363

MICMAC SEMANTICS:
MEDIALS FOR N O U N CLASSES
J. Peter Denny
University of Western Ontario

Micmac is a notoriously deviant and complex member of the Algo-

nquian language family spoken in the Atlantic Provinces of Canada.
So far as I know, no anthropological semantic studies of it have been
published. I have chosen to begin with noun classes because they are
familiar to m e in other languages, but the morphological difficulties and
semantic innovations I encountered make the current report only a ten-
tative examination of certain core areas of the noun class system. Noun
classes are expressed in Micmac, as in other Algonquian languages, by
classificatory medials incorporated into verbs: e.g., /maq-apsk-e-k/1
'it (three-dimensional) is big' is a verb /maq-e:-k/ 'it is big' incorpo-
rating the classificatory medial /-apsk-/ which classifies the subject of
the verb as a three-dimensional object, i.e., an object of roughly equal
size in all three spatial dimensions such as a child's block or a ball.2
Dimensionality classes
Micmac seems to have four medials expressing the dimensionality
of a concrete object: /-apsk-/ 'three-dimensional (3D)', /-ak-/ 'two-
dimensional (2D)', /-taq-/ 'one-dimensional (ID)flexible',and /-oq-/
'one-dimensional (ID) rigid'. This set of four dimensional classes in
Micmac is the same semantic sub-system found in other Algonquian
languages such as Cree and Ojibwa; objects are classed as either roughly
equal-sized in all three dimensions (3D) such as a square rubber eraser
or an apple, or much larger in two dimensions than the third (2D) such
as hide or bark, or much larger in one dimension than the other two
(ID) such as a stick or string. The ID objects are further classed as
eitherflexiblelike a string or rigid like a stick. Of the four Micmac me-
dials, the last three have the same meanings as Proto-Algonquian (PA)
*/-ye:k-/ '2D', */-a:pje±-/ 'IDflexible',and */-a:xkw-/ 'ID rigid', but

A hyphen (-) indicates a morpheme boundary within a stem; an equals sign ( •)

indicates a stem-inflection boundary.
2
Algonquian languages have grammatical animate gender, so that throughout this
paper some objects that are thought of in English as having inanimate gender will
be described by verbs inflected for animate gender.
 364 Peter Denny

only the second, /-ak-/ '2D', and fourth, /-oq-/ 'ID rigid' seem to be
reflexes of Proto-Algonquian forms. Thefirstmight correspond to P A
*/-a:peOk-/ 'mineral solid'; in Ojibwa and Cree the meaning '3D' is
conveyed by /-minak-/.
The Micmac forms appear as medials in verbs with number roots
and other roots describing the classified object: /nan-apsk-e-kl/ (inan-
imate intransitive (II) verb) 'five 3D S T A T I V E 3rd plur - they (3D) are
five'.e.g., sugar cubes; /piptoqo-psk-asi-t/ (animate intransitive (A
verb) 'round 3 D S T A T I V E 3rd sing - he (3D) is round', e.g., potato;
/nes-ak-i-kl/ (II) 'three 2D A B S T R A C T F I N A L 3rd plur - they (2D)
are three', e.g., furs; /toq-aq-sa:-k/ (II) 'short 2D be cut 3rd sing -
it (2D) is (cut) short', e.g., piece of cloth; /nes-oq-te-kl/ (II) 'three
ID rigid placed 3rd plur - there are three (ID rigid)', e.g., pieces of
stovewood; /pit-oq-si-t/ (AI), long ID rigid S T A T I V E 3rd sing - he
(ID rigid) is tall', e.g., a tree; /tapu-taq-pi=cik/ (AI) 'two ID flexible
placed 3rd plur - there are two (IDflexible)',e.g., ropes; /epun-taq-a:l/
(transitive animate (TA) verb) 'lower IDflexiblecause - put him (ID
flexible) lower', e.g., a rope.
Morphological Problems
One hardly knows where to begin with the morphological irregular-
ities and extensive morphophonemic alterations these four morphemes
can undergo or trigger in adjacent morphemes. Let us lookfirstat the
stative abstractfinals(AF). For the II verbs it is clear that /-apsk-/
'3D' takes /-e-/ as its stative A F , e.g., /nes-apsk-e-kl/ 'they (3D) are
three', and /-ak-/ '2D' takes A F /-i-/, e.g., /nes-ak-i-kl/ 'they (2D) are ,
three'. The reduced form /newt-taq=q/ 'it (IDflexible)is one' might be
analyzed as /-taq-/ 'IDflexible'taking A F /-i-/: /newt-taq-i-k/ 'one ,
IDflexibleA F 3rd'. In the plural, vowel harmony is usually added:
/nan-taq-i-k-d/ /nantaqqal/ 'they (IDflexible)are 5'. Further com-
plexities appear for the fourth morpheme /-oq-/ 'ID rigid', since in
II statives it has allomorph /-aq-/ which commonly yields /-a:-/, per- '
haps by some productive process: /nes-a:«qd/ 'they (ID rigid) are 3' is
usually used but a less altered form /nes-aq-qal/ [note vowel harmony \
again] is possible, both presumably from /nes-aq-i-k-d/ 'three ID rigid
A F 3rd plur'. From the above it seems that /-oq-/ 'ID rigid' takes II
A F /-i-/ but appears in its /-aq-/ allomorph. W e have now seen that )
both /-ak-/ '2D' and /-oq-/ 'ID rigid' have homonymous allomorphs t
/-aq-/ which can cause confusion to speakers unless the context makes |
 Micmac Semantics 365

