Mood, Aspect, Modality Revisited: New Answers to Old Questions

By Anastasia Giannakidou

Over the past several decades, linguistic theorizing of tense, aspect, and mood (TAM), along with a strongly growing body of crosslinguistic studies, has revealed complexity in the data that challenges traditional distinctions and treatments of these categories. Mood, Aspect, Modality Revisited argues that it’s time to revisit our conventional assumptions and reconsider our foundational questions: What exactly is a linguistic category? What kinds of categories do labels such as “subjunctive,” “imperative,” “future,” and “modality” truly refer to? In short, how categorical are categories?

Current literature assumes a straightforward link between grammatical category and semantic function, and descriptions of well-studied languages have cultivated a sense of predictability in patterns over time. As the editors and contributors of Mood, Aspect, Modality Revisited prove, however, this predictability and stability vanish in the study of lesser-known patterns and languages. The ten provocative essays gathered here present fascinating cutting-edge research demonstrating that the traditional grammatical distinctions are ultimately fluid—and perhaps even illusory. Developing groundbreaking and highly original theories, the contributors in this volume seek to unravel more general, fundamental principles of TAM that can help us better understand the nature of linguistic representations.
 

• Language: English
• Publisher: University of Chicago Press
• Release date: Jan 12, 2017
• ISBN: 9780226363660

Book preview

Index

Preface

The chapters in this volume were presented at the international TRAIT workshop How Categorical Are Categories? The workshop was organized at the University of Wrocław in January 2013 as part of the project Understanding Linguistic Categories funded by the Foundation for Polish Science within the FOCUS program.

As the workshop title suggests, the driving force behind this volume is the question of what constitutes a linguistic category. Each chapter explores this question as it relates to the three traditional categories of tense, aspect, and mood. We call these categories traditional because they have long been acknowledged and studied by philologists and linguists and are by now classic in the sense that consensus has been reached that these categories have certain properties and manifest themselves in predictable ways that include patterns of expected interaction. Given this broad agreement, what new can be said about them?

In fact, a lot can be said—if one addresses seriously what the labels subjunctive, imperative, future, aspect, and modality refer to. The literature tends to assume a straightforward mapping between grammatical category—for example, future ‘tense’, perfective ‘aspect’, modal ‘verb’—and semantic function; and the description of well-studied languages (e.g., Slavic with respect to aspect, Romance with respect to mood, English with respect to modals) cultivates a sense of stability and predictability in patterns that, however, vanishes once we branch out to lesser-known patterns and languages. Our goal in this volume is to reveal some of these lesser-known patterns and raise awareness about the consequences the study of these patterns have for our concept of linguistic category and the mapping between morphology and syntax on the one hand, and semantics and pragmatics on the other.

Consider, as a brief illustration, the labels ‘tense’ and ‘mood’. In the literature, tense indicates the tense morphology used for temporal orientation. Tense is thus a morphological category—that is, the grammatical category realized in the system of a language. In a language such as English, the tenses are bound morphemes on the verbs; but in Mandarin morphemes that can be characterized as tenses are not attached to the verb, and they can be particles or adverbs, operating at the sentence rather than the verb level. In Modern Greek, the future tense appears to be a particle, in contrast with the other morphological tenses in Greek, which appear on the verb. Similar variation exists in the morphological realization of mood: in Romance languages, mood is a verbal category; but in Greek, mood distinctions appear external to the verb in the form of particles. In Native American languages, prospective aspect can play the role of mood or future; and in African languages, there is considerable syncretism in the morphological categories of tense and aspect. Often it is only by appealing to semantic functions that we can make meaningful distinctions.

Hence we have already identified two important levels of analysis for what constitutes a linguistic category: the morphological level and the notional or semantic level. Crucially, the grammatical category of a particular expression—that is, whether it is a bound morpheme, a verb, an adverb, or a particle—does not predict its semantic function or its role within the notional system of the language. For instance, adverbs and bound morphemes can carry temporal information; likewise, aspectual markers can be particles, verbs, or suffixes. In many languages, modality is expressed by both verbs and adverbs. In Greek and German, modal particles have meaning that is typically assigned to modal verbs in English, and in Italian an apparent tense suffix (the future ‘tense’) gets a modal semantics. Such apparent mismatches are in fact quite common, suggesting that the mapping between morphological and notional categories is not isomorphic or universal, and that it needs to be addressed within the systems of specific languages.

