... has a need for a. 4 See Mauri and van der Auwera (2012), and Evans and Levinson (2009). ..... Evans and. Levinson ...... Nicolle, Steve & Billy Clark. 1999.
Page 1 of 45
Cross-linguistic differences in expressing time and universal principles of utterance interpretation
Kasia M. Jaszczolt University of Cambridge
This chapter addresses the question how linguistic diversity and universalism in the domain of representing time can be reconciled. It is demonstrated how the contextualist theory of Default Semantics (Jaszczolt 2005, 2009, 2010a) accounts for cross-linguistic differences in conveying temporal location, utilising at the same time universal pragmatic principles. This is achieved by allocating information about temporality to different but universally available sources of information about meaning and to different universal processes which interact in producing a representation of the primary, intended meaning. It offers some examples of how lexicon/grammar/pragmatics trade-offs in conveying temporality can be represented in merger representations of Default Semantics, at the same time reflecting the underlying universal principles of the composition of meaning.
Keywords: Default Semantics, epistemic modality, linguistic relativity, semantic and pragmatic universals, temporality
Page 2 of 45 1. Time concept and time talk
Languages afford a variety of ways in which referring to the past, present, and future can be accomplished. Equally, relative temporal ordering of events can be conveyed in a variety of ways. One temporal concept can be expressed by different means across languages, and equally, there are various means available for speaking about time within one single language. These ways range from the use of lexical and grammatical markers of time, through automatically assigning salient interpretations to overtly tenseless expressions, to relying on the addressee’s active, conscious inference of the temporality of the situation in the particular context. In the context of these significant cross-linguistic differences in conveying temporal location (see also other contributions to this section), this chapter addresses the question whether there are universal pragmatic principles beyond the linguistic diversity of expressing temporality and if so, how the diversity and the universalism can be reconciled. It is widely acknowledged that the conceptualization of time has both universal and culture-dependent aspects (see e.g. Nishi, Yoshioka and HilberinkSchulpen, or Gladkova – all in this collection).1 In what follows I further develop the view, following my earlier enquiry in Jaszczolt (2009), that the human concept of time is universal but it is not primitive. The property of temporality supervenes, in the sense of its dependence, providing definitional characteristics, on the property of epistemic commitment, where the latter is taken to be a psychologically basic concept. I defend a contextualist approach to meaning and demonstrate how the contextualist theory of Default Semantics (Jaszczolt 2005, 2009, 2010a) accounts for cross-
1
On the co-existence of universal and language-specific aspects of conceptualisation in the domain of space see e.g. Levinson (2003), Filipović (2010), and Marotta and Meini, this collection, Filipović and Geva in this collection.
Page 3 of 45 linguistic differences in conveying temporal location, utilising at the same time universal pragmatic principles. This is achieved by allocating information about temporality to different but universally available sources of information about meaning and to different universal processes which interact in producing a representation of the primary, intended meaning. I give some examples of the application of this analysis. Memories, current experiences, and anticipations, as well as mental ordering of events and states, constitute a large part of the subject matter of human discourse. We talk about what happened or will happen, or we talk about current states of affairs, either as known facts or as possibilities. What is fascinating in the ‘time talk’ from the linguistic point of view is that languages afford a diverse array of means for referring to the past, the present, and the future. Moreover, even within one natural language there are often various options for expressing temporality. These means include grammatical markers of time such as tense, aspect, or modality, lexical markers such as temporal adverbs, temporal connectives, and other particles, as well as evidential markers.2 In addition to overt devices present in the lexicon and grammar, there are also what we can call pragmatic devices. These pertain to the principles on which discourse is organised. Overtly tenseless expressions can obtain tensed interpretations either due to the fact that such interpretations are salient to the interlocutors in the particular context or salient and default for that construction in that language in general. Securing a recovery of such a reading can be achieved either by relying on the addressee’s active, conscious inference of the temporality of the situation in the particular context or by relying on the automatic, unconscious assignment of the reading by the addressee.
2
See (8) below for an example of the interaction between evidentiality and tense.
Page 4 of 45 These pragmatic devices can be of two types: some are universal, and some are specific to the particular language and culture. In this chapter I address the question of the universal status of the mechanisms governing the recovery of temporality of a speaker’s expression, but rather than focusing on pragmatic devices per se, as contrasted with lexical and syntactic ones, I ask whether one can discern universal discourse principles on which such a selection, or trade-off, between the types of devices can proceed. English, for example, relies predominantly on tense and temporal adverbials in expressing temporality. Thai, on the other hand, has optional markers of tense and aspect, random use of adverbials, and relies largely on situated meanings, inferred from the shared background assumptions or assigned subconsciously as default interpretations.3 In the context of the significant cross-linguistic differences in conveying temporal location, one would naturally start by asking whether the human concept of time is a universal concept, and if so, whether it is primitive. Having adopted a universal but modality-based approach, this universal concept of time is confronted with the intra- and inter-language diversity of expressing temporality. The question to pose is then how the empirically demonstrable linguistic diversity reconciles with the idea of language universals. In order to address this question, I defend a contextualist approach to meaning and demonstrate, with the help of some examples from diverse languages, how the contextualist theory of Default Semantics allows for representing temporality as the result of processing of various overt and covert linguistic devices discussed above. I demonstrate how we can account for cross-linguistic differences in conveying temporal location by allocating information about temporality to different sources of information about meaning and to different processes that interact in
3
See Srioutai (2006); Jaszczolt and Srioutai (2011).
Page 5 of 45 producing a representation of the primary, intended meaning. It is the universal applicability of these sources, their mutual trade-offs, as well as the universal status of the pragmatic processes that allow us to reconcile diversity and universalism about language. In the process I also defend the methodological assumption of the compositionality of utterance meaning, also known as pragmatic compositionality (Recanati 2004; Jaszczolt 2005). The chapter is structured as follows. In section 2, I discuss the diversity of means for expressing temporality in discourse and illustrate it with pertinent examples. In section 3, the question of universals is taken up and in particular the proposal of pragmatic universals and processing universals. In section 4 I argue in favour of a contextualist approach to meaning in which the linguistic diversity and universal pragmatic principles find an explanation. Section 5 follows with a brief presentation of sources of information about meaning and types of processes that interact in meaning production and recovery, as proposed in the contextualist theory of Default Semantics. It is argued that these sources and processes allow us to adequately represent the diversity in time reference. The question that follows is that of the principles of composing meaning when such sources and interacting processes are involved. I address it in section 6, where I contribute to the ongoing discussion on compositionality by supporting the view that a truly compositional approach to meaning has to account both for overt and covert means, continuing in section 7 by representing this variety of means in merger representations of Default Semantics. Section 8 sums up the argument and makes some general remarks on the ‘depth’ of diversity when confronted with the existence of universal principles of utterance interpretation.
