From ‘clubs’ to ‘clocks’: lexical semantic extensions in Dene languages

1 Introduction

The Dene languages form a large and geographically extended family indigenous to the North American west. Numerous and diverse in the subarctic north, but more populous in the southwest, Dene languages have drawn the attention of linguists since before the time of Boas and Sapir. However, the majority of the scholarship on these languages has focused on grammatical and lexicographic description, predominantly from structuralist viewpoints (Rice 2017). The diachronic study of Dene languages has focused on sound change and phonological reconstruction. This study builds on this research, but focuses on lexical semantics drawing on Cognitive Linguistic insights into meaning change, especially the theory of metonymy. The Cognitive Linguistic theory of metonymy provides a powerful tool for the description of semantic change (Blank 1999; Koch 2011; Paradis 2011; Radden and Kövecses 1999). Metonymy has been found to be active across many domains of language and to take many different forms (Panther and Thornburg 2007). The following discussion, however, will illustrate the particular usefulness of Ruiz Mendoza Ibáñez’s reformulation of metonymy for the description of semantic change in Dene verbs (Ruiz de Mendoza Ibáñez 2003).

At the center of the analysis is the semantic evolution of the morphological root *x̣ʷɑˀɬ meaning ‘club’. Originating in Proto-Na-Dene, the root has evolved to form part of a network of nominal and verbal constructions centered on schematic imagery derived from ‘clubs’ as elongated, wooden objects used for striking. The diachronic trajectory of this Dene root can be described as repeated extension through source in target and target in source metonymies. The physical and functional features of clubs have been exploited semantically by Dene languages to form lexical expressions ranging from striking and drumming to rolling motions associated with cylindrical or round objects. Along the way, *x̣ʷɑˀɬ has schematized, ultimately only expressing particular aspects of shape and motion. Developing into an abstract root conflating semantic properties of form and motion, some of the reflexes of this root form patterns resembling classificatory verbs. As such, the semantic extensions of the root described here can be seen as strengthening the hypothesis that classificatory verbs emerged from other verb forms (Fortescue 2006).

The paper aims to draw attention to the importance of bringing Cognitive Linguistic analyses to semantic and diachronic problems in Indigenous languages of the Americas – and, in turn, contribute to strengthening the role of Indigenous languages for theoretical discussion in the field. The analysis of the polysemic extensions of *x̣ʷɑˀɬ reflexes follows a discussion of the relevant descriptive and theoretical background.

2 The Dene languages

The Dene languages form part of the Na-Dene (also called Athapaskan-Eyak-Tlingit) family, whose members are spread over a large, discontinuous territory in western North America. The largest continuous area where these languages are spoken stretches across interior Alaska, the Yukon, and across much of the Canadian subarctic boreal forestland west of the Hudson Bay. South of the subarctic, Dene languages are spoken in interior British Columbia and central Alberta. Intriguingly, languages belonging to this family are also found in two other geographically distant and separate regions: along the Pacific Coast in Oregon and Northern California, and in the southwestern United States. Historical relationships among the Dene languages are complex and scholars tend to make reference to broad geographic regions instead (Mithun 1999; Snoek et al. to appear). The geographic regions that Dene languages are found in are also associated with culture areas. In Alaska and the Canadian Interior, for instance, speakers of Dene languages traditionally lived the nomadic lifestyles of subarctic hunter-gatherers forming family-sized bands with little or no hierarchical structures beyond the family unit. In this paper, the subarctic Dene are represented by Koyukon, which was once spoken in the Alaskan interior, but is now on the verge of extinction, and South Slavey, which still has many speakers in the Deh Cho Region of the Northwest Territories in Canada. The Pacific Coast Dene languages are represented by Hupa, which is spoken by riverine peoples of northern California who lived sedentary lives and congregated in long-houses typical of the region. The Navajo, far to the south, represent the culture area today known as the American Southwest. Relative newcomers to the region, the Navajo adopted pastoralist economic modes as well as growing maize. Finally, the Tsuut’ina represent Dene groups that came to live on northern reaches of the Great Plains in Alberta, Canada. Today, all the Dene languages are severely endangered and, with the exception of Navajo, have on average less than a few hundred fluent speakers remaining. While the Navajo are one of the most numerous groups of Indigenous America, few children are learning the language as a first language today.

