interpretation. Therefore, we propose that the morphological type shifter does not shift the animacy of the noun, but rather it shifts the type of the case-marker. The bridging morpheme -ni extends the function of the ablative case-marker from an inanimate source meaning to an animate cause meaning.
This kind of overt type shifters can also be found in core case marking. A case in point is the case marking on transitive subjects in the Papuan language Fore (Scott 1978; Donohue 1999; Malchukov 2008). Use of the subject case-marker -ma in Fore is conditioned by a number of factors. For human and animate arguments, it is dependent on both the relative position of arguments in the animacy hierarchy and the order of arguments. The suffix -ma is used to flag a lower-ranked argument as the subject or to flag an object-initial order in case of arguments of equal animacy. When the A-argument is inanimate, case marking becomes obligatory, independently of the animacy features of the object. According to Scott (1978) the reason for this obligatoriness is that the use of subject case marking turns inanimate nouns into potential agents, the only elements allowed as transitive subjects. Although potential agent is a somewhat vague notion, evidence for Scott’s claim can be found in the form of the case-marker used. Human nouns occur with the marker -ma, but in the case of animate and inanimate nouns we find –wama. This marker seems to be composed of the case-marker -ma and the noun wa ‘man’.
‘The stick injures me.’
The case suffix -ma in Fore thus seems to select for human nouns. We could say it is categorized as selecting an argument of type ehuman. When confronted with an argument lacking this value this would normally result in type incompatibility. This mismatch in types is solved through the insertion of wa “man” to saturate the humanness restriction, and allowing the composition to proceed. The compound case-marker wama can thus be assigned to a transitive subject of type einanimate. Note that the conceptual animacy of the stick in (14) does not shift (it is not conceptualized as being human). Rather, the inanimate noun phrases are made compatible with the case-marker for A-arguments through the bridge morpheme wa that carries the feature [+human].
Another example of such an overt type shifting (bridging) operation is the insertion of a preposition in Dutch with inanimate objects of verbs of contact, as discussed in the introduction (cf. de Swart 2014). Recall (2) above, repeated as (15) below, for convenience.
‘The dog bit the bread.’
In this case, a preposition is inserted to overcome the mismatch between the selectional restrictions of the verb, requiring an animate (sentient) object, and the inanimate noun phrase the bread. Again, the overt type shifter, the preposition, does not change the conceptual animacy of the argument, but it changes the verb. Whereas transitive verbs such as bijten “bite”, schoppen “kick”, and slaan “hit” are of type <eanimate, <eanimate, t>>, hence require animate objects, their complex counterparts bijten in “bite in”, schoppen tegen “kick against”, and slaan op “hit at” are of type <einanimate, <eanimate, t>>, and thus require an inanimate object. Hence, the locative preposition that combines with the verb, changes the selectional restrictions of the verb. Additional evidence that the preposition functions to make the verb and object compatible comes from the fact that the preposition may change depending on properties of the object (e.g. bijten op “bite on” when the surface of the object is less penetrable).
The overt type shift that comes with the insertion of a preposition is radically different from the covert type shift in which the conceptual animacy status of an argument is changed, often via the selectional restrictions of the verb. The examples (3) and (4) above, repeated below as (16) and (17), illustrate our point.
As pointed out above, the verb slaat ‘hits’ in Dutch requires an animate object of type eanimate, whereas the complex verb beet in “bit in” requires an inanimate one, type einanimate. Clearly, de man “the man” in (16) would normally denote in type eanimate, yet its occurrence as the object of beet in “bit in” induces a shift in animacy. Hence, de man “the man” is conceptualized as being inanimate, for example a corpse of type einanimate. Similarly, het kopje “the cup” would normally denote in type einanimate, but its occurrence as the direct object of slaat “hits” causes a shift in animacy. As a consequence, het kopje “the cup” is conceptualized as animate, for example in the context of a fairy-tale.