Page 6 of 45
2. Expressing temporality: Lexicon, grammar, and pragmatics
It can be easily observed that the diversity of ways of expressing meaning does not only pertain to temporal reference but rather persists throughout the conceptual system, affecting even those domains that at first glance, judging by common sense, should not differ from language to language because they contain concepts that are in frequent use across cultures. In short, meanings that are expressed overtly in one language, by the lexicon or grammar, may be expressed in another through pragmatic inference or default assignment of meaning to a construction. What is important is that this diversity of expression is a common fact rather than an exception and therefore should be regarded as such by any explanatorily adequate theory of meaning. For example, basic knowledge of propositional calculus entrenches in many people the conviction that concepts such as conjunction, disjunction, implication, equivalence, and negation are so basic that they are necessarily lexicalised in all languages. And yet, not all languages have clear equivalents of the English and, or, if…then, only if, or not. In Wari’, a Chapacura-Wanham language of the Amazon, and in Tzeltal, Mayan language spoken in Mexico, there is no direct equivalent of or. In Maricopa, a Yuman language of Arizona, there is no direct equivalent of and. In Guugu Yimithirr, an Australian aboriginal language, there is no direct equivalent of if.4 Now, every student of linguistics introduced to the facts of linguistic diversity knows that the lack of a word for a concept does not necessarily mean that the language is unable to express that concept; if the relevant culture has a need for a
4
See Mauri and van der Auwera (2012), and Evans and Levinson (2009).
Page 7 of 45 particular concept, it can express it in some other ways. Or it can have various coexisting means for expressing it. But what is still lacking in linguistics is a comprehensive theory of meaning that would take this diversity seriously. When a concept can be inferred or just ‘assigned’, so to speak, to a particular contextdependent interpretation of a sentence in context, there is no reason not to give it equal attention in a formal representation to that we give to words and structures.5 While some progress in this area has been achieved since the late 1970s, with the rising awareness of the need to account for pragmatic ‘intrusions’ into the logical form when we want the semantic representation to be in line with the intention of the person who produced that sentence, this awareness has largely stopped at this proposal of ‘intrusions’, ‘pragmatic enrichment’, or ‘pragmatic modulation’, or resolving underspecification or underdetermination.6 What we need to do instead is to address the questions: (i) if the language doesn’t have a word or grammatical structure for expressing a certain concept, does it still have other means for expressing it, and if so, (ii) how can we make sure that these means are given adequate attention in a theory of meaning? In order to exemplify the need for these research questions, let us continue with the example of conjunction, this time closer to home, and interwoven with the pertinent question of mental temporal ordering of eventualities (events and states, see also Wallington, this collection). English language has lexicalised conjunction, most commonly in the form of the word and, but the meaning expressed by and is not in a bi-unique mapping with the logical connective of conjunction in propositional logic. Example (1) presents a common and widely discussed scenario where and means
5
On this topic, see also Apresjan, this collection and Gladkova, this collection. Literature on this topic is ample. For an overview see for example Recanati (2005); Jaszczolt (2002); and Jaszczolt (forthcoming). 6
Page 8 of 45 more than conjunction; it means a temporal conjunction and then. The symbol ‘pr’ stands for, to use a theory-neutral term, ‘pragmatic elaborations’.
(1)
Tom finished the chapter and closed the book. pr Tom finished the chapter and then closed the book.
Various names have been given to this pragmatic inference of temporality that takes place in the process of utterance interpretation. There have been intense discussions in the literature concerning the status of such additional meanings. However, all these discussions start with the presumption that since such additions are in principle cancellable, they must be regarded, precisely, as ‘additions’. But it is not at all certain that the criterion of cancellability is the appropriate one. To invoke example (1) again, the temporal meaning may in fact be well entrenched in a particular context, to the extent that, in some contexts, cancelling it may not be a feasible conversational move at all. So perhaps it is not the different provenance and strength of lexical and pragmatically conveyed meanings that we should be focusing on but rather the fact that they are all present in, so to speak, ‘one basket’: all contributing to the meaning of an utterance in a discourse situation. Moreover, it has to be noted that English allows for the explicit expression of temporality as well, through and then, next, and so forth. It would take a strong argument to maintain that there is a difference, relevant for the interpretation of discourse, between lexical and pragmatic means of expressing temporal ordering of events. To continue with example (1), temporality is not even most accurately described as a form of enrichment of and, nor is it well introduced when we speak
Page 9 of 45 about it as a form of shift, modulation, of the meaning of (1). It arises equally easily in (2), where there is no connective.
(2)
Tom finished the chapter. He closed the book.
A less controversial case would be that of disjunction, where the exclusive or inclusive interpretation is often context- or content-driven as evidenced in the obviously inclusive reading of (3).
(3)
In order to qualify you have to be a citizen or a permanent resident.
Continuing on the topic of conjunction, contrary to English, in Swahili and then is fully grammaticalised as the consecutive tense marker ka that replaces the tense marker in all but the first clause. For example, in (4), the past time reference is signalled in the first phrase by li, after which it is assumed, or inferred,7 from the presence of the consecutive ka.
(4)
a.
…wa-Ingereza
wa-li-wa-chukua
wa-le
maiti,
3PL-British
3PL-PAST-3PL-take
3PL-DEM
corpses
‘…then the British took the corpses,
b.
wa-ka-wa-tia
katika bao
moja,
3PL-CONS-3PL-put.on
on
one
board
put them on a flat board, c. 7
wa-ka-ya-telemesha
maji-ni
kwa
utaratibu w-ote…
I shall not discuss the big question of inference vs automatic meaning assignment at this point. See for example Carston (2007); Jaszczolt (2010b).
Page 10 of 45 3PL-CONS-3PL-lower
water-LOC
with
order
3PL-all
and lowered them steadily into the water…’
(adapted from Givón 2005: 154)
To summarise the options, conjunction can be lexicalised or not, and when it is, one lexical item may carry more meaning than the meaning of conjunction alone, just as the Swahili ka does. Or it may contribute to triggering more meaning, such as in the English example (1), without, so to speak, being ‘enriched’ by this meaning as a lexical item, as evidenced by (2). The temporality is rather expressed pragmatically by the juxtaposition of two sentences pertaining to events, as is well summarised in Asher and Lascarides’ (e.g. 2003) rhetorical structure rule of Narration, which states that if both sentences refer to events, then the event expressed by the first sentence takes/took/will take place before the event expressed by the second sentence, as in (5):
(5)
Tom finished the chapter.
e1
He closed the book.
e2
There does not seem to be any reason, methodological or epistemological/metaphysical, for giving more representation to the lexical or grammatical way of expressing the same concept than to the pragmatic ones; as long as the concept is expressed and conveyed, it has to figure in the representation of meaning. Next, let us come back to the example of Wari’, which does not have a direct equivalent of the logical connective of disjunction, neither does it have a word closely
Page 11 of 45 related to it like the English or.8 However, the concepts of the alternatives or disjunction are present there no less than in languages with the connective. The absence of a disjunctive marker is compensated by the use of some irrealis marker as for example in (6).
(6)
’am
’e’
perhaps live
ca
’am
3SG.M. perhaps
mi’
pin
give
complete 3SG.M.
ca
‘Either he will live or he will die.’
(from Mauri and van der Auwera 2012: 391).
