You are on page 1of 27

Semantics of the Inuktitut (Eskimo) Spatial Deictics

Author(s): J. Peter Denny


Source: International Journal of American Linguistics, Vol. 48, No. 4 (Oct., 1982), pp. 359-384
Published by: University of Chicago Press
Stable URL: http://www.jstor.org/stable/1264840
Accessed: 24-11-2015 15:51 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to International Journal of
American Linguistics.

http://www.jstor.org

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF THE INUKTITUT (ESKIMO) SPATIAL DEICTICS'

J. PETER DENNY

UNIVERSITY OF WESTERN ONTARIO

0. Introduction 6. Objects in-field and out-of-field


1. Places classified for extent 7. A summary of the system
2. Divisions of 'there' 8. Complex forms
3. Other deictic fields 9. Pragmatics
4. Word classes and case suffixes 10. Related dialects
5. Extent classification and the 11. General implications for loca-
word classes tive semantics

0. One thing that most linguists know about the Eskimo languages is
that they have an enormous number of spatial deictic words (Ruhlen
1975). This article offers a detailed semantic analysis of the spatial
deictic system in Eastern Eskimo as spoken in the Aivilingmiut dialect at
Rankin Inlet, North-west Territory. It supports the analysis given for
neighboring dialects by Gagne (1966; 1968) and extends it by describing
the special semantic properties of the three classes of spatial deictic
words, demonstratives, adverbs, and predicative particles. In particular,
it is shown that: (1) the essential semantic difference between the
demonstratives and the adverbs lies in what is located: an object for the
demonstratives and a place for the adverbs; (2) a parallel difference
obtains between what is classified as restricted or extended by the pairs
of deictic roots: the object for demonstratives and the place for adverbs;
(3) the predicative particles act semantically like either the demon-
stratives or the adverbs under appropriate conditions of discourse, either
locating and classifying an object or locating and classifying a place; and
(4) episodes, that is, states, events, and processes, are only located
secondarily by a local case relation to a primarily located object or
place. A quite different aspect of the spatial deictic words which is
I I am grateful to the many Inuktitut speakers who have helped in this work, especially
Luke Issaluk who collaborated in a grueling survey of all the words in the system. A lot of
this work was done at the finest research station imaginable, the Artic Research and
Training Center, at Rankin Inlet, created by Bob Williamson of the University of
Saskatchewan. Crucial linguistic advice was given by S. T. Mallon, and vital logistic
support by Peter Bait, then principal of Rankin Inlet School. Partial funding was obtained
from the Department of Indian Affairs and the Government of the North-West Territories.

[JAL, vol. 48, no. 4, October 1982, pp. 359-84]


? 1982 by The University of Chicago. All rights reserved.
0020-7071 /82/4804-0001 $01.00

359

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
360 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

emphasized in this article is the syntactic nature of much of their word-


internal structure. The article ends with a section on general issues in
locative semantics, as they are affected by the analysis of the elaborate
Inuktitut deictic system, which introduces several novel views. First of
all, the basic locative relation, 'at', often called "locative" case, is shown
to be absent in many locatives-instead they have an 'away' relation,
usually further differentiated by some orientation to the spatial reference
point. Second, it is shown that places are involved in some locatives but
not others: deictics permit objects to be located directly with respect to
the reference point, but they require that episodes be located indirectly
at a place which is in turn located with respect to the reference point
(objects may also play this intermediary role). Third, the dynamic
locative cases-source, path, and goal-are reinterpreted as operators
that select a part of a movement episode and then specify a locative case
relation for that part of the movement.

1. One of the most interesting features of the Inuktitut system is that


places in deictic space are classified according to their extent. This was
first systematically described by Gagne (1966; 1968). Places may either
be classed as unitary spots which the speaker does not break down more
finely, or they may be classed as areas or stretches of space which are
combinations of several spots. In Gagne's example, a large tabletop
would be an area, that is, a composite of several places at which an
object could be located, but any of the minimal places where the object
might be placed would be a spot. Similarly, a house might be seen by the
speaker as located at a spot, whereas a street of houses would occupy a
stretch of space. Places which are minimal spots are termed "restricted,"
whereas places which are composites of spots, whether stretches or areas
of space, are termed "extended."2 Jacobsen (1977:42) nicely summarizes
the classification by saying that restricted locations are those "whose
entire extent is comprehensible to the eye in a single glance," whereas
extended locations are those "whose entire extent is not comprehensible
in a single glance." To express this classification Inuktitut provides a
pair of roots for most deictic locations, one for a restricted and one for
an extended place. Thus, for 'here' there are the roots uv-3 'restricted,

2 In previous papers (e.g., Denny 1978) I have used the terms "nonextended" and
"extended." Since the Alaskan linguists (Reed et al. 1977 and Jacobson 1977) seem to have
adopted my term "extended," I have adopted their term "restricted" as an improvement
over "nonextended," in hopes of slightly reducing the terminological confusion rife in
Eskimo linguistics.
3 lnuktitut words are given in the official native orthography except t is used for
voiceless / not the official &. Unusual graphs are: j, voiceless palatal fricative; g, voiced
velar fricative; r, voiced uvular fricative; q, voiceless uvular stop; and nng, geminate velar
nasal.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 361

here' and maj- 'extended, here', so that uv-unga 'restricted. here-to = to


right here' could refer to a place right beside the speaker, whereas ma-
unga 'extended here-to = to around here' could refer to the whole room
the speaker was in. Although the extended category normally includes
stretches and areas of space that are well defined (such as a room, a bed,
a beach, a berry-patch, etc.), it is also commonly used for indefinite and
unknown locations: for example, un-ani 'extended down there-at =
somewhere down there' could be used for some unknown location down
at the seashore. It is important to remember that what counts as
restricted and extended depends very heavily on the particular situation
being spoken about. A minimal spot for locating a berry-box will be
much smaller than the minimal spot for locating the berry-picker!

2. The next way in which Inuktitut enriches its spatial deictic cate-
gories is by dividing 'there' into five more specific categories: 'up-there',
'down-there', 'in-there', 'out-there', and 'over-there' (i.e., there in the
horizontal plane). Since separate roots are provided for these five
categories, the larger category 'there' is not expressible. Speakers have to
decide whether the location, which is other than their own (i.e., not
'here'), is in addition 'up', 'down', 'in', 'out', or in the horizontal plane
(glossed approximately by English 'over'). For all of these five locations,
except 'in-there' which has only the one root qav-, pairs of roots are
provided indicating whether the place is extended or restricted: for
example, pik- 'restricted, up-there' and pag- 'extended, up-there' could
be used, respectively, in pik-unga 'to right up there' to refer to the place
defined by a hook on the wall, and in pa-unga 'to around up there' to
refer to the place defined by a shelf along the wall. It is important to
remember that all five 'there' categories are only expressible by a speaker
who is not at the location in question. Thus, to say ki-unga 'to right out
there' (root kig- 'restricted out-there'), the speaker must be inside.

