Place-naming strategies in Inuit-Yupik and Dene languages in Alaska

Draft 2013‐07‐21 Place‐naming strategies in Inuit‐Yupik and Dene languages in Alaska 1. Introduction1 The two major Alaskan language families, Inuit‐Yupik (Eskimo) and Dene (Athabascan), share a border which extends in an arc nearly 2000 km long from Cook Inlet in the Gulf of Alaska, paralleling the coast of Alaska all the way to the Canadian border.2 Both families extend further into Canada; Inuit‐Yupik extends all the way to the east coast of Greenland. Along this shared border many thousands of places have been named, and these names—and the place‐naming strategies which underlie them—provide insight into indigenous conceptualizations of the landscape. Inuit‐ Yupik place‐naming is grounded in human affordance; names are assigned based on people’s relationship to the land. In contrast, Dene place‐naming is highly deterministic, based on a generative geographic directional system. There are of course plenty of exceptions which prove these rules, but broadly speaking these generalizations hold across the two language families. Here I suggest that this difference in place‐naming strategies can be explained in part in terms of differences in the demonstrative systems of the two language families. Both Inuit‐Yupik and Dene languages have elaborate systems of words expressing spatial relations, allowing a much finer distinction than is possible with the proximal ‘this’ and distal ‘that’ in English. However, the function of the demonstrative system differs greatly in the two language families. In Inuit‐Yupik languages the demonstrative system functions primarily on the local level and have limited application to the larger landscape domain. In Dene languages the demonstrative systems are fundamental to the conceptualization of landscape, playing a key role in place naming strategies. To a certain extent this result is not unexpected. Landscape is a semantic domain whose categorization is known to vary across languages. As Levinson (2008:257) notes, “from a geological point of view [landscape] is mere deformation of a continuous surface, so that discrete units and categories must be the construction of the cognizer.” In other words, concepts such as “mountain” are not universal in either denotation or connotation. Put another way, “different language groups/cultures have different ways of conceptualizing landscape, as evidenced by different terminology and ways of talking about and naming landscape features” (Mark et al. 2007). Further evidence from specific languages can be found in the various case‐studies contained in Mark et al. (2011). However, landscape categorization is not just limited to feature terminology. Place names also provide insight into the categorization of landscape, and these names may also be deeply embedded within orientation systems. For example, a language which employs a riverine orientation system embodies a very different approach to landscape than does a system which employs a cardinal system based on compass directions, even though both are “absolute” systems in the sense of Levinson (2003). In a riverine system movement and location are contextualized within the parameters of upstream‐downstream and landward‐waterward. The valley system is “burned in” to a speaker’s relationship to the land. In contrast, in cardinal system locations and movement can be described without any reference to the notion of valley. In comparing Inuit‐Yupik and Dene languages the relevance of orientation systems is easily overlooked. On first glance, the two language families appear to have very similar orientation systems, both essentially riverine in nature (though coastal languages substitute upcoast‐ downcoast for upstream‐downstream). The geographic dimension is based on either a riverine or coastal template, consisting at its core of an orthogonal distinction between an upstream‐ downstream axis (or upcoast‐downcoast) and a landward‐waterward axis. The basic geographic template is superficially similar in the two language families. This can be illustrated by comparing the basic orientation roots in Alaskan Yup’ik (Figure 1) and Koyukon (Figure 2). 1 Draft 2013‐07‐21 Figure 1: Coastal orientation roots in Yup’ik (Inuit‐Yupik) piavet kanavet kiavet uavet Figure 2: Riverine orientation roots in Koyukon (Dene) -naane -naa’e -tlene -do’u -negge Such systems of orientation are quite common in the world’s languages, being found commonly for example in Austronesian languages (Adelaar 1997). However, what is of interest for the present purposes is that these systems of orientation derive ultimately from larger systems of demonstratives, and the paths by which these larger demonstrative systems have come to be reduced to orientation systems differs significantly between Inuit‐Yupik and Dene languages. In the remainder of this chapter I begin by first describing the demonstrative and orientation systems in Inuit‐Yupik (§0) and Dene (§3) languages. Place naming strategies are compared in §4. 2. Inuit‐Yupik orientation systems Inuit‐Yupik languages are notably famous for their complex systems of demonstratives. The precise realization varies across individual Inuit‐Yupik languages. It is most elaborated in Alaskan Yup’ik, which contrasts three dimensions corresponding roughly to distance from the deictic center; a dimension termed indicability; and a dimension term accessibility (Jacobson 1984). Although the structure of the system varies greatly across individual languages, the forms correspond regularly, permitting the entire system to be reconstructed at the level of Proto Inuit‐Yupik (PIY) by application of the standard tools of the linguistic comparative method (see Table 1). Although we think of orientation systems in the modern languages as being based on relationship to water (river or coast), the PIY demonstrative system can be better described as an elevation‐based system which distinguishes up, down, and same level. To these basic elevations are added proximal (near deictic center) and distal (away from deictic center) terms which are independent of elevation. Such elevation‐based systems are not uncommon in the world’s languages (Diessel 1999). 