Covert animacy shifts are expected to occur under the proposed approach to animacy, and in fact can be taken as further evidence for it. According to Dahl (2008) these types of shifts can be easily made because the types are so different there is little risk of confusion. Such covert animacy shifts, driven by selectional restrictions of the verb, or by the context in general, have been discussed in the literature. Well-known is the ham sandwich example by Nunberg (1979; 1995). Uttered in a restaurant context, the subject of the sentence “the ham sandwich is sitting at table 7” should be interpreted as the person who ordered the ham sandwich. This kind of meaning transfer (or sense shift) is based on a contextually salient relationship between the animate and inanimate entity and are arguably less “severe” than the cartoon-like shifts discussed above. Living toothbrushes, talking trains and dancing peanuts all involve the crossing of a sortal boundary. Notwithstanding some initial surprise, language users seem relatively at ease with such interpretations. Nieuwland and van Berkum (2006) investigated the effect of cartoonlike contexts on on-line interpretation. They embedded inanimate entities in a context in which they acquired animate characteristics, e.g. a peanut which was dancing, singing, and smiling. They tested whether hearers showed a processing difference when these inanimate words occurred with a predicate showing a “lexical fit”, e.g. salted, in comparison to a predicate showing a “contextual fit”, e.g. in love. The results show an eventual increase in brain activity for predicates with a lexical fit in comparison to those with a contextual fit. This indicates that the lexical animacy of a noun phrase can be overruled by context and that the conceptual animacy of a noun phrase is used by hearers to determine whether the selectional restrictions of a verb are violated.
Studies like that of Nieuwland and van Berkum (2006) (see also Schumacher 2011 for a similar study with “ham sandwiches”) attest the flexibility of language users in coming up with a shifted animacy interpretation in conflicting contexts.
5 Gradience and animacy
In the previous sections we have argued for a view on animacy in terms of discrete types. This approach contrasts with an approach based on an animacy hierarchy. We claim that for the phenomena discussed in this paper (and probably many more) such a discrete approach is preferred. However, as stated above, there are clear cases in which the influence of a hierarchy is evident.
A view of animacy as discrete types seems directly at odds with the general perception of animacy in the linguistic literature as a gradient notion. Typically, animacy is represented as a category that is ordered along a continuum (or hierarchy) (cf. Bossong 1984; Aissen 2003). Animacy is thus not a direct reflex of the biological distinction between living and non-living entities, but is also affected by other dimensions. According to Yamamoto (1999) animacy values assigned by language users are closely related to the empathy they feel for entities. She proposes a radial category prototype model with humans as the central category. Other types of entities can be positioned closer or further away from the central prototype depending on their similarity to it. Thus, anthropomorphized entities will be closer and abstract entities further to the outside (see Dahl 2008, for a related proposal). Also, inanimate characters or narrators in a story can be perceived as more or less human, dependent on various physical and psychological properties (cf. Bernaerts et al. 2014; Looser and Wheatly 2010) as well as linguistic clues (cf. Vogels et al. 2013; Trompenaars et al. 2016).
How should we resolve this opposition between gradience and discreteness? We maintain that we shouldn’t, as both views can hold at the same time. There is indeed ample evidence that our conceptual organization of animacy is gradient (as any other type of conceptualization), see Radanović et al. (2016) for recent psycholinguistic evidence. However, this conceptual gradedness is (in many cases) grammatically irrelevant. As Ortmann (1998) pointed out, most languages have a binary system of grammatical animacy. The reason for this is to be found in the fact that animacy is involved in linguistic phenomena that are binary themselves. Animacy affects, for instance, whether or not a speaker should use case marking, agreement marking or plural marking, and whether they use the third person singular pronoun he or it. In other words, the discrete view is enforced by the language itself. A speaker then has to make a mapping from the graded conceptual representation to a discrete linguistic manifestation (see Lestrade 2012 for an analogue mapping in the domain of spatial language). This mapping is a source for variation within and between languages, as conceptualization will be affected by speaker beliefs (e.g. some speakers may see cats as more animate than dogs, cf. Yamamoto 1999), cultural beliefs (e.g. in some linguistic communities higher animals will be treated grammatically as human, cf. Heath 1980) or plain context (as in the cartoon contexts discussed above).
