‘Default’ can mean many different things in theories ofmeaning. It is so not only because of the multiplicity of approachesand dimensions from which meaning can be studied but also due to thefact the theoretical landscape is changing swiftly and dynamically.First, there is the dimension of the ongoing debates concerning thedelimitation of explicit content (e.g., Jaszczolt 2009a, 2016a).Second, discussions about ‘defaultness’ are propelled bythe debates concerning the literal/nonliteral vis-à-vissalient/nonsalient distinction (e.g., Giora & Givoni 2015; Ariel2016). Next, defaultness plays an important role in computationallinguistics that develops statistical models for learningcompositional meaning using ‘big data’ (Jurafsky &Martin 2017 [Other Internet Resources]; Liang & Potts 2015). Morerecently, the ongoing ‘revolution’ in philosophy oflanguage noticeable in the progressing departure from analyticphilosophy has brought in the debates on socially expected, orsocially and politically correct – and as such, standard, ordefault – meanings in the context of discussions on epistemicnorms for assertion (e.g., Goldberg 2011, 2015), theories ofargumentation (e.g., Macagno 2022), and, relatedly, commitment andaccountability for one’s linguistic actions (e.g., Haugh 2013;Borg 2019). Defaults are also relevant in the discussions of theconventional import of lexical items such as expressives in that theirstandard expressive (often offensive) meaning may not arise in certaintypes of context; the view on their truth-evaluability (or at least onwhat aspects of their meaning are truth-evaluable) is then closelyrelated to the value one attaches to this context-dependence (Potts2007, 2012; Richard 2008; Geurts 2007).

All in all, the term ‘default meaning’ has been used in avariety of ways in the literature, including statistically commoninterpretation, predictable meaning, salient meaning, or automaticallyretrieved meaning. To begin with a common-sense definition, defaultinterpretation of the speaker’s utterance is normally understoodto mean salient meaning intended by the speaker, or presumed by theaddressee to have been intended, and recovered (a) without the help ofinference from the speaker’s intentions or (b) without anyconscious inferential process whatsoever. Default interpretations,interpretations producing the standard content, are defineddifferently depending on how ‘default’ is defined: as adefault for the lexical item, a default for the syntactic structure, adefault for a particular construction, or even a default for aparticular context (where, in addition, there is a necessarycorrelation with the adopted definition of ‘context’). Thedelimitation of such defaults can proceed according to differentmethods that, again, can affect the results, and as such furthercontribute to the definition of defaults. For example, thepsychological route is associated with automatic, inference-freeinterpretations, while the statistical route appeals to quantitativeanalyses of data, where the latter can pertain to corpora ofconversations or big databases of word co-occurrence as used instatistical, distributional approaches in computational semantics.

In what follows, I attend to such seminal conceptualisations ofdefaultness, their provenance, and their relative merits (Sections1–3). Section 4 follows with some remarks on what can be called‘dynamic defaultness’ in the theories and models of ajoint construction (‘co-construction’) of meaning that aresteadily gaining ground across different strands of pragmatic research(Lewis 1979; Asher and Lascarides 2013; Elder and Haugh 2018). Next, Ibriefly move to the role of defaults on the crossroads of philosophyof language with disciplines such as ethics, epistemology, and law(Section 5).

The overview of the major perspectives and debates makes it clear thatthere is no consensus in the literature as to the unique set ofproperties that default interpretations should exhibit, opening up thediscussion as to whether the term has only an intra-theoretic utility.Next, of course, comes the question of the utility of theterm and the utility of the concept –something I touch upon in the concluding section in the context ofconceptual engineering.

• 1. Default Interpretations in Semantics and Pragmatics
  • 1.1 Defaults, the Said, and the Unsaid
  • 1.2 Default Reasoning
  • 1.3 Presumptive Meanings and Cancellability
  • 1.4 Rhetorical Structure Rules
  • 1.5 Computational Semantics Landscape
  • 1.6 Defaults in Optimality-Theory Pragmatics
  • 1.7 Defaults in Truth-Conditional Pragmatics
  • 1.8 Types of Defaults in Default Semantics
• 2. Definitional Characteristics of Default Interpretations
• 3. Defaults, Salience, and Literalness
• 4. Co-construction of Meaning: Towards Dynamic, Interactional Defaults
• 5. Defaults and Accountability
• 6. Concluding Remarks and Future Prospects
• Bibliography
• Academic Tools
• Other Internet Resources
• Related Entries

1. Default Interpretations in Semantics and Pragmatics

1.1 Defaults, the Said, and the Unsaid

In post-Gricean pragmatics it has been accepted that communicatorsconvey more information than is contained in the expressions theyutter. For example, sentences (1a)–(2a) normally convey(1b)–(2b).

(1a)

Tom finished writing a paper and went skating.

(1b)

Tom finished writing a paperand then went skating.

(2a)

Picasso’s painting is of a crying woman.

(2b)

The painting executed by Picasso is of a crying woman.

Such additions to the content of the uttered sentence were called byGrice (1975)generalized conversational implicatures (GCIs),that is, instances of context-independent pragmatic inference.Subsequently, the status of such context-independent additions hasbecome the subject of heated debates. Some post-Griceans stay close toGrice’s spirit and propose that there are salient, unmarked,presumed meanings that occur independently of context (Horn, e.g.,2004, 2012; Levinson 1995, 2000; Recanati 2004). Some identify defaultmeanings as those arising automatically in a given situation ofdiscourse (Jaszczolt, e.g., 2005, 2010, 2016b; Elder & Jaszczolt2016). Others reject defaultstout court and subsume suchsalient meanings under a rather broad category of context-dependentpragmatic inference (Sperber & Wilson 1986; Carston 2002).

Next, some, following Grice, consider such pragmatic contributions toutterance meaning to be generalized conversational implicatures(Levinson), others classify them as pragmatic input to what is said,albeit using a variety of theory-specific labels (Recanati, Carston),reserving the term ‘implicature’ for meanings that can berepresented by a separate logical form and that function independentlyfrom the content of the main utterance in reasoning. Others definethem as contributions to primary meanings where the latter cut acrossthe explicit/implicit divide (Jaszczolt). Yet another possibility isto regard them as a separate level of what is implicit in what is said(Bach 1994, 2007; Horn 2006). In short, the status of such‘default’ meanings is still far from clear. However, atleast in general terms, there is a reason for drawing a distinctionbetween salient, automatic enrichments and costly pragmatic inferencesince some of these pragmatic contributions go throughnormally, unnoticed, as a matter of course. As Horn (2004:4–5) puts it,

Whatever the theoretical status of the distinction, it is apparentthat some implicatures are inducedonly in a special context(…), while others go throughunless a special contextis present ….

In the above, the differences in using the term ‘default’consist of the acceptance or rejection of at least the followingproperties:

1. cancellability (also known asdefeasibility) of preferredinterpretations;
2. availability of preferred interpretations without making use ofconscious inference;
3. shorter time required for their formation by the speaker andrecognition by the addressee as compared with that required for themeanings induced through inference;
4. the availability of preferred interpretations prior to thecompletion of the processing of the entire proposition (local,pre-propositional defaults).

When analysed in standard truth-conditional semantics, defaults cancontribute to the truth-conditional content or affect what is implicit– presupposed or implicated (see e.g., Potts 2015). The side onwhich we find defaults in this distinction is largely dictated by theorientation concerning the semantics/pragmatics boundary, where thechoice ranges from traditional semantic minimalism to radical versionsof contextualism. I discuss these in more detail in the followingsections. But it has to be remembered that the category is tangentialto such concepts as what is said, conversational implicature,conventional implicature, presupposition, or, to use a more generalterm, projective content (on universals in projective content seeTonhauseret al. 2013). For example, presuppositions arestronger than defaults: presupposition triggers such as‘know’, ‘regret’, ‘again’ or‘manage’ do not give the hearer much option ofinterpretation, save admitting some form of metalinguistic orquotative reading when these are negated as in (3).

(3)

I didn’t forget about your birthdayagain; it is the first time it happened.

What is said can rely on various types of defaults (Section 2) andcontextually salient interpretations (Section 3), but likewise it canrely on effortful pragmatic inference from a variety of sourcesavailable in the situation of discourse. Relevant implicatures can beconventional (Grice 1975; Potts 2005) and conversational generalised,the latter understood either as grammar-driven (Chierchia 2004) or,more loosely, language-system-driven (Levinson 2000 and Section 1.3),but implicatures can also be entirely context-dependent(particularised). To add to this multi-dimensionality,context-dependent implicatures can on some occasions ariseautomatically, so when our definition of defaults relies on thedefinitional criterion of the automaticity of the process, asdiscussed above, then, by this definition, such implicatures can alsobe dubbed ‘defaults’ (Giora & Givoni 2015; Jaszczolt2016a). In short, pursuing the standard route of analysing the typesof content will not get us far with analysing defaults.Said/implicated, at-issue, or question-under-discussion-drivenanalyses (e.g., Roberts 2004) will encounter defaults on either side ofthe pertinent dichotomies.

A further complication in linking defaultness with the categories ofthe said or the unsaid is the fact that even weak implicatures orpresuppositions adopted through accommodation can enjoy either status.In (4), we can accommodate globally the presupposition in (5) –either via inference or automatically.

(4)

Tom says that Ian hasn’t finished writing a novel.

(5)

Ian is writing a novel.

As a result, (5) can enjoy the default status according to some of thestandard understandings of defaults as automatic, or more frequent,more salient, or even more ‘literal’ interpretations, or,alternatively, it can simply be an interpretation that is easier toprocess – arguably, in itself a plausible criterion for‘defaultness’.

Next, conventional implicatures, that is lexical meanings that,according to Grice (1975), do not contribute to what is said, have, atfirst glance, less to do with defaultness: they are entrenched,non-cancellable and form-dependent (detachable), and they cannot becalculated from maxims, principles or heuristics (Horn 1988: 123).However, more recent inquiries into the related category ofexpressives gives more scope for pursuing defaultness. Slurs are,arguably, offensive by default but their derogatory import does notcarry across to context of banter and camaraderie. As to whether theexpressive content is an implicature or part of what is said, thematter is still hotly discussed (see e.g., Richard 2008; Sileo 2017).In what follows, I try to bring some order into this unwieldy termvis-à-vis the semantics/pragmatics distinction and finish withsome reflections on its usefulness for semantics and pragmatics.

1.2 Default Reasoning

Be that as it may, default meanings come from default reasoning.According to Kent Bach (1984), in utterance interpretation we use‘jumping to conclusions’, or ‘defaultreasoning’. In other words, speakers know when context-dependentinference from the content of the sentence is required and when it isnot. When it is not required, they progress, unconsciously, to thefirst available and unchallenged alternative. This step is cancellablewhen it becomes obvious to the addressee that the resulting meaning isnot what the speaker had intended. What is important in this view isthe proposed distinction between (conscious) inference and theunconscious act of ‘taking a step’, as Bach (1984: 40)calls it, towards the enriched, default interpretation. Such a move tothe default meaning is not preceded by a conscious act of deliberationas to whether this meaning was indeed intended by the speaker. Rather,it just goes through unless it is stopped by some contextual or otherfactors that render it implausible.