Again, the presence of the concept of an alternative is sufficient ground for including disjunction in the representation of meaning of such a sentence – realised as a juxtaposition of modal phrases. To repeat, the reason is simple and overwhelming. To use von Fintel and Matthewson’s (2008: 170) words,
while perhaps none of the logical connectives are universally lexically expressed, there is no evidence that languages differ in whether or not logical connectives are present in their logical forms.
Examples of linguistic diversity in connectives can be continued endlessly: not only do we have cultural and linguistic diversity, but also frequently one language affords us choices of means. In addition, every discourse situation creates a locum for new pragmatic inferences as well as automatic, default meaning assignments.9 I have addressed here the issue of overt and covert temporal ordering of eventualities by
8 9
‘Closely related’ because English or can adopt inclusive () or exclusive meaning. I discuss this further in section 5.
Page 12 of 45 using an example of conjunction and consecutive tense. It goes without saying that the phenomenon of Sequence of Tense (SOT) in English occupies the central place in this topic. For our current purpose, the ambiguities created by the SOT phenomenon such as in (7) are yet another domain in which we find corroboration for the argument that grammatical, lexical, and pragmatic means have to be regarded on a par, on an equal footing, in representing meaning.10 In (7), the temporal location of Mary’s state with respect to the time of speaking can vary depending on the pragmatic inference or default assumptions for the context.
(7)
John will think that Mary is pregnant.
(from Hornstein 1990: 86).
Therefore, the same theory of meaning that accounts for the lexicon/pragmatics mix in the consecutivity marker ka, the English or, or, less obviously, the English and, will have to account for the grammar/pragmatics mix in SOT. Two more examples discussed in the remainder of this section will add more flesh to this argument: the evidentiality/temporality mix and the tense-time mismatches. To address the first: in Matses, a Panoan language spoken in the Amazon region, there is an evidential system that requires that the source of information is overtly specified whenever a past event is reported on the basis of inferential evidence (see Fleck 2007). In particular, it is specified in a sentence how long ago the event took place, as well as how long ago the speaker obtained evidence. Fleck calls this inflectional solution double tense. A relevant verbal inflectional suffix in Matses 10
I stay clear of the discussion as to whether this theory of meaning should be called semantics – in the contextualist sense, as in Relevance Theory, Sperber and Wilson (1995), or my Default Semantics, Jaszczolt (2005) – or pragmatics, such as truth-conditional pragmatics, see Recanati (2010). In sections 4 and 5 below I adopt the first option but this choice makes no difference to the argument.
Page 13 of 45 combines temporal and evidential information as in (8). ERG stands for ‘ergative’, and DIST.PAST.INF-REC.PAST.EXP
for ‘distant past inferential’ combined with ‘recent past
experiential’. The conveyed temporal information is that the speaker discovered the hut a short time ago, while it was made a long time ago.
(8)
mayu-n
bëste-wa-nëdak-o-şh.
non.Matses.Indian-ERG
hut-make-DIST.PAST.INF-REC.PAST.EXP-3
‘Non-Matses Indians (had) made a hut.’
(from Fleck 2007: 590).
Matses has three past tenses: recent, distant, and remote, and three evidential distinctions: direct experience, inference, and conjecture (see Fleck 2007: 589). The marker nëdak expresses the distant past, referring to any temporal interval from between about a month ago to the speaker’s infancy, combined with the inferential evidence, while o marks recent past, normally from immediately before the time of speaking to up to a month ago (although a more extended scale is also used in some contexts), combined with experiential source. Altogether we obtain nëdak-o which combines two items of temporal reference: distant past, that of the making of the hut, and recent past, that of the discovery or obtaining information. Analogously, markers for other combinations, such as that for recent past inferential or distant past experiential, are available in the language, making up nine markers for the past time reference in total. Now, Fleck suggests that the fact that these distinctions are present only for referring to the past can be partly ascribed to the obvious fact that these distinctions are less important (although perhaps not unnecessary, as Fleck suggests) in the case of the future or the present. What is interesting for our purpose is that what is expressed
Page 14 of 45 as compulsory double tense in Matses would normally be achieved through a grammar/lexicon means in English, or grammar/pragmatics when the time of obtaining evidence is obvious from the context. In addition, when we also consider the three-way distinction in marking past tense, we require an additional grammar/lexicon or grammar/pragmatics mix in English, which may result, for example, in a grammar/pragmatics/lexicon mix for expressing the combination from (8), as in the rather crude attempt in (9).
(9)
Tom built a house [a long time ago]. [I [have just] realised/deduced from what you were saying that he did].
It is easy to envisage discourse scenarios in which the material in square brackets, outer or inner, could be redundant. All such numerous differences in what languages grammaticalise or lexicalise as far as temporal reference is concerned add more fuel to the argument that, to put it in the form of a slogan, whatever information content there is in an utterance has to be present in its meaning representation, independently of how it got there, i.e. independently of what sources or what processes are responsible for it. Finally, let us address tense-time mismatches. These are abundant in many languages, and it is so for good reasons. For example, when a speaker uses present tense with future-time reference, a phenomenon that is sometimes called tenseless future, the choice is likely to be motivated by the intention to emphasise the high probability of the future event or potential difficulty in altering the plan or schedule as in (10).
(10)
On Monday the Prime Minister is in Glasgow.
Page 15 of 45 The so-called futurate progressive as in (11) exemplifies a similar phenomenon.
(11)
On Monday I am planting my hedge.
Vivid, or historic, present obtained through present tense forms used with past-time reference as in (12) is yet another example of the discussed mix.
(12)
This is what happened to me yesterday: I enter the office and see this guy standing by my desk and smiling to me. I say to him, …
From the conceptual, or semantic, point of view, it is more appropriate to call the phenomenon the past of narration in that the grammatical form is used to refer to past eventualities – on analogy with the classification of (10) and (11) as examples of the future. In short, as is well known, future or past temporality need not map onto future or past verb forms in English. Moreover, there are languages in which temporal reference is not grammatically marked (Mandarin) or this marking can be optional (Thai). I emphasise this fact as it serves as further supporting evidence that all information about temporal reference has to be treated on an equal footing, no matter what its provenance. In section 7 I exemplify how this representation can be executed. What needs to be established first, in the logical order of explanation, is (i) whether there is a ‘theoretical whip’ that would tame and subjugate this diversity (section 3); (ii) the theoretical framework for the theory of meaning that should be adopted in order to give us the required object of study and scope (section 4); as well as (iii) the identification of sources and processes that contribute meaning information
Page 16 of 45 (section 5), followed by (iv) a disclaimer concerning the use we make of the principle of compositionality (section 6).