Evidence for the dissociation between conceptual and grammatical animacy can be found in the psycholinguistic study by Radanović and Milin (2011) on animacy effects in Serbian case marking. Case marking on Serbian masculine nouns is differentiated based on animacy. For animate nouns, syncretism is observed between genitive and accusative case, but for inanimate nouns syncretism occurs between nominative and accusative case. We thus observe a binary grammatical split (animate vs. inanimate). Given that animate concepts are considered conceptually more accessible than inanimate ones (e.g. Bock & Warren 1985) one might expect the grammatical opposition to be mirrored in reaction times, if there is indeed a one to one mapping between grammatical and conceptual animacy. Radanović and Milin (2011), however, found no effect of this binary grammatical animacy distinction on reaction times in a lexical decision task and conclude that this distinction is hence not “cognitively relevant” (Radanović and Milin 2011: 352). This suggests that the distinction between conceptual and grammatical animacy is indeed real, as at least on one interpretation of the findings it is not surprising to find no effect when the underlying conceptual representation is not discrete but gradient (see Radanović et al. 2016 for a more recent discussion on the psycholinguistic implications of animacy categorizations).
Another piece of evidence comes from languages with grammaticalized animacy genders. Prime examples are the Algonquian languages which have a so-called animate and inanimate gender. All nouns in these languages are classified as animate or inanimate. For conceptually animate nouns classification is clear, but for inanimate nouns gender is not fully predictable, even though clear tendencies exist (Mithun 2001: 93). The fact that animacy categorization leaks is expected under our view. A parallel can be drawn to the notions of feminine and masculine gender. These genders are named after their prototypical animate members, which bear the same gender (sex) in the real world. At the same time, most members of these categories, the inanimates, do not have a corresponding extra-linguistic sex. The same thing may be happening in animacy classification. The categories are named after the most prominent members, but at the same time members that do not fit this conceptual profile are included based on other semantic or formal criteria, e.g. grammatical behaviour similar to that of the prototypical members (see e.g. Welch 2016). As a result, categories bear a specific name but are a mixed bag. Labelling these categories as “animate” or “inanimate” obscures the fact that they cannot be directly mapped onto a conceptual classification. Indeed, it may be more productive to use nonspecific category names (e.g. category numbers as in the Bantu tradition) without any association to the related conceptual dimension.
We thus adopt a two-tiered view on animacy in which we distinguish conceptual animacy from grammatical animacy (summarized in Figure 1). The two notions are clearly related, but not identical. Let us illustrate this (again) with a final example.
We have discussed the phenomenon of differential object marking in Spanish. As shown in example (8), repeated below as (18), the use of the object marker a is determined by animacy, grammatical animacy that is. Following López (2012) i.e., we can model this in terms of the binary opposition [±animate]. What entities end up in the grammatical [+animate] class will be (partially) determined by conceptual animacy (the down arrow in Figure 1).
‘I know this film.’
‘I know this actor.’
We may also find the object marker with inanimate objects, which is associated with a personified reading, as in example (20) from Wiltschko and Ritter (2015: 904).
‘S/he called out to Death.’
Here, “death” is represented as an animate being. We can analyse this in two ways. From the hearer’s perspective, it represents another instance of the covert shift that we have seen in the Dutch examples (16) and (17) above. The conceptual shift here is induced by the linguistic context. The object marker comes with a selectional restriction for animate arguments. The complement in its basic sense does not meet this requirement and the hearer has to make a conceptual shift. This analysis represents the up arrow in Figure 1. Alternatively, from the speaker’s perspective, the referent is conceptualized as animate from the start and as such the object ends up in the grammatical [+animate] category, as represented by the down arrow in Figure 1. Being grammatically animate it requires the object marker. Either way, the example testifies to the flexibility of animacy in language.