3 Dene verb constructions

Typologically, Dene languages can be described as fusional and polysynthetic with complex derivational and inflectional morphology. Across the family, the grammatical description of Dene languages recognizes three major part of speech categories: nouns, postpositions, and verbs. Nouns and postpositions feature limited inflectional affixation marking person and number of possessors or postpositional objects. The most complex part of speech in Dene languages, however, are the verbs which are composed of numerous prefixes and a handful of suffixes attaching to a root form. The prefixes encode grammatical mood and aspect distinctions, a wide range of adverbial meanings, the noun class of clausal participants, as well as number and person of the clausal participants themselves. These prefixes occur in two different groups: those closer to the stem mark inflectional categories and are referred to as the conjunct; prefixes to the left of the conjunct, the disjunct, form a more heterogeneous group encoding adverbial meanings, the objects of postpositions, incorporated noun stems, and thematic or derivational affixes (Kari 1989; McDonough 2000; Young and Morgan 1987). While there is some controversy about the morphological categories expressed by these two groups of prefixes across the languages, there are good phonological and phonetic grounds for upholding the distinction between conjunct and disjunct prefix groups (McDonough 2000).

For the sake of clarity, and since the focus here is on the semantics, some of the morpho-phonological details of the conjunct prefixes are omitted (for a fuller treatment of Dene verbal morphology see Rice 2000). Example (1) shows the typical structure of a Dene verb. Since the data presented in this paper come from different Dene languages, the origin of each example is indicated by the language name to the right of the example number.^[1]

The voice/valency marker ł- (glossed as vv) and the noun classifier dí- (glossed as nc) stay constant in all inflected forms of this verb. Together with the root these prefixes form the verb theme (Kari 1979; Young and Morgan 1987: 140), which can be understood as one of the basic units of the Dene lexicon (Rice 1989). The verb themes form word families of often closely related and polysemic senses based around meanings encoded in the roots. Incorporated objects and postpositional phrases can be added to these themes in creative ways to meet specific communicative demands. In (2), for example, the verb root expresses the movement of an object specified in terms of its form, i.e. a small round object. This object – the speaker’s head, is explicitly represented as an incorporated noun. At the far left end of the verb complex, the verb features an incorporated postposition (against) with a prefix indicating a third-person object.

The verb root -ʔón occurs in combination with many different prefixes. Strings of prefixes form stable patterns with the root to derive lexical meanings. In example (2), the root is coupled with the voice/valency marker d- as well as the incorporated noun ts’í- ‘head’. Even though these elements are discontinuous, they form a lexical unit known as a verb theme in the literature on Dene languages (Kari 1979; Rice 2000). As stable form-meaning pairings, these can be considered constructions in the sense of Langacker (2008: 161), and this terminology is used in the discussion of the lexical semantics of club-verbs in the following sections. Viewing Dene verb themes as lexical constructions also provides a theoretical basis on which to apply a Cognitive Linguistic approach to semantic change to these data. The relevant aspects of the theory are presented in the next section.

4 Metonymy and lexical semantic change

Describing the diachronic trajectory of the root form with the meaning ‘club’ in Dene languages requires taking into account the rich conceptual structure associated with this item of material culture. The relevant conceptual structures for the meaning of the root *xʷɑˀɬ are those of a thing in the sense of Langacker (2008: 105), as the oldest identified form of this root is a noun form (see discussion in Section 5). The conceptual structure associated with nominals can be described in terms of qualia roles (Paradis 2005: 553), especially as developed for lexical semantic analysis by Pustejovsky (1998). The qualia structure is a useful descriptive tool because it allows an encyclopedic approach to meaning to be operationalized through an emic grid of categories that support more detailed analysis. The categories of qualia serve to describe aspects of nominals such as shape and form of a denoted object (formal role), the way it came into being (agentive role), its function (telic role), and its meronymic structure (constitutive role) (Pustejovsky 1998: 85–86). For the object club, the formal role can be described as: roughly cylindrical in shape, and thicker at one end; the agentive role as that of a man-made object; the telic role as that of an implement made for striking – in the Dene case especially as used for killing certain types of land animals and fish; the constitutive role as that of a simple object typically made of wood with no other sub-parts (Figure 1).