Bach founds his account on the Gricean theory of intentionalcommunication and therefore he has a ready explanation for the factthat different meanings come with different salience. He makes anassumption that intentions allow for different degrees of strength(Bach 1987). He also adds that salience has a lot to do withstandardisation (Bach 1995; 1998) which consists of interpreting anutterance according to a pattern that is established by previous usageand as such short-circuits the process of (conscious) inference. Inshort, ‘jumping to conclusions’ is performed unconsciouslyand effortlessly.

For Bach, such default meanings are neither implicatures nor what issaid (or explicatures): they are implicit in what is said, orimplicitures. They are a result of ‘fleshing out’the meaning of the sentence in order to arrive at the intendedproposition, or ‘filling in’ some conceptual gaps in thesemantic representation that, only after this filling in, becomes afull proposition. An example of ‘fleshing out’ is given in(6b), where the minimal proposition is expanded. ‘Fillingin’ is exemplified in (7b), where a so-called propositionalradical is completed.

(6a)

Tom is too young.

(6b)

Tom is too young to drive a car.

(7a)

Everybody likes philosophy.

(7b)

Everybody who reads the SEP likes philosophy.

But default meanings do not exhaust the category membership of theimpliciture: implicitures can be a result of default reasoning as wellas a context-dependent process of inference. Analogous to thedistinctions discussed before, default meanings are orthogonal to thedistinction between what is said, impliciture, and implicature: thedefault/inferential distinction cuts across all three.

1.3 Presumptive Meanings and Cancellability

Stephen Levinson (1995, 2000) argues for default interpretations thathe callspresumptive meanings and classifies as implicatures.He uses the term borrowed from Grice, generalized conversationalimplicatures (GCIs), but ascribes some properties to them thatdifferentiate them from Grice’s GCIs. For Levinson, GCIs areneither properly semantic nor properly pragmatic. They should not beregarded as part of semantics as, for example, in DiscourseRepresentation Theory (Kamp and Reyle 1993), nor should they be seenas a result of context-dependent inference performed by the hearer inthe process of the recovery of the speaker’s intention. Instead,“they sit midway, systematically influencing grammar andsemantics on the one hand and speaker-meaning on the other.”(Levinson 2000: 25).

Such presumed meanings are the result of rational, communicativebehaviour and arise through three assumed heuristics: (1) ‘Whatisn’t said, isn’t’; (2) ‘What is expressedsimply is stereotypically exemplified’, and (3)‘What’s said in an abnormal way isn’t normal’,called Q, I, and M heuristics (principles) respectively.Levinson’s GCIs, unlike their Gricean progenitors, can arise atvarious stages in utterance processing: the hearer need not haveprocessed the whole proposition before arriving at some presumedmeanings. Also, unlike Grice’s GCIs that are taken to bespeaker’s intended meanings, Levinson’s presumptivemeanings seem to behearer’s meanings, obtained by thehearer as a result of the assumptions he or she made in the process ofutterance interpretation (see Saul 2002 and Horn 2006 for discussion).On the other hand, like Grice’s GCIs, they are cancellablewithout contradiction.

Now, when defaults are delimited by contextual salience, arguably,cancellation may not occur except for cases of miscommunication. Inother words, when the meaning is salient in a given context, it islikely that it had been meant by the speaker unless the speakermisjudged the common ground. But when they are understood aslanguage-system-driven meanings, à la Levinson’s GCIs,cancellability constitutes direct evidence of such defaultness.Salient components of meaning added to the overtly expressed content(in the form of additional information or choices of interpretation)tend to be entrenched and as such difficult to cancel. But asJaszczolt (2009a, 2016a) demonstrates, cancellability is a propertythat does not side with implicit as opposed to explicit content butrather with salience. If the main intended message is communicatedindirectly, as in (8b), then it is the implicature (8c) that isdifficult to cancel.

(8)

(Fred and Wilma talking about Wilma’s piano recital)

(8a)

Fred: Was the recital a success?

(8b)

Wilma: Lots of people left before the end.

(8c)

The recital was not a success.

The presence or absence of cancellation in utterance interpretation isstill a matter of dispute. It is difficult at present to decidebetween the rival views (i) that a particular GCI arose and wassubsequently cancelled or (ii) that it did not arise at all due tobeing blocked by the context. There is not sufficient experimentalevidence to support either stance. The answer to this question isclosely dependent on the answer to the so-called globalism-localismdispute. If, as Levinson claims, default interpretations arise‘locally’, out of the processing of a pre-propositionalunit such as a word or a phrase, then they have to be subjected tofrequent cancellation once the proposition has been processed. If,however, despite the incrementality of the interpretation process theyarise post-propositionally, or ‘globally’, in accordancewith Grice’s original assumption, then utterance interpretationcan proceed without costly backtracking (see Geurts 2009; Jaszczolt2008, 2009a, 2016a; Noveck & Sperber 2004).

1.4 Rhetorical Structure Rules

Gricean pragmatics is not the only approach in which defaults arediscussed. Defaults and nonmonotonic reasoning are also wellentrenched incomputational linguistics. Defaults are distinguished there with respect to various units ofmeaning, from morphemes and words to multisentential units (Asher& Lascarides 1995; Lascarides & Copestake 1998). In thissection I focus on intersentential default links and in the next Iplace ‘glue logic’ in the context of some otherunderstandings of defaults in computational semantics.

The tradition of defaults in nonmonotonic reasoning can be traced backto Humboldt, Jespersen and Cassirer, and more recently toReiter’s (1980) default logic and his default rules of theform:

A:B
C

whereC can be concluded ifA has been concluded andB can be assumed (andnot B cannot be proven). Suchdefaults can be built into standard logic:

It is just as valid to conclude ‘Presumablyx isB’ from ‘x isA’ and‘A’s are normallyB’ as it is toconclude ‘x isB’ from ‘xisA’ and the ‘AllA’s areB’. One does not have to set one’s mind to adifferent mode of reasoning to get the former. Veltman (1996: 257).

But the resulting logic will become nonmonotonic because there aredefault rules and default operators in the language. The literature onthe topic is vast and is best considered as a separate topic from ourcurrent concern (see e.g., Thomason (1997) for an overview).

A good example of how default interpretations can be accounted for informal semantic theory is Segmented Discourse Representation Theory(SDRT, e.g., Asher & Lascarides 2003). SDRT is an offshoot ofDiscourse Representation Theory, a dynamic semantic approach tomeaning according to which meaning arises incrementally throughcontext change. In SDRT, defaults are regarded as highly probableroutes that an interpretation of a sentence may take in a particularsituation of discourse. There are rules of discourse, so-calledrhetorical structure rules, that produce such defaultinterpretations. These rules spell out the overall assumption thatdiscourse is coherent and that this coherence can be furtherelaborated on by proposing a set of regularities. For example, twoevents represented as two consecutive utterances are presumed to standin the relation ofNarration, where the event described inthe first utterance precedes the one from the second utterance. If thesecond utterance describes a state, then it stands in the relation ofBackground to the first one. There are many other types ofsuch relations, among them Explanation and Elaboration. Axioms preventa relation from being of two incompatible types at the same time. Therelations between states and events are computed as strongprobabilities, in the process called defeasible reasoning. The laws ofreasoning are ‘defeasible’ in the sense that if theantecedent of a default rule is satisfied, then its consequent isnormally, but not always, satisfied. The inferencenormally,but notalways, obtains: ceteris paribus, the relationpredicted by the law obtains, but in certain circumstances it may not.It is also nonmonotonic in that the relation may disappear with thegrowth of information.

SDRT includes the following components: (i) the semantics of sentencesalone, that is the underspecified output of the syntactic processingof the sentences; (ii) the semantics of information content, that is,further addition to these underdetermined meanings, including defaultadditions summarised by rhetorical structure rules; and (iii) thesemantics of information packaging that ‘glues’ suchenriched representations by means of the rules of the rhetoricalstructure of discourse. This ‘gluing together’ isdefeasible, in that the rules result in the dependencyA>B, that is ‘ifA, then normallyB’, whereA andB stand for theenriched propositional representations of two sentences. In otherwords, they stand for the meanings of two consecutive utterances.

The main strength of this approach is that it is fully formalized andit allows for computational modelling of discourse that takespragmatic links between utterances seriously and incorporates them inthe semantics. Next, it also aspires to cognitive reality and althoughthe cognitive reality of the particular rules can be disputed, theview of discourse processing that they jointly produce is highlyplausible. Finally, as the authors often stress, SDRT allows them, formost part, to model discourse without recourse to speakers’intentions. However, a direct comparison with Gricean accounts ofdefaults is precluded by the fact that we would not be comparing likewith like. In SDRT, the default interpretations are the defaults thatare formalized with respect to the actually occurring discourse: thereare rules that tell us how to take two events represented in twoconsecutive sentences, there are also rules that specify the relationbetween them depending on some features of their content. Griceandefaults are, on the contrary, defaults for speakers’ overallknowledge state: they may arise because the speaker didnotsay something he or shecould have said or because thespeaker assumed some cultural or social information to be sharedknowledge. For example, we cannot formalize the interpretation of (9a)as (9b) by means of rhetorical structure rules. The interpretation of(9a) as (9b) fits under the SDRT component (ii) rather than (iii)above, i.e., the semantics of informationcontent rather thanpackaging.

(9a)

Pablo’s painting is of a crying woman.

(9b)

Picasso’s painting is of a crying woman.

Finally, it has to be mentioned that the discourse relations that forAsher and Lascarides belong to the ‘glue logic’ canalternatively be conceived of as part of the grammar proper: Lepore& Stone (2015), for example, incorporate conventions intominimalistically understood, grammar-driven semantics, and a fortioriinto grammar proper; following Lewis’s (1979) ideas onconvention and ‘scorekeeping’, they propose that“semantics describes interlocutors’ social competence incoordinating on the conversational record” (Lepore & Stone2015: 256). Merits of putting conventions into grammar are, however,not easy to find (for a review see Jaszczolt 2016b).

1.5 Computational Semantics Landscape

The computational semantics landscape contains a few landmarks inwhich the concept of a default figures prominently, albeit underdifferent labels. I have already discussed the role of defaults andinheritance reasoning in artificial intelligence research in theexample of SDRT. This kind of research in computational linguistics isarguably the closest to theoretical linguistic semantics andpragmatics in that it directly appeals to human practices inreasoning. Pelletier & Elio (2005) refer to this characteristic asthe psychologism of nonmonotonic logics, and thus a property that wasso fiercely banished from logic by Frege as a form of a‘corrupting intrusion’, in that ‘being trueis quite different frombeing held as true’ (Frege1893: 202). Pelletier and Elio write:

Unlike most other reasoning formalisms, nonmonotonic or defaultreasoning, including inheritance reasoning,is“psychologistic” – that is, it is defined only aswhat people do in circumstances where they are engaged in“commonsense reasoning”. It follows from this that suchfundamental issues as “what are the good nonmonotonicinferences?” or “what counts as ‘relevance’ toa default rule?”, etc., are only discoverable by looking atpeople and how they behave. It isnot a formalexercise to be discovered by looking at mathematical systems, nor isit to be decided by such formal considerations as“simplicity” or “computability”, etc.Pelletier & Elio (2005: 30).