The volume provides a forum for discussion of the nature of linguistic diversity in the syntax and semantics of irrealis moods (subjunctive, imperative), modality, and aspectual markers. Are these universally separate notional or morphological categories with clear-cut distinctions, or should they be understood as less discrete, conceptually related categories? If the latter option should turn out to be correct or more convincing, then how categorical is the semantics of the respective morphemes? The overarching question we are interested in is how the status of the traditional grammatical distinctions changes in light of linguistic variation, with emphasis on the growing body of evidence from languages that have previously been under-investigated. We approach these questions within a unified perspective (the question of what a linguistic category is) and by bringing together theoretical as well as experimental perspectives. The contributions in this volume can be grouped into three thematic sections:

Section 1: Tense, Aspect, and Modals: Their Categorial Status and Cross-linguistic Variation

Section 2: Irrealis Moods: Subjunctive and Imperative

Section 3: Aspectual Recursion and Aspectual Coercion

We offer more commentary in separate introductions to each section.

We want to thank the authors for their valuable contributions, as well as for working carefully with us to revise and finalize their chapters. We also want to thank our editor, Chris Rhodes, for his help and guidance, and three anonymous reviewers for reading the manuscript and offering generous and insightful feedback on each chapter. Finally, we acknowledge the support we received from the Foundation for Polish Science as part of the FOCUS program, without which this publication would not be possible.

The Editors

PART I

Tense, Aspect, and Modals

Their Categorial Status and Cross-linguistic Variation

This section addresses the ‘universality’ of the categories indicated in the title, namely tense, aspect, and modality. There have been important discussions in the formal semantics literature about the categorial status of expressions traditionally described as carrying temporal, aspectual, or modal information, and the chapters included here attempt to identify sources of universality by studying cross-linguistic variation and systematic patterns of interaction.

It is important, at the outset, to make clear that the terms tense, aspect, and modality have been used as both morphological and notional (i.e., semantic) categories. To illustrate the difference, consider tense. In the literature, the term tense is used to indicate tense morphology, which typically, though maybe not exclusively, is used for temporal orientation. Tense, in this sense, is a morphological category: that is, the grammatical category realized in the system of a language. In contrast, the term tense system is taken to mean the mechanisms of temporal interpretation assumed to be common, perhaps even universal, to natural languages—for instance, a Reichenbachian tense diagram (Reichenbach 1947) or a Priorian system (Prior 1967). These two definitions correspond to the grammatical tense distinction as opposed to the notional partitions of time represented in a natural language (Jespersen 1924, p. 255).

Crucially, the grammatical category of a particular expression—that is, whether it is a bound morpheme, a verb, an adverb, or a particle—does not predict its semantic function or its role within the notional system of the language. For instance, adverbs and bound morphemes can carry temporal information, and likewise aspectual markers can be particles, verbs, or suffixes. Modality is expressed by both verbs and adverbs in many languages. In Greek and German, modal particles have meaning that is typically assigned to modal verbs in English, and in Italian an apparent ‘tense’ suffix (the ‘future’) gets a modal semantics, as becomes clear in the discussion by Anastasia Giannakidou and Alda Mari (Chapter 3). Such apparent mismatches are in fact quite common, suggesting that the mapping between morphological and notional categories is not universal, and that it needs to be addressed within the systems of specific languages.

