1 Introduction - University of Essex

The Typology of Nominal Tense
Rachel Nordlinger and Louisa Sadler ¡
University of Melbourne, Australia and University of Essex, UK ¡
August 6, 2002
1 Introduction
In this paper we discuss the marking of tense/aspect/mood (henceforth TAM) on nouns, pronouns,
determiners and other NP/DP constituents (henceforth ‘nominals’). 1 That nominals may be inflected
for TAM has been noted in the grammatical descriptions of individual languages for some
time (e.g. Guasch (1956) and Gregores and Suárez (1967) on Guaraní, Firestone (1965) on Sirionó,
among others), and, more recently, in various typological works (e.g. Mel’čuk 1994, Evans 2000,
Lehmann and Moravcsik 2000, Raible 2001). Nevertheless, the categories of TAM remain generally
considered to be inflectional features of verbal classes only, with the possibility of TAM as
an inflectional category of nominals largely omitted from general linguistic discussion. 2 Furthermore,
the extent of TAM-inflection of nominals and its properties in the world’s languages has,
until now, remained largely unexplored. The purpose of this paper is to provide a detailed survey
of the phenomenon and its implications for linguistic theory. In fact, as we show, there are a surprising
number of languages which associate some type of TAM information with NP constituents,
suggesting that the phenomenon is far less marginal than the general paucity of discussion in the
literature might lead one to expect. 3 The encoding of TAM on nominals can have one of two broad
functions. In one, it functions essentially like verbal TAM to provide TAM information for the whole
proposition. Propositional TAM on nominals is found in Lardil (Australia), for example, in which
language an extra level of case marking occurs on non-subject nominals encoding propositional
1 Some parts of this paper have appeared previously as Nordlinger and Sadler (2000) and Sadler and Nordlinger
(2001). For comments, suggestions and many interesting leads we thank Alexandra (Sasha) Aikhenvald, D.N.S. Bhat,
Joan Bresnan, Alec Coupe, Matthew Dryer, Nick Evans, Brent Galloway, John Hajek, Jacqueline Lecarme, Hitomi
Ono, Bill Palmer, Tom Payne, Nick Piper, Joachim Sabel, Ruth Singer, Andy Spencer, Judith Tonhauser, audiences
of LFG00 and LFG01, and members of the LINGTYP discussion list. We are especially grateful to Joan Bresnan,
Nick Evans and Judith Tonhauser for extensive comments on an earlier version of this paper, and to Sasha Aikhenvald,
Sebastiana Ertel (via Dagmar Jung), Nick Evans, Brent Galloway, P.J. Mistry, Hitomi Ono, Tom Payne and Joachim
Sabel for providing access to unpublished data. Needless to say, none of these people are to be held responsible for
any remaining inadequacies of this paper, and we apologise if we have not always been able to do justice to their
comments. Nordlinger would like to acknowledge the financial support of the Australian Research Council (APD
F9930026) and the University of Melbourne, and Sadler the University of Essex for a period of sabbatical leave during
which this paper was prepared.
2 Opinions differ as to whether tense, aspect and mood should be considered properties of the clause itself, or of
the clausal head (i.e. the verb). Foley and Van Valin (1984: 224) and Van Valin and LaPolla (1997:47) treat all three
differently: for them, aspect is a property of heads (‘nuclear layer’), tense a property of the clause, and (root) modality
a property of the ‘core’ (including the head and its core arguments). It is important, however, to distinguish between
the semantic/syntactic scope of a category, and its morphological realisation. Tense, aspect and mood, when encoded
morphologically in a language, are usually (inflectional) properties of verbs, despite the fact that their syntactic and
semantic scope may be clausal (or otherwise).
3 Throughout this paper we use the terms ‘nominal’ and ‘nominal constituent’ to refer to either nouns or other
constituents of NP/DPs (e.g. articles, pronouns, etc.). The term ‘nominal TAM’ thus refers to the inflectional encoding
of tense, aspect or mood information on nouns or any other NP constituents.
1

tense in agreement with the verb (see section 2.2 for further discussion). 4
(1) Ngada niwentharr maarn-arr wu-tharr.
1SG.NOM 3SG.NFOBJ spear-NFOBJ give-NFUT
‘I gave him a spear’ (Klokeid 1976:476, 56b)
(2) Ngada marndi-thu niwentha niwen-kur-u
1.SG.NOM rob-FUT 3SG.FOBJ
‘I will steal his wife for him.’ (Hale 1997:201)
kerndi-wur-u.
3SG.GEN-INSTR-FOBJ wife-INSTR-FOBJ
Alternatively, nominal TAM inflection may specify TAM information intrinsic to the NP itself,
independent of any propositional tense. This is what is found in languages such as Tariana (Brazil),
in which nouns are inflected with tense markers to temporally locate the nominal predicate internal
to the NP, independently of the TAM of the clause (which in the example below is marked with the
‘remote past reported’ tense/evidentiality clitic attached to the verb). Non-propositional nominal
TAM is discussed further in section 3.
(3) di-ma¢ thepi
eta-miki-¢ e=pidena
i-nuku
to.water 3SG.NF-throw.CAUS=REM.P.REP eagle-NOM.PST-NF-TOP.NON.A/S
‘He threw the remains of the eagle (lit. what used to be the eagle) into water.’ (Aikhenvald
(to appear))
In this paper we present a typological survey of the phenomenon of nominal TAM inflection, examining
its functions, properties and realisations cross-linguistically. The focus of this study is on
tense/aspect/mood as an inflectional 5 category of nominals. 6 There are a number of other ways in
which nominals can combine with elements encoding TAM information which therefore fall outside
the scope of the present study. These include TAM clitics that are phonologically attached to
nominal constituents. Such clitics are, by definition, syntactically and morphologically independent
of their hosts. They cannot, therefore, be considered an inflectional property of the nominals;
their attachment to nominal constituents is a property of the phonology instead. Languages which
use TAM clitics attached (optionally) to nominal constituents to encode propositional TAM include
the Australian language Garrwa (Furby and Furby 1977), and the Arawak languages Apurinã (Facundes
2000) and Tariana (Aikhenvald (to appear)).
Nor do we consider as examples of nominal TAM such vestiges of verbal tense/aspect/mood marking
as may be retained in deverbal nominalisations. In Polish, for example, the aspectual distinction
imperfective/perfective encoded with verbs is retained in derived action nominals. Thus
corresponding to the verbal pair czytać (imperfective)/przeczytać (perfective) ‘to read’ are the derived
action nominals czytanie/przeczytanie. While both could be translated into English as ‘the
reading’ (e.g. ‘The reading of the book gave me much pleasure’), czytanie refers to the process
of reading, while przeczytanie refers to the totality of the act of reading (Comrie and Thompson
4
NFOBJ = nonfuture object, NFUT = nonfuture, FOBJ = future object
5Though it is clearly inflectional, we shall see that nominal TAM is not always obligatory in languages in which it
occurs.
6We do not, however, rule out the possibility that TAM within NPs could be expressed periphrastically, involving
separate particles or function words in some languages. Exploring this periphrastic expression of nominal TAM is
beyond the scope of the present paper, although our research efforts to date have not revealed any such examples.
2

1985:363). This aspectual distinction is clearly primarily an inflectional category of the original
verb, rather than of the nominal word class to which the derived forms belong. It can therefore be
held in direct contrast with true cases of nominal TAM on deverbal nominalisations where a TAM
distinction encoded on all types of nominals is extended (predictably) to deverbal nominals also
(see the discussion of Hixkaryana in section 3).
Nominal TAM inflection has implications for many aspects of linguistic theory. The fact that both
nouns and verbs can be inflected for the same morphosyntactic features has interesting ramifications
for the categorization of word classes and for the properties associated with nouns crosslinguistically.
It also raises interesting questions for formal theories of grammar. In particular,
examples such as (1), in which a dependent nominal is inflected for a clausal property, challenge
the common assumption that clause-level properties are contributed only by clausal heads (i.e.
verbs). Such implications are explored in further detail in section 5.
This paper is organised as follows. We begin by surveying the scope of the phenomenon and
its various properties across the world’s languages. In this empirical survey we have endeavoured
to bring together data from many lesser known languages, and discuss the properties for
each language in sufficient detail to illustrate the interaction of nominal TAM inflection with the
rest of the grammatical system. As noted above, the functions of nominal TAM inflection can be
categorised according to whether the TAM inflection serves to encode propositional or independent
TAM information. Section 2 discusses nominal TAM inflection which functions to encode the
propositional TAM. Then in section 3 we discuss the use of nominal TAM inflection to encode
non-propositional TAM, that is, to temporally locate the nominal (or other NP constituent) itself independent
of any clausal TAM specification (see example 3 above). Having presented the primary
data, we turn in section 4 to a more detailed discussion of the various types of nominal TAM found
cross-linguistically and the relationships between them. Then, in the final section we discuss the
broader implications of the nominal TAM data for various aspects of linguistic theory, and identify
some avenues for further research.
2 Propositional TAM on Nominals
In this section we discuss TAM inflection on nominals which functions to encode TAM information
for the proposition as a whole. The most straightforward examples of this type of nominal TAM
inflection involve nominal predicates in verbless clauses which, in some languages, may carry (a
subset of) TAM inflections like their verbal counterparts. We discuss these cases in section 2.1. In
some languages, however, we find non-predicate nominals – nominals functioning as arguments or
adjuncts – encoding propositional TAM. Thus, in these languages, (TAM) properties of the whole
proposition may be encoded on clausal dependents. We discuss this type of nominal TAM inflection
in section 2.2.
2.1 Propositional TAM on Nominal Predicates
In some languages nominals that function as clausal predicates are morphologically marked for
categories of tense, aspect and/or mood. In many ways this is the nominal equivalent of regular
3

verbal TAM on verbal heads: the nominal is the clausal predicate and therefore is encoded for
propositional TAM.
One language with only this type of nominal TAM is Bininj Gun-wok (non-Pama-Nyungan, formally
known as Mayali), a dialect chain of Arnhem Land in Northern Australia. 7
In Bininj Gun-wok (Evans (to appear)) predicate nominals and adjectives are inflected for a subset
of the regular verbal TAM markers: the past imperfective -ni (which in this context simply marks
past), and the irrealis mood marker. Examples include (4)-(6) with the past tense and (7) with both
the past tense and the irrealis mood marker niwirrinj: 8
(4) Mayh na-mekke nakka bininj-ni.
bird MASC-DEM MASC.DEM human-PAST
‘Those birds, they were human then,’ (Evans to appear:680, 13.27b)
(5) Na-mak-ni.
MASC-good-PAST
‘He was a good man.’ (ibid:682, 13.37c)
(6) . . . nga-wurdurd-ni
. . . 1.SG-child-PAST
. . .
‘. . . (when) I was a child . . . ’ (ibid:437, 8.95)
(7) Yawkyawk bokenh na-wu bene-berd-djenj-ni yimankek kun-dad-niwirrinj.
young.girl two MASC-REL 3.DU-tail-fish-PAST CTRFAC NEUT-leg-IRR
‘There were two young girls who had tails like fish, they didn’t have legs.’ [lit. ‘there were
no legs’] (ibid:437, 8.96)
Note that some predicate nouns also take verbal pronominal prefixes (e.g. nga- ‘1.SG’ in (6), bene-
‘3.DU’ in (7)). However, this is not a requirement for the addition of TAM suffixes. In (4) we see a
tense-inflected nominal with no prefixes (bininj- ‘human’), while (5) shows the use of a tense affix
with a nominal retaining regular nominal gender marking (na-mak ‘MASC-good’). The retention
of regular gender morphology in such nouns is just one indication that they retain their nominal
properties and have not undergone morphological derivation to verbs.
A particularly common source of nominal predicates marked with TAM information is the use of
nominalisations as predicates in subordinate clauses, as in the following from Turkish:
asaˇg£ (8) Çocuk-lar-a ya inip
kendisi-ni sokak-ta bekle-dik-leri-ni
child-PL-to
söyle-di.
downwards having.descended her-DO street-in await-VN:NF-their-DO
say-3SG.PAST
‘She told the children that they went (had gone) down and waited for her in the street.’
(Comrie and Thompson 1985: 362, ex 50)
7Other languages with this type of nominal TAM inflection include Tundra Nenets (Salminen 1997) and Turkish
(Lehmann and Moravcsik 2000:742).
8
IRR = irrealis, CTRFAC = counterfactual, MASC-REL = masculine relative pronoun.
4

asaˇg£ (9) Çocuk-lar-a ya inip
kendisi-ni sokak-ta bekle-yecek-leri-ni
child-PL-to
söyle-di.
downwards having.descended her-DO street-in await-VN:FUT-their-DO
say-3SG.PAST
‘She told the children that they would go down and wait for her in the street.’ (ibid, ex 51)
In each of the above examples the subordinate clause is headed by a deverbal noun, the form of
which differs on the basis of tense: deverbal nouns in -dik (8) encode non-future tense, while those
in ecek (9) mark future tense. This is quite distinct from the tense marking found on verbs in
Turkish which mark a past/nonpast distinction.
Such TAM encoding on nominalised verbs is different from the other types of nominal TAM discussed
in this paper in a number of respects. Firstly, in some languages deverbal nominals are the
only type of nominal which can carry tense (or other TAM) information. The encoding of TAM
information cannot be considered a property of nominals in general in these languages, but only
a property of nominalised forms derived from verbs. Secondly, TAM in this context usually encodes
relative rather than absolute TAM. Thirdly, as with the Turkish example above, this type of
nominal TAM is often associated with the nominalisation morphology itself, and is therefore not
always clearly a separate inflectional category. However, despite these differences it can be clearly
distinguished from the vestiges of verbal TAM retained in deverbal nominals in languages like Polish
(see section 1 for discussion) – often encoding different TAM distinctions than verbs, as in the
Turkish case – and therefore constitutes a particular type of nominal TAM marking.
An interesting twist is common in Australian languages, where case marking morphology is used
with subordinate nominalised predicates to encode relative tense information (this is referred to by
Dench and Evans (1988) as ‘t-complementizing case’). 9 The following examples from Wambaya
(non-Pama-Nyungan) are typical:
(10) Bungmaji gi-n mirra yanduji-ni barrawu.
old.man 3.SG.S-PROG sit mind-LOC house
‘The old man’s staying (here) looking after the house.’ (Nordlinger 1998:213, 8-2)
(11) Yarru ng-amany ngaji-nka ngaya.
go 1SG.S-PST.TWD see-DAT her.DAT
‘I came to see her.’ (ibid:214, 8-9)
(12) Gumarra g-u nyagaj-ba yarru-nnga.
calf 3SG.S-FUT be.tired-FUT go-ABL
‘His calves will be tired from walking.’ (ibid, 8-8)
The use of the locative case in (10) indicates that the subordinate clause is simultaneous with the
main clause; the dative case in (11) marks the subordinate clause as following (and being a purpose
of) the main clause event, while in (12) the ablative case is used to mark the subordinate clause
event as preceding that of the main clause. In Kayardild, the proprietive and consequential case
suffixes can be used with derived nominals to mark present role or responsibility and prior action,
respectively (Evans 1995:464):
9 The use of case markers as subordinators is also reported in the literature for the Kartvelian language Laz (Harris
and Campbell 1995) and the Tibeto-Burman language Newari (Genetti 1991).
5