Figure 1:

Fish club used among the Hupa (Goddard 1903–1904: 24).

From a Cognitive Linguistic point of view, qualia structure can be seen as capturing lexically relevant elements of conceptual structures such as frames, Idealized Cognitive Models, or domains. These conceptual structures are mental models that emerge as an organization of perceptual, sensorimotor, experiential, and learned cultural knowledge and ‘serve as the background for interpreting the meaning of linguistic forms’ (Cienki 2007: 183). The term idealized is drawn from the Cognitive Linguistic literature and is intended to draw attention to the schematic and generalized nature of conceptual structures relevant to metonymy. However, the term Cognitive Cultural Model or CCM (Blount 2014) may be more appropriate, since it explicitly draws attention to the shared conceptual content metonymies need to presuppose in order to signify successfully. The cultural nature of mental models is especially apparent when considering instances of mental models more closely tied to specific cultural contexts as when considering the conceptual structure underlying lexical expressions denoting lightning discussed in Section 5.5. The term CCM will generally be used here.

All these elements form contiguous parts of a conceptual model of the item and associated activities or events that is accessible to speakers. The conceptual contact or contiguity of the qualia elements as part of a CCM makes them possible sources and endpoints of processes of semantic change (Blank 1997, 2002; Koch 2011). In particular, the term denoting the whole CCM can be used for some of its parts or vice versa through metonymy. Metonymy involves a conceptual leap from the semantic information that is explicitly and conventionally encoded by the morphological vehicles (roots, affixes, and their combinations) to the intended meaning. The conceptual leap afforded by metonymic inference is guided by two factors: the metonymic inference schema or cognitive operation and the conceptual domain which provides the elements for the inference. In Radden and Kövecses’s definition:

Metonymy is a cognitive process in which one conceptual entity, the vehicle, provides mental access to another conceptual entity, the target, within the same idealized cognitive model. (Radden and Kövecses 1999: 21)

The term vehicle (also known as source) is here intended to designate the overtly encoded meaning of the given form while the target denotes the meaning actually realized through the expression. Metonymy thus involves an associative inference from a vehicle to a target, as in the whole for part metonymy Germany won the world cup, where Germany is a vehicle denoting the national football team. In contrast to other forms of linguistic association, the vehicle and target must be conceptually related, forming part of a unified mental structure – the ICM or CCM. In the case of the football team, the vehicle can stand for the target only on the basis of a presupposed CCM of international athletic competition. Many individual metonymies have been identified giving rise to the definition above, as well as to competing definitions differing in important descriptive and theoretical details (Croft 2006; Littlemore 2015; Peirsman and Geeraerts 2006). However, all the theories of metonymy put forward within the theoretical framework of Cognitive Linguistics share the idea that metonymy abounds in language (Panther and Thornburg 2007).

Ruiz de Mendoza Ibáñez (2011) has proposed that the large number of documented metonymies can be seen as instantiations of two general types: those metonymies in which the vehicle forms a salient element of a complex CCM and those in which the vehicle denotes a CCM with only one part being salient. The former are called source in target metonymies, as for example in the piano has been drinking, where the instrument denotes the musician. The latter are called target in source metonymies as for example in the sentence Germany won the world cup, where the football team forms the salient sub-part of the CCM of the Nation. Importantly, Geeraerts and Peirsman have drawn this idea out further, and argued that target in source metonymies can also account for cases of meaning facetization – those cases where particular semantic sub-parts or aspects of a conceptual structure become the operative meanings (Geeraerts and Peirsman 2011). For example, when book is taken to refer only to the text within, rather than the tome, the source expresses only a part of its meaning range. This development of the idea of target in source and source in target metonymies provides an effective theoretical tool for the description of the diachronic development of the root denoting ‘club’ in Dene languages. The semantic extensions of the root discussed in Section 5 emerge from the application of both of the kinds of metonymies posited by Ruiz Mendoza Ibáñez, but they produce different effects over time. While source in target metonymies lead to the lexicalization of complex events in which clubs or stick-like elements play central roles, target in source metonymies bring semantic elements to the fore while backgrounding or eliding others.