2 Draft 2013‐07‐21 Table 1: Proto Inuit‐Yupik demonstrative roots (Fortescue et al. 1994) RESTRICTED EXTENDED OBSCURED *uv‐ *mað‐ *im‐ PROX DIST LEVEL DOWN UP ACC NON‐ACC ACC NON‐ACC ACC *kiv‐ *iŋ‐ *kan‐/*kað‐ *piŋ‐ *kiɣ‐ *ik‐ *uɣ‐ *pik‐ *qav‐ *av‐ *un‐ *pav‐ *qaɣ‐ *aɣ‐ *unəɣ‐ *paɣ‐ *qam‐ *am‐ *cam‐ *pam‐ NON‐ACC *qakəm‐ *akəm‐ *cakəm‐ *pakəm‐ The PIY demonstrative system adds two additional dimensions, indicability and accessibility. The dimension labeled indicability by Jacobson (1984) has to do with visibility and contrasts restricted (confined to limited extent), extended (moving or unconfined), and obscured (blocked from view). The semantics of the dimension of accessiblity are less consistent but nonetheless clearly defined for each accessible (ACC) and non‐accessible (NON‐ACC) pair of terms. The precise semantics of the system need not concern us here. Rather, what is of interest for the present purposes is the way this system is realized in the individual Inuit‐Yupik languages and in particular how the system maps into the landscape domain. Not all of the original PIY demonstrative roots survive in the modern languages, and the modern orientation systems make use of only a small subset of the larger demonstrative system. Moreover, the modern orientation systems are based not on the up‐level‐down elevation distinction found in the reconstruction PIY system but rather on an orthogonal coordinate system. To derive the modern orientation systems from the original PIY demonstrative system the modern languages employ a subset of the original demonstratives and then reassign their semantics to form an orthogonal grid. Each modern Inuit‐Yupik language achieves this in a slightly different way. Consider first the Inupiaq (North Slope dialect) demonstrative system, as shown in Table 2. The table is laid out here to parallel the organization of the PIY demonstratives in Table 1. Gaps indicate PIY demonstratives which lack a reflex in Inupiaq. The highlighted cells indicated terms which are used in the orientation system, to be discussed below. Table 2: Alaskan Inupiaq demonstrative roots. Those used in the orientation system are shaded. (MacLean To appear) RESTRICTED EXTENDED uv‐ PROX OBSCURED ma‐ sam‐ ACC NON‐ACC ACC NON‐ACC ACC NON‐ACC DIST LEVEL kiv‐ kig‐ ik‐ qag‐ ag‐ DOWN kan‐ qav‐ av‐ un‐ qam‐ am‐ sam‐ pam‐ qakim‐ akim‐ sakim‐ pakim‐ UP pik‐ pag‐ Comparing the Inupiaq demonstratives with their PIY counterparts two things are immediately evident. First, both the forms of the Inupiaq roots and their structural distribution are very much like those found in PIY. Only some minor sound changes have occurred, such as PIY *c > Inupiaq s. (Note that in the Inupiaq practical orthography <g> represents [ɣ], so is unchanged from PIY.) Second, there are some gaps in the table, reflecting PIY demonstrative roots which have been lost in modern Inupiaq. In general, as one moves east across the arctic fewer of the original PIY demonstrative roots survive in the modern languages. In Inupiaq these gaps lead to the partial collapse of the accessibility dimension with the restricted and extended terms. 3 Draft 2013‐07‐21 The demonstrative system provides the basis for and coexists with an orientation system which contrasts the orthogonal dimensions of up‐down the coast versus waterward‐landward. The full orientation system also includes terms deriving from winds, with the choice of wind term varying greatly by location (Fortescue 1988). However, if we ignore the wind terms for moment we can posit a kind of intermediate orientation system based only on the demonstrative system, as in Table 3. Table 3: Alaskan Inupiaq orientation system (wind terms ignored) UPCOAST DOWNCOAST WATERWARD LANDWARD RESTRICTED EXTENDED OBSCURED kiv‐ ik‐ kan‐ pik‐ qav‐ av‐ un‐ pag‐ qam‐ am‐ sam‐ pam‐ The Inupiaq orientation terms are precisely those which occur shaded in Table 2. Of the six restricted Inupiaq demonstrative roots in Table 2, only four are employed in the orientation system. As in all Inuit‐Yupik languages the proximal term is not employed in the orientation system. The DOWN and UP terms kan‐ and pik‐ are used for the waterward‐landward axis, i.e., ‘down toward coast’ vs. ‘up away from coast’. The accessibility distinction is irrelevant here since these terms have no counterpart in the accessibility parameter in modern Inupiaq. The single restricted level term ik‐ is used to mean ‘down the coast’ or ‘to the left facing the water’. The accessible distal term kiv‐ is used to mean ‘up the coast’ or ‘to the right facing the water’. This latter term retains as well its demonstrative sense of ‘inside’, which contrasts with the non‐accessible form kig‐ meaning ‘outside’. This results in homophony between the orientation system sense of kiv‐ meaning ‘down the coast’ and the more localized sense meaning ‘inside’. This ambiguity is clearly the result of the original demonstrative system being extended for use as part of the orientation system. A general rule for mapping the demonstrative system onto the orientation system is that wherever an accessible term exists it is the one employed in the orientation, and thus like kiv‐ becomes polysemous between its larger orientation sense and its more localized demonstrative sense. The corresponding non‐accessible term is not used as in the orientation system but maintains its demonstrative sense. In particular, none of the obscured non‐accessible terms are employed in the orientation system, but they continue to be used as demonstratives: qakim‐ ‘out there, not visible’; akim‐ ‘over there across, not visible’; sakim‐ ‘out there in the arctic entry, not visible’; and pakim‐ ‘up there on the roof, not visible’. Quite a different picture emerges in the neighboring Central Alaskan Yup’ik language. Here the PIY demonstrative system is preserved almost wholly intact, as shown in Table 4. Unlike the Inupiaq system there are no gaps to facilitate choice of accessible or non‐accessible term for use in the orientation system. 