(13) dara-n-kuru dangkaa
break-NMLZR-PROP person
‘man who has to do the circumcising’
(14) dara-n-ngarrba dangkaa
break-NMLZR-CONS person
‘man who has circumcised (someone) before’
These restricted TAM systems on deverbal nominals can be contrasted with the more robust systems
of languages such as Hixkaryana, discussed in section 3, which mark TAM on all nominal types,
not simply deverbal nominalisations.
2.2 Propositional TAM on Dependent Nominals
Above we discussed the encoding of propositional TAM on nominals which function as the predicate
of the clause. There are a surprising number of typologically, genetically and geographically
diverse languages, however, which encode propositional TAM on nominals which are not clausal
predicates, but which function as arguments and adjuncts of the verbal predicate. We refer to this
function of nominal TAM inflection as propositional TAM on dependent nominals. This type of
nominal TAM inflection is particularly problematic for many formal theories of language structure
which assume that such clause-level information (i.e. that which pertains to the whole clause or
proposition) must necessarily be associated with clausal heads, not dependents (see Nordlinger
and Sadler 2000, and section 5 for discussion).
This section presents an empirical overview of this type of nominal TAM cross-linguistically. The
basic organisation of the section is functional in that we have attempted to systematize the data in
terms of the role of the nominal TAM marking, especially vis-à-vis any verbal TAM inflection. In the
majority of cases we have encountered, the propositional TAM on nominal dependents essentially
supplements that expressed morphologically in the verbal system, although the two systems may
not always encode identical categories (see for example the discussion of Kayardild). In a minority
of cases, however, a TAM distinction is only or primarily expressed within the nominal system.
In Lardil, a Tangkic (non-Pama-Nyungan) language of northern Australia, the case markers on
non-subject complements are dependent on the tense properties of the clause, as expressed on the
verb. Objects (and other non-subject dependents, see below) are inflected with one of three possible
objective case markers corresponding to the tense inflection of the verb: -(i)n ‘objective (OBJ)’ is
used with the plain, unmarked verbal inflection -¤ /-tha (GNF)); -(w)ur ‘future object (FOBJ)’ is
used with the future tense verbal inflection -thur (FUT)); and -(ng)arr ‘marked non-future object
(NFOBJ)’ is used with the marked non-future tense verbal inflection -tharr (NFUT)) (Klokeid 1976,
Hale 1997). 10 In addition, objects of imperative clauses take the nominative (unmarked) case.
Consider the following examples illustrating the use of plain, future, non-future and nominative
object marking respectively:
10 Richards (2001) discusses the many grammatical differences that exist between present-day Lardil (his ‘New
Lardil’) and the Lardil of the 1960s when Ken Hale did his early fieldwork (his ‘Old Lardil’). The case/tense properties
discussed here are features of Old Lardil.
6

(15) Ngada niween maarn-in wu-tha.
1SG.NOM 3SG.OBJ spear-OBJ give-GNF
‘I gave him a spear’ (Klokeid 1976:476, 56a)
(16) Ngada bilaa wu-thur ngimbenthar diin-kur wangalk-ur.
1SG.NOM tomorrow give-FUT 2SG.FOBJ this-FOBJ boomerang-FOBJ
‘I’ll give you this boomerang tomorrow.’ (ibid:493)
(17) Ngada niwentharr maarn-arr wu-tharr.
1SG.NOM 3SG.NFOBJ spear-NFOBJ give-NFUT
‘I gave him a spear’ (ibid:476, 56b)
(18) (Nyingki) ne-tha kiinda.
(2SG.NOM) hit-GNF that.person(NOM)
‘(You) hit that person.’ (Hale 1997:21)
These case markers also appear on other non-subject complements, following and outside of the
regular semantic case inflection (so that these examples involve a form of the case stacking well
known in Australian languages (Dench and Evans 1988, Dench 1995, Nordlinger 1998b)). In
(19) the instrumental NP ‘his wife’ is inflected with two case markers: first the instrumental case
marker encoding its grammatical function within the clause, and then by the future objective case
in agreement with the future tense of the verb.
(19) Ngada marndi-thu niwentha niwen-kur-u
1.SG.NOM rob-FUT 3SG.FOBJ
‘I will steal his wife for him.’ (Hale 1997:201)
kerndi-wur-u.
3SG.GEN-INSTR-FOBJ wife-INSTR-FOBJ
Note that the case markers are not limited to appearing at the periphery or on the head of the
nominal constituents they mark. (20) exemplifies future marking participating in case concord
within a NP and (21) illustrates penetration into a nominal compound (following Hale (1998) the
compound is marked with #).
(20) Ngada were-thur kiin-kur karnan-kur maarn-kur
1.SG.NOM throw-FUT that-FOBJ long-FOBJ spear-FOBJ
‘I will throw that long spear’ (Hale 1998:200, 8)
(21) Ngada
1.SG.NOM
kurri-thur
see-FUT
mang-kur#dangka-r
child-FOBJ#person-FOBJ
‘I will see the child’ (ibid:201, 9b)
Nominal TAM in Lardil, then, is effected through the use of a particular set of case markers which
interact with the TAM inflections of the verb. These case markers combine expression of temporal
information with regular relational case for direct objects, and are placed outside semantic case
inflections for other non-subject complements. Although, for the most part, the TAM-related case
markers appear to simply agree with the TAM inflection of the verb, the two can be shown to operate
independently of each other. Firstly, the verbal TAM markers have distinct forms marking negative
polarity, while the corresponding nominal TAM/case suffixes appear in the same form irrespective
of modality, indicating that they cannot simply be a copy of the verbal forms. Thus:
7

(22) Ngada bule-thur yak-ur.
I catch-FUT fish-FOBJ
‘I will catch a fish.’
(23) Ngada bule-nengkur yak-ur.
I catch-NEG.FUT fish-FOBJ
‘I will not catch a fish.’ (Hale 1997:36)
Secondly, the same verb form (glossed GNF for ‘general non-future’) is used with both general nonfuture
tense and imperative mood. The two are differentiated however, through the case marking
of the object which is nominative for the imperative use (see (18) above) and (plain) objective
otherwise (see (15) above). Such examples show clearly the interaction between the nominal and
verbal morphology in fully specifying the propositional TAM in Lardil.
The existence of TAM as an inflectional category of nominals is shown even more clearly by the
related language Kayardild which displays an even more elaborate system of TAM-sensitive case
marking. There are five such ‘modal’ cases in Kayardild (Evans 1995), which co-occur with
different verbal inflections to encode the tense/mood value for the clause. As in Lardil, the modal
case is attached directly to objects, but otherwise appears outside the regular case marker for the
NP. Consider the following examples (all taken from Evans (1995)): 11
(24) Ngada yalawu-jarr yakuri-na mijil-nguni-na.
I(NOM) catch-PST fish-M.ABL net-INST-M.ABL
‘I caught fish with the net.’ (Evans 1995:108, 3-30)
(25) Ngada yalawu-ju yakuri-wu mijil-nguni-wu.
I(NOM) catch-POT fish-M.PROP net-INST-M.PROP
‘I will catch fish with the net.’ (ibid:109, 3-31)
In (24) and (25) both the direct object yakuri ‘fish’ and the instrumental adjunct mijil-nguni- ‘net-
INST-’ are inflected with case markers encoding information about clausal tense and mood. In (9),
the modal ablative case signals clausal past tense, while in (10) the modal proprietive case is used
with the ‘potential’ verbal suffix to encode futurity. On the basis of these examples it may appear
that this modal case is simply concordial agreement with the verbal inflection. However there are
several reasons for thinking that this is not the case.
Firstly, note that modal case marking is not limited to non-subject complements, but is also found
on sentential adjuncts such as ‘tomorrow’ and ‘yesterday’. This indicates that it is not simply a
VP-based phenomenon and moreover would be very unusual in a concord phenomenon.
(26) Ngada kurri-nangku mala-wu (balmbi-wu).
1.SG(NOM) see-NEG.POT sea-M.PROP morrow-M.PROP
‘I won’t be able to see the sea (tomorrow).’ (Evans 1995:404, 10-12)
(27) Ngada kurri-nangku mala-y (barruntha-y).
1.SG(NOM) see-NEG.POT sea-M.LOC yesterday-M.LOC
‘I could not see the sea (yesterday).’ (Evans 1995:404, 10-13)
11 The modal case suffix is marked with an M. in its gloss: M.ABL ‘ablative case in modal function’.
8

Secondly, as the above examples also illustrate, Kayardild shows even more clearly than Lardil
that there is not always a one-to-one relationship between the modal case category and the verbal
tense/mood inflection that it co-occurs with. Rather, the two are independent systems with different
semantic values which work together to provide in combination a composite tense/mood value for
the whole clause (see Evans (1995: Chapter 10) for a full discussion). Thus, in examples (26) and
(27) the verbal inflection remains constant, and it is through the different modal case inflections
that the clausal tense/mood distinction is encoded. The ‘negative potential’ verbal inflection is used
here with its meaning of ‘inability’: combining with the “future” meaning of the modal proprietive
case marker in (26) places this inability in the future, while combining with the “instantiated”
meaning of the modal locative in (27) expresses that there was a real occasion, yesterday, when
the inability existed (Evans 1995:404). The following examples, with the apprehensive verbal
inflection, exemplify this point even further with three possible alternatives depending on the form
of modal case:
(28) Warrjawarri ngada barrbiru-tha manarr-iy, kurri-nyarra ngijin-inj,
slowly(NOM) 1.SG(NOM) lift-ACT
kala-nyarr, rabi-nyarr.
torch-M.LOC see-APPR 1.SGPOSS-M.OBL
fly-APPR arise-APPR
‘Unhurriedly I lifted the bark torch, in case (the diver birds) should see me and fly off.’
(Evans 1995: 404, 10-14)
(29) Nying-ka ngudi-na wangalk, ngada ngumban-ju burldi-nyarr.
2.SG-NOM throw-NEG.IMP boomerang(NOM) 1.SG(NOM) 2SG-M.PROP throw-APPR
‘Don’t you throw the boomerang, or I’ll throw one at you.’ (ibid:405, 10-15)
(30) Thararra kali-nyarra wambal-iya, naa-nyarr.
ember-NOM jump-APPR bush-M.LOC burn-APPR
‘(Look out), the embers are jumping into the bush, it might burn.’ (ibid:405, 10-16)
The basic use of the apprehensive verbal inflection APPR is to express that the event is (or would
be) unpleasant, and for this the oblique modal case is used, as with ngijin-inj (1.SGPOSS-M.OBL),
the object of kurri-nyarra (see-APPR) in (28). In (29) on the other hand, the modal proprietive
(which expresses “future” meanings) is used to stress the speaker’s certainty of being able to effect
an unpleasant retaliation. Finally, in (30), the unpleasant event is actually taking place, and so the
modal locative is used to indicate the reality of the occurrence (Evans 1995:405).
Further evidence of the distinction between the verbal TAM system and that of the modal case
markers is the fact that modal case can appear in clauses which have no verb at all, yet still encoding
the same TAM information: 12
(31) Ngada dathin-kiring-ku kamarr-iring-ku.
1.SG(NOM) that-ALL-M.PROP stone-ALL-MPROP
‘I will (go) to that stone.’ (Evans 1995:403, 10-7)
12 It is not clear what the best analysis of these clause types is. For present purposes we assume that in a sentence
such as (31) the allative case marker has predicative semantics and serves as the main predicate of the clause. Another
possible analysis is that the main verb (e.g. ‘go’) has been ellipsed.
9

(32) Jina-na darr-ina nying-ka jirrka-an-kina?
where-M.ABL time-M.ABL 2.SG(NOM) north-FROM-M.ABL
‘When did you (come back) from the north?’ (ibid:403, 10-8)
The fact that the modal case system in Kayardild is distinct from the system of verbal TAM inflection
shows clearly that TAM (as encoded by modal case) is a meaningful inflectional category of
nominals in this language.
A well-known instance of TAM/case interaction is found in many South Asian languages, including
for example Urdu/Hindi, Punjabi and Gujarati, all of which exhibit an aspectually-based ergative
system. 13 These languages differ from Lardil and Kayardild in that it is the case of the subject that
is sensitive to TAM contrasts, as is clear from the following discussion of Gujarati. 14
Gujarati has 6 nominal cases: nominative (unmarked), ergative/locative (-e), accusative/dative (ne),
genitive (-n-), instrumental/ablative (-thi) and locative (-mãã). The case of the subject varies
according to clausal TAM properties. When the verb is in the past tense (33), the subject has
nominative case and when it is in the perfective aspect the subject is marked with the ergative case
(34). There is also a dependency between subject case marking and other semantic distinctions at
the clausal level: in (35) the subject is ergative in a recommendative clause; in (36) it is a form
homophonous with the accusative/dative in the desiderative mood; 15 and in the abilitative mood
the subject is marked with the instrumental case (37) (Mistry 2001).
16¥ 17
13 In Urdu/Hindi and Punjabi case is not expressed strictly morphologically, since there is good evidence that some
case markers are clitic particles and so independent elements within the syntax of NP. In contrast, Mistry (2001)
is explicit in treating the Gujarati case markers as inflectional and in the following we discuss the Gujarati data. For
discussion of the Urdu/Hindi data the reader is referred to Mohanan (1994) and Butt and King (to appear). The Punjabi
facts are discussed in Bhatia (1993).
14 A similar but simpler system is found in Guaymi (Chibchan, Costa Rica and Panama), in which the ergative case
is found only in the past tense (we are grateful to Tom Payne for providing this data):
(1) Tom Dori dëma-e
T. D. greet-NPST
‘Tom greets Doris.’
(2) Toma-gwe Dori dëma-ini
T.-ERG D. greet-PST
‘Tom greeted Doris.’
15 Mistry (2001) himself doesn’t gloss the case suffix in these examples, but argues that it is different from both the
accusative and dative forms with which it is homophonous. Evidence for its distinctness includes the fact that the first
person pronoun usually used in accusative and dative cases cannot be used in these contexts where the unglossed form
appears (p. 342).
16 F = feminine, INF = infinitive, III = third person.
17 Similar pairs can be given for Urdu/Hindi also, in which the modality appears to come from the case clitic. The
choice of ergative or dative case determines sentential features — the ergative introduces a clear desiderative modality,
the dative case functions more as a default marker which could be interpreted either in terms of obligational force
or with the desiderative modality, depending on the particular context of the utterance. The verbal complex itself is
invariant across this pair of sentences).
(1) nadya=ne
Nadya.F=ERG
zu
zoo.M.LOC
ja-n-a
go-INF-M.SG
h¦ i
be.PRES.3.SG
‘Nadya wants to go to the zoo.’
10