From a diachronic perspective, the cognitive process involved in target in source metonymies is close to schematization which Talmy defined as: ‘a process that involves the systematic selection of certain aspects of a referent scene to represent the whole while disregarding the remaining aspects’ (Talmy 1983: 225). The result of this process is what traditional historical semantics identifies as a broadening of meaning. Broadening or generalization typically involves a taxonomic change in which an older meaning becomes a hyponym of a newer meaning, and consequently loses some of its semantic features (Blank 1997: 201). In his important discussion of the role of contiguity and metonymy in semantic change, Koch sharply distinguished between broadening and metonymy on the grounds that, conceptually, taxonomic relations and relations of contiguity should be clearly distinguished (Koch 2011: 275–276). The analysis of the semantic changes undergone by the root denoting ‘club’ in Dene languages presented in Section 5 suggests that repeated target in source metonymies can lead to broadening in certain cases. This point will be returned to following the discussion of the data.

5 The polysemic network around ‘club’

The lexical network described in this section originates in the semantics associated with the ancient root *x̣ʷɑˀɬ. Cognates of this root are found in all branches of Na-Dene. The reflexes of the Proto-Dene root *x̣ʷɑˀɬ in the sample of Dene languages considered in this study are listed in Table 1. The attested root forms are written in the orthographies for each language which consistently encode the distinction between voiced and unvoiced back velars as gh and x respectively.

The qualia structure of ‘club’ described above forms the conceptual substrate on which new meanings evolve – predominantly through processes of metonymy. The stages of the evolution of this root are shown schematically in Figure 2: the development from stage 1 to stage 2 comes about through a target in source where the telic role of the club as a material object is the target. The shift from stage 1 to stage 2 involves a sub-event for whole event metonymy (which is a type of target in source metonymy). The shift from stage 3 to stage 4 involves highlighting the movement of the shape of club-like or stick-like objects – again through a target in source metonymy.

Figure 2:

Stages of the polysemic extensions of *x̣ʷɑˀɬ.

As indicated in Figure 2, the meaning evolution of this root can be grouped into four stages. These stages reflect diachronic phases in evolution with the earlier stages being shared by all the Dene languages considered here, and the later ones evolving only in some of the languages. These stages and the lexical items that instantiate them in the Dene languages are described in Sections 5.1–5.4.

6 Metonymy, schematization, and Dene classificatory verbs

Dene languages have evolved a highly schematic set of constructions known as classificatory verbs which describe objects being handled, being in position, or in moving independently. The verbs are matched to particular objects on the basis of properties of form, shape or consistency. The classificatory verbs are a distinct sub-set of the Dene verbal roots that have received sustained attention in the linguistic literature (most notably Davidson et al. 1963; Hoijer 1945a; de Reuse 2001; Rice 1998). The roots in this lexical subsystem encode information about a salient object as well as the type of event this object is involved in. The kind of events described by this system fall into four types, according to whether the object is at rest or in a location, being handled or manipulated, under active control, or moving independently (Davidson et al. 1963). Along with nature of the event, the semantics of the verb roots foreground the physical and constitutive properties of the object. To describe the handling of objects of a certain sort, a verb root is chosen whose semantics match the properties of the object, as in (28).

The verb is appropriate in this case, since it matches the properties of a gun as being long and rigid or stick-like (the description common in the literature for the meaning of this verb root in other Dene languages is ‘slender stiff object’). A mismatch between the verb semantics and the properties of the object is typically deemed ungrammatical. Consequently, there are classes of object that are matched with the verb. However, some objects can occur with multiple roots. The semantic properties of the root then play a role in the interpretation of the nominal object.

In (29) the speaker has used the classificatory root -’a which is used when handling compact objects (or co). The tea is therefore construed in typical compact forms such as a teabag or a box of tea. Different classificatory stems produce different construals of the noun, as in (30) and (31).

In (30), the root specifies the object as being a shallow open container. The target meaning ‘cup’ (or other contextually relevant container) is inferred from the CCM via a target in source metonymy (Ruiz de Mendoza Ibáñez 2011). The term lidí denotes a cultural model of tea as the substance introduced by Europeans and occurring in the form of chopped leaves, bags, tins, or as a beverage imbibed from shallow containers. Lidí is the source that attains the precise target meanings through metonymic inference.

Beyond triggering facet readings in overt nominals, classificatory verbs can also have discourse tracking functions. This secondary property acts as a constraint on creative uses of classificatory verbs since the referent needs to be made evident by contextual cues (Axelrod 2000; Rushforth 1991). Nonetheless, metaphorical and other extensions of the verb used do occur.

