On the tense-aspect/modality interaction in non-PamaNyungan languages: A comparative account of the
Anindilyakwa, Iwaidja and Murrinhpatha modal systems
Patrick Caudal / Robert Mailhammer / James Bednall / Rachel
Nordlinger*
1.
Introduction
The topic of this paper1 is a detailed comparative investigation of the modal
systems of three non-Pama-Nyungan languages of the Top End region of
Australia: Iwaidja, Anindilyakwa and Murrinh-Patha. Capitalizing on recent
in-depth data collection and analysis of modal systems of Anindilyakwa (cf.
Bednall 2020) and Iwaidja (including substantial fieldwork on the Iwaidja
modal system conducted over the past few years) which update and extend
previous analyses of these two languages (particularly (Pym & Larrimore
1979)’s description of Iwaidja and (van Egmond 2012)’s description of
Anindilyakwa), we examine and compare the TAM systems of these languages
alongside examining preexisting analyses of the Murrinh-Patha tense-aspectmodality (TAM) system, in particular (Nordlinger & Caudal 2012).
A key objective of this paper is in the examination of Verstraete's (2005;
2006)’s seminal concept of ‘composite mood marking’ (which has had a
significant influence among linguistic researchers of the ‘Top End’2). Given the
fact that the three languages in our comparative sample, Anindilyakwa,
*
CNRS, LLF & U. Paris-Cité, pcaudal@linguist.univ-paris-diderot.fr / Western Sydney University,
r.mailhammer@westernsydney.edu.au / Charles Darwin University & Australian National
University, james.bednall@anu.edu.au / The University of Melbourne, SOLL,
racheln@unimelb.edu.au
1
We gratefully acknowledge the support of multiple institutions and projects who funded the
present research over the years: the Labex Empirical Foundations of Linguistics (Agence
Nationale de la Recherche programme Investissements d’Avenir, ANR–10LABX–0083),
subprojects GD4, GL3 and MEQTAME (Strands 3 and 2) (CI: Patrick Caudal) (2010-), the
CNRS SMI project Complexité morphologique et sémantique de la modalité en Iwaidja
(2018–2019) (CI: Patrick Caudal), the CNRS FEMIDAL (‘Formal/Experimental Methods
and In-depth Description of Australian Indigenous Languages’) International Research
Project (2021-) (CI: Patrick Caudal), Discovery Grant (DP130103935, CI: Robert
Mailhammer) by the Australian Research Council, and Western Sydney University Seed
Grant ‘Iwaidja Oral History & Ethnomedicine’ (CI: Robert Mailhammer).
2
Fieldwork linguists specializing in languages spoken in the Top End of Australia, especially
non-Pama-Nyungan languages, frequently use this term when affectionately referring to
fellow linguists having worked in remote communities of Northern Australia.
2
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
Iwaidja and Murrinh-Patha, possess discontinuous morphological marking for
their TAM forms, we will try and determine whether this concept of ‘composite
mood marking’ could, or could not, be legitimately applied to these languages,
and by extension to other languages with a similar morphological make-up,
both within and outside of Australia. The key question we ask is – what type of
morphology to semantics interface should be applied to such forms? One
obvious major issue is compositionality – does morphological discontinuity in
modal inflections rhyme with compositionality? And more specifically, can we
ascribe separate aspectuo-temporal vs. modal meanings to certain segments of
said discontinuous forms, and construe the overall meaning of modal
inflections by combining the meanings of these hypothetical sub-elements?
E.g., can we treat them as separate tense-aspect vs. modal morphemes? And if
it turns that ‘composite mood marking’ actually does not apply to the
languages of our sample, then what of the semantics of the relevant modal
inflections: does it nevertheless exhibit distinct aspectuo-temporal and modal
ingredients, and what can we say about their interactions?
Answering these questions is particularly relevant in understanding how
tense-aspect and modality are connected in the type of languages investigated
here, both formally (with respect to the morphology to semantics interface)
and semantically/pragmatically.
With respect to the morphological complexity and compositionality of the
three languages in our sample, we will establish the following key facts:
•
•
•
Iwaidja has a morphologically complex inventory of modal prefixes and
suffixes used in combination with each other as well as together with
modal adverbs/particles, similar to Ilgar/Garig (Evans 2000), without any
re-entrance of realis past vs. present tense suffixal exponents into modal
forms (i.e. there is no ‘past vs. present modal’ formal opposition), so that
any compositionality is ruled out in the tense-aspect/modality interaction
whereas
the Anindilyakwa modal system crucially involves a (partial) combination
of past vs. present-related suffixed exponents, along with an irrealis modal
prefixed exponent, essentially used to produce present vs. past irrealis
readings, in a seemingly transparent way. In addition to this,
Anindilyakwa offers a dedicated negative circumfixal modal paradigm,
and a deontic-imperative prefix only combining with a present temporal
suffix (thus undermining the system’s compositionality)
at first sight, the Murrinhpatha modal system, like Anindilyakwa, seems to
offer a modicum of compositionality in the tense-aspect interaction,
notably due to the re-entrance of the imperfective augment (=dha) in
irrealis and past imperfective paradigms; however, as we will see, there are
several facts relating to other paradigms (in particular other irrealis
paradigms) which militate against such a compositional approach.
Our first main claim will be that despite these surface differences, all three
TAM systems should be analyzed as involving single discontinuous TAM
morphs (not a combination of two separate tense-aspect vs. modality morphs),
with some amount of apparent semantic transparency in the tenseaspect/modality interaction in Anindhilyakwa and Murrinhpatha, but not real
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
3
compositionality (pace (Verstraete 2005; Verstraete 2006)). From the
perspective of the semantic diversity and semantic structure of inflectional
modal meanings in these languages, we will demonstrate that although some
significant divergences can of course be observed, especially among ‘present’
inflectional modals, it is more than counterbalanced by a striking number of
semantic convergences among ‘past’ inflectional modals, and the interaction
with negation. We will also show that aspectual parameters and actualistic
readings of modals constitute key factors when trying to better analyze the
tense-aspect/modality interaction in such languages. We will suggest that this
is also reflected on the existence of recurrent trends in the semantic
organization of modal categories in other Arnhem Land languages, and
beyond.
1.1. Theoretical background
We will assume in what follows a standard semantic approach to modality,
whereby the semantics of modals should comprise two main semantic
ingredients (cf. e.g. (Matthewson & Truckenbrodt 2018): (i) force/strength
(which, assuming a Kratzerian (Kratzer 1991) type of quantificational account
of modals, would be implemented using weak vs. strong quantifiers; this can
also be captured resorting to e.g. scalar-probabilistic accounts of modals, e.g. à
la (Lassiter 2014) or à la (Portner & Rubinstein 2016)) and (ii) flavor (which
corresponds to the non-quantificational/scalar, lexical meaning of modals; cf.
the Kratzerian noton of ‘modal bases’). Although this will not be investigated
here in detail, it is worthwhile noting that Australian languages have been
claimed (cf. e.g. (Bednall 2023)) to pattern like e.g. Indigenous American
languages (and unlike many Indo-European languages) in that they do not
lexicalize modal force/strength; i.e., force contrasts as found in English must
(necessity) vs. might (weak possibility) are primarily a contextual matter in
Australian languages. We will also assume that modal inflections as well as
modal verbs and auxiliaries are stative event predicates, as the case in
numerous works stressing the importance of viewing modals as bona fide event
predicates (Homer 2011; Caudal 2012; Ferreira 2014; Homer 2021).3
We will here follow a view pioneered in (Bybee 1998), according to
which irrealis should not necessarily be conceived of as a single monolithic
crosslinguistic category, opposed to another, single realis category also found
across languages. Rather, it should be considered as a wide semantic notional
domain with possibly some language-specific variation. It can be potentially
embodied by multiple grammatical markers (e.g. different inflections)
endowed with meanings pertaining to the domain of irrealis meaning,
standing in opposition to another multiplicity of marker endowed with realis
meanings (also possibly realized as different inflections).
3
The main difficulty for such an approach lies in handling the scope relations between tense
and the event denotation for various modal flavours; some authors claim that epistemic
modals ‘outscope’ tense-aspect operators, while others claim that they don’t. See e.g.
(Hacquard 2010; Homer 2013; Rullmann & Matthewson 2018) for a discussion of a very
complex issue, which we will here set aside.
4
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
Like in much of the existing theoretical literature, we will in particular
distinguish between two main types of irrealis grammatical markers: those
capable of contributing so-called ‘foreclosed’, ‘past irrealis’ or counterfactual
meanings, and those capable of contributing ‘open’, ‘present irrealis’, noncounterfactual meanings. By counterfactual, and assuming a possible-world
type of semantic model, we really mean possible worlds that are given as
absolutely inaccessible given a certain current context and/or past history (i.e.,
not merely very unlikely, but still attainable worlds); all other irrealis meanings
refer to accessible possible worlds. Note that in many formal semantic works,
‘counterfactual’ has really been used as a cover term for irrealis – see e.g. (Lewis
1973; Ippolito 2003; Arregui 2009; Romero 2014), among others. Our definition
is broadly similar to that of ‘present’ or ‘open counterfactual’ vs. ‘past’ or
‘foreclosed counterfactual’ in the latter terminological tradition, or ‘the
possible’ vs. ‘the counterfactual’ in (von Prince, Krajinović & Krifka 2022). By
foreclosed counterfactuals, we will refer to the meaning of irrealis marked
expressions, such that the proposition under the scope of such a modal belongs
with utterly inaccessible worlds, and is implied to be untrue (1).4 Vice versa,
open counterfactual meanings are associated with modalized propositions
which, while they are nevertheless untrue for the contextually relevant worldtime pair, may be accessible at some ulterior possible world (however unlikely
this might be) (2). Note that in e.g., contemporary Germanic and Romance
languages, two-past counterfactuals always qualify as foreclosed, and one-past
counterfactuals (often) qualify as open counterfactuals.5 However, one-past
counterfactuals also seem to be able to have foreclosed interpretations,
including in structures where they normally don’t have such readings; this is
particularly true when they mark individual-level stative clauses as in (3). Of
course, it is worthwhile bearing in mind that Australian languages – and
languages in our sample – do not offer similar contrasts between one and twopast counterfactuals; as we will see, this is in fact a key fact.
(1) If I had been rich, I would have bought a huge mansion.
(2) If I were rich, I would buy a huge mansion.
(3) If I were you, I would be thrilled. (Karawani, Kauf & Zeijlstra 2019:
212)
4
We will in fact argue that ‘past’ is, in fact, a somewhat unfortunate label, as not all foreclosed
counterfactual modals can be associated with a bona fide past anchoring; rather, past
conditions in modals (if present diachronically or synchronically) seem to serve the
purpose of somehow rendering some possibility irrevocably inaccessible.
5
This generalization does not apply across the diachrony of both Romance and Germanic
languages, and does not apply to some current modal constructions (e.g. biclausal
conditionals with imperfetto protases in Italian, or conditional structures with imparfait
marking other than si P,Q biclausal constructions in French; see (Caudal 2018a) for a
discussion).
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
5
We will assume that ‘irrealis’ is an appropriate cover category for
modal meanings as found in our small language sample, in that it can be
distinguished from a set of non-modal realis meanings. It should also be noted
that we regard mood as a cover term referring to modal markers, but not as a
notional category, nor their semantic content – for which we will rather use
the term modality.
We will claim that in our small sample (and in fact, in very many
Indigenous Australian languages), present vs. past irrealis categories (or open
vs. foreclosed modals, if you will) usually point semantically to two distinct
clusters6 of modal meanings. However there are some exceptions to this, where
one single modal inflection exhibits readings pertaining to both clusters. This
is for instance the case of the Bininj Gun-wok irrealis inflection, which admits
both open (4) and foreclosed irrealis (5) readings (Evans 2003a: 372–376), as
well as the ‘irrealis 2’ (I2) inflection in Mawng, which admits present irrealis,
deontic/directive readings (6), on top of past irrealis, counterfactual readings,
past admonitive readings, avertive readings, negative past readings and
avertive readings (7) (Singer 2006: 62).
(4) Nungka wanjh ∅-ra-yi
He
then 3P-gO-IRR
'He should go soon.'
werrkwerrk.
quickly
(5) bi-ma-yi
Na-burlanj gun-mak.
(Bininj Gun-wok)
3/3hP-marry-IRR ma-[skin] IV-good
‘She should have married straight, to a Naburlanj man.’ (Evans
2003:375)
(6) "An-kakujpi-na! Karlapuk! La arukin
arruni-ngartpanpu-ø!"
2sg-be.silent-I2 shoosh! CONJ snake
3MA/1pl.in-attack-I1
"Be quiet! Shoosh! Or else the serpent might attack us!." (Singer
2006 :62)
(7) Ja
karrkpin
ja
alakaraj
ing-errka-nyi.
MA
big
MA fishing.spear
3FE/3MA-spear-I2
'She tried to spear it with a big spear' ((Capell & Hinch 1970: 80) in
(Singer 2006:62))
It is worthwhile noting that across Australian languages, the open irrealis
cluster seems to be covered by multiple synthetic inflections, plus some
periphrastic ones, whereas the foreclosed irrealis cluster is generally covered
by a single synthetic inflection, also appearing in periphrastic inflections and
sometimes complemented with unrelated periphrastic modal inflections cf.
6
See (Caudal 2023) for a related use of the concept, where the ‘irrealis-avertive cluster’
designates what we call here the foreclosed irrealis, which seems to overwhelmingly
comprise an avertive meaning. through one or several (inflectional) forms.
6
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
(Caudal 2023), notably some sort of past indicative combined with a modal
particle or clitic.
We take the realis/irrealis divide not to align strongly with illocutionary
force and clause typing, in that both irrealis and realis can in principle mark
any type of clause, and any speech act type. However, modal markers can, in
some languages, impact discourse contextual and discourse structural
parameters connected to speech act types; this is particularly obvious if one
regards e.g. rhetorical relations as functions over speech act referents, as in the
case of the SDRT formalism (see (Asher & Lascarides 2003) For instance, if one
considers (Roberts 1989)’s notion of ‘modal subordination’ illustrated in (8),
where speech act referent 𝛽 appears to be modally dependent from speech act
referent 𝛼 (assuming that speech acts can be reified in the SDRT fashion, cf.
(Asher & Lascarides 2003), then it appears natural to view conditionals, and
from there, all inferential modals – e.g. epistemics – and evidentials, as
essentially discursive issues in need of a treatment at the
semantics/pragmatics interface. Interestingly, this is exactly how (Ciardelli
2022) proposes to re-cast (Kratzer 1977; Kratzer 1981; Kratzer 1991)’s analysis of
modals and conditionals as Generalized Quantifiers, i.e. as operators
combining an intensional contextual parameter (the restrictor) and a
(modalized) proposition (the nuclear scope) analysis inspired by the notion of
modal subordination à la (Roberts 1989), (9) ; see also (Fintel & Gillies 2015) for
a somewhat similar discursive twist on Krater’s original idea.
(8) A wolf might come in (𝛼). He would eat you first (𝛽).
(9) Modal_operator [intensional_context parameter]restrictor
[proposition]nuclear scope
Although we will not discuss evidentiality at length here, as it seems to be
mostly lacking specific very clear grammatical expressions in our sample
(unlike in e.g. Arandic or Ngumpin-Yapa languages, see Browne & Ennever,
this volume), it is important to specify that we do not wish to here endorse a
strong theoretical view concerning the relationship between evidentiality and
modality – a well-known theoretical hurdle for specialists of these categories.
Some argue for a complete disjointness of the two categories (see e.g. : (De
Haan 1999; Aikhenvald 2004), some for some kind of overlap (some modals are
evidentials, or some evidentials have a modal component of meaning) (Faller
2002; Squartini 2004; Matthewson, Davis & Rullmann 2008), while others
argue for a principled inclusion of one category into the other – notably by
treating e.g. epistemic modals as a subset of evidentials (von Fintel & Gillies 2010)
– or a principled identification of epistemic modals with evidentials, and vice
versa (Matthewson, Davis & Rullmann 2008)
This paper focuses on the tense-aspect / modality interaction. We will
notably investigate how temporal and aspectual conditions can be identified
in the various readings found for at least some present vs. past irrealis forms.
Counterfactual meanings are a notoriously difficult phenomenon in this
respect, as is evidenced by strong variations (and frequent divergences) within
relevant theoretical and formal analyses of irrealis forms across languages of
the world, particularly in that they frequently (but not systematically) involve
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
7
a joint aspectual, temporal and modal marking. Or at least a joint aspectual
and modal marking, or joint temporal and modal marking, depending on the
languages. In the case of the three languages here studied (and in fact, in many
other non-Pama-Nyungan languages presenting similar discontinuous
morphs), our main point will be that morphological structure – which often,
turns out to be unanalyzable as involving different morphs – does not mirror
the semantic complexity of the TAM meanings involved. Semantic complexity,
involving separate temporal, aspectual and modal conditions, need not project
on morphological complex units. Complex morphologization processes
combined with semantic change, can result in a single (possibly
discontinuous) morph endowed with a complex semantic content
But we should also stress that while such a discontinuous marking is
a common morphological configuration in non-Pama-Nyungan languages (cf.
e.g. Jaminjung, (Schultze-Berndt 2000), Dalabon (Evans & Merlan 2003),
Mawng (Singer 2006), Ngarnka (Osgarby 2018: 20), Ngan'gityemerri (Reid 2011)
a.o.), it is by no means systematic. See for instance, Amurdak (Mailhammer
2009), which possesses mere TAM prefixes even though they are related to
languages with discontinuous TAM morphs of the type investigated here; see
also Maningrida languages, which seem to only involve TAM suffixes. Such
systems, we believe, only differ from the type here studied by the nature of
morphology involved – in both cases, we have a simplex morphology (in spite
of appearances, in the case of discontinuous morphs), encoding a complex TAM content. But the complexity of the meaning associated with these forms is
fairly similar, and also involves significant interactions between aspectuotemporal meaning and modal meaning. Or to put it differently, that semantic
complexity does not map onto morphological complexity (i.e., does not
involve separate morphemes, with completely autonomous contributions).
Many non-Pama-Nyungan languages possess several distinct
morphological modal paradigms, described using various terms in the
literature (e.g., ‘future’, ‘potential’, ‘optative’, ‘irrealis’...). While we will not?
delve deep into the semantics of any of them in particular here, nor offer a
theoretical semantic overview of modality in these languages, we will offer
some general observations about the ‘semantic mapping’ of each TAM system.
This is notably relevant to understanding how tense and aspect information
combines with modal information in said systems.
1.2. Main objective of the paper
Following notably (Verstraete 2005)’s seminal study of ‘composite mood
marking’ in non-Pama-Nyungan languages, the concept appears to have
gained traction in the literature, and to have become relatively commonly
used. However, its precise definition, from the perspective of (i) morphological
analysis and (ii) the morphology to semantics interface, remains problematic.