As pointed out by von Fintel and Matthewson (2008), in studying semantic universals one should focus on the interpretation of functional categories, which are hypothesized to be innate. In the discussion of the universality of functional morphemes, often the definitional criteria used to establish the categorial status of functional morphemes are not clear, and it consequently becomes difficult to decide whether a given functional category exists in a given language. Another important question is whether the meaning of the grammatical categories tense, aspect, and modality is cross-linguistically universal. Even if we find evidence that the meaning of some functional category is not universal, it is still worth investigating whether the observed variation in its semantics is random or subject to predictable constraints or semantic parameters. There is a growing body of evidence—coming especially from languages that look very different from English and other European languages—suggesting that interesting semantic parameters can be found, not only when we analyze tense, aspect, and modality as three separate categories, but also when we investigate how the meanings of the three categories interact in the expression of temporality, which in turn may suggest that they potentially should be understood as a continuum of a more abstract broader temporal/modal category.

The chapters in this section address central aspects of these issues. Chapter 1, by Malte Zimmermann and Anne Mucha, discusses the temporal interpretation in two West African languages: Hausa and Medumba. These two languages differ significantly in their grammatical coding of tense, aspect, and modality (TAM). Hausa is a tenseless language, in which tense is not grammatically marked, whereas Medumba is a graded-tense language and as such it has a rich system of temporal morphemes expressing fine-grained temporal distinctions such as recent past, remote past, current future, and remote future. Zimmermann and Mucha present novel fieldwork data and make a number of important claims. First, they say that Hausa does not have a linguistic category of tense. Rather, the temporal interpretation in Hausa is determined by Aktionsart, aspect, and pragmatic principles. Second, in Medumba the graded-tense effects are not due to the presence of multiple tense morphemes, and the apparent multiple temporal markers belong to different categories, including tense, aspect, and modality, which interact in complex ways to give rise to many temporal interpretations. Third, both Hausa and Medumba express future-oriented readings with a modal element, showing that there is no category ‘future tense’ in these languages. This finding agrees with a similar conclusion drawn by Giannakidou and Mari, that there is no future tense even in more familiar European languages such as Modern Greek and Italian.

The chapter has a number of theoretical implications that address directly the ‘mismatches’ between notional and morphological categories. For example, the authors hypothesize that, cross-linguistically, the presence of the categories ‘aspect’ and ‘modality’ in a language is independent of the presence or absence of the morphological category ‘tense’. In addition, they contribute new evidence in support of the idea that the pragmatic principles under lying the process of temporal interpretation are universal, as postulated by Smith et al. (2007). They also emphasize the importance of using proper methodology when conducting formal semantic fieldwork.

Chapter 2, by Valentine Hacquard, begins with the observation that the words used to express root and epistemic modality in many different languages differ systematically in their interactions with tense and aspect. More specifically, epistemic modals tend to scope above, but root modals below, tense and aspect. Moreover, only root modals (with perfective aspect) trigger actuality entailment—that is, a veridicality entailment that the event denoted by clause took place in the actual world. (Giannakidou and Staraki [2013] show that aspect is not the crucial factor, and that the phenomenon of actuality entailment may also involve causality.) Hacquard, in line with her earlier work, reveals patterns of interaction between modals and aspect, and argues that they are not a consequence of their different meaning (i.e., root vs. epistemic meaning), but result from scopal interactions between tense and aspect. According to Hacquard, there are three possible events that a modal can be anchored to: (i) the VP event when it appears in the low position and is bound by the Aspect Phrase, (ii) the speech event when it is realized high in the matrix clause (above the Tense Phrase), or (iii) an attitude event when it appears in the high position of a clause embedded under an attitude verb. The first situation results in a root interpretation of the modal, and the last two situations result in an epistemic interpretation of the modal. There is also a new observation here, namely that the pattern of interaction between modals and tense or aspect is restricted to grammatical root and epistemic modal words and does not generalize to equivalent lexical root and epistemic modal words. Even though epistemic modals share clear meaning components with verbs and adjectives that express epistemic attitudes, these verbs and adjectives do not appear to be subject to the same constraints in their interaction with tense and aspect. Hacquard thus concludes that the constraints are limited to ‘grammatical modality’; hence, if a given modal expression enters into the discussed patterns of interaction with tense and aspect, we can take this as evidence that it is a ‘grammatical modal’ and not a ‘lexical modal’.