|En esta receta,||la leche||puede sustituir||a-l||huevo.|
|in this recipe||the milk||can.3.SG replace||a-the||egg|
‘In this recipe, egg can be replaced by milk.’ (García García 2007: 64)
The absence of an animacy shift in (21) can be explained by a difference in case marking strategy (cf. de Hoop and Malchukov 2008). The examples in (18) and (19) demonstrate a prominence marking strategy, which is directly related to the animacy feature of the direct object. The use of the case-marker in (21) is probably better seen as an instance of ambiguity avoidance, in which the animacy of the object is only involved through reference to semantic roles (cf. Primus 2012; Verhoeven 2014).
Animacy properties reflect ontological subtypes in the domain of entities, in particular ehuman for human beings, eanimate for animate entities and einanimate for inanimate ones. We have argued that whereas conceptually animacy may be a gradient notion, in that for example a dancing peanut may still be perceived as less animate than a dancing man, linguistically animacy is a discrete phenomenon that can be modelled in terms of binary features [±human] and [±animate]. Evidence for the existence of animacy shifts can be gleaned from cases in which a language employs morphology to overtly signal a type shift (cf. Aristar 1997). However, we have argued that overt type shifts are never used to mark a conceptual shift in animacy. Indeed, such marking would be superfluous in language, as pointed out by Malchukov (2008). Grammatical restrictions can lead to overt type shifting operations, but these pertain to semantic roles or grammatical functions. They sometimes involve a meaning extension of the predicate, but crucially, they do not show a concomitant shift in conceptual animacy. By contrast, covert animacy shifts do conceptually shift an animate referent into an inanimate one, or the other way around. Covert animacy shifts have been shown to be psycholinguistically real, and they can be quite easily elicited by selectional restrictions or other properties of the linguistic or extra-linguistic context.
Aristar, Anthony R. 1996. The relationship between dative and locative: Kuryìowicz’s argument from a typological perspective. Diachronica 13. 207–224. Google Scholar
van Bergen, Geertje. 2011. Who’s first and what’s next? Animacy and word order variation in Dutch language production. Radboud University Nijmegen PhD dissertation.Google Scholar
Bornkessel-Schlesewsky, Ina, Andrej Malchukov & Marc Richards. 2015. Scales and hierarchies: A cross-disciplinary perspective. Berlin: de Gruyter. Google Scholar
Bossong, Georg. 1984. Animacy and markedness in Universal Grammar. Glossologia 2/3. 7–20. Google Scholar
Bril, Isabelle. 2011. ‘And’ and ‘with’ conjunctive strategies in some Austronesian languages: Syntax, semantics, pragmatics. Language and Linguistics 12. 239–272. Google Scholar
Bugaiska, Aurélia, Alain Méot & Patrick Bonin. 2016. Do healthy elders, like young adults, remember animates better than inanimates? An adaptive view. Experimental Aging Research 42. 447–459. CrossrefGoogle Scholar
Carnie, Andrew. 2006. Some remarks on markedness hierarchies: A reply to Aissen 1999 and 2003. In D. Siddiqi & B. V. Tucker (eds.), Coyote Papers: Working papers in linguistics, 37–50. Arizona: University of Arizona. Google Scholar
Comrie, Bernard. 1989. Language universals and linguistic theory. Chicago: University of Chicago Press. Google Scholar
Corbett, Greville G. 2010. Implicational hierarchies. In J.J. Song (ed.), The Oxford Handbook of Linguistic Typology, 190–205. Oxford: Oxford 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 T. Fretheim & J.K. Gundel (eds.), Reference and referent accessibility, 65–88. Amsterdam: John Benjamins. Google Scholar