Verstraete (2005:224), observes that “In the majority of non-Pama-Nyungan
languages, mood is not marked in one specific slot in the morphological
structure of the verb, but spread over at least two slots, marked by a
combination of morphemes in a prefix and a suffix slot”. Verstraete (2005:225)
further argues that “most non-Pama-Nyungan languages show a primary
subdivision between an irrealis-type prefix and a realis-type prefix (or absence
of a prefix […])”, the latter being used in most “non-modalized structures” –
8
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
i.e., indicative clauses, denoting actual events. Finally, Verstraete (2005:225)
claims that “within the broad semantic range covered by the irrealis prefix,
moreover, there is usually a further subdivision of modal meanings in terms of
combinations with different types of tense suffixes: […] the counterfactual
type is usually set apart from the other types of modality in terms of a basic
contrast between past and non-past tense suffixes.” He also very specifically
claims that it is endowed with “formal compositionality” (p. 225) – but at the
same time observes that in some languages, modal suffixes can also appear in
the same slot as past vs. present tense suffixes (such as the potential -yan suffix
in Wardaman), which we believe to be difficult to reconcile with a
compositional account of the tense-aspect / modality interaction in Australian
languages – so that we believe that Verstraete’s (2005) original notion of
“composite mood marking” should be revised in significant ways.
We will argue that Verstraete’s (2005) analysis should be revised with
respect to (i) compositionality and (ii) the role of aspectual conditions in the
semantics interplay between aspectuo-temporal and modal meanings of
irrealis forms. Concerning (i), we will suggest that while some languages in our
small sample exhibit what seems to be compositionality, it is really at best
limited to a small number of tense-aspect/modality paradigms, and that as a
result, it is best seen as a matter of relative transparency of said paradigms (but
not real compositionality). Concerning (ii), we will show that a number of
semantic phenomena (in particular what we will refer to as postmodal readings
of modal forms, using a term coined in (van der Auwera & Plungian 1998))
cannot be accounted for without incorporating aspectual meaning into a
semantic account of irrealis forms in our sample.
To establish (i) and (ii), the paper will proceed as follows: section §2 will
provide a detailed review of the morphology to semantics interface of
synthetic inflections in Iwaidja, Murrinh-Patha and Anindilyakwa; §3 will then
focus on some periphrastic modal inflections identified for these languages.
Finally, §4 will offer some tentative theoretical observations about modal as
well as postmodal readings, and will suggest that they should be indeed
grounded in a theory of the tense-aspect / modality interaction. The picture
emerging from this inventory will be one of a series of morphologically ‘atomic’
categories – namely modal categories – in the sense of (Blevins 2016), rather
than complex/composite signs, whose semantics should nevertheless
comprise multiple semantic ingredients, including – and crucially so – tense
and aspect. In particular, we will argue that this semantic complexity spanning
the aspectuo-temporal and modal (as well as postmodal) domains, stems from
the fact that modals, regardless of their morphological nature (periphrastic or
synthetic) behave crucially like event predicates, i.e. require temporal (and
aspectual) information to be processed. Some tentative diachronic
observations will be made to further substantiate this claim, as we will
additionally claim that semantic change tends to deploy in ways further
shedding light on the importance of the tense-aspect/modality interaction in
the modal domain.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
9
2. Synthetic verbal inflections in Iwaidja, Anindilyakwa and MurrinhPatha, and the modality / tense-aspect interaction
We will begin our theoretical journey through the modal landscape of our
language sample by describing and then analyzing some key properties of
synthetic modal inflections of Iwaidja, Anindilyakwa, and Murrinhpatha. To
this effect, we will take a broad look at the respective synthetic
tense/aspect/mood (TAM) systems of these languages, in order to determine
whether or not some manner of ‘composite mood marking’ à la (Verstraete
2005) can be ascribed to them, i.e. with two defining properties: (i) a tendency
(at least) to encode aspectuo-temporal meanings via suffixed exponents, and
modal meanings via prefixes exponents and (ii) compositionality.
This study should be placed in the broader context of discontinuous TAM
marking – a term we will use in lieu of Verstraete’s (2005) concept of
‘composite mood marking’ – found across many non-Pama-Nyungan,
Northern Australian languages. It should first be noted that many non-PamaNyungan languages possess several distinct morphological modal paradigms.
Various terms have been coined in the Australianist literature to refer to the
corresponding categories; although obviously, one should not assume that
identical modal categories should be found across non-Pama-Nyungan
languages (let alone Australian languages as a whole), it may be possible to
come up with useful comparative cover terms to relate them (while indicating
as well to what extent each language-specific instantiation of each
comparative category, semantically differs from said comparative category).
This will be very much our assumption here, as we will make apparent. For
instance, an irrealis present category capable of receiving both
hypothetical/future/futurate and deontic/imperative readings, and commonly
found across non-Pama-Nyungan languages, has been dubbed ‘future’,
‘potential’, ‘optative’, ‘irrealis’, or some other term, depending on authors. We
will propose to refer to these various inflections as ‘present
irrealis/counterfactual’, or ‘open counterfactual’ inflections.
Paradigms typically involve discontinuous morphological marking, with
two exponents realized on two disjoint positions in the verb template. In
Anindilyakwa (10) and Iwaidja (11) and in this involves a pronominal
portmanteau combining TAM and pronominal marking (Pro), found in
position 1 – a type of morphological configuration commonly observed across
non-Pama-Nyungan languages. In Murrinhpatha (12) this TAM and subject
pronominal marking is portmanteau with a classifier stem providing event
classifying semantics (see (Nordlinger 2015; Mansfield 2019) for detailed
discussion):
(10) Anindilyakwa verb template7
1
2
3
4 5
6
7
8 9
Pro.(TAM1) QUANT BEN IBP stem CAUS/REFL/RECIP TAM2 ma CASE
7
TAM: ‘tense, aspect, modality’; ‘Pro’: pronominal; QUANT: quantificational marker ; BEN :
beneficiary, IBP : incorporated body part, CAUS: causative; REFL: reflexive; RECIP:
reciprocal
10
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
(11) Iwaidja verb template:
1
Pro.(TAM1)
2
stem
3
REDuplication
4
TAM2
(12) Murrinh-Patha verb template 8
1
CS.SUBJ.
/
TAM1
2
SUBJ.NUM
/ OBJ
3
R
R
4
IBP/
APP
L
5
LEX
S
6
TAM
2
7
AD
V
8
SUBJ.NUM
/ OBJ.NUM
9
AD
V
But are these TAM1/2 exponents separate morphemes? Or does each of these
pairings constitute an instance of discontinuous morpheme (cf. the concept of
‘distributed exponence’ in (Carroll 2016), or that of ‘constructional
morphology’ in (Booij 2010))? This is really a key question we need to answer
for each of the languages in our sample.
2.1 Iwaidja synthetic inflections
Our description of the Iwaidja TAM system is based on extensive fieldwork
conducted over a decade (2013-2023) by R. Mailhammer and P. Caudal, mostly
in the Minjilang community (Croker Island, Northern Territory). It departs
from (Pym & Larrimore 1979)’s earlier description in several important ways,
but for want of space, we will not discuss them here.
We can decompose the Iwaidja verb template as in (13) – where RED stands
for ‘REDuplication’, and TAM1 and TAM2 are the prefixed vs. suffixed
exponents conveying TAM information:
(13) [Portmanteau prefix (Deixis+TAM1+Subject (+Object))9]-[Verb
Root]-[RED]-[TAM2]
Iwaidja is a head-marking language. Two types of verbs are differentiated in
terms of their morphology, namely transitive and intransitive verbs. Transitive
verbs express the Subject and Object in a portmanteau prefix, whereas the
prefixes of intransitive verbs only express the Subject even if the verbs are
divalent and have a second argument, cf. (14).
(14) a. nga-
ngartbuni
-Ø
8
CS.SUBJ.TENSE: Portmanteau encoding classifier stem, subject agreement and tense;
SUBJ.NUM: Subject number marker; OBJ: Object agreement marker; RR:
Reflexive/Reciprocal marker; IBP: Incorporated body part; =APPL: Applicative marker;
LEXS: Lexical stem; ADV: Adverbial; OBJ.NUM: Object number marker
9
‘+’ indicates here morphological fusion in a single portmanteau. Note that the ordering
provided here is not particularly significant – for instance, elements diachronically
reconstructable as TAM1, can also intervene in between Subject and Object.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
1.SG.PREF1‘I fall.’
fall
11
-PRES
b. rilda-Ø
3SG.M>3SG.PRED1- eat-PRES
‘He eats.’
c. warrnganyu -Ø nuwung
3SG.PRED1carry in arms -PRES2SG.OBL.PRON
‘He/she/it is carrying you in his/her/its arm’
(Pym & Larrimore
1979: 84)
The following table gives a synthetic overview of the various sets of prefixed
(TAM1) and suffixed (TAM2) paradigms, where PREFm- and -SUFn are arbitrary,
provisional names for these exponents:10
Table 1: overview of TAM prefixes & suffixes in Iwaidja
TAM2 -PR
-ANT
-PIPFV
TAM1
PREF1- Present Anterior Past
imperfective
PREF2- ✗
✗
✗
PREF3-
✗
✗
✗
-OPT
-FUT
-RMOD
-PCF
Present
irrealis
✗
✗
✗
✗
✗
Present ✗
irrealis
Present irrealis Past irrealis
✗
Seven meaningful synthetic paradigms involving prefix / suffix components
were identified during our fieldwork:
1. PREF1- ~ -PR: the indicative present
2. PREF1- ~ -ANT: an aspectually underspecified anterior tense, fairly
similar to the English simple past in that it can receive imperfective
viewpoint readings with atelic verbs, vs. perfective viewpoint
readings with telic verbs; in the latter case, it can also have
10
Semantically grounded names of TAM exponents are nevertheless provided on the suffixed
part, for ease of reference. These are:
-PR : present
-ANT: anterior
-PIPFV: past imperfective
-OPT : optative
-FUT: future
-RMOD:
root modal
-PCF: past counterfactual
See below for further morphological and semantic details.
12
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
resultative, perfect flavour, given appropriate modifiers or
contexts)
3. PREF1 ~ -PIPFV: an indicative past imperfective
4. PREF1- ~ -OPT: a present modal inflection, with directive (hortative
and imperative) uses, apprehensive-epistemic uses, and bona fide
optative-volitional uses (cf. English ‘I wish P’)
5. PREF2- ~ -FUT : a present modal inflection with directive/deontic
uses, as well as proximative/predictive uses
6. PREF3- ~ -RMOD : a present modal inflection also with directive
deontic uses, as well as capacitative uses
7. PREF3- ~ -PCF : a past modal inflection, with volitional, proximative
and avertive uses (Caudal 2023), as well as past hypothetical and
past counterfactual uses, including admonitive uses (‘X should have
V-ed’).
An obvious theoretical question at this stage is – do the prefixes and suffixes
contribute separate meanings, or do they form a single, though discontinuous,
morphological unit, i.e. constitute an instance of so-called distributed
exponence (Carroll 2016) ?
Table 1 makes it clear that:
1. Pref2- is not re-entrant anywhere in the inflectional system, so there is
no analytical advantage to assume that its combination with -Suf5 (-fut)
does not form a unique but discontinuous exponent (a circumfix, if you
will)
2. Pref1- vs. Pref2- / Pref3- cannot be understood as marking an
indicative/irrealis split, as Pref1- can associate both with an indicative
(e.g. Pref1-~ -SUF1 = Pref1-~ -pr) and with an irrealis interpretation (Pref1~ -SUF4 = Pref1-~ -opt), while Pref2- / Pref3- only ever associate with
irrealis; therefore, the indicative/irrealis divide cannot be constrained
solely by means of prefixes, and must be determined at the level of
PREFm- ~ -SUFn combinations
3. Furthermore, as we will see, the two PREF3- paradigms (PREF3- -~ -SUF6
and PREF3- -~ -SUF7 ) do not have the same irrealis meanings; this
demonstrates that prefixes are not only unable to discriminate irrealis
vs. indicative meanings, but also to discriminate between different
subtypes of irrealis meanings
4. None of the suffixed exponents can combine with more than one
prefixed exponent, so that there is no analytical advantage to assuming
a compositional account; on the contrary, such an analysis sounds very
ad hoc, while assuming that we are here dealing with distributed
exponence (i.e., circumfixal paradigms) appears to be the only natural
option.
Let us go through a specific illustration. The series of prefixes (PREF1-) that
are found in the present tense are also found in the optative, for instance.
Therefore, the prefix they share cannot be marked for the indicative/irrealis
opposition, i.e., modality. Additionally, (5) also contrasts with (6), which
expresses the past counterfactual (-pcf) in terms of modal (but also temporal)
content. This further confirms that prefixes are non-discriminating for
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
13
modality. In effect, suffixes alone are sufficient for all TAM discrimination
purposes (i.e. w.r.t. to both tense-aspect and modality), and given that none of
them appears in more than one prefix + suffix combination, these
combinations are the only legitimate locus for encoding TAM information, and
the Iwaidja inflectional TAM morphology is an instance of distributed
exponence.
(15) ri3SG.M>3SG.PREF1‘He might eat.’
lda-ng
eat-OPT
(16) nanilda-∅
3SG.M>3SG.PREF3EAT-PCF
‘He was going to eat it’.
To summarize, Iwaidja prefix and suffix combinations are obviously non
compositional since each prefix involved in a compositional marking would
contribute a separate meaning)., the best analysis is to treat those prefixed and
suffixed exponents as forming a unit from the point of view of form-meaning
pairings, i.e. as a ‘minimal sign’ (Blevins 2016) in this language.
2.2 Anindilyakwa synthetic inflections
The next TAM system we turn to is that of Anindilyakwa. Our description
and analysis of the Anindilyakwa TAM system is based on Bednall (2020),
which draws from extensive fieldwork on the Groote Eylandt archipelago
between 2016-2019. Bednall (2020) extends and refines previous analyses of the
verbal inflectional system (particularly van Egmond 2012) in a number of ways.
While there is not space in this paper to discuss the departures of this vs.
previous analyses, the reader is directed to Bednall (2020) for further
discussion.
Anindilyakwa is a head marking language, with core arguments crossreferenced via portmanteau prefixes on the verb. Verbs comprise stems that
can be simple or complex, with simple stems comprising just the verb root,
while complex stems historically consist of an uninflecting root plus inflecting
element, but are synchronically monomorphemic. The verb stem obligatorily
inflects with circumfix-like morphs; two non-adjacent slots of the verb
template. Similar to Iwaidja, this discontinuous morph involves a portmanteau
prefixed exponent showing pronominal and TAM information, and a suffixed
TAM exponent. The make-up of the Anindilyakwa TAM system is, however,
substantially different from that of Iwaidja. Four different series of TAMpronominal prefixal exponents (REAL-, IRR-, DEON-, NEG.NPST-) combine with
five series of TAM suffixal exponents (-NPST, -PST, -USP, -POT, -NEG.NPST). Out of
twenty theoretically possible combinations, only eleven are attested. Unlike
Iwaidja, the synthetic Anindilyakwa TAM inflectional system exhibits a
substantial re-entrance of both prefixes and suffixes. In particular, the REAL-,
IRR- and DEON- prefixes combine respectively with three, four and three
suffixes (assuming a ‘zero’, phonologically null suffix: -USP).
14
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
The following table shows the TAM prefixal and suffixal exponents
that make up the inflectional TAM system.
Table 2: overview of TAM prefixes & suffixes in Anindilyakwa
REALIRRDEON-
NEG.N
PST-
-NPST
-PST
-∅
-POT
Present
Indicative
Present
Irrealis
Present
deontic/
directive
✗
Past
Indicative
Past Irrealis
Present+Past
Indicative
Present+ Past
Irrealis11
Present deontic /
directive (*Past)
✗
✗
✗
✗
Present
Irrealis
Present
deontic
/directive
✗
NEG.NPST
✗
✗
✗
Present
(Negation)
The eleven meaningful prefixal-suffixal morphs are:
1) REAL- ~ -NPST: indicative present
2) REAL- ~ -PST: indicative past
3) REAL- ~ -USP: a temporally and aspectually underspecified inflectional
category, where aspectuo-temporal properties are sensitive to event
structural parameters
4) IRR- ~ -NPST: a present modal inflection, expressing epistemic, deontic
and base modal readings, and future readings
5) IRR- ~ -PST: a past modal inflection, with past hypothetical and
counterfactual, volitional, and avertive readings, and marking past
negative polarity
6) IRR- ~ -USP: a present modal inflection, expressing epistemic, deontic,
base modal and intentional readings; a past modal inflection in a
number of specific constructions (narrowly averted readings (in
combination with the purposive clitic), frustrative constructions, and
in dependent clauses (consequent clauses of conditionals, volitional
constructions, etc.))
7) IRR- ~ -POT: a present modal inflection, expressing epistemic, deontic
and base modal readings
8) DEON- ~ -NPST: a present modal inflection, with deontic and directive
(imperative/hortative) readings
9) DEON- ~ -USP: a present modal inflection, with deontic and directive
(imperative/hortative) readings
10) DEON- ~ -POT: a present modal inflection, with deontic and directive
(imperative/hortative) readings
11
Past anchoring occurs only in specific constructions: narrowly averted readings (in
combination with the PURPOSIVE clitic), frustrative constructions, and in dependent
clauses (consequent clauses of conditionals, volitional constructions, etc.).
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
15
11) NEG.NPST- ~ -NEG.NPST: negative present modal (proximative,
epistemic, deontic/necessity
The system appears more transparent than the other languages examined in
this paper, Iwaidja and Murrinh-Patha, however does that mean that it should
be analysed as being compositional?
While there are aspects of the Anindilyakwa TAM inflectional system
that are seemingly relatively transparent, such as the -NPST and - PST suffixal
forms, which occur only with present and past temporal anchoring
respectively, there are some shortcomings to applying a compositional
approach across the inflectional TAM system as a whole, as we examine below.
As shown in Table 2, only 11 of the 20 formally possible combinations
of prefixal and suffixal exponents are possible. This is largely due to the unique
neg.npst- ~ -neg.npst circumfix, a clear instance of a discontinuous modal
morph, where both the prefixed element and the suffixed element do not
combine with any other exponent to form a separate inflectional paradigm
(e.g. *REAL- ∼ -NEG.NPST and *IRR- ∼ -NEG.NPST are not grammatical inflectional
markings) (i.e. only one of the possible eight prefixal-suffixal paradigms
involving these two exponents are possible). This circumfix is also notable in
that, unlike the other prefix elements, the neg.npst- prefixal exponent doesn’t
mark person or number (i.e. it marks only tense-aspect-modality-polarity, with
a free pronoun required to specify person/number information, if necessary).
The NEG.NPST- ~ -NEG.NPST circumfix is used in all non-past contexts to express
negative polarity (aside from situations emphasising the intensity of the
negative polarity – i.e. for situations that do not occur under any condition (e.g.
never P), which instead inflect IRR- ∼ -NPST to an obligatorily reduplicated verb
stem).
If we turn to the other two disallowed paradigms, the ungrammaticality of
*REAL- ~ -POT is prima facie semantically explicable, given that the -POT suffixal
exponent occurs only with modal readings (i.e. IRR- ~ -POT and DEON- ~ -POT;
*REAL- ~ -POT). However, the final disallowed paradigm, *DEON- ~ -PST, is
semantically possible, despite it being grammatically unacceptable. deon- ~ npst expresses present deontic modal readings and directives, as in (17):
(17) Ø-lhukwe-n
deon.2-dance-npst
‘You should dance’
alhəkwanja
neut.dance
(CW, JRB1-049-01, 00:50:53.911-00:50:56.506)
IRR- ~ -NPST can similarly express present deontic modal readings, as in (18),
while on the other hand IRR- ~ -PST can express past deontic readings, as in (19).