3. A third enlargement of the system comes from the provision for


shifting the deictic reference point. Normally, the reference point of a
spatial deictic system is the speaker's location, so that the system
describes the 'speaker's field' in space. However, various languages
provide for spatial fields with other reference points, such as an
'addressee's field' for space organized around the hearer (Fillmore 1975;
Denny 1978). In Inuktitut, there is one general alternative category for
spatial fields having any reference point other than the speaker's loca-
tion, which I shall call 'other field'. This category includes such things as
the previous speaker's field (i.e., locations 'here' or 'there' with respect to
some location at which someone previously spoke), as well as the
addressee's field (i.e., 'here' or 'there' for you the hearer) and some third
person's field (i.e., 'here' or 'there' for some other person or thing in the

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
362 JOURNALOFAMERICANLINGUISTICS
INTERNATIONAL

situation, which need not have been referred to before). This category of
'other field' is expressed by what is customarily regarded as the only
prefix in Inuktitut, ta-, which occurs before the root. Thus, pik-unga 'to
right up there, for the speaker' becomes tak-pik-unga 'to right up there,
for something else'. All forms in the system occur with and without this
field-shifting prefix. Although I have not investigated the topic, my
impression is that speakers perceive the basic and the prefixed roots as
separate lexical items, so I believe that ta- is joined to the root by a
derivational process producing a new root.
Some examples of 'other field' usage will help to clarify the situation.
Reference to a previous speaker's field is found in ta-ik-unga agjaruk
'take it right over there, where we previously mentioned'-this is the
usage often called anaphoric. A common situation illustrating the use of
the 'other field' prefix to refer to an 'addressee's field' occurs when the
object to be located is more saliently related to the addressee, perhaps in
part because he is interacting with it. Thus, one might say ta-ik-unga
illiguk 'put it right over there, from you'. Finally, any other salient
object in the situation may be made the reference point for the spatial
field (even if never before mentioned), for example, a position above a
table may be described as tak-pik-ani 'other-field-restricted'up-there-at
= right up there, from it'. It is important to note that the meaning of ta-
is simply 'other field', that is, a spatial field calculated from some
reference point other than the speaker's location. The further specifica-
tion of 'addressee's field', etc., is done pragmatically, not semantically,
by factors in the context of the utterance. These could include many
things, such as looking at the addressee while saying taikunga illiguk
'put it right over there, from you' or pointing to the reference object
while saying takpikani'right up there, from it'.4 This shifting of reference
field applies to all deictic locations including 'here': for example, a
position on a table might be tagvani (root uv- 'restricted, here') 'right
here, for it', that is, 'r;ght at the reference point' with the table indicated
as the reference point.
Now that we have recognized the existence of both 'speaker's field'
and 'other field' within the Inuktitut spatial deictic system, we have to
revise our understanding of the basic semantic variable that provides the
contrast between 'here' and 'there'-'here' must be understood as 'at
reference-point' and 'there' as 'away from reference-point', since we must
allow for 'here' to be either the speaker's location if the reference point is

4 Both Gagn6 (1966:unit V, p. B) and Bergsland (1955:64) recognize that ta- 'other field'
includes both other reference points in the present situation and previous reference points.
Thus ta- has an anaphoric use but is not basically an anaphoric element.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 363

the center of a speaker's field, or the location of something else if the


reference point is the center of some other field.

4. Based on the locative roots (with or without the prefix ta-),


Inuktitut forms not only adverbs and demonstratives but also a third
word class of uninflected predicative particles (e.g., pik-ka 'right up
there'). All three of these word classes seem to be formed by syntactic
rules which use morphological operations, a type of process described in
Dowty (1978). The various morphemes-roots, demonstrative suffixes,
and case suffixes-belong to distinct syntactic categories and are
combined via the operation of suffixation to form derived members of
some third syntactic category-either adverb, demonstrative, or predica-
tive. The operation which combines a root with a suffix is morpho-
logical, that is, the roots and suffixes are combined into a single
phonological word with typical morphophonemic adjustments and allo-
morphic irregularities. However, the syntactic character of the combina-
tions is seen in the fact that the meanings remain compositional-the
meaning of a demonstrative is a regular combination of the meaning of
its parts and does not drift away from this over time, as does the
meaning of a derived word. This process is the normal one for word
formation in Inuktitut and most of its products are not lexical items: for
example, tusar-uma-rqau-rquu-nngi-kkaluar-i-vuq 'he, also, actually,
didn't, I think, just now want to hear' (from Spalding 1969) is hardly
likely to be stored in anyone's lexicon. However, given the high
frequency of the deictic forms and their membership in a paradigmatic
system of contrasting forms, it is only reasonable to expect that speakers
will handle many of them as unitary lexical items in ordinary sentence
production. Fortescue (1979) calls this phenomenon in Inuktitut semi-
lexicalization.
1 shall now discuss the semantics of the three word classes. The deictic
adverbs are those that have appeared in all previous sections of this
article, consisting of a root and a local case suffix. There are four local
case suffixes: -ani 'at', -anngat 'from', -uuna 'via', and -unga 'to',
corresponding to the four local cases posited in recent analyses of
English-locative, source, path, and goal (Bennett 1975). The adverbial
deictics serve to locate states, events, and processes, which I shall
collectively call episodes: for example, ik-ani iksivajuq 'he is sitting right
over there' gives the location of the sitting episode. From our knowledge
of sitting we can infer that the participant is also located there. However,
it is important to recognize that adverbial deictics do not specify the
participant's location, they only give the location of the episode (Sond-
heimer 1978). This is shown by the many verbs for which the participant
would not usually share the location of the episode (e.g., he looks over

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
364 INTERNATIONALJOURNAL OF AMERICAN LINGUISTICS

there, he shouts down there, etc.). The adverbial deictics specify a


locative relation between the episode and a place; the place in turn is
deictically located with respect to the reference point. The example ikani
iksivajuq means 'he is sitting at a restricted place which is over there
with respect to the speaker'. The four local case suffixes indicate what
part of the episode is located. Either the whole episode is at the place, as
in the above example, or some part of the episode-the beginning,
middle, or end-is at the place. If the first part of the episode is at a
place, then the whole episode is from it (e.g., pik-anngat tigusiviit? 'did
you pick up something from right up there?'). If the middle part of an
episode is at a place, then the whole episode is via the place (e.g., un-
uuna pisukpuq 'he is walking via around down there'). If the last part of
an episode is at a place, then the whole episode is to it (e.g., av-unga
illiguk 'put it to around over there'). This analysis of the four local cases
is a revised version of that developed by Sondheimer (1978); the
revisions are discussed in 11. Since the latter three relations involve
changes of location within an episode, they occur when the episode is a
movement.
In contrast to the adverbials which locate episodes, the demonstratives
locate objects (e.g., ingna 'the one over there'). Another difference is that
the demonstratives refer to the object being located, whereas reference to
the episode located at a place by an adverbial is given by the verb (e.g.,
ikani iksivajuq 'he is sitting right over there'). A third difference is that
demonstratives locate objects directly with respect to the reference point
(ingna 'the one over there'), whereas the adverbials indirectly locate an
episode at a place which has itself been located with respect to the
reference point (e.g., ikani 'at a place over there'). The demonstratives
are formed with a locative root, a demonstrative suffix (most often
-sum- 'the one') which refers to the object located, and a case suffix (e.g.,
pik-sum-inga 'the one up there [accusative case]'). Since they are
nominals, not adverbs, they take the full set of cases, that is, the four
grammatical cases as well as the four local cases. Also, they are inflected
for dual and plural number by changing the demonstrative suffix (e.g.,
ik-sum-inga 'the one over there', ik-kungn-inga 'the two over there', ik-
kun-inga 'the ones over there').5
The main job of the demonstratives is to locate objects, referred to by
the demonstrative suffix, at the deictic location given by the root. The
four grammatical case suffixes permit expression of grammatical rela-
tions of this object to various episodes and objects as is done for any