4 Draft 2013‐07‐21 Table 4: Central Alaskan Yup’ik demonstrative adverbs (terminalis case). Those used in the orientation system are shaded. (after Jacobson 2012) RESTRICTED EXTENDED wavet maavet PROX ACC DIST LEVEL DOWN UP NON‐ACC kiavet keggavet yaavet ikavet kanavet uavet piavet pikavet ACC NON‐ACC qavavet avavet unavet pavavet qagaavet agaavet un’gavet pagaavet OBSCURED ACC NON‐ACC qamavet qakmavet amavet akmavet camavet cakmavet pamavet pakmavet The Yup’ik demonstrative system does not make use of level demonstratives in the orientation system. Rather, both the accessible and non‐accessible down terms are used. The accessible term kana‐ ‘down there’ is used for the direction toward water, while the non‐accessible term ua‐ is used for the ‘downriver’ direction. The toward water term kana‐ is paired with the UP accessible term pia‐ ‘up there’ to mean away from water, while the downriver term ua‐ is paired with the DISTAL accessible term kia‐ ‘inside’ to mean upriver. This yields an orthogonal riverine directional, illustrated in Figure 3 with restricted terms inflected for terminalis case. Figure 3: Yup’ik orientation system in a riverine system (restricted, terminalis case) kanavet DOWN, ACC kiavet rivermouth DIST, ACC uavet DOWN, NON-ACC piavet UP, ACC The same Yup’ik orientation terms can also map onto a coastal system in which the downward non‐ accessible term denotes not downriver but rather down the coast or to the right facing the water, and the distal accessible term denotes not upriver but rather up the coast or to the left facing the water. The Inupiaq and Yup’ik systems represent but two of the many ways in which the PIY demonstrative system is realized in the modern Inuit‐Yupik languages and in which those demonstratives systems are extended to wider‐scale orientation. A more extreme example can be found in West Greenlandic. As shown in Table 5 the Greenlandic demonstrative system is greatly reduced from PIY. In no dimension other than the proximal is an entire series of roots preserved. 5 Draft 2013‐07‐21 Table 5: West Greenlandic demonstrative roots. Those used in the orientation system are shaded. RESTRICTED EXTENDED OBSCURED u‐ ma‐ (im‐) PROX ACC DIST LEVEL DOWN UP NON‐ACC ACC kig‐ ik‐ qav‐ av‐ pik‐ pav‐ NON‐ACC ACC NON‐ACC qam‐ kan‐ sam‐ The lack of terms in the extended and obscured domains has led to an orientation system which no longer makes this distinction. Rather, the Greenlandic orientation system uses terms drawn from both the restricted and extended subsystems, and terms which may have originally belonged to different dimensions of the demonstrative paradigm. Thus, an originally accessible demonstrative kan‐ ‘down (toward the coast)’ is now opposed to an originally non‐accessible demonstrative pik‐ ‘up (away from the coast)’. The original non‐accessible ‘down’ demonstrative has been lost, as has the original accessible ‘up’ demonstrative. With the accessibility dimension thus extinguished the juxtaposition of av‐ and qav‐ is now unproblematic. Figure 4: Greenlandic coastal orientation roots pikkanaqav- av- While the Greenlandic system is not directly relevant to the Alaskan languages considered here, it serves to illustrate the significant variation among the Inuit‐Yupik languages in both the realization of the demonstrative system and the use of the demonstratives to form an orientation system. While the demonstrative terminology may be readily reconstructable to PIY, the individual demonstrative systems themselves function quite differently. These differences are greater still when we move to larger geographic scales beyond a single village. As one moves toward larger distance scales the undulations of the local coastline vary, and the need for less locally‐dependent terminology increases. The geographic integrity of the system is maintained by employing wind terms in lieu of some of the demonstrative roots. This strategy is found throughout the Inuit‐Yupik languages, but the particular implementation varies greatly not only by language but also by geographic location within a given language (Fortescue 1988). This variation can be illustrated by comparing Yup’ik and Inupiaq (North Slope) directional terms (Figure 5). Both Yup’ik and Inupiaq employ reflexes of the wind terms PIY *nəɤəʀ and *uŋalaʀ. In Yup’ik these wind terms negeq and ungalaq are paired with the upriver/downriver (or upcoast/downcoast) terms. In Inupiaq the wind terms nigiq and uŋalaq are paired with the toward/away from coast terms. 6 Draft 2013‐07‐21 Figure 5: Yup'ik (left) and Inupiaq (right) directional terms compared The variation in the realization of Inuit‐Yupik directional systems can be explained in terms of geography (Fortescue 1988; 2011). In Yup’ik negeq is a north wind, hence orthogonal to the prevailing east‐west trending rivers and their concomitant downstream/upstream terms. In North Slope Inupiaq nigiq is an east wind, hence orthogonal to the toward/away from water direction. So the choice of the downstream/upstream axis in Yup’ik versus the toward/away from water axis in Inupiaq is readily explained. However, the ability of Inuit‐Yupik languages to essentially pick and choose among demonstratives has significant consequences for the conceptualization of landscape. The reification of these orientation terms into an essentially cardinal directional system decouples the terms from the landscape, depriving them of their potential function as guides to the topography and sources for place naming. In practice Inuit‐Yupik orientation terms may have very little to do with the wider landscape. In my own field work with speakers of Yup’ik I have noted a great reluctance to use these orientation terms on any scale beyond the immediate vicinity. Travel along rivers is much more likely to be described either in terms of cardinal directions (e.g., negeq ‘north’) or in terms of movement either with or against the current (e.g., asgur‐ ‘move against the current’) than with the orientation system. So while the Yup’ik demonstrative and orientation systems may be extremely rich and complex, they have little practical relevance to the domain of landscape. Nor, as we shall see in §4, do they play major roles in place naming. 3. Dene demonstrative systems A very different situation is found in the Dene languages. The Proto‐Dene demonstrative system is reconstructed as in Table 6. There are two paradigms corresponding to motion away (allative) and static (punctual). The modern Dene languages add additional dimensions of motion toward the deictic center and static location in an area. 7 Draft 2013‐07‐21 Table 6: Proto Dene demonstrative roots (Leer 1989) UPSTREAM DOWNSTREAM LANDWARD WATERWARD AHEAD ACROSS AWAY ABOVE BELOW ALLATIVE PUNCTUAL *niʔ *daʔ *nəɢ‐ə *tsənʔ *nəs‐ə *ɲaˑnʔ *ʔɑnʔ *‐ə *dəɢ‐ə *ni’‐d *da’‐d *nəχ *tsį’‐d *nəs *ɲą’ˑ‐d *ʔą’ˑ‐d *‐d *deχ Rather than a three‐way elevation‐based contrast up‐level‐down, as in Inuit‐Yupik, the Proto‐Dene system contrasts the four basic demonstratives of upstream, downstream, landward, and waterward, forming a two‐dimensional coordinate system. To these basic terms are added additional terms indicating ahead into open country or water; across water; away