*DEON- ~ -PST, however, is disallowed and thus doesn’t allow for past deontic
readings (cf. (20)). IRR- ~ -PST is the only grammatically acceptable inflection
available to express past deontic readings. It is therefore unclear why this
*DEON- ~ -PST prefixal-suffixal combination is deemed unacceptable, at least
on semantic grounds.
16
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
(18) kə-warri-na
m-akən
malharra
IRR.2-throw-NPST VEG -that
VEG.stone
‘You should throw that stone away’
(JL, JRB1-049-01, 00:51:41.589-00:51:44.291)
(19) nungkwa kə-lhəke-nə=ma
2.PRO
IRR.2-go-PST=MUT
‘You should have gone’
(JL, JRB1-060-02, 00.17.46-00.17.50)
(20) *Ø-lhukwe-nə=ma
DEON.2-dance-pst=MUT
alhəkwanja
NEUT.dance
‘You should have danced’ [constructed]
Thus, while much of the Anindilyakwa TAM system is fairly transparent and
semantically conventional, where (partial) combinations of past vs. present
suffixed exponents with irrealis modal prefixed exponents mark past vs.
present irrealis readings, in several areas – such as the NEG.NPST- ~ -NEG.NPST
negative circumfixal modal paradigm and the disallowance of the *DEON- ~ PST paradigm – the system appears more arbitrary. Therefore the most
consistent and straightforward overall approach to take is to analyse each
combination of prefix and suffix exponents as a single discontinuous TAM
morph.
2.3 Murrinhpatha synthetic inflections
The third TA/M system we need to discuss is that of Murrinhpatha; our
description will be primarily based on data discussed in (Nordlinger & Caudal
2012). The Murrinthpatha TAM system (see Table 3) involves two main
morphological TAM slots, in position 1 (classifier stem alternation) and
position 6 of the verb template (12) (TAM suffix). Six different classifier
stem/‘prefix’ series in position 1 combine with four different TAM infixes in
position 6 (they can be followed by other morphological exponents, so we
cannot treat them as suffixes in the general case)– we are assuming that the
morphologically null suffix constitutes a morphological class in its own right;
for the sake of simplicity, we will treat it as a zero exponent, as this enables us
to treat it on a par with exponents contrasted with what really constitutes a
reduced form of the verb.
Out of 24 theoretically possible combinations, only 8 are attested in our data.
5 out of 6 prefixes associate with a single suffix; the 6th combines with 3 out of
4 suffixes in position 4. Two position 6 suffixes (-nu and -nukun) associate with
a unique classifier stem (TAM3- and TAM6-), the two other suffixes (-Ø and dha) combine with two prefixes.12
12
The present analysis of the Murrinhpatha TAM system differs from the analysis put forth in
Nordlinger & Caudal (2012) w.r.t. to the so-called ‘future’ and ‘future irrealis’ paradigms
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
17
Table 3: overview of discontinuous TAM marking in Murrinh-Patha
Slot 6
Slot 1
TAM1
TAM2
TAM3
TAM4
TAM5
TAM6
-dha
-Ø
Temporally
underspecified
tense (-USP)
✗
✗
✗
Present deontic
/ directive (DEON)
-nukun
-mani13
✗
✗
✗
✗
✗
Volitional/
deontic
✗
✗
✗
✗
✗
✗
✗
Apprehensive
(deontic+
epistemic) (-APPR)
✗
Aspectually
✗
underspecified
past (-PST)14
✗
Future (FUT)
Irrealis (-IRR)
(Epistemic /
deontic / future /
directive/
hypothetical)
✗
Past irrealis (PSTIRR)
Existential15
*
✗
-nu
Given the above picture, it seems difficult to invoke overall compositionality
for such a morphological TAM system, as at least two series of exponents in
position 6, associate with a unique classifier stem form in position 1. Reentrance is non-existent for two suffixes, and most classifier stem forms in
position 1 also require a unique suffix in position 6 – only TAM6 combines with
several suffixes. This exponent exhibits at once the widest distribution, both
formally and semantically. While associated with multiple modal meanings
in the latter reference.
13
Although (Nordlinger & Caudal 2012) does not list this paradigm, (Mansfield 2020: 155, 178)
mention -mani/=mani as being part of a modal paradigm, since it cannot combine with
any other TAM suffix.
14
Following (Mansfield 2020), we are assuming that this past indicative inflection is in fact
aspectually underspecified, rather than intrinsically imperfective. This makes the
Murrinhpatha tense-aspect system very similar to that of Anindilyakwa, with a
temporally underspecified tense (called ‘non-future’ in Nordlinger & Caudal 2021)
constrasting with an aspectually underspecified past tense.
15
(Mansfield 2019) calls this inflection ‘authoritative’, and argues that it conveys the speaker’s
commitment to the validity of the proposition uttered (i.e., in terms of ‘epistemic
stance’).
18
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
(epistemic, deontic, directive and apprehensive), it seems to only mark a
present type of modal meaning.
(21) mere
NEG
na-ngi-mathputh-nukun – thurru (Murrinhpatha)
2SGS.HANDS(8).APPR-1SGO-interrupt-APPR– 2SGS.GO (6).
APPR
‘Don’t you continually interrupt me.’ (Street & Street 1989)
It is also unclear why certain position 1 / position 6 combinations of
exponents are ruled out, at least on semantic grounds. For instance, it’s unclear
why the seemingly apprehensive infix -nukun does not combine with other
exponents in position 1 – e.g. TAM3 or TAM4. Apprehensive morphology is
thus crosslinguistically known to be usable in counterfactual contexts and in
combination with forms expressing past irrealis meanings, cf. the Nakkara
periphrastic apprehensive inflection in (22) combining particle korla with the
past counterfactual/past irrealis inflection. However, in Murrinhpatha, suffix nukun does not seem to combine with TAM4 (or any other prefix) to encode
such a reading.
(22) (we built a huge fire ...) korla minja
namunja ya-bburba-ma
APPR flies
3>3.IRR+follow.food-PCT
‘We built a huge fire, otherwise the flies would have hung around’
(Nakarra) (Eather 1990: 347)
This suggests that overall, some combinations are semantically
conventional and well-established, while others are not, on arbitrary grounds.
While it is possible that such combinations are in fact rare, and not
represented in our data rather than impossible, this remains potentially a
genuine concern for a compositional theory of ‘composite mood marking’ in
Murrinh-Patha. Especially other published works on Murrinh-Patha does not
seem to provide any data illustrating said missing combinations; see e;g.
(Mansfield 2014; Mansfield 2019; Mansfield 2020).
In spite of the above argumentation, one could try and maintain that the
distribution of suffixes -Ø and -dha with certain position 1 markers is
semantically motivated, and could constitute a ‘core’ compositional sub TAM
system – as has been suggested above already for Anindilyakwa. However, this
becomes quickly problematic once we try to generalize the type of meaning
that should be ascribed to position 1 vs. 6. Consider for instance the position 6
suffix -dha. It marks a variety of both modal and non-modal inflectional
paradigms: the past imperfective (TAM2 + -dha) (23), the past irrealis (TAM4
+ -dha) (24) and what seems to be some kind of present irrealis/present
counterfactual paradigm (TAM6+ -dha) (25).
(23) ngarde-rerte-dha-ngime
1DUS.BE(4).PST-hit(RDP)-PST-PC.F
(Murrinh-Patha)
‘We were knocking (shellfish off the reef).’ (Nordlinger & Caudal 2012:98)
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
19
(24) ngay-dha ngatha-ka
me-mawatha-dha-wa
1SG-PSTIRR
if-FOC
1SGS.HANDS(8).PSTIRR-rectify-PSTIRREMPH
‘If it had have been me, I would have rectified it.’ (Street & Street 1989)
(25) ku
CLF:ANIM
thina-thi-dha-ya
2SGS.HEAT(27).DEON-cook-DEON-DM
‘Why don't you cook it?’ (Nordlinger & Caudal 2012:108)
It follows from this distribution that -dha cannot encode any temporal
information, as it marks both present and past paradigms.
In turn, this implies that all the relevant position 1 markers must have
temporal anchoring; for instance, TAM2 must encode a past anchoring, and
TAM6 a present anchoring. This would mean that position 1 cannot be a purely
modal position, and it must be at least temporo-modal (and possibly aspectuotemporo-modal). In the best-case scenario, we could argue that -dha is only
endowed with aspectual meaning. But then, this cannot be true of all elements
in position 6: some must have modal meaning, if we insist on treating them as
semantically autonomous morphemes – this is notably the case of the
apprehensive suffix -nukun, cf. (26). This is a serious problem if we want to
main some kind of ‘composite mood marking’ analysis, which requires a
compositional analysis of TA/M interaction at the morphology to semantics
interface. Both position 1 and position 6 appear to play in determining
temporal (or aspectuo-temporal) and modal meaning. Additionally, if we take
-dha to have a purely aspectual meaning, then it becomes really unclear why it
shouldn’t combine with the future classifier stem (TAM3). If we analyze
futures as a presently anchored type of modal of the predictive/doxastic type,16
such a combination should be perfectly licit. But that doesn’t seem to be the
case, and semantically, this does not seem to make sense. This, too, is a serious
problem for a ‘composite mood’-style, compositional approach to the
distribution of position 1 vs. position 6 TAM markers.
(26) ke-nhi-bath-nukun!
3SGS.poke:RR(21).APPR -2SGO-cook-APPR
‘It might burn you!’ (Nordlinger & Caudal 2012:84)
Last but not least, in combination with negation the following uses are
attested (all other combinations are unattested in Rachel Nordlinger’s field
data): TAM4 + -dha can convey negative past events (‘you didn’t P’) OR
16
This the currently predominant assumption concerning the formal/theoretical analysis of
futures, cross-linguistically; cf. e.g. (Bochnak 2019; Cariani 2021) (as opposed to
‘temporal’ analyses, such as e.g. (Kissine 2008)), see also clear diachronic facts such as
the development of Romance futures, where historically present imperfective
morphology seems to be re-entrant in their development. This even still transparent to
speakers in contemporary French.
20
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
negative admonitives (‘you shouldn’t have P’) – cf. (27), and its (a) vs. (b)
readings.
(27) marda the-na-mut-tha
palngun.
(Murrinh-Patha)
NEG
2SGS.POKE(19).PSTIRR-3S.M.BEN-give-PSTIRR
FEMALE
a.‘You didn’t give him that girl.’
b.‘You shouldn't have given him that girl.’ (Nordlinger & Caudal
2012:106)
-
TAM6 + -Ø conveys prohibitive (‘don’t P’!) or negative deontic (‘you
shouldn’t P’)
(28) mere
thu-ngi-bat-∅!
2SGS.SLASH(23).IRR-1SGO-hit-IRR
‘Don't hit me !’ (Nordlinger & Caudal 2012:107)
NEG
-
TAM6 + -nukun conveys single clause deontic apprehensives (‘don’t do
P [or else Q]’, with undesirable possibility Q left implicit)
TAM6 + -nu conveys negative future (predictive modal)
The fact that negation combines with certain paradigms to convey meanings
normally associated with other paradigms (for instance, NEG + TAM6 + -nu
conveys future meaning, which should be expressed by NEG+TAM3 + -nu), is
also highly suggestive that positions 1 and 6 do not contribute meaningful,
separate TAM morphemes, but really constitute discontinuous morphs very
much those found in Iwaidja and Anindilyakwa.
2.4 Interim conclusion
The above review of the TAM systems of Iwaidja, Anindilyakwa and
Murrinhpatha seems to suggest that they cannot be characterized as
illustrating ‘composite mood marking’ in the sense proposed in (Verstraete
2005; Verstraete 2006). While some markers (e.g. -dha in Murrinhpatha, or the
Anindilyakwa present vs. past suffixes) appear to have some kind of
transparent function in marking particular types of meanings (imperfective
meaning for -dha, present vs. past temporal anchoring for the Anindilyakwa
suffixes), one cannot argue for a principled, general compositional analysis of
the underlying TAM morphology as a whole. Notably because it is impossible
to characterize the relevant discontinuous positions in the verb templates of
these languages as purely aspectuo-temporal (or even aspectual) vs. modal.
3. Periphrastic modal inflections: a first glimpse into the modality /
tense-aspect interaction
In addition to their polysynthetic modal inflections, at least some of the
languages of our sample appear to offer what seems to constitute periphrastic
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
21
modal inflections. This will offer us our first glimpse into how tense-aspect and
modal information interact w.r.t. morphology in these languages.
3.1 Iwaidja periphrastic inflections
Recently collected data17 indicates that Iwaidja possesses constructions that
appear to form a periphrastic modal system complementing its wellestablished inflectional system. (Pym & Larrimore 1979) only identifies
particles maju and mana (which they respective treat as an adverbial vs. a
conjunction; while maju is clearly a preverbal particle, mana can be either a
particle (preverbally) or an adverb (postverbally) depending on its syntax).
Maju is most striking with respect to its distribution with tenses; with the
future, it can take on an evidential/pretense reading (‘X looks like P’) (29), or
an open volitional reading (30). With either the anterior (31)-(32) or past
counterfactual (33), it can take a past volitional/avertive reading; it can even
combine with the present tense to form what seems to be a present avertiveconative (‘trying to do something, but so far in vain’) inflection (36).
Interestingly, while past avertives (31)-(33) seem to correlate with a past
proximative-volitional modal (‘tried to’ translations are an implication of the
volitional component of meaning, ‘nearly V-ED’ a variant of the proximative),
it seems difficult to characterize it in aspectual terms; one could argue that it
derives from a past imperfective viewpoint meaning, but at the same time,
whatever modal state held in the past, it was soon interrupted.
(29) maju
MOD
abana-wirradbi
1SG>3SG.FUT-knead-FUT
‘I’ll pretend to knead it.’ (Pym & Larrimore 1979:244)
(30) maju
abana-marta
nganduwulang
MOD
1SG>3S-FUT-save.for-FUT 1SG-3SG=PERS-mother
‘I’m going to save (some) for my mother’ (Pym & Larrimore 1979:238)
(31) malany maju nganba-ldakani-ny
Why
MOD
3F.SG>1SG.ANT-make.sad-ANT
baraka, ngaldalmalangkajangkaj.
DEM
1SG.PR-feel.betrayed.by.spouse -PR
ngara
1SG.PR-go-PR
‘Why has she tried to/did she want to make me sorry, I’m (already)
feeling sick in the stomach about what she did.’ (Iwaidja dictionary)
(32) maju
birdirlkbu-ny.
Nganduka a-bi-ny?
3SG.ANT-struggle.free-ANT INT
3SG.ANT-do-ANT ?
‘He tried to struggle free but in vain.’ [= ‘and to what effect?’]
(Iwaidja Dictionary)
MOD
17
Long-term fieldwork was conducted by P. Caudal and R. Mailhammer on Iwaidja modal
categories in 2013-2014, 2018-2019, and 2023, thanks to funding through a variety of joint
projects.
22
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
(33) maju
ngan-ambija-na
MOD
1SG.PCF-laugh-PCF
‘I was going to laugh (but I didn’t’)’ (Pym & Larrimore 1979: 76)
The above data demonstrates that maju-based periphrastic modals can have
a variety of temporal anchorings. The fact that maju can combine with the
future inflection (itself arguably a present modal inflection in isolation, with
predictive-doxastic, futurate and deontic/directive uses18) to convey an
evidentiality-related,19 pretense-play use or a present volitional/futurate use,
contrasts with its appearing in combination with the past counterfactual
inflection to encode avertives (31) (where speaker bemoans a vain attempt at
causing him to feel bad, as he was already feeling bad), including without
volitional content. Note that the very peculiar nature of at least some of the
meanings encoded by those <maju + inflection> combinations, particularly the
pretense-play, rather suggests that they are conventionalized uses, not
straightforwardly and compositionally derivable from the particle and the
inflection: indeed, maju does not seem to have a very clear meaning on its
own.20 (33) demonstrates that maju hasn’t retained an volitional content in
such structures, and that indeed, the combination of maju plus the past
counterfactual inflection has been reanalyzed as a general avertive inflection
with past proximative/futurate flavour (and not a volitional-only avertive, i.e.
a frustrative in (Caudal 2023)’s terms). Given the latter fact, it does not seem
reasonable to assume that maju-based periphrastic structures can be regarded
as compositional.
From particle maju and ngamin (the first person of the present tense of the
‘say’ root), a complex particle ngamin maju with non-compositional, arbitrary
meanings (no saying event is ever involved)21 has been construed. Like maju, it
18
We are here following a now predominant theoretical approach to futures, which does not
treat them as temporal but as modal inflections, whereby so-called ‘temporal’ uses
involve a predictive/doxastic modal meaning, anchored in the present (vs. in the past
for ‘future-in-the-past’ uses); cf. e.g. (Cariani 2021).
19
Pretense-play uses of modal/evidential forms are crosslinguistically well-known, and also
found in Australian languages – including Iwaidja in our sample. See also Mparntwe
Arrernte, whose evidential/doxastic particle akwala also has pretense-play uses. Cf.
(Caudal, Henderson & Faller 2011). Pretense-play uses of modals/evidentials can be
analyzed as instances of deliberate ‘self-deception’ (Gendler 2007) – see e.g (Kaiser
2022) and (Caudal, Henderson & Faller 2011) for independently proposed, similar
analyses.
20
While (Pym & Larrimore 1979: 76–77) gloss maju as ‘intent’, they provide examples where
maju cannot have a volitional meaning (73). The Iwaidja dictionary describes its lexical
meaning as being obviously difficult to pin down, listing multiple, rather unrelated
modal meanings: ‘perhaps, hopefully; intention’.
21
Moreover, ngamin maju being perceivable as a present tense-marked structure, its
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
23
is compatible with both past and present inflections, whether indicative or
modal. In combination with the anterior inflection (i.e., an aspectually
underspecified past) (34) and the past counterfactual inflection (35),22 it
encodes so-called ‘mistaken thoughts’ (or ‘mistaken beliefs’, cf. (McGregor
2023)), a postmodal reading with an actualistic content insofar as it entails an
actual negative past event.
(34) maju ngamin
MOD1
mardngalk
COV.drown
ba
warrki
the man
ruka rtadbihi
that creek
angkuwany
ANT.3SG.drown.ANT
‘I thought that he was drowning at that creek’ [but he didn’t]
[speaker’s translation]
(TAIM 190604MM_Modality_2 00:31:00.772 –
00:31:27.717)
(35) ngamin maju
nani-ldalku-nyi
arlirr
3SG>3SG.PCF-cut-PCF
tree
‘I thought he was gonna cut that arlirr’ [but he didn’t]
[speaker’s translation] (TAIM-20230709-Avertivity#2 – 00:08:24.881 00:08:26.913)
MOD
An interesting fact is that according to dictionary data, both maju and
ngamin maju can combine with the so-called ‘root modal’ inflection23. With
maju, this seems to encode some kind of present avertive, (36); the
combination with the root modal makes sense insofar it has capacity readings.
With ngamin maju, the resulting combination seems to involve some kind of
‘wishful thinking/hope about P’, attitudinal meaning24 – possibly with a
doxastic modal dimension, as it seems to encode a belief that some possibility
might materialize. It has an implicative meaning comparable to English verb
hope (hence its rendering as ‘hope, hopefully’) – this could explain why it
combines with a capacitative inflection, as ‘hope’ verbs are known to select for
specific moods/modal marking (Silk 2018).
(36) maju
MOD
aldakijba-Ø
1SG.IRR-pass-IRR
lda kalmu
CONJ HIGH.QUANTITY
mudika
caR
combination with past modal inflections (cf. (35)) would be temporally problematic.