(33) kaaga§
ˇsilaa
lakh-t-i.
Sheela(F).NOM letter(M) write-PST-F
‘Sheela used to write a letter.’ (Past) (Mistry 2001:335, 6)
(34) cop¨ ˇsilaa-e
i kharid-y-i.
Sheela(F)-ERG book(F) buy-PERF-F
‘Sheela bought a book.’ (Perfective) (ibid:338, 15a)
(35) cop¨ ˇsilaa-e
i kharid-w-i.
Sheela(F)-ERG book(F) buy-INF-F
‘Sheela should buy the book.’ (Recommendative) (ibid:339, 15b)
(36) cop¨ ˇsilaa-(n)e
i kharid-w-i che.
Sheela(F)-ACC/DAT? book(F) buy-INF-F is
‘Sheela wishes to buy the book.’ (Desiderative) (ibid:339, 15c)
(37) cop¨ ˇsilaa-thi
i kharid-aa-ˇs-e.
Sheela(F)-INST book(F) buy-ABIL-F-III
‘Sheela could buy the book.’ (Abilitative) (ibid:338, 15e)
A particularly interesting system is found in Pitta Pitta (Pama-Nyungan, Australia) in which the
case markers for subjects, objects and instruments all encode a distinction between future and nonfuture
tense (Blake 1979). Furthermore, the case marking system itself differs according to the
tense of the clause: future tense involves a nominative/accusative case distinction, and non-future
a three-way distinction between intransitive subject (S), transitive subject (A) and object (O). The
forms are shown in Table 1 ( after Blake 1987:59). 18
Table 1: Pitta Pitta case/tense suffixes
S A O Inst
Non-Future -¤ -lu -nha -lu
Future -ngu -ngu -ku -ngu
Tense information is also expressed on the verb in Pitta Pitta, but the system is different to the
future/non-future contrast encoded by nominals. Verbs distinguish the three core tenses: present
(-ya), past (-ka) and future tense, which is unmarked (Blake 1979:201-2). The case alternations on
subjects (38-39) and objects (40-41) and their interactions with the tense of the verb are illustrated
by the following examples.
(38) Ngamari karnta-ya ngartu-nga kankari-marru.
mother(NOM) go-PRES nardoo-PURP knife-having(NOM)
‘Mother’s going for (to get) nardoo (edible plant sp.) with a knife.’ (Blake 1987:59, 4.11)
(2) nadya=ko zu
h¦ ja-n-a i
Nadya.F=DAT zoo.M.LOC go-INF-M.SG be.PRES.3.SG
‘Nadya has/wants to go to the zoo.’
18 Blake (1979) does however note that the non-future object form -nha is used by some of his language consultants
for future tense also, alongside the specifically future tense form -ku.
11

(39) Ngamari-ngu karnta ngartu-nga kankari-marru-ngu.
mother-NOM.FUT go nardoo-PURP knife-having-NOM.FUT
‘Mother will go for (to get) nardoo with a knife.’ (ibid:60, 4.13)
(40) Ngamari-lu ngunytyi-ka ngali-nha mangarni-marru-nga-nha kathi-nha.
mother-ERG give-PAST we.DU-ACC bone-having-GEN-ACC meat-ACC.
‘Mother gave us the doctor’s meat.’ (ibid, 4.12)
(41) Ngamari-ngu ngunytyi ngali-ku mangarni-marru-nga-ku
mother-NOM.FUT give
kathi-ku.
meat-ACC.FUT.
we.DU-ACC.FUT bone-having-GEN-ACC.FUT
‘Mother will give us the doctor’s meat.’ (ibid, 4.14)
In all of the examples of nominal TAM that have been dicussed so far, the TAM information associated
with the nominal morphology works in conjunction (perhaps even in agreement) with the
verbal TAM to encode the TAM for the clause as a whole. There are some languages, however, in
which the TAM information encoded on an NP constituent can be the sole exponent of a clause-level
TAM category. Since the verb does not encode the relevant distinction at all, in these examples, it
is quite clear that the propositional TAM information is introduced solely by the dependent NP.
In Sirionó (Firestone 1965) (Tupí-Guaraní, Bolivia) it seems that propositional temporal distinctions
can be expressed by using the same TAM markers on nouns as are used on verbs. Nouns
inflected for such features can serve as nominal predicates, that is, as the heads of complete propositions
as in (42) —- the quite widespread phenomenon of nominal TAM on nominals in predicative
function has already been discussed in section 2.1 above.
(42) Ñéd¸a-he-rae.
road-REFL-FUT
‘It will be a road.’ (Firestone 1965:24)
Firestone (1965) establishes that such TAM inflected words are indeed nominal. On the basis of
the distinctive inflectional affixes which stems take he distinguishes a class of ‘nominals’, which
take tense and aspect but nothing else (e.g. j´vku ‘turkey’, áe j´vku-ke-rv (he turkey-PST-PERF)
‘He was a turkey’) as well as a class of ‘nouns’, which take tense and aspect and also some other
inflections, such as a possessive prefix. What is crucial here, of course, is that both these classes are
distinguished from ‘verbs’ which, as well as taking tense and aspect inflections also take subject
agreement. 19
In (42) above we illustrated the ability of nominal elements to serve as (sentential) predicates,
taking tense and aspect affixes. It is clear that dependent nominals in Sirionó may also take tense
and aspect markers in the presence of a verbal predicate. In (43) the tense marking appears on the
subject nominal and not on the verb; in (44) both tense and aspect appear on the verb; in (45) the
aspect marking appears both on the verb and the indirect argument ‘water’ 20 and finally in (46) the
19 Verbs also differ in permitting two alternative orderings of affixes, differing in whether Tense and Aspect are
suffixes or prefixes, with no consequences for interpretation — see Firestone (1965:27) for details.
20 There is an interesting syntactic restriction at work here. Firestone remarks that indirect objects either precede
or follow the V. When they precede, the aspect marker appears only on the verb (as in (44)); when they follow the
aspectual suffix is attached to the indirect object as in (45).
12

aspect appears only on the subject nominal. These intriguing examples illustrates that tense and
aspect are truly optional inflectional categories of both verbs and nouns.
(43) Ési-ke óso ñá ií-ra.
woman-PST go near water-to(LOC)
‘The woman went near the water.’ (Firestone 1965:37-8)
(44) Áe ií osó-ke-rv.
he water go-PAST-PERF
‘He went to the water.’ (ibid:35)
(45) Áe osó-ke-rv ií-rv.
he go-PAST-PERF water-PERF
‘He went to the water.’ (ibid:35)
(46) E¸ v¸gvtú¸i¸-rv b¸áe b¸u¸kiacáa¸.
tapir-PERF thing steal.not
‘The tapir did not steal from others.’ (ibid:33)
Finally, there is interesting further evidence that the tense and aspect markers distributed across
the verb and nominal elements are acting as a single system.
(47) j¸´v¸kv-ke úke-rv.
tiger-PST sleep-PERF
‘The tiger slept.’
Thus in Sirionó we appear to have an example of a language in which the same TAM markers can
appear either on the verb, or on a dependent NP, or on both.
In Chamicuro, a moribund Arawak language of Peru, propositional tense information is encoded
on the definite article which must usually accompany (definite) nominal subject and object arguments
(Parker 1999). There are two forms of the definite article: na, used in present and future
tenses, and ka which marks past tense. In contrast there is no obligatory tense morphology on
verbs: there is no present tense marker, and the past and future tense markers (-kati and -ye© respectively)
are optional. Thus, in most examples, it is the definite marker alone which signals the
tense information for the clause: 21
(48) a. I-nis-kána
3-see-PL
na
THE(NPST)
čamálo.
bat
‘They see the bat.’ (Parker 1999:552, 2)
b. Y-alíyo ka ké:ni.
3-fall THE(PAST) rain
‘It rained’ (the rain fell). (ibid:552, 3)
21 The facts are complicated by the fact that these articles do not bear stress, and since all lexical words in Chamicuro
must contain two syllables, they are not completely independent phonological items (see Parker 1999:556 and
following). Parker (1999) shows that they encliticize phonologically to C final, but not to V final, preceding words.
This behaviour is completely predictable on purely phonological grounds and he establishes that the definite articles
are structurally part of the NP, even when encliticized to a preceding V.
Parker notates encliticization by use of a morpheme boundary - we have used = instead.
13

U-© c. =na Pámpa Hermosa-ˇsána.
-yé©
1-go-FUT=THE(NPST) Pampa Hermosa-LOC
‘I will go to Pampa Hermosa.’ (ibid:554, 9)
That the tense distinction for the whole proposition is determined by the form of the definite marker
alone is most evident in the following pair of examples in which the present/past tense distinction
is reflected only in the form of the article.
(49) P-aˇskala© a. t-ís=na čamálo.
2-kill-2.PL=THE(NPST) bat
‘You (plural) are killing the bat.’
P-aˇskala© b. t-ís=ka čamálo.
2-kill-2.PL=THE(PAST) bat
‘You (plural) killed the bat.’ (Parker 1999:553, 7,8)
While there are optional verbal tense markers in the language, Parker notes that he has no examples
in which the past tense definite article co-occurs with a verb overtly marked for past tense (Parker
1999:553, fn. 5). Consequently, in all examples in which it occurs, the past tense definite article
is the only exponent of past tense for the clause. Co-occurrence of the non-past article is possible
(although optional) with the future tense verbal marker, as shown in (48c)) above (of course the
non-past definite article does not distinguish present and future).
That these tense-marked elements are indeed definite articles and not part of the verbal complex is
shown by the fact that they appear within NPs: 22
aná© (50) =na čmeˇs� óna
this=THE(NPST) man
‘this man’ (ibid:554, 13)
(51) Y-ahkaˇsamustá-wa
3-scare-1.OBJ
ka ma© póhta
THE(PAST) two
‘The two jaguars scared me.’ (ibid, 14)
ka ma© náli.
THE(PAST) jaguar
Finally, it should be noted that definite articles encoding temporal contrasts occur with all nominals
in the full range of syntactic functions. The examples below show past tense marking by means of
a definite article associated with a nominal in adverbial function.
(52) i-ˇs� ak-kána
3-dance-PL
‘They are dancing.’ (ibid:555, 19)
i-ˇs� (53) ak-kána
likahpé© ka
ta
3-dance-PL THE(PAST) yesterday
‘They danced yesterday.’ (ibid:555, 20)
22 In (51), the definite article appears twice within the NP: once before the numeral and once before the head noun,
as is typical for Chamicuro NPs containing numerals and demonstratives (Parker 1999: 554).
14

In Gurnu, a dialect of the moribund Pama-Nyungan Ba:gandji language of south-eastern Australia,
it is the pronouns and demonstratives which are used to encode clausal tense, showing a three-way
tense distinction between unmarked (and present tense) (marked with initial � ), future (marked
with initial g-), and past (marked with initial w-) (Wurm and Hercus 1976, Hercus 1982). As in
Chamicuro, verbs have no obligatory tense marking, and usually show none at all. The form of
the pronoun/demonstrative alone signals the tense for the clause. The forms of the pronouns and
demonstratives are given in Table 2 (Wurm and Hercus 1976: 40). 23
Table 2: Gurnu Pronouns
UNM PAST FUTURE
1.SG.S � aba waba gaba
1.SG.A � adhu wadhu gadhu
2.SG.S � imba wimba gimba
2.SG.A � indu windu gindu
3.SG.S nadhu wadhu gadhu
3.SG.O � a:nha wa:nha ga:nha
1.DU � ali wali gali
2.DU � uba wuba guba
3.DU � adhu:lu wadhu:lu gadhu:lu
1.PL � ina wina gina
2.PL � urda wurda gurda
3.PL � adi wadi gadi
‘this’ idhi widhi gidhi
‘that’ � a� a wa� a ga� a
Except for in cases of special emphasis, only one pronoun in the clause is marked for tense; the
other pronouns appear in the present/unmarked form. 24 The tense marked pronoun is usually either
the subject or the pronoun referring to the main topic. 25 While (tense-marked) subject pronouns
usually follow the verb (54), it is also possible for them to follow other parts of speech (as in (56)
and the second tensed pronoun in (58) which follows a nominal) or appear clause-initially (59),
showing that the tense marking (at least synchronically) is truly a property of the pronoun itself. 26
23 We have slightly altered the orthographies used by Wurm and Hercus (1976) and Hercus (1982). Specifically, we
have replaced the diacritics marking lamino-dental and retroflex consonsants, marking lamino-dentals with a following
/h/ (e.g. /dh/, /nh/) and retroflexes with a preceding /r/ (e.g. /rn/, /rl/). We also use /rr/ for the apico-alveolar tap/trill
and : to mark a long vowel. In Table 2 we have relabelled Wurm and Hercus’s ‘present tense’ form as ‘unmarked’
(UNM) since this is the form that is used in clauses already containing a tense-marked pronoun.
24 Note that (58) is one example with two tensed forms: a pronoun and a demonstrative. We have no further
information about this sentence; perhaps it is an example of special emphasis.
25 Although note the presence of an object form for third person singular. There are no examples of its use in the
grammatical description. We assume it must be used for object topics.
26 Wurm and Hercus (1976) provide a number of phonological and morphosyntactic arguments supporting this
analysis; details can be found therein. Note that the verbal suffix glossed TOP in (58) is a stem-forming affix which
is called a ‘topicalising’ suffix by the authors because it focuses attention on the aims of an action, making it definite
rather than haphazard (Hercus 1982:191). For example, from the verb bami- ‘to see’, it derives bami-la ‘to look at’. It
therefore has nothing to do with the pragmatic topic of a clause.
15