3. Pragmatic universals?
It can be assumed that the composition of meaning must follow some universal principles that have their provenance in the structure of the brain and operations the brain is capable of performing, as well as in the broadly similar external environment and purpose associated with language use. One way to proceed would be to begin by ascribing generative capacity to syntax, following the generative syntax school, and attempting to address all meaning composition through proposing adequate syntactic operations. However, there are two categories of problems with this method. Firstly, these operations would either have to be assumed to be cognitively real, or would have to be discovered and thereby awarded a clear empirical status. If we manipulate them in syntactic theory just to make them account for the data, we are in danger of overgeneralising. Arguably, not all languages exhibit recursion, and even the assumption of the universality of constituent structure poses some problems of analysis as free-order languages such as Latin exemplify (see Evans and Levinson 2009 for a convincing argument in support of this claim). On the other hand, conceptual universals seem to be beyond dispute: there are general patterns on which semantic composition is founded. To quote Evans and Levinson (2009: 444),
Page 17 of 45 although recursion may not be found in the syntax of languages, it is always found in the conceptual structure, that is, the semantics or pragmatics – in the sense that it is always possible in any language to express complex propositions.11
Compositionality can be safely accepted as a property of conceptual structure, and thereby, on the definition of semantics we will be accepting below, also as a property of semantics, as it is understood on contextualist accounts. What it means is that generative power is ascribed to semantics/pragmatics rather than to syntax. This can be understood in a variety of ways. Richard Montague strove for a formal representation of sentence meaning in terms of intensional logic. His followers in the general tradition of dynamic semantics, such as Discourse Representation Theory and Dynamic Predicate Logic, are striving for an adequate formal representation of meaning, where meaning includes pragmatic addition such as the resolution of anaphora and presupposition, allowing for various degrees of representationalism while retaining the commitment to compositionality. The next step in the direction of pragmatics is the so-called pragmatic, interactive compositionality adopted by some post-Gricean contextualists and notably in Recanati’s (2004, 2010) Truth-Conditional Pragmatics and Jaszczolt’s Default Semantics (2005, 2010a) and defended in section 6 below. Let us first look at the proposal of semantic universals. Von Fintel and Matthewson (2008) begin by invoking the Strong Effability Hypothesis and Translatability Thesis, both attributed to Jerrold Katz. The Strong Effability Hypothesis says that ‘Every proposition is the sense of some sentence in each natural language.’ This is, needless to say, wishful thinking for a formal semanticist and it doesn’t take much evidence to disprove it. Next, the more relaxed Translatability 11
On conceptual universals see also Pinker and Jackendoff (2009); on a neurobiological account of universals see Müller (2009).
Page 18 of 45 Thesis says that ‘For any pair of natural languages and for any sentence S in one and any sense σ of S, there is at least one sentence S' in the other language such that σ is a sense of S’. This thesis, again, is much too strict and only workable when we substitute the term ‘utterance’ for ‘sentence’ and construe semantics in a radical contextualist way. On the other hand, von Fintel and Matthewson (2008: 191) note the following:
We found that languages often express strikingly similar truth conditions, in spite of non-trivial differences in lexical semantics or syntax. We suggested that it may therefore be fruitful to investigate the validity of ‘purely semantic’ universals, as opposed to syntax-semantics universals.
The obvious follow-up question is, what would such ‘purely semantic’ universals have to be like? Von Fintel and Matthewson suggest (i) some universal semantic composition principles, which, however, to sum up crudely, are problematic to construe, or (ii) Gricean principles of utterance interpretation. If we select the latter, then we opt for the semantics/pragmatics mix and the follow-up question is what role universal processing principles play in such semantic/pragmatic universals. Evans and Levinson (2009) contend that processing principles are the universals sought.12 The issue is this. What is important for the current discussion of the provenance of universals is whether we choose to look for them in the domain of formal semantic/pragmatic generalisations per se or rather, or also, in the domain of processing generalisations. For example, if we were to go along with dynamic semantics and incorporate more and more information from context into formal representation, we would be opting for the first strategy. If we were to go along with post-Gricean, and therefore intention-based, contextualism in semantics and 12
On universals in processing see also Hawkins (2004), (2009).
Page 19 of 45 incorporate results of pragmatic inference in the truth-conditional representation, we could still opt for either focus on the final representation or focus on the kinds of processes that lead to this pragmatics-rich representation. Evans and Levinson go with the latter. As they say, “[f]or our generativist critics, generality is to be found at the level of structural representation; for us, at the level of process” (ibid., p. 475). To sum up, the methodological question is whether universal principles should include generalisations about processing. There are multiple arguments in favour of the affirmative answer.13 We need methodological assumptions about the theory of meaning but we also have to see how they govern the production and comprehension of meaning because, as was amply exemplified in section 2, and as is also widely acknowledged in various current processing models, comprehension of meaning involves the synthesis of chunks of information that come from different domains.14 Now, the main methodological assumption about the theory of meaning is the Principle of Compositionality, which says that the meaning of a complex expression is determined by the meanings of its parts and the structure in which they are combined. Pursued in formal semantics by Richard Montague, Barbara Partee, and armies of acolytes, the principle remains a Holy Grail of semanticists. There are intensional constructs that refuse to succumb unless sophisticated changes or additions are allowed in their logical form. Temporal reference is also in this problematic category. The contextualist orientation in post-Gricean pragmatics is the most suitable approach
13
I argued extensively for the inclusion of processing consideration in semantic/pragmatic theory in Jaszczolt (2008). See also Saul (2002). 14 Needless to say, the question is orthogonal to the question of lexical universals pursued in lexical semantics: the proposal that words such as I, you, big, small stand for universal human concepts because they are universally lexicalised feeds into the overall issue of meaning universals but can fare independently of it as well. The universality of semantic types t,e is also orthogonal in that when semantic/pragmatic universals allow for interaction across the domains of lexicon, grammar, and pragmatics, neither domain has to separately exhibit universals of this type.
Page 20 of 45 to look into for pursuing this pragmatic/semantic universal. This is the topic to which I now turn.
4. A contextualist approach to expressing temporal distinctions
In the post-Gricean approach to utterance, and, more recently, also discourse meaning, the focus of the debate has been on the pragmatic constituents of what is said. In other words, since the late 1970s, the main debates have circled around the delimitation of the propositional content as contrasted with what is truly implicit. Some pragmatic constituents, such as the enrichment of the meaning of sentential connectives as in (1) above or (13), the specification of the domain of quantification as in (14), the precisification of the meaning of negation, and many others, are said to contribute to the propositional content of the uttered sentence.
(13)
You will pass the test if you practise a lot. pr You will pass the test if and only if you practise a lot.
(14)
Everybody read Frege. pr Every member of the research group read Frege.
