Linguistic Context and Contextualism

The Philosophical Problem of Context

A speaker utters “It is raining.” The proposition assessed by an ordinary hearer concerns precipitation at some place and time. Neither the isolated orthographic sentence nor an abstract inventory of its words identifies every element needed for the assessment. Yet communication ordinarily succeeds: hearers locate the relevant weather situation, distinguish the report from a theatrical line, and infer the speaker's practical point. The philosophical issue is how the content supplied by linguistic convention relates to the information recruited from the circumstances of use. Perry's discussion of weather reports makes the unexpressed location an especially influential test case (Perry, 1986).

This problem is more exacting than the observation that speakers rely on background knowledge. A semantic theory aims to specify what linguistic expressions contribute to the content of a sentence, how contributions compose, and under what conditions the resulting content is true. A theory of language use explains why a particular speaker chooses an expression, what the audience recognizes, and which further commitments or suggestions become communicatively available. Contextualism contests the distribution of explanatory work across these theories. Strong versions hold that truth-conditional contents generally depend on pragmatic interpretation; moderate versions identify specific forms of incompleteness; minimalists restrict context's semantic contribution while admitting extensive contextual effects on what speakers communicate (Recanati, 2004; Borg, 2004; Cappelen & Lepore, 2005).

The issue therefore involves three distinct questions. First, which expressions have context-sensitive meanings by their linguistic character? Second, can complete truth-conditional propositions require contextual contributions for which no overt expression or syntactic position supplies a value? Third, how do we individuate what an utterance asserts when communicative practice supports multiple accurate reports of a single episode? These questions intersect with metaphysics, because propositions are candidates for truth and identity; with epistemology, because the wording of knowledge attributions may shift with conversational demands; and with social philosophy, because shared assumptions and interpretive authority determine what can be publicly communicated.

A useful investigation begins with the ordinary distinction between sentence type, utterance token, propositional content, and communicative act. A sentence type can recur unchanged across occasions. An utterance token is an event of using it. A proposition specifies a way the world is presented as being. An assertion is an act that places a proposition before an audience with characteristic commitments. Even when one utterance carries one asserted proposition, its implications and practical significance can exceed that content. Treating all four as a single object is a principal source of confusion in debates about context.

Definition and Scope of Linguistic Context

Linguistic context denotes those features of an occurrence of language that are relevant to its interpretation according to a specified linguistic, pragmatic, or discourse-theoretic account. Some are immediate: speaker, addressee, utterance time, location, and demonstratively indicated objects. Some are sequential: the preceding utterances, salient alternatives, conversational questions, and anaphoric antecedents. Others are institutional or normative: the standing of participants, conventions governing the interaction, accepted evidence, and purposes that determine which contrasts matter. It is more precise to speak of context as a structured explanatory variable than as an undifferentiated surrounding environment.

A co-text is the neighboring linguistic material that helps determine an expression's interpretation. In “Maria entered. She smiled,” the first sentence supplies a candidate antecedent for “she.” A physical setting can anchor “here,” whereas a socially organized conversation can determine the comparison class of “tall.” A semantic index is a formally defined tuple of values relative to which an expression is evaluated. A common ground is information treated, for conversational purposes, as jointly accepted or presupposed. These are different models of context. They may interact, but none can substitute automatically for the others (Kaplan, 1989; Lewis, 1980; Stalnaker, 1978).

A contextual parameter must be selected by some principle. The relevant value for “I” is normally supplied by a conventional speaker rule. The relevant comparison class for “tall” may be recovered from the conversational question, a salient domain, or an implicit standard. The location relevant to “It is raining” can be the location under discussion rather than the position of the speaker. A theory that merely says “context determines the value” has named the problem without solving it. It needs to explain which contextual facts matter, why they matter, and how disagreements over their selection should be resolved.

Contextualism, in the principal linguistic sense of this dossier, is the family of views on which the content or truth conditions of an utterance are significantly sensitive to contextual determination beyond the stable lexical and compositional meanings of its words. This definition covers degrees. Moderate contextualists can accept fixed semantic contributions and isolate specific pragmatic operations; radical contextualists argue that occasion-specific understanding extends broadly through ordinary predication. The broader term can also describe semantic contextualism about particular expressions, such as knowledge ascriptions, without implying a globally contextualist theory of language. Each claim must be assigned its appropriate scope.