in a non‐specific direction; above vertically; and below vertically. The resulting system is thus three‐dimensional and highly descriptive of the riverine valley which characterizes much of the Alaskan Dene landscape. Another major difference between the Inuit‐Yupik and Dene demonstrative systems is that the Proto‐Dene system is realized homologously across the Alaska Dene languages, augmented to varying degrees with prefixes specifying distance and suffixes specifying motion or area. That is, the ancient Proto‐Dene system is robustly preserved in all the modern languages. The Tanacross system in Table 7 is typical in that it includes a four‐way distinction between allative (movement away from deictic center), ablative (movement toward the deictic center), punctual (static location at specific point), and areal (static location in general area). These four paradigms derive ultimately from suffixation patterns which have been historically obscured. Table 7: Tanacross (Dene) demonstrative roots (Arnold et al. 2009)3 UPSTREAM DOWNSTREAM INLAND WATERWARD AHEAD ACROSS AWAY ABOVE BELOW ALLATIVE ABLATIVE PUNCTUAL AREAL ‐ndéʔe ‐ndáˑʔa ‐ndeg ‐tθɛ́ nʔ ‐nɛð ‐náˑnʔ ‐ʔɛ́ nʔ ‐deg ‐ʒégʔ ‐ndîˑdz ‐ndâˑdz ‐ndêdz ‐ndéˑ ‐ndaˑ ‐ndég ‐tθíˑ ‐ndíˑg ‐ndáz ‐ʔáz ‐dêdz ‐ʒêz ‐náˑn ‐déˑ ‐ʒéˑ ‐ndóg ‐tθúg ‐noð ‐ndás ‐ʔóg ‐ndóg ‐ʒóg The forms in Table 7 are stems and must be inflected in order to form a demonstrative word. As in other Dene languages the demonstratives are preceded by a prefix indicating distance from the deictic center. In Tanacross these prefixes are a‐ NEUTRAL, da‐ PROXIMAL, na‐ INTERMEDIATE, ja‐ DISTAL, jaʔa DISTANT. 8 Draft 2013‐07‐21 Figure 6: Tanacross (Dene) demonstratives (distal, allative paradigm) yanáan’ yandedh yatthén’ yandé’e rivermouth yandá’a yandeg ya’én’ As in Inuit‐Yupik languages this three‐dimensional paradigm of demonstratives allows very precise orientation. However, unlike Inuit‐Yupik this extends across the entire language family, robustly attested in each of the Alaskan Dene languages.4 Moreover, the system operates at all levels, being equally relevant applied at the large‐scale geographic domain, within a house, or locally on the human body (Table 8). This contrasts with Inuit‐Yupik languages, where the demonstrative system functions only at a very local scale, while the more generalized orientation system functions at larger scales relevant to the landscape domain. In Dene languages the riverine‐based system permeates all aspects of orientation, independent of scale. Table 8: Examples of Tanacross demonstratives at various scales Example yandá’a Fairbanks ts’ı̨́ tíhhaay ‘I’m going down to Fairbanks.’ dandee didhindah ‘Sit down on the upstream side (of the table).’ nandôg shtthí’ tah sháʔ xúnłee ‘I have lice in my hair.’ Demonstrative distal, downstream, allative proximal, upstream, punctual intermediate, above, areal To understand just how pervasive the Dene riverine orientation system is, consider the usage of the demonstrative system within a house. The extension of demonstratives within a house is based on a conventionalization in which the front door of the house is orientated facing the river. Thus, ‘upstream’ within a house is the direction to the left or right of the door, depending on the direction of flow of the river.5 The upstream‐downstream and inland‐waterward axes are reflected throughout Dene languages in all aspects of language use. The robustness of the riverine demonstrative system within the family underscores the importance of the riverine valley in Dene. As discussed in the following section, it also provides the motivation for place naming strategies. 4. Place naming strategies Although the Inuit‐Yupik and Dene orientation systems are superficially similar, they are reflected quite differently in the toponymic systems for the two language families. The Dene demonstrative 9 Draft 2013‐07‐21 roots define a streamscape based on the orthogonal dimensions of upstream‐downstream and toward‐away from water. This streamscape is used regularly to generate toponymic clusters based on shared generic terms. The core set of generics is composed of *kæq’ ‘stream mouth’, *tł’at ‘stream headwaters’, *wən ‘lake’, and *naʔ/*niq’ə ‘stream’ (where the asterisk indicates a reconstructed Proto‐Dene form), as in Figure 7.6 These terms are not related to the demonstrative system, but they are determined by that system. That is, the riverine structure of the demonstratives delineates a linear valley template to which these landscape terms are assigned. As with the demonstrative system, reflexes of the proto‐Dene streamscape generic terms are robustly attested in all the modern Alaskan Dene languages. Figure 7: Proto‐Dene streamscape generic terms Proto-Dene streamscape terms ,pc顕"1",pksÓ殴 ,vŽcv ,y殴p *kæq’ The system is GENERATIVE in the sense that for any given specific term, all generics can (and usually do) occur (Levinson 2003; Kari 2010b). As an example consider the Tanacross word ch’inchedl ‘nose ridge’. This word occurs as the name for a prominent ridge rising some 500 m to the north the Tanana River. It is used as a specific term to generate a cluster of names in that locality, including: Ch’inchedl Ndiig ‘nose creek’ (<*niq’ə), a creek which drains the back side of Ch’inchedl; Ch’inchedl Menn’ ‘nose lake’ (< *wən), the lake from which the creek flows; Ch’inchedl Tl’aa ‘nose headwaters’ (< *tł’at), the headwaters of the creek; and Ch’inchedl Teyy’ ‘nose hill’, a peak above the headwaters. This last generic term teyy’ ‘hill’ augments the basic streamscape system. Crucially, a given specific term may be repeated only if it is not used to generate name clusters. Thus, the Tanacross name Ch’inchedl is a singleton, a unique name which is not repeated. The singleton Ch’inchedl can be contrasted with the specific term ch’endaag ‘mineral lick’. This latter term occurs in the name Ch’endaag Menn’ ‘mineral lick lake’, which is repeated fully five times. This is rather striking in that the Tanacross territory is among the smallest of any Athabascan language in Alaska, and the five tokens of the name ‘mineral lick lake’ are located within 10‐50 km of each other. However, none of these names participates in a larger generative naming pattern. That is, the specific term ch’endaag does not occur in any other derived forms—either referring to neighboring or distant features. There is simply no ‘mineral lick mouth, ‘mineral lick creek’, ‘mineral lick headwaters’, etc. This