Whether we call them parts of explicature (Sperber and Wilson 1995), what is said (Recanati 1989), or retain the term implicature (Levinson 2000), the common-sense judgement remains that they are in some sense more basic and important than implicatures, where the latter constitute separate messages to the addressee. This pragmatics/semantics mix approach has been called radical pragmatics, sense-
Page 21 of 45 generality (in that the structure of the sentence affords only the general, underdetermined representation of meaning), or more recently contextualism (see e.g. Recanati 2004, 2005, 2010, 2012). The latter term is what we shall adopt.15 To sum up the idea, semantic analysis takes us only part of the way towards the recovery of utterance meaning and pragmatic enrichment completes the process. The standard logical form of the sentence is enriched, or modulated, as a result of pragmatic processing, either inferential or automatic in kind, and the entire semantic/pragmatic product becomes subjected to the truth-conditional analysis. Next, the question that has frequently been addressed, since it was brought to the fore by Carston (1988, 1998), is how far the logical form can be extended. Or, in other words, ‘how much pragmatics’ is allowed in the representation of the main intended meaning of an utterance. According to Default Semantics, a contextualist approach I am going to use in what follows (see Jaszczolt 2005, 2009, 2010a), the answer is quite radical: as much as is required to faithfully depict the main, primary message that the speaker intends to convey. This answer, albeit commonsensically obvious, is not so easy to defend for a compositional and truth-conditional theory of meaning because ‘as much as required’ also includes situations where the main message is conveyed indirectly, and this happens surprisingly often and is true of different languages and cultures (see e.g. Nicolle and Clark 1999; Pitts 2005; Schneider 2009). Indirect main meaning means that the logical form of the intended message may be significantly different from the logical form of the uttered sentence and this pulls the rug from under the feet of formal semanticists who wish to retain the backbone of sentence structure. But at the same time it opens a window for those of a
15
But note that contextualism can also be understood more broadly, to include pragmatics in two-dimensional semantics such as Stalnaker’s approach. See Jaszczolt (forthcoming) for discussion.
Page 22 of 45 more cognitive orientation who want to retain compositionality and at the same time recognise the fact that meaning comes from a variety of sources and through a variety of processes and is merged into one cognitive representation. It is this cognitively real representation that is compositional and is the object of study of Default Semantics (DS). It is called there merger representation and its most distinctive feature, differentiating it from primary objects of other contextualist accounts, is that it does not undergo a syntactic constraint. DS does not recognise the level of meaning at which the logical form is pragmatically developed, enriched, or modulated as a real, cognitively justified construct. To do so would be to assume that syntax plays a privileged role among various carriers of information, which is considered to be a contextualists’ mistake. In (15), the main intended message would normally be as in (15c) and (15c) is therefore modelled as the merger representation of the primary meaning in preference to its alternatives.
(15)
Child to mother: Everybody has a bike.
(15a)
All of the child’s friends have bikes.
(15b)
Many/most of the child’s classmates have bikes.
(15c)
The mother should consider buying her son a bike.
(15d)
Cycling is a popular form of exercise among children.
Interlocutors frequently communicate their main intended content through a proposition that is indirect and therefore not syntactically restricted by the uttered sentence. It is this intended proposition that post-Gricean, intention-based contextualism should attend to. Understood as being about meaning that is intended by the speaker and recovered by the addressee, it is important not to conflate it with the view that meaning is to be analysed from the position from which it is assessed.
Page 23 of 45 The latter is a relativist stance (see MacFarlane 2005, 2011); the first is a contextualist one, Gricean in spirit.
5. Sources of temporal information
In section 2 I gave several examples of sentences in the case of which temporal information had to be retrieved from the lexicon/pragmatics, grammar/pragmatics, or even lexicon/grammar/pragmatics mix. In order to account for this diversified provenance and merger of outputs, DS has to be able to assign information to clearly defined sources. In the current version of DS (Jaszczolt 2009, 2010a), the following sources have been identified:
(i)
world knowledge (WK)
(ii)
word meaning and sentence structure (WS)
(iii)
situation of discourse (SD)
(iv)
properties of the human inferential system (IS)
(v)
stereotypes and presumptions about society and culture (SC)
I will not elaborate on these sources here. In what follows some basic knowledge of DS will prove useful but not essential. Suffice to say, as a point of explanation, that source (iv) accounts for interpretations that stem from the standard properties of human intentional mental states. For example, when a speaker uses a definite description, it will be, by default, understood as associated with the strongest referential role a definite description can play, namely the referential rather than the
Page 24 of 45 attributive one, unless some other source of information intervenes and stops this default from arising. The outputs of sources of information about meaning are treated on an equal footing. To repeat, the syntactic constraint present in other contextualist accounts, such as relevance theory or truth-conditional pragmatics, is abandoned. Merger representations have the status of mental representations. They have a compositional structure: they are proposition-like constructs, integrating information coming from various sources that interacts according to the principles established by the intentional character of discourse. While the sources of information are delineated clearly, they are not a claim to originality in DS. Many pragmatic approaches emphasise the role of cultural and social assumptions and the role of the architecture of the brain in meaning production and comprehension. Levinson (2000) makes out of them a rigid theory of generalised implicatures. In his socio-cognitive approach Kecskes (e.g. 2010) emphasises that pragmatics has to take into account both societal (cooperation) and individual factors (egocentrism) as context-dependency is counterbalanced by individual tendencies. He proposes that a dialectal model of pragmatics should present both the speaker’s and the addressee’s perspective (Kecskes 2008) – a claim that is clearly in line with the Model Speaker–Model Addressee perspective of DS (Jaszczolt, e.g. 2005). Next, the model of sources of information can be mapped onto types of processes that produce the merger representation of the primary meaning, as well as the additional (secondary) meanings. Lexicon and grammar represent both the source and the type of processing that is unique to them (WS). Next, structure and properties of the brain discussed as IS above, lead to so-called cognitive defaults (CD) – automatic interpretations that trigger for example the strong referential reading of definite
Page 25 of 45 descriptions. Sources WK and SC are responsible for social, cultural, and world knowledge defaults (SCWD).16 But they can also pertain to a process of conscious pragmatic inference (CPI): information from world knowledge (WK) or knowledge of society and culture (SC) can be accessed automatically (SCWD) or they can be consciously inferred (CPI). Situational context of discourse (SD) will, when active, lead to CPI. A CD is exemplified in (16). Sentence (17) exemplifies the fact that SC can result in either an SCWD or CPI process. PM stands for ‘primary meaning’.
(16)
IS CD The author of Presumptive Meanings is coming to Cambridge next spring. PM: Stephen Levinson is coming to Cambridge next spring.
(17)
SC SCWD or CPI A Rembrandt was sold at Christie’s last week. PM: A painting by Rembrandt was sold at the Christie’s auction house in
London last week.
In constructing merger representations DS makes use of the processing model and it indexes the components of the representation (symbolised by , for summation of information) with a subscript standing for the type of processing. ‘Default’ is a notorious term in the literature in that it is used in a wide variety of ways and for a variety of purposes.17 Therefore it is important to stress that SCWDs and CDs are defaults for the situation, for the Model Speaker and Model Addressee. SCWDs depend on societal and cultural factors, as well as on shared knowledge of the physical laws of the world. CDs, when present, are triggered by the structure and 16
These are marked as SCWDpm when they pertain to primary meaning and SCWDsm when they result in secondary meanings. Analogously for CPI discussed below. 17 See Jaszczolt (2010b) for an encyclopaedic overview.