5 These examples are in the accusative case. This case and the local cases are formed
fairly regularly as described, with the local case suffixes as given previously for the
adverbial forms. The others have irregularities.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 365

noun (e.g., accusative -inga in pik-sum-inga takujuq 'he sees one up


there'). However, when inflected with the local case suffixes the demon-
stratives take on a second role-the object for which they give a deictic
location becomes a secondary reference point for locating an episode
(e.g., iksivajuq pik-sum-ani, with locative case -ani, means 'he is sitting
at the one up there'). This is similar to the way that any object referred
to by a noun can be a locative reference point (e.g., illu-mi'house-Loc =
at the house').6
To summarize, a demonstrative always expresses a location for an
object and often also expresses a local case relation between the object
and some episode. For example, pik-sum-unga agjaruk 'take it to the
one up there' says that the 'taking' episode will end at 'the one' (the goal
case relation expressed by -unga), and that 'the one' is located 'up-there'
(the locative relation).
The predicative particles simply express a location, which may apply
either to an object or a place, but not to an episode. They are much
more commonly used to express the locations of objects than of places,
as in predicative sentences such as Piita uvva 'here is Peter'. As we shall
see, their use to express the location of places seems to require some
special emphasis in discourse on the place; nonetheless, uvva can be used
somewhat elliptically to express 'at a place here'. They cannot apply
directly to episodes, since episodes can only be located indirectly by
having a local case relation to a located place or a located object. The
predicative particles consist only of a deictic root, which is altered into a
phonological word by a morphological process which doubles the final
consonant and adds -a: for example, pik-k-a 'up there', kan-n-a 'down
there'.
Besides the predicative sentences illustrated above, the predicative
particles, as applied to objects, are used in reply to questions such as
nauk Luki? to which the answer might be ikka 'over there'. In this case,
the object to be located, Luke, has already been specified in the
discourse so that the predicative particle is all that is needed to give its
location. In addition, they can appear when both the location and the
object have already been specified in the discourse, that is, to reiterate the
location: to the question ping-n-a Lukii? 'up *there-one-ABsLuke INTER-
ROGATIVE = is the one up there Luke?' one can reply ii, tak-pik-a 'yes
other field-up - there-PREDICATIVE= yes, up there from you'. This little
example shows a lot of morphological and semantic detail that may be

6 It
may appear that there could be other primary combinations such as 'the one from
up there'-this is, however, elliptical for 'the one moving from up there' and is expressed
by a complex form discussed in a later section pik-annga-ar-tuq 'right up there-from-
move-one = the one coming from right up there'.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
366 JOURNALOF AMERICANLINGUISTICS
INTERNATIONAL

worth reviewing. For the question, the root pik- 'up there' is used, in an
absolutive demonstrative, to form an equational sentence in which the
relation of 'being' between the one up there and Luke is unexpressed-
the yes/no question is formed by lengthening the final vowel of the last
word. The same root pik- is reiterated in the reply, preceded by the
'other field' prefix to indicate that it is a location previously specified in
the question uttered by the present listener, and expressed in a predica-
tive form since only the location is being echoed.
The reiterative function of the predicative particles is often seen in a
single sentence such as pikka pinga arsaq niuvirilauqpanga 'I bought
that ball up there' in which 'up-there' is specified both by the predicative
pikka 'up there' and the demonstrative pingna 'the one up there'. The
pragmatic reasons for such reiteration are presently obscure, as they are
for similar English forms such as this one here, contrasting with this one
and the one here.

5. We have seen that the Inuktitut deictic adverbs specify the location
of a place (and also a local case relation between an episode and the
place), whereas the demonstratives specify the location of an object (and
often also a local case relation of an episode to that object). This basic
difference is coordinated with an important difference in the application
of the extent classification provided by the two members of each pair of
roots. When introducing the extent classification we looked only at
adverbs such as pik-unga 'restricted*up - there-to = to right up there' and
pa-unga 'extended up there-to = to around up there', where the roots
classify the place as either a restricted spot, as in the first example, or an
extended stretch or area in the second. However, the extent classifica-
tion does not apply to a place when the roots are used in demon-
stratives, but instead to the object which is being located. Thus, the
demonstrative pik-sum-anngat classifies an object referred to by the
demonstrative suffix -sum- as restricted, giving the meaning 'from the
restricted object up there', whereas the adverbial pik-anngat classifies a
place as restricted, giving the meaning 'from the restricted place up
there'. Typical examples of objects classed as extended by demon-
stratives are: (1) referring to a curtain rod, pangna takuvara 'I see the
[extended] one up there'; (2) referring to a hockey stick, taungna piujuq
'the [extended] one down there from you is good'; (3) referring to a pole,
kiksumunga taguguk 'hit it with the [extended] one out there'; and
(4) referring to a wall, pangna minguaruk 'paint that [extended] one up
there'.
The classification of objects is somewhat more complex than that of
places. Basically an object is restricted if it is roughly equidimensional,
like a ball or a box (i.e., there is no substantial difference in length,

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 367

width, and thickness). Conversely, an object is extended if it is much


larger in one or two dimensions than in the other(s). For example, a
harpoon or a rope is very much longer than it is wide or thick-such
"one-dimensional" objects are classed as extended. In addition, a hide or
blanket is very much longer and wider than it is thick-such "two-
dimensional" objects are classed as extended. To count as extended, one
or two dimensions must be on the order of five times larger than the
other(s)-although humans and animals like bears and dogs are longer
than they are wide or thick, the differences are not very great and so
they are all classed as restricted. There is a size limit on this basic
classification-small objects such as pencils, needles, scissors, or small
towels are usually classed as restricted. To be classifiable as extended, an
object seems to have to be at least three or four feet long. Typical
extended objects are: sled, pole, water pipe, airplane, boat, long table,
hose, rope, board, ladder, broom, wall, and big square rug.
Besides this basic or intrinsic classification based on the spatial
characteristics of the object itself, there is also a temporary, acquired
classification based on the momentary configuration of the object in
space. For example, three intrinsically restricted dogs would be classed
as extended if they are spread out in space. Similarly, a rope which is
intrinsically extended may be classed as restricted if it is coiled up. One
interesting application of the principle that objects spread out in space
are extended is to one member of a pair. A single mitten or boot which
is 'here' is manna 'the [extended] one here', because it is spatially separate
from its mate.
By far the most important cases of acquired extent are those of
moving objects-anything moving, whether human, animal, or thing, is
classed as extended, presumably because the path of its movement is
extended in space. For example, talking about a hockey player skating
along during a game one could say ung-na piujuq 'extended down there-
one ABShe is good = the [extended in motion] one down there is good'.
The contrast between stationary and moving is seen in pairs such as
ingna iksivajuq 'the [restricted] one there is sitting' and amna pisuktuq
'the [extended in motion] one there is walking'. Again, there is a lower
limit for application of the extended category-the slow movements of a
caribou grazing are not sufficient for it to be classified as extended.
In contrast to adverbs and demonstratives, the classification achieved
by a predicative particle is variable-it depends on whether the locative
predicate expressed is being applied to an object or a place: if to an
object, then the object is classified, if to a place, then the place is
classified. This optionality is perhaps not surprising when we remember
that the predicatives only express a bare locative predicate, without any
pronominal or case element-all other aspects of their meaning must be

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
368 INTERNATIONALJOURNAL OF AMERICAN LINGUISTICS

adverbs:
A _, B

00o 0r 0'
; 0
\O , uvani

C D

___ _ _^ , _ ,_
,, , a
0 0 O % maani
1 0
, , \1

predicatives: uvva majja

FIG. I

determined by context, including the entity which they classify. Since


they most commonly express the locations of objects, they usually
classify the extent of the object, as do the demonstratives. This can be
seen in examples such as asking the location of a ball, arsaq nauk?,
when the predicative used in reply will usually classify the ball as
restricted (e.g., pikka 'up there [of restricted object]'). Similarly, if one
asks where a rope is, ak#unaaq nauk?, the predicative will usually
classify it as extended (e.g., pagga 'up there [of extended object]').
The contrast between the predicatives normally used to classify the
object and the adverbs which classfy the place can be seen in the
following situations. There are three dogs which are either lying together,
so that they are a restricted object, or spread out, so that they are an
extended object. The dogs are located either with respect to a patch of
grass, which is a restricted place, or an area of gravel, which is an
extended place. Four situations result, as shown in figure 1.
If the speaker is using a predicative to reply to a question such as
nauk qimmiit? 'where are the dogs?' the predicative will be determined
by the object, so that uvva 'here [of restricted object]' will be chosen for
situations A and C in which the dogs lie close together as a restricted
object, without regard to whether the place is extended or not. Similarly,
predicative majja 'here [of extended object]' will be chosen for the dogs
lying apart as an extended object in situations B and D. On the other
hand, if the speaker is using an adverb to reply to a question such as
naniippat qimmiit?'where are the dogs located?', the choice is determined
by the place, so that uvani 'right here' is chosen for the patch of grass
whether the dogs are grouped at it as in A or spread out around it as in
B, and similarly maani 'around here' is chosen for the area of gravel in C
and D.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 369