Historical Formation: Sense, Use, and Occasion

Frege's “Über Sinn und Bedeutung” (1892) offers a foundational distinction between an expression's mode of presentation and what it designates. The distinction arose in the analysis of identity, informative judgment, and propositional composition. Frege's discussion of indexical and temporal components also shows that the same words can require features of an utterance's occurrence in order to communicate a complete thought. This does not establish contemporary radical contextualism; rather, it supplies a formal problem: how can recurring linguistic items express determinate thoughts when their contribution is sensitive to circumstances? The original German terms Sinn and Bedeutung require care, since Bedeutung is translated in different editions as “reference” or “meaning.” The discussion here uses the standard analytic distinction sense/reference, with the original German supplied to avoid equating the two English terms (Frege, 1892).

In the twentieth-century ordinary-language tradition, the explanatory center shifted toward what people do with words on actual occasions. Austin's lectures, published posthumously as How to Do Things with Words (1962), challenged the assumption that the essential role of utterances is always to represent facts. A promise, warning, verdict, or appointment may depend on an institutionally constituted situation in which an utterance performs an action. Austin's attention to felicity conditions made the setting of use essential to the classification and success of an act. This historical strand concerns the social effectiveness of speech as well as its descriptive content.

Wittgenstein's Philosophical Investigations, first published in 1953, develops language-games and the relation of linguistic expression to human practice. In §43 he proposes, for a large class of cases, an explanation of a word's meaning through its use in language. The qualifying formulation matters: it is a methodological redirection, not the simplistic thesis that any effect of a word in any situation is its meaning. Sections on following rules and criteria for correct application also indicate why public practices help stabilize what expressions can be taken to mean. A later contextualist can use this orientation to contest the ideal of an interpretation fully detached from the situations in which words operate; a semantic minimalist can still ask whether such practices generate regularities that a systematic semantics should describe (Wittgenstein, 1953/2009, §§7–43, 138–242).

Fregean analysis and Wittgensteinian attention to practice address complementary explanatory demands: the recurrence of logical and linguistic structure across uses, and the ability of situated agents to apply that structure in changing activities. Their historical relationship should be reconstructed through their texts rather than represented as a contemporary debate already fully formed. The later controversy over contextualism emerged when formal semantics, speech-act theory, and pragmatic inference developed precise and partially competing definitions of what a sentence contributes and what an utterance communicates.

From Conventional Meaning to Speaker Meaning: Grice

Grice's “Logic and Conversation,” delivered in the William James Lectures and published in 1975, transforms the treatment of context by distinguishing what is said from what is conversationally implicated. A speaker can communicate more than the proposition her words conventionally express because the audience reasons about the speaker's cooperative purposes, the informational needs of the exchange, and apparent departures from expected conversational behavior. A remark can carry a warning, a criticism, or a suggestion without encoding that content as part of the sentence's literal truth conditions.

The classical example of an indirect answer clarifies the distinction. Asked whether a colleague is reliable, a speaker replies that the colleague arrives punctually. The response literally concerns punctuality but may implicate a reserved assessment of the colleague's reliability. To explain the inference, the hearer considers what answer would be relevant, why a cooperative speaker offered something apparently less direct, and what the speaker presumably intended to convey. The implicated assessment is defeasible: further context can cancel it without rendering the words contradictory. The question of whether a content is cancellable helps distinguish conversational inference from a linguistically mandated contribution, although the test is less decisive than some early presentations suggest.

Grice's framework presupposes that we can identify a suitable content for what is said. Contextualists challenge how much work this identification requires. Suppose “Everyone is ready” is uttered before a seminar. The relevant domain is usually everyone expected to participate, not every human being. Is the restriction part of the proposition said, a hidden linguistic argument, or something communicated while the sentence retains an unrestricted content? Gricean reasoning can explain how the audience recovers the intended group, but that alone does not settle whether the recovery changes truth conditions. Pragmatic interpretation and semantic composition may cooperate at the same interpretive stage (Grice, 1975; Carston, 2002).