(54) Barridji-rri dharni gaba.
far.away-ALL go 1.SG.FUT
‘I’ll go a long way off.’ (Wurm and Hercus 1976:40)
� (55) Bami adhu.
see 1.SG.A.PRES
‘I can see.’ (ibid:41)
(56) Wilgawilga ngadi.
hungry 3.PL.PAST
‘They’re hungry.’ (Hercus 1982:124)
dhi� (57) Gila ga-rri waba.
not rise-INC 1.SG.PAST
‘I didn’t get up.’ (Wurm and Hercus 1976:41)
(58) Dhiga-la gadi gi:rra gidhi-nha miri.
return-TOP 3.PL.FUT country this.FUT-GEN towards
‘They’ll go back to their country.’ (ibid:41)
(59) Wadhu ga:ndi barlu-barlu.
3.SG.PAST carry small.child
‘It was him that carried the small children.’ (ibid:42)
In the rare event that a sentence contains no pronominal forms to encode the tense, a verbal suffix
-dji is available to mark past tense on the verb.
� (60) Wi:mbadja urlarrdji dhunga-maldha-dji.
man many bury-REFL-PAST
‘A lot of people buried themselves (in quicksand).’ (Hercus 1982:203, 550)
Tense-inflected subject pronouns are not restricted to one language family: a further example is
Ya� g Dii, a Niger-Congo language spoken in Northern Cameroon. 27 Subject pronouns of the mí
series are inflected according to a future/non-future tense distinction, as shown below in Table 3
(/ separates forms which are in complementary distribution: -n (1S) and -m (2S) are suffixed to
certain preceding vowel-final elements). The discontinuous pronouns (1+2 PL) are separated by
the verb, its pronominal object and even a set of serial verbs. 28
Other TAM categories are expressed with the use of separate particles (called ‘construction markers’
by Bohnhoff (1986)). Crucially, however, these do not encode a future/non-future distinction,
which is encoded only by the tense information associated with the subject pronominals. This is
27 The pronominal system of Ya� g Dii is complex, and only the aspects relevant to our discussion have been presented
here. The reader is referred to Bohnhoff (1986) for a more complete discussion of the pronominal system.
28 The mí set of subject pronouns is used in both perfective and imperfective aspects in factative mood in main
clauses, and is also used in several types of subordinate clauses. A second set of pronouns, the ’ à� `n series, is used in the
hortative mood and in most subordinate clauses. In main clauses pronouns from this latter series have an imperative,
cohortative or jussive meaning and a sense of obligation may be associated with them in dependent clauses. The ’ à� `n
series of pronouns, however, are not inflected for tense, and so will not be discussed further here. Note however, that it
appears as if the distinction between the mí and ’à� `n sets of pronouns may encode a complex type of clause-level mood
information. Other languages in which pronouns more clearly encode mood information for the clause are discussed
below.
16

Table 3: Ya� g Dii Tensed Pronouns
Person, Number mí set mí NFUT mí FUT
1S -n/mí mín míń
1+2 D ba ba`n báń
2S -m/m´ � m´ � n m´ � ń
3S ø ø w´ � ń
1PL v´ � v´ � n v´ � n
1+2PL ba .... ví ba`n .... ví báń .... ví
2PL ví vín víń
3PL v � v � `n v´
illustrated by the following examples. Examples (61) and (62) illustrate the combination of the
future with the perfective and the imperfective respectively. (63) illustrates the combination of the
non-future with the imperfective, this time with a past construction marker in initial position.
Y´� (61) ghó míń s´� ’´� lúú
tomorrow 1SG.FUT leave PERF
‘I (have decided to) leave tomorrow’ (Bohnoff 1986:108, 10)
(62) Míń kaal´£ làà
1.FUT leave town-to
‘I will go to town’ (ibid, 11)
m´� t� (63) Ba n laa la?
PAST 2SG.NFUT go where
‘Where were you (recent past)?’ (ibid:109,12)
Interestingly, the future tense subject pronominal can itself undergo morphological derivation to
give an emphatic form, parallel to the morphological derivation of an emphatic form of the nontensed
mí pronominal. The relevant forms are given in Table 4 below (after Bohnhoff 1986:126,
Table 10). The fact that these tensed pronominals can be input to what is most plausibly analysed
as a nominal morphological process illustrates clearly that this tense inflection is a morphological
property of nominals in this language.
In another Niger-Congo language, Supyire, pronouns alone may encode clause-level mood information.
In Supyire (Carlson 1994), first and second person pronouns have two distinct forms
depending on whether the mood of the clause is declarative or non-declarative. The two sets of
forms are shown in Table 5 below (after Carlson 1994:152, 154). 29
29 Mood distinctions encoded by pronouns are also found in /Gui (Central Khoisan, Botswana), in which subject
pronouns mark imperative mood. Example (2) below illustrates an imperative clause with a first person subject. Note
that the verb remains in the same form in both examples: the imperative mood of the clause is encoded by the subject
pronoun alone (Hitomi Ono, pers. comm.).
(1) Cire
1.SG.NOM
!koõ.
go
� ń
17

Table 4: Emphatic Subject Pronouns
PN mí series mí EMPH mí FUT mí FUT EMPH
1S -n/mí míi míń míńno
1+2 D ba baà báń báńno
2S -m/m´ � móo m´ � ń m´ � ńno
3S ø w� w´ � ń w´ � ńno
1PL v´ � vóo v´ � ń v´ � ńno
1+2PL ba .... ví baà .... ví báń ....ví báńno ....ví
2PL ví víi víń víńno
3PL v� v��� v´ � ń v´ � ńno
Table 5: Supyire Pronouns
DECL. NON-DECL.
1.SG mìi na
2.SG mu ma
1.PL wùu wu
2.PL yìi yi
The declarative set is possible with all sentence types including non-declaratives, and is thus perhaps
better labelled ‘unmarked’. The non-declarative set, however, is only possible in clauses with
non-declarative mood such as commands (65a, 65b) and questions (65c). The distinction between
these two pronoun sets is shown in the following examples. Note that the imperative mood is
signalled only by the choice of the non-declarative pronoun in (65a) and (65b).
(64) a. mìi `� kùù� i
my chicken.DEF
‘my chicken’ (Carlson 1994:152, 1a)
b. Mìi à pa.
I PERF come
‘I have come.’ (ibid:152, 1b)
c. Mu a mìì kánhá.
You PERF me tire
‘You have annoyed me.’ (ibid:152, 2b)
‘I go.’
(2) Da
1.SG.IMP
!koõ.
go
‘Let me go.’
18

(65) a. Ma taha na fyè e!
you.NONDECL follow my.NONDECL footprints in
‘Follow me (lit. in my tracks), please!’ (polite command). (ibid:522, 8a)
b. Na wìì.
me.NONDECL look.at
‘Look at me.’ (imperative) (ibid:154, 7a)
c. Na cevoo kùu, taá ma
`�
my.NONDECL friend chicken where you.NONDECL
‘My friend chicken, where are you going?’ (ibid, 7c)
´� gé k´� ke?
go.IMPV LOC.Q
These examples illustrate both the declarative and non-declarative pronoun sets functioning as
subjects (64b, 64c), objects (64c, 65b) and even as possessive pronouns (64a, 65a, 65c). The fact
that the clausal mood distinction is encoded even when the pronoun is functioning as a possessor
is particularly interesting as it shows that such clause-level information can be embedded within
complex argument NPs.
This mood distinction encoded by pronominals is completely independent of other TAM systems
in the language. Supyire has three other ways of encoding TAM information (Carlson 1994:307ff.).
The most common of these involves a series of auxiliaries which appear between the subject and
the direct object (Supyire is essentially an SOV language). These auxiliaries encode a large variety
of TAM distinctions, including remote and recent past tenses, future tense (66), perfect aspect (67),
prohibitive mood (68) and many others (see Carlson 1994:308 for a full list), and interact with the
non-declarative pronouns as semantically appropriate: the prohibitive mood auxiliary co-occurs
with a non-declarative pronoun in (68).
(66) U ´� sí -kàn.
it FUT FUT-give
‘It will be given.’ (Carlson 1994:335, 55a)
(67) U à tá� pa jáà
s/he PERF come yesterday
‘S/he came yesterday.’ (ibid:338, 63b)
(68) Ma hà a Kàrája
you.NONDECL PROH SUBJ.IMPV Karaja
‘Don’t insult Karaja!’ (ibid:371, 136)
d� cyera à !
insult.IMPV NEG EXCL
Other ways of encoding TAM distinctions include the use of verbal affixes to contrast perfective
(unnmarked) with imperfective aspect (68), and to encode future tense (co-occuring with a future
tense auxiliary) (66). None of these other means of encoding TAM information, however, encodes
the distinction between declarative and non-declarative mood as do the first and second person
pronouns.
Languages which encode clause-level TAM information on subject pronouns, such as we have
seen above for Gurnu, Ya� g Dii, and Supyire, raise an interesting theoretical problem. Namely,
how are we to distinguish between such TAM-inflected pronominals on the one hand and cases of
pronominal incorporation into verbal auxiliaries on the other? That is, for each putative example
19

of, say, tensed pronominals in a language we must rule out the analysis whereby these elements
are not independent pronominals inflected for tense, but are instead themselves pronominal affixes
attached to a verbal (auxiliary) constituent. In some cases, this will amount to establishing that the
putative tensed pronominal is an independent nominal lexical element, rather than simply a verbal
affix which expresses both pronominal and tense features. In other cases, the status of the putative
tensed pronominal as an independent (syntactic) word might not be in doubt, but an analysis under
which it is a verbal auxiliary morphologically incorporating a pronominal element must be ruled
out if it is to be viewed as a true tensed pronominal. These last two possibilities in particular may
be very difficult to distinguish, but at a theoretical level the distinction between a tensed pronominal
and an incorporated pronominal is rather significant. The former involves the encoding of
clause-level TAM information on dependents, while the latter simply involves the incorporation of
a (pronominal) argument into a verbal (head) constituent encoding TAM and is therefore largely
unproblematic for standard theoretical approaches. Likewise in strictly morphological terms, the
coding of (pronominal) argument features on verbal heads is both widely attested and well understood,
while the relevance of clausal temporal properties to the inflectional morphology of nominal
categories (such as nouns, adjectives and determiners) is rather less well-known.
What is needed in each case is a way of establishing the categorial status of the “composite”
element, in order to determine the appropriate analysis. In Gurnu, for example, the fact that these
tense distinctions are found with demonstratives such as widhi ‘this.PAST’ and gidhi ‘this.FUT’
supports their analysis as tense-inflected pronominals, since demonstratives are not usually found
incorporated into verbal auxiliaries cross-linguistically. This analysis is further supported by the
fact that these tense-inflected forms can be further inflected for case, a typical nominal category, as
in (69) (repeated from above), showing them to be NP constituents rather than verbal auxiliaries.
(69) Dhiga-la gadi gi:rra gidhi-nha mi˜ri.
return-TOP 3.PL.FUT country this.FUT-GEN towards
‘They’ll go back to their country.’ (Wurm and Hercus 1976:41)
Similarly, in Supyire this mood distinction amongst pronominals is found even within complex NP
constituents which are not plausibly related to a verbal auxiliary node, as shown by the contrast in
the possessive phrases in (64a) and (65c) above. This fact demonstrates clearly that such moodmarked
pronouns are NP constituents, and that an analysis treating them as verbal auxiliaries would
be untenable.
While the situation may be relatively clear-cut in these cases, there are languages for which the
argumentation required to distinguish the two analyses is much more subtle. Such is the case for
the English non-syllabic reduced auxiliaries such as ’ll in the following:
(70) You’ll [/l/ , */� l/] be leaving soon.
I’ll [/l/ , */� l/] be leaving tomorrow.
The non-syllabic reduced pronunciation occurs only with a pronominal. The correct analysis of
these English auxiliaries has been the subject of some discussion in the recent theoretical literature.
Sadler (1998) (also Nordlinger and Sadler 2000), following Spencer (1991), argues on the basis of
detailed phonological and morphological evidence that they are tense inflections suffixed to subject
pronominals (see also Barron 1998). Thus, on this account you’ll and I’ll in the above examples are
20

tense-inflected pronominals on a par with Gurnu gimba ‘2.SG.FUT’ and gaba ‘1.SG.FUT’. Bender
and Sag (2001), on the other hand, argue that in fact the subject pronominal has incorporated into
the tense-marked auxiliary. On this view, these are not subject NPs at all, but auxiliary heads
carrying both tense and subject information in a clause with no subject NP. Further discussion
of this issue is beyond the scope of the present paper. However, the existence of such tenseinflected
pronominals in the languages discussed above show them to be a typological possibility
irrespective of whether or not they are attested for English. 30
2.3 Summary
The languages discussed in this section all make use of nominal morphology to encode information
about propositional TAM. In many of these languages, the nominal TAM marking operates in
conjunction with verbal TAM, although the distinctions made in the nominal system are often distinct
from those made with verbs (see particularly Pitta Pitta, Kayardild, Chamicuro and Supyire).
In one language, Gurnu, pronominal TAM is the usual way of encoding TAM information about
the clause, with the verb carrying minimal (usually no) TAM information at all. In one language,
Sirionó, the same TAM inflections can be used on verbs and nouns, either together or independently.
3 Non-Propositional TAM
In section 2 we discussed the use of nominals and other NP constituents to encode TAM information
for the proposition as a whole. In this section we turn to another type of nominal TAM inflection,
in which the TAM information encoded by the nominal is completely independent of the TAM of
the proposition. 31 Nominal TAM inflection in this function serves to locate the time at which the
property denoted by the nominal holds of the referent or, in the case of possessive phrases, the time
at which the possessive relation holds.
A straightforward example of independent nominal TAM inflection is provided by Tariana, an
Arawak language from north-west Amazonia, Brazil. 32 Nouns in Tariana can be inflected for
either past or future tense (unmarked nouns are unspecified for tense). 33 The occurrence of tense
morphology on nominals is very widespread, indeed Aikhenvald (to appear) reports that around
40% of nouns in texts are tense-inflected.
There is a single form for nominal future tense, -pena, which specifies that the property denoted
by the nominal holds in the future, for example: wa-� ima¢ i-pena (1PL-son.in.law-FUT) ‘our future
son-in-law’; pi-ya-dapana-pena (2SG-POSS-house-FUT) ‘your future house’. Nominal past tense
has three forms: miki-¢ i for masculine singular nouns, -miki-¢ u for feminine singular nouns, and
30Languages with similar “composite” forms which seem best analysed as auxiliaries with incorporated pronominals
include Hausa (Burquest 1986) and Iai (Tryon 1968).
31Parts of this section draw substantially on Sadler and Nordlinger (2001).
32The data provided is courtesy of Sasha Aikhenvald, and is taken from her forthcoming grammar (Aikhenvald (to
appear)).
33Aikhenvald describes the nominal tense markers as clitics. However, they appear inside oblique case markers,
which she considers to be suffixes (Aikhenvald 1999) and so for our purposes are clearly part of the inflectional
morphology of nouns.
21