Examples of the more infrequent use of the predicatives to classify a


place are the following. If the query is unaaq nauk? 'where is the
harpoon?', the usual reply would be majja 'here, extended', classifying the
harpoon as extended. However, it would be possible to say uvva 'here,
restricted' to classify the place as restricted, indicating that the harpoon
was 'right here' at a delimited spot. Another example shows the role of
previous discourse in biasing the classificatory effect of the predicative
particle. The first query concerns an extended location expressed by an
adverb: avani iksivavaa? 'is he sitting somewhere over there?', which is
echoed in the first reply, ii, tauvani (root av-) 'yes, somewhere over
there'. The second query concerns the identity of the (restricted) person
under discussion, imna Lukii? 'the [restricted] one over there, [is it]
Luke?'. However, the respondent still has in mind the extended location
and replies with a predicative form confirming the location of the person
but classifying the place as extended: ii, tauva 'yes, somewhere over
there'.
There is a small amount of evidence to suggest that even an adverbial
form can sometimes reverse its classificatory effect so that instead of
indicating whether or not a place is extended, it indicates that a thing is
extended. Gagne (1966:23) has an example involving the question nani
sininniarpisi? 'where are you [pl.] going to sleep?'. In the reply, instead
of the expected demonstrative such as paksumani 'with the [extended in
motion] one up there', the corresponding adverbial paani is given, with
the meaning 'with him [moving] who is up there'. The adverbial appears
to be classifying the one up there as extended in motion just as the
demonstrative would do. The same kind of discourse is possible in
English: Q: where are you sleeping? A: right here, in which the informa-
tion that the reply means 'with the person right here' might be indicated
by some gesture such as touching the person. In Inuktitut, reference to
the person is somewhat stronger since the root pag- classifies the person
as extended due to his motion. One thing which allows for this kind of
flexible functioning is the fact that the linkage between the root pag- and
the case suffix -ani is a syntactic not a derivational one. Since pa-ani is a
syntactic construction meaning literally 'up *there-at', it can be elliptical
for pak-sum-ani 'up there-one-at', and it becomes more understandable
that the root would classify the object in the reduced (adverbial) form as
it does in the full (demonstrative) form. Another factor leading to the
choice of an adverbial form is the context provided by the inquiry nani
sininniarpisi? 'where are you going to sleep?', which asks for the location
of an activity to which the normal reply is an adverbial form. The
respondent, however, does not know where he is going to sleep but only
with whom. The compromise is to reply with an adverb which refers in a
classificatory way to the person with whom he is going to sleep. This
sort of compromise formation is discussed in Lakoff (1977).

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
370 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

manna pisuktuq mauna

'the [eCcxtended in motion] one here is ( ^


walking via [the area] here'

manna pisuktuq uvuuna

'the Cextended in motion] one here is walking


via Ethe spot] here'

FIG. 2

We have seen that the classificatory meanings of the deictic roots


apply to objects in the demonstratives and to places in the adverbs. This
allows separate classification of object and location within a sentence, as
in figure 2. Both sentences in figure 2 have the absolutive case demon-
strative using the root maj- 'extended, here' for the moving person. The
first one has the path case adverb using the same root, to indicate that
the movement is via an extended place. The second one has the path
case adverb using the root uv- 'restricted, here' for movement via a
restricted place.
This section and the previous one have sought to achieve a new
understanding of the semantics of the three word classes found in the
Inuktitut deictic spatial system. One implication is that the grouping of
the three into two classes, as is done in many grammars, will falsify the
situation. The two-class views always group the predicatives with either
the demonstratives (e.g., Dorais 1978:19) or the adverbs (e.g., Bergsland
1955:153, who calls the adverbs and predicatives together "number
indifferent forms," i.e., nondemonstratives). Both associations are tempt-
ing and both are misleading because, as we have seen, a predicative form
may express part of the meaning of either a demonstrative or an adverb.

6. The last semantic variable needed to describe the Inuktitut spatial


deictic system is one which distinguishes locations that are 'in-field',
normally meaning in view, from those that are 'out-of-field'. This
distinction only applies to the locations of objects specified by the
demonstratives and the predicative particles, not to the locations of
episodes specified by the adverbs. Thus, while ikani 'at a [restricted]
place over there' can be anywhere out to infinity, iksumani 'at the
[restricted] one over there' refers to an object which is in-field. Usually
this means it is in view, but it might be just out of view or have only
recently disappeared. There is an additional root, ip-, which forms

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 371

demonstratives for objects out-of-field: im-na 'the one out of field


[ABSOLUTIVE]', ip-sum-unga 'to the one out of field', etc. Although the
'in-field/out-of-field' distinction applies to the predicative particles, so
that they can only be used for an object which is in-field, there is no
out-of-field predicative form. Where a predicative would normally be
used to reply to nauk Luki?'where is Luke?', if the object is out-of-field,
then an adverb must be used instead: for example, ikani 'at a [restricted]
place over there'-the fact that the spot is out-of-field is not expressed,
but might be inferred from the use of an adverb where a predicative
would be usual. A possible reason for this restriction on the use of the
predicative forms lies in the fact that their main role is to directly locate
objects-if an object is out of view, one cannot directly know its
location, but one might infer it from some episode which it is involved
in, such as making a noise. The location of this episode can then be
specified by an adverbial form. It should be emphasized that the 'in-
field' restriction applies to all deictic locations for objects, even 'in-there'
expressed by qav-. One can only say qamna 'the one in there' of objects
that are in view or could be easily brought into view by shifting one's
position. This usually did not present a problem in traditional days
because Inuit interiors were all quite open (Carpenter 1973:25) com-
pared to the closed interiors of the industrial world. One could see into
another room of an igloo, see in the door or under the edge of a skin-
tent, and of course see someone standing in a stone-built fish trap. The
more complex form used if one cannot see into an interior is given in a
later section.

7. I have now introduced all of the semantic factors expressed by the


Inuktitut spatial deictics, and a summary view of the system can be
given.
(1) There are twelve basic deictic roots, each of which yields a derived
root using the prefix ta-. These twenty-four roots have the following
semantic structure, shown in figure 3. (I use Gagne's abbreviations for
the classificatory variable: X for restricted and / / for extended. Note too
that 'here' and 'there' do not appear in figure 3 because they are
combinations of 'speaker's field' with 'at reference-point' and 'away from
reference-point', respectively.) The semantic variables are split between
the locations themselves, shown on the vertical axis in figure 3, and the
reference point and class of the located object or place, shown on the
horizontal axis. It may help us to understand how these variables
function together to determine the speaker's choice of a deictic root if we
follow a hypothetical sequence of steps toward the selection of the right
form for a given situation. It seems clearest to start with the choice of
speaker's field or other field. Depending on this choice of reference