Page 26 of 45 operations of the human brain and in particular by the property of intentionality exhibited by the relevant mental states. All in all, what we have in DS is a set of universally utilised sources of information and a set of universally applicable processes in meaning construction. We are still far from knowing how exactly the merger of information proceeds; it is possible that for this we require evidence from neurolinguistics. Evans and Levinson (2009: 429) define universal principles as “stable engineering solutions, satisfying multiple design constraints, reflecting both cultural-historical factors and the constraints of human cognition”. In DS, as we have seen, these universal principles translate into, respectively, (i) social, cultural, and world-knowledge defaults (SCWD) or conscious pragmatic inference from societal and cultural knowledge or knowledge of physical (world) facts (CPI); and (ii) cognitive defaults (CD), grounded in the properties of the human inferential system (IS), and in particular in the property of intentionality of mental states. After a brief explanation in section 6 of how the requirement of compositionality of meaning fares under this DS-theoretic account, in section 7 I put the above categories to use in representing the lexicon/grammar/pragmatics mix in temporal reference.
6. Covert and overt devices and the question of compositionality
The requirement of compositionality is often considered to be a necessary condition for any theory of meaning. However, sometimes it is overtly acknowledged that some types of constructions, namely intensional contexts, evade it, and the claim that natural language semantics is compositional is lifted. Schiffer’s (e.g. 1991, 1992,
Page 27 of 45 1994, 2003) way of approaching the problem is to stipulate what it would take for semantics to be construed as compositional – for example, what it would take to construct formal representations of propositional attitude reports that observe the compositionality requirement (Schiffer 1992), but, nevertheless, conclude that the natural language semantics in fact lacks these required features that would make intensional contexts compositional. Instead, it is possible that the composition of meaning reflects compositional reality; the structure of meaning supervenes on the structure of the world. Next, Jerry Fodor, in his second and substantially rethought attempt at the Language of Thought in LOT2 (Fodor 2008), proposes that compositionality be sought on the level of referential properties. Now, if compositionality is not to be found in traditionally understood semantics but semantics must ‘stay close’ to the compositionality requirement by observing referentiality and, generally, the world- dependence, then, equally, we can bring these extra semantic components into a theory of meaning as long as we redefine semantics as a semantics/pragmatics mix. In other words, we do not postulate slots in the logical form of the expression that have to be filled from some extra-sentential source like indexicalists do (see e.g. Stanley and Szabó 2000; Stanley 2002, 2007), but we say that these sources contribute according to some still unknown principles of interaction of processes of meaning construction and meaning recovery, as in DS. Recanati’s ‘top-down’ modulation of meaning is a middle station in this conceptual shift. It denies indexical slots, allows free modification of sense, but does not address the question of sources and interaction of processes. However, Recanati (2004) does address the question of lifting compositionality to such an interaction: compositionality belongs to modulated propositions, it is called ‘interactionist’, or ‘Gestaltist’ compositionality. Similarly, DS assumes compositionality of utterance
Page 28 of 45 meaning rather than sentence meaning. ‘Assumes’ because compositionality is here, like in Recanati’s approach, a methodological assumption. It is at the same time an empirical assumption, a necessary characteristic of all possible human languages, in agreement with Szabó (2000). What remains is a colossal task of understanding the interaction itself, through work on corpora, neurolinguistic evidence, and formal theories. At present we can only demonstrate that a pragmatic compositional account such as DS adequately accounts for the lexicon/grammar/pragmatics mix, and for the syntactic-constraint-free primary meanings, keeping the identification of processes, as well as the units on which they operate, perspicuous in the meaning representation.
7. Representing the diversity
Temporal reference that is of interest for our mixed-sources and interaction-ofprocesses account was exemplified by, among others, (i) adding a temporal (and then) meaning to a non-temporal connective in (1); (ii) adding temporal reference to contiguous sentences in discourse in (2), juxtaposed with grammatical marking of this relation in Swahili by the consecutive tense marker ka in (4); (iii) SOT phenomenon with future-time reference in the main clause in (7); (iv) double-tense in Matses in (8), contrasted with English (9) and normally analogous to a mere (9') below; and (v) tense-time mismatches in English in (10)–(12), all repeated below for convenience.
(1)
Tom finished the chapter and closed the book.
(2)
Tom finished the chapter. He closed the book.
Page 29 of 45 (4)
a.
…wa-Ingereza
wa-li-wa-chukua
wa-le
maiti,
3Pl-British
3Pl-Past-3Pl-take
3Pl-Dem
corpses
‘…then the British took the corpses,
b.
wa-ka-wa-tia
katika bao
moja…
3Pl-Cons-3Pl-put.on
on
one
board
put them on a flat board…’
(7)
John will think that Mary is pregnant.
(8)
mayu-n
bëste-wa-nëdak-o-şh.
non.Matses.Indian-ERG
hut-make-DIST.PAST.INF-REC.PAST.EXP-3
‘Non-Matses Indians (had) made a hut.’
(9)
Tom built a house [a long time ago]. [I [have just] realised/deduced from what you were saying that he did. ]
(9')
Tom built a house.
(10)
On Monday the Prime Minister is in Glasgow.
(11)
On Monday I am planting my hedge.
(12)
This is what happened to me yesterday: I enter the office and see this guy standing by my desk and smiling to me. I say to him, …
In (1), what is of interest is that the temporal meaning of and comes from the source SCWD: since there are two events juxtaposed, the addressee automatically adds the relation of temporal ordering 1 < 2; see Figure 1. Subscripts stand for the type of process responsible for the contribution to utterance meaning and the square brackets contain the material on which the process operates. To repeat, the processes can then be easily, albeit not bi-uniquely, mapped onto sources of information. For clarity of
Page 30 of 45 presentation, the representation of past-time reference is omitted (but will be attended to in other examples).
Page 31 of 45
x y z 1 2 [1 2]WS Tom (x) chapter (y) book (z) 1: 2:
[1 < 2]SCWDpm
[x finished y]WS [x closed y]WS
Figure 1: for example (1)
Similarly, in (2), the juxtaposition of two sentences referring to events results in [1 < 2]SCWDpm. The difference is that the conjunction is present in the merger representation qua conceptual representation but is not overtly present in the sentence, hence we do not have [1 2]WS as in Figure 1. Instead, there is [1 2]CD as in Figure 2. We assume that it is CD that is responsible for the conceptual conjunction because the natural order of events happening in the world is mirrored here in the mental states.
Page 32 of 45
x y z 1 2 [1 2]CD Tom (x) chapter (y) book (z) 1: 2:
[1 < 2] SCWDpm
[x finished y]WS [x closed y]WS
Figure 2: for example (2)
Example (4) demonstrates that the temporal order of events can be conveyed with recourse to neither the lexicon (and then) nor to pragmatics (enrichment of and with two events juxtaposed), but instead to grammar. In other words, there is considerable variation in how the ‘order of narration’ is externalised. In Swahili, it is the marker ka that conveys this meaning. In DS there is no difficulty in varying this source of information about temporal sequence of events, as Figure 3 demonstrates. The meaning 1 < 2 is ascribed there to WS. Upper case ‘X’ stands for a plural referent with collective reading.