Korta and Perry's Critical Pragmatics (2011) further develops the relationship between reference, speaker intentions, and communicative interpretation, clarifying why an analysis of the reference of an expression and an analysis of what its user communicates must be jointly pursued (Korta & Perry, 2011).

The enduring achievement of the Gricean program is its demand that an account distinguish conventions governing expressions from inferences governing communicative acts. Its unresolved question is where to place contextually derived information that appears necessary for the proposition the speaker directly asserts. The subsequent positions in the contextualism debate offer different answers to that placement problem.

Formal Context Sensitivity: Kaplan's Character and Content

Earlier formal analysis of expressions whose reference depends on the occasion appears in Bar-Hillel's “Indexical Expressions” (1954). The historical continuity is methodological: formal context-sensitive interpretation requires rules relating recurring signs to occurrences of use.

Kaplan's “Demonstratives,” presented in an early version in 1977 and published in 1989, introduces the influential distinction between character and content. An indexical such as “I” has a stable linguistic rule that takes a context of use to an individual. The content of a particular token can vary as the person occupying the speaker role varies. Thus the sentence “I am standing” retains a fixed character across occasions while expressing different propositions when different people speak. The context sensitivity is systematic because the changing contribution follows a stable rule (Kaplan, 1989, especially the discussion of character and content).

Kaplan's framework also distinguishes a context of use from a circumstance of evaluation. The former selects the content of a context-sensitive expression. The latter is what the resulting content is assessed against, typically with a possible world and, in suitable formal systems, additional parameters. Confusing these dimensions obscures how indexicals can be tied to their use while the resulting propositions can be considered at alternative worlds. The contrast is central to the philosophical interest of direct reference: a particular use of “I” picks out its actual utterer for the purpose of the content, even when one assesses that person's possible conditions elsewhere.

The framework yields a simple analytical representation. Let S be a sentence, c a context of use, and w a world of evaluation. Then a character function assigns a content to S at c; the content, in turn, is evaluated for truth at w. Schematically, Content(S,c) = Character(S)(c), and Truth(Content(S,c),w) expresses evaluation. These expressions summarize the architecture; they are not asserted as quotations of Kaplan's own notation. They reveal where a semantic rule can use context without surrendering the idea of stable meaning.

A contextualist may accept this model but insist that the inventory of context-sensitive expressions is much larger than paradigmatic pronouns and demonstratives. A minimalist may accept the framework as a bounded explanation of semantic context dependence. A dispute about the extent of context sensitivity is accordingly different from a dispute about its possibility. No serious participant needs to deny that “I,” “here,” and “now” change their contribution with use; the controversy concerns whether a similarly structured operation is required for apparently ordinary predicates, quantified sentences, and descriptions of everyday situations.

Lewis, Conversational Score, and the Architecture of Evaluation

Lewis's “Scorekeeping in a Language Game” (1979) treats discourse as a practice whose conversational score can change as participants make moves. Presuppositions, standards, and other contextual coordinates are often accommodated, rather than mechanically found in a pre-existing list. A conversational contribution can alter which later contributions are acceptable, what is presupposed, and what counts as satisfying a relevant standard. This dynamic conception is essential because context changes during communication.

In “Index, Context, and Content” (1980), Lewis further distinguishes the role of contextual features in specifying which semantic objects are expressed from the indices used to evaluate contents. A theory may include possible worlds, times, places, speakers, addressees, or other coordinates, but their formal appearance does not settle the metaphysical or linguistic interpretation of those parameters. Some coordinates belong to the evaluation of a content, while others help determine content in the first place. Whether to represent a contextual dependence as variation in what is said or as variation in a truth-evaluation parameter is a substantive modeling choice.

Dynamic scorekeeping explains the phenomenon of accommodation. Imagine a participant saying, “The chair of the committee has arrived,” when the audience did not previously discuss any committee. Hearers may provisionally treat there as being a salient committee and chair, depending on the institutional setting. The utterance can prompt a repair or adjustment of common ground. The process is more organized than private guessing: participants monitor the compatibility of new assertions with what they can jointly treat as presupposed.