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
372 INTERNATIONALJOURNAL OF AMERICAN LINGUISTICS

POINT & CLASS


REFERENCE

speaker's
Ao other
field field
/ \
X // X //

LOCATION
uv- maj- tagv- tamaj -
at reference-point
ik- av- taik- tauv-
!ori zonta
/super i or pik- pag- takpik- takpag-
n field I K t--Icai 1
,^..
/

i X"
....
"
away from reference-
rX
inferior
i
kan- ug- takan- tauk-

poi nt
^ nterior
bounded /i qav- taqqav-

rboundedi e
\ exte r i or kig- qag- takkig- taqqag-

out-of-field taip-

FIG. 3

point, one might determine the location by deciding whether it is in or


out of that field. If it is in-field, is it at the reference point for the field or
away from the reference point? If it is away from the reference point, it
is located either horizontally, vertically, or across some boundary inter-
vening between it and the reference point. Finally, the object or place
being located may be classified as extended or restricted.
(2) Based on these roots, three classes of words are built. First, there
are adverbs, which are syntactic combinations of root and local case
suffix-they specify the location of a place, classify the place for
extendedness, and express a local case relation of an episode to the
place. The 'in-field'/'out-of-field' distinction does not apply to them.
Second, there are demonstratives consisting of a syntactically combined
root, a demonstrative suffix, and a case suffix-they specify the location
of an object, classify its extent, and may express a local case relation of
an episode to the object. Third, there are predicative particles, consisting
only of a root (and morphology needed to form a phonological word),
which simply express the deictic location of either a place or an object
and give its classification. If occasional homonyms are still counted as
separate words, there appear to be 686 words in the total system.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 373

8. Since the three deictic word classes are formed by syntactic rules,
although using morphological processes, it is not surprising that further
syntactic combinations are possible, still employing morphological pro-
cesses within the phonological word. Some of the most frequently used
are verbs of motion and location. The suffix -aq- 'move' can be
combined with the three local cases which are specialized for movement,
goal, source, and path, but not the stationary locative case. Since -aq- is
a verb, the adverbial forms are used to locate the episode described by
the verb. For example, pa-unga-ar-puq 'he is going to around up there'
uses the goal case adverb paunga to locate the end portion of the
movement. In pa-una-ar-puq 'he is going via around up there' the path
case adverb pauna is used to locate the middle portion of the episode. In
pik-annga-a-qatau-juq 'he is coming, with others, from right up there' the
source case adverb pikanngat is used to locate the beginning of the
movement expressed by -aq- (which is followed by the suffix -qatau-
'with another, others').
In contrast to -aq- 'move', used with the dynamic local cases, -it- 'to
exist' is used only with the static local case, the locative: for example,
pik-ani-it-tuq 'he is right up there' using the locative case adverb pikani.
As this example illustrates, 'to be at a place' is one particular kind of
condition of existence-other adverbs can express other such conditions,
for example, taima-it-tuq 'it is thus'. This verb -it- seems to be the basic
verb of existence since it is used as an independent verb root, at least in
Greenlandic Eskimo, for example, ipput 'there are' in kuyatiwtinni
urpikSuwit napaarturSuwit ipput 'to-the-South-of-us trees straight there-
are' (Bergsland 1955:81). One special value of -it- for the deictic locative
system is that, in its noun participle form, it allows reference to an
object at a location under two conditions not permitted by the demon-
stratives. First, the location may be classified, rather than the object, for
example, pa-ani-it-tu-mik qiniqpuq 'extended up there-at-be located-
one-ACCUsATIvE he-seeks = he is looking for the one who is somewhere
up there'. In this example, classifying the place as extended is an
essential part of the message about seeking the person. Second, the
locations of objects out-of-field can be specified, rather than using the
nonspecific out-of-field root ip-, for example, taqqav-ani-it-tu-mik
isumajuq 'he is thinking about the one who is in there [where we
mentioned]', said when the latter person is out of view. Since -it- occurs
following the locative case, it can occur with locative case demon-
stratives as well, for example, paksumani-it-tuq 'it is at the one [extended
in motion] up there' referring to the location of luggage left in a plane
which has just taken off with it.
There is another verb of being which can be used with all the local
case adverbs, as well as the predicative forms; this is -u-, which has the

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
374 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

meaning 'to be something, be characterized by something'. This verb is


normally used with nouns and pronouns (e.g., aputi-u-vuq 'it is snow, it
is snowy'). When used with predicative demonstratives (e.g., tagva-u-juq
'it is here for you'), it expresses the meaning that the location indicated
by the predicative characterizes the object referred to. In these forms,
the predicative classifies a place (e.g., pagga-u-juq 'it is somewhere up
there), therefore, we know that it is a place which characterizes the
object described. In this example, the word means that an object is
characterized by an extended place up there, just as an object is
characterized by snow in aputi-u-vuq 'it is snowy'.
When -u- combines with the adverbial forms very different structures
result, as shown in these examples:
LOCATIVE: ik-ani-u-juq 'he lives over there'
SOURCE: pa-annga-u-jut 'they come from around up there'
PATH: av-uuna-u-juq 'he is going via around over there'
GOAL: pik-unga-u-juq 'he is on his way to right up there'

Notice that none of these forms, including especially the locative case
one, describes the location of the participant at the time referred to, as
was done by the two previous kinds of forms, LOCATIVE ADVERB+ -it-:
pik-ani-it-tuq 'he is +
right up there', and PREDICATIVE -U-: tagva-u-juq 'it
is here, for you'. Instead, each form conveys a particular kind of activity.
For the locative case adverb it is 'dwelling' (including permanent loca-
tion for inanimate things), for the source case adverb it is 'originating',
and for the path and goal case adverbs it is 'traveling'. Only one of these
can be mirrored by a parallel English construction, the source case one,
he is from there, which involves origin in the same way as in the
Inuktitut. Various analyses are possible, but at the moment I am
inclined to believe that each case suffix and -u- have combined deriva-
tionally to yield a new verb: -aniu- 'dwell, be permanently located at',
-anngau- 'originate', -ungau- 'go to', -uunau- 'go by way of'. Apart from
the idiosyncratic meanings of the combinations, a derivational rather
than a syntactic linkage seems likely because the parallel combinations
of case suffix and -u- which combine with noun roots rather than
demonstrative roots seem to be treated by speakers as lexical items;
Smith's (1978) Inuit consultants separated out items such as -muu- 'to go
to', which is derived from -mut 'to' and -u- and appears in nuna-muu-
vuq 'he goes to the land'. The verbs of traveling to a deictic location,
-ungau- 'go to' and -uunau- 'go by way of', have somewhat broader
meaning than the verbs of movement formed with -aq- 'move', such as
pik-unga-ar-puq 'he is going to right up there'. The former, such as pik-
unga-u-juq 'he is on his way to right up there', refer to a journey which
may not at all times involve movement in the specified direction.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 375

What we have seen in this section is that Inuktitut has two verbs of
being, -it- 'to exist' and -u- 'be something, be characterized by some-
thing'. Both can express location: -it- 'to exist' when combined with an
adverbial deictic, indicating the location of the existing, and -u- 'be or be
characterized by' when combined with a predicative deictic, characteriz-
ing the object by a place. In addition, -u- can combine with the local
case suffixes of the deictic adverbs, apparently forming derived verbs of
idiosyncratic meaning.7