Page 33 of 45
X 1 2 [wa-Ingereza]CD (X) 1: 2:
[wa-chukua wa-le maiti [X]CD]WS [wa-tia katika bao moja [X]CD]WS [1 2]WS
Figure 3: for example (4)
Example (7) obtains two interpretations: one where is pertains to the future, as indicated by the preceding future-tense marker will, and the other one without a shift, with the present-time grounded is. Following the theory of time adopted in DS, the future is represented by a modal acceptability operator ACC qualified by the degree and superscripted by the source of information (see e.g. Jaszczolt 2009). So, ACCrf ├ Σ' means ‘it is acceptable to the degree pertaining to the regular future form that it is the case that Σ'’, where the sources of information are WS and CD – that is the structure, lexicon, and the default sense of will. On the shifted reading, Σ'' remains in the future, as the CD subscript on ACC indicates. This reading is represented in Figure 4. The superscript rf stands for ‘regular future’.
Page 34 of 45
x y Σ' [John]CD (x) [Mary]CD (y) [ACCrf ├ Σ']WS,CD ': Σ'':
[x think '']WS [y be pregnant]WS
Figure 4: for the default shifted reading of example (7)
The present-time reading of Σ'' is obtained via context-triggered inference that Mary’s state of pregnancy obtains at the time of discourse. The process is then CPIpm as in Figure 5. The superscript rn stands for ‘regular present’.
Page 35 of 45
x y Σ' [John]CD (x) [Mary]CD (y) [ACCrf ├ Σ']WS,CD ': Σ''
[x think '']WS [ACCrn├ Σ'']WS, CPIpm [y be pregnant]WS
Figure 5: for the contextually triggered reading of example (7)
Example (8) has a conceptual representation as in Figure 6, where the information ‘a long time ago’ and ‘as I have just deduced’ (or something to that effect) is obtained via the grammatical source, and hence WS. In the English equivalent uttered as (9'), this double temporality and source of evidence is either not intended or, in some circumstances, may be inferred via CPIpm or taken for granted. For the clarity of argument, we did not attend to the nuances of the ergative structure and regarded ‘bëste-wa’ as a unit and ‘mayu-n’ as the ‘subject’ in the sense of the actor.
Page 36 of 45
X Σ' mayu-n (X) [ACCREC.PAST.EXP[ACCDIST.PAST.INF├ Σ']]WS Σ'
[bëste-wa (X)]WS
Figure 6: Σ for example (8)
Sentences (10) and (11) normally present a problem for formal semantic accounts, where temporal reference is dictated by the tense of the sentence, as for example in Kamp and Reyle’s (1993) account of Discourse Representation Theory. In DS, however, all sources of information are treated on an equal footing and are equally able to override or prevent the potential output of other sources. So, tensetime mismatches are not an exception but simply a case where CPI is at work. (10) obtains the representation in Figure 7. The superscript tf stands for ‘tenseless future’.
Page 37 of 45
x t Σ' [the Prime Minister]CD (x) on Monday (t) [ACCtf ├ Σ']WS, CPIpm Σ'
[x be in London]WS
Figure 7: Σ for example (10)
Futurate progressive in (11) works analogously, with the temporality conveyed through the CPIpm as in Figure 8. The superscript fp stands for ‘futurate progressive’.
Page 38 of 45
x t Σ' [the speaker]CD (x) on Monday (t) [ACCfp ├ Σ']WS, CPIpm Σ'
[x plant x’s hedge]WS
Figure 8: Σ for example (11)
Finally, the past of narration in (12) is analogously represented by taking into account the pragmatic source of information that, in merger representations of DS, can override the information potentially carried by the grammar. Hence, (12') obtains the past-time reference via CPIpm as in Figure 9.
(12')
I enter the office.
For the purpose of the current argument it is sufficient to represent how the past-time reference is assigned to the present-tense form. However, we must remember that merger representations, being the ‘pragmaticky’ offspring of the discourse representation structures of Discourse Representation Theory (Kamp and Reyle 1993), represent entire discourses. Hence (12') is normally part of a for a larger chunk of discourse as in (12) and the discourse condition ‘yesterday (t)’ is already
Page 39 of 45 present there by virtue of processing the previous sentence. The superscript pn stands for ‘past of narration’.
Page 40 of 45
x t Σ' [the speaker]CD (x) yesterday (t) [ACCpn ├ Σ']WS, CPIpm Σ'
[x enter the office]WS
Figure 9: Simplified for example (12') as part of (12)
All in all, having the processes identified in DS at our disposal, aided by the requirement of treating them all on an equal footing as far as their contribution to the compositional mental merger representation is concerned, allows us to represent the linguistic diversity of means of expressing temporal reference and temporal ordering, at the same time making use of universal principles of utterance interpretation.
8. Concluding remarks: The depth of diversity
In discussing the various solutions to expressing temporal reference and temporal ordering I have demonstrated how adopting merger representations allows for representing the linguistic diversity of means of conveying information (the lexicon/grammar/pragmatics trade-offs) by allocating them to contributing sources and processes. An adequate representation of this diversity requires that all contributing sources and processes are treated on an equal footing. The commonsense
Page 41 of 45 principle of efficiency in communication may require leaving some aspects of intended meaning to pragmatic means such as the capturing of defaults or situated inference. In searching for universal sources of information and universal processes of discourse interpretation it was proposed that the contextualist orientation, and in particular its radical version in DS, withstands the test for cross-linguistic applicability set up in this chapter. The final question to address is whether in explaining language use one ought to focus on language diversity or rather on universal patterns. Evans and Levinson, (2009: 436), in their debate with generative syntacticians, opt for the first. They make a programmatic statement that languages and cultures have adaptive character, reflect the cultural and ecological interests of the community, and it is these diversified solutions that languages adopt for this task that should be the core of linguistic theory. Mapping languages onto formal (whether conceptual or not) structures disposes of this core point of interest of language study. I hope to have demonstrated in this chapter, in the case of some simple examples of the variety of means available for expressing temporality, that neither focusing on the variety of means nor stressing the universality of semantic/pragmatic principles and processes is a preferable option. They have to be considered in tandem, à la Kantian percepts and concepts, in an adequate account of intended meaning in communication – and thereby in any adequate, compositional theory of meaning.
Page 42 of 45
References
Asher, Nicholas & Alex Lascarides. 2003. Logics of Conversation. Cambridge: Cambridge University Press. Carston, Robyn. 1988. Implicature, explicature, and truth-theoretic semantics. In R. M. Kempson, ed., Mental Representations: The Interface between Language and Reality, 155–181. Cambridge: Cambridge University Press. Carston, Robyn. 1998. Postscript (1995) to Carston (1988). In A. Kasher, ed., Pragmatics: Critical Concepts. Vol. 4, 464–479. London: Routledge. Carston, Robyn. 2007. How many pragmatic systems are there? In M. J. Frápolli, ed., Saying, Meaning and Referring: Essays on François Recanati’s Philosophy of Language, 18– 48. Basingstoke: Palgrave Macmillan. Evans, Nicholas & Stephen C. Levinson. 2009. The myth of language universals: Language diversity and its importance for cognitive science. Behavioral and Brain Sciences 32: 429–492. Filipović, Luna. 2010. Thinking and speaking about motion: Universal vs. language-specific effects. In G. Marotta, A. Lenci, L. Meini, & F. Rovai, eds., Space in Language, 235-248. Pisa: University of Pisa Press. Fleck, David W. 2007. Evidentiality and double tense in Matses. Language 83: 589–614. Fodor, Jerry A. 2008. LOT 2: The Language of Thought Revisited. Oxford: Clarendon Press. Givón, Talmy. 2005. Context as other Minds: The Pragmatics of Sociality, Cognition and Communication. Amsterdam & Philadelphia: Benjamins. Hawkins, John A. 2004. Efficiency and Complexity in Grammars. Oxford: Oxford University Press. Hawkins, John A. 2009. Language universals and the performance-grammar correspondence hypothesis. In M. H. Christiansen, C. Collins, & S. Edelman, eds., Language Universals, 54 – 78. Oxford: Oxford University Press. Hornstein, Norbert. 1990. As Time Goes By: Tense and Universal Grammar. Cambridge, MA: MIT Press. Jaszczolt, Katarzyna M. 2002. Semantics and Pragmatics: Meaning in Language and Discourse. London: Longman. Jaszczolt, Katarzyna M. 2005. Default Semantics: Foundations of a Compositional Theory of Acts of Communication. Oxford: Oxford University Press.
Page 43 of 45 Jaszczolt, Katarzyna M. 2008. Psychological explanations in Gricean pragmatics and Frege’s legacy. In I. Kecskes & J. Mey, eds., Intentions, Common Ground, and the Egocentric Speaker-Hearer, 9–45. Berlin: Mouton de Gruyter. Jaszczolt, Katarzyna M. 2009. Representing Time: An Essay on Temporality as Modality. Oxford: Oxford University Press. Jaszczolt, Katarzyna M. 2010a. ‘Default Semantics’. In B. Heine & H. Narrog, eds., The Oxford Handbook of Linguistic Analysis, 215–246. Oxford: Oxford University Press. Jaszczolt, Katarzyna M. 2010b. Defaults in semantics and pragmatics. Second edition. In E. N. Zalta, ed., Stanford Encyclopedia of Philosophy.
http://plato.stanford.edu/contents.html Jaszczolt, Katarzyna M. forthcoming. Semantics and pragmatics: The boundary issue. In K. von Heusinger, P. Portner, & C. Maienborn, eds., Semantics: An International Handbook of Natural Language Meaning. Berlin: Mouton de Gruyter. Jaszczolt, Katarzyna M. & Jiranthara Srioutai. 2011. Communicating about the past through modality in English and Thai. In A. Patard & F. Brisard, eds., Cognitive Approaches to Tense, Aspect, and Epistemic Modality, 249–278. Amsterdam & Philadelphia: Benjamins. Kamp, Hans & Uwe Reyle. 1993. From Discourse to Logic: Introduction to Modeltheoretic Semantics of Natural Language, Formal Logic and Discourse Representation Theory. Dordrecht: Kluwer. Kecskes, Istvan. 2008. Dueling context: A dynamic model of meaning. Journal of Pragmatics 40: 385–406. Kecskes, Istvan. 2010. The paradox of communication: Socio-cognitive approach to pragmatics. Pragmatics and Society 1: 50–73. Levinson, Stephen C. 2000. Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, MA: MIT Press. Levinson, Stephen C. 2003. Space in Language and Cognition: Explorations in Cognitive Diversity. Cambridge: Cambridge University Press. MacFarlane, John. 2005. Making sense of relative truth. Proceedings of the Aristotelian Society 105: 321–339. MacFarlane, John. 2011. Relativism and knowledge attributions. In S. Bernecker & D. Pritchard, eds., Routledge Companion to Epistemology, 536-544. London: Routledge. Mauri, Caterina & Johan van der Auwera. 2012. Connectives. In K. Allan & K. M. Jaszczolt, eds., The Cambridge Handbook of Pragmatics, 377-401. Cambridge: Cambridge University Press.
Page 44 of 45 Müller, Ralph-Axel. 2009. Language universals in the brain: How linguistic are they? In M. H. Christiansen, C. Collins & S. Edelman, eds., Language Universals, 224 –252. Oxford: Oxford University Press. Nicolle, Steve & Billy Clark. 1999. Experimental pragmatics and what is said: A response to Gibbs and Moise. Cognition 69: 337–354. Pinker, Steven & Ray Jackendoff. 2009. The components of language: What’s specific to language, and what’s specific to humans. In M. H. Christiansen, C. Collins & S. Edelman, eds., Language Universals, 126 –151. Oxford: Oxford University Press. Pitts, Alyson. 2005. Assessing the evidence for intuitions about what is said. M.Phil. essay, University of Cambridge. Recanati, François. 1989. The pragmatics of what is said. Mind and Language 4. Reprinted in S. Davis, ed., 1991, Pragmatics: A Reader, 97–120. Oxford: Oxford University Press. Recanati, François. 2004. Literal Meaning. Cambridge: Cambridge University Press. Recanati, François. 2005. Literalism and contextualism: Some varieties. In G. Preyer & G. Peter, eds., Contextualism in Philosophy: Knowledge, Meaning, and Truth, 171–196. Oxford: Clarendon Press. Recanati, François. 2010. Truth-Conditional Pragmatics. Oxford: Oxford University Press. Recanati, François.2012. Contextualism: Some varieties. In K. Allan & K. M. Jaszczolt, eds., The Cambridge Handbook of Pragmatics, 135-149. Cambridge: Cambridge University Press. Saul, Jennifer M. 2002. What is said and psychological reality; Grice’s project and relevance theorists’ criticisms. Linguistics and Philosophy 25: 347–372. Schiffer, Stephen. 1991. Does Mentalese have a compositional semantics? In B. Loewer & G. Rey, eds., Meaning in Mind: Fodor and his Critics, 181–99. Oxford: Blackwell. Schiffer, Stephen. 1992. Belief ascription. Journal of Philosophy 89: 499–521. Schiffer, Stephen. 1994. A paradox of meaning. Noûs 28: 279–324. Schiffer, Stephen. 2003. The Things We Mean. Oxford: Clarendon Press. Schneider, Anna. 2009. Post-Gricean Pragmatics without the Syntactic Constraint: A Study with Reference to Requests in Russian and British English. PhD thesis, University of Cambridge. Sperber, Dan & Deirdre Wilson 1995. Relevance: Communication and Cognition. Oxford: Blackwell. Second edition. First published in 1986. Srioutai, Jiranthara. 2006. Time Conceptualization in Thai with Special Reference to d1ay1II, kh3oe:y, k1aml3ang, y3u:I and c1a. PhD thesis. University of Cambridge. Stanley, Jason. 2002. Making it articulated. Mind and Language 17: 149-68. Stanley, Jason. 2007. Language in Context: Selected Essays. Oxford: Clarendon Press. Stanley, Jason & Zoltán Gendler Szabó. 2000. ‘On quantifier domain restriction’. Mind and
Page 45 of 45 Language 15: 219–261. Szabó, Zoltán Gendler. 2000. Compositionality as supervenience. Linguistics and Philosophy 23: 475– 505. von Fintel, Kai & Lisa Matthewson. 2008. Universals in semantics. The Linguistic Review 25. 139– 201.