In Chapter 3, Anastasia Giannakidou and Alda Mari contribute new evidence from Greek and Italian showing that the ‘future tense’, despite its name, does not belong to the notional category of tense. They analyze Greek and Italian future morphemes as modal operators, with the force of necessity, and with the ability to associate with both metaphysical and epistemic modal bases, thus resulting in future and epistemic readings respectively. In Greek and in Italian, future morphemes can be used with purely epistemic readings, similar to English ‘will’ in That will probably be the mailman. Giannakidou and Mari define the category epistemic future and show that, in this use, the future morpheme is literally equivalent to epistemic ‘must’ as in John must have been sick. They then address the issue of weakness of future and ‘must’ and argue that there are two dimensions of weakness. The first has to do with nonveridicality: MUST p and FUT p do not entail actual truth; the second dimension of weakness is epistemic nonveridicality: that is, MUST/FUT p does not entail that the speaker knows p. In the state of full knowledge—for example, if I see that it is raining, I cannot utter It must be raining. The so-called evidential component of MUST, therefore, is merely a reflex of the fact that it does not entail knowledge of p. At the same time, Giannakidou and Mari argue that universal modals are biased, unlike mere possibility modals, which are in nonveridical equilibrium. All modals that come with ordering sources are biased, and therefore stronger than mere possibility modals. This analysis renders universal epistemic modals both strong (because they are biased) and weak (because they are nonveridical). Regarding the future, it becomes clear that the future as a notional category of tense is redundant.

References

Giannakidou, A., and E. Staraki. 2013. Rethinking ability: Ability as modality and ability as action. In Genericity, ed. Alda Mari et al., 250–275. Oxford: Oxford University Press.

Jespersen O. 1924. The Philosophy of Grammar. London: Allen and Unwin.

Prior, A. N. 1967. Past, Present and Future. Oxford: Clarendon Press.

Reichenbach, H. 1947. The tenses of verbs. In Elements of Symbolic Logic. New York: Macmillan.

Smith, C. S., E. T. Perkins, and T. B. Fernald. 2007. Time in Navajo: Direct and indirect interpretation. International Journal of American Linguistics 73(1): 40–71.

von Fintel, K., and L. Matthewson. 2008. Universals in semantics. Linguistic Review 25(1–2): 139–201.

CHAPTER ONE

TAM Coding and Temporal Interpretation in West African Languages

Anne Mucha and Malte Zimmermann

1   Introduction

This chapter discusses the temporal interpretation and the grammatical coding of tense, aspect, and modality (TAM) in two West African languages Hausa (Chadic, Afro-Asiatic) and Medumba (Grassfield Bantu, Niger-Congo). We chose these two languages because their tense-marking systems are radically different. As demonstrated in Mucha (2012, 2013), Hausa belongs to the typological class of tenseless languages, in which tense is not grammatically marked and temporal interpretation is pragmatically resolved, relying on aspectual and contextual cues. By contrast, Medumba belongs to the typological class of graded-tense languages (Comrie 1985; Cable 2013), which are able to express more fine-grained distinctions in past- and future-oriented interpretations, such as recent past or remote past. As a result, temporal interpretation in Medumba is expected to rely less on aspectual and contextual cues. Despite this striking difference in tense coding, we will show that the two languages are surprisingly similar in other respects. First, both languages have clearly identifiable linguistic categories of aspect and modality, which shows that the presence of these categories in a language is in dependent of the presence or absence of the category of tense. Second, both languages express future-oriented readings with a modal element, showing that there is no future tense in these languages and, perhaps, cross-linguistically (Enç 1996; Matthewson 2006; Copley 2009). Third, both languages rely on aspectual and contextual cues for temporal interpretation, providing support for the analysis of tense as a deictic category, relating the reference time (RT) to the utterance time (UT).