9. Although the topic of this article is the semantics of the Inuktitut


spatial deictics, there is a special need to say something about their
pragmatics. The pragmatics of these terms are more salient to speakers
and often to students of the language and may obscure from view the
semantics. It may help readers who work with Inuktitut to penetrate to
the less-consciously experienced semantics if the more-conscious prag-
matics are described. The pragmatics may be thought of as additional
information, beyond that contained in the semantic meaning of the
term, which is conveyed when the term is used in a particular situation.
Consider first a frequently occurring pragmatic amplification of the
classificatory variable, using ik- 'restricted, over-there' and av- 'extended,
over-there'. Suppose the first speaker asks nauk pilaut? 'where is the
knife?' and receives the reply ikka 'right over there' with the predicative
form classifying the place as restricted. The first speaker then moves
away from the second, looking for the knife, but stops at a location
which is not far enough away. The second person can then alter his reply
to suggest a place farther away, by increasing the size of the place
referred to-this can be done by switching to the extended form, avva
'around over there'. The meaning conveyed is simply that the place,
within which the knife is located, is bigger; however, the pragmatic
information, in this context of moving away while looking, is to look
even farther away. This pragmatic connotation of distance attached to
the extended class deictics is so frequently elicited by the context, that it
is sometimes mistaken for their meaning. Both speakers and analysts are
encouraged to do so because the pragmatics are more consciously
experienced than the semantics. Schneider (1967) and Ittuq (reported in
Dorais 1971) thought that 'near' and 'far' were the meaning of certain
restricted/extended pairs such as pik- 'restricted, up-there' and pag-
'extended, up-there'. However, Gagne (1968) and this investigation have

7There is a third verb of being (Mey 1968): -qaq- 'to have' is used to express existential
quantification (e.g., aputi-qaq-tuq 'there is snow'). Although this does not seem to
combine with the deictics by morphological processes, the deictic adverbs can of course
modify it by syntactic processes: ikani mirquti-qar-paa?'is there a needle over there?'.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
376 INTERNATIONALJOURNAL OF AMERICAN LINGUISTICS

shown that the semantic distinction is 'restricted' versus 'extended',


because this contrast still obtains at any particular distance: for two
equally near places, or two equally far places, the restricted one such as
a spot will be referred to by pikka 'up there, of a restricted place', and
the extended one such as an area will be referred to by pagga 'up there,
of an extended place'.
A rich variety of pragmatic messages can be conveyed by the spatial
deictics. Another frequently occurring message concerns perceptibility.
It is common to use a predicative form if the object can be seen at its
location, but to switch to an adverbial form if the object is not visible.
This is a somewhat more complex case because although neither kind of
form has visibility as a part of its meaning, the predicatives do have
visibility as a semantic implicature-whatever predicative is used will
imply that the object is visible because there is no predicative form for
out-of-field (in this case invisible objects), that is, no predicative form
using root ip- 'out-of-field'. If the discourse is such as to lead to a
predicative form, typically by beginning with the question word nauk?
'where is?' which seems to elicit predicatives, then giving an adverbial
instead will set up a contrast in this particular context which conveys
pragmatically that the object in question is out of view. For example,
nauk arsaq? 'where is the ball?' could have the predicative reply pikka
'up there' if it were visible. If instead the adverbial pikani 'up there' is
used, it will pragmatically convey that the ball is out-of-sight, perhaps
hidden in a box. The frequency with which visibility is conveyed
pragmatically by deictic forms has tempted some investigators to regard
it as part of their meaning; Dorais (1971) suggests that degree of
perceptibility is the semantic factor separating the classificatory pairs of
roots such as pik- and pag-. This study argues against this because two
objects which are equally perceptible or equally imperceptible will be
classed according to their extendedness: pingna 'the restricted one up
there' will be used for a stationary person and pangna 'the extended one
up there' for a moving person when both are clearly perceptible or when
both are equally hard to perceive. If the speaker judges the person to be
actually imperceptible, he will not use either the pik- or pag- root, but
will switch to the root for imperceptible objects, ip- 'out-of-field' (e.g.,
imna 'the one out-of-field'). Thus, perceptibility does enter into the
semantics of the spatial deictics, but it only helps to discriminate ip-
'out-of-field' from the other eleven roots which all have the feature
'in-field' or 'perceptible'. There are no semantic differences involving per-
ceptibility among these latter roots.
A third kind of pragmatic information seems quite salient for speakers
of Inuktitut. This is the use of extended class forms for the location of
sets of things, for example, av-ani-it-tu-t qimmi-it 'extended'over there-
at-located-MODE-PLURAL dog-PLURAL = the dogs are around over there',

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 377

using root av- 'extended, over-there' to classify the place. Nonetheless,


speakers also recognize that sets may be concentrated at a restricted
location, so that ikaniittut qimmiit 'the dogs are right over there' is also
said. On first reflection, however, plurality is sometimes taken to be part
of the meaning of the extended forms.

10. What I have described applies to the Aivilingmiut dialect of the


northwestern shore of Hudson's Bay. To broaden the study brief
comments are offered based on semantic studies of two other Canadian
dialects of Eastern Eskimo.
Baffin Island. Gagne (1966) is thought to be based on Baffin Island
data. The main dialectal difference is simply that phonological assimila-
tion produces some homonyms so that both student and researcher have
to be alert to context. Although all eleven predicatives are the same as
reported herein, in the absolutive case demonstratives Aivilingmiut
ingna 'the [restricted] one over there' and imna 'the one out-of-field'
have both become Baffin Island inna. Similarly, Aiv. qangna 'the
[extended] one out there' and Aiv. qamna 'the one in there' have both
become qanna.
Quebec. The above phonological changes are also found in the closely
related Quebec dialects. However, they seem to have led to the dis-
appearance of qanna (Aiv. qangna) 'the [extended] one out there' so that
there is no extendedness classification for locations out-there and kinna
(Aiv. kingna) simply means 'the one out-there'. The other homonyms
inna 'the [restricted] one there' and inna 'the one out-of-field' apparently
still exist since both meanings are given in Schneider's (1970) dictionary;
however, in his grammar (Schneider 1967) he seems to regard them as
one form, yielding a count of ten demonstratives, rather than the actual
eleven. Dorais (1977) leaves inna 'the one out-of-field' out of his
grammar thereby also getting a count of ten.

11. The main purpose of this article has been to portray for Amer-
indianists the semantics of the very large spatial deictic system of
Aivilingmiut Eskimo. In the course of doing so a number of general
problems in locative semantics were encountered, and proposed solu-
tions developed, so that the description of the system might be com-
pleted. In this section I discuss these problems and solutions in the
context of a general theory of locative semantics to show how the
analysis of the Inuktitut system can contribute to our understanding. To
express my views more precisely I also sketch them in a many-sorted
predicate logic.
In general, spatial locatives express a locative relation between an
entity to be located and a reference point, which we may symbolize as
L(x,r) 'an entity is located with respect to a reference point'. For deictic

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
378 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

locatives, the reference point is not fully specified in the utterance, but is
given by knowledge of the speech situation. For the Inuktitut prefixless
forms the hearer must be able to identify the speaker as the reference
point. For the prefixed forms the hearer must be able to identify either
some other object in the present situation, or an object previously
referred to, as the reference point. In the logical symbolization the
speaker as reference point will be indicated by subscript s: L(x,rs) 'an
entity is located with respect to the speaker', and any other reference
point by o: L(x,ro) 'an entity is located with respect to something other
than the speaker'.
The first major theoretical question I discuss is the nature of the
locative relation. In recent treatments of locative semantics, primarily
concerned with nondeictics, it has been claimed that the only locative
relation is the locative case (so-called by Bennett 1975, called 'AT'by
Lyons 1977, and called 'Place' case by Sondheimer 1978). However,
study of Inuktitut and other deictic systems suggests that there are two
basic locative relations, 'at' and its opposite which I shall call 'away'.8
These are all that is expressed in minimal deictic systems such as the
English one: 'an entity is here' At(x,rs), 'an entity is there' Away(x,rs).
Moreover, the 'away' relation is often further differentiated by recogniz-
ing that a located object which is 'away' from the reference point may
bear various orientations to that reference point, for example, 'an entity
is ahead of a reference point' means it is both 'away' from the reference
point and 'anterior' to it, Away(x,r) and Anterior(x,r).
In the Inuktitut deictics, the five divisions of 'there' are all 'away'
locations bearing differing orientations to the reference point. These
orientations are (1) vertical: superior, pik-/pag- 'up-there', inferior, kan-
lug- 'down-there'; (2) horizontal, ik-/av- 'over-there'; or (3) across a
boundary: interior, qav- 'in-there', and exterior, kig-/qag- 'out-there'.
These combinations of 'away' and some orientation can be symbolized
as single units, for example, Up(x,rs) 'an entity is up with respect to the
speaker' indicates the combination of 'away' and 'superior' orientation.
Also, it is worthwhile for the symbolization to recognize that the
locative relation and the reference point are always expressed together in
the root of the Inuktitut deictic word, so I adopt the standard option of
having the second argument of the relation appear as a part of a one-
place predicate, for example, Ups(x) 'an entity is up with respect to the
speaker'. To symbolize the meaning of the roots in isolation, the
argument variable is shown as a superscript indicating the kind of
argument the predicate takes: for example, pik-/pag- Ups 'up with