22
(35) is the only occurrence of that complex modal particle with reverse order maju ngamin.
This might be consequence of the fact that ngamin is clearly perceived by speakers as an
inflected verb, so that maju being a particle, could appear on its left.
23
It is unclear why we find another root modal marking on ganang-urrwu (3FSG>3SG.RMOD-seeRMOD). This could be a matter of agreement within a complement clause, or clauselinking construction under a perception verb.
24
One of our language consultants confirmed that ngamin maju could combine with the root
modal inflection, but it was a bit unclear what the intended meaning was in his mind.
24
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
‘I'm trying to pass, but there's too much traffic.’ (Iwaidja dictionary).
(37) ngamin maju
nani-winybu-Ø...
but arlarrarr
3SG>3SG.RMOD-wash-RMOD
but no
‘I thought he was going to wash it but ..karlu [no]’ [speaker’s
translation] (TAIM-20230709-Avertivity#1+ANG-V 00:35:59.425 00:36:10.713)
MOD
(38) ja-wani
ju-ka-n
alan, ngamin maju
1SG.DIST.PR-sit-PR 1SG.DIST.PR-look-PR
ROAD MOD
ana-yanyjing
ana-wulaku-Ø.
1SG>3SG.RMOD-see-RMOD 3SG.RMOD.PROX-come.down-RMOD
‘I'm sitting there looking at the road, hoping to see someone come
down.’ (Iwaidja dictionary) [= hoping I might be able to see
someone]
(39) nga-wani
ng-uka-n
jumung,
ngamin maju
1SG.PR-sit-PR
1SG.PR-WAIT-PR
OBL.3SG,
MOD
an-aya-nyjing
ganang-urrwu.
3FSG>3SG.RMOD-see-RMOD
3FS.G>3SG.RMOD-burn-RMOD
‘I'm sitting here waiting for her. Hopefully I might be able to see her
lighting the fire.’ (Iwaidja dictionary)
It should be noted that if assumed maju and ngamin maju to
compositionally combine with inflections in the above configurations, some
of them would be difficult to reconcile with a unique semantic content
applicable to these particles in all of their uses. For instance, maju and ngamin
maju cannot be taken to encode past avertive meanings, as they both also
appear in structures with a non-actualistic, non-postmodal, present modal
reading. The shift from a present modal to a past postmodal, (negative)
actualistic readings of both mjau and ngamin maju could be attributed to the
inflections they combine with in such readings, but the specifics of said
reading are difficult to predict on the basis of a monosemous reading ascribed
to each of these particles – e.g., why does ngamin maju + RMOD encodes the
reading it has? Also, and quite tellingly, our language consultants rejected the
combination of ngamin maju with the future inflection; this does not seem
semantically explainable in a straightforward way, since maju can combine
with the future (30). Or to put in a nutshell – it seems quite plausible that most,
if not all of the above instances of periphrastic modals, have an arbitrary
semantics, alongside with conventionalized association with certain
inflections.
Let us now turn to two other particles capable of combining with past
inflections to encode negative actualistic, avertive readings, namely wurrkany
and wartuj. Wurrkany (‘was about it/seemed to’) only marks modal utterances
with a past temporal anchoring – in effect a past postmodal, avertive reading.
It preferentially associates with verbs in the past counterfactual (PCF), cf. (40),
and sequence-of-tense effects can appear on serial verbs combined with
wurrkany, cf. (41). It can combine with untensed adjectival forms (cf. burruli
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
25
‘good’ in (42)), in which case it encodes a mistaken belief/perception sort of
reading (‘X looked Adj, but wasn’t Adj’). We suspect this reading obtains in
combination with stative verbs as well, but this needs to be investigated more
closely. Additionally, it can combine with a verb in the anterior tense (ANT) (an
English simple past-like past tense) and still produce an utterance avertive
reading (Kuteva 1998) (43).
(40) wurrkany nanilda
ba walij ba
MOD
3MSG>3SG.PCF-eat-PCF DET food CONJ
karlu
riwany
NEG
3MSG>3SG.ANT-eat-ANT
‘[The dog] looked (deceptively) like he was going to eat the food, but
he didn’t eat it’ (TAIM_190604MM_Modality_1.eaf@ 00:30:21.204)
(41) wurrkany nanimalanma
janara
MOD
3SG>3SG.PCF-drive-PCF
3SG.PCF.DIST-go-PCF
‘He was going to/tried to go fishing [by driving] [but he didn’t go]’
(TAIM_190604MM_Modality_1.eaf@00:14:58.321
(42) wurrkany ruka mudika
burruli
MOD
that
car
good
‘This car looked deceptively good [as it's broken now]’
(TAIM_190604MM_Modality_2.eaf@00:05:23.344)
(43) wurrkany awukung
ba walij rardudban
MOD
1SG>3SG.ANT-GIVE-ANT DET FOOD 3MSG>3SG.ANTLEAVE.BEHIND-ANT
‘I tried to give him food but he left it behind.’
(TAIM_190604MM_Modality_1@27:43)
Wurrkany does not combine with verbs in the future inflection (pace claims
made in the Iwaidja dictionary by B. Birch); speakers systematically rejected
our attempts, and corrected the corresponding utterances to a past
counterfactual form, cf. (44) and (45). The reason for this incompatibility
might be temporal, as in contrast, our language consultants appear to accept
combinations of wurrkany with the future inflection, and endow the resulting
structure with an avertive meaning.
(44) A: Can you say ‘*wurrkany
banimalamanma [mudika]?
MOD
3SG>3SG.FUT-drive-FUT [a car]
(TAIM 190604MM_Modality_1 – 00:13:41.731 - 00:13:44.631)
B: wurrkany nani-malamanma [informant corrected linguist A]
MOD
3SG>3SG.PCF-drive-PCF
‘he nearly drove/wanted to drive/tried to drive the car’
(TAIM 190604MM_Modality_1 – 00:14:53.741 - 00:14:57.271)
26
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
(45) A: Can I say ‘*nanguj
wurrkany banildalkung
[arlirr]?’
yesterday MOD
3SG>3SG.FUT-cut-FUT [a tree]
(TAIM-20230709-Avertivity#2 – 00:09:21.980 - 00:09:27.352)
B: wurrkany
corrected A]
nani-ldalku-nyi
arlirr [consultant
3SG>3SG.PCF-cut-PCF
TREE
‘he nearly cut/wanted to cut/tried to cut the tree’ (TAIM-20230709Avertivity#2 – 00:09:27.376 - 00:09:29.684)
MOD
Wartuj appears to have a negatively oriented semantics close to that of
wurrkany, with a distinct negative evidential twist, as it mostly expresses (i) a
doubtful possibility (46)-(47) or (ii) that something deceptively looked like it
was going to happen, but did not happen (48). Wartuj differs from wurrkany in
that it can combine with the future inflection, and can have present modal
readings. For want of more detailed data we will not say more here; however it
should be noted that this an additional periphrastic modal with postmodal,
(negative) actualistic readings – and contributes to making this type of
phenomenon extremely salient in the Iwaidja grammar of modality.
(46) wartuj
yabanara...
mana
3SG.DIST.FUT-go-FUT maybe
'Maybe he's going to go'
(TAIM-20230709-Avertivity#2 – 00:12:46.407 - 00:12:49.280)
MOD
(47) wartuj
yabanara
or imalda
yawaran
3SG.DIST.FUT-go-FUT ALREADY 3SG.DIST.ANT-GO-ANT
‘Maybe [doubtful] he's going… or he's already gone’ [speaker’s
translation]
(TAIM-20230709-Avertivity#2 – 00:13:03.142 - 00:13:04.625
00:13:06.186 - 00:13:07.846)
MOD
(48) wartuj
naniwuni
[murlk] lda karlu. (Iwaidja)
MOD
3SG>3SG.PCF-hit.kill-PCF [fly]
CONJ NEG
‘It [deceptively] seemed he was going to kill the fly…but no’
(TAIM-20230709-Avertivity#2 – 00:15:06.083 - 00:15:07.708)
To summarize the most important observations in this subsection, it
appears that Iwaidja offers a very rich series of periphrastic modal paradigms,
where it is impossible to tease apart modal and aspectuo-temporal meanings.
One cannot associate either of these types of meanings with separate
morphological units. On the contrary, from the onset, at least some of these
modal particles (cf. wurrkany) being morphologically past,25 the resulting
25
Wurrkany could well be derived from an -ANT inflected verb form, possibly cognate with
burrka- ‘dream’. Interestingly, ‘dream’ verbs are common crosslinguistic lexical cradles
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
27
periphrases had a rigid temporal anchoring on top of having various types of
modal/postmodal meanings. Indeed, even synchronically, wurrkany is
homophonous with a verb form - the third singular anterior (past) form of
burrka- ‘dream’; if it is actually cognate with it, this might explain its
‘unreal/misguided perception’ readings. Numerous actualistic readings –
mostly negative ones, plain avertives, or ‘mistaken belief/thoughts’ with an
avertive flavour – were identified in our survey of periphrastic modals in
Iwaidja. This is perfectly in line with hypothesis that aspect (here, imperfective
aspect, as ‘proximative’ renderings are common in speaker’s translation, cf. e.g.
(35), (37), (40), (47), a.o.) – not just tense – must play a key role in the semantics
of foreclosed counterfactuals in Australian languages.
The morphosyntactic and semantic properties of wurrkany are probably the
most significant to our analysis. They very clearly support Osgarby’s (2018) idea
that non-Pama-Nyungan inflectional modals derive from morphologized
modal particles/clitics via lexically separate ‘auxiliaries’ (rather than a single
verb template, contra e.g. (Evans 2003b)). But they also suggest that Osgarby’s
analysis cannot apply as is to Iwaidjan languages, as wurrkany forms a clearly
past modal periphrasis. In order to apply to an item like wurrkany, Osgarby’s
theory should be amended so as to incorporate temporal parameters in the
original meaning associated with such an auxiliary. The systematic pastness of
wurrkany periphrases contrasts with the temporal variation observed with
maju, ngamin maju and wartuj-based periphrases, where both present and
past-anchored readings can be found.
3.2 Murrinhpatha periphrastic modals
According to (Mansfield 2014: 446), Murrinhpatha also exhibits
combinations of modal particles or clitics with inflections – which seem to
constitute potential periphrastic modal inflections. The most striking of those
is clitic =nukun, which also appears in its morphologized form as suffix -nukun
(position 6 of the verb template) as part of the discontinuous apprehensive
inflection (in combination with the TAM6 ‘future irrealis’ exponent in position
1). Very strikingly, as a position 6 exponent, -nukun appears both on verbs
conveying apprehensive-epistemic clauses (it then associates with an implicit
order) and deontic-epistemic apprehensive clauses (it then associates with
implicit negative possibility; it conveys an order and a threat, cf. English don’t
you dare P!). In the latter case, it is claimed in (Nordlinger & Caudal 2012;
Mansfield 2014) that the apprehensive inflection must be associated with a
negation. Rather strikingly, clitic =nukun can directly mark negation mere in
(51), which achieves a reading apparently identical to that of its morphologized
realization in (50)). This is a perfect illustration of the frequent ‘fluidity’ of the
delineation between particles /clitics vs. affixes (or morphological exponents
on the verb template itself).
for modal/evidential/mirative idioms, cf. Engl. ‘pinch me I’m dreaming!’, Fr. ‘mais je
rêve!’ (negatively oriented mirative) vs. ‘pincez-moi je rêve !’ (positively oriented
mirative), etc. See also (Delancey 2012), which mentions the use of mirative forms in
Kham when reporting dreams.
28
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
Semantically speaking, if the apprehensive inflection clitic or suffix can
mark either deontic-apprehensive mono-clausal structures, or epistemicapprehensive mono-clausal structures, then its modal meaning is
underspecified. But if it must associate with an overt negation and negative
directive/prohibitive meanings (and cannot encode ‘positive directive
imperative meanings) in mono-clausal deontic-apprehensive uses, then these
cannot be compositional: we are then here faced with an entrenched type of
meaning. Finally, it should be stressed that the distribution of -/=nukun in
illustrates the evolution cline whereby particles can become clitics, and then
morphologized into affixes; cf. (Osgarby 2018).
(49) ke-nhi-bath-nukun!
3SGS.poke:RR(21).APPR -2SGO-cook-APPR
(Murrinh-Patha)
‘It might burn you!’ (Nordlinger & Caudal 2012:84)
(>Implicit order: ‘don’t touch it!’)
(50) mere na-ngi-mathputh-nukun=thurru
(Murrinh-Patha)
NEG 2SGS.HANDS(8).APPR-ISGO-interrupt-APPR=2SGS.GO(6).APPR
‘Don’t you continually [go(6)=’keep.on’] interrupt me.’
(>Implicit threat: ‘or I’ll punish you’) (Nordlinger & Caudal
2012:104)
(51) mere=nukun
thurru
(Murrinh-Patha)
NEG=APPR
2SG.GO(6).FUTIRR
(Mansfield 2014:446)
‘you better not go!’
(>Implicit warning: ‘or you’ll regret it’)
3.3 What the literature on the tense-aspect interaction can tell us about
the tense-aspect/modality interaction w.r.t. periphrastic vs. synthetic
forms
At this stage, to further make sense of our observations about synthetic vs.
periphrastic modal inflections in our sample, it is useful to actually step out,
and replace our investigations within a broader Australian and typological
picture. The first important fact to note, is that crosslinguistically, periphrastic
forms are often (and even most generally) non-compositional from the
perspective of the morphology to semantics interface.26 From this, it would
naturally follow that synthetic forms derived from such periphrastic forms are
even less likely to be compositional w.r.t. the morphology to semantics
interface, as they have undergone further re-analysis. We believe that this
generalization makes it reasonable to hypothesize that compositionality is at
best an accidental outlier in such morphologisation scenarios.27
26
For further considerations about analogies between the morphology to semantics interface
of periphrastic TAM forms and that of polysynthetic ones, see (Caudal 2022a).
27
Note that even if (a) the initial periphrastic form could be deemed compositional (which, to
be fair, seems very unlikely, given that most originate in constructions / collocational
structures, which, by definition, have diverged from a normal, compositional use,
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
29
Considering the literature on discontinuous TAM forms in general,
including e.g., periphrastic tense-aspect forms, could be helpful to clarifying
the matter at stake here. We believe that the literature of discontinuous tenseaspect forms, because it is so much more developed in many respects, can help
us grasp why the general assumption that all types of discontinuous TAM
inflectional forms (including modal forms) should ideally be analyzed
compositionally, is maybe not as wise as it may seem.
To understand this, let us turn to morphologically discontinuous perfects,
and tenses deriving from such perfects. This comprises so-called ‘perfectivized’
perfect inflections, such as the French passé composé (Caudal & Roussarie
2006; Caudal 2015) or German Perfekt, or even the perfect in some varieties of
English. Their meaning has expanded towards perfective viewpoint uses, cf.
(Squartini & Bertinetto 2000)’s notion of aoristic drift. There is an increasingly
large body of evidence showing that Romance and Germanic perfects
originated in a variety of constructions whose meaning was noncompositional (as it was a matter of lexification/entrenchment) from the
onset, and that their subsequent semantic changes involved various reanalysis
processes (Öhl 2009), possibly furthering their non-compositional nature (cf.
e.g. (Pinkster 1987; de Acosta 2006; Bourova 2007; Haverling 2010a; Haverling
2010b; Öhl 2014; Pieroni 2016), where multiple syntactic and semantic variants
of said construction are listed).28 They ended up being morpho-syntactic atoms
towards some kind of entrenched, lexicalized meaning) and (b) if its morphologisation
as (poly)synthetic morphology preserved said compositionality, there is no guarantee
that it will not be whittled away into non-compositionality through subsequent
evolutions. Semantic change is a constant in the evolution of tenses; it is very unlikely
that in due time, it would not affect its purported ‘compositionality’. A good case in
point for this, is the development of perfective readings for perfects: these are obviously
difficult to reconcile with the (rather common place) idea that some element in said
perfect should be endowed with a resultative meaning.
28
Romance perfects can be argued to have in fact derived from a complex network of resultative
possessive constructions; while it has often been claimed that these ultimately derive
from a precursor possessive constructions with habere and should therefore be regarded
as initially compositional, detailed corpus work has proven this hypothesis to be
incorrect: (i) a wide array of formally and semantically distinct constructions already
existed in Latin (Acosta 2011; Acosta 2013; Hertzenberg 2015), and (ii) these did not derive
from a possessive reading of habere, but from a more convoluted and complex
development path, rather involving a set of attained-state constructions (not possessive
constructions). And interestingly, the same appears to hold true of Germanic ‘have’
perfects cf. e.g. (Fischer 2020) and (Johannsen 2016), including the English perfect (Acosta
2013) – again, a complex network of constructions with many lexical collocational effects
(see e.g. (McFadden 2016) and (Alexiadou & McFadden 2006) for complexity found in the
development of ge- particle morphology in Germanic languages) could be observed,
without any possessive dimension – possibly coined by analogy with other attained state
constructions based on other verbs Through ‘gang effects’, reanalysis (and corresponding
semantic change), those multiple resultative constructions gradually ‘merged’ into more
30
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
with a discontinuous realization – ‘minimal signs’ in the sense of (Blevins
2016). And most importantly, their semantics continuously evolved through
time – in terms of e.g. their interaction with various Aktionsart parameters
(Alexiadou & McFadden 2006; McFadden 2015; Rebotier 2017) or agentivity (cf.
(Carey 1994) w.r.t. subject agentivity marking the real shedding line between
Old English resultative constructions, and a nascent perfect gram). The
semantic evolution of tenses is even mirrored in their ability to coerce some
aspectual types of verbs (Caudal 2020) – where it exhibits numerous signs of
being arbitrary in nature, as said changes appear to be ‘collostructional’
(Stefanowitsch & Gries 2003) in nature. None of the above changes can be
evidently ascribed to an alleged morphological sub-element of said forms; it
should obviously affect the entirety of the analytical tense – and if change is
driven by collostructional, then it involves a collocation with particular lexical
classes (which makes it even less amenable to a matter of compositionality
within the analytical structure itself).
Yet, in spite of these rather overwhelming diachronic observations, there
has been a steady (though limited) flow of compositional analyses of perfects
over the past decades (cf. e.g. (Grewendorf 1995; Musan 2001) for the German
Perfekt, (Caudal 2015: 182) for the French passé compose, (Klecha 2016) for the
English perfect, or most recently, (Wegner 2019; Zhao 2022) for a variety of
perfect forms). Some works seem to happily turn a blind eye to the increasing
large set of evidence that perfects are probably best analyzed as noncompositional forms, and do not involve smaller meaningful morphological
elements. Others assume that reanalysis somehow prompted a
decompositional reanalysis, where various sub-elements are endowed with
particular meanings. See (Öhl 2009: 294) for a discussion focusing on the
Perfekt, and (Caudal 2015; Wegner 2019) for diachrony-informed, or even
diachronic accounts perfects along such lines. And even if subsequent changes
alter the meaning of a well-established analytical tense, one could postulate
that the semantic change should be ascribed to a particular formal ingredient
in some compositional analysis, as was done in e.g. (Caudal 2015). However,
both those moves somehow seem self-serving, in the sense that they look like
stipulations, supporting in an aprioristic manner a compositional analysis of
analytical tenses; compositionality is regarded as a priori desirable for its
elegance.29 And when semantic change seems to affect only certain
collocations/constructions, i.e., exhibits clear evidence of being arbitrary,
or less united have perfect paradigms, with a novel conventional meaning. Some perfects,
in spite of over two thousand years of evolution, still exhibit striking conservative
features; cf. e.g., the persistence of agreement properties of the ‘avoir’ passé compose in
Modern French (where agreement with the object is warranted by its appearing before
the auxiliairy, in line with its deriving from an attributive construction).