Dik, Simon C. 1997. The Theory of Functional Grammar. Part 1: The structure of the clause (2nd revised edition, edited by K. Hengeveld). Berlin: Mouton de Gruyter. Google Scholar
Donohue, Cathryn. 1999. Optimizing Fore case and word order. Unpublished manuscript, Stanford University. Google Scholar
Egger, Julia. 2016. Asking the magic mirror: Fairytales and animacy in Malayalam. Nijmegen: Radboud University term paper. Google Scholar
García García, Marco. 2014. Differentielle Objektmarkierung bei unbelebten Objekten im Spanischen. Berlin: de Gruyter. Google Scholar
García García, Marco, García. 2007. Differential object marking with inanimate objects. In G.A. K. & M. Leonetti (eds.), Proceedings of the Workshop “Definiteness, Specificity and Animacy in Ibero-Romance Languages” (=Arbeitspapier des Fachbereichs Sprachwissenschaft 122), 63–84. Konstanz: Universität Konstanz. Google Scholar
Gelman, Susan A. & John E. Opfer. 2002. Development of the animate–Inanimate distinction. In U. Goswami (ed.), Blackwell handbook of childhood cognitive development, 151–166. Malden, MA: Blackwell Publishers. Google Scholar
Givón, Talmy. 2011. Ute Reference Grammar. Amsterdam/Philadelphia: John Benjamins. Google Scholar
Haspelmath, Martin. 2015. Descriptive scales versus comparative scales. In I. Bornkessel-Schlesewsky et al. (eds.), Scales and hierarchies: A cross-disciplinary perspective, 45–58. Berlin: de Gruyter. Google Scholar
Heath, Jeffrey. 1980. Basic materials in Ritharngu: Grammar. texts, and dictionary. Canberra: Pacific Linguistics. Google Scholar
Hoekstra, Teun. 1992. Aspect and theta theory. In I. M. Roca (ed.), Thematic structure: Its role in grammar, 145–174. Berlin/New York: Mouton de Gruyter. Google Scholar
de Hoop, Helen. 2009. On (in)animate noun phrases. In A. Giannakidou & M. Rathert (eds.), Quantification, Definiteness & Nominalization, 234–250 Oxford: Oxford University Press. Google Scholar
de Hoop, Helen. 2012. Type shifting. In C. Maienborn et al. (eds.), Semantics. An International Handbook of Natural Language Meaning. 3, 2259–2271. Berlin: de Gruyter. Google Scholar
de Hoop, Helen & Erica Kemperman. 2015. ‘A relieved Obama’ won’t last long. In B. Köhnlein & J. Audring (eds.), Linguistics in the Netherlands 32, 75–87. Amsterdam: John Benjamins. Google Scholar
van der Horst, Luuc. 2007. De markering van kwantoren in het Nederlands: Taalnorm vs. taalgebruik. Nijmegen: Radboud University BA thesis.Google Scholar
Lestrade, Sander. 2012. A linguistic ontology of mode: The use of locations in spatial language. In C. Stachniss et al. (eds.), Spatial Cognition VIII, 319–337. Heidelberg: Springer. Google Scholar
López, Luis. 2012. Indefinite objects. Scrambling, choice functions, and differential marking. Cambridge: MIT Press. Google Scholar
McCawley, James D. 1971. Where do noun phrases come from? In D.D. Steinberg & L.A. Jakobovits (eds.), Semantics. An interdisciplinary reader in philosophy, linguistics and psychology, 217–231. Oxford: Oxford University Press. Google Scholar
Minkoff, Seth. 2000. Animacy hierarchies and sentence processing. In A. Carnie & E. Guilfoyle (eds.), The syntax of verb initial languages, 201–209. Oxford: Oxford University Press. Google Scholar
Mithun, Marianne. 2001. The languages of native North America. Cambridge: Cambridge University Press. Google Scholar
Nicol, Janet L. 1998. The production of agreement in English and Japanese: Animacy effects (or lack thereof). In D. Hillert (ed.), Sentence Processing: A Crosslinguistic Perspective (Syntax and Semantics 31), 113–130. Bingley: Emerald Group Publishing Limited. Google Scholar