The limit of the scorekeeping metaphor is its generality. It describes a flexible public mechanism but must be combined with a theory of which semantic and pragmatic transitions are licensed. Some changes concern demonstrative assignment; others concern the adoption of a comparison class, an accommodation of background assumptions, or a change in standards of precision. A sophisticated contextual theory must identify the specific coordinate moved by each conversational development rather than treating all shifts as the same operation (Lewis, 1979, 1980).

Perry: Essential Indexicals and Unarticulated Constituents

Perry's “The Problem of the Essential Indexical” (1979) shows that action-guiding cognition sometimes depends on a first-person or otherwise situated way of thinking that cannot be replaced, without explanatory loss, by a purely impersonal description. Knowing that a certain person is making a mess and recognizing that I am the person making the mess can lead to different behavior despite the apparent co-reference. The philosophical point concerns cognitive significance and practical reasoning, with consequences for the semantics of self-reference.

In “Thought Without Representation” (1986), Perry develops the idea of an unarticulated constituent: an element of the relevant thought or proposition that is not represented by a corresponding component in the articulated sentence or thought. The local place relevant to “It is raining” supplies the standard example. It may be an indispensable element of what the utterance says truly or falsely even though no phrase in its overt form names the location. This account makes context do more than fill an explicitly marked indexical slot (Perry, 1986).

The central argumentative difficulty is underdetermination. The rain report can have different truth values depending on location, but the grammatical surface supplies no ordinary locative phrase. One solution identifies a contextually supplied constituent of the proposition. A rival solution posits a hidden locative position in the logical or syntactic structure. Another treats the sentence's minimal content as incomplete in some technical sense or assigns contextually enriched truth conditions only to the communicative act. Evidence for location sensitivity therefore supports a problem that multiple theories can solve; it does not on its own identify the unique metaphysical status of an unarticulated constituent.

The case also exposes the limits of stipulating the relevant context. A weather presenter who says “It is raining” while sitting in a dry studio may be reporting conditions in a distant region. The value cannot always be extracted from the speaker's physical location. It is often set by the topic of the conversation, a map, a camera image, or an announced forecast area. Context contributes through relevance, discourse organization, and shared attention; the utterance event alone is too thin to settle the matter.

Stalnaker: Common Ground and Context Change

Stalnaker's “Assertion” (1978) models communication through propositions treated as presupposed by participants and the possibilities that remain compatible with their conversational assumptions. Assertion proposes to reduce the set of possibilities left open in the common ground. The content of an assertion depends on a context, and the act of assertion aims to change that context. This dual direction of dependence provides one of the clearest frameworks for understanding discourse as an evolving interpretive situation (Stalnaker, 1978, pp. 315–332).

A common ground is not equivalent to an inventory of what each person actually and privately believes. It is the information participants treat as jointly available for present purposes. A speaker may cooperate temporarily with a stipulated fictional world, an assumption for argument, or an institutional convention without endorsing all its propositions as personal beliefs. The analytical advantage is that conversation can be described at the level of public commitments and presuppositions rather than requiring direct access to individual mental states.

The framework illuminates presupposition and accommodation. If a speaker says, “Her brother returned,” the possessive and the verb can invite assumptions about a salient woman, a brother, and an earlier presence or absence. Listeners can challenge those assumptions or accommodate them. But the fact that common ground changes does not by itself tell us whether the presupposed material is semantically encoded, conventionally triggered, or pragmatically induced. The theory is a framework for describing updates that other semantic mechanisms must specify.

Stalnaker's later collection Context and Content (1999) develops these relations between intentionality, context, and speech. Contextualism benefits from the model because content determination and conversational updating can be discussed separately and then related systematically. The central methodological result is that context belongs both before and after an utterance: it helps make an act interpretable, and the completed act alters the resources through which subsequent speech is interpreted.

Semantic Minimalism and the Limits of Encoded Sensitivity

Semantic minimalism aims to preserve a systematic, compositionally constrained account of sentence meaning while acknowledging that an utterance may communicate much more. Cappelen and Lepore's Insensitive Semantics (2005) defends the distinctive conjunction of semantic minimalism and speech-act pluralism. A sentence can possess a relatively stable minimal semantic content while its different utterances make distinct claims in context. The number of things appropriately said by a speaker need not determine the number of semantic contents attached to an expression type.