distinction between specific terms which generate name clusters and those that do not is clearly functional. Because those specifics which generate name clusters are not repeated outside the cluster, these singleton specifics essentially denote a region or 10 Draft 2013‐07‐21 territory. Names for individual parts of the territory can be generated readily even by those unfamiliar with the territory by drawing the generative principles of the Dene naming system. The generative capacity of the Dene naming system is so deeply entrenched as to seem almost deterministic. This is particularly true for the generic term *kæq’ ‘mouth’. Once one knows the name of a particular river, the name of its mouth is also known. This is not simply a matter of specifying a location using a geographic term. Rather, if the mouth is named its name is almost invariably based on *kæq’; alternate names are simply not possible. These mouth names are often highly lexicalized and often borrowed into English with the generic term. Thus, at the mouth of the Kantishna River is located a village known in English as Crossjacket. The Lower Tanana name for the Kantishna River is K’osr No’, a binominal name composed of the specific k’osr ‘polishing stone’ and the generic no’ (< *naʔ). Thus, the village at its mouth must be K’osr Chaget (< *kæq’), which is readily seen to be the etymological source of the English name. Examples like this abound across the Dene territory in Alaska (Table 9). Table 9: Some common village names with Dene etymologies based on *kæq’ English Salcha Bearpaw Chena Healy Lake Ketchumstuck Holikachuk Anvik Stony River Chistochina Copper Center Allakaket Hughes McGrath Dene Name Soł Chaget Ch’edzaya’ Chaget Ch'eno' Khwdochaget Mendees Cheeg Saages Cheeg Holjichak’ Gitr’ingith Chagg K’qizaghetnu Hdakaq’ Tsiis Tl’edze’ Caegge Tl’aticae’e Aalaa Kkaakk’et Hut’odlee Kkaakk’et Tochak’ Language Lower Tanana Lower Tanana Lower Tanana Tanacross Tanacross Holikachuk Deg Xinag Dena’ina Ahtna Athna Koyukon Koyukon Upper Kuskokwim The generative capacity also has synchronic relevance. New names are rarely coined in Dene languages, as most of the country is already named, obviating the need to assign new names. However, where new names are coined the riverine system provides the template. Thus, a new name near a mouth of a creek will almost invariably be named using the generic ‘mouth’. There are exceptions to this rule, but these arise only when there is an overriding influence from a competing naming strategy. There is a single such example in the list of 2436 Ahtna names. The name Naghilden, located at the mouth of Canyon Creek. Rather than the generic cae’e ‘mouth’ it contains a generic den ‘place, area’ and means literally ‘waterfall place’. In this single case the prominence of a nearby hydrologic feature took precedence, but in the vast majority of cases the system exhibits a constrained productivity in which new names must follow the generative strategy. The use of generics in Inuit‐Yupik languages is quite different. In particular, Inuit‐Yupik place naming is not generative. To see just how different the Inuit‐Yupik strategy is it is useful to compare the use of the Inuit‐Yupik generic *paðə ‘mouth’ with Dene *kæq’ə. As in Dene languages the Inuit‐ Yupik generic ‘mouth’ can be used in place names. For example, the Central Yup’ik name for the village of Stony River is Teggalqum Kuigan Painga, incorporating the Yup’ik word pai ‘mouth’ (<*paðə). This village is located in a bilingual region bordering Yup’ik and Dena’ina (Dene) territory, so it also has a Dena’ina name, K’qizaghetnu Hdakaq’, which also incorporates the 11 Draft 2013‐07‐21 Dena’ina generic kaq’ ‘mouth’ (< *kæq’ə). Yet the name for Stony River is actually quite exceptional in this regard. Most Yup’ik names for villages located at river mouths do not in fact contain the generic ‘mouth’. For example, Egegik, located at the mouth of the Egegik River is known simply as Igyagiiq, a generic term meaning ‘throat’ and referring metaphorically to ‘the area of a river a little ways back from the mouth’ (Jacobson 2012:279). This name contains no specific component but is simply a landscape generic. It is descriptive but not generative. The contrast between Inuit‐Yupik and Dene extends to features beyond river mouths themselves. A large mountain above the Cheeneetnuk River, known locally as Swift River Mountain, is called in Deg Xinag (Dene) Jonetno’ Xidochagg Deloy Chux, literally ‘big mountain at mouth of Jonetno’ (chagg < *kæq’ə). Jonetno’, literally ‘clear water creek’, is the Deg Xinag name for Cheeneetnuk River. But the Yup’ik name has nothing to do with either the creek or its mouth. Instead this mountain is known in Yup’ik by the highly descriptive name Kiturciigalnguq, meaning ‘place one cannot pass’. Gusty Mikhail explains the name as follows: “That means ‘we can’t pass mountain’. You see the river is so crooked that that mountain when you go up, you go sometimes behind like that, sometimes it hit us. Sometimes sideways. You can’t pass it. That’s why they call him that way. You can’t pass that mountain.” (Mikhail, quoted in Kari 1980) While the Deg Xinag language anchors the name generatively in the landscape via generics ‘river’, ‘mouth’ and ‘mountain’, the Yup’ik name forgoes landscape terminology in favor of a name based on human affordance. This difference is fundamental to understanding place naming strategies in the two languages—a point to which we will return below. Not only is the usage of the Inuit‐Yupik and Dene ‘mouth’ generic quite different, the terms also have fundamentally different semantics. Inuit‐Yupik *paðə has broad semantics referring to an ‘opening’ or ‘entrance’. This broad semantics is preserved in most of the languages of the family, including Yup’ik (Fortescue et al. 1994). Thus, Yup’ik pai (variant paa) is can refer not only to the ‘mouth of river’ but also to ‘opening of den, bottle, etc.’ or the ‘cockpit of kayak’ (Jacobson 2012). In contrast, the Dene generic *kæq’ is restricted to the landscape domain, referring only to ‘river mouth’. It is distinguished from roots such as du ‘orifice’ and zaq’ ‘mouth (anatomical)’. As I have argued previously, this Dene generic serves to delineate a prototypical Dene streamscape centered around a valley. The term *kæq’ is not just ‘river mouth’ but more properly ‘mouth of a valley’, as evidenced for example by the Lower Tanana name Dradlaya Chaget, which is located