-miki for plural nouns. It is used more with animates than inanimates, but possible with both.
Examples include correio-miki-¢ i (post office-PST-NF) ‘old/former post office’; du-sa-do-miki-¢ u
(3SG.NF-spouse-FEM-PST-FEM) ‘his late spouse’.
In contrast, propositional tense is encoded via tense/evidentiality clitics, which attach to the verb
or any other focused constituent (including nominals). The nominal tense system is much simpler
than that of the propositional tense/evidentiality clitics, and the forms are quite distinct from their
propositional counterparts, which are provided in Table 6. table.
Table 6: Propositional Tense/Evidentiality in Tariana
Present Recent Past Remote Past
Visual =naka =ka =na
Nonvisual =mha =mahka =mhana
Inferred N/A =sika =sina
Reported =pida =pidaka =pidana
Examples of the use of these nominal tense markers in regular verbal clauses are given below. The
fact that the temporal reference of the nominal can be independent of that of both the verb and
the clause within which it appears, is illustrated by (72) in which a future tense nominal co-occurs
with a verb carrying the reported remote past clitic: 34
(71) di-ma¢ Thepi
eta-miki-¢ e=pidana
i-nuku.
to.water 3SG.NF-throw.CAUS=REM.P.REP eagle-PST-NF-TOP.NON.A/S
‘He threw the remains of the eagle (lit. the ‘ex-eagle’, what used to be the eagle) into
water.’
wa¢ ipe¢ (72) Kayu-maka hĩ
e unyane-pena di-kakwa=pidana.
so-AFF DEM:ANIM Walipere flood-FUT 3sg.NF-plan=REM.P.REP
‘Thus Walipere was planning the future flood.’
Such data raises the question of how (or even whether) this type of nominal tense marking is
to be distinguished from derivational affixes such as the English ‘ex-’ (e.g. ‘ex-husband’, ‘ex-
President’). The distinction is not at all straightforward – the two types of affixes clearly cover
some of the same semantic ground – but it is possible to identify some significant differences. 35
Firstly, nominal tense markers are inflectional rather than derivational affixes; in some languages
they even form portmanteaux with other inflectional nominal categories such as possession and
definiteness (see the discussions of Hixkaryana and Somali below, respectively). Secondly, due
to their inflectional nature, they are fully productive. The suffix ‘ex-’ in English, on the other
hand, is quite restricted in its semantics. It is most common with nouns denoting occupations
(‘ex-President’, ‘ex-director’, ‘ex-teacher’) and non-kin relationships (‘ex-wife’, ‘ex-boyfriend’).
34 Non-obvious abbreviations in these examples include: AFF ‘affix’, CL.ANIM ‘animate classifier’, DEM.ANIM
‘demonstrative animate’, NF ‘non-feminine’, PAUS ‘pausal’, PRES.VIS ‘present visible’, REL ‘relative’, REM.P.REP
‘remote past reported’, TOP.ADV ‘topic advancement’, TOP.NON.A/S ‘topical non-subject clitic’.
35 We are grateful to Matthew Dryer for discussion of some of these issues.
22

It is substantially less appropriate with common nouns such as ‘dog’ and ‘house’ (?ex-dog, ?exhouse).
Such restrictions are not found in the true nominal tense examples, as can be verified in
the examples throughout this section.
As mentioned above, propositional tense/evidentiality clitics can appear on any focused constituent
in the clause. Naturally then, it is possible for these clitics to attach to nominals, along with independent
nominal tense marking as illustrated above. This is shown in the following example in
which the final word ‘person’ is inflected with nominal past tense as well as carrying the propositional
tense clitic (marking ‘present non-visual’).
(73) Kayu-maka
so-AFF
diha
he
nawiki-nha ñamu na-nite
person-PAUS evil.spirit 3PL.say-TOP.ADV+CL:ANIM
i=mha.
nawiki-miki-¢
person-PST-NF=PRES.NONVIS
‘So this man called evil spirit ñamu, he is the one who used to be a person (lit. he is an
‘ex-person’).’
The contrast between the two nominal tense markers is highlighted in the following pair of nominals,
both also carrying the ‘present visual’ propositional tense clitic:
(74) Pi-ya-dapana-pena=naka.
2SG-POSS-house-FUT=PRES.VIS
‘This is your future house (I can see it).’
(75) Pi-ya-dapana-miki-¢ i=naka.
2SG-POSS-house-PST-NF=PRES.VIS
‘This is what used to be your house (I can see it).’
In possessive examples such as these above, there are two semantic predicates with respect to
which the tense marker may logically be interpreted. One possibility is that the tense marker
temporally locates the nominal referent itself (e.g. ‘former/future house’). Another possibility is
that the tense marker refers not to the nominal, but rather provides the time at which the possessive
relation holds (e.g. ‘formerly/to be possessed’). In the large majority of languages with this type of
non-propositional nominal TAM marking, such examples are ambiguous between the two logically
possible readings. This is true for the Tariana examples above: 74, for example, can mean either
‘this is your thing which will be a house’, in which the tense marker is interpreted with respect to
the nominal itself, or ‘this is a house which will be yours’, in which it refers to the time at which
the possessive relation holds. Below we will see many more such ambiguous examples in other
languages with non-propositional nominal TAM.
A similar nominal tense constrast to Tariana is found in Guaraní, a Tupí-Guaraní language which is
widely spoken in Paraguay (Guasch 1956, Gregores and Suárez 1967, de Canese 1983, see also the
brief discussion in Mel’čuk 1994:54). Guaraní has the nominal tense suffixes -kwé PST (sometimes
-ré), � � ra FUT and -rangue for a future which is not to be realized (that is, an irrealis future), which
occur on nominal arguments and on nominalisations. Examples include:
(76) h-óga-kwé
his-house-PST
23

‘his former house’ (Gregores and Suárez 1967:127))
(77) h-emi-.apò-r �
� a
his-work-FUT
‘his future work’ (ibid)
(78) aò-apò-hà-ré
maker of clothes-PST
‘the one who made the clothes’ (ibid)
As with the Tariana examples discussed above, the possessive examples in (76) and (77) are ambiguous:
(76), for example, can mean either ‘my thing which used to be a house (e.g. it has burned
down)’ or ‘the house which used to be mine (but now belongs to somebody else)’.
The independence of the nominal tense from the TAM of the whole proposition is illustrated by
the following examples, in which the propositional and nominal tense vary independently of each
other. 36
(79) O-va-ta che-roga-kue-pe.
3-move-FUT 1SG-house-PST-in
‘He will move into my former house.’
(80) A-va-vaekue hoga-rã-pe.
1SG-move-PST 3.house-FUT-in
‘I have moved into his future house.’
In the following interesting example there are two tense inflections on the nominal, the first nonpropositional
and the second propositional. The final tense suffix is from the verbal tense series
which includes six different past markers akue, mi, kuri, va’ekue, raka’a and ra’e and five distinct
future markers ta, pota/mbota, va’erã, ne and arã.
(81) Che-roga-rã-ta.
1SG-house-FUT-FUT
‘It is my future house, it will be my future house.’
(81) is therefore an example of a type of ‘tense-stacking’, in which a single nominal is inflected
with two tense markers. In this example, the two have quite different functions: the inner tense
marker encodes non-propositional tense, temporally locating the referent of the nominal itself (i.e.
‘future house’). The second tense marker, on the other hand, marks the clausal tense which, in this
case, is also future (although could potentially have been something else). As noted above, this
example involves the use of what is otherwise described as a verbal tense marker on a nominal
predicate, in addition to the nominal future marker for a nominal dependent (non-propositional
tense). Future research is needed to determine just how widespread this phenomenon of tensestacking
is.
36 The following Guaraní examples (including (81)) were kindly collected for us by Dagmar Jung from Sebastiana
Ertel, a native speaker of Guarani from Asuncion now resident in Cologne, Germany. The examples from Gregores
and Suárez ((76) - (78)) are written in what was at the time standard Guaraní orthography while the informant data is
represented in the current standard orthography.
24

In the Guaraníexample, the two stacked tense markers have different functions. A further interesting
question is whether or not it is possible to find tense-stacking in which the two tense markers
have the same nominal TAM function – i.e. when they both encode non-propositional tense. Thus,
do we find morphologically encoded equivalents of English complex phrases such as ‘future exhusband’
or ‘former future President’? The descriptions of the languages discussed above provide
no suggestion that such tense-stacking is possible in these languages. However, it may be possible
in (at least) one Tupí-Guaranílanguage, Tupinamba, which is mentioned briefly by Lehmann
and Moravcsik (2000). The nominal TAM system in Tupinamba appears largely similar to that of
Guaraní, as described above. Interestingly, however, Tupinamba is reported to have no TAM marking
on verbs, showing that nominal TAM inflection can exist in a language independently of verbal
TAM inflection. Lehmann and Moravcsik (2000:742) provide the following examples showing the
use of past and future tense with nominals: rók-a ‘house’, rók-wér-a ‘former house’, rók-wám-a
‘future house’. In addition they provide one example which appears to contain two tense markers,
future and past : rók-ám-wér-a ‘what was to be a house, ex-future house’. The translation
implies that the outer suffix (past) has scope over the inner tense suffix (future), that is, the property
of being a future house is temporally situated at some point in the past. Thus it appears as if
tense-stacking of non-propositional tense markers may be possible, albeit rare. Further research is
clearly needed.
Halkomelem (Salish) (Burton 1997, Gerdts 1988, Galloway 1993) is a further language in which
non-propositional tense is marked on nominal elements. As in Tariana and Guaraní, nominals
can be inflected with one of two tense markers: future tense, expressing ‘will be’, and past tense
encoding meanings such as ‘former, ex-, late (dead)’. In contrast to these languages, however, the
set of affixes used to mark non-propositional tense on nominals are the same as those used for
marking propositional tense with verbal predicates, a point to which we return below. 37
The following examples are from Burton (1997:67), who discusses the various interpretations
available for past tense nominals: 38
(82) tel mál
my father
(83) tel má:l-elh
my father-PST
‘my late father’
(84) te sqwemá:y
the dog
(85) te sqwemá:y-elh
the dog-PST
‘the dead dog’
The ‘deceased’ reading exemplified above occurs only with animate nouns. When the noun refers
to a non-cancellable (or ‘lifetime’) property such as ‘father’ or ‘dog’ (i.e. one cannot cease to be a
father or a dog without ceasing to exist) then past tense animate nouns always have the deceased
reading. When the noun refers to a cancellable property, however, Burton notes that an alternative
reading of ‘former, ex-’ is possible (Burton 1997:74):
37 The same is reported for some other native American languages such as Kwakw’ala (Anderson 1985) and
Potawatomi, from which Hockett (1958:238) gives the following examples: nk� ˇsat� s ‘I am happy (verb)’, n čiman
‘my canoe (noun)’ vs. nk� ˇsats� p� n ‘I was formerly happy (but not now)’, n čimanp� n ‘my former canoe, now lost,
destroyed, or stolen’.
38 Some of the interlinear glosses have been simplified for ease of exposition. The reader is referred to Galloway
(1993) for further details.
25