Other landmarks include research on default feature specification insyntactic theory and default lexical inheritance (e.g., Gazdaretal. 1985; Boguraev & Pustejovsky 1990; Lascarides et al.1996), where default inheritance comes from a simple idea pertainingto all taxonomies: regular features belonging to an entity of acertain type are inherited from the categories higher up in thetaxonomic hierarchy, that is simply by virtue of the membership of acertain ontological type. As a result, only the non-default featureshave to be attended to (on various semantic networks in computationallinguistics see also Stone 2016). To generalize, this line of researchcan lead to incorporation of information into logical forms,including, as can be seen in the example of SDRT, dynamic logicalforms of discourses. In a different camp there are statistical,distributional approaches to meaning where meaning is derived frominformation about co-occurrence of items gleaned from corpora and thenquantitatively analysed. This orientation gave rise to currentvector-based approaches (see, e.g., Jurafsky & Martin 2017 [OtherInternet Resources]; Coeckeet al. 2010 and for discussionLiang & Potts 2015). Vector semantics exploits the finding thatdates back at least to Harris (1954) and Firth (1957) that the meaningof a word can be computed from the distribution of the words in itsimmediate context. The term ‘vector semantics’ derivesfrom the representation of the quantitative values in thisdistribution called a ‘vector’, where the latter isdefined as a distributional model that presents information in theform of a co-occurrence matrix. Vectors have been around since the1950s but it is only recently that such distributional methods havebeen combined with logic-based approaches to meaning (see Liang &Potts 2015). Vectors can measure the similarity of texts with respectto a lexical item, the similarity of lexical items with respect tosources, or, what interests us most, the co-occurrence of selectedwords in a selection of contexts (using additional methods to rule outco-occurrence by chance). In distributional semantics therefore thesalient or default meaning is the meaning given by the observed highco-occurrence or, in other words, delimited by the high conditionalprobability of its occurrence in the context of other words.

Current compositional semantics is beginning to combine compositionalsemantic theory (logic-based approaches discussed above) withstatistical models, conforming to the standard view ofcompositionality on which complex meanings are a function of lexicalmeanings and the mode of combination, arrived at through a recursiveprocess, but at the same time aiming at capturing the generalizationfrom (finite) past experiences that would inform machine learning.Defaults arise in this context in several different forms: (i) asshortcuts to standard meanings of more semantically predictablecategories, that is, closed-class words such as determiners, pronounsor sentential connectives. (This can be extended perhaps to types ofpredictable projective content such as various types of implicature orpresupposition; see Tonhauseret al. 2013); (ii) aspredictable cross-sentential discourse relations; (iii) as predictablediscourse-anaphoric links; (iv) as meaning arising from frequentsyntagmatic associations; (v) as meaning arising from frequentconversational scenarios, to name a few salient concepts. In this new,positively eclectic orientation in computational linguistics thatcombines logical and statistical approaches, the label‘default’ is likely to lead to more confusion than utilityin that it can pertain to either of the two contributing orientations.On the other hand, if the findings lead to the same set of what we cancall ‘shortcuts through possible interpretations’, theconfusion may be of merely a methodological rather than ontologicalimportance.

1.6 Defaults in Optimality-Theory Pragmatics

Optimality-Theory pragmatics (OT pragmatics, Blutner 2000; Blutner andZeevat 2004; ) is another attempt at a computational modelling ofdiscourse but unlike SDRT it makes use of a post-Gricean,intention-based account of discourse interpretation. The process ofinterpretation is captured in a set of pragmatic constraints. Thepragmatic additions to the underdetermined output of syntax aregoverned by a rationality principle called an optimization procedurethat is spelled out as a series of constraints. These constraints areranked as to their strength and they are defeasible, that is, they canbe violated (see Zeevat 2000, 2004). The resulting interpretation ofan utterance is the outcome of the working of such constraints. OTpragmatics formalizes and extends the Gricean principles ofcooperative communicative behaviour as found in Horn (1984) andLevinson (1995, 2000). For example, STRENGTH means preference forreadings that are informationally stronger, CONSISTENCY meanspreference for interpretations that do not conflict with the extantcontext, FAITH-INT stands for ‘faithful interpretation’,that is interpreting the utterance without leaving out any aspect ofwhat the speaker says. The ordering of these constraints is FAITH-INT,CONSISTENCY, STRENGTH. The interaction of such constraints, founded onLevinson’s heuristics, explains how the hearer arrives at theintended interpretation. At the same time, this model can be regardedas producing default, presumed interpretations. With respect tofinding an antecedent for an anaphor, for example, the interaction ofthe constraints explains the general tendency to look for the referentin the immediately preceding discourse rather than in the more remotefragments or, rather than constructing a referentad hoc. Inother words, it explains the preference for binding over accommodation(van der Sandt 1992, 2012).

Defaults in OT pragmatics combine the precision of a formal accountwith the psychological reality of Gricean intention-basedexplanations. The main difference is that they don’t seem to bedefeasible: OT pragmatics tells us how an actual interpretation arose,rather than what the default interpretation could be. Constraints areranked, so to speak,post hoc: they explain what actuallyhappened and why, rather than what should happen according to therules of rational communicative behaviour. In other words, context isincorporated even sooner into the process of utterance interpretationthan in Gricean accounts and allows for non-defeasible, albeitstandard, default, interpretations. With respect to this feature theyresemble defaults of Default Semantics discussed in Section 1.8.

1.7 Defaults in Truth-Conditional Pragmatics

In truth-conditional pragmatics (Recanati, e.g., 2004, 2010), themeaning of an utterance consists of the output of syntactic processingcombined with the output of pragmatic processing. Pragmaticprocessing, however, is not necessarily fulfilled by consciousinference: processes that enrich the output of syntax aresub-doxastic, direct, and automatic. The resulting representation ofutterance meaning is the only representation that has cognitivereality and it is subject to truth-conditional analysis. On thisaccount, the content of an utterance is arrived at directly, similarto the act of perception of an object. Recanati calls this viewanti-inferentialist in that “communication is as direct asperception” (Recanati 2002: 109): the processing of thespeaker’s intentions is (at least normally) direct, automatic,and unreflective. Such processes enriching the actually utteredcontent are called primary pragmatic processes. Some of them make useof contextual information, others are context-independent. So, theyinclude some cases of Grice’s GCIs as well as someparticularised implicatures (PCIs; on implied content see alsoTonhauseret al. 2013) – but only the ones whichfurther develop the logical form of the uttered sentence. When thepragmatic addition constitutes a separate thought, it is, on thisaccount, an implicature proper, arrived at through a secondary,conscious, and reflective pragmatic process.

There are two kinds of enrichment of the content obtained through thesyntactic processing: (i) completing of a semantically incompleteproposition as in (10b), calledsaturation, and (ii) furtherelaboration of the meaning of the sentence that is not guided by anysyntactic or conceptual gaps but instead is merely triggered by thehearer’s opinion that something other than the bare meaning ofthe sentence was intended, as in (11b). The latter process is calledfree enrichment.

(10a)

The fence isn’t strong enough.

(10b)

The fence isn’t strong enough to withstand the gales.

(11a)

John hasn’t eaten.

(11b)

John hasn’t eaten dinner yet.

Default interpretations are here defaults forprocessing of anutterance in a particular context. Automatic and unconsciousenrichment produces a default interpretation of the utterance and“[o]nly when there is something wrong does the hearer suspend orinhibit the automatic transition which characterizes the normal casesof linguistic communication”. (Recanati 2002: 109). To sum up,such defaults ensue automatically, directly, without the effort ofinference. They are cancellable, they can make use of contextualclues, but they are not ‘processes’ in any cognitivelyinteresting sense of the term: they don’t involve consciousinference, albeit, in Recanati’s terminology, they involveinference in the broad sense: the agent is not aware of performing aninference but is aware of the consequences of this pragmaticenrichment of the interpreted sentence.

1.8 Types of Defaults in Default Semantics

One of the main questions to ask about any theory of utteranceinterpretation is what sources information about meaning comes from.In Default Semantics, (Jaszczolt, e.g., 2009, 2010, 2016a, 2021),utterance meaning is the outcome of merging of information that comesfrom five sources: (i) word meaning and sentence structure (WS); (ii)situation of discourse (SD); (iii) properties of human inferentialsystem (IS); (iv) stereotypes and presumptions about society andculture (SC); and world knowledge (WK). WS is the output of thesyntactic processing of the sentence, or its logical form. SD standsfor the broadly understood context in which the discourse is immersed.IS pertains to properties of mental states which trigger certain typesof interpretations. For example, the property of intentionalityensures that we normally use referring expressions with a referentialintention that is the strongest for the given context. SC pertains tothe background knowledge of societal norms and customs and culturalheritage. WK encompasses information about physical laws, nature,environment, etc. It is important to stress that the four sources thataccompany WS do not merely enrich the output of the latter. All of thesources are equally powerful and can override each other’soutput. This constitutes a substantial breakaway from the establishedboundary between explicit and implicit content.

The identification of the sources also allows us to propose aprocessing model in Default Semantics in which three types ofcontribution to utterance interpretation are distinguished: (i)processing of the sentence (called combination of word meaning andsentence structure, WS); (ii) conscious pragmatic inference (CPI) fromthree of the sources distinguished above: SD, SC, and WK; and (iii)two kinds of default, automatic meanings: cognitive defaults (CD)triggered by the source IS, and social, cultural and world-knowledgedefaults (SCWD).

The primary meaning is arrived at through the interaction of theseprocesses and therefore need not bear close resemblance to the logicalform of the sentence; the output of WS can vary in significance ascompared with the output of other types of processes. For example, toborrow Bach’s (1994) scenario, let us imagine little Johnnycutting his finger and crying, to which his mother reacts by uttering(12a).

(12a)

You are not going to die.

Thewhat is said/explicature of (12a) is something to theeffect of (12b). There may also be other communicated meanings butthose fall in the domain of implicatures.

(12b)

You are not going to die from this cut.

In Default Semantics, the primary content of an utterance is its mostsalient meaning. This is so even when this meaning does not bear anyresemblance to the logical form derived from the syntactic structureof the uttered sentence. In other words, CPI can override WS andproduce, say, (12c) as utterance meaning (called primarymeaning, represented in amerger representation) for thegiven context. The explicit content of the utterance need not be evenpartially isomorphic with the meaning of the uttered sentence: it neednot amount to the development of the sentence’s logicalform.

(12c)

There is nothing to worry about.

CDs and SCWDs are default interpretations. Similar to Recanati’sautomatic free enrichment, these default meanings cut acrossGrice’s GCI/PCI divide. Some of them arise due to the propertiesof words or constructions used and are present by defaultindependently of the context of the utterance, while others aredefault meanings for the particular situation of discourse. CDs aredefault interpretations that are triggered by the properties of mentalstates. For example, when speakers use a definite description in anutterance, they normally use it referentially (about a particular,known, intersubjectively recognisable individual) rather thanattributively (about whoever fits the description). This defaultreferential use can be given a functional as well as a cognitiveexplanation. Firstly, it can be explained in terms of the strength ofthe referential intention associated with the act of utterance:ceteris paribus, humans provide the strongest information relevant andavailable to them. At the same time, in cognitive terms, it can beexplained through the property of mental states that underlie thespeaker’s speech act: this is the property ofintentionality oraboutness, in the sense in whichthe mental state is about a particular object, be it a person, thing,or situation. Like the strongest referring, so the strongestaboutness, is the norm, the default. For example, the description‘the architect who designed St Paul’s cathedral’ in(13a) is likely to be interpreted as ‘Christopher Wren’,as in (13b).