not at a river mouth, as the term chaget (< *kæq’) might imply, but at the place where the river leaves a steep‐ walled valley and spills onto the Minto Flats (Holton 2011:234). The Dene examples above reflect the fundamental importance of the riverine orientation system for Dene place naming. Although the grammar of demonstratives is extremely complex in both Dene and Inuit‐Yupik languages, only in Dene is the demonstrative system so fully embedded within place names. This becomes especially apparent when place naming strategies are compared quantitatively. In order to do this we must consider comprehensive name inventories, since selective name lists could potentially skew the results. Within Alaskan Dene the most comprehensive published place name inventories are those for Ahtna (Kari 2008) and Lower Tanana (Kari et al. 2012), listing 2208 and 1064 names, respectively.7 No study of similar scope has yet been published for Inuit‐Yupik languages in Alaska; however, we are fortunate to have available a comprehensive list of 1007 names for the Inuinnait of Western Canada, which can be used as a proxy for Alaskan Inuit‐Yupik languages (Collignon 2006). The Ahtna and Inuiannait territories are comparable in size, and the name inventories are similarly exhaustive.8 The Ahtna name density is 12 Draft 2013‐07‐21 thus roughly twice that of the Inuinnait, but the two systems can nevertheless be compared without undue risk of sampling error. As we expect given the claimed generative capacity of Dene naming, more than 60% of Ahtna names are binominal (or trinomial) and headed by one of 22 landscape generics. In contrast, only 21% of Inuinnait names are based on a landscape generic. Moreover, nearly half of these names (94 of 207) are duplicates, so that the percentage of unique Inuinnait names based on a landscape generic is more like 11%. In fact, name duplication is much more prevalent in Inuinnait than in Dene. Fully 26% (257 of 1007) of Inuinnait names are duplicates, compared to only 6% (155 of 2436) of Ahtna names and just 4% (44 of 1064) of Lower Tanana names. Even if we ignore name duplication the percentage of landscape‐based names in Ahtna is three times that in Inuinnait. However, this figure ignores grammatical structure of Dene binomial names. Inuinnait names based on landscape terms include many which are simply a landscape term or a landscape term modified by an adjectival suffix (postbase). Table 10: Examples of Inuinnait names based on landscape generic (Collignon 2006) Name Kuunayuq Kuugaluk Kuugaaryuk Palliq Qikiqtahuk Tahialuk Literal ‘long river’ ‘big river’ ‘small river’ ‘bay’ ‘small island’ ‘lake’ This is also true of the Yup’ik (Inuit‐Yupik) names on Nunivak Island, one of the few subregions of Inuit‐Yupik territory in Alaska for which comprehensive published name data are available. A large number of Nunivak names consist only of a generic name with a modifying adjectival suffix. This includes 21 single‐word names consisting of the generic root kuik‐ ‘river’ together with one or more derivational suffixes. Table 11: Yup’ik place names (tokens) based on generic kuik‐ ‘river’ on Nunivak Island (Drozda 1994) name Kuicungar Kuigaar (7) Kuigaarag Kuigaaremiut Kuiggavluar (2) Kuigglugar Kuigglugarmiut Kuigkaun Kugimiutuli Kuigpii Kuiguar (2) Kuileg literal ‘dear little river’ ‘little river’ ‘two little rivers’ ‘village of little river’ ‘just a little river’ ‘poor old river’ ‘village of poor‐old‐river’ ‘future river’ ‘one who stays at the river’ ‘its big river’ ‘imitation river’ ‘one with a river’ Frequent use of generic names leads naturally to a high incidence of name repetition. The seven tokens of Nunivak Kuigaar is one example of such repetition. We also find on Nunivak Island five tokens of Pengur ‘dune’ (as well as fifteen more names derived from the same root); 4 tokens of 13 Draft 2013‐07‐21 Penarrat ‘small cliffs’ (as well as twenty other names derived from penat ‘cliffs’); and 4 tokens of Qemirrlag ‘major hill/ridge’ (as well as fourteen other terms based on the root qemir ‘hill/ridge’). Names comprised solely of a landscape generic are impossible in Ahtna, and names based on adjectival modification of a landscape generic are extremely rare, comprising less than 2% of the inventory. Such names tend to refer to major features, such as Dghelaay Ce’e for Mt. McKinley, literally ‘big mountain’. The more common generative pattern can be exemplified by the Ahtna names based on yidateni ‘jaw trail’. The nine names in Table 12 make use of landscape generics referring to ‘canyon’, ‘mountain’, ‘river mouth’, ‘hill’, ‘creek’, ‘headwaters’, ‘lake’, and ‘uplands’. In addition, the specific term itself occurs as a name Yidateni, denoting a convex landform. The landscape generics themselves do not occur as names. Table 12: Ahtna names based on specific term Yidateni (Kari 2008:27) Name Yidateni Dyii Yidateni Dyii Dghelaaye’ Yidateni Caek’e Yidateni Caek’e Tes Yidateni Na’ Yidateni Tl’aa Yidateni Tl’aa Bene’ Yidateni Dghelaaye’ Yidateni Na’ Ngge’ Literal ‘jaw trail canyon’ ‘jaw trail canyon mountain’ ‘jaw trail mouth’ ‘jaw trail mouth hill’ ‘jaw trail creek’ ‘jaw trail headwaters’ ‘jaw trail headwaters lake’ ‘jaw trail mountain’ ‘jaw trail creek uplands’ The names in Table 12 form what Kari has described as a place name CLUSTER built upon a single specific term. Examples of such clusters abound in Alaska Dene languages. Within a cluster names are generated by addition of one or more landscape generics. Crucially, the domain of application of the cluster is the river valley. All but one of the names in Table 12 include generics referring to the riverine valley: ‘canyon’, ‘creek’, ‘river mouth’, and ‘headwaters’. The sole exception is Yidateni Dghelaaye’ which contains only the generic ‘mountain’. This name refers to mountains on either side of the headwaters of Yidateni Na’. The generative nature of Dene naming has important functional implications. The most striking feature of the system is its near predictive value. The major creek in the vicinity of Yidateni must almost obligatorily be named Yidateni Na’, and the pass located at the headwaters of Yidateni Na’ is similarly known as Yidateni Tl’aa. Such statements must of course be qualified, for exceptions do exist, and the fact that such Ahtna names “make sense” in terms of the local geography should not be confused with a claim that those same names are predetermined. For example, where two lake exists at the headwaters