Nieuwland, Mante S., Andrea E. Martin & Manuel Carreiras. 2013. Event-related brain potential evidence for animacy processing asymmetries during sentence comprehension. Brain and Language 126. 151–158.CrossrefGoogle Scholar
Ortmann, Albert. 1998. The role of [+/-animate] in inflection. In R. Fabri et al. (eds.), Models of inflection, 60–84. Tübingen: Niemeyer. Google Scholar
Partee, Barbara H. 1987. Noun phrase interpretation and type shifting principles. In J. Groenendijk et al. (eds.), Studies in Discourse Representation Theory and the theory of generalized quantifiers, 115–143. Dordrecht: Foris. Google Scholar
Primus, Beatrice. 2012. Animacy, generalized semantic roles, and differential object marking. In M. Lamers & P. de Swart (eds.), Case, word order and prominence: Interacting cues in language production and comprehension, 65–90. Dordrecht: Springer. Google Scholar
Pustejovsky, James. 1991. The generative lexicon. Computational Linguistics 17. 409–441. Google Scholar
Radanović, Jelena & Petar Milin. 2011. Morpho-semantic properties of Serbian nouns: Animacy and gender pairs. Psihologija 4. 343–366. Google Scholar
de Schepper, Kees & Joost Zwarts. 2009. Modal geometry. Remarks on the structure of a modal map. In L. Hogeweg et al. (eds.), Cross-linguistic semantics of tense, aspect and modality, 245–269. Amsterdam/Philadelphia: John Benjamins. Google Scholar
Schumacher, Petra B. 2011. The hepatitis called … : Electrophysiological evidence for enriched composition. In J. Meibauer & M. Steinbach (eds.), Experimental Pragmatics/Semantics, 199–219. Amsterdam/Philadelphia: John Benjamins. Google Scholar
Scott, Graham K. 1978. The Fore Language of Papua New Guinea. Canberra: Research School of Pacific Studies, The Australian National University. Google Scholar
Siewierska, Anna. 1988. Word Order Rules. London: Croom Helm. Google Scholar
Silverstein, Michael. 1976. Hierarchy of features and ergativity. In M. Robert & W. Dixon (ed.), Grammatical Categories in Australian Languages, 112–171. Canberra: Australian Institute of Aboriginal Studies. Google Scholar
de Swart, Peter. 2007. Cross-linguistic variation in object marking. Radboud University Nijmegen PhD dissertation. Utrecht: LOT Publications. Google Scholar
de Swart, Peter. 2014. Prepositional inanimates in Dutch: A paradigmatic case of differential object marking. Linguistics 52. 445–468. Google Scholar
de Swart, Peter & de Hoop. Helen 2007. Semantic aspects of differential object marking. In E. Puig Waldmüller (ed.), Proceedings of Sinn und Bedeutung 11, 568–581. Barcelona: Universitat Pompeu Fabra. Google Scholar
Trompenaars, Thijs, Lotte Hogeweg, Wessel Stoop & de Hoop. Helen 2016. The language of an inanimate narrator. Unpublished manuscript (submitted). Radboud University Nijmegen. Google Scholar
Vogels, Jorrig, Emiel Krahmer & Alfons Maes. 2013. When a stone tries to climb up a slope: The interplay between lexical and perceptual animacy in referential choices. Frontiers in Psychology 4. doi: CrossrefGoogle Scholar
Whaley, Lindsay J. 1997. Introduction to Typology: The unity and diversity of language. London: Sage Publications. Google Scholar
Wiltschko, Martina & Elizabeth Ritter. 2015. Animating the narrow syntax. The Linguistic Review 32. 869–908. Google Scholar
Yamamoto, Mutsumi. 1999. Animacy and reference: A cognitive approach to corpus linguistics. Amsterdam/Philadelphia: John Benjamins. Google Scholar
Yoshida, Hanako & Linda B. Smith. 2003. Shifting ontological boundaries: How Japanese- and English-speaking children generalize names for animals and artifacts. Developmental Science 6. 1–34. CrossrefGoogle Scholar