29
Such an aprioristic view is extremely commonly defended among semanticists, especially
formal semanticists – ‘sense enumeration’, or ‘homonymy’ has had a bad name in the
domain for several decades, see e.g. (Pustejovsky 1995). For a specific illustration of this
belief concerning perfects, see (Klein 2000: 362), and for a detailed discussion of how
problematic such a view can be, see (Caudal 2018b)
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
31
assuming a compositional analysis seems clearly unjustified (see e.g. (Rebotier
2017; McFadden 2015)). Or to put it in a nutshell: diachrony very much stands
in the way of compositional approaches to analytical tenses.
Much the same can be said of the development of e.g., Romance synthetic
futures and conditionals, which derive from Latin constructions having
morphologized. Interestingly, in spite of the existence of an obvious reanalysis
step in their evolution (when from being analytical forms, they became
synthetic), some works nevertheless claim that they exhibit semantic
compositionality. This is for instance what (Iatridou 2000; Camussi-Ni 2006)
claim about the French conditionnel présent (‘present conditional’), at least to
some extent; its imparfait-like ending should contribute aspectuo-temporal
information similar to that of the imparfait itself. Thus, constituting an
instance of ‘fake tense’ for Iatridou, as it would need to give rise to present
modal meanings in combination with modal verbs (cf. il devraitCOND être malade
meaning ‘he should be sick’, which is a present epistemic modal). As shown in
(Caudal 2018a) though, such an analysis faces insurmountable empirical
obstacles, as the ill-named conditionnel présent also marks some structures
with past modal uses, notably of the ‘future-in-the-past’ type (53) – including
with non-epistemic modal verbs (54), in contrast with (52). Note that here too,
semantic change can proceed in a collocation-driven, arbitrary manner, so that
assuming a compositional account of change changes seems impossible ; see
e.g. (Caudal 2017) for further observations along these lines.
(52) Marie
devrait
Marie have.to-CONDPR.3SG
‘Marie should go.’
partir.
leave-INF.
(53) Jean déclara
que Marie tomberait
malade.
Jean declare-3SG.PS that Marie fall-CONDPR.3SG sick
‘John declared that Marie would become sick’.
(French)
(French)
(54) Jean
déclara
que Marie devrait
partir.
(French)
Jean
declare-3SG.PS that Marie have.to-CONDPR.3SG leave-INF
‘John declared that Marie would have to go.’
There is, in fact, widespread crosslinguistic evidence for modal inflections to
exhibit mixed present/past readings, and for a general evolution cline from
past modal readings to present modal readings, generally affecting deontic
uses of said modals first. Such semantic discrepancies are very problematic for
a compositional conception of the tense-aspect/modality interaction.
For some general observations concerning this cline, and in general the
tendency of past modals to drift towards present uses, see (Hogeweg 2009;
Patard 2019); this evolution appears to start most generally with
deontic/interactional deontic uses of modal, due to ‘politeness effects’ of
priority modals being used with past marking, but with a presently valid
relevance. I.e., it is a common byproduct of ‘politeness’ uses of past priority
32
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
modals; for a detailed description and account of a ‘politeness’ use of a priority
modal, see (Caudal 2017).30
While similar diachronic evidence w.r.t. the morphological and semantic
evolution of modal inflections is of course not accessible for Australian
languages, some of the facts uncovered here could be of import to attempts at
reconstructing the evolution of modal inflections, at least for non-PamaNyungan languages. Thus, (Osgarby 2018) argues that Ngarnka derived its
current modal inflectional paradigms from the morphologisation of modal
particles or auxiliary, in a preverbal position. He tentatively suggests that such
an evolution cycle could be found in other Mirndi languages, and even be
common across non-Pama-Nyungan languages. This obviously makes sense
for all languages possessing so-called ‘pronominal portmanteaus’ on some left
slot of their verbal morphology, if said portmanteau also plays a part in
encoding TAM information – regardless of whether or not said portmanteau
must ‘team up’ with a verb-final TAM suffix.
Now if our observations about Iwaidja periphrastic modal forms are correct,
these would naturally look like precursors to Osgarby’s morphologisation
cycle: more specifically, they would constitute analytical modal inflection
prior to modal particles/auxiliaries being ‘fused’ with the initial slot of the verb
template. Then it would follow that in such a subsequent stage, regardless of
whether or not the tense-aspect / modality interaction was compositional at
the periphrastic stage, tense-aspect information would end up being
constrained by exponents found in two distinct slots in the verb template. As
reanalysis is obviously required to transition from the periphrastic to the (poly)synthetic stage of the inflection, there is no reason to postulate that a previous
compositionality would be preserved. Which, again, given what we know from
the semantic diachrony of discontinuous inflections, sounds very unlikely in
the first place.
While such a temporal disparity is not obviously the case for any form in our
sample (it comprises several instances of temporally underspecified modal
forms, cf. the Anindilyakwa USP (Ø) paradigm), Mawng – a language closely
related to Iwaidja – exhibits datapoints strikingly similar to that of the French
conditionnel, in that it illustrates the very same tortuous, partial evolution from
past to present modal meanings. Mawng possesses an irrealis inflection (the
so-called ‘irrealis 2’, glossed I2 below) whose various readings do not have the
same temporal anchoring. Most strikingly, its deontic readings are all present
(55), whereas its semantics is otherwise relatively similar (and seems to be
formally cognate with) the Iwaidja past counterfactual – it notably has both
avertive (56) and past counterfactual readings (57).
(55) nuyimung
30
anng-arntakpu-ni mata warlk.
(Mawng)
Quite significantly, (Caudal 2018a: 58–59) observes that emphatic ‘politeness’ uses of the socalled past conditional in French, started emerging at the end of the 18th century; these
uses have now become fairly widespread to mark actual polite uses of priority modal
verbs and constructions in French, as present conditional have become the expected
inflectional marking of such structures (so that a novel emphatic present modal
structure was needed, due to linguistic erosion).
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
2SG.CONTR
2SG-chop.down-I2 VE
‘Now you chop the tree’.
(56) ja
tree
(Singer 2006: 113)
karrkpin
ja
jalakaraj
big
MA
fishing.spear
‘She tried to spear it with a big spear’ (Mawng)
MA
33
ing-errka-nyi.
3FE/3MA-spear-I2
(Singer 2006: 63)
(57) Kun-pu-ni
ma-warlka-nya.
(Mawng)
1SG/2SG-hit-I2
VE-fall-I2
‘I would have knocked you down’ (if I was a little younger). (Mawng)
(Singer 2006: 148)
Such a fact is impossible to reconcile with a compositional analysis of the
Mawng inflections.31 And it strongly suggests that as a rule of thumb, temporal
anchoring should be regarded as a potential matter of uses of modals (or modal
constructions), rather than necessarily as an inherent, rigid semantic feature
of a particular modal inflection. This is why providing a thorough description
of TAM systems, including in their distribution with particles, and also
investigating their lexical/constructional quirks (in effect, some frozen
structures which they happen to mark, and which many scholars nevertheless
try to explain on a compositional semantic basis), is essential in order to make
certain we are drawing appropriate conclusions as to the tenseaspect/modality interaction in any given language. Our description of Iwaidja
modal markers (be they synthetic or periphrastic) seems to suggest that they
are either rigidly past (or at least foreclosed), present, but that none of them is
currently undergoing a temporal shift in their semantics. At the same time,
some facts uncovered in our exploration of periphrastic modal inflections in
this language rather suggests a pattern of conventionalized TAM agreement.
For instance, it is rather striking that the future in Iwaidja seems to be unable
to combine with modal particles apparently associating with rigid past
temporal anchoring. However for want of space (and for want of sufficiently
clear data for some combinations of forms), we will leave this issue open to
future research.
4. Theorizing the interaction between aspectuo-temporal meanings and
modal meanings of modal inflections in our sample (and in other
Australian languages)
The more theoretical part of our paper will try and determine how aspectuotemporal and modal meanings effectively combine to construe the meanings
31
Moreover, the Mawng portmanteau prefix contrasts future-marking prefixes with
realis+irrealis marking prefixes (which can be either past or present); it can also bear
additional present tense marking. A ‘composite mood marking’ account of the Mawng
TAM system therefore seems difficult to adopt anyway, as this portmanteau must
encode both modal and temporal information
34
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
one can ascribe to modal inflections in the languages of our sample, and
possibly in other non-Pama-Nyungan / Australian languages.
4.1 Some (meta-)theoretical preliminaries
While this paper cannot cover such a wide-ranging and complex issue as the
role played by temporal and aspectual parameters in the semantics of modal
inflections, we will here offer some tentative insights concerning their role in
the semantics of Australian modal inflections (and in particular those found in
non-Pama-Nyungan languages).
As is well known from the theoretical literature (see e.g. (Mizuno &
Kaufmann 2022) for a quick overview), two major approaches to the role of
temporal parameters w.r.t. counterfactuality have been so far put forth:
(i)
the so-called ‘Past-as-Modal’ approach, which comes in several brands.
The most famous version of this approach is undoubtedly the so-called
‘Fake-Past’ whereby past conditions in modals play a modal part by
‘excluding’ some proposition from accessible worlds – it conveys in effect
some of kind of modal remoteness condition; pioneered by (Iatridou
2000) in the formal literature, it is actually a fairly old idea (an early
instantiation of the concept can be found in (Damourette & Pichon 1911)’s
‘toncal’), but has receive several fairly precise technical treatments, cf. e.g.
(Schulz 2014; Mackay 2019). Other, alternatives of this type include e.g.
(Cipria & Roberts 2000) (which take the Spanish imperfetto to denote a
modal, whose denotation can be trivialized to a temporal expression), or
more recently (Karawani, Kauf & Zeijlstra 2019) (which adopt a
technically different, but analytically similar approach, whereby tenses
essentially makes reference to world-time pairs (and possibly involving
the actual world), but not the time of utterance; as a last resort, if a tense
is not embedded under any modal context, past tense will get indexed to
the actual world w0, and will receive a straightforward past
interpretation).
(ii) the so-called ‘Past-as-Past’ approach, whereby past conditions in modals
typically signal some actually past world, where the modalized
proposition and its relevant possible world of validation might (still) have
been accessible; or, under a less ‘realistic’ understanding of pastness, as
involving some manner of temporal shift.
(Formal and) theoretical works focusing on the interaction between tenseaspect and modality in conditional-counterfactuals, tend to assume that either
type of approach should hold about all flavours of modals.
At this stage, it could be worthwhile observing that a considerable number
of theoretical and formal analyses of modality:
(i) primarily focus on a well-documented European language – even though
there is an already considerable and ever-growing body of e.g. formal
semantic analyses of modals in endangered/minorized/under-described
languages based on first hand descriptive work in the field (see e.g. work
by L. Matthewson and past and present associated), this is not your
mainstream theoretical paper on modality;
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
35
(ii) when works actually focus on language diversity, and take on a
comparative or typological perspective (such as in (Verstraete 2005;
Verstraete 2006; Van Linden & Verstraete 2008a)), they generally do not
investigate a substantial part of the grammar of modality in each studied
language – or at least, do not investigate various types of modal bases in
a systematic way.
As a result, empirical bases of numerous theoretical works remain narrower
than they should be, in a double sense.
This is potentially very problematic, as shown in (Mizuno & Kaufmann
2022). It is argued in this work that desiderative modals with an optativecounterfactual meaning (expressing wishes) with past marking in Japanese,
require a ‘Past-as-Past’ approach in a way quite different from the manner in
which temporal marking seems to interact with e.g. English counterfactuals
such as (1)-(2) – this indicative of the fact that much of the discussion about
these two approaches could benefit from exposition to a greater empirical
coverage of relevant facts. (Mizuno & Kaufmann 2022) suggest that a lack of
concern for the diversity of linguistic facts both language-internally, and
crosslinguistically, can have a very significant impact on our understanding of
the semantics of modality in general, and of the tense-aspect / modality
interaction in particular. We couldn’t agree more, of course; our investigations
need to be descriptively wide-sweeping both language internally and language
externally. The empirical and theoretical coverage (three languages with an indepth analysis of some facts, combined with wide-ranging typological and
even diachronic observations) of the present paper is an attempt at finding
some way of striking a compromise solution between both types of coverage,
while keeping our investigations manageable within the confines of a single
paper. As we will see, this broad coverage will lead us to notably conclude that
temporality has been given too much prominence in too many works w.r.t. our
understanding of foreclosed counterfactual forms, and aspect not enough
4.2 Aspect: an important, and under-rated parameter in the study of the
tense-aspect/modality interaction
It is striking to note that in the theoretical literature on modality, temporal
conditions have received a disproportionate importance in the investigation
of foreclosed vs. open counterfactual meanings. It is notably significant that
many works descriptively refer to utterances like (1) as ‘double past
counterfactuals’; the label is unfortunate, as in effect, it rather combines an
aspectual gram – a perfect – with a past tense.
It is worth recalling now that Australian languages do not exhibit such
‘double-past’ marking; indeed, as a consequence, they tend not to have socalled ‘fake past’ marking (i.e., past marking of an otherwise open
counterfactual modal meaning; cf. (58), where the possibility of the relevant
vehicle is a open possibility at speech time, where the relevant evaluation
modal event is anchored in the present)32 ; the semantic representation of such
an utterance arguable offers a present temporal anchoring.
32
It must be recalled that we assume modals to be event predicates with stative variables, à la
36
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
(58) “If it [the vehicle] broke down we would probably just have to patch it
back together with duct tape.” (“Richardsons keeping Starbuck on ice
– Our Communities”, Winnipeg Free Press, 10.06.2011)
So given the (apparently widespread) absence of so-called ‘fake past’ in
Australian languages, it seems natural that many descriptive and theoretical
works focusing on foreclosed counterfactual meanings in said languages,
ascribe a central role to temporality/pastness – for past modals seem to be
overwhelmingly ‘truly’ past, and to have foreclosed meanings. This is most
obvious in (Verstraete 2005; Verstraete 2006; Van Linden & Verstraete 2008a),
as we will see.
However, outside of the Australianist community, a substantial number of
scholars now object to temporality-centered theories of modality (including
theories resorting to the concept of ‘fake past’), arguing that aspect too should
play an important role in the semantic analysis of modal forms, whether
foreclosed or not. See e.g. (Homer 2011; Ferreira 2014; Homer 2021), or to a lesser
extent, (Halpert & Karawani 2012).33 In particular, it is widely hypothesized in
such works that ‘run-of-the-mill’ modal meanings require imperfective
aspectual conditions – which is reflected in the crosslinguistic frequency of
imperfective forms, or aspectually underspecified forms, marking modal
forms, including complex modal constructions such as e.g. conditional or
biclausal counterfactuals of various types; cf. e.g. (Boogaart & Trnavac 2011;
Ferreira 2014). In contrast to such commonplace modal meanings, postmodal
readings of modal forms are arguably marked – and are the most revealing
readings w.r.t. the importance of aspect in the semantics of modal forms. The
most pervasive and well-known type are so-called ‘actuality entailment’
readings (59) (Bhatt 1999; Mari 2016; Homer 2021), and avertive readings of
modal verbs verbs (60) – which convey in fact, inactuality entailments, (Caudal
2023: 166–167)), i.e. entailments that some negative event actually took place
(we are here assuming the existence of negative events in our ontology,
following (Bernard & Champollion 2018)). In Romance and Germanic, such
readings of modals overwhelmingly associate with perfective viewpoint
marking – and a perfective viewpoint semantics.
Could related phenomena also cast some doubt on the efficaciousness of
temporality-centered accounts of modality in Australian languages?
Remarkably, while actuality entailments appear to be fairly rare in the
grammar of Australian languages (or to be limited to some very specific
modals, such as purposives and apprehensives). inactuality entailments of
many types (especially those associated with avertive meanings) are very
(Ferreira 2014; Homer 2021) – this is the event here mentioned.
33
We are assuming that ‘fake past’ is nothing more than a morphological reflex without any
semantic significance in languages exhibiting it, and is entirely due to the normal
evolution cycle of past modal forms towards present meanings, cf. e.g. (Patard 2019). For
want of space, we will not discuss here the role played by so-called ‘fake aspect’ (see e.g.
(Grønn 2013)) in relation to ‘fake past’ in some theories, but it certainly reflects on the
realization that aspect does matter in the tense-aspect/modality interaction.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
37
common. (Caudal 2023: 169–170) attributes this contrast to the fact that
Australian languages do not exhibit the type of perfective viewpoint marking
of modal structures found in e.g. Romance (and possibly altogether lack
‘strong’ perfective viewpoints, as found in e.g. Romance languages, and only
possess tenses with ‘weak’ perfective readings, in the sense of (Martin &
Demirdache 2020); cf. (Caudal 2022b)).
(59) Il a
pu
partir.
(French)
He have.PR.3SG
be.capable.PP
go-INF
‘He was able to leave.’ (= he managed to leave OR was allowed to
leave OR seized an opportunity and left)
(60) Il a
voulu
partir.
(French)
He have-PR.3SG
want-PP leave-INF.
‘He tried to leave (and failed)’ (lit.: ‘he wantedperfective to leave’)
It is also significant that much like in many other parts of the world, one can
reconstruct recurrent patterns of (especially past) irrealis marking formally
deriving from past imperfective markers in Australian languages (cf. e.g.
(Caudal 2023: 157, note 87)). Murrinhpatha itself is quite significant of this very
pattern, as its morphology appears to bear lingering traces of such an origin.
The-dha exponent re-entrant in past imperfective, past irrealis and present
deontic/directive forms, is obviously connected with imperfective morphology
being a common denominator to all these forms.
By itself, the general role of imperfective morphology in the development of
irrealis forms is hardly surprising, given that this is crosslinguistically common.
But it is probably relevant as to why Australian modal inflections lack actuality
entailments: as their aspectual components are morphologically ‘frozen’, but
commonly reflect on former imperfective, their semantics is more likely to be
imperfective. And even more importantly, in Australian languages where
modality is mostly inflectionally encoded, one cannot morphologically
construe aspectual contrasts between e.g., imperfective modal inflections, and
perfective modal inflections – unlike e.g. Romance modal verbs, which can
receive an overt past perfective or past imperfective marking. This is an
important argument against a true ‘composite mood marking’ approach to
modality in Australian languages: aspect as a grammatical category cannot
play any part in their modal structures because these are primarily based on
inflections, rather than verbs (or at least so-called ‘auxiliaries’ or other lexicogrammatical modal categories with a significant lexical verb-like content) –
while vice versa, such an approach appears far more applicable to e.g.
,Romance modal verbs.
Now the reason why Australian languages present so many inactuality
entailment readings probably stems from the fact those easily associate with
imperfective viewpoint meanings (contrary to actuality entailment readings,
which seem to have clear affinities with perfective viewpoint meanings).
Indeed, inactuality entailments are crosslinguistically common with overt
imperfective viewpoint marking. They are even found in e.g. Romance
languages – this is notably the case in Romanian, where a bona fide
semantically avertive construction with imperfective marking can be found
38
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
(Pahonțu forthcoming). It will become apparent in the remainder of this
paper, just how important postmodal meanings (especially inactuality
readings of modals) are in our own study of the semantics of modal inflections
in Iwaidja, Anindilyakwa and Murrinhpatha – and that such data poses
significant challenges to temporality-based accounts of modality, including to
J.C. Verstraete’s accounts.