The phrase in (32) is an idiom used to talk about news or gossip, but nonce formations are also possible and technically ungrammatical uses occur, especially in humorous contexts (Sapir 1932).

The system of classificatory verbs is found in all Dene languages (Hoijer 1945b; de Reuse 2001; Young and Morgan 1987). This system has grown over time. The evidence for this growth comes from the observation that different Dene languages have different classificatory verbs, while sharing a core set (de Reuse 2001; Vajda 2019). In addition, classificatory verbs are not found in the closest sister languages to the Dene stock: Eyak and Tlingit. From a semantic perspective, all classificatory verbs specify two things: (1) the properties of a salient event participant and (2) whether the object is being handled, moving independently, or at rest (de Reuse 2001: 76). Structurally, the classificatory verbs are expected to occur only with specific strings of derivational affixes. As de Reuse notes, this latter criterion serves to divide the category of classificatory verbs into better and less clear instances. For instance, some verbs clearly fit the semantics, while not participating in all the same patterns as the clearest cases. There is some variation in the criteria used and some authors include stems which others don’t. For example the root denoting ‘club’ or ‘to club’ is considered a classificatory verb by dint of its semantics in the Koyukon language (or some analyses of the system in those languages), but not in others. As the analysis in Section 5 shows, the roots denoting ‘club’ can also be used to express much more schematic meanings combining properties of a form that is an elongated object with various states and movements. In many derivations, constructions involving the ‘club’ root have semantic properties very much in line with those of true classificatory verbs. Classificatory verbs are, therefore, best understood as constituting a prototypical category as argued by de Reuse (2001).

The emergence of this class of verbs can be traced to Proto-Dene (Vajda 2019), but the growth and evolution of this system of verbs remains an open research question. While all Dene languages have classificatory verb stems, their numbers differ from language to language (Davidson et al. 1963; de Reuse 2001; Vajda 2019). Vajda has suggested that the origin and growth of the classificatory system may have come about through contact with other languages (Vajda 2019), especially the Wakashan languages spoken in British Columbia. Classificatory verbs are rare typologically (Aikhenvald 2003: 153) and are sometimes thought to have originated in incorporated nouns Aikhenvald (2003). However, Fortescue, working on the origins of classificatory verbs in Wakashan argued that they evolved from other verbs: ‘Arguably, then, some classificatory roots developed in the Wakashan languages from earlier ordinary verbs undergoing either collocation restrictions or semantic widening and generalization’ (Fortescue 2006: 280). As this seems to be a process very similar to the evolution of *x̣ʷɑˀɬ in some Dene languages, semantic changes through target in source metonymies might be a productive future area of research in the diachronic study of classificatory verbs.

7 Conclusion

The classification of the wide range of possible metonymies into just two types by Ruiz Mendoza Ibáñez provides a powerful but elegant tool for the description of semantic extensions from a diachronic point of view. In conjunction with detailed descriptions of Cultural Cognitive Models, they can help to model patterns of association leading to lexical changes. The reflexes of the root *x̣ʷɑˀɬ share a deep history dating back to Proto-Na-Dene, diversifying only in the more recent past in three evolutionary stages. The further evolution of reflexes of *x̣ʷɑˀɬ are suggestive of the development of a schematic form of the verb resembling members of the classificatory verbs, which specify the nature of the objects they denote in terms of shape and constitution. Taking metonymic change as an analytical tool revealed extensions of meanings through source in target and target in source metonymies in a language family that is typologically quite distinct from those in which these theoretical tools were developed. It is hoped that this study contributes to the greater application of Cognitive Linguistics to data from other languages, and that, in turn, the theoretical framework may benefit from a broader typological scope.

Data availability statement

All the data used in this paper are listed in the examples (1)–(32). With the exception of examples (1), (2), (5), and (9), the data are drawn from publicly available, published dictionaries and grammars. Examples (1), (2), (5), and (9) are drawn from an unpublished manuscript. This manuscript may be available on request from the Tsuut’ina Gunaha Institute (https://www.tsuutinagunahainstitute.com/). However, the data taken from that manuscript are represented in their entirety in the examples above.