(86) stó:les-elh
wife-PST
‘dead wife, ex-wife’
(87) siyó:ye-lh
friend-PST
‘dead friend, former friend’
When the past tense marker is used with a possessed inanimate noun, it exhibits the ambiguity we
have seen above for other languages with this type of nominal tense system. Thus, the past tense
marker can either mark the possessed item as being destroyed or lost (i.e. temporally locating the
nominal is in all of the examples above), or it may mark the possessive relation as holding in the
past (Burton 1997:67-68):
(88) tel xeltel-elh
my pencil-PST
‘my former pencil, used to be my pencil,
my destroyed pencil’
(89) tel pukw-elh
my book-PST
‘my former book, used to be my book,
my destroyed book’
In (90) the suffix -elh is attached to a pre-verbal auxiliary and marks clausal past tense, thus illustrating
the fact that the same tense affix can attach indiscriminately to nouns and (auxiliary) verbs.
Clausal tense marking is not obligatory in Halkomelem; it is possible for the verb and auxiliary to
remain uninflected for tense, in which case the tense of the clause is determined by contextual and
pragmatic considerations (Burton 1997:68), as exemplified in (91) below.
(90) i-lh imex tel sí:le.
AUX-PST walk my grandfather
‘My grandfather walked.’
A number of influential papers have argued that the Salish languages lack a distinction between the
lexical categories, including the basic distinction between noun and verb (see e.g. Kinkade 1983,
Jelinek and Demers 1994, Julinek 1996). Clearly, if this were the case, the use of TAM markers
with words such as mál ‘dog’ would be no less expected than with words such as imex ‘walk’,
and it would therefore be misleading to include it here as an example of nominal TAM inflection.
However, this analysis is argued for mainly on the basis of data from Straits Salish and Upper
Chechali (and to a lesser extent Lillooet (St’át’imcets), Lushootseed and Twana). The position is
refuted for St’át’imcets by Demirdache and Matthewson (1995) and Matthewson and Demirdache
(1995) and for Halkomelem by Galloway (1993). Galloway argues that a distinction between nouns
and verbs can be made in Halkomelem on both morphological and syntactic grounds: for example,
affixes encoding possession and diminutives are possible only with nouns (verbs must first be
nominalised with a prefix such as s- or sw� -); while only verbs may be inflected with subject and
object pronominal affixes and valency-changing morphs (Galloway 1993:238). Furthermore only
nouns may be preceded by the demonstrative article (once again, verbs must be nominalised first),
and in simple sentences the usual order is for verbs to precede nouns. These arguments against
the category neutral view of Salish languages, suggest that the Halkomelem examples discussed
in this section are indeed cases of tense affixes occurring on nominals to encode non-propositional
tense.
Despite the fact that they are encoded with the same morphological marker, propositional tense and
26

nominal tense are completely independent categories in Halkomelem, and can vary independently
of each other. In (91) and (92) a past-tense inflected nominal (with the deceased reading) co-occurs
with clausal past time reference (non-overt in (91) and overt in (92)). But in (93) a past-tense
nominal occurs in a future tense clause, indicating that nominal and clausal tense marking are
distinct. Such examples establish that tense marking on nominals in Halkomelem is semantically
distinct from propositional tense, otherwise we would expect such cases of conflict in tense values
to result in ungrammaticality.
(91) kw’étlexwes tel má:l-elh te sqwemá:y.
see my father-PST the dog
‘My late father saw the dog.’ (Burton 1997:68)
(92) éwe-lh kw’étslexw the-l sí:l-á:-lh.
NEG.be-PST see the(f)-my grandparent-PST
‘He didn’t see his late grandmother.’ (Brent Galloway, p.c.)
(93) El-’éliyemet-tsel-cha
the-l sí:l-á:-lh.
REDUP-dream.about-1SG.SUBJ-FUT the(f)-my grandparent-PST
‘ I’ll be dreaming about my late grandmother.’ (Brent Galloway, p.c.)
The role of future tense marking on nominals is rather different. In combination with the future
suffix -cha, which is also used to mark future tense with verbs (see (93) above), the noun functions
as a (future) stative predicate. In these cases, the tensed nominal constitutes the clausal predicate,
and thus what we have here is a case of tense inflection of a nominal predicate (see section 2.1). 39
(94) Swíyeqe-cha
man-FUT
‘It will be a man.’
(95) Swíyeqe-cha
man-FUT
kw’-a’-s hákw’eles
the-your-NMZR remember
‘It will be a man that you remember.’
In the languages discussed above, we’ve seen examples in which the interpretation of the nominal
tense inflection is ambiguous between the temporal location of the nominal referent itself,
or the temporal location of the possessive relation of which the nominal is the object. In some
languages this latter use has been grammaticised such that it is the only interpretation possible
for nominal tense inflection. This is the case for many Carib languages, including Hixkaryana
(Derbyshire 1979, Derbyshire 1999). 40
In Hixkaryana independent nominal tense is expressed with a series of portmanteau nominal suffixes
which mark present, past and remote past possession, and depossession. 41 The possessor is
coded by a prefix which expresses person, and which co-occurs with the suffixal possession/tense
markers. Here the tense marker serves to temporally locate the possessive relation, rather than the
39These examples are from Brent Galloway, (p.c.). The suffix -s in (95) functions to nominalize the following phrase
so that it can be relativized.
40Similar facts to the Hixkaryana ones discussed here pertain to the other Carib languages including Apalai (Koehn
and Koehn 1986), Macushi (Abbott 1991), Wai Wai, Carib, Dekwana, Trio, and Wayana (Derbyshire 1999).
41According to Derbyshire (1999) the depossession suffix appears (although rarely) on inalienably possessed nouns
to indicate more general reference.
27

property denoted by the nominal, so that ro-kanawa-r£ ‘1-canoe-POSS.PAST’ means ‘canoe which
used to be mine’, rather than ‘my thing which used to be a canoe’. The possessor/tense suffixes
are tabulated in Table 7. The forms in parentheses are phonologically conditioned allomorphs: the
reader is referred to Derbyshire (1979) for details. 42
Table 7: Hixkaryana Possessor/Tense Suffixes
Features Form
Poss -��� , -n� , -t� , -t� e, -¤
Past Poss -th� r� (t��� r� ), -tho (-t� ho), nh� r� ,-nho
Remote Poss nh� r�
Depossession -nano
The following examples are provided by Derbyshire (1979:98-99). 43
¢ o-kanawa-¢�£
(96)
1-canoe-POSSD
my canoe
(97) ¢ o-kanawa-tho
1-canoe-POSSD.PST
‘the canoe that used to be mine’
(100) £ -he-t� e
3-wife-POSSD
his wife
£ -he-t� e-nh£ r£
(101)
3-wife-POSSD-PST
his former wife
ow-ot-t£ (owot£ (98)
2-meat-POSSD
your meat
)
ow-wo-t£ -th£�¢�£
(99)
2-meat-POSSD-PST
‘that meat which used to be yours’
(102) ¢ o-katxho-ø
1-things-POSSD
my things
£ -katxho-ø-th£ r£
(103)
3-things-POSSD-PST
his old things
Derbyshire (1979) reports that the forms -nh£ r£ and -nho also inflect (some) non-possessed words,
and gives the examples waha-nho “one who had been a killer”, and toto-tho-nh£ r£ (human-DEV-
PST) “one who had been a human being” (Derbyshire 1979:99). 44
42 The choice of basic possessive suffix (expressing present possession) is lexically conditioned, with most nouns
taking -��� , and the alternative forms occurring with few stems. The distribution of the past possessive suffixes depends
on the type of the possessor: -tho and -nho occur with the prefixes for 1, 1+3 and 3 with preceding overt NP possessor,
and -th� r� and -nh� r� with the remaining person-marking prefixes. -nho and -nh� r� in the past tense meaning occur only
with a small set of lexically specified stems.
There is some slight uncertaintly in Derbyshire’s description concerning the remote past reading of -nho since he
omits it from his initial listing of forms (Derbyshire 1979:98), but later refers explicitly to its remote past reading
(page 99), and see also (Derbyshire 1999). We take it that this form does indeed have a remote past reading.
43 In the examples in this section, some changes have been made in the glosses in order to increase consistency
across the languages discussed. Non-obvious abbreviations in these examples include DEV ‘devalued’, REM ‘remote’.
Note that the simple past suffix normally follows all allomorphs of the possessor suffix except r� , which it replaces.
When this occurs we gloss it as POSS.PST.
44 This example contains the suffix -tho glossed as a devalued marker (DEV). This modifying element can co-occur
28

Past and remote possession marking is also found on deverbal nominals, where the person prefix
(possessor morphology) codes one of the core arguments (either subject or object), and is very
often obligatory. The following action and result nominalizations illustrate: 45
ro-to-th£ r£ -nh£ r£
(104)
1-go-PST-REM
my going long ago (Derbyshire 1979:99)
e-karyma-t� h£ r£ -nh£ r£
(105)
3-tell-PST-REM
its telling long ago (ibid:99)
o-n-menho–th£ r£ -nh£ r£
(106)
2-OBJ NOMZ-write-PST-REM
the thing you wrote long ago’ (ibid:99)
The following examples show nominal and verbal tense marking varying independently – present
possession co-occuring with verbal past tense. 46
ro-kanawa-r£
(107) mar-yako
£ uro ryhe ro ha rokanawar£ .
1-canoe-POSS 2S.3O.take-REC.PST.COMPL mine EMPH that INT 1-canoe-POSS
‘You took my canoe, the canoe that belongs to me’ (ibid:129, 288)
y-ow-t£ £ (108) Waraka
yakaro -to-no.
Waraka 3-brother-POSS with 1S.3O-go-IMM.PST
‘I went with Waraka’s brother’ (ibid:11, 22b)
In all of the languages with non-propositional nominal TAM systems that we have discussed above,
the only TAM category encoded is the category of tense. The category of aspect appears to be
one which can only hold of the proposition; there are no languages in which aspect is marked
on nominals with non-propositional function. It is possible, however, for languages to encode
(some) non-propositional mood information on nominals. We have found two such languages, one
which marks possibility, and the other evidentiality. Crucially, however, these languages also mark
nominals for non-propositional tense.
Iate is a Macro-Jê language spoken in the vicinity of Pernambuco (Brazil). According to Lapenda
(1968) nouns in Iate can be inflected for one of three tenses (past, present and future) and one of
two moods (‘realis’ and ‘possible’). The realis mood (which is unmarked) has three tenses (with
the present also unmarked), and the possible mood has just two: present and past. The different
possibilities and their meanings are given in Table 8, using the noun seti ‘house’. Note that the
possible mood forms involve taking the realis equivalent and adding the suffix -këá (Lapenda
1968:77). 47
with past marking, indicating that it is distinct (although homophonous).
45 Derbyshire notes that (104) may have a relative reading in which the going is previous to some other action
marked with -tho as in ro-to-tho 1-go-POSS.PST.
46 The system of verbal tense marking in Hixkaryana is considerably complex, see Derbyshire (1979: 136) for
details.
47 The past tense morph in the possible past is is a regularly conditioned allomorph of the past tense suffix -s ê.
29

Table 8: Nominal tense and mood suffixes in Iate
realis
pres seti that which is a house or serving as a house
past se‘ti-sê that which was once a house;
that which stopped being a house
fut sêti-he future house, will be a house;
house which is being built
possible
pres se‘t-këá a possible house; something which has the
possibility of being a house
past se‘ti-s -këá something which would have been a house but wasn’t;
something which had the possibility of being a house
Non-propositional evidentiality (in combination with tense) on nominals is reported by Lowe
(1999) for Nambiquara, a small family of dialects from the Northern Mato Grosso, Brazil. Nouns
in Nambiquara are optionally suffixed for definiteness, and definite nouns may be further inflected
for tense and evidentiality. 48 The relevant suffixes found on Nambiquara definite nouns are listed
in Table 9 (after Lowe 1999:282), and examples of the use of these affixes are given below. 49
Table 9: Nambiquara Definite Nominal Endings
definite, unmarked
-a�
na� definite, current
-ai�
ti� observational,recent past, given
-in�
ta� li� oservational, mid past, given
-ait�
tã� observational, mid past, new
-ait�
tã inferential, definite, unmarked
-nũ�
tai� na� inferential, current
-nũ�
t˜e� � tã� quotative, mid past, given
-au�
wa� lin� -su� -n� ti�
(109)
manioc-CL:BONE-LIKE-OBSERVATIONAL.RECENT.PAST.GIVEN
‘This manioc root that both you and I saw recently.’ (p. 282, ex. 32)
wa� lin� -su� -nũ� tã�
(110)
manioc-CL:BONE-LIKE-INFERENTIAL.DEF.UNMARKED
‘The manioc root that must have been at some time past, as inferred by me (but not by
you).’ (p. 282, ex. 35)
48Although Lowe notes that only a limited number of tense and evidentiality combinations are attested in his (very
large) corpus (1999:282).
49The superscripted numbers are tone markers.
30

In Nambiquara indicative main verbs in the non-future are obligatorily inflected for evidentiality
and newness. The verbal forms are distinct from the nominal forms, however, as shown by Table
10 (after Lowe (1999:275)).
Table 10: Nambiquara new-information verbal suffixes
imperfective aspect
-wa�
perfective aspect
-ra�
— -nĩn� inferential
-nũ�
observed circumstances
ta�
observed action
-na�
quotative
-ta�
mid past tense external
-h˜e�
present tense internal support
-nha�
recent past tense internal support
-h˜e�
mid past tense internal support
-h˜e�
-na� action currently observed by both speaker and hearer
The nominal TAM systems of Iate and Nambiquara demonstrate that the category of mood may be
non-propositional, modifying a nominal independently of the mood of the proposition as a whole.
However, these languages exhibit only a small subset of the range of moods associated with verbal
constituents cross-linguistically. An important question for further research is whether there are
any restrictions on the moods that can function non-propositionally, and what these restrictions
may be.
A particularly intricate example of non-propositional tense is found in the Cushitic language of
Somali, in which it interacts with the nominal category of definiteness. There is extensive discussion
of this phenomenon in Lecarme (1996, 1999), which we draw on substantially in this section.
In Somali, definite determiners (which are nominal affixes) can encode past tense on the heads of
noun phrases (non-past determiners are unmarked for tense). Affixal determiners involve an initial
consonant: k (and its allomorphs) with masculine stems, and t (and its allomorphs) with feminine
stems. Non-past determiners also distinguish nominative and non-nominative case. The forms of
the determiners are shown in Table 11 (Lecarme 1999: 335).
Table 11: Somali Definite Articles
Case NON-PAST PAST
NOM -ku/-tu -kii/-tii
NON-NOM -ka/-ta -kii/-tii
The tensed determiners shown in the table above are in paradigmatic opposition with a separate
deictic system involving near/far demonstratives, which do not have a temporal interpretation (see
Table 12, after Lecarme 1999: 335).
31