The chapter is structured as follows. In the remainder of the introduction, we lay out the background assumptions concerning the semantic concepts under lying tense, aspect, and modality, as well as the notion of linguistic categories in natural language. We also identify a number of problems that are often encountered in semantic field research on TAM categories. Section 2 presents our analysis of Hausa as a tenseless language, which follows the more detailed account by Mucha (2013). The central claims are that Hausa does not have a linguistic category of tense, such that temporal interpretation is determined by the interaction of Aktionsart, aspect, and context, and that the so-called future marker is in fact a modal marker. Section 3 presents the first ever sketch of a formal-semantic analysis of tense-aspect marking in Medumba, comparing it with another graded-tense language, Gĩkũyũ, as discussed by Cable (2013). We propose that some of the temporal markers in Medumba are indeed tense markers. In addition, Medumba has two sets of aspectual markers located in different structural positions. From a theoretical point of view this means that what is perceived as a unified typological class of graded-tense languages splits up into various subclasses that make use of different formal means for expressing temporal information. In addition to tense, Medumba has two aspectual categories, and, as in Hausa, future-oriented readings are expressed by means of a modal marker. Section 4 concludes.

1.1   Semantic Concepts Under lying Tense, Aspect, and Modality

We adopt the following definitions for the semantic concepts underlying the notions of tense, aspect, and modality, respectively. Following Reichenbach (1947) and Klein (1994), tense markers express the temporal relation between the utterance time (UT) and the reference time (RT) relative to which the proposition expressed is evaluated. There are three basic relations between UT and RT (Comrie 1985): simultaneity (UT = RT, or UT ⊆ RT) for present time reference, anteriority (RT < UT) for past time reference, and posteriority (UT < RT) for future time reference. As for grammatical coding, we follow Partee (1973) and Kratzer (1998) in treating tense as introducing a variable expression ti in T. This indexed tense variable gets a value g(i) by means of a contextual assignment, where g(i) corresponds to RT. Possible value assignment to ti in a given context is restricted by tense specifications. The tense specification adjoins to ti and denotes a partial identity function giving back a value if and only if g(i) stands in the relevant relation to UT as illustrated for the past tense in (1) (see Heim 1994):

Aspect refers to the temporal relation between RT and the event time (ET), which is the time at which the described event takes place (Klein 1994). The temporal relations involved are subset (⊆) or precedence relations (<). Basic aspects frequently found in natural languages are imperfective/progressive (RT ⊆ ET), perfective (ET > RT), perfect (ET < RT), and prospective (RT < ET) aspect (Kratzer 1998; Cable 2013). Syntactically, aspectual heads are located in a position above vP. Semantically, aspect maps an event property of type onto a property of times (), as illustrated for progressive and perfective aspect.¹

The semantic concept of modality is frequently subdivided into the classes of epistemic modality and root modality (Hoffmann 1966; Hacquard 2009; Kratzer 2012a). Epistemic modality includes evidential interpretations, and root modality subsumes all deontic, bouletic, and inertial interpretations. In addition, some scholars assume a metaphysical modality, which they argue shows up with future marking in natural languages; see, for example, Thomason (1984), Condoravdi (2002), and Giannakidou and Mari (2014 and this volume). Semantically, modal elements denote quantifiers over possible worlds, which are restricted by a realistic or metaphysical modal base (MB), depending on the theory, and by a stereo typical, deontic, bouletic, or inertial ordering source (O) (Kratzer 2012a). With respect to future marking, we follow Copley (2009) and assume that the commonly observed interpretive effects of intention and prediction are triggered by bouletic or inertial ordering sources. By way of illustration, the bouletic modal in (3) takes a proposition and a time-world pair as arguments and yields the value ‘1’ if all worlds w’ that satisfy the circumstantial (realistic) MB and that are ranked highest on the scale of bouletic preferences, specified by MAXBOULETIC, are also p worlds. Syntactically, we locate root modals in a functional projection between AspP and TP, such that we end up with the functional architecture of the extended vP projection in (4):

1.2   Linguistic Categories

A major problem for any formal semantic fieldwork on tense, aspect, and modality consists in the fact that the coding of the semantic concepts under lying tense, aspect, and modality in natural language is not always categorical and transparent. Because of this, fieldworkers will often have difficulty identifying tense, aspect, and modality as linguistic categories. Here, we define a linguistic category as a class of form-and-meaning pairs in which a paradigm of subforms expresses a paradigm of submeanings from the same conceptual space. In the ideal case, there is a clear-cut 1:1 correspondence between a given meaning and a (possibly covert) grammatical marker expressing it, such that the expression of the meaning requires the presence of the marker, and vice versa. Typically, grammatical markers expressing tense, aspect, or modality come in the form of syntactic heads, affixes, or particles. Another reliable means of establishing the existence of two different categories is to check whether formal markers from each category can co-occur. Both identification strategies are employed in our investigations of Hausa and Medumba in sections 2 and 3.