4.3 Discussion of Verstraete’s (2005, 2006) and Van Linden and
Verstraete’s (2008) temporal account of irrealis interpretations in
Australian languages
Let us now turn to a discussion of (Verstraete 2005; Verstraete 2006; Van
Linden & Verstraete 2008a)’s semantic and pragmatic theory of the interaction
between tense-aspect and modality in Australian languages – for simplicity of
reference, we will simply refer to the ‘V&V account’, as the three relevant
references have a lot in common, theoretically. Our critique will tap three
particular areas where we believe potential trouble awaits: (i) the manner in
which ‘negative entailments’ are rigidly associated with pastness in Australian
modals (we will see that this causes serious issues of empirical adequacy) by
the V&V account and (ii) the absence of any function ascribed to aspect in the
semantic analysis of Australian past modals in the V&V account. If our above
meta-theoretical musings are founded, then aspectual meaning should matter
in the semantic modelling of Australian modals, including in our sample.
The V&V account is based on four main empirical generalizations
concerning foreclosed counterfactual meanings in Australian language, where
those meanings can be encoded via four means:
a. Dedicated morphology (but how dedicated is actually somewhat unclear)
– this is deemed rare, on the basis of the samples investigated by V&V
b. A past tense marker (perfect or perfective, according to V&V – although
it’s not entirely clear whether they actually mean what is generally meant by
these terms in the mainstream aspectual literature)
c. A purely aspectual marker – which in fact, has a crucially temporal effect,
according to V&V (it is taken to be a proxy for a temporal marker)
d. Some combination of a modal element and a past or past-inducing tense
marker (perfect or perfective) – this is the most common pattern in V&V’s
sample; so double marking, but no ‘double past’ marking for foreclosed
counterfactuals in Australian languages.
The interpretative part of the V&V account focuses on pattern (d),
highlighting the importance of temporal parameters (again, aspect is very
much reduced to its temporal effects). It is crucially based on the idea that
counterfactuality originates as an implicature (or entailment, V&V are not
quite clear about this)34 derived from the combination of modal and past
34
Note that this obviously a question relating to what is known as ‘pragmatic intrusion’;
(Chierchia 2004)’s concept of locally computed implicatures would be a natural
explanation for the difficulty to identify these phenomena as properly semantic, or as
pragmatic. See (Lee 2008) a for a detailed discussion.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
39
temporal conditions – aspectual conditions are ignored or reduced to
temporal conditions, and so-called perfect or perfective markers are taken to
essentially contribute a pastness condition. Relevant implicatures are
generated qua ‘Horn scales’ (Horn 1989) – i.e., as scalar implicatures (Carston
1998). In a nutshell, Horn scales predict that given expressions appearing on an
informative scale of strength, Grice’s Maxim of Quantity will trigger some kind
of implicature (or entailment), such that some weaker expression
entails/implicates that some stronger expression does not hold – hence the
negative entailment/implicature in (61). As according to V&V, past modal
utterances are informationally weaker than their non-modalized, indicative
past tense-marked counterparts, they entail/implicate that said past-marked
indicative counterpart does not hold true, i.e., this gives rise to the negative
entailment/implicature (62).
(61) some<all
THEREFORE
(62) a. John should have left < John left.
b. John should have left ® ¬John left.
some® ¬all
THEREFORE
But why is pastness so central to the V&V account of counterfactual modals?
Because pastness is taken to be a necessary ingredient for negative entailments
such as (61). Observing that the present modal utterance (63) does not entail
the negative present indicative utterance (64) – contrary to the past modal
utterance (65), which entails indicative negative past utterance (66) – (Van
Linden & Verstraete 2008b: 1879) conclude that such negative
entailments/implicatures are only part of the interpretative content of modals
with past meaning/anchoring – not of that of modal forms with present
meaning/anchoring.
Two problems are already worth noting at this stage: (i) (64) is in fact a
modal utterance (as it receives a proximative/futurate reading, which arguably
constitutes a bona fide type of modal meaning; cf. e.g. (Copley 2009)), and (ii)
(64) and (66) involve different aspectual parameters. As a result, the
connection (63) and (64) do not differ from (65) and (66) on mere temporal
grounds, and the conclusion drawn is not warranted and (Van Linden &
Verstraete 2008b) fails to establish that pastness plays a key role in the
pragmatic explanation it puts forth. Intuitively, part of the problem lies in the
different semantic roles played by Aktionsart parameters in present vs. past
indicative tenses bearing on telic utterances – a difference not found in present
vs. past modal utterances. Had V&V used atelic utterances (e.g. Jack should be
sick / Jack is sick vs. Jack should have been sick / Jack was sick), the problem
might have gone unnoticed. We will get back to this question further down in
our argumentation, but it already suggests that aspect needs to play a role in
our understanding of the semantics of present vs. past modals, and that it
obviously plays different roles in modal vs. non-modal forms. But an
immediate side effect of this is that there is in fact no way we can construe
evidently parallel non-modal vs. modal utterances for all Aktionsart types of
verbs – so that the very empirical foundations of the V&V account is already in
jeopardy; temporality alone cannot explain everything.
(63) Jack should come to the party.
40
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
(64) Jack is not coming to the party.
(65) Jack should have come to the party.
(66) Jack did not come to the party.
But let us put this problem aside for the sake of pursuing our critique. To
substantiate the idea that pairings of non-modal and modal utterances can
form the type of quantity implicatures shown in (62), (Van Linden & Verstraete
2008b: 1877) resorts to the following set of examples in order to demonstrate
the existence of a scale of certainty, ranging from (67) (indicative utterance) to
the weakly potential (69) (modal utterance conveying a mere epistemic
possibility) – the crucial point being that (67) is stronger (‘more certain’) than
either (68) or (69).
(67) John is coming.
statement of certainty ; present indicative
(68) John must come. statement of strong potentiality
(69) John may be coming
(necessity)
statement of weak potentiality (possibility)
We here find again some of the issues already identified with (63)-(66). First,
like (64), (67) (i) stands out by its aspectual marking and (ii) is in fact a modal
utterance: regardless of whether it has a progressive reading or a proximative
one, (67) indicates that in some possible, ulterior world, John is expected to
have reached whatever location is involved in the current deictic centre. In
other words, (67), (68) and (69) are all modal utterances, albeit with different
modal flavours (and some modal strength differences). The use of an indicative
tense marking does not necessarily imply an absence of modal meaning.
Therefore, the above data does not warrant at all the validity of the purported
‘scalar implicature’ assumed in (62) between modal and non-modal
utterances: (67)-(69) does not oppose a non-modal utterance with two less
informative modal utterances (68)-(69). One could finally observe that it is not
very clear that (67) is less certain than (68), if (67) is taken to express that John
intends to come.
Furthermore, the very notion of certainty as applied by V&V to both modal
and non-modal expressions is debatable. The non-modalized assertion of P in
some perfective, past tense does contrast with some past modalized possibility
or necessity assertion of P, but in terms of actuality, not certainty. In other
words, asserting P in a perfective tense makes P actual and irrevocable – this is
not an epistemic notion, pace the V&V account.
But the most serious challenges faced by the V&V account lie elsewhere: the
account seems to make incorrect predictions as soon as we start introducing
some empirical complexity. Thus, once we start looking at a broader set of
modal meanings than V&V did to ground their theory of scalar implicatures
stemming from past modals, it becomes obvious that the type of negative
entailment/implicature illustrated in (62) does not always hold – far from it.
Should have V-ed actually involves a rather peculiar type of
reproachative/admonitive (Olmen 2018) modal meaning (not a plain past
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
41
deontic meaning), whose relations to negative entailment are dependent on
the modal flavour at stake. The following sequence of examples is quite
revealing of the problem:
(70) John had to leave.
(‘was under the obligation of leaving, but
did not necessarily leave’; past deontic)
(71) John had to leave.
entailment reading)
(‘was compelled to leave’; actuality
(72) John could have left.
leave)
(foreclosed counterfactual: John did not
(73) John was able to leave.
(‘actuality entailment’ reading: John left)
(74) John would have gladly left.
not leave)
(foreclosed counterfactual: John did
A first issue is immediately visible in (70): it does not evidently entail either
John left or John did not leave – and this contradicts the V&V account. Worse
yet: while (72) and (74) entail John did not leave (which is line with the V&V
account) (71) and (73) entail John left – which, if we apply the V&V account to
French, results in an under-generation problem (the V&V account only
predicts negative entailments for modals, not positive ones).
This series of examples is enough to disprove the notion that past modals
systematically entail the negation of the corresponding utterance in the past,
i.e., the fundamental tenet behind the V&V account, connecting pastness
(temporality) with foreclosed counterfactual meanings. The above datapoints
suggest that negative vs. positive (or the absence of any entailment of that
type) depend on the lexical semantics of particular modals, as well as their
interaction with temporal and aspectual conditions (and marking).
It is important to note that so-called ‘actuality entailments’, cf. (59), or
‘inactuality entailments’ cf. (60) readings of e.g. French modals clearly
illustrate how aspectual meaning can play a key role in such matters. Indeed,
if we mark these examples with a past imperfective tense instead of a past
perfective one, then the observed actuality/inactuality entailments vanish –
(75)-(76) entail neither ‘he left’ nor ‘he did not leave’. Such datapoints clearly
suggest that aspectual parameters should not be overlooked, or reduced to
temporal ‘proxies’ (as is the case in the V&V account), in a theory of the
interpretation of past modals.
(75) Il
pouvait
partir.
(French)
He have.PR.3SG be.capable.IPFV go-INF
‘He had the ability/had been given permission to leave.’
(76) Il
He
voulait
want-IPFV
partir.
leave-INF.
(French)
42
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
‘He was experiencing a desire to leave.’35
Related issues were mentioned in (McGregor 2009). It was observed in this
paper that the V&V account could not properly explain why some Nyulnyulan
past modal inflections encoded no uncertainty about some proposition P
holding in the past. And that on the contrary, such utterances semantically
encoded that ‘not P’ was the case. As a result, such utterances cannot be
informationally ‘weaker’ than a non-modalized/non-modal past utterance.
McGregor illustrates this problem with an avertive-denoting utterance (77).
Crucially, (Van Linden & Verstraete 2008b: 1879) did cite a similar
(Gooniyandi) avertive example in support of their account. This means that
they would claim that (77) has a negative interpretation because it is less
certain (i.e., informationally ‘weaker’) than a past indicative, non-modalized
assertion corresponding to (78). Such an analysis is evidently misguided; (77)
has a negative semantic content, as it is an instance of semantic avertive
(Caudal 2023) – (77) is not semantically ‘less certain’ than (78), but it is
contradictory with it. Avertive utterances involving a semantically avertive
inflection like in (77) conventionally involve actual, negative past events
(albeit possibly in a secondary dimension of meaning, as presuppositions or as
conventional implicatures à la Potts (2007)). (77) and (78) utterances are
inherently contradictory, without there being any need to invoke a Horn scale.
It should be noted that the label ‘inactuality entailment’ might seem a tad
misleading; avertives like (77) semantically encode that some negative event
actually took place. Following (Bernard & Champollion 2018), we will regard
such negative events as ontologically existent, but endowed with a negative
polarity. In other words, ‘inactuality entailments’ are in fact the negativelyoriented counterpart of actuality entailments; they are also an actualistic type
of (post)modal statement, albeit one bearing on a negative event.
(77) miliyarri nga-l-jamba-na
kinya
juurru ngayu-na
(Warrwa)
long.ago 1MIN.NOM.IRR-step-PST this
snake I-ERG
‘I nearly stepped on a snake [but I didn’t]’
(McGregor 2009: 158)
(78) I stepped on a snake.
Moreover, and even more problematically, McGregor (2009) notes that
under its epistemic-evidential uses, the past ‘subjunctive’ inflection found in
Gooniyandi entails that some proposition P actually holds in the past. In
contrast, the V&V account incorrectly predicts that (79) should entail ‘they did
not eat here’. This is a similar problem to the V&V account incorrectly
predicting (if applied to French modals) that actuality entailment readings
should not exist – but with datapoints in an Australian language.
35
While we agree that this translation is sub-optimal, it highlights the fact that (76) reports a
truly imperfectively-viewed mental state/desire.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
43
(79) Ngab-ja-wirra
ngamoo-nyali
(Gooniyandi)
eat-SUBJ-3PL.NOM/PST+A before-REP
‘They were apparently eating here not long ago’ (McGregor 2009:
160)
Interestingly, a similar problem arises with actualistic interpretations of
apprehensive modals with a past interpretation. While such postmodal
readings seem to be rare, they are attested in at least some languages where
apprehensive modals with a doxastic-epistemic reading can be found – note
that the apprehensive modal inflection in Yidiny is temporally underspecified;
in (80), it interpretation must comprise a past temporal condition, so the V&V
analysis should apply – and incorrectly predicts that (80) entails ‘I did not fall’:
36
(80) ngayu
bama
wawa:l
wandanḑinggu.
(Yidiny)
I-SA
person-ABS
see-PST
fall-APPR-ERG
‘I saw the person as I was (unfortunately) falling over’. (Dixon 1977:
352)
Last but not least, the V&V account (Van Linden & Verstraete 2008b: 1878)
also makes a problematic claim in relation to the impact of negation on the
above purported ‘certainty scales’. In particular, it is argued that negation
simply preserves those scales, (81)-(84):
(81) potential p < p
(82) potential p ® ¬p
(83) potential ¬p < ¬p
(84) potential ¬p ® ¬(¬p), i.e. potential ¬p ® p
However, as was shown in (81) about Murrinhpatha, non-Pama-Nyungan
past irrealis inflections (including those found in Iwaidja and Anindilyakwa)
often exhibit a striking ambiguity when combined with negation, where the
relevant readings are mutually contradictory, cf. (85). This means that
negation would not systematically preserve those so-called ‘certainty scales’
(regardless of whether or not they are legitimate in the first place); on the
contrary, reading (b) is the equivalent of an actual negative past event, i.e. of a
36
Past narrative, actualistic uses of purposives found in numerous Australian languages might
constitute another similar class of problematic datapoints for the V&V analysis, but we
will leave this as an open question for future research. One might also take into
consideration past dispositional modals/habituals – which are generally assumed to
constitute a class of modal meanings (see e.g. (Carlson & Pelletier 1995) for a standard
modal quantificational analysis, and (Cohen 2012) for a more innovative one). It is
unclear how the V&V account would handle such modals.
44
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
non-modalized negative past assertion (which, again, is suggestive that no
such ‘certainty scales’ are involved in the interpretation of modal inflections in
Australian languages).
(85) marda the-na-mut-tha
(Murrinhpatha)
NEG 2SGS.POKE(19).PSTIRR-3S.M.BEN-give-PIMP
a.‘You didn’t give him that girl.’
palngun.
female
b.‘You shouldn't have given him that girl.’
To conclude our little critique, it appears that the V&V account (i.e., (Van
Linden & Verstraete 2008b) and its precursors (Verstraete 2005; Verstraete
2006)) is plagued with numerous empirical adequacy problems, and is
theoretically flawed in several important respects:
(i)
it is dubious that modalized and non-modalized assertions form
any kind of scale upon which quantity implicatures could be built
in relation to temporality alone, as past non-modalized utterances
are not certain, they are actual (i.e. pas and irrevocable, which
modalized propositions are not); moreover, parallel modal/nonmodal utterances cannot always be found without resorting
aspectual markers (which can make it impossible to actually find
actually acceptable correspondents)
(ii)
it is incorrect to posit that PAST(MOD(P)) in general entails ¬P; this
depends on lexical modal meanings, and how some positive or
negative actualistic readings (actuality vs. inactuality entailments)
may have conventionalized with certain aspectuo-temporal
conditions/markings (and also on contextual parameters, in case
some modal is not conventionally biased towards either P or ¬P
being implicated); the V&V account is clearly unable to explain the
contrast between actuality vs. inactuality entailment patterns, and
any form of past modal without any negative
entailment/implicatures (cf. e.g. (79)).
(iii)
it is incorrect to assume that negation preserves the connection of
modal forms with negative entailments/implicatures (and of course
that it preserves ‘scales’ connecting modalized vs. non modalized
statements, as said scales are non-existent anyway), in the light of
NEG + past irrealis marking being ambiguous in numerous nonPama-Nyungan languages;
(iv)
combining modal and temporal parameters is obviously
insufficient when trying to account for the semantics of modal
forms – aspectual parameters must also come into play (and not
merely because they can impact temporal anchoring; imperfective
viewpoint meanings have special affinities with modal meanings,
which obviously need to be paid attention to).
We believe that point (iv) should be really crucial to a proper semantic
theory of the tense-aspect/modality interaction.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
45
4.4 Theoretical and formal consequences for modelling tenseaspect/modality in languages of our sample
Now that we have established that aspect does indeed play a role in
determining the semantic behavior of Australian past irrealis forms (or
foreclosed counterfactual forms, if you will), what could we possibly conclude
from this about the semantic makeup of modal inflections in our sample?
Prior to this, we have established that the discontinuous TAM morphology
found in our sample could not be analyzed in the compositional terms of
‘composite mood marking’ à la V&V. But does this mean that modal and
aspectuo-temporal conditions need to be treated as unanalyzable semantic
atoms? We believe that this is not the case. Indeed, if synchronic
‘morphological atomicity’ is warranted for such forms, this does not mean that
their semantics should not comprise distinct aspectuo-temporal and modal
ingredients. And the fact that the V&V account fails at explaining how and why
pastness seems to play an important role in encoding foreclosed
counterfactuality, does not mean that past temporal anchoring should not be
a necessary semantic ingredient for modal inflections – or at least for some of
them.
According to V&V, Australian languages do not exhibit instances of
presently-anchored instances of otherwise ‘past’ modal inflections – contra e.g.
‘single past’ counterfactuals (2)-(3), or even (arguably) ‘two past’
counterfactuals involving individual-level states (1). We will not venture here
far enough in the realm of hypothetical conditional to ascertain whether (1)(3) can have correspondences in Australian languages.
But there is one important theoretical move we can make, namely that
tense-aspect conditions appear to systematically outscope modal conditions
in the semantic representation ascribed to the denotation of modal inflections
in our sample. This means that for these languages, a scope hierarchy such as
(86) must hold between aspectuo-temporal and modal operators/functions in
the semantic representation ascribed to modal inflections (whether synthetic
or periphrastic).
(86) Aspectuo-temporal conditions > modal conditions > lexical
conditions
How general (86) should be, is an obviously delicate theoretical question:
should it apply to all types of modal meanings, i.e. flavors? In particular, could
it apply to non-root modal meanings? There is indeed a vast body of literature
arguing that epistemic modals should not be subjected to such a scope
hierarchy, as they should outscope tense; see e.g. (Hacquard 2006) for
references and a detailed review of the classic literature on this topic. While
settling such a mightily complex question cannot be achieved within the
confines of our paper, and inspired by descriptive and theoretical insights
found in works such as (Homer 2013; Rullmann & Matthewson 2018; Homer
2021), we will suggest that (86) should also apply to non-root modals. Our main
motivation can be found in the numerous ‘mistaken thoughts/beliefs’ readings
associated with a variety of modal forms in our sample (and in Australian
languages in general, see e.g. (Caudal 2023; McGregor 2023)). Those meanings
obviously correlate with evidential-epistemic flavors. And being past and
46
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
actualistic, they must require some manner of past aspectuo-temporal
‘binding’ the modal flavor they derive from.