clear what sort of object one is dealing with. For instance /toq-aq-ci:c-
i-t/ (containing diminutive final /-ci:c-/) 'he is short' is probably an
occurence of /-oq-/ 'ID rigid' classifying the person's body, but one
speaker felt it might be /-ak-/ '2D' because it seemed to be 'low' or
'flat', this latter meaning coming from the root /toq-/ 'short'.
If we look just beyond this basic sub-system of classificatory medials
for the dimensionality of objects we find another medial whose form is
/-aq-/, this one meaning 'together'. Some morphological adjustments
seem to have developed to keep /-ak-/ '2D', with its allomorph /-aq-/,
separate from /-aq-/ 'together'. The expected forms of the II and AI
statives appear for /-aq-/ 'together': /nes-aq-e-kl/ (II) 'they (together)
are three', /nes-aq-si-cik/ (AI) 'they (together) are three', e.g., poles
tied together. The abnormal forms appear for /-ak-/ '2D': /nes-ak-
i-kl/ (II) 'they (2D) are 3', /nes-ak-i-cik/ (AI) 'they (2D) are 3'. These
forms are unusual in two important ways: 1) the general phonological
process by which k - q/ a is not applied (Fidelholtz 1968:64-74);
and 2) the normal stative A F -asi- is not used, although it appears with
the other three medials for dimensionality: /nes-apsk-asi-cik/ 'they
(3D) are three' /nesu-taq-si-cik/ 'they (IDflexible)are three', /nes-oq-
si-cik/ 'they (3D rigid) are three'.
T H e conditions of alternation between /-oq-/ and /-aq-/ 'ID rigid'
are certainly obscure. However, it may be conditioned by the final
which follows it. W e have seen that AI stative abstractfinal/-asi-/
takes /-oq-/, II abstractfinal/-i-/ takes /-aq-/, and the diminutive fi-
nal /-ci:c-/ takes /-aq-/ (AI form given above; II: /nes-aq-ci:c-kl/ 'they
(ID rigid) (small) are three'). Otherfinalstaking /-oq-/ appear in /nes-
oq-pi-cik/ (AI) 'there are three (ID rigid)', /nes-oq-te-kl/ (II) 'there
are three (ID rigid)'. If /-oq-/ comes as expected from P A */a:xkw/,
metathesis of the */w/ may somehow be involved. Perhaps the neces-
sary rules are in Fidelholtz but I haven't been able to apply them.
Relational classes
So far we have had a look at the four basic noun classes for ob-
jects which Micmac shares with Central Algonquian languages. More
interesting is the hint of other noun classes which might be Micmac in-
novations - we saw a medial /-aq-/ 'together', e.g., /wel-aq-sa-k/ (AI)
'good together S T A T I V E 3rd sing - he is plentiful' referring to a good
growth of trees or blueberries. Such noun classes are concerned with
the spatial relations among objects rather than with their dimension-
 366 Peter Denny