1.3   Methodological Considerations

Research on tense, aspect, and modality in semantically under-researched languages confronts a number of methodological, theoretical, and empirical problems. Beginning with methodological issues, cross-linguistic semantic research can rarely rely on existing traditional descriptions of a given language, as these tend to focus on the interpretation of paradigms of overt forms, without checking systematically for the ways in which a given interpretation can be coded in a language. Descriptive grammars typically do not control for the bi-directionality between form and meaning, which we take to be the defining criterion of a linguistic category. Nor do such grammars provide negative semantic evidence in order to control for the limits of interpretive possibilities.

The theoretical problems revolve around the question of the universality of TAM categories, which can take either of two forms. First, do the semantic concepts under lying tense, aspect, and modality constitute cross-linguistic universals, and do languages differ only in which categorical distinctions are overtly grammaticalized (Jakobson 1959; von Fintel and Matthewson 2008)? Second, are the relevant TAM concepts universally coded in particular functional positions, which are projected in every language (Ritter and Wiltschko 2004; von Fintel and Matthewson 2008)? In this chapter we adopt the universalist position regarding the first question and assume that the semantic concepts under lying the temporal interpretation of Hausa, Medumba, and English sentences are essentially the same. Regarding the second question, we postulate the possibility of cross-linguistic variation. In particular, we argue that tense is a linguistic category in Medumba, but not so in Hausa, and that Medumba has two different linguistic categories of aspect, where Hausa has but one.

The empirical problems are of crucial concern for our present undertaking of identifying TAM categories in Hausa and Medumba. The following problems may arise in connection with attempts to tie a particular linguistic coding to a particular interpretation: (i) underspecification—a single form can be used to express several temporal, aspectual, or modal concepts; (ii) covert marking—a specific conceptual meaning is coded by the absence of formal marking. A well-known example is the covert non-future tense marker in St’át’imcets (Matthewson 2006); (iii) overcoding (periphrasis)—a single concept is linguistically coded by more than one expression; (iv) undercoding—two meaning components are coded in one form. An example in point is the English will-future, which expresses a modal shift (inertial or bouletic ordering source) and a temporal shift simultaneously, according to some analyses (e.g., Enç 1996; Copley 2009); (v) meaning similarities—separate linguistic categories may share some conceptual building blocks, which makes it sometimes difficult to differentiate the categories on empirical grounds. For instance, the temporal anteriority relation ‘<’ is involved in the expression of past tense (RT UT) and perfect aspect (ET RT). The notions of underspecification, overcoding, and meaning similarities are of crucial importance in our discussion of temporal and aspectual interpretation in Hausa and Medumba.

In response to the empirical problems discussed, we adopted the methodological principles for semantic fieldwork advocated in Matthewson (2004, 2011). Semantic data were collected in controlled elicitations with at least four consultants for each language. The elicited data consist of grammaticality judgments, truth-value judgments, and felicity judgments of object language sentences relative to a given context, and they also come in the form of contextualized translations into and from the object language. The elicitation methods employed allow not only for the identification of negative evidence in the form of impossible readings for a given sentence, but also for the uncovering of ambiguities and marked interpretations, which are difficult or impossible to detect using corpora consisting only of naturally occurring language.