5. Conclusion
We must now conclude our comparative study of modal inflections in
Iwaidja, Anindilyakwa and Murrinhpatha.
Let us turn first to the negative side of our results: hypotheses and ideas we
have refuted. We have established that Verstraete’s concept of ‘composite
mood marking’ could not be applied as is to these three languages, as they do
not exhibit the type of compositionality in the tense-aspect/modality
interaction it would require. Of course, one cannot a priori exclude the
existence of bona fide composite mood-marking in non-Pama-Nyungan
languages, i.e., languages where real compositionality involving two separate
tense-aspect and modality morphemes would take place. But we have
suggested that it is rather unlikely, given the highly inflectional encoding of
modality in these languages. It could more fruitfully apply to languages with
overt, semantically productive tense-aspect marking of modal
verbs/auxiliaries such as Romance languages, than to Australian languages.
Our detailed review of the various discontinuous paradigms found in the
languages of our sample, and of the various semantic reasons why they cannot
be regarded as supporting a ‘composite mood marking’ compositional analysis,
remains our first and most solid argument against such a view. Even if we leave
aside obviously non-compositional paradigms (e.g., if their prefixed/suffixed
elements are unique), why certain combinations of prefixes/suffixes (or more
to the point, exponents found in the left vs. right positions in the verb
template) are not possible is generally not explainable on semantic grounds.
Nor can we explain numerous readings associated with certain discontinuous
paradigms in a compositional manner. This is an important methodological
point to us: synchronically, one cannot make any informed hypotheses about
the tense-aspect/modality interaction, if one does not consider a TAM system
globally.
In lieu of ‘composite mood marking’, we would like to speculate that
discontinuous TAM morphs as well as periphrastic TAM inflections (modal
inflections in our data) found in our sample, could be better thought of as a
matter of conventionalized pairings of exponents – collocations, if you will; see
(Bonami 2015). One could for instance compare them to so-called asymmetric
marking on conditional structures (Molencki 2000); they have been analyzed
as a matter of a purely formal, conventionalized marking in works insisting
upon the fact that they were compositional, while they really are the ‘frozen’
end-product of diachronic processes.37 While French conditional structures
are a perfect test-bench for this type of approach, cf. (Caudal 2018a; Patard
2019), Australian languages also have potential in this respect, since they also
37
Although one could always claim that reanalysis could introduce compositionality where
there was none, initially, but this might well be compositional wishful thinking:
reanalysis can be an entirely formal, without any significant semantic change – at least
at first.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
47
exhibit features suggestive of the same diachronic processes being at play
(such as, for instance, the tendency for certain kinds of past modals to acquire
present readings over time; cf. the ‘irrealis 2’ in Mawng, (55)-(57)).
We have also shown that (Van Linden & Verstraete 2008b) pragmatic
analysis of foreclosed counterfactual meanings in Australian languages was
highly problematic in several respects: (i) it is dubious that Horn Scales
connecting modal utterances and their non-modal utterances can be invoked
to explain why foreclosed counterfactual utterances entail negation of the
propositional content they mark and (ii) such an approach cannot account for
a variety of phenomena, especially postmodal, actualistic interpretations of
modal inflections (dubbed ‘actuality entailments’, and ‘inactuality
entailments’ above), and is therefore empirically inadequate. We have
suggested that the primary explanation behind this rather radical diagnostic,
is that the theory places far too much emphasis on pastness/temporality alone,
and connects it to negative entailments in a way that is theoretically and
empirically inadequate. It also fails to acknowledge the importance of aspect
while investigating the interplay between tense-aspect and modality – a fact
now well-known from the study of actualistic readings of modal forms, both as
positive ‘actuality entailments’ (Homer 2021) or as negative ‘inactuality
entailments’ (Caudal 2023), as in the case of e.g. avertive utterances. We have
suggested that the extreme prominence of the latter in Australian languages,
has to do with the fact that they can easily emerge from imperfective
viewpoints (an idea already defended in Caudal 2023) and that Australian
pasts irrealis inflections are rigidly associated with an imperfective viewpoint
meaning. Vice versa, the latter fact would explain the near absence of positive
‘actuality entailment’ readings of modal inflections in Australia: those would
require perfective viewpoint meanings that Australian modal inflections do
not really possess.38 Given that Australian languages do not allow for overt,
distinct tense-aspect marking of modals (which are essentially flectional in
these languages), and do not offer compositional tense-aspect/modality
interactions even in their discontinuous TAM inflections, we conclude that
even an aspectually-enriched and revised theory of ‘composite mood marking’
(so not the original Verstraetian analysis, but see e.g. (Ferreira 2014; Homer
2021) for possible alternatives) cannot apply to non-Pama-Nyungan Australian
languages. But, vice versa it would be very-well suited to Romance languages,
as they express modality via still strongly lexical modal verbs. Their numerous
actualistic readings of modal verbs with perfective tense marking seem to be a
perfect playground for such a theory 39
38
How some rare actuality entailment readings of modal inflections nevertheless emerged in
Australian languages, is an independent question we must leave to future research. But
it is possibly connected with some special semantic properties of said modals.
39
This being said, actuality and inactuality entailment readings of such verbs may well involve
conventional implicatures à la (Potts 2007) (cf. (Caudal 2023: 168)), or presuppositions
(Mari 2016), or maybe some more complex semantic configuration, combining
entailments with presuppositions (Homer 2021), so it’s unclear how well this would
connect with these readings being fully compositional. Some caution is probably
48
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
Turning now to the positive results of our investigations, we have uncovered
substantial evidence supporting (Osgarby 2018)’s hypothesis that non-PamaNyungan combinations of prefixed (or near prefixed) ‘pronominal
portmanteaux’ with TAM suffixes (or other TAM positions on the other end of
the verb template), might be reconstructed as deriving from the joint
morphologisation of respectively preverbal modal auxiliary/particles/clitics
and postverbal modal auxiliary/particles/clitics. Iwaidja was shown to offer
very solid evidence in favor of a similar origin for its TAM prefixes, as it
currently presents a complex system of preverbal modal particles already
forming a periphrastic inflectional modal system, while the Murrinhpatha /=nukun (more or less) verb final affix/clitic, clearly illustrates the postverbal
part of Osgarby’s hypothesis (i.e., that postverbal particles could also
morphologize), cf. (49)-(51).
At the same time, our data demonstrates that quite a few wrinkles (not to
say serious revisions) should be added to Osgarby’s original hypothesis. In
particular, Osgarby did not take into consideration the fact that TAM
information – not just modal information – could be spread across both ends
of the verb; he assumed tense-aspect meanings had to be postverbal. Our
description of wurrkany modal periphrases suggests that this is a necessary
amendment for Iwaidja, even when a preverbal modal particle is
morphologized: indeed, wurrkany appear to rigidly associated with past modal
meanings, and is transparently a past verb form. So diachronically, this means
that it probably must have been perceived from the onset as a combination of
modal and temporal meanings. We cannot stress enough how important this
particular datapoint is, when trying to untangle the morphology to semantics
interface behind TAM forms in some non-Pama-Nyungan languages: it very
strongly suggests that even from their earliest diachronic stages, at least some
languages do not separate aspectuo-temporal and modal conditions on their
‘pronominal-TAM’ portmanteaus. This places (again!) in a very favorable light
the idea that tense-aspect compositionality should not be invoked for such
systems: given the chance, they can (and will) ‘bundle up’ aspectuo-temporal
and modal meanings in a single morphological slot, on either end of the verb.
Of course, as non-compositionality in the tense-aspect/modality interaction is
already visible in the periphrastic, modal particle/auxiliary-based forms such
as those identified in e.g. Iwaidja, it seems very unlikely that actual
compositionality could surface at later stages, in systems derived from such
paradigms – especially if one considers the overall distribution of TAM
exponents in the entirety of a TAM system in any given language, rather than
a couple of forms.
Still on the positive side, another important idea we have tried to push
forward, is that tense-aspect/modality interaction is semantically complex
even for morphologically complex forms, due to numerous datapoints
demonstrating that temporal and aspectual semantic conditions interacted
with modal conditions in the semantics of Iwaidja/Anindilyakwa/MurrinhPatha modal inflections – the abundance of (mostly negative) actualistic
interpretations of modal inflections found in our sample is our best argument
in support of this idea. This is also reflected in the fact that different temporal
advised here.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
49
anchoring can be associated with different modal inflections in Australian
languages (languages in our sample exhibit a lot of rigidly past vs. present
morphemes) – and that a single modal inflection can have different temporal
anchoring (cf. again the Mawng ‘irrealis 2’). We have suggested a relatively
straightforward set of scope relations between tense-aspect and modality,
whereby tense-aspect systematically outscopes modal operators, even in the
case of non-root modal meanings, i.e. epistemic readings.
6.
References
Acosta, Diego de. 2006. <Have + perfect participle> in Romance and English: Synchrony
and Diachrony. Cornell University PhD Thesis.
http://ecommons.cornell.edu/handle/1813/2929. (4 October, 2017).
Acosta, Diego de. 2011. Rethinking the genesis of the Romance periphrastic
perfect. Diachronica 28(2). 143–185. https://doi.org/10.1075/dia.28.2.01dea.
Acosta, Diego de. 2013. The Old English Have-Perfect and Its Congeners. Journal of
English Linguistics 41(1). 33–64. https://doi.org/10.1177/0075424212441706.
Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: Oxford University Press.
Alexiadou, Artemis & Thomas McFadden. 2006. Pieces of the be Perfect in German
and Older English. In Proceedings of the 25th West Coast Conference on Formal
Linguistics, 270–278. Somerville, MA.: Cascadilla.
Arregui, Ana. 2009. On similarity in counterfactuals. Linguistics and Philosophy 32(3).
245–278.
Asher, Nicholas & Alex Lascarides. 2003. Logics of Conversation. Cambridge:
Cambridge University Press.
Auwera, Johan van der & Vladimir Plungian. 1998. Modality’s semantic
map. Linguistic Typology 2(1). 79–124.
Bednall, James. 2020. Temporal, aspectual and modal expression in Anindilyakwa, the
language of the Groote Eylandt archipelago, Australia. Canberra / Paris: ANU &
Université de Paris PhD Thesis.
Bednall, James. 2023. Modality and mood. In Claire Bowern (ed.), The Oxford Guide to
Australian languages, 392–410. Oxford: Oxford University Press.
Bernard, Timothée & Lucas Champollion. 2018. Negative events in compositional
semantics. Semantics and Linguistic Theory 28(0). 512–532.
https://doi.org/10.3765/salt.v28i0.4429.
Bhatt, Rajesh. 1999. Covert Modality in Non-finite Contexts. Berlin, Boston: University
of Pennsylvania PhD Thesis. https://doi.org/10.1515/9783110197341.
Blevins, James P. 2016. The minimal sign. In Andrew Hippisley & Gregory Stump
(eds.), The Cambridge Handbook of Morphology, 50–69. Cambridge: Cambridge
University Press.
Bochnak, M. Ryan. 2019. Future reference with and without future
marking. Language and Linguistics Compass 13(1). e12307.
https://doi.org/10.1111/lnc3.12307.
Bonami, Olivier. 2015. Periphrasis as collocation. Morphology 25(1). 63–110.
https://doi.org/10.1007/s11525-015-9254-3.
Boogaart, Ronny & Radoslava Trnavac. 2011. Imperfective aspect and epistemic
modality. In Adeline Patard & Frank Brisard (eds.), Cognitive Approaches to Tense,
Aspect, and Epistemic Modality (Human Cognitive Processing), vol. 29, 217–248.
Amsterdam: John Benjamins Publishing Company.
https://benjamins.com/#catalog/books/hcp.29/main. (1 October, 2014).
Booij, Geert. 2010. Construction Morphology. Oxford, New York: Oxford University
Press.
50
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
Bourova, Viara. 2007. Les constructions latines Infinitif + habebam vs. Infinitif +
habui et le développement du conditionnel roman. In David Trotter (ed.), Actes
du XXIV Congrès International de Linguistique et de Philologie Romanes:
Aberystwyth 2004, Vol. 2, 2007, ISBN 978-3-484-50502-8, págs. 461-474, 461–474.
Tübingen: Max Niemeyer.
https://dialnet.unirioja.es/servlet/articulo?codigo=5053184. (26 April, 2018).
Bybee, Joan. 1998. Irrealis as a Grammatical Category. Anthropological
Linguistics 40(2). 257–271.
Camussi-Ni, Marie-Armelle. 2006. Analyse formelle et conceptuelle des formes verbales
du français contemporain : A la croisée du passé simple et de l’imparfait, du futur et
du conditionnel, les concepts « ±potentiel » et « ±défini ». Université Rennes 2 PhD
Thesis.
Capell, Arthur & H. E. Hinch. 1970. Maung Grammar (Janua Linguarum, Series
Practica). Vol. 98. The Hague / Paris: Mouton de Gruyter.
Carey, Kathleen. 1994. The Grammaticalization of the Perfect in Old English: an
Account Based on Pragmatics and Metaphor. In William Pagliuca
(ed.), Perspectives on Grammaticalization, 103–117. Amsterdam: John Benjamins.
Cariani, Fabrizio. 2021. The Modal Future: A Theory of Future-Directed Thought and
Talk. Cambridge, U.K.: Cambridge University Press.
Carlson, Gregory & Francis Jeffry Pelletier. 1995. The Generic Book. (Ed.) Gregory
Carlson & Francis Jeffry Pelletier.
Carroll, Matthew Jay. 2016. The Ngkolmpu Language with special reference to
distributed exponence. Canberra: Australian National University PhD Thesis.
https://openresearch-repository.anu.edu.au/handle/1885/116801. (12 March,
2020).
Carston, R. 1998. Informativeness, relevance and scalar implicature. In R. Carston & S.
Uchida (eds.), Relevance Threory: Applications and Implications, 179–236.
Amsterdam: John Benjamins. http://discovery.ucl.ac.uk/22922/. (24 March, 2016).
Caudal, Patrick. 2012. Relations entre temps, aspect, modalité et évidentialité dans le
système du français. Langue française 173(1). 115–129.
https://doi.org/10.3917/lf.173.0115.
Caudal, Patrick. 2015. Uses of the passé composé in Old French: evolution or
revolution ? In Jacqueline Guéron (ed.), Sentence and Discourse, 178–205. Oxford:
Oxford University Press.
Caudal, Patrick. 2017. Les « usages atténuatifs » de l’imparfait entre
conventionnalisation locale et compositionnalité globale – vers une analyse
constructionnelle. In Eta Hrubaru, Estelle Moline & Anca-Marina Velicu
(eds.), Nouveaux regards sur le sens et la référence. Hommages à Georges Kleiber,
179–256. Cluj: Echinox.
Caudal, Patrick. 2018a. Vers une approche constructionnelle des structures au
conditionnel. Langue francaise 200(4). 49–61.
Caudal, Patrick. 2018b. De la théorie du sens, à celle des appariements formes/sens :
synthèse de quinze ans de recherche sur le TAM(E). Université de Paris 7 Mémoire
d’habilitation à diriger les recherches.
Caudal, Patrick. 2020. Coercion for the ages? A thousand years of parallel inchoative
histories for the French passé simple and passé composé. Presented at the LSA
2020 (94th annual meeting of the LSA), “Formal approaches to
grammaticalization” Workshop, New Orleans.
Caudal, Patrick. 2022a. Que peut nous apprendre la morphologie flexionnelle des
langues polysynthétiques australiennes sur la relation temps / aspect ? In Tatiana
Milliaressi (ed.), Typologie des aspects, 193–220. Lille: Presses Universitaires du
Septentrion.
https://www.septentrion.com/fr/livre/?GCOI=27574100044630&fa=author&pers
on_ID=10953.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
51
Caudal, Patrick. 2022b. Culmination/telicity and event delineation in Australian
Languages : phonology, morphosyntax, semantics and pragmatics. In Construire
sur les décombres de Babel (Sciences du langage), 193–227. Saint-Denis: Presses
universitaires de Vincennes. https://doi.org/10.3917/puv.boneh.2022.01.0193.
Caudal, Patrick. 2023. Avertive/frustrative markers in Australian languages: blurring
the boundaries between aspectuo-temporal and modal meanings. In Kasia M.
Jaszczolt (ed.), Understanding Human Time (Oxford Studies of Time in Language
and Thought), 107–173. Oxford: Oxford University Press.
Caudal, Patrick, John Henderson & Martina Faller. 2011. On the Arrernte “quotative”
akwele. Conference talk presented at the 42nd Annual Conference of the
Australian Linguistics Society (ALS 2011), Australian National University,
Canberra.
Caudal, Patrick & Laurent Roussarie. 2006. Brands of perfects : semantics and
pragmatics. In Pascal Denis, Eric McCready, Alexis Palmer & Brian Reese
(eds.), Proceedings of the 2004 Texas Linguistics Society Conference: Issues at the
Semantics-Pragmatics Interface, 13–27. Somerville, MA.: Cascadilla.
Chierchia, Gennaro. 2004. Scalar Implicatures, Polarity Phenomena, and the
Syntax/Pragmatics Interface. In Adriana Belletti (ed.), Structures and Beyond: The
Cartography of Syntactic Structures, Volume 3, 39–103. Oxford: Oxford University
Press. https://doi.org/10.1093/oso/9780195171976.003.0003.
Ciardelli, Ivano. 2022. The restrictor view, without covert modals. Linguistics and
Philosophy 45(2). 293–320. https://doi.org/10.1007/s10988-021-09332-z.
Cipria, Alicia & Craige Roberts. 2000. Spanish Imperfecto and Pretérito: Truth
Conditions and Aktionsart Effects in a Situation Semantics. Natural Language
Semantics 8(4). 297–347.
Cohen, Ariel. 2012. Generics as Modals. Recherches linguistiques de Vincennes (41). 63–
82. https://doi.org/10.4000/rlv.2068.
Copley, Bridget. 2009. The Semantics of the Future. London: Routledge.
Damourette, Jacques & Emile Pichon. 1911. Des mots à la pensée. Essai de grammaire
de la langue française. Paris: D’Artrey, tome V.
De Haan, Ferdinand. 1999. Evidentiality and epistemic modality: Setting
boundaries. Southwest journal of linguistics 18(1). 83–101.
Delancey, Scott. 2012. Still mirative after all these years. Linguistic Typology 16(3).
529–564. https://doi.org/10.1515/lity-2012-0020.
Dixon, R. M. W. 1977. A Grammar of Yidiny (Cambridge Studies in Linguistics).
Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9781139085045.
Eather, Bronwyn. 1990. A Grammar of Nakkara (Central Arnhem Land Coast).
Canberra: Australian National University Ph.D. Thesis.
Egmond, Marie-Elaine van. 2012. Enindhilyakwa phonology, morphosyntax and genetic
position. University of Sydney PhD Thesis.
Evans, Nicholas. 2000. Towards a grammar of Ilgar/Garig.
Evans, Nicholas. 2003a. Bininj Gun-wok: a pan-dialectal grammar of Mayali,
Kunwinjku and Kune. Australian National University, Canberra: Pacific
Linguistics.