8 The relation
'away' is not realized in English simply by away, since Bennett (1975)
shows that this word involves a goal case. The closest English might be be away from.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 379

respect to the speaker, for some entity', takpik-/takpag- Up' 'up with
respect to some object other than the speaker, for some entity'.
A second important question in locative semantics is what sort of
entities can enter into a locative relation. Lyons (1977:694) has suggested
that the locative relation (called AT)obtains between an object (X) and a
place (Y): AT(X,Y). Quite differently, Sondheimer (1978:243-44) has
suggested a two-part structure in which, first, the locative relation
(called PLACEcase, but synonymous with AT)holds between an episode
(event or state) and a place: PLACE(episode,place), and, second, the
place has a 'part of' relation to the reference object: Part-of(place,
reference-object). For example, running in the park would be analyzed
as PLACE(running, place) A IN(place, the park). The analyses of Inuktitut
deictics only partially confirm these views. Objects can be located, as
Lyons suggests, but never in relation to places; episodes are located at
places as Sondheimer suggests, but, in addition, the places are located at
the reference object, requiring recognition that a locative relation is a
component of his 'part-of' relation. However, episodes can also be
located at objects which have in turn been located at the reference
object. Let us turn now to a systematic review of the sorts of entities that
are located by deictic locative systems.
In general, locative semantic relations allow three sorts of entities to
be located-objects, episodes, and places. For example, John is playing
on the table locates the episode of playing on the table, but does not
specify whether John shares the location. In the deictic systems dis-
cussed in this article, places are located only in order to serve as
secondary reference points for locating something else. Thus here, which
can be paraphrased at this place, means 'at the place which is at the
speaker'. The primary reference point, 'the speaker', is used to character-
ize the secondary one, 'a place which is at the speaker'. In logic we have
to use a special operator to indicate that we are going from a proposi-
tion 'a place is at the speaker' Ats (p), to a secondary object 'a place
which is at the speaker'. The operator we need is the epsilon-operator
for indefinite descriptions, e (Bartsch 1972), because the descriptive
phrase 'which is at the speaker' only identifies some indefinite place
having the description not a definite place: (ep)At,(p) 'some place which
is at the speaker'. In discourse a particular place will be identified by
other factors, and this can be indicated by an index, i, on (ep) 'some
place': (ep)YAts(p)'a particular place which is at the speaker'.
Different kinds of locative systems have different restrictions on what
sort of entities may be located. Sondheimer (1978) argues that for
nondeictics such as in the park only episodes (called events by him) can
be located. This is because any such location is limited temporally, so
that to have a location (as in John is in the park) is to participate in a

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
380 INTERNATIONALJOURNAL OF AMERICAN LINGUISTICS

temporally limited episode. Thus, it is an episode which he calls BEING-


AT which has the location, not the object, John. Another point is that,
for some episodes, it would be unusual for the participant to have the
same location as the episode: John is shouting across the table, John is
looking into the hole. All of this suggests that the nondeictic locatives
only locate the episode. However, deictic systems seem to have different
restrictions. First of all, they allow objects to be located: for example,
English demonstrative this and Inuktitut demonstrative una both mean
(co)'Ats(o) 'an object at the speaker'. The reason this is possible is that
the time limitation of the location is provided by the deictic index and
therefore need not be given in the semantics of the lexical items.
Semantically this means 'an object at the speaker' independently of
temporal changes in location, but when the hearer consults his knowl-
edge of who is speaking he will, of course, find that the particular
identity of the speaker is time-limited and therefore the location is also
time-limited.
A second restriction found in deictic systems is that episodes cannot
be directly located but are only located secondarily at objects and places
which have themselves been deicticly located. The Inuktitut evidence for
this restriction is very clear, but I believe it holds also for English. Since
here can be paraphrased at this place I think its meaning is probably
Ate( ..., [(cp)'Ats(p)]) 'at a place which is at the speaker, for some episode'
(the ellipsis indicates the other unspecified argument of the relation, and
the superscript e on At indicates that this argument can only be an
episode). If, as just pointed out, this means 'an object at the speaker',
then both it fell from this and it fell from here involve secondary
locating of the episode: it fell from this means 'it participated in an
episode of falling, the beginning of which was at an object which is at
the speaker' and it fell from here means 'it participated in an episode of
falling, the beginning of which was at a place which is at the speaker'. In
the first case, the falling is at an object which is at the speaker, and in
the second case the falling is at a place which is at the speaker, but in
neither case is the falling at the speaker. I am less sure about the reason
for this second restriction on the semantic structure of deictics, but it
may be related to another unusual property of spatial deictics, namely,
that they seem to involve two indices. The first is the familiar reference
point I have been discussing (which is represented in the logical
symbolizations), either the speaker or some other object salient in
discourse. The second (not symbolized herein) is pointing-by pointing
it is possible to limit the secondary reference point beyond its specifica-
tion by the primary reference point. For example, one can point at one
spot on a set of shelves and say it fell from here, and then change one's
mind and, pointing to another spot, say no, it fell from here. In both
sentences here refers to the place at the speaker, but the pointing picks

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 381

out two different subplaces within that place. In order to be able to limit
reference points by pointing it seems likely that (1) there have to be
secondary reference points which can be so limited, since no amount of
pointing at a primary reference point such as the speaker can further
define it, and (2) the secondary reference points have to be concrete
entities which can be accurately selected by pointing. This seems to
restrict them to objects and places, eliminating episodes since they are
often too amorphous to be pointed at-one cannot point at the shouting
which fills the space between a mother and the child she is shouting at.
As a consequence, deictics permit the primary localization of objects and
places, but only secondary localization of episodes.
How, then, are these patterns realized in Inuktitut? A point to
remember is that the two sorts which are directly located, objects and
places, are further subcategorized as extended (subscript e) or restricted
(subscript r). The first pattern, the primary localization of objects, is
achieved by the demonstratives in any of the four grammatical cases
(not the local cases): pingna 'the [restricted] one up there', symbolized
(eor)'Ups(or)'a restricted object which is up with respect to the speaker'.
The second pattern, the secondary locating of episodes, has two variants
depending on whether the secondary reference point is an object or a
place. Locating an episode at an object which is already located is done
by the demonstratives in any of the four local cases: piksumani 'at the
[restricted] one up there', Ate(...o[(Eor)iUps(Or)])'at a restricted object
which is up with respect to the speaker, for some episode'. The other
variant of the second structure, the secondary locating of episodes at an
already located place (rather than an already located object), is ex-
pressed by the adverbials: pikani 'at the [restricted] place up there',
Ate(...,[(E?pr)Ups(p)])'at a restricted place which is up with respect to the
speaker, for some episode'. One thing to note about secondary localiza-
tion is that only the 'at', not the 'away', relation holds between the
episode and the place or object serving as secondary reference point.
This is true of Inuktitut but may not be for all languages. Finally, two
different subparts of these patterns may be expressed by the predica-
tives. The first is the primary localization of an object, including
specification of its sort, restricted or extended, but excluding reference
to the object: pikka 'up there [of restricted object]', Upr 'up with respect
to the speaker, for some restricted object'. The second is primary
localization of a place, giving its sort without reference to it: pikka 'up
there [of restricted place]', UpPr 'up with respect to the speaker, for some
restricted place'. As shown previously, only discourse chooses between
these two patterns of meaning for a predicative form.
A third problem in locative semantics concerns the status of the
dynamic locative cases, source, path, and goal. Prior to Bennett (1975)
these were often thought to be primary cases in their own right;