2   TAM Categories and Temporal Interpretation in a Tenseless Language: Hausa

Drawing on the detailed discussion by Mucha (2012, 2013), we make the following three central assumptions concerning temporal interpretation in Hausa. First, Hausa is a genuinely tenseless language: temporal interpretation is not restricted by means of overt or covert tense marking. Second, temporal interpretation in Hausa is pragmatically resolved, depending on a number of factors, such as the Aktionsart of the predicate, grammatical aspect, and contextual information. Third, future time reference is expressed by a combination of two elements, neither of which belongs to the category of tense: a modal marker taking a bouletic or inertial ordering source used for the expression of plans and predictions, respectively, and a prospective aspect marker, which shifts the event time relative to the reference time. In sum, the analysis provides evidence for the linguistic categories of modality and aspect in Hausa, but not tense.

2.1   Hausa: Background and Questions

The Chadic language Hausa is spoken by more than 30 million people, mainly in the northern parts of Nigeria and in Niger. Two excellent reference grammars (Newman 2000; Jaggar 2001) provide ample information on its aspectual and modal markers. Hausa is a tone language with two tones (H, L`), pro-drop, and the basic word order SVOX as illustrated in (5). The verbs in (5) and (6) are preceded by the so-called person-aspect complex (PAC) (Newman 2000), marked in bold, which consists of a weak subject pronoun and an aspectual marker often in the form of a portmanteau morpheme.

With instances of A’-movement, the PAC takes a special relative form in progressive and perfective clauses, as illustrated for focus-fronting in (6).

Notice that PAC marking, and hence explicit aspect marking, is obligatory in Hausa. Next to the progressive and the perfective aspect shown in (5) and (6), other aspects in (7) are the habitual (kàn) and a form that is traditionally called ‘subjunctive’ in the Hausa literature (e.g., Newman 2000), and which is consistently expressed by L-tone on the subject pronoun. Following Schuh (2003), we analyze this form as a prospective marker in section 2.5 and therefore gloss it as such throughout the chapter. By contrast with the other TAM markers, the future marker zaa in (8), which is regularly found in future-oriented sentences, is located in a position preceding the subject pronoun, indicating that it is not an aspectual marker. Zaa must co-occur with prospective aspect marking. Its combination with progressive or perfective aspect leads to ungrammaticality (Mucha 2013):

Most important, there is no evidence for tense marking in the surface realization of Hausa sentences, with the exception of a future marker zaa in (8), making them superficially tenseless sentences (STSs) in the sense of Matthewson (2006). This raises the question of (i) whether Hausa is a genuinely tenseless language, without overt and covert tense morphology that would restrict the location of RT relative to UT, as has been convincingly argued for the unrelated languages Mandarin (Smith and Erbaugh 2005), Navajo (Smith, Perkins, and Fernald 2007), Yucatec Maya (Bohnemeyer 2002, 2009), and Guaraní (Tonhauser, 2011a); or (ii) whether Hausa is only superficially tenseless and does in fact possess covert tense morphology restricting the location of RT relative to UT. On the latter analysis, Hausa would pattern with the Salish languages St’át’imcets and Gitxsan, for which Matthewson (2006) and Jóhannsdóttir and Matthewson (2008) have argued that a covert non-future tense morpheme requires RT to be located at a time prior to or simultaneous to UT.

2.2   Covert Tense in Superficially Tenseless Languages: St’át’imcets

Matthewson (2006) argues for the existence of a covert tense morpheme with the semantic specification in (9) in St’át’imcets. This morpheme restricts RT to a point in time before or simultaneous to the utterance time tc, where the restriction comes in form of a presupposition on possible values for the temporal variable provided by TENSE.

Effectively, the temporal restriction in (9) makes the tense-morpheme act as a non-future tense. A similar informal proposal for a covert non-future tense marker in Gungbe (Kwa, Niger-Congo) is found in Aboh (2004).

Matthewson (2006) provides the following empirical arguments in favor of the covert tense analysis. First, past tense and present tense readings are equally available for STSs in St’át’imcets, irrespective of aspectual marking (IPFV vs. PFV) (cf. (10)). A future interpretation is categorically ruled out for superficially tenseless sentences (STSs) in St’át’imcets (cf. (11)).