of the stream it is not possible to know a priori which will be named with the generics ‘headwater lake’. However, where both lakes are named, the typical pattern would be to distinguish them with the directional terms ‘upstream’ and ‘downstream’, as in the Ahtna names Hwdaandi Taltsogh Bene’, literally ‘downstream yellow‐water lake’, and Hwniindi Taltsogh Bene’, literally ‘upstream yellow‐water lake’. The overwhelming tendency toward deterministic naming practices in Dene languages is very real, both to observers and the speakers themselves. As Kari notes, “Ahtna geographic names are so informative and learnable that they facilitate the understanding and recognition of the landscape” (2010a:xv, emphasis added). Ahtna names index the landscape in a reciprocal fashion. On the one hand the names literally describe the landscape, providing knowledge of places with which one is not familiar; on the other hand the landscape imposes the names, providing a physiogeographic structure which allows ready memorization and 14 Draft 2013‐07‐21 usage of names. Knowledge of a small number of specific terms can be readily extended to a large geographic area using the generative naming system. The robustness of this system is further attested by the widespread agreement in linguistically cognate names across language boundaries (Kari 2010b). The contrast with Inuinnait could not be more stark. There is no way to know in advance whether a particular river will be known as ‘big river’ or ‘long river’ or simply as ‘river’. Given this ambiguity it is perhaps not surprising that knowledge of Inuinnait names is not considered a prerequisite to ability to travel or hunt on the land (Collignon 2006:107). Rather, Inuinnait names connect people to the landscape and serve to create a human dimension to this landscape. Of course, the same could be said for Ahtna names. The difference is that where Inuinnait names are deliberately chosen, Ahtna names are largely imparted by the landscape itself; they are inseparable from the landscape. That is not to say that naming is completely unconstrained in Innuinnait: one would presumably be unlikely to name a lake using the Inuinnait generic for ‘mountain’. Nor is naming completely constrained in Ahtna: the choice of specific terms such as yidateni reflects speaker creativity. But these observations are secondary to the basic distinction in the role of landscape in Inuit‐Yupik and Dene place naming. Inuit names are much more likely to be based on human experience (Collignon’s uumajuit), with no reference to landscape. One thus finds Inuinnait names such as Alliakhaqhiurvik ‘place to search for material to make sledges’ and Ihurvik ‘place where hunters wait for game’. For this reason Inuit names are also readily coined. This is true in Alaska among the Yup’ik just as much as with the Inuinnait. Although Yup’ik names are sometimes claimed to be of great antiquity, Fienup‐Riordan cites numerous examples of recently coined names, noting that “some places were named simply to make us smile” (2011:xxix). Thus the Yup’ik place Kass’aq, literally ‘white person’, is so named simply because a white person lived there. Such whimsical names are almost entirely absent in Dene languages. Rather, Dene names are predominantly landscape based, generated in clusters within the domain of the riverine valley. 5. Discussion The comparisons presented here lend some support to the hypothesis that Alaska’s two major language families conceptualize the landscape in very different ways. Though both groups are nomadic hunters sharing a common border across subarctic, their linguistic relationships to this landscape are quite different. The primary contribution of this chapter is to suggest a relationship between demonstrative systems and place naming strategies. Although both Inuit‐Yupik and Dene languages have extremely rich demonstrative systems, the Inuit‐Yupik systems operate primarily at a local scale. At larger scales relevant to landscape the systems have been reduced and altered in language‐specific ways. There is no overarching Inuit‐Yupik landscape demonstrative system. In contrast, the Dene demonstrative system is preserved intact in all of the Alaskan Dene languages, giving special prominence to the linear valley. This valley system can be thought of as a semantic template or “semplate,” that is, a semantic system which is reflected in more than one area of the grammar (Levinson and Burenhult 2009). The linear valley also serves as the organizing principle for generative place naming based on a shared specific term combined with a suite of landscape generics. The existence of the linear valley semplate provides evidence for a deep‐rooted Dene conceptualization of the valley as central to the landscape. This concept is further reinforced by the reciprocal nature of Dene place naming, through which the landscape essentially names itself. 15 Draft 2013‐07‐21 Place naming strategies in Inuit‐Yupik and Dene languages draw on different linguistic resources, rooted in the underlying differences in their demonstrative systems. As a result, Alaska’s two major language families, which seem at first glance to have very similar demonstrative systems, approach the naming of the landscape in very different ways. Whether or not this difference in naming strategies reflects different ways of conceptualizing the landscape, or simply different linguistic designs, remains an outstanding question. Of course, any conclusions drawn here must necessarily be considered tentative, as they rely on disparate sources, many incomplete, from a variety of languages. Inadequate documentation remains a major barrier to the analysis of the landscape domain in Alaska. Research on indigenous toponymy requires exhaustive documentation in order to avoid sampling bias. Yet, most place names studies in Alaska have been opportunistic or guided by etic territorial boundaries. Place name documentation driven by indigenous communities tends to focus on single communities rather than entire language areas, and research driven by government agencies tends to impose artificial boundaries. More popular and widely‐distributed names lists are often redacted, including only a subjective sampling of names for more prominent features. While these materials may be informative regarding the names they do contain, they do not admit a larger synthesis. For example, without comprehensive coverage one cannot extract information