Table 12: Somali Demonstratives
proximal remote
-k/t- -án ‘this’ -áas ‘that’
-k/t- éer ‘that (far away)’ -óo ‘that (very far away)’
Somali also has a set of possessive determiners which are suffixed to nominal heads, and undergo
a set of sandhi rules similar to those affecting definite determiners. These also show past/nonpast
distinctions: gúri ‘your house’ gúrigàagii ‘your house.PST’ (Saeed 1999:115), with the meaning
‘your former house’. 50
The following examples illustrate the basic system. In (111) the subject noun is unmarked for
tense, leading to the reading that it still exists. In contrast, in (112) the head noun is inflected with
a past tense determiner, which locates the nominal reference in the past.
(111) dhibaatá-da Khalíij-ku welí way
problem-DET.F Gulf-DET.M.NOM still
‘The crisis of the Gulf still persists’
FOC.3S
taagán tahay.
permanent is
(112) dhibaatá-dii Khalíij-ku wáy dhammaatay.
problem-DET.F.PST Gulf-DET.M.NOM FOC.3S end.PST
‘The (past) crisis of the Gulf ended’ (Lecarme 1999:335)
Lecarme also provides some examples of the choice of determiner leading to specific presuppositions:
(113) is only appropriate if the speaker believes the exhibition is closed at utterance time,
and (114) only appropriate if the speaker believes it is open. Likewise, the choice of determiner
in (115) is determined by whether the journey is still in progress or not. These examples also
demonstrate the independence of the nominal and propositional tense systems. In (113) and (114),
for example, the propositional tense (as marked on the verb) remains the same while the nominal
tense varies to signal the change in meaning.
(113) bandhíg-gii máad daawatay?
exhibition-DET.M.PST Q.2S see.PST
Have you seen the exhibition (closed at UT) (Lecarme 1999:338)
(114) bandhíg-ga máad daawatay?
exhibition-DET.M Q.2S see.PST
Have you seen the exhibition (still running at UT) (ibid:338)
(115) búugganu sáfarkayga/-ii
buu tilmáanayaa
book.DET.M.NOM journey.DET.M.POSS1S./PST FOC.3S relates
This book relates to my journey (Lecarme 1996:7)
50 It is not clear from Saeed’s description whether these constructions have the same ambiguity–i.e. between ‘house
which used to be mine’ and ‘my thing which used to be a house’–discussed above for many other languages with
non-propositional nominal tense.
32

Nominal tense is implicated in the syntax of Somali in several intriguing ways. One of these
concerns tense agreement within the noun phrase. Adjectives used attributively also inflect for
tense, sharing the tense endings of the highly irregular verb ‘be’ (ø -PST, -aa PST.M, -ayd PST.F).
Adjectives agree in gender and tense, with adjectival tense only appearing when the noun is definite
(and therefore tensed) (Lecarme 1996:4, Lecarme 1999:343). 51 The following examples show
gender and tense agreement with masculine singular, feminine singular and feminine plural nouns
respectively. This tense concord between nouns and adjectives clearly demonstrates that the tense
marking is an inflectional property of nouns in Somali.
(116) árday-ga wanaagsan
student-DET.M good
(117) árday-gii wanaagsan-aa
student-DET.M.PST good-PST
the good student
(120) ardáy-da wan-wanaagsan
students-DET.F PL-good
the good students
(118) ardayád-da wanaagsan
student(f)-DET.F good
(119) ardayád-dii wanaagdan-ayd
student(f)-DET.F.PST good-PST.F
the good student
33
(121) ardáy-dii wan-wanaagsan-aa
students-DET.F.PST PL-good-PST
the good students
Such tense concord between nouns and adjectives is not found in the other languages discussed
above with non-propositional nominal tense systems.
The past tense marking on Somali determiners interacts with the discourse in ways that extend
what we have reported for other languages in the discussion so far, and which raise many open
questions for future research. In (123), for example, the past tense marked noun does not have
the interpretation ‘ex-students’, but rather is used anaphorically to refer to a past time already
mentioned in the discourse and taken as the reference point (Lecarme 1999).
(122) ardáy-da baan kasin su’áash-aadii.
students-DET.F FOC.NEG understand.PST question-DET.F.POSS2S.PST
‘The students (who are present/I am telling you about) did not understand your question.’
(123) ardáy-dii wáy joogaan.
students-DET.F.PST FOC.3P are present.NPST
‘The students (I told you about) are present.’ (ibid:335)
The discourse interaction of nominal TAM inflection in Somali is presumably linked with its association
with definite determiners, which have, by definition, a strong discourse function and are
frequently anaphoric (Givón 2001). 52 That these determiners in Somali retain primarily a temporal
function is clear from their interaction with overt temporal modifiers, which must occur with
51 Number may be marked through optional reduplication in Somali adjectives.
52 A full discussion of the semantic relationship between nominal tense and determiners is beyond the scope of this
paper, but it is a topic in need of much further research. Other languages which show a type of association between
nominal determiners and tense marking include Mao Naga (Tibeto-Burman), in which nominal suffixes marking spatial
deixis can also be used to encode certain nominal tense distinctions (Giridhar 1994 pp. 118-9); Iraqw (Cushitic), in
which determiners encode tense distinctions when used discourse-anaphorically (Mous 1993:90); and Jingulu (non-
Pama-Nyungan) in which clausal tense markers have grammaticalised into nominal suffixes encoding spatial deixis

a matching tense marking. In the examples below, the temporal modifier ‘next year’ selects a
non-past determiner, while ‘last year’ selects a past determiner.
(124) sánnad-ka/*-kii dambe
year-DET.M
next year
next
(125) sánnad-kii/*-ka hore
year-DET.M.PST before
last year (Lecarme 1999:342)
Furthermore, the addition of a temporal modifier such as hore ‘before’ may unambiguously fix the
event or the reference time. Consider (126) in contrast to (123).
(126) ardáyday-dii
hore dhammáan-t-ood (waa ilá soo xariiran).
students-DET.F.POSS1S.PST before entirety-DET.F-POSS3P (are in contact with me)
‘All my ex-students (are in contact with me).’ (Lecarme 1999:342)
In (126), ardáyday-dii excludes the possibility of the individuals still being students at the time of
utterance (that is, the nominal predication or ‘event’ itself is temporally restricted).
This section has explored the use of nominal morphology to temporally locate the referent of the
nominal predicate, independent of the temporal location of the clausal predicate. As we have
shown, this is quite a widespread phenomenon. Distinctions in tense are found in all of the languages
we have discussed, and in a couple of languages we find distinctions in mood also (Iate and
Nambiquara). Interestingly, in no language have we found examples of non-propositional aspect.
We have encountered several cases which combine the inflectional expression of possession and
tense, and these cases raise interesting issues concerning precisely what is temporally located (the
nominal referent or the relation of possession itself). We have also found cases of tense stacking
on nominals in Guaraníand Tupinamba: the former involving the stacking of propositional tense
outside non-propositional tense on a nominal predicate, and the latter involving the stacking of two
non-propositional tense markers. These instances of tense stacking are particularly intriguing; it is
hoped more such instances can be revealed through future research on the issue.
4 Discussion
In the preceding sections we have illustrated the range of nominal TAM systems found amongst
the world’s languages. In doing so, our intention has been to provide enough information for
each individual language to give the reader some sense of the interaction of the nominal TAM
inflection with the rest of the grammatical system. In this section we discuss some of the issues
and generalisations that can be drawn from from this empirical discussion.
(Pensalfini 1997, 2002). Of course, since the deictic and anaphoric functions of tense and definiteness are rather
similar, finding a direct relationship between tense and the determiner system is not surprising, but we leave further
discussion of this issue for future research.
34

As we have seen, there are essentially two broad functional types of nominal TAM inflection: that
which encodes the TAM value for the clause (‘propositional’); and that which temporally situates
the dependent nominal independently of the clausal TAM (‘non-propositional’). These two types
have different properties, and interact quite differently with the rest of the grammatical system.
Within the class of propositional nominal TAM inflection we distinguish TAM inflection of dependent
nominals (section 2.2) from TAM inflection of predicate nominals (section 2.1). The latter
type is the nominal equivalent of standard tense inflection on verbs. TAM inflection on nominal
predicates, however, also shares characteristics with non-propositional TAM inflection (section 3)
in that it functions to temporally situate the nominal to which it is attached. Consider the following
examples from Bininj Gun-wok and Tariana respectively (repeated from above).
(127) Mayh na-mekke nakka bininj-ni.
bird MASC-DEM MASC.DEM human-PAST
‘Those birds, they were human then,’
(128) di-ma¢ Thepi
eta-miki-¢ e=pidana
i-nuku.
to.water 3SG.NF-throw.CAUS=REM.P.REP eagle-PST-NF-TOP.NON.A/S
‘He threw the remains of the eagle (lit. the ‘ex-eagle’, what used to be the eagle) into
water.’
In example (127) the tense-inflected nominal is the predicate of a verbless clause, and thus constitutes
an example of propositional TAM inflection. In (128) on the other hand, the tense-inflected
nominal is a dependent of a verb-headed clause, and this then constitutes an instance of nonpropositional
TAM inflection. Semantically, however, the two past tense markers have identical
functions. In each case, the tense marker specifies that the property denoted by the nominal holds
(of the nominal referent) in the past.
In spite of this similarity of semantic function, the division into two distinct types of nominal
TAM inflection is justified on a number of grounds. Firstly, there are languages which allow only
one and not the other. Bininj Gun-wok, for example, allows TAM inflection on nominals only
when they function as clausal predicates. And the nominal tense system in Tariana functions
only independently of clausal tense — that is, it cannot be used with nominals used as clausal
predicates to encode propositional tense. This is evidenced by the fact that nominal predicates
take propositional TAM clitics (totally distinct from the nominal TAM system, see section 3 for
discussion) to encode propositional tense, which can co-occur with non-propositional nominal
TAM inflections. Consider the following Tariana examples repeated from above.
(129) Pi-ya-dapana-pena=naka.
2SG-POSS-house-FUT=PRES.VIS
‘This is your future house (I can see it).’
(130) Pi-ya-dapana-miki-¢ i=naka.
2SG-POSS-house-PST-NF=PRES.VIS
‘This is what used to be your house (I can see it).’
These factors demonstrate that the encoding of propositional TAM on nominal predicates is logically
independent of the inflection of (dependent) nominals for non-propositional tense, and support
the categorization presented above.
35

The encoding of propositional TAM on nominal predicates is perhaps not surprising given that these
elements function as the predicate head of the clause and, as such, are functionally equivalent to
verbs in regular verbal clauses. More surprising, however, are the number of languages which
mark propositional TAM categories on clausal dependents (section 2.2). The unexpected nature
of this phenomenon, given standard assumptions about the properties of TAM inflection crosslinguistically,
is illustrated by Lehmann and Moravcsik (2000) who, in their discussion of tense
marking on nouns, claim that
“[w]hile there may be agreement between a nominal dependent and its verb in other
categories, tense is not an agreement category. Even where both the noun and the verb
have tense, tense is selected independently for a verb and its nominal dependents, as
in My ex-wife is visiting me, my future wife visited me, etc.” (p. 742)
Thus, Lehmann and Moravcsik (2000) explicitly rule out the possibility that tense marking on
nominals may encode propositional tense along with (or even in place of) tense marking on the
verb. But in fact, this is just what we find in the languages discussed in section 2.2 above, and
illustrated by the following examples from Lardil.
(131) Ngada niween maarn-in wu-tha.
1SG.NOM 3SG.OBJ spear-OBJ give-GNF
‘I gave him a spear’ (Klokeid 1976:476, 56a)
(132) Ngada bilaa wu-thur ngimbenthar diin-kur wangalk-ur.
1SG.NOM tomorrow give-FUT 2SG.FOBJ this-FOBJ boomerang-FOBJ
‘I’ll give you this boomerang tomorrow.’ (ibid:493)
(133) Ngada niwentharr maarn-arr wu-tharr.
1SG.NOM 3SG.NFOBJ spear-NFOBJ give-NFUT
‘I gave him a spear’ (ibid:476, 56b)
The object inflection in these examples varies in agreement with the tense category of the verb.
This is a case, then, in which both the nominal dependent and the verb do encode the same tense
function, both encoding the tense of the proposition.
Propositional TAM on nominal dependents usually interacts in this way with TAM marking on the
verb, together determining the TAM category for the clause. In a small number of languages, however,
the nominal inflection can be the only locus for the encoding of a particular propositional TAM
category. This is found in Chamicuro, for example, in which verbs are only optionally inflected for
tense. In many examples, the only instance of propositional TAM marking is that associated with
the definite article, as in the following example repeated from above.
(134) Y-alíyo ka ké:ni.
3-fall THE(PAST) rain
‘It rained’ (the rain fell).
Such examples are particularly interesting as they cannot even be treated as agreement between a
verb and its dependents. Rather, the propositional tense inflection must be analysed as a property
of the NP itself.
36

Within the category of non-propositional TAM inflection (section 3) the basic distinction is that
between past tense and non-past tense. All languages with this type of nominal TAM inflection
encode this distinction, and for some languages it is only this two-way distinction which is marked
(e.g. Hixkaryana, Somali). Other languages mark a three-way contrast, including a future tense
inflection also (e.g. Tariana, Guaraní, Halkomelem). While a couple of languages encode some
mood distinctions (i.e. Iate and Nambiquara), we have found no languages which encode aspectual
distinctions in this function. There appears to be no principled reason as to why this should be the
case, given that such meanings can be encoded by adjectives in languages without inflectional
nominal TAM systems (e.g. my ongoing journey, my completed journey, that which continues to be
a house), and may simply be an accidental gap in the data.
Finally, we have found many languages which have more than one type of nominal TAM function.
Modal case in Kayardild, for example, encodes propositional TAM on both nominal dependents
(e.g. (24, 25)) and nominal predicates (e.g. (31, 32)). And in Halkomelem, tense inflections
are used with nominals to encode both propositional tense on nominal predicates (e.g. (94)) and
non-propositional tense on nominal dependents (e.g. (92)). We have found no language, however,
which encodes both propositional and non-propositional tense on dependent nominals: languages
chose either to temporally locate the nominal itself (non-propositional), or to use the nominal to
(help) temporally locate the clausal predicate (propositional), but no language does both.
5 Further Implications
Such nominal TAM inflection as discussed in this paper has wide-ranging implications for linguistic
theory. Firstly, the recognition of tense as a possible inflectional category for nouns has
implications for theories of word class categorization which treat nouns as inherently time-stable,
manifesting stactic concepts that change little over time (Givón (1979: 320-322), (2001: 51-54)).
The analysis of nouns as inherently time-stable leads to the prediction that it should not be possible
for them to be temporally modified independently of the verbal predicate. This, however, is
exactly what we find in languages with non-propositional nominal TAM inflection (discussed in
section 3). While it is certainly true that nouns are less prototypically marked for tense than verbs
cross-linguistically, their being more inherently time-stable than verbs can not be interpreted as
precluding them from being temporally modified at all.
The treatment of nouns as time-stable is contradicted by work in semantics which has argued that
nouns, being semantic predicates, are time sensitive and therefore need to receive a temporal interpretation
independently of that of the verbal predicate (Enç (1981, 1986), Musan 1985, Lecarme
(1996, 1999), Tonhauser (2000, 2002)). The majority of this work has centred on languages (like
English) in which tense is not encoded morphologically on nouns (with Lecarme’s work on Somali
a notable exception). The existence of languages with explicit tense marking on nominals would
appear to be overt morphological evidence for these semantic approaches, although it remains to
be seen how easily such overt morphological systems can be integrated with semantic theory in
this respect.
A further implication related to the issue of word class typology is the fact that tense (and to a
lesser extent, aspect and mood) must now be seen as a possible inflectional category for nouns as
well as for verbs. Traditionally, inflectional categories are partitioned between the two major word
37