Evans, Nicholas. 2003b. Comparative Non-Pama-Nyungan and Australian historical
Linguistics. In Nicholas Evans (ed.), The non-Pama-Nyungan languages of
northern Australia: comparative studies of the continent’s most linguistically
complex region, 3–25. Canberra: Pacific Linguistics, Australian National
University.
Evans, Nicholas & Francesca Merlan. 2003. Dalabon verb conjugations. In Nicholas
Evans (ed.), The non-Pama-Nyungan languages of northern Australia: comparative
studies of the continent’s most linguistically complex region, 269–283. Canberra:
Pacific Linguistics, Australian National University.
52
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
Faller, Martina. 2002. Semantics and Pragmatics of Evidentials in Cuzco Quechua.
Stanford, CA.: Stanford University Ph.D. dissertation (unpublished).
www.mpi.nl/world/persons/private/marfal/talks2.html.
Ferreira, Marcelo. 2014. Displaced Aspect in Counterfactuals: Towards a More Unified
Theory of Imperfectivity. In Luka Crnič & Uli Sauerland (eds.), The Art and Craft
of Semantics Vol 1 (MIT Working Papers in Linguistics 70), 147–164. Cambridge,
MA.: MITWPL.
Fintel, Kai von & Anthony S. Gillies. 2010. Must . . . stay . . . strong! Natural Language
Semantics 18(4). 351–383. https://doi.org/10.1007/s11050-010-9058-2.
Fintel, Kai von & Anthony S. Gillies. 2015. Hedging Your Ifs and Vice Versa. MIT,
Cambridge, MA., ms.
Fischer, Hanna. 2020. The development of the perfect in selected Middle and New
Germanic languages. In Robert Crellin & Thomas Jügel (eds.),Perfects in IndoEuropean Languages and Beyond (Current Issues in Linguistic Theory), 95–122.
John Benjamins Publishing Company. https://doi.org/10.1075/cilt.352.04fis.
Gendler, Tamar Szabó. 2007. Self-Deception as Pretense. Philosophical
Perspectives 21(1). 231–258. https://doi.org/10.1111/j.1520-8583.2007.00127.x.
Grewendorf, Günther. 1995. Präsens und Perfekt im Deutschen. Zeitschrift für
Sprachwissenschaft. De Gruyter 14(1). 72–90.
https://doi.org/10.1515/zfsw.1995.14.1.72.
Grønn, Atle. 2013. Aspect and tense in counterfactual main clauses: Fake or real? In
Folke Josephson & Ingmar Söhrman (eds.), Diachronic and Typological
Perspectives on Verbs, 133–158. Amsterdam: John Benjamins Publishing.
Hacquard, Valentine. 2006. Aspects of Modality. MIT PhD Thesis.
Hacquard, Valentine. 2010. On the event relativity of modal auxiliaries. Natural
Language Semantics 18(1). 79–114.
Halpert, Claire & Hadil Karawani. 2012. Aspect in Counterfactuals from A(rabic) to
Z(ulu). In Jaehoon Choi, E. Alan Hogue, Jeffrey Punske, Deniz Tat, Jessamyn
Schertz & Alex Trueman (eds.), Proceedings of the 29th West Coast Conference on
Formal Linguistics, 99–107. Somerville, MA: Cascadilla Proceedings Project.
Haverling, Gerd V. M. 2010a. Actionality, tense, and viewpoint. In Cuzzolin, Pierluigi
& Baldi, Philip (eds.), New perspectives on historical Latin syntax, vol. 2, Constituent
syntax: adverbial phrases, adverbs, mood, tense, 277–523. Berlin / New York:
Mouton de Gruyter.
Haverling, Gerd V. M. 2010b. Sur l’expression du temps et de l’aspect grammatical en
latin tardif. De Lingua Latina 3(5). 1–23.
Hertzenberg, Mari Johanne. 2015. Habere + pp and the Origin of the Periphrastic
Perfect. Journal of Latin Linguistics. De Gruyter Mouton 14(1). 33–63.
https://doi.org/10.1515/joll-2015-0003.
Hogeweg, Lotte. 2009. What’s so unreal about the past: Past tense and
counterfactuals. In Anastasios Tsangalidis, Roberta Facchinetti & Frank Robert
Palmer (eds.), Studies on English Modality in honour of Frank R. Palmer, 181–208.
Bern: Peter Lang. http://repository.ubn.ru.nl/handle/2066/78856. (10 November,
2016).
Homer, Vincent. 2011. French Modals and Perfective: A Case of Aspectual Coercion.
In Mary Byram-Washburn, Katherine McKinney-Bock, Erika Varis, Ann Sawyer &
Barbara Tomaszewicz (eds.), Proceedings of the 28th West Coast Conference on
Formal Linguistics (WCCFL 28), 106–114. Somerville, MA.: Cascadilla.
Homer, Vincent. 2013. Epistemic Modals : High, ma non troppo. In Seda Kan, Claire
Moore-Cantwell & Robert Staubs (eds.), Proceedings of NELS 40, vol. I, 273–290.
UMass., Amherst: GLSA.
Homer, Vincent. 2021. Actualistic interpretations in French. Semantics and
Pragmatics 14. 12:1-56. https://doi.org/10.3765/sp.14.12.
Horn, Laurence R. 1989. A Natural History of Negation. Stanford, CA.: CSLI.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
53
Iatridou, Sabine. 2000. The Grammatical Ingredients of Counterfactuality. Linguistic
Inquiry 31(2). 231–270.
Ippolito, Michela. 2003. Presuppositions and Implicatures in
Counterfactuals. Natural Language Semantics 11(2). 145–186.
Johannsen, Berit. 2016. From possessive-resultative to perfect? Re-assessing the
meaning of [hæbb- + past participle] constructions in Old English prose. In
Valentin Werner, Elena Seoane & Cristina Suárez-Gómez (eds.), 1. From
possessive-resultative to perfect? Re-assessing the meaning of [hæbb- + past
participle] constructions in Old English prose, 23–42. Berlin: De Gruyter Mouton.
https://doi.org/10.1515/9783110443530-004.
Kaiser, Elsi. 2022. Evidentiality in Finnish: On the communicative functions of the
reportative evidential kuulemma and the dubitative marker muka. Journal of
Uralic Linguistics. John Benjamins Publishing Company 1(1). 67–120.
https://doi.org/10.1075/jul.00004.kai.
Karawani, Hadil, Carina Kauf & Hedde Zeijlstra. 2019. The Asymmetry of Past Tense.
In Julian J. Schlöder, Dean McHugh & Floris Roelofsen (eds.), Proceedings of the
22nd Amsterdam Colloquium, 211–217. Amsterdam: ILLC.
https://archive.illc.uva.nl/AC/AC2019/uploaded_files/inlineitem/1AC2019_Procee
dings.pdf. (30 March, 2023).
Kissine, Mikhail. 2008. Why will is not a modal. Natural Language Semantics 16(2).
129–155. https://doi.org/10.1007/s11050-008-9028-0.
Klecha, Peter. 2016. The English Perfect is Past. In Nadine Bade, Polina Berezovskaya
& Anthea Schöller (eds.), Proceedings of Sinn und Bedeutung 20, 393–411. Eberhard
Karls Universität. Tübingen.
Klein, Wolfgang. 2000. An Analysis of the German Perfekt. Language 76(2). 358–382.
https://doi.org/10.2307/417660.
Kratzer, Angelika. 1977. What “Must” and “Can” Must and Can Mean. Linguistics and
Philosophy 1(3). 337–355.
Kratzer, Angelika. 1981. The notional category of modality. In Hans-Jürgen
Eickmeyer & Hannes Rieser (eds.), Words, worlds, and contexts: New approaches
in word semantics, 38–74. Berlin: Walter de Gruyter.
Kratzer, Angelika. 1991. Modality. In Arnim von Stechow & Dieter Wunderlich
(eds.), Semantics: An International Handbook of Contemporary Research, 639–650.
Berlin: De Gruyter.
Kuteva, Tania A. 1998. On Identifying an Evasive Gram: Action Narrowly
Averted. Studies in Language 22(1). 113–160. https://doi.org/10.1075/sl.22.1.05kut.
Lassiter, Daniel. 2014. Modality, Scale Structure, and Scalar Reasoning. Pacific
Philosophical Quarterly 95(4). 461–490. https://doi.org/10.1111/papq.12045.
Lee, C. 2008. Scalar implicatures: Pragmatic inferences or grammar? Proceedings of
the 22nd Pacific Asia Conference on Language, Information and Computation,
PACLIC 22 30–45.
Lewis, David. 1973. Counterfactuals. Oxford: Basil Blackwell.
Mackay, John. 2019. Modal interpretation of tense in subjunctive
conditionals. Semantics and Pragmatics 12. 2:1-34. https://doi.org/10.3765/sp.12.2.
Mailhammer, Robert. 2009. Towards an Aspect-Based Analysis of the Verb Categories
of Amurdak. Australian Journal of Linguistics 29(3). 349–391.
Mansfield, John. 2014. Polysynthetic sociolinguistics: The language and culture of
Murrinh Patha youth (PhD thesis). Canberra: Australian National University PhD
Thesis.
https://www.academia.edu/9080041/Polysynthetic_sociolinguistics_The_languag
e_and_culture_of_Murrinh_Patha_youth_PhD_thesis_. (4 December, 2014).
Mansfield, John. 2019. Epistemic authority and sociolinguistic stance in an Australian
Aboriginal language. Open Linguistics 5(1). 25–48.
54
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
Mansfield, John. 2020. Murrinhpatha Morphology and Phonology. Murrinhpatha
Morphology and Phonology. Berlin: De Gruyter Mouton.
https://www.degruyter.com/view/title/514377. (28 October, 2020).
Mari, Alda. 2016. Actuality Entailments: When the Modality is in the Presupposition.
In Proceedings of the 9th International Conference on Logical Aspects of
Computational Linguistics. Celebrating 20 Years of LACL 1996—2016 - Volume
10054 (LACL 2016), 191–210. Berlin, Heidelberg: Springer-Verlag.
https://doi.org/10.1007/978-3-662-53826-5_12.
Martin, Fabienne & Hamida Demirdache. 2020. Partitive accomplishments across
languages. Linguistics 58(5). 1195–1232. https://doi.org/10.1515/ling-2020-0201.
Matthewson, Lisa, Henry Davis & Hotze Rullmann. 2008. Evidentials as Epistemic
Modals: Evidence from St’át’imcets. In Jeroen Van Craenenbroeck & Johan
Rooryck (eds.), Linguistic Variation Yearbook 2007, 201–254. John Benjamins.
Matthewson, Lisa & Hubert Truckenbrodt. 2018. Modal flavour/modal force
interactions in German:: soll, sollte, muss and müsste. Linguistische Berichte(255).
259–312.
McFadden, Thomas. 2015. Preverbal ge- in Old and Middle English. In André
Meinunger (ed.), Byproducts and side effects : Nebenprodukte und
Nebeneffekte (ZAS Papers in Linguistics : ZASPiL), vol. 58, 15–48. Berlin: Zentrum
für Allgemeine Sprachwissenschaft. http://www.zas.gwz-berlin.de/zaspil.html.
McFadden, Thomas. 2016. Preverbal ge- in Old and Middle English.
http://publikationen.ub.uni-frankfurt.de/frontdoor/index/index/docId/38232. (18
February, 2016).
McGregor, William B. 2009. Another view of the Gooniyandi and its implications to
the Van linden-Verstraete typology. Journal of Pragmatics 41(1). 157–162.
McGregor, William B. 2023. On the expression of mistaken beliefs in Australian
languages. Linguistic Typology. De Gruyter Mouton. https://doi.org/10.1515/lingty2022-0023.
Mizuno, Teruyuki & Stefan Kaufmann. 2022. Past-as-Past in counterfactual desire
reports: a view from Japanese. Semantics and Linguistic Theory 83–103.
https://doi.org/10.3765/salt.v1i0.5345.
Molencki, Rafał. 2000. Parallelism vs. asymmetry: The case of English counterfactual
conditionals. In Olga Fischer, Anette Rosenbach & Dieter Stein (eds.), Pathways of
Change: Grammaticalization in English, 311–328. Amsterdam: John Benjamins
Publishing.
Musan, Renate. 2001. The Present Perfect In German: Outline Of Its Semantic
Composition. Natural Language & Linguistic Theory 19(2). 355–401.
https://doi.org/10.1023/A:1010698022098.
Nordlinger, Rachel. 2015. Inflection in Murrinh-Patha. In, 491–519.
Nordlinger, Rachel & Patrick Caudal. 2012. The tense, aspect and modality system in
Murrinh-Patha. Australian Journal of Linguistics 32(1). 73–113.
https://doi.org/10.1080/07268602.2012.657754.
Öhl, Peter. 2009. Die Entstehung des periphrastischen Perfekts mit haben und sein
im Deutschen – eine längst beantwortete Frage? Zeitschrift für
Sprachwissenschaft 28(2). 265–306. https://doi.org/10.1515/ZFSW.2009.027.
Öhl, Peter. 2014. Periphrasis as precursor of analytic inflection. Auxiliation in the
(pre-)history of German. In Rolf Kailuweit & Malte Rosemeyer (eds.), Auxiliary
Selection Revisited, Gradience and Gradualness, 181–205. Berlin / Munich / Boston:
Walter De Gruyter. http://www.degruyter.com/view/product/248250.
Olmen, Daniël Van. 2018. Reproachatives and imperatives. Linguistics 56(1). 115–162.
https://doi.org/10.1515/ling-2017-0033.
Osgarby, David. 2018. Reconstructing proto-Mirndi verbal morphology: from particles
and clitics to prefixes. Australian Journal of Linguistics 38(2). 223–292.
https://doi.org/10.1080/07268602.2018.1400504.
On the tense-aspect/modality interaction in non-Pama-Nyungan languages
55
Pahonțu, Beatrice. forthcoming. Périphrases progressives/proximatives et avertivité en
roumain. Université de Paris-Cité / Universitatea din Bucureşti PhD Thesis.
Patard, Adeline. 2019. To the Roots of Fake Tense and ‘Counterfactuality.’ In Adeline
Patard, Rea Peltola & Emmanuelle Roussel (eds.), Cross-Linguistic Perspectives on
the Semantics of Grammatical Aspect, 176–212. Leiden ; Boston: Brill.
Pieroni, Silvia. 2016. Habeo plus perfect participle in Cicero. Pallas. Revue d’études
antiques. Presses universitaires du Mirail (102). 127–135.
https://doi.org/10.4000/pallas.3626.
Pinkster, Harm. 1987. The Strategy and Chronology of the Development of Future and
Perfect Tense Auxiliaries in Latin. In Martin Harris & Paolo Ramat
(eds.), Historical Development of Auxiliaries, 193–224. Berlin: Walter de Gruyter.
Portner, Paul & Aynat Rubinstein. 2016. Extreme and Non-extreme Deontic Modals.
In Nate Charlow & Matthew Chrisman (eds.), Deontic Modality, 0. Oxford
University Press. https://doi.org/10.1093/acprof:oso/9780198717928.003.0010.
Potts, Christopher. 2007. Into the Conventional-Implicature Dimension. Philosophy
Compass 2(4). 665–679. https://doi.org/10.1111/j.1747-9991.2007.00089.x.
Prince, Kilu von, Ana Krajinović & Manfred Krifka. 2022. Irrealis is real. Language.
Linguistic Society of America 98(2). 221–249.
Pustejovsky, James. 1995. The Generative Lexicon. Cambridge (MA.): MIT Press.
Pym, Noreen & Bonnie Larrimore. 1979. Papers on Iwaidja Phonology and
Grammar (Work Papers SIL-AAB, Series A). Darwin: Summer Institute of
Linguistics.
Rebotier, Aude. 2017. The grammaticalization of tenses and lexical aspect – the case
of German and French perfects. In Kees Hengeveld, Heiko Narrog & Hella Olbertz
(eds.), The Grammaticalization of Tense, Aspect, Modality and Evidentiality: A
Functional Perspective, 241–272. Berlin: Walter de Gruyter.
Reid, Nicholas J. 2011. Ngan’gityemerri: A Language of the Daly River Region, Northern
Territory of Australia. München: LINCOM Europa.
Roberts, Craige. 1989. Modal Subordination and Pronominal Anaphora in
Discourse. Linguistics and Philosophy 12. 683–721.
Romero, Maribel. 2014. “Fake Tense” in Counterfactuals: A Temporal Remoteness
Approach. In Luka Crnič & Uli Sauerland (eds.), The Art and Craft of Semantics Vol
1 (MIT Working Papers in Linguistics 70), 47–63. Cambridge, MA.: MITWPL.
Rullmann, Hotze & Lisa Matthewson. 2018. Towards a theory of modal-temporal
interaction. Language. Linguistic Society of America 94(2). 281–331.
https://doi.org/10.1353/lan.2018.0018.
Schultze-Berndt, Eva. 2000. Simple and Complex Verbs in Jaminjung - A study of event
categorisation in an Australian language. Universiteit Nijmegen Phd Thesis.
Schulz, K. 2014. Fake Tense in conditional sentences: a modal approach. Natural
Language Semantics 22(2). 117–144. https://doi.org/10.1007/s11050-013-9102-0.
Silk, Alex. 2018. Commitment and states of mind with mood and modality. Natural
Language Semantics 26(2). 125–166. https://doi.org/10.1007/s11050-018-9144-4.
Singer, Ruth. 2006. Agreement in Mawng: Productive and lexicalised uses of agreement
in an Australian language. Melbourne: The University of Melbourne PhD Thesis.
Squartini, Mario. 2004. Disentangling evidentiality and epistemic modality in
Romance. Lingua 114(7). 873–895.
Squartini, Mario & Pier Marco Bertinetto. 2000. The Simple and Compound Past in
Romance languages. In Östen Dahl (ed.), Tense and Aspect in the Languages of
Europe, 403–440. Berlin: De Gruyter Mouton.
https://doi.org/10.1515/9783110197099.3.403.
Stefanowitsch, Anatol & Stefan Th. Gries. 2003. Collostructions: Investigating the
interaction of words and constructions. International Journal of Corpus
Linguistics 8(2). 209–243. https://doi.org/10.1075/ijcl.8.2.03ste.
Street, Chester S. & Lyn Street. 1989. Murrinh-Patha vocabulary (electronic version).
Wadeye, N.T., ms.
56
Patrick Caudal / Rob Mailhammer / James Bednall / Rachel Nordlinger
Van Linden, An & Jean-Christophe Verstraete. 2008a. The nature and origins of
counterfactuality in simple clauses: Cross-linguistic evidence. Journal of
Pragmatics 40(11). 1865–1895.
Van Linden, An & Jean-Christophe Verstraete. 2008b. The nature and origins of
counterfactuality in simple clauses: Cross-linguistic evidence. Journal of
Pragmatics 40(11). 1865–1895.
Verstraete, Jean-Christophe. 2005. The semantics and pragmatics of composite mood
marking: The non-Pama-Nyungan languages of northern Australia. Linguistic
Typology 9(2). 223–268.
Verstraete, Jean-Christophe. 2006. The Nature of Irreality in the Past Domain:
Evidence from Past Intentional Constructions in Australian
Languages. Australian Journal of Linguistics 26(1). 59–79.
Wegner, Dennis. 2019. The properties of perfect(ive) and (eventive) passive
participles: An identity approach. Glossa 4(1). https://doi.org/10.5334/gjgl.751.
Zhao, Ruoying. 2022. Decomposing Perfect Readings. Languages 7(4). 251.
https://doi.org/10.3390/languages7040251.