about name density or the relative frequency of certain naming strategies. To date comprehensive place names lists have only recently been published for just three Alaskan languages: Ahtna (Kari 2008), Lower Tanana (Kari et al. 2012), and Tlingit (Thornton 2012). Even the best reference dictionaries provide little information regarding the semantics of generic landscape terms. There is still much to learn, and ongoing documentation efforts must also be supplemented by experimental work. NOTES This work was supported by National Science Foundation Alaska EPSCoR award 335863 and grant OPP‐1203194. Thanks to James Kari, Robert Charlie, Dora Andrew‐Ihrke, and Evelyn Yanez for their assistance with field work and to Ken Pratt, Robert Drozda, and Lawrence Kaplan for helpful feedback on an earlier version of this manuscript. They are of course not responsible for any 2 Though dispreferred in Canada, the term Eskimo is still widely used in Alaska to refer to both major branches of the language family; I use the hyphenated term Inuit‐Yupik synonymously with Eskimo. Likewise, the term Athabaskan is increasingly dispreferred in Canada but still widely used in Alaska; I use the terms Athabaskan and Dene synonymously. 3 Gaps in the table reflect forms which have not been elicited, ostensibly due to language attrition. 4 Notably, the riverine system does not reconstruct to the higher level branch of the larger Na‐Dene family. Rather, the riverine system is an innovation within the Dene branch (Leer 1989:602). 5 In practice local river direction will also be conventionalized. Thus, in Tanacross village houses are treated as if they were facing the river flowing from right to left as one looks out the door. This remains the case even though only one house is actually situated in this fashion today. Nonetheless, demonstrative terms are applied unambiguously within the house based on this conventionalization. 6 For the difference in distribution of reflexes of *na and *niq’e see Kari (1996). 7 The list published in 2008 includes 2208 names; a revised and updated list available at the Alaska Native Language Archive includes a total of 2436 names. 1 16 Draft 2013‐07‐21 Kari (2008) estimates the size of the Ahtna territory as 50,000 square miles. No ready estimate of the size of the Inuinnait territory is available, but it can be approximated by the size of Victoria Island, which is approximately 84,000 square miles, on the same order of magnitude as the area of the Ahtna territory. 8 REFERENCES Adelaar, K.A. 1997. An exploration of directional systems in West Indonesia and Madagascar.In: Senft, G., ed. Referring to Space: Studies in Austronesian and Papuan Languages. Oxford: Oxford University Press. 53‐82. Arnold, I., Thoman, R., and Holton, G. 2009. Tanacross Learners' Dictionary: Dihtâad Xt'een Iin Anděg Dínahtlǎa'. Fairbanks: Alaska Native Language Center. Collignon, B. 2006. Knowing Places: The Inuinnait, Landscapes, and the Environment: Canadian Circumpolar Institute. Diessel, H. 1999. Demonstratives: Form, function, and grammaticalization. Amsterdam: John Benjamins. Drozda, R.J. 1994. Qikertamteni Nunat Atrit Nuniwarmiuni: The Names of Places on Our Island Nunivak. Anchorage: U.S. Bureau of Indian Affairs, ANCSA Office. Fienup‐Riordan, A., ed. 2011. Qaluyaarmiuni Nunamtenek Qanemciput: Our Nelson Island Stories. Seattle: University of Washington Press. Fortescue, M.D. 1988. Eskimo orientation systems. Meddelelser om Grønland, Man and Society 11:3‐30. Fortescue, M.D. 2011. Orientation Systems of the North Pacific Rim. (Meddelelser om Grønland 352). Copenhagen: Museum Tuscalanum Press. Fortescue, M.D., Jacobson, S.A., and Kaplan, L. 1994. Comparative Eskimo Dictionary, with Aleut Cognates. (Alaska Native Language Center Research Papers no. 9). Fairbanks: Alaska Native Language Center. Holton, G. 2011. Differing conceptualizations of the same landscape: The Athabaskan and Eskimo language boundary in Alaska.In: Mark, D.M., Turk, A.G., Burenhult, N., and Stea, D., eds. Landscape in Language. (Culture and Language Use: Studies in Anthropological Linguistics). Amsterdam: John Benjamins. 225‐237. Jacobson, S.A. 1984. Semantics and morphology of demonstratives in Central Yup'ik Eskimo. Inuit Studies:185‐192. Jacobson, S.A. 2012. Yup'ik Eskimo Dictionary, 2nd edition. Fairbanks: Alaska Native Language Center. Kari, J. 1980. Kuskokwim River Placenames by Gusty Mikhail. Ms. CY980K1980b. Alaska Native Language Archive, Fairbanks. Kari, J. 1996. A preliminary view of hydronymic districts in Northern Athabaskan prehistory. Names 44(4):253‐271. Kari, J. 2008. Athna Place Names Lists, 2nd edition revised edition. Fairbanks: Alaska Native Language Center. Kari, J. 2010a. Ahtna Travel Narratives: A Demonstration of Shared Geographic Knowledge among Alaska Athabascans. Fairbanks: Alaska Native Language Center. 17 Draft 2013‐07‐21 Kari, J. 2010b. The concept of geolinguistic conservatism in Na‐Dene prehistory.In: Kari, J., and Potter, B.A., eds. The Dene‐Yeniseian Connection. (Anthropological Papers of the University of Alaska, vol. 5). Fairbanks: Department of Anthropology, University of Alaska Fairbanks. Kari, J., Holton, G., Parks, B., and Charlie, R. 2012. Lower Tanana Athabaskan Place Names. Fairbanks: Alaska Native Language Center. Leer, J. 1989. Directional systems in Athapaskan and Na‐Dene.In: Cook, E.‐D., and Rice, K.D., eds. Athapaskan Linguistics: Current Perspectives on a Language Family. Berlin: Mouton de Gruyter. 575‐622. Levinson, S.C. 2003. Space in Language and Cognition. Cambridge: Cambridge University Press. Levinson, S.C. 2008. Landscape, seascape and the ontology of places on Rossel Island, Papua New Guinea. Language Sciences 30(3):256‐290. Levinson, S.C., and Burenhult, N. 2009. Semplates: A new concept in lexical semantics? Language 85(1):153‐174. MacLean, E.A. To appear. Iñupiatun Uqaluit Taniktun Sivunniuġutiŋit; North Slope Iñupiaq to English Dictionary. Fairbanks: Univeristy of Alaska Press. Mark, D.M., Turk, A.G., Burenhult, N., and Stea, D., eds. 2011. Landscape in Language. Amsterdam: John Benjamins. Mark, D.M., Turk, A.G., and Stea, D. 2007. Progress on Yindjibarndi ethnophysiography.In: Winter, S., Duckham, M., Kulik, L., and Kuipers, B., eds. Proceedings of the 8th International Conference on Spatial Information Theory. Melbourne: Springer. 1‐19. Thornton, T., ed. 2012. Haa Léelk'w Hás Aaní Saax'ú: Our Grandparents' Names on the Land. Seattle: University of Washington Press. 18