Montague's “Universal Grammar” (1970) exemplifies the kind of compositional semantic ambition that provides the background for this demand: a systematic account relates syntactic construction to semantic interpretation, even when different formulations of contextual dependence remain possible (Montague, 1970).

The attraction of this position is explanatory discipline. If the semantics of every ordinary expression changes with all conversational purposes, a semantic theory risks becoming a redescription of interpretation rather than an account of language-specific rules. Stable compositional contributions help explain how a finite linguistic competence supports novel expressions, how speakers can disagree using recurring sentences, and how explicit semantic inference works across conversational changes. Minimalism therefore asks contextualists to provide independent evidence that a purported contextual contribution belongs to literal truth conditions rather than to a communicative elaboration.

Borg's Minimal Semantics (2004) and Pursuing Meaning (2012) develop a closely related but individually distinctive defense of formal semantic theorizing. The rule-governed context sensitivity of pronouns, tense, and demonstratives is acknowledged; optional pragmatic elaboration is not automatically included in semantic content. A sentence such as “Ted is ready” can be given a minimal proposition while an ordinary hearer retrieves readiness for a relevant activity. The minimalist claims that the enriched interpretation is crucial to communication and often to assertion, but linguistic semantics need not encode the added purpose as an unspoken constituent.

The challenge concerns explanatory adequacy. If the minimal proposition is too thin to be the proposition a competent speaker actually asserts, semantic theory may preserve formal tidiness at the cost of separating itself from ordinary truth evaluation. The minimalist reply is that literal sentence meaning and asserted content are different explanatory targets. This is a substantive division of labor, not a denial of communication. The strongest form of the debate asks which predictions become possible only when a disputed enrichment is placed in semantics, and which predictions can already be recovered by an independent pragmatic account.

Indexicalism, Hidden Structure, and Stanley's Challenge

An alternative to unconstrained pragmatic enrichment holds that apparent context dependence is grammatically or logically mediated. Stanley's “Context and Logical Form” (2000) defends the thesis that the truth-conditional effects of extra-linguistic context are traceable to elements of a sentence's actual syntactic structure. He discusses examples of purported unarticulated constituents and apparently non-sentential assertions as potential counterexamples, resisting the conclusion that contextual effects must enter a proposition without linguistic mediation (Stanley, 2000).

The contrast is between an interpretation licensed by an articulated or unpronounced structural position and an addition produced entirely by conversational reasoning. Consider “Every student passed.” In ordinary use the domain is restricted to some class or institution; the question is whether a covert contextual restrictor participates in the semantics of the quantifier. If it does, the structure contains a position whose value context selects. If it does not, the literal universal generalization can be pragmatically understood as a claim about the intended group. Both explain surface variability; they disagree over the syntax–semantics interface.

Argument reconstruction clarifies the disagreement. Premise one: semantic composition should proceed from linguistic structure and the interpretation assigned to its parts. Premise two: utterances appear to vary in truth conditions with contextual choices. Premise three: if contextual choices alter the truth conditions while composition remains structurally constrained, the relevant values should be associated with positions in logical form. The conclusion supports hidden structure where independent linguistic and explanatory considerations establish it. The inference is strongest when binding, scope, anaphora, or systematic interaction with operators provides evidence beyond the original intuition of enrichment.

The contextualist objection is that a general license to posit unpronounced structure can make the thesis difficult to falsify. If every contextual influence is matched by an invisible variable, one has preserved compositionality by stipulation. The structural theorist can answer by demanding grammatical diagnostics and independent constraints. The dispute should therefore be decided case by case through predictions about binding, embedding, scope, and cross-linguistic expression, rather than by the bare assertion that language must or cannot hide syntactic material.

Moderate Contextualism, Impliciture, and Completion

Bach's “Conversational Impliciture” (1994) occupies a significant intermediate position. He identifies cases in which a speaker communicates an enriched proposition closely related to what is said. Completion supplies material required to articulate an otherwise incomplete semantic representation; expansion makes what is communicated more specific than a formally complete sentence. Impliciture is distinguished from Gricean implicature because the additional information forms part of a proposition the speaker communicates directly, instead of being a separate proposition inferred from what was said (Bach, 1994).