ality, and they occur in many language families. However, they do not
seem to be present in Cree or Ojibwa, except for one medial given by
Baraga (1878) for Ojibwa which I have not found for current speakers:
/-oshk-/ 'side by side' in /nij-oshk-ad/ (II) 'it is two (side by side)'
describing a double-barreled gun. O n the other hand, Micmac seems
to have three others expressed by classificatory medials in c o m m o n
use besides /-aq-/ 'together'. One is the opposite of this latter one,
/-ampk-/ 'apart', e.g., /nes-ampk-ate-kl/ (II) 'there are three (spread
out)'. This one interacts with the "collective" and "distributive" plu-
rals in an interesting way. With the collective (or "internal") plural it
is the individual objects that are spread out: e.g., /nes-ampk-api-cik/
(AI) 'there are three (spread out)' describes three rose bushes planted
apart; with the distributive (or external) plural it is the groups of ob-
jects that are spread out: e.g., /nes-ampk-ap-ulticik/ (AI) 'there are
three's (spread out)' describes groups of three rose bushes each, the
groups being planted apart. These two plurals were discussed by Hew-
son (1976).
The other two relational noun classes are concerned with objects
joined together, either as "layers" expressed by /-anq-/, or as "tiers"
expressed by /-coq-/. The layers seem usually to be 2D objects, e.g., /el-
anq-e-k/ (II) 'it is a layer' [of wallpaper], /el-anq-api-cik/ (AI) 'there
is a layer' [of clouds], /nes-anq-api-cik/ (AI) 'there are three layers'
[plywood piled on top of each other], /nes-anq-ate-kl/ (II) 'there are
three layers' [deer hides, piled-up empty potato bags], /nes-anq-asi-t/
(AI) 'it is three-ply' [clothing]. O n the other hand, the tiers usually
seem to be 3D objects, e.g., /tapu-coq-te-kl/ (II) 'there are two tiers'
[small cubes, boxes, dishes,filledpotato bags], /newkt-icoq-a:si-t/ (AI)
'he is one (object with tiers)' [a cup placed in its saucer], /nesu-coq-
te-kl/ (II) 'there are three tiers' [books]. For some speakers books
seemed to be a borderline item between '3D tier' and '2D layer', so
that avoiding both classificatory medials seemed best, /nesu-te-kl/ (II)
'there are three'. Thin books like scrapbooks could be layers, /nesu-
anq-ate-kl/ (II) 'there are 3 layers' [scrapbooks piled up]. Layers and
tiers seem to involve things which are joined along reasonablyflatsides.
Consequently many ID objects such as logs and ropes cannot be so
joined. Those ID objects that do haveflatsides seem to be included
in the tiers class presumably because they are not thin enough to be
layers, e.g., /nesu-coq-pi-cik/ (AI) 'there are three tiers' could be used
 Micmac Semantics 367

for three piled up 2X4s or boards. Although horizontal orientation

seems most frequent, layers and tiers can be in any orientation, e.g.,
layers of wallpaper.
Problems of cultural interpretation
The discovery of four relational noun classes in Micmac not found
in Ojibwa or Cree provides some support for the view originated by
Hoffman (1955) and developed by M c G e e (1978,1980) that the Micmacs
are not hunter-gatherers on the pattern of the Cree-Montagnais and the
Northern Ojibwa. According to this theory, the wealthyfisheryof the
Atlantic provinces allowed the development of a more complex hunter-
gatherer society involving permanent villages, economic specialization
and trade, and an emerging social hierarchy. Relational noun classes
are reported for m a n y different agricultural societies around the world,
but they are not found for hunting groups such as the Inuit, nor for
the Cree-Montagnais w h o have the most exclusively hunting past of
the Algonquians, nor are they reported for the Australian languages
(Dixon 1982; Worsley 1954). They are, however, found among the
Haida, w h o are also wealthy fishermen with permanent villages and
social stratification (Enrico 1981).
The reasons for this association of relational noun classes with the
slight increase in societal complexity found in these wealthyfishingso-
cieties are obscure. It m a y be that the construction and maintenance
of permanent structures such as homes andfishingweirs focuses at-
tention on the spatial relations among things serving as parts of these
assemblages. There might also be a more indirect effect of permanent
settlement and the permanent division of the landscape into economic
sites such asfishingweirs - a new emphasis on spatial relations within
the landscape m a y be required which is reflected in how objects are
classified. Future investigations need to determine whether relational
classes are also found among adjacent hunting tribes such as the Male-
cite and Penobscot, and among agricultural Algonquians such as the
Delaware and Shawnee.

ACKNOWLEDGEMENTS
Thanks are due to my Micmac language consultants: Bernie Francis at Member-
tou Reserve in Nova Scotia, Manny Metallic at Restigouche Reserve in Quebec, and
Jane Simon and Michael Francis at Big Cove Reserve in New Brunswick. Good ad-
vice came from linguists specializing in Micmac, Watson Williams and Paul Proulx.
Mildred Milliea and Rebecca Simon helped with the arrangements at Big Cove. This
368 Peter Denny

research was supported by a grant from the Social Science and Humanities Research
Council of Canada.

REFERENCES

Baraga, F.
1878 A Dictionary' of the Otchipwe Language. Montreal: Beauchemin et Valois

Dixon, R.M.W.
1082 Where Have All the Adjectives Gone? and Other Essays in Semantics
Syntax. Berlin: Mouton.
Enrico, J.
1081 The Haida Classifiers. Manuscript.

Fidelholtz, James
1068 Micmac Morphophonemics. Ph.D. dissertation, MIT.

Hewson, John
1076 The So-called Dual in Micmac. Paper given at 1076 Algonquian Confer-
ence, Montreal.
Hoffman, B.G.
1055 Historical Ethnography of the Micmac of the 16th and 17th Centuries.
Ph.D. dissertation, University of California, Berkeley.
McGee, H.
1078 The Micmac Indians — the Earliest Migrants. In Banked Fires — Ethnics
of Nova Scotia. D. Campbell, ed. Port Credit, Nova Scotia: Scribbler's
Press.
1080 Micmac Residence Patterns and House Types. (Report prepared for Pa
Canada).
Worsley, P.M.
1054 Noun Classification in Australia and Bantu: Formal or Semantic? Oceania
24:275-288.