The ungrammaticality of (11) constitutes a strong argument that the resolution of RT is subject to grammatically hard-wired constraints, and not solely dependent on contextual resolution.²

In addition, St’át’imcets sentences with the future marker kelh can receive both a present-oriented will-reading and a past-oriented would-reading (Matthewson 2006, p. 689). Again, this is expected since the future marker is evaluated relative to RT, and since RT is located either before or at UT because of the meaning of covert tense. The future marker itself is analyzed as a modal marker with a prospective meaning component locating ET at a time following RT, as shown in (12) from Rullmann, Matthewson, and Davis (2008); see also Tonhauser (2011b):

2.3   Hausa as a Tenseless Language

An investigation of temporal interpretation in Hausa shows that STSs in Hausa do not pattern with their St’át’imcets counter parts on the crucial criteria put forward in favor of the covert tense account in Matthewson (2006). From this, we conclude that Hausa STSs do not contain a covert non-future TENSE morpheme, and that the temporal location of RT is not restricted by grammatical means: Hausa is a truly tenseless language.

First, unlike in St’át’imcets, Hausa STSs receive default interpretations depending on their aspectual marking. Progressive marking triggers default present readings (13a), while perfective marking triggers default past readings (14b). By contrast, PROG sentences with past-oriented adverbials (13b) and PFV sentences with present-oriented adverbials (14a) were judged to be degraded without a context. This argues against the presence of a covert tense morpheme covering both past and present time intervals.

If necessary, though, the default temporal interpretations can be overwritten by contextual information. This is illustrated for a progressive clause with past-oriented interpretation (RT < UT) in (15a), and for a perfective clause with present-oriented interpretation (RT = UT) in (15b):

The context-dependency of the temporal interpretations in (15a, b) provides evidence against the grammatical coding of tense distinctions in Hausa. What is more, future interpretations for STSs in Hausa are available in appropriate contexts, again unlike in St’át’imcets. The felicity of (16a, b) with future reference shows that the temporal interpretation of Hausa STSs is not restricted by a covert non-future tense marker of the kind in (9).

The context-induced availability of future interpretations in Hausa shows clearly that tense distinctions are not grammatically coded in that language.³

Finally, zaa-marked sentences give rise not only to present-oriented (will) readings, as in (8), and past-oriented (would) readings, as in (17), but also to a third reading, on which RT is located at a future point in time following UT. For conceptual reasons, such readings are generally difficult to obtain, but the story-teller scenario in (18) provides a plausible context for the temporal configuration UT < RT < ET.

Since sentences containing the future-oriented marker zaa can be interpreted relative to an RT that is variably located before, during, or after UT, we conclude that zaa is not a tense marker, which would restrict the location of RT relative to UT. This is in accordance with our general claim that Hausa lacks the linguistic category of tense and that the temporal interpretation of Hausa sentences is pragmatically resolved in the absence of grammatical tense distinctions.

2.4   Pragmatic Resolution of Temporal Interpretation

On the tenseless analysis of Hausa, the value for RT can be pragmatically inferred on the basis of lexical information (Aktionsart, temporal adjuncts), grammatical information (aspect), and contextual information, making crucial use of the following pragmatic principles for temporal interpretation put forward in Smith, Perkins, and Fernald (2007, pp. 44, 45, 60):

The BEC follows from the fact that speakers follow a tacit convention that communication is instantaneous. Equating RT with an instantaneous UT—that is, adopting a present perspective—is incompatible with the report of a bounded event, as the bounds of the event would go beyond the moment of UT.⁴ The DP captures the empirical fact that RT is preferably identified with UT in the absence of grammatical or contextual evidence to the contrary. The SPI demands that the simplest possible interpretation compatible with grammatical and contextual information be chosen from the following hierarchy of simplicity: present (RT = UT) >> past (RT < UT) >> future (UT < RT), where past interpretations are more complex than present interpretations, as they require a temporal shift, and future interpretations are more complex than past interpretations, as they require a modal displacement in addition to temporal shift. As we have stated repeatedly here, these default principles can be overruled by contextual information in accordance with the following