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
382 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

however, he showed that each of them embeds a locative case, raising


the question about what extra information they convey. To answer this
it helps to recognize that they are only used for the location of
movement episodes. The difficulty in locating movement episodes lies in
the fact that at different times the movement is in different places. I
believe that the general solution adopted by semantic systems is to break
the episode into its temporal parts, beginning, middle, and end. These
temporal parts may then be located reasonably precisely. The temporal
parts of a movement episode are expressed in the Inuktitut deictics by
the three dynamic local case suffixes, -anngat 'from' AtFirst'part(e) 'at a
reference point, for the first part of some episode', -uuna 'via' AtMid-part(e)
'at a reference point, for the middle part of some episode', and -unga 'to'
'at a reference point, for the last part of some episode'. Thus,
AtLast-part(e)
the location of a movement involves locating a temporal portion of it:
for example, ikuuna pisukpuq 'he is walking via the [restricted] place
over there' (3e)( o)At([Mid-part(e)],[(epr)'Ths(p)]) A W(e) A Abs(o,e)
'there is an episode, and an object such that: the middle part of the
episode is at a restricted place which is over-there with respect to the
speaker, and the episode is walking, and the object is in an absolutive
case relation to the episode'. This treatment of the three dynamic local
cases was suggested by a similar approach used by Sondheimer (1978).
In following Sondheimer I have made one major change-I have the
local cases pick out a part of the episode whereas he has them pick out a
part of the place occupied by the whole movement episode. His view
seems a bit more complex because a place for the whole movement must
be recognized, whereas in my treatment only the located place at which
some part of the episode occurs need be recognized. Does this view of
the three dynamic cases mean that the 'locative' is the only local case?
Not within the theory outlined in this section, since, besides the 'locative'
or 'at' relation, an opposite relation, 'away', has been recognized-only
here involves 'at'; there involves 'away'. Furthermore, as we have seen,
locations 'away' from a reference point may be further specified by
orientations such as 'superior' in up there. Instead of one locative case,
we seem to have a set of locative relations.
To summarize, starting from a general notion of a locative relation,
L(x,r) 'an entity is located with respect to a reference point', we have
derived a number of more specific structures which are expressed by the
Inuktitut deictic locatives. The locative relation is a set of relations
consisting of 'at' (expressed by uv-/maj-/tagv-/tamaj-), a set of 'away'
relations, and 'out-of-field'. The 'away' relations are the five kinds of
'there', each combining 'away' from the reference point with an orienta-
tion to it, for example, 'up-there' (expressed by pik-/pag-/takpik-/
takpag-). The entities to be located include episodes, objects, and places.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
SEMANTICS OF INUKTITUT SPATIAL DEICTICS 383

However, only objects and places are directly located with respect to the
reference point-episodes are only secondarily located with respect to an
already located object or place. Objects may be located in their own
right, for example, pingna (Or)Ups(Or) 'a restricted object which is up
with respect to the speaker', but places are only located so as to serve as
secondary reference points for locating episodes. Movement episodes
have temporal parts which are indicated by the three dynamic local case
suffixes. The primary reference points are either the speaker or some
other object given in the discourse.

REFERENCES

BARTSCH, RENATE.1972. The Grammar of Adverbials. Amsterdam: North-Holland.


BENNETT, DAVIDC. 1975. Spatial and Temporal Uses of English Prepositions. London:
Longman.
BERGSLAND, KNUT.1955. A grammatical outline of the Eskimo language. Mimeographed.
[Microfiche available from Inter Documentation, Zug, Switzerland.]
CARPENTER, EDMUND.1973. Eskimo Realities. New York: Holt.
DENNY,J. PETER.1978. Locating the universals in lexical systems for spatial deixis. Papers
from the Parasession on the Lexicon, pp. 71-84. Chicago: Chicago Linguistic Society.
DORAIS,LOUIS-JACQUES. 1971. Some notes on the semantics of Eastern Eskimo localizers.
Anthropological Linguistics 13:91-95.
1977. Inuit Kupaimmiut Uqausingit: The Language of the Northern Quebec Inuit.
Quebec: Laboratoire d'Anthropologie, Universit6 Laval.
.1978. Iglulingmiut Uquasingit: The Inuit Language of Igloolik, N.W.T. Quebec:
Laboratoire d'Anthropologie, Universite Laval.
DOWTY,DAVID. 1978. Applying Montague's views on linguistic metatheory to the structure
of the lexicon. Papers from the Parasession on the Lexicon, pp. 97-137. Chicago:
Chicago Linguistic Society.
FILLMORE, CHARLES. 1975. Santa Cruz Lectures on Deixis. Bloomington: Indiana Univer-
sity Linguistic Club.
FORTESCUE,MICHAEL. 1979. The relationship of derivation to external syntax in West
Greenlandic. Proceedings of Fifth Scandanavian Conference of Linguistics, pp. 23-38.
Frostvallan.
GAGNt, RAYMOND. 1966. Eskimo language course. Mimeographed. Ottawa: Department
of Indian Affairs and Northern Development.
.1968. Spatial concepts in the Eskimo language. Eskimo of the Canadian Arctic,
ed. Victor F. Valentine and Frank G. Vallee, pp. 30-38. Toronto: McClelland &
Stewart.
JACOBSON,STEVEN.1977. A Grammatical Sketch of Siberian Yupik Eskimo. Fairbanks:
Alaska Native Language Center.
LAKOFF, GEORGE. 1977. Linguistic gestalts. Papers from the Thirteenth Regional Meeting,
pp. 236-87. Chicago: Chicago Linguistic Society.
LYONS,JOHN. 1977. Semantics. Cambridge: Cambridge University Press.
MEY, JACOB. 1968. On the notion 'to be' in Eskimo. The Verb 'Be' and Its Synonyms, pt. 2,
ed. J. W. M. Verhaar, pp. 1-34. Dordrecht: Reidel.
REED,IRENE,ETAL. 1977. Yup'ik Eskimo Grammar. Fairbanks: Alaska Native Language
Center.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions
384 INTERNATIONALJOURNAL OF AMERICAN LINGUISTICS

RUHLEN,MERRITT.1975. A Guide to the Languages of the World. Stanford, Calif.:


Stanford University Language Universals Project.
SCHNEIDER,LUCIEN. 1967. Grammaire Esquimaude du Sous-Dialecte de L'Ungava.
Quebec: Direction Generale du Nouveau-Quebec.
.1970. Dictionnaire Esquimau-Francais du Parler de L'Ungava. Quebec: Presses de
L'Universite Laval.
SMITH,LAWRENCE R. 1978. A Survey of the Derivational Postbases of Labrador Inuttut
(Eskimo). Ottawa: National Museums of Canada.
SONDHEIMER,NORMANK. 1978. A semantic analysis of reference to spatial properties.
Linguistics and Philosophy 2:235-80.
SPALDING,ALEXE. 1969. Salliq: An Eskimo Grammar. Ottawa: Department of Indian
Affairs and Northern Development.

This content downloaded from 160.36.178.25 on Tue, 24 Nov 2015 15:51:22 UTC
All use subject to JSTOR Terms and Conditions

You might also like