classes: nouns are inflected for categories such as case, gender and number and verbs for those like
tense, mood, aspect and person/number agreement (e.g. Lyons (1968: Ch. 7), Hopper and Thompson
(1984:703), Givón (2001: Ch. 2)). In fact, the categories of tense, aspect and mood may be
categories of nouns also (Evans 2000, Lehmann and Moravcsik 2000). The existence of propositional
TAM inflection on nouns, as discussed in section 2.2, also has potential implications for
Bybee’s principle of relevance which predicts that a semantic element will only have inflectional
expression if its meaning is “highly relevant” to the stem to which it attaches (1985: 13). Since
propositional TAM values are clearly not relevant to the interpretation of a dependent nominal,
the principle of relevance predicts that propositional TAM should not be an inflectional category
of nouns. The languages discussed in section 2.2 clearly contradict this prediction, and thus are
particularly interesting for theories of typology which assume it.
Such inflection of dependent nominals for propositional TAM properties also has significant implications
for grammatical architectures which assume that clause-level information must necessarily
be associated with the clausal head, and not with nominal dependents. According to the standard
notion of headedness, clausal properties (including propositional TAM) must be associated with
clausal heads (i.e. verbs and auxiliaries). Dependent nominals, especially subject nominals (which
are not even part of the VP), are not clausal heads and, therefore, cannot (according to such formal
theories) encode clasual properties. Thus, the data that we have presented in section 2.2 poses real
challenges for a number of formal theories of grammatical structure. 53
In this paper we have surveyed data from a wide range of typologically and genetically diverse
languages and identified two major types of nominal TAM inflection – propositional (appearing
on both dependent nominals and nominal predicates) and non-propositional (i.e. operating independently
of the TAM of the clause). In fact, there are also languages in which tense is encoded
on other parts of speech, such as adverbs and prepositions (e.g. Malagasy (Sabel 2002)). The
extent of TAM inflection on these (and other) word classes remains to be seen, and we leave the
investigation of its properties and its relationship to the typology of nominal TAM inflection as a
topic for further research. Clearly, however, the breadth of data from languages which encode TAM
information on their nominals and other NP constituents shows that TAM can no longer be thought
of as an inflectional property solely of verbs.
Bibliography
Abbott, M. 1991. Macushi. In D. C. Derbyshire and G. K. Pullum (Eds.), Handbook of Amazonian
Languages, Vol. 3, 33–127. Berlin: Mouton de Gruyter.
Aikhenvald, A. Y. 1999. Multiple marking of syntactic function and polysynthetic nouns in Tariana.
In Papers from the 35th Chicago Linguistic Society. CLS.
Aikhenvald, A. Y. To appear. The Tariana language of Northwest Amazonia. Cambridge: Cambridge
University Press.
Anderson, S. R. 1985. Typological distinctions in word formation. In T. Shopen (Ed.), Language
Typology and Syntactic Description, Vol. 3, 3–56. Cambridge: Cambridge University Press.
53 For fuller discussion of this issue, the reader is referred to Sadler and Nordlinger (in preparation).
38

Barron, J. 1998. ‘have’ contraction: explaining ‘trace effects’ in a theory without movement.
Linguistics 36(2):223–251.
Bender, E., and I. Sag. 2001. Incorporating contracted auxiliaries in English. In R. Cann, C. Grover,
and P. Miller (Eds.), Grammatical Interfaces in HPSG. Stanford: CSLI Publications.
Bhatia, T. K. 1993. Punjabi. London: Routledge.
Blake, B. 1979. Pitta Pitta. In R.M.W. Dixon and B. Blake (Ed.), Handbook of Australian Languages
Volume 1, 183–242. Amsterdam: Benjamins.
Blake, B. 1987. Australian Aboriginal Grammar. London: Croom Helm.
Bohnhoff, L. E. 1986. Ya.g Dii (Duru) Pronouns. In U. Wiesemann (Ed.), Pronominal Systems,
103–129. Gunter Narr Verlag.
Burquest, D. 1986. The pronoun system of some Chadic Languages. In U. Wiesemann (Ed.),
Pronominal Systems, 71–101. Gunter Narr Verlag.
Burton, S. C. 1997. Past Tense on Nouns as Death, Destruction, and Loss. In K. Kusomoto (Ed.),
NELS 27, 65–77.
Bybee, J. L. 1985. Morphology. Amsterdam: John Benjamins.
Carlson, R. 1994. A Grammar of Supyire. Berlin: Mouton de Gruyter.
Comrie, B., and S. A. Thompson. 1985. Lexical nominalization. In T. Shopen (Ed.), Language
Typology and Syntactic Description, Vol. Volume 3, 349–398. Cambridge: Cambridge University
Press.
da S. Facundes, S. 2000. The language of the Apurina People of Brazil(Maipure, Arawak). PhD
Dissertation, SUNY Buffalo.
de Canese, N. K. 1983. Gramatica de la Lengua Guarani. Asuncion: Coleccion Nemit˜y.
Demirdache, H., and L. Matthewson. 1995. On the universality of syntactic categories. In NELS
25.
Dench, A., and N. Evans. 1988. Multiple Case-Marking in Australian Languages. Australian
Journal of Linguistics 8:1–47.
Derbyshire, D. 1979. Hixkaryana. Amsterdam: North Holland.
Derbyshire, D. 1999. Carib. In R. M. W. Dixon and A. Aikhenvald (Eds.), The Amazonian Languages,
23–64. Cambridge, UK: Cambridge University Press.
E. S Furby and C. E. Furby. 1977. A Preliminary Analysis of Garawa phrases and clauses. Canberra:
Pacific Linguistics.
Enç, M. 1981. Tense without Scope: An Analysis of Nouns as Indexicals. PhD Dissertation,
University of Wisonsin, Madison.
39

Evans, N. 1995. A Grammar of Kayardild: With Historical-Comparative Notes on Tangkic. Berlin:
Mouton de Gruyter.
Evans, N. 2000. Word Classes in the world’s languages. In G. Booij, C. Lehmann, and J. Mugdan
(Eds.), Morphologie/Morphology: ein internationales Handbuch zur Flexion und Wortbildung/an
International Handbook on Inflection and Word-Formation, Vol. 1, 708–32. Berlin: Walter de
Gruyter.
Evans, N. to appear. A Pan-dialectal Grammar of Bininj Gun-Wok (Arnhem Land): Mayali, Kunwinjku
and Kune. Canberra: Pacific Linguistics.
Firestone, H. L. 1965. Description and Classification of Sirionó. London: Mouton.
Foley, W. A., and R. D. V. Valin. 1984. Functional Syntax and Universal Grammar. Cambridge:
CUP.
Galloway, B. 1993. A Grammar of Upriver Halkomelem. Berkeley: University of California Press.
Genetti, C. 1991. From postposition to subordinator in Newari. In E. Traugott and B. Heine (Ed.),
Approaches to Grammaticalisation, Vol. Volume 2, 227–55. Amsterdam: John Benjamins.
Gerdts, D. B. 1988. Object and Absolutive in Halkomelem Salish. New York: Garland.
Giridhar, P. P. 1994. Mao Naga Grammar. Mysore: Central Institute of Indian Languages.
Givón, T. 1979. On Understanding Grammar. New York: Academic Press.
Givón, T. 2001. Syntax, Vol. 1. Amsterdam: John Benjamins.
Gregores, E., and J. Suárez. 1967. A Description of Colloquial Guaraní. The Hague: Mouton.
Guasch, A. 1956. El idioma guarani. Gramática y antología de prosa y verso. Asunción: Casa
Américas.
Hale, K. 1998. On endangered languages and the importance of linguistic diversity. In L. A.
Grenoble and L. J. Whaley (Ed.), Endangered languages: Language loss and community response,
192–216. Cambridge: Cambridge University Press.
Harris, A., and L. Campbell. 1995. Historical syntax in cross-linguistic perspective. Cambridge:
Cambridge University Press.
Hercus, L. 1982. The Bagandji Language. Canberra: Pacific Linguistics.
Hockett, C. F. 1958. A Course in Modern Linguistics. New York: MacMillan Company.
Hopper, P. J., and S. A. Thompson. 1984. The discourse basis for lexical categories in Universal
Grammar. Language 60(4):703–752.
Jelinek, E. 1996. Quantification in Straits Salish. In E. Bach, E. Jelinek, A. Kratzer, and B. Partee
(Eds.), Quantification in Natural Languages. Dordrecht: Kluwer.
40

Jelinek, E., and R. Demers. 1994. Predicates and Pronominal Arguments in Straits. Language
70:697–736.
Ken Hale. 1997. Remarks on Lardil phonology and morphology. In Ngakulmungan Kangka
Leman (Ed.), Lardil Dictionary, 12–56. Queensland, Australia: Mornington Shire Council.
Kinkade, M. D. 1983. Salish evidence against the universality of ‘noun’ and ‘verb’. Lingua
60:25–40.
Klokeid, T. 1976. Topics in Lardil Grammar. PhD Dissertation, MIT.
Koehn, E., and S. Koehn. 1986. Apalai. In D. C. Derbyshire and G. K. Pullum (Eds.), Handbook
of Amazonian Languages, Vol. 1, 33–127. Berlin: Mouton de Gruyter.
Lapenda, G. 1968. Estrutura da língua Iatê: falada pelos índios Fulniôs em Pernambuco. Recife:
Unversidade Federal de Pernambuco.
Lecarme, J. 1996. Tense in the nominal system. In J. Lecarme, J. Lowenstamm and U. Shlonsky
(Ed.), Studies in Afroasiatic Grammar. The Hague: Holland Academic Graphics.
Lecarme, J. 1999. Nominal Tense and Tense Theory. In F. Corblin, C. Dobrovie-Sorin, and J.-
M. Marandin (Eds.), Empirical Issues in Formal Syntax and Semantics 2, 333–354. The Hague:
Thesus.
Lehmann, C., and E. Moravcsik. 2000. Noun (Article 73). In C. L. Geert Booij and J. Mugdan
(Eds.), Morphologie/Morphology, 732–757. Berlin: Mouton de Gruyter.
Lowe, I. 1999. Nambiquara. In A. Y. Aikhenvald and R. Dixon (Eds.), The Amazonian Languages,
269–292. Cambridge: Cambridge University Press.
Lyons, J. 1968. Introduction to Theoretical Linguistics. Cambridge: Cambridge University Press.
Matthewson, L., and H. Demirdache. 1995. Syntactic categories in St’át’imcets (Lillooet Salish).
In 30th International Conference on Salsih and Neisghboring Languages, 54–68. University of
Victoria, Victoria.
Mel’čuk, I. 1994. Cours de Morphologie Générale. Vol. 2. Montréal: Les Presses de l’Université
de Montréal.
Mistry, P. 2001. Subjecthood of nonnominatives in Gujarati. In Proceedings of the International
Symposium on ‘Non-nominative subjects’, 333–359. ILCAA, Tokyo.
Mous, M. 1993. A Grammar of Iraqw. Hamburg: Helmut Buske Verlag.
Nordlinger, R. 1998a. A Grammar of Wambaya, Northern Territory (Australia). Canberra: Pacific
Linguistics.
Nordlinger, R. 1998b. Constructive Case: Evidence from Australian Languages. Stanford: CSLI
Publications.
Nordlinger, R., and L. Sadler. 2000. Tense as a Nominal Category. In M. Butt and T. H. King
(Eds.), Proceedings of LFG 2000. CSLI Publications.
41

Parker, S. 1999. On the behavior of definite articles in Chamicuro. Language 75(3):552–562.
Pensalfini, R. J. 1997. Jingulu Grammar, Dictionary and Texts. PhD thesis, MIT, Boston, MASS.
Pensalfini, R. J. 2002. Verbs as spatial deixis markers in Jingulu. Paper presented at the Workshop
on subordinate clauses in Australian languages, Victoria Australia, March 2002.
Plank, F. (Ed.). 1995. Double Case: Agreement by Suffixaufnahme. Oxford: OUP.
Richards, N. 2001. Leerdil Yuujmen bana Yanangarr (Old and New Lardil). In Forty Years On:
Ken Hale and Australian languages, 431–445. Canberra: Pacific Linguistics.
Sabel, J. 2002. Tense-Agreement effects in Malagasy. Paper presented at AFLA 9, Cornell University,
27th April 2002.
Sadler, L. 1998. English auxiliaries as tense inflections. Essex Research Reports 1–16.
Sadler, L., and R. Nordlinger. 2001. Nominal Tense with Nominal Scope: A Preliminary Sketch.
In M. Butt and T. H. King (Eds.), Proceedings of LFG 2001. CSLI Publications.
Sadler, L., and R. Nordlinger. In preparation. Clausal tense on sentential dependents. MS, University
of Essex and University of Melbourne.
Saeed, J. 1999. Somali. Amsterdam: John Benjamins.
Salminen, T. 1997. Tundra Nenets Inflection. Helsinki: Suomalais-Ugrilainen Seura.
Spencer, A. 1991. Morphological Theory. Oxford: Blackwells.
Tonhauser, J. 2002. A Dynamic Semantic account of the temporal interpretation of Noun Phrases.
Paper presented at SALT 12, San Diego.
Tryon, D. T. 1968. Iai Grammar. Canberra: Pacific Linguistics.
Valin, R. D. V., and R. J. LaPolla. 1997. Syntax: structure, meaning and function. Cambridge:
CUP.
Wurm, S., and L. Hercus. 1976. Tense-marking in Gurnu pronouns. Papers in Australian Linguistics
10:33–49.
42