Suppose a person says “She has finished.” One reasonably asks: finished what? In some contexts a completion is needed to identify the intended propositional content. In a different example, “I have eaten” may communicate that the speaker has eaten dinner, even though the sentence literally states a much more general event. Whether these are logically parallel depends on the analysis of the predicate, the verbal construction, and how discourse participants use the sentence. A moderate theory need not classify all such cases as semantically identical.

Bach's intervention matters because it rejects a rigid binary between encoded semantic proposition and detachable conversational implication. Communication involves propositional contents situated at different degrees of explicitness. A speaker can exploit linguistic material as a schematic contribution and leave recognizable features to contextual completion without thereby producing an implicature unrelated to the direct statement. Yet identifying an intermediate communicative layer creates another question: do the truth conditions of an utterance attach to the conventional proposition, to the impliciture, or to a structured relation between the two?

A defensible response preserves separate explanatory questions. What do the words and syntax conventionally determine? What proposition does the speaker intentionally communicate as the primary act? Which further contents can the hearer derive? The answers can differ. The concept of impliciture demonstrates that this difference need not be dismissed as an inconsistency; it may describe how communication exploits a partially specified linguistic contribution.

Relevance Theory and the Content of Explicit Communication

Sperber and Wilson's Relevance: Communication and Cognition (first published 1986; second edition 1995/1996) explains communication through the audience's inferential search for an interpretation worth processing. The hearer uses linguistic decoding in conjunction with available assumptions and expectations about relevance. The account places contextual selection within a broader theory of cognition, attention, and interpretation rather than treating it solely as an exceptional repair mechanism for incomplete sentences.

Carston's Thoughts and Utterances (2002) develops the relevance-theoretic conception of explicature. Explicitly communicated propositions can be pragmatically developed from linguistically encoded logical forms through disambiguation, saturation, and enrichment. This framework challenges the idea that the explicit/implicit boundary coincides neatly with the semantic/pragmatic boundary. A proposition can be explicitly communicated even when pragmatic inference contributed to its full articulation (Carston, 2002).

A typical example is “There is nothing to eat.” Taken as a maximally unrestricted existential claim, it is almost certainly false whenever any edible item exists anywhere. In an everyday setting the speaker reports an absence of suitable food in an accessible or relevant domain. The interpretive process draws on expectations about the speaker's purpose and the audience's practical situation. Relevance theory can explain why the restricted reading arises quickly and without an explicit act of calculation, but it still faces the challenge of specifying which pragmatic contributions are integrated into the explicature and which remain implicatures.

The framework also motivates empirical questions. Comprehension involves processing effort, expectations about informative usefulness, and rapidly updated contextual assumptions. Behavioral experiments and computational models may help determine which interpretations are psychologically preferred; however, a measured preference does not by itself establish whether the selected interpretation is semantic, pragmatic, or conventionalized through repeated use. Philosophical analysis must connect psychological evidence to a discriminating theory of representational structure.

Truth-Conditional Pragmatics and Pragmatic Modulation

Recanati's Literal Meaning (2004) defends a contextualist treatment in which what is said—the truth-conditional content of an utterance—depends on processes of pragmatic interpretation. In Truth-Conditional Pragmatics (2010), he develops an account of pragmatic modulation including narrowing, broadening, and related changes in the contribution of expressions. His position asks whether context dependence extends beyond a small list of indexical words and whether pragmatic competence participates in determining the truth conditions speakers actually express (Recanati, 2004, 2010).

Suppose someone says “The city is asleep.” The utterance can be used as a metaphor, an ordinary description of reduced activity, or a claim about the inhabitants rather than the urban area as a physical object. Context influences which property is attributed and how the predication should be evaluated. Some cases are recognizable figures of speech; others look like routine applications of a word whose occasion-specific extension is adjusted without any conspicuous metaphor. A truth-conditional pragmatist argues that these adjustments can contribute to what the speaker says, rather than merely adding secondary suggestions to a fixed proposition.

The core philosophical issue is whether an interpreted predicative contribution can be both constrained by lexical meaning and freely modulated by pragmatic purposes. A theory that includes arbitrary ad hoc narrowing loses predictive force. A theory that excludes all pragmatic modulation risks treating the content of natural utterances as unrelated to ordinary evaluations of their truth. Recanati's program locates interpretation within the constitutive production of utterance content; the central remaining task is to explain how default lexical constraints, salient contrasts, and speaker intentions organize that production.