(13a)

The architect who designed St Paul’s cathedral was a genius.

(13b)

Sir Christopher Wren was a genius.

Next, SCWDs are default interpretations that arise due to the sharedcultural and social background of the interlocutors. To use a wellworn example, in (14a), it is the shared presumption that babies areraised by their own mothers that allows the addressee to arrive at(14b).

(14a)

The baby cried and the mother picked it up.

(14b)

The baby cried and the baby’s mother picked it up.

In CDs and SCWDs, no conscious inference is involved. The naturalconcomitant of reducing the role of the logical form (WS) to one offour equally potent constituents of utterance meaning is a revisedview of compositionality. The compositional nature of meaning isretained as a methodological assumption but this compositionality isnow sought at the level of the merger of information from the fivesources, arrived at through the interaction of the four identifiedprocesses. The output of these processes is called mergerrepresentation and is expected to be a compositional structure.Current research focuses of providing an algorithm for the interactionof the output of the identified processes.

2. Definitional Characteristics of Default Interpretations

It is evident from the sample of approaches presented above that thenotion of default meaning is used slightly differently in each ofthem. We can extract the following differences in the understanding ofdefault interpretations:

[1a] Defaults belong to competence.
vs.
[1b] Defaults belong to performance.

[2a] Defaults are context-independent.
vs.
[2b] Defaults can make use of contextual information.

[3a] Defaults are easily defeasible.
vs.
[3b] Defaults are not normally defeasible.

[4a] Defaults are a result of subdoxastic, automatic process.
vs.
[4b] Defaults can involve conscious pragmatic inference.

[5a] Defaults are developments of the logical form of the utteredsentence.
vs.
[5b] Defaults need not enrich the logical form of the sentence but mayoverride it (which is orthogonal to the question as to whetherdefaults have to be literal meanings)

[6a] Defaults can all be classified as one type of pragmatic process.
vs.
[6b] Defaults come from qualitatively different sources in utteranceprocessing.

There is also disagreement concerning the following properties, to bediscussed below:

[7a] Defaults are always based on a complete proposition.
vs.
[7b] Defaults can be ‘local’,‘sub-propositional’, based on a word or a phrase.

[8a] Defaults necessarily arise more quickly than non-defaultmeanings. Hence they can be tested for experimentally by measuring thetime of processing of the utterance.
vs.
[8b] Defaults do not necessarily arise more quickly than non-defaultmeanings because both types of meaning can be based on conscious,effortful inference. Hence, the existence of defaults cannot be testedexperimentally by measuring the time of processing of theutterance.

The ‘whose meaning’ question also give rise tocontroversies, such as those discussed in Section 4:

[9a] Default meanings are static: they are intended, or recovered, or(normally) both.
vs.
[9b] Default meanings are dynamic, interactional, and as such are theresult of the joint construction of meaning in conversation.

Questions on the boundary of linguistics and ethics, linguistics andepistemology, or linguistics and law (to name a few) also give rise torival positions, such as:

[10a] Speakers are accountable only for the minimal semantic contentof their utterances.
vs.
[10b] Speakers are accountable for standard, default (in one of theabove senses) interpretations or their utterances.

This may lead to more specific controversies, such as:

[10a′] What counts as a lie or an insult pertains to default (inone of the above senses) content.
vs.
[10b′] What counts as a lie or an insult pertains to minimalsemantic content.

These are addressed in Section 5.

[1]–[8] are the most standardly accepted characteristics ofdefault interpretations in theoretical semantics and pragmatics. Weshall not include here definitional characteristics of defaults incomputational linguistics as these are a subject for a separate study.Some of the properties in [1]–[8] are interrelated, some of theothers just tend to occur together. Levinson’s presumptivemeanings, for example, are defeasible, i.e., fulfil [3a], local [7b],pertain to competence [1a], and are faster to process than inferentialmeanings [8a]. They are competence defaults of the type [1a] becausethey arise independently of the situation of discourse and aretriggered by the construction alone, due to the presumed defaultscenario that it pertains to. For example, scalar inference from‘many’ to ‘not all’ is a case of acompetence-based, context-independent, local default. Similarly, therhetorical structure rules of SDRT give rise to competence defaults.(15b) is a result of the common, shared knowledge that pushingnormally results in falling.

(15a)

You pushed me and I fell.

(15b)

You pushed me and as a result I fell.

As regards feature 7, it is at least conceivable that presumedmeanings arise as soon as the triggering word or construction has beenprocessed by the hearer. For Levinson (1995, 2000), salient meaningshave this property of arising even before the processing of thesentence is completed but can subsequently be cancelled if furthercontext witnesses against them.. In other words, they arisepre-propositionally orlocally. Discourseinterpretation proceeds incrementally and similarly the assignment ofdefault meanings to the processed segments is incremental. Forexample, the scalar termmany in (16a) triggers the presumedmeaningnot all as soon as it has been processed. Thesubscript_d in (16b) stands for the default meaning and isplaced immediately after the triggering construction.

(16a)

Many people liked Peter Carey’s new novel.

(16b)

Many (_d many but not all) people liked Peter Carey’s new novel.

Similarly, ‘paper cup’ and ‘tea cup’ give riseto presumed meanings locally, as in (17b) and (18b) respectively.

(17a)

Those paper cups are not suitable for hot drinks.

(17b)

Those paper cups (_d cups made of paper) are not suitable for hot drinks.

(18a)

I want three tea cups, three saucers and three spoons please.

(18b)

I want three tea cups (_d cups used for drinking tea), three saucers and three spoons please.

Inferences such as those in (17b) and (18b) are very common. They are,however, substantially different from the inference in (16b) in thatthe resulting meaning is the lexical meaning of the collocation,similar to that of a compound. Other examples include ‘pocketknife’ vs. e.g., ‘bread knife’, and ‘coffeespoon’ vs. e.g., ‘silver spoon’. It is worthremembering that on Levinson’s account, presumed, salientinterpretations can be explained through the principles of rationalcommunicative behaviour summed up as his Q, I and M heuristics (seeSection 1.2 and Levinson 1995, 2000). (16b) arises through theQ-heuristic, ‘What isn’t said isn’t’, while(17b) and (18b) arise through the I-heuristic, ‘What isexpressed simply is stereotypically exemplified’. Mostgenerally, the defaults that arise through the Q-heuristic exploit acomparison with what was not, but might have been, said. For example,‘most’ triggers an inference to a denial of a strongeritem ‘all’; ‘believe’ triggers an inference to‘not know’. At the same time, they are all easilycancellable, as (16c) illustrates.

(16c)

Many, and possibly all, people liked Peter Carey’s new novel.

I-heuristic exploits only what there is in the sentence: it is aninference to a stereotype and as such is not so easily cancellable.For example, (19) and (20) seem rather bizarre.

(19)

Those paper cups, I mean cups used for storing paper, are full.

(20)

I want three tea cups, I mean cups used for storing tea leaves.

Perhaps the fact that these defaults are not so easily cancellablecomes from their property of resembling lexical compounds and, like inthe case of compounds, the link between the juxtaposed lexemes is verystrong in their case. If indeed it is plausible to treat them on a parwith compounds, then they are not very useful as a supporting argumentfor local defaults: instead of defaults, we have lexical meaning ofcompounds.

Local defaults allow us to dispose of the level of an underspecifiedpropositional representation in semantic theory. Since the inferencesproceed incrementally, then as soon as the triggering expression isencountered, there is no level of a minimal proposition that wouldconstitute a foundation for further inferences. If there is one, it isjust accidental, in that the triggering item may just happen to beplaced at the end of the sentence, for example ‘tea cups’in the first clause of (20) above. But it is also important to notethat the status of such defaults is still far from clear. For example,Levinson’s defaults are local, but at the same time‘cancellable’ to the extent that the context may preventthem from arising. This leads to a difficulty in examples such as(21)–(22).

(21)

You are allowed five attempts to get the prize.

(22)

You are allowed to do five minutes of piano practice today because it is late.

It is clear that in (21) ‘five’ is to be understood as‘at most five’. How are we to model the process ofutterance interpretation for this case? Are we to propose that theinference from ‘at least five’ to ‘exactlyfive’ takes place and is then cancelled? Or are we to proposethat ‘five’ is by default ‘at least five’ (orunderdetermined five, or ‘exactly five’, depending on theorientation (see Horn 1992; Koenig 1993; Bultinck 2005) and becomesaltered in the process of pragmatic inference to ‘at mostfive’ in the context of ‘allow’? But then,‘allow’ is also present in (22) and the inference to‘at most’ is not at all salient: doing a longer pianopractice is generally preferred but may not be what the addresseelikes doing and ‘five’ may end up, in this context, tomean ‘as little as five’ or ‘five or more’,stressing that more than five is not expected but allowed. In (23),the problem is even more salient. If ‘five’ triggerslocally the ‘exactly’ meaning, then the default has to becancelled immediately afterwards when ‘are needed’ hasbeen processed and the ‘at least’ interpretation becomesobvious.

(23)

Five votes are needed to pass the proposal.

Alternatively, we can stipulate that the first inference takes placeafter the word ‘needed’. It is clear that a lot needs tobe done to clarify the notion of local defaults: most importantly, (i)what counts as the triggering unit, (ii) to what extent context isconsulted, and (iii) how common cancellation is. But it seems that ifdefaults prove to be so local as to arise out of words or evenmorphemes, then they are part of the computational power of grammarand they belong to grammar and lexicon rather than to semantics andpragmatics. Chierchia (2004) and Landman (2000) represent this view.Chierchia argues that since scalar implicatures do not arise indownward-entailing contexts (contexts that license inference from aset to its subset), there is a clear syntactic constraint on theirbehaviour (but see Chemlaet al. 2011). Jaszczolt (2012)calls such units that give rise to defaults or inferentialmodification ‘fluid characters’, employing Kaplan’s(1989) content-character distinction, to emphasise the fact that theunit of meaning that leads to inference or to a default meaning variesfrom context to context and from speaker to speaker: characters are‘fluid’ because they correspond to ‘flexibleinferential bases’ or ‘flexible default bases’. Butmuch more theorizing and substantial empirical support are needed toestablish the exact size of such local domains and the correspondingfluid characters.

As far as feature [8] is concerned, experimental work helps decidebetween [8a] and [8b], measuring the recovery time for the defaultmeaning as opposed to the non-default one. The development of theability to use scalar inferences has also been tested (Noveck 2001,2018; Papafragou & Musolino 2003; Musolino 2004; Noveck andSperber 2004; Geurts 2010; see also contributions to Cummins andKatsos 2019 ). It has been argued on the basis of some evidence thatdefault interpretations are not faster to produce and can be absentaltogether from processing in the case of five-year old subjects.Noveck (2004) provides the following evidence against Levinson’sautomatic and fast defaults. Children were presented with somedescriptions of situations in which the order of the events wasinverted in narration. They had to assess whether the description wastrue or false. The outcome was that the children who agreed with theinverted description reacted faster than the ones who disagreed. Itwas then concluded that enriching ‘and’ to ‘andthen’ is not automatic: it takes time. And, if pragmaticallyenriched responses take longer, then they cannot be the default ones(see Noveck 2004: 314). Similarly, with scalar terms, if one coulddemonstrate that the enriched readings, such as ‘some but notall’ for ‘some’, arise faster than ‘some butnot necessarily not all’, one would have strong evidence insupport of the defaults view.

The problem is that all these experiments assume Levinson’snotion of a fast and inference-free default while this is, as we haveseen, by no means the only understanding of default interpretations,and, arguably, not even the most commonly assumed one. Theexperimenters talk of arguments for and against ‘the DefaultView’, ‘the Default Model’ (see also Bezuidenhoutand Morris 2004, Breheny et al 2006), while, in fact there is no suchunique model to be falsified: there are very different understandingsof defaultness even in post-Gricean pragmatics alone, as is evidentfrom Section 1. The list of possible defining characteristics ofdefault interpretations in [1]–[8] shows that one cannot talkaboutthe default meaning. At the same time, it is muchharder to provide any experimental evidence for or against salientmeaning that draw on some contextual information, arise late inutterance processing, and are not normally cancellable. The latteralso seem much more intuitively plausible than Levinson’s rigiddefaults in that they are simply shortcuts through costly pragmaticinference and as such can be triggered by the situation itself ratherthan by the properties of a lexical item or construction at large.They are just normal, unmarked meanings for the context at hand and itis not improbable that such default, salient interpretations willprove to constitute just the polar end of a scale of degrees ofinference rather than have qualitatively different properties fromnon-default, clearly inference-based interpretations. They will occupythe area towards the ‘zero’ end of the scale of inferencebut will not trigger the dichotomy ‘default vs. inferentialinterpretation’. But since it is debatable whether salienceought to be equated with defaultness in he first place, ourterminological quandary comes back with full strength (see Section3).

It is also difficult to pinpoint the boundary between default andnon-default interpretations when we allow context and inference toplay a role in default meanings, that is when we allow [2b] and [8b].This does not mean, however, that we should pour them out with thebath water and resort to proposing nonce-inference in the case ofevery single utterance produced in discourse. When context-dependenceof defaults is allowed, then the main criterion for such meanings istheir subdoxastic arrival. When conscious inference is allowed, thenthe main criterion is the fact that only minimal contextual input isallowed, such as, say, the co-text in (24). In (24), the definitedescription ‘the first daughter’ has the attributiverather than referential reading.

(24)

The first daughter to be born to Mr and Mrs Brown will be called Scarlett.

On a traditional Gricean view of post-propositional, sentence-basedpragmatic inference, we have here the default attributive reading: theexpression ‘to be born’ and the future auxiliary‘will’ signal that no particular, extant, known individualis referred to. This is also the view followed in Default Semantics(Section 1.8) where both inference and defaults are ‘assumed tobe global’ – ‘assumed’ in the sense of amethodological assumption put in place until we have the means to testthe actual length of fluid characters and the content of thecorresponding default bases. In other words, information arrived atthrough WS merges with that from CPI, CD and SCWD whenall ofthe WS is ready. But in Default Semantics there is no default involvedin (24): we have WS merging with CPI to produce the attributivereading. On Levinson’s presumptive meanings account (Section1.3), it can be stipulated that (24) would fall in-between GCIs andPCIs: the only context that is required is the sentence itself, so theexample is not different from any other cases of GCIs. But thelocality of the GCI is the problem: depending on how we construct thelength of the triggering expression, we obtain a GCI or a PCI. When weconstrue it as ‘the first daughter’, the sub-part of thedefinite noun phrase, then we obtain the referential reading as thedefault, to be cancelled by ‘to be born’. In short, wedon’t know yet, at the current state of theorizing andexperimenting, which of the potential defining characteristics ofdefaults to employ. Neither are we ready to propose the demarcationline between default and non-default interpretations. We can conceiveof the first as shortcuts through inference but such a definition willnot suffice for delimiting a category. We can, however, concede thatdefault interpretations are governed by principles of rationalbehaviour in communication, be it Gricean maxims, neo-Griceanprinciples or heuristics, the logic of information structuring ofSDRT, or a version of defeasible logic as presented above.

All in all, it appears that the diversity of default interpretationspertains not only to their features listed in [1]–[8] but alsoto their provenance. This diversified use makes the term heavilytheory-dependent. Next, we can move to orthogonal, albeit not lessimportant, discussions of defaults vs. salience and defaults vs.literal meaning.

3. Defaults, Salience, and Literalness

In pragmatic theory, the term ‘default’ is often used inassociation with the term ‘salience’, so it is importantto clarify the similarities and differences between them. For Giora(e.g., 2003; Giora & Givoni 2015), salience and defaultness are twodifferent concepts. Salience depends merely on ‘accessibility inmemory due to such factors as frequency of use or experientialfamiliarity’ (Giora 2003: 33). For her, salience concernsmeanings, while defaultness concerns interpretations:

Defining defaultness in terms of an automatic response to a stimuluspredicts the superiority of default over non-default counterparts,irrespective of the degree of non-salience, figurativeness, contextstrength…. (Giora & Givoni 2015: 291)

So, salience is here a graded phenomenon (see here Giora 2003 for herexperimentally supported Graded Salience Hypothesis). Salience is alsoindependent from the literalness of the interpretation: the highlyaccessible meaning is not always the literal one. According to Giora(2003: 33), ‘literality is not a component of salience’.The latter is caused by experience, frequency of use, and as such isreflected in the accessibility in memory. Her experiments demonstratedthat ‘familiar instances of metaphor and irony activated theirsalient (figurative and literal) meanings initially, regardless ofcontextual information’ (p. 140).

Reconciling defaultness with salience may, however, be problematic inparticular cases, leading to the proposal of the degrees ofdefaultness (Giora & Givoni 2015), and to a rathercounterintuitive diluting of the concept of a default. For example,sarcasm can rely onnon-salient but default interpretation– non-salient when the interpretation is compositionally puttogether instead of being processed as a conventionalised unit.

On the other hand, for Jaszczolt (e.g., 2016a) defaultness reliesprecisely on salience that leads to automatic meaning retrieval. Shecalls this view Salience-Based Contextualism: the meaning of anutterance is derived through a variety of interacting processes, someof which rely on automatic interpretations such as cognitive defaultsor socio-cultural and world knowledge defaults identified in thetheory of Default Semantics and discussed in Section 1.8. Suchdefaults are defaults for the context and for the speaker, butsalience that predicts them is not. According to Salience-BasedContextualism, words and structures can trigger salient, automaticallyretrieved meanings. This is guaranteed by the fact that language is asocio-cultural as well as a cognitive phenomenon and as such is shapedby its common use in discourse on the one hand, and by the structureand operations of the brain on the other (see Jaszczolt 2016a: 50).Salience is situation-free (although not always co-text free –see above on fluid characters), defaultness is not: it is easy toimagine a speaker who does not make use of the available salientinterpretation because he or she lacks the necessary socioculturalbackground, knowledge of the laws of physics, or is guided by thecontext towards a different interpretation. Defaultness applies to aflexible unit on the basis of which an interpretation is formed (afluid character). Defaults in conversation result from emergentintentionality (Haugh 2008, 2011): they rely on the process ofco-construction of meaning discussed in Section 4. So, conversationaldefaults subsume salient meanings – literal or not, such asGiora’s salient meanings understood as conventional,prototypical, familiar, or frequent.

One of the important corollaries of this salience-based defaultness isthe possible redundancy of the literal/nonliteral distinction. Ifwords exhibit strong influences on each other in a string as well asinfluences from the situation of discourse (sometimes called‘lateral’ and ‘top-down influences’,respectively – see e.g., Recanati 2012), then we have littlereason for postulating literal meaning. For example, in (25), we havelittle reason for postulating ‘literal’ meanings of‘city’ or ‘asleep’: either of the words mayaccommodate to fit the other, and the appropriate interpretation willfollow.

(25)

The city is asleep. (from Recanati 2012: 185).

If ‘city’ means ‘inhabitants of the city’,‘asleep’ applies to it directly. If it means the placewith its infrastructure, ‘asleep’ has to adjust to meansomething to the effect of ‘quiet’, ‘dark’ or‘showing no movement’. But neither of the processes has aclear point of departure: there is no obvious, clearly definable movefrom literal to non-literal. Such co-occurrence comes with degrees ofsalience of certain meanings. Such options can also be viewed as‘probabilistic meanings’ that are ‘contextuallyaffirmed’, to use Allan’s (2011: 185) terminology, throughnonmonotonic inferences either in the co-text or by some other factorin the common ground.

However, probabilistic meanings presuppose ambiguity, while in generalthe salience- and default-oriented semantics and pragmatics relies onthe assumption of underspecification. While in the case of (26)–(27)an ambiguity account appears justified when viewed from theperspective of the inventory of the lexicon in a language system (inthat one would expect an analogy from ‘leopard’ and‘fox’ to ‘lamb’ and ‘goat’, whilein the context of the sentence this analogy is not present, as shownin (26a) and (27a)), most cases of lexical adjustment cannot be tracedback to properties of entries in the lexical inventory.

(26)

Jacqueline prefers leopard to fox.

(27)

Harry prefers lamb to goat.

(26a)

Jacqueline prefers leopard skin to fox fur.

(27a)

Harry prefers eating lamb to eating goat.

(from Allan 2011: 180). Probabilistic meaning, according to Allan,ought to be included in the lexicon: a lexeme ought to be listed inthe lexicon with different interpretations, annotated for theirprobability and circumstances in which this meaning is likely tooccur. He calls such probabilistic meanings ‘grades ofsalience’.

It is evident that this proposal brings us to the territory ofvector-based semantics in computational linguistics, while, on theother hand, accounts based on intention- and context-driven adjustmentsuch as Recanati’s and Jaszczolt’s pull in the directionof post-Gricean pragmatics. But as Sections 1.1–1.8 demonstrate, thetwo traditions are not necessarily incompatible. Lexical salience andradical contextualism about the lexicon point in the same direction asdistributional accounts in computational semantics: we haveprobabilities of certain meanings because these are meanings derivedfrom the ‘company a word keeps’, to adapt Firth’s(1957: 11) famous dictum. All in all, content words are stronglycontext-dependent – to the extent that perhaps indexicalityought to be viewed not as a defining feature of some lexical items butas a gradable feature of the entire lexicon. But what counts asindexicality is a separate theoretical question that cannot be pursuedhere.

Next, there is one more reason why salience has to be clearlydistinguished from defaultness. Let us consider demonstratives. Theobject referred to by using ‘that’ can be located with thehelp of (i) the recognition of the speaker’s intention, or (ii)the act of pointing, or even (iii) the presence of a particularprominent object in the visual field of the interlocutors. All thesecombine to delimit the concept of salience: such an object has to be(made) salient for the linguistic demonstration to succeed. Heresalient meaning is entirely, oralmost entirely (allowing forthe grammatical rendering of e.g., the proximal/distal distinction)determined by the given context and by the speaker’s knowledgethat itis the relevant context (see Lewis 1979 onscorekeeping; Cappelen and Dever 2016 for a discussion). What isimportant for us here is that salience can beproduced by theuse of a context-dependent term: objects are brought to salience bythe use of an indexical. Salience so understood is still compatiblewith the situation-independent concept of salience discussed above (todistinguish it from defaultness) in that it is the semantic meaning,the character (Kaplan 1989) of the demonstrative that triggers thebringing-to-salience process.

Now, on the one hand, linguistic research informs us that expressionswith thin semantic content such as anaphorically used demonstrativepronouns or personal pronouns are employed for referents whosecognitive status is high. In other words, they are used when theobject is in focus of attention or at least activated in memory(Gundel et al. 1993; see Jaszczolt 2002: 140–149 for adiscussion). On the other hand, when combined with an act ofdemonstration, the object can bemade salient. Cappelen andDever (2016) discuss here two types of successful referring bydemonstratives: pointing and intending (i.e. (i) and (ii) above),contrasted with prominence (i.e. (iii) above). The first createssalience, while the latter pertains to extant salience; the firstbrings entities into focus, while the latter exploits their in-focuscognitive status. What is of particular interest to semanticists (ofboth orientations discussed here, Gricean and computational) is thatthe first type allows for accommodation (Lewis 1979): objects are mademore salient when communication requires it.

To conclude, salience clearly differs from defaultness but forexpressions (words, phrases, sentences) for which delimiting defaultinterpretations makes sense, salience can provide an explanans.

4. Co-construction of Meaning: Towards Dynamic, Interactional Defaults

It is not a new observation that interlocutors can have collective, orjoint, intentions in communication. Joint construction of meaning hasbeen attended to in approaches as different as game theory (Lewis1979; Parikh 2010), action theory (Searle 1990) and conversationanalysis (Sacks, Schegloff and Jefferson 1974). But it is only morerecently that the joint construction (or co-construction) of meaninghas begun to come to the forefront of pragmaticstout court.For example, Arundale’s so-called Conjoint Co-constituting Modelof Communication focuses on the interactive construction ofimplicatures (Arundale 1999, 2010; Haugh 2007, 2008; Haugh andJaszczolt 2012). Next, Elder and Haugh (2018) propose a model ofcomplex inferential work, continuing for several conversational turns,that can account for mismatches between the expected and the actualinferences, as well as for the fact that speakers may not have preciseintentions while issuing an utterance and be open to negotiatingmeaning. These aspects of communication had often been neglected, oreven denied, in post-Gricean research.

It is no surprise that understanding communication as emergentmeanings, as negotiating intentions and commitments (see e.g., Geurts2019; Elder 2021) is taking the concept of default to the newterritory of dynamic, co-constructed meanings. Elder and Haugh, forexample, aim at modelling the joint construction of the meaning thatis settled on by the interlocutors as the main content – as theysay, akin to the ‘primary meaning’ of Default Semantics(Jaszczolt 2005) or to Ariel’s (2002) ‘privilegedinteractional interpretation’ – free from any formalconstraints that the logical form of the uttered sentence mightimpose, but instead, and in addition, flowing freely, as conversationprogresses, with dynamic, changing intentions. Such co-constructedmeaning is steered by default interpretations that are assignedprivately by each side but also sometimes rejected in favour of whatprogressively emerges as a default inference – an interactivelyconstructed meaning that is sensitive to what is fit to be taken forgranted, and as such sensitive to what other parties would, and do,take for granted. In short, dynamic, interactional pragmatics comeswith dynamic, interactional defaults.

5. Defaults and Accountability

Interactional pragmatics discussed in the previous section highlightsanother important aspect of communication that was largely neglectedin early pragmatic theories, namely speaker accountability. Haugh(2013: 53) presents it as follows:

First, what a speaker is held accountable for goes beyond the veracityof information to include other moral concerns, such as social rights,obligations, responsibilities and the like. Second, to be heldinteractionally accountable differs from inferringcommitment. The former is tied to an understanding of speaker meaningas arising through incremental, sequentially grounded discourseprocessing …, while the latter is tied to a punctuated view ofspeaker meaning that arises at the level of utterance processing.

As he says, it is important to view speaker meaning as adeontological concept – that is, view it from theperspective of moral philosophy and norms of what actions arepermitted or not permitted in conversation. Viewing speaker meaning asdeonotological puts emphasis on practical applications of speech, onsuch moral norms and on the consequences of conversational behaviour.And this is where defaultness comes to the fore, propelled by thequestion as to what kind of content speakers ought to be heldaccountable for. Should it be the kind of meaning that speakers feelcommitted to, or the kind of meaning that they, sometimesinadvertently, communicated? Or, perhaps, merely the minimal semanticcontent? Emma Borg (2019), for example, distinguishes between what shecalls (i) strict liability, that is liability for the minimalproposition pertaining to one’s utterance, delimited accordingto the principles of her Minimal Semantics (Borg 2004, 2012), and (ii)conversational liability (a gradable concept), applied to potentialinterpretations of an utterance in the context of conversation. Butsuch philosophical solutions don’t go far enough in that theymerely offer a more fine-grained classification rather thanimplementable answers. Next, Elder and Haugh (2018) focus on‘the most salient propositional meaning that isostensively made operative between interlocutors’ (p. 595),which allows them to offer a proposal with more direct utility fordiscussions of moral responsibility. The concept of dynamic, operativemeaning allows them to argue that it is the addressee’s responsethat makes the speaker accountable for the emergent meaning –the meaning that the interlocutors settle for in the next turn whenthe original speaker becomes aware of their ‘reflexiveaccountability’. But, again, feasibility of implementation hasto wait for a model with some predictive power.

All in all, the concepts of commitment, liability (including legalliability) and accountablity are beginning to get to the forefront ofphilosophy of language, pragmatics, and contextualist semantics (andas such, necessarily, also metasemantics and metapragmatics, seeJaszczolt 2022), and take the concept of defaultness with them to thenew dynamic, interactive dimension. The offshoots of those debates areample. For example, they open up new perspectives on such questions aswhat counts as an assertion, insult, slur, or a lie: what kind ofcontent ought we to include, and, on the other hand, what norms oughtwe to adopt? Here are a few snapshots.

Sanford Goldberg asks what constitutes the speech act of assertion,and what epistemic norm assertion should satisfy. He says thatassertions are “speech acts in which a given proposition ispresented as true, where this presentation has certainforce (what we might callassertoric force)”(Goldberg 2015: 5) and that “[i]t is mutually manifest toparticipants in a speech exchange that assertion has a robustlyepistemic norm; that is, that one must: assert that p, only if E(one,[p])” (p.96), where ‘E’ stands for a description ofa mutually manifest epistemic standard. But this opens up the questionas to what counts as asserted content, for linguistic as well associal, legal, and ethical purposes. Is it merely minimal content?But, perhaps, there is no such thing as minimal content in that hereare no core meanings, no context-free concepts (cf. Rayo 2013)? Or,perhaps, one ought to opt for Levinson-style presumptive meanings(Section 1.3), or context-driven defaults (Sections 1.7–1.8), or eveninteractive meanings that emerge thanks to salience and defaultness ofcertain interpretations (Section 4)? Scope for theorizing andexperimentation is still ample.

Relatedly, default meaning is an important concept for the discussionof deception, such as lying and misleading. Insincerity has recentlyengendered heated debates in linguistic pragmatics in that it is amoot point what exactly counts as lying. For example, is lying sayingsomething that the speaker believes to be false? Or is it sayingsomething with an intention to deceive but the content of which neednot necessarily be false (see e.g., Stokke 2018)? And, what exactlycounts as ‘saying’ for the purpose of lying (see e.g Saul2012)? Further, can one lie through implicatures and presuppositions(Meibauer 2014)? And perhaps we ought to venure even further. AsHeffer (2020: 6) says, “…we need to extend thescope of untruthfulness both from utterance insincerity(lying and misleading) to discursive insincerity (withholding), andfrom intentional insincerity to epistemic irresponsibility” (seealso Carson 2010). Epistemic sincerity is the default in ordinary,‘non-strategic’ contexts, so, lying, misleading, andbullshitting (that is, speaking without regard for truth or falsehood)can easily go through. But here the question as to what exact meaningthe speaker is accountable for is intimately related to the questionof conversational salience and defaults.

Now, it appears that for the debates at a crossroads of theory ofmeaning and ethics, the crucial understanding of defaultness is thatof salient emergent meaning in conversation. As we have seen here,such defaults can be abused – a topic that is of interest topragmatics as well as to theories of argumentation. Let us considerone more snapshot, this time an example of an argumentative strategyof decontextualisation, where instead of inferring the defaultinterpretation (in the sense of the automatic, most salientinterpretation for the context), the addressee reacts to thedecontextualized semantic content of the speaker’s utterance.This is traditionally called ‘the fallacy of ignoringqualifications’ (secundum quid orsecundum quid etsimpliciter, loosely translatable as ‘what is true in a wayand what is true absolutely’. It is a form of deceit that, asMacagno (2022) argues, is best explicated in a broader pragmaticcontext of ignoring not only what is explicit but also of what isimplicit:

According to this view, thesecundum quid does not resultfrom the ‘suppression’ or ‘ignoring’ ofexplicit qualifications or evidence (…); rather, it iscommitted when an implicit qualification is reconstructed in a waythat was not intended and could not be presumed. (Macagno 2022: 6)

As he says, post-Gricean pragmatic theories are indispensable here forexplaining how this form of manipulation works, in that they caninvoke principles and heuristics of how the intended meaning isrecovered by the addressee. He focuses on contextualist accounts thatcapture the proposition expressed, or Recanati’s what is saidand Sperber and Wilson’s explicature.Secundum quid isexplained as a strategy of manipulation that makes use of thepragmatic process of enrichment, but relyingnot on theassumptions shared by the interlocutors but instead substitutingunwarranted, strategically chosen ones:

The qualified or absolute interpretations are fallacious because theyare different from the ‘default’, ‘salient’,or more generally ‘plain’ meaning. (Macagno 2022: 7)

The importance of such defaults for argumentation theory requires nofurther defence.

6. Concluding Remarks and Future Prospects

This comparison of various selected approaches to defaultinterpretations in semantics and pragmatics allows for somegeneralizations. Firstly, it is evident from the surveyed literaturethat, contrary to the assumptions of some experimental pragmaticists,there is no one, unique ‘default model’ of utteranceinterpretation. Instead, default (and salient) meanings are recognisedin many approaches to utterance interpretation but they are defined bydifferent sets of characteristic features. Next, in the present stateof theorizing, data-based analyses and experimenting, while therationale for default interpretations is strong, some of theproperties of such interpretations are still in need of furtherinvestigation. For example, the discussions of locality of defaultsand their subdoxastic arrival are in need of empirical support beforethey can be taken any further. In other words, a ‘fluidcharacter’ is in need of empirical identification – andeven more so when we add the dynamic perspective of interactivelyachieved, co-constructed meanings.

Moreover, the principle and method for delimiting a defaultinterpretation as distinguished from an inferential interpretation isstill a task for the future. The existence of a shortcut throughcostly inference is an appealing and hardly controversial thesis butthe exact properties of such meanings are still subject todisputes.

Next, the automatic arrival at context-dependent meanings has to bediscussed as part of the debate between the direct access view and themodular view of language processing. Direct access predicts thatcontext is responsible for activating relevant senses to the extentthat the salience of the particular sense of a lexical item does notplay a part. According to the modular view, lexical meanings that arenot appropriate for the context are also activated, only to besuppressed at a further stage of processing. With the rise of theoriesthat sit between these polar views, the question of the compatibilityof the salience in the lexicon and the default status of utteranceinterpretations requires more attention. What can be attributed to thelexicon and what to the context of utterance remains an unresolvedquestion.

Finally, whether we approach defaults through distributionalcomputational semantics, theoretical truth-conditional semantics,post-Gricean pragmatics, or some version of a combined view, progressin research on defaults in human reasoning will necessarily requireprogress in technology. No matter how powerful our theories are, theywill have to be tested either on large corpora, or throughneuroimaging: more traditional methods of psycholinguistic experimentsor small databases will always leave a wide margin of doubt as to‘is it really how human reasoning works’? If there are biggeneralizations to be made regarding how we jump to conclusions, thesewill have to be modern equivalents of the 19th centuryphenomenological ideas, founded on intentionality of mental states,informativeness of acts of communication, all predicted by assumedefficiency (but not necessarily cooperation), but with access tomodern methods of empirical corroboration (or falsification).

In this context, can we safely assume that the concept of defaultmeaning will sail with the most successful approaches to humancommunication, or is it in need of precisification now, when it is sodiversely used (and sometimes misused)? Conceptual engineering isoften a risky business. But going along with the common-sense use isnot so: meanings that ‘spring to mind’ are defaultmeanings – it is just that for some purposes we want to talkabout meanings and associations that spring to mindin the contextof a particular discourse, and for others,in the languagesystem alone. This degree of flexibility, further justified bydiverse objectives in employing the concept – as diverse as,say, making an interlocutor accountable for a strongly conveyed but’unsaid’ insult on the one hand, and devising algorithmsfor training machines to use language to communicate on the other, isprobably not so difficult for theorists of meaning to live with.

Bibliography

• Allan, K., 2011, “Graded Salience: Probabilistic Meaning inthe Lexicon”, inSalience and Defaults in UtteranceProcessing, K. M. Jaszczolt & K. Allan (eds.), Berlin: DeGruyter Mouton, 165–187.
• Ariel, M., 2002, “Privileged InteractionalInterpretations”,Journal of Pragmatics, 34:1003–1044.
• –––, 2016, “Revisiting the Typology ofPragmatic Interpretations”,Intercultural Pragmatics,13: 1–35.
• Arundale, R. B., 1999, “An Alternative Model and Ideology ofCommunication for an Alternative to Politeness Theory”,Pragmatics, 9: 119–153.
• –––, 2010, “Constituting Face inConversation: Face, Facework, and Interactional Achievement”,Journal of Pragmatics, 42: 2078–2105.
• –––, 2013, “Conceptualizing‘Interaction’ in Interpersonal Pragmatics: Implicationsfor Understanding and Research”,Journal of Pragmatics,58: 12–26.
• Asher, N. & A. Lascarides, 1995, “Lexical Disambiguationin a Discourse Context”,Journal of Semantics, 12:69–108.
• –––, 2003,Logics of Conversation,Cambridge: Cambridge University Press.
• Bach, K., 1984, “Default Reasoning: Jumping to Conclusionsand Knowing When to Think Twice”,Pacific PhilosophicalQuarterly, 65: 37–58.
• –––, 1987,Thought and Reference,Oxford: Clarendon Press.
• –––, 1994, “Semantic Slack: What Is Saidand More”, inFoundations of Speech Act Theory:Philosophical and Linguistic Perspectives, S. L. Tsohatzidis(ed.), London: Routledge, 267–291.
• –––, 1995, “Remark and Reply.Standardization vs. Conventionalization”,Linguistics andPhilosophy, 18: 677–686.
• –––, 1998, “Postscript (1995):Standardization Revisited”, inPragmatics: CriticalConcepts (Volume 4), A. Kasher (ed.), London: Routledge,712–722.
• –––, 2007, “Regressions in Pragmatics (andSemantics)”, inPragmatics, N. Burton-Roberts (ed.),Basingstoke: Palgrave Macmillan, 24–44.
• Bezuidenhout, A. L. & R. K. Morris, 2004, “Implicature,Relevance and Default Pragmatic Inference”, inExperimentalPragmatics, I. A. Noveck & D. Sperber (eds.), Basingstoke:Palgrave Macmillan, 257–282.
• Blutner, R., 2000, “Some Aspects of Optimality in NaturalLanguage Interpretation”,Journal of Semantics, 17:189–216.
• Blutner, R., & H. Zeevat, 2004, “Editors’Introduction: Pragmatics in Optimality Theory”, inOptimality Theory and Pragmatics, R. Blutner & H. Zeevat(eds.), Basingstoke: Palgrave Macmillan, 1–24.
• Boguraev, B. & J. Pustejovsky, 1990, “Lexical Ambiguityand the Role of Knowledge Representation in Lexicon Design”,Proceedings of the 13th International Conference on ComputationalLinguistics COLING ’90, Helsinki, 36–41.
• Borg, E., 2004, Minimal Semantics, Oxford: ClarendonPress.
• –––, 2012, Pursuing Meaning, Oxford:Oxford University Press.
• –––, 2019, “Explanatory Roles for MinimalContent”,Noûs, 53: 513–539.
• Breheny, R., N. Katsos & J. Williams, 2006, “AreGeneralised Scalar Implicatures Generated by Default? An On-lineInvestigation into the Role of Context in Generating PragmaticInferences”,Cognition, 100: 434–463.
• Bultinck, B., 2005,Numerous Meanings: The Meaning of EnglishCardinals and the Legacy of Paul Grice, Amsterdam: Elsevier.
• Cappelen, H. & J. Dever, 2016,Context andCommunication, Oxford: Oxford University Press.
• Carson, T., 2010,Lying and Deception: Theory andPractice, Oxford: Oxford University Press.
• Carston, R., 1988, “Implicature, Explicature, andTruth-Theoretic Semantics”, inMental Representations: TheInterface Between Language and Reality, R. M. Kempson (ed.),Cambridge: Cambridge University Press, 155–181.
• –––, 2002,Thoughts and Utterances: ThePragmatics of Explicit Communication, Oxford: Blackwell.
• Chemla, E., V. Homer & D. Rothschild, 2011, “Modularityand Intuitions in Formal Semantics: The Case of Polarity Items”,Linguistics and Philosophy, 34: 537–570.
• Chierchia, G., 2004, “Scalar Implicatures, PolarityPhenomena, and the Syntax/Pragmatics Interface”, inStructures and Beyond: The Cartography of SyntacticStructures (Volume 3), A. Belletti (ed.), Oxford: OxfordUniversity Press, 39–103.
• Coecke, B., M. Sadrzadeh & S. Clarke, 2010,“Mathematical Foundations for a Compositional DistributionalModel of Meaning”,Linguistic Analysis, 36:345–384.
• Cummins, C. & N. Katsos (eds.), 2019,The Oxford Handbookof Experimental Semantics and Pragmatics, Oxford: OxfordUniversity Press.
• Elder, C.-H., 2021, “Speaker Meaning, Commitment andAccountability” inThe Cambridge Handbook ofSociopragmatics, M. Haugh, D. Z. Kádár & M.Terkourafi (eds.), Cambridge: Cambridge University Press,48–68.
• Elder, C.-H. & M. Haugh, 2018, “The InteractionalAchievement of Speaker Meaning: Toward a Formal Account ofConversational Inference”,Intercultural Pragmatics,15: 593–625.
• Elder, C.-H. & K. M. Jaszczolt, 2016, “Towards aPragmatic Category of Conditionals”,Journal ofPragmatics, 98: 36–53.
• Firth, J. R., 1957.Papers in Linguistics 1934–53,Oxford: Oxford University Press.
• Frege, G., 1893,Grundgesetze der Arithmetik (Volume 1),references to the reprint inThe Frege Reader, M. Beaney(ed.), 1997, Oxford: Blackwell, 84–191..
• Gazdar, G., E. Klein, G. Pullum, & I. Sag, 1985,Generalized Phrase Structure Grammar, Oxford: B.Blackwell.
• Geurts, B., 2007, “Really Fucking Brilliant”,Theoretical Linguistics, 33: 209–214.
• –––, 2009, “Scalar Implicature and LocalPragmatics”,Mind and Language, 24: 51–79.
• –––, 2010,Quantity Implicatures,Cambridge: Cambridge University Press.
• –––, 2019, “Communication as CommitmentSharing: Speech Acts, Implicatures, Common Ground”,Theoretical Linguistics, 45: 1–30.
• Giora, R. 2003.On Our Mind: Salience, Context, and FigurativeLanguage, Oxford: Oxford University Press.
• ––– & S. Givoni, 2015, “DefaultnessReigns: The Case of Sarcasm”,Metaphor and Symbol, 30:290–313.
• Goldberg, S. C., 2011, “Putting the Norm of Assertion toWork: The Case of Testimony”, inAssertion: NewPhilosophical Essays, J. Brown & H. Cappelen (eds.), Oxford:Oxford University Press, 175–195.
• –––, 2015.Assertion: On the PhilosophicalSignificance of Assertoric Speech, Oxford: Oxford UniversityPress.
• Grice, H. P., 1975, “Logic and Conversation”, inSyntax and Semantics (Volume 3), P. Cole & J. L. Morgan(eds.), New York: Academic Press; references to the reprint in H. P.Grice, 1989,Studies in the Way of Words, Cambridge, Mass.:Harvard University Press, 22–40.
• Gundel, J. K., N. Hedberg & R. Zacharski, 1993,“Cognitive Status and the Form of Referring Expressions inDiscourse”,Language, 69: 274–307.
• Harris, Z., 1954, “Distributional Structure”,Word, 10: 146–162.
• Haugh, M., 2007. “The Co-constitution of PolitenessImplicature in Conversation”,Journal of Pragmatics,39: 84–110.
• –––, 2008, “The Place of Intention in theInteractional Achievement of Implicature”, inIntention,Common Ground and the Egocentric Speaker-Hearer, I. Kecskes &J. Mey (eds.), Berlin: Mouton de Gruyter, 45–85.
• –––, 2011, “Practices and Defaults inInterpreting Disjunction”, inSalience and Defaults inUtterance Processing, K. M. Jaszczolt & K. Allan (eds.),Berlin: De Gruyter Mouton, 189–225.
• –––, 2013, “Speaker Meaning andAccountability in Interaction”,Journal of Pragmatics48: 41–56.
• Haugh, M., & K. M. Jaszczolt, 2012, “Speaker Intentionsand Intentionality” inThe Cambridge Handbook ofPragmatics, K. Allan & K. M. Jaszczolt (eds.), Cambridge:Cambridge University Press, 87–112.
• Heffer, C., 2020,All Bullshit and Lies? Insincerity,Irresponsibility, and the Judgment of Untruthfulness, Oxford:Oxford University Press.
• Horn, L. R., 1984, “Toward a New Taxonomy for PragmaticInference: Q-based and R-based Implicature”, inGeorgetownUniversity Round Table on Languages and Linguistics 1984, D.Schiffrin (ed.), Washington, D.C.: Georgetown University Press,11–42.
• –––, 1992, “The Said and theUnsaid”,Ohio State University Working Papers inLinguistics, 40 (SALT II Proceedings): 163–192.
• –––, 1988, “Pragmatic Theory”, inLinguistics: The Cambridge Survey, Vol. 1. Linguistic Theory:Foundations, F. J. Newmeyer (ed.), Cambridge: CambridgeUniversity Press, 113–145.
• –––, 2004, “Implicature”, inTheHandbook of Pragmatics, L. R. Horn & G. Ward (eds.), Oxford:Blackwell, 3–28.
• –––, 2006, “The Border Wars: A Neo-GriceanPerspective”, inWhere Semantics Meets Pragmatics, K.von Heusinger and K. Turner (eds.), Oxford: Elsevier,21–48.
• –––, 2012, “Implying and Inferring”,inThe Cambridge Handbook of Pragmatics, K. Allan & K. M.Jaszczolt (eds.), Cambridge: Cambridge University Press,69–86.
• Jaszczolt, K. M., 1999,Discourse, Beliefs, and Intentions:Semantic Defaults and Propositional Attitude Ascription, Oxford:Elsevier Science.
• –––, 2002,Semantics and Pragmatics: Meaningin Language and Discourse, London: Longman.
• –––, 2005,Default Semantics: Foundations ofa Compositional Theory of Acts of Communication, Oxford: OxfordUniversity Press.
• –––, 2008, “Psychological Explanations inGricean Pragmatics and Frege’s Legacy”, inIntention,Common Ground and the Egocentric Speaker-Hearer, I. Kecskes &J. Mey, (eds.), Berlin: Mouton de Gruyter, 9–44.
• –––, 2009a, “Cancellability and thePrimary/Secondary Meaning Distinction”,InterculturalPragmatics, 6: 259–289.
• –––, 2009b,Representing Time: An Essay onTemporality as Modality. Oxford: Oxford University Press.
• –––, 2010, “Default Semantics”, inThe Oxford Handbook of Linguistic Analysis, B. Heine & H.Narrog (eds.), Oxford: Oxford University Press, 193–221.
• –––, 2012, “‘Pragmaticising’Kaplan: Flexible inferential bases and fluid characters”,Australian Journal of Linguistics, 32: 209–237.
• –––, 2016a,Meaning in LinguisticInteraction: Semantics, Metasemantics, Philosophy of Language,Oxford: Oxford University Press.
• –––, 2016b, “On Unimaginative Imaginationand Conventional Conventions: Response to Lepore and Stone”,Polish Journal of Philosophy, 10: 89–98.
• –––, 2021, “Default Semantics”, inOxford Bibliographies in Linguistics, M. Aronoff (ed.), NewYork: Oxford University Press (online).
• –––, 2022, “Metasemantics andMetapragmatics: Philosophical Foundations of Meaning”, inThe Cambridge Handbook of the Philosophy of Language, P.Stalmaszczyk (ed.), Cambridge: Cambridge University Press,139–156.
• Kamp, H. & U. Reyle, 1993,From Discourse to Logic:Introduction to Modeltheoretic Semantics of Natural Language, FormalLogic and Discourse Representation Theory, Dordrecht:Kluwer.
• Kaplan, D., 1989, “Demonstratives: An Essay on theSemantics, Logic, Metaphysics, and Epistemology of Demonstratives andOther Indexicals”, inThemes from Kaplan, J. Almog, J.Perry, & H. Wettstein (eds.), New York: Oxford University Press,481–563.
• Koenig, J.-P., 1993, “Scalar Predicates and Negation:Punctual Semantics and Interval Interpretations”,ChicagoLinguistic Society, 27(2): The Parasession on Negation,140–155.
• Landman, F., 2000,Events and Plurality, Dordrecht:Kluwer.
• Lascarides, A., T. Briscoe, N. Asher & A. Copestake, 1996,“Order Independent and Persistent Type DefaultUnification”,Linguistics & Philosophy, 19:1–89.
• Lascarides, A., & A. Copestake, 1998, “Pragmatics andWord Meaning”,Journal of Linguistics, 34:387–414.
• Lepore, E. & M. Stone, 2016,Imagination and Convention:Distinguishing Grammar and Inference in Language, Oxford: OxfordUniversity Press.
• Levinson, S. C., 1995, “Three Levels of Meaning”, inGrammar and Meaning. Essays in Honour of Sir John Lyons, F.R. Palmer (ed.), Cambridge: Cambridge University Press,90–115.
• –––, 2000,Presumptive Meanings: The Theoryof Generalized Conversational Implicature, Cambridge, Mass.: MITPress.
• Lewis, D., 1979, “Scorekeeping in a Language Game”,Journal of Philosophical Logic, 8: 339–359.
• Liang, P. & C. Potts, 2015, “Bringing Machine Learningand Compositional Semantics together”,The Annual Review ofLinguistics, 1: 355–376.
• Macagno, F., 2022, “Ignoring Qualifications as a PragmaticFallacy: Enrichments and Their Use for ManipulatingCommitments”,Languages, 7(13).doi:10.3390/languages7010013
• Meibauer, J., 2014,Lying at the Semantics-PragmaticsInterface, Berlin: Mouton de Gruyter.
• Musolino, J., 2004, “The Semantics and Acquisition of NumberWords: Integrating Linguistic and Developmental Perspectives”,Cognition, 93: 1–41.
• Noveck, I. A., 2001, “When Children are More Logical thanAdults: Experimental Investigations of Scalar Implicature”,Cognition, 78: 165–188.
• –––, 2004, “Pragmatic Inferences Relatedto Logical Terms”, inExperimental Pragmatics, I. A.Noveck & D. Sperber (eds.), Basingstoke: Palgrave Macmillan,301–321.
• –––, 2018,Experimental Pragmatics,Cambridge: Cambridge University Press.
• Noveck, I. A., & D. Sperber (eds.), 2004,ExperimentalPragmatics, Basingstoke: Palgrave Macmillan.
• Papafragou, A. & J. Musolino, 2003, “ScalarImplicatures: Experiments at the Semantics-PragmaticsInterface”,Cognition, 86: 253–282.
• Parikh, P., 2010,Language and Equilibrium, Cambridge,MA: MIT Press.
• Pelletier, F. J. & R. Elio, 2005, “The Case ofPsychologism in Default and Inheritance Reasoning”,Synthese, 146: 7–35.
• Potts, C., 2005,The Logic of Conventional Implicatures,Oxford: Oxford University Press.
• –––, 2007, “The ExpressiveDimension”,Theoretical Linguistics, 33:165–198.
• –––, 2012, “Conventional Implicature andExpressive Content”, inSemantics: An International Handbookof Natural Language Meaning, vol. 2, C. Maienborn, K. vonHeusinger and P. Portner (eds.), Berlin: De Gruyter Mouton,2516–2535.
• –––, 2015, “Presupposition andImplicature”, inThe Handbook of Contemporary SemanticTheory, S. Lappin & C. Fox (eds.), Oxford: Wiley-Blackwell,168–202.
• Rayo, A., 2013, “A Plea for Semantic Localism”,Noûs, 47: 647–679.
• Recanati, F., 2002, “Does Linguistic Communication Rest onInference?”,Mind and Language, 17: 105–126.
• –––, 2004,Literal Meaning, Cambridge:Cambridge University Press.
• –––, 2010,Truth-ConditionalPragmatics, Oxford: Clarenson Press.
• –––, 2012, “Compositionality, Flexibility,and Context Dependence”, inThe Oxford Handbook ofCompositionality, M. Werning, W. Hinzen & E. Machery (eds.),Oxford: Oxford University Press, 174–191.
• Reiter, R., 1980, “A Logic for Default Reasoning”,Artificial Intelligence, 13: 81–132.
• Richard, M., 2008,When Truth Gives Out, Oxford: OxfordUniversity Press.
• Roberts, C., 2004, “Context in DynamicInterpretation”, inThe Handbook of Pragmatics, L. Horn& G. Ward (eds.), Oxford: Blackwell, 197–220.
• Sacks, H., E. A. Schegloff, & G. Jefferson, 1974, “ASimplest Systematics for the Organization of Turn-Taking forConversation”,Language, 50: 696–735.
• van der Sandt, R. A., 1992, “Presupposition Projection asAnaphora Resolution”,Journal of Semantics, 9:333–377.
• –––, 2012, “Presupposition andAccommodation in Discourse”, inThe Cambridge Handbook ofPragmatics, K. Allan & K. M. Jaszczolt (eds.), Cambridge:Cambridge University Press, 329–350.
• Saul, J. M., 2002, “What Is Said and Psychological Reality;Grice’s Project and Relevance Theorists’Criticisms”,Linguistics and Philosophy 25,347–372.
• –––, 2012,Lying, Misleading, and What IsSaid: An Exploration in Philosophy of Language and Ethics,Oxford: Oxford University Press.
• Sperber, D. & D. Wilson, 1986,Relevance: Communicationand Cognition, Oxford: Blackwell; reprinted in 1995, secondedition.
• Stokke, A., 2018,Lying and Insincerity, Oxford: OxfordUniversity Press.
• Stone, M., 2016, “Semantics and Computation”, inThe Cambridge Handbook of Formal Semantics, M. Aloni & P.Dekker (eds.), Cambridge: Cambridge University Press, 775–800.
• Thomason, R. H., 1997, “Nonmonotonicity inLinguistics”, inHandbook of Logic and Language, J. vanBenthem & A. ter Meulen (eds.), Oxford: Elsevier Science,777–831.
• Tonhauser, J., D. Beaver, C. Roberts, & M. Simons, 2013,“Towards a Taxonomy of Projective Content”,Language, 89(1): 66–109.
• Veltman, F., 1996, “Defaults in Update Semantics”,Journal of Philosophical Logic, 25: 221–261.
• Zeevat, H., 2000, “Demonstratives in Discourse”,Journal of Semantics, 16: 279–313.
• –––, 2004, “Particles: PresuppositionTriggers, Context Markers or Speech Act Markers”, inOptimality Theory and Pragmatics, R. Blutner & H. Zeevat(eds.), Basingstoke: Palgrave Macmillan, 91–111.

Academic Tools

	How to cite this entry.
	Preview the PDF version of this entry at theFriends of the SEP Society.
	Look up topics and thinkers related to this entry at the Internet Philosophy Ontology Project (InPhO).
	Enhanced bibliography for this entryatPhilPapers, with links to its database.

Other Internet Resources

• Jurafsky, D. & J. H. Martin, 2017,Speech and Language Processing, 3rd draft (in progress); 2nd edition, London: Pearson Education,2009.

Related Entries

discourse representation theory |Grice, Paul |implicature |linguistics: computational |logic: non-monotonic |logical form |meaning, theories of |pragmatics |reasoning: defeasible |speech acts

Acknowledgments

This entry draws on some sections of my ‘DefaultInterpretations’, published inHandbook of PragmaticsOnline, vol. 10, 2006, ed. by J.-O. Ostman and J. Verschueren. Iowe thanks to John Benjamins Publishing Co., Amsterdam, for permissionto use the material.