A useful distinction separates saturation from modulation. Saturation fills a position that a chosen semantic analysis makes obligatory—for example, a parameter associated with an overt indexical or a demonstrably encoded contextual slot. Modulation adjusts the interpretation of a contribution even where no such required slot has been independently established. Different theories disagree about which examples fall into which category. That disagreement is substantive, because saturation ordinarily appears more amenable to compositional rule specification than unrestricted modulation.

Occasion Sensitivity and Travis's Challenge

Travis's “On What Is Strictly Speaking True” (1985) advances a powerful case for occasion sensitivity. The same words, used of the same objects, can contribute to different assessments of truth depending on what is at issue. A description is applied under standards of relevance and correctness fixed by the occasion. The challenge reaches beyond speaker intentions to the conditions under which a representation counts as true. The familiar idea that a standing sentence has independently fixed truth conditions is thereby questioned (Travis, 1985).

The philosophical force of occasion sensitivity can be reconstructed through ordinary predicates. Someone describes a painted leaf as green. In a botanical discussion concerned with natural coloration, the description may be rejected; in a conversation about the visible colors of a design, it may be accepted. The stipulated object and the conventional adjective remain stable, yet the criteria for its application differ with the question under discussion. A critic might attribute the difference to implicit comparison classes, different propositions, or pragmatic standards for assertion. Travis's challenge presses for a reason to locate the crucial variation in one of those places rather than simply assuming that the sentence already bears an invariant condition of truth.

Travis's Thought's Footing (2006) relates these questions to Wittgenstein and Frege. Its concern is how thought and judgment achieve a determinate bearing on the world through forms of human practice. The account gains strength from examining cases in which interpretation is not reducible to choosing a referent for a particular indexical. It faces a corresponding theoretical demand: if occasion sensitivity is pervasive, what explains reproducible understanding and valid inferential transitions across occasions? A theory cannot rely entirely on the assertion that standards shift; it must specify how competent speakers recognize a shift and preserve the relevant continuities.

The significance of the Travis challenge is methodological. It demands that philosophers resist projecting the ideal of mathematically complete semantic interpretation onto ordinary speech without argument. Equally, it invites the contextualist to show how situated judgment operates with enough stability to support communication, correction, and reasoned disagreement. Occasion sensitivity and systematic semantics must be evaluated against the same explanatory phenomena.

Contextual Variation in Different Linguistic Constructions

Context dependence has different sources across constructions. With “I,” the ordinary rule selects the utterance speaker, though quotation and special uses can complicate the picture. With “that,” the demonstrative target can depend on gestures, intentions, attention, and conversational salience. With “here,” a place can be a building, a city, or a region selected by context. These are recognizable candidates for conventional indexicality and can be studied using Kaplanian character (Kaplan, 1989; Stanford Encyclopedia of Philosophy, “Indexicals,” revised 2026).

Gradable adjectives introduce a distinct problem. “Tall,” “expensive,” and “near” are often assessed against comparison classes and standards. A child may be tall among eight-year-olds and short among professional basketball players. The words carry stable dimensions of comparison, yet the threshold for a correct use is variable. A formal account may associate a degree argument or standard parameter with the adjective; a pragmatically oriented account can instead model how an assertion selects a practically appropriate comparison. Both need to explain embedded uses, negation, and disputes over the relevant class.

Quantifier restriction poses another question. “Every book is damaged” usually means every book in an identifiable collection, not every book in existence. Restriction can be represented through a contextual domain parameter; alternatively it can be viewed as part of an utterance's pragmatically specified content. The interpretation is constrained by the topic, preceding descriptions, and practical purposes. A comparison of these approaches requires independent evidence from syntax, binding behavior, and inferences to other quantified claims.

The apparent incompleteness of “ready,” “finished,” or “enough” is different again. Such expressions invite questions about an activity, task, or threshold. Some constructions include semantically required complements, while others allow complete grammatical sentences whose utterance conveys an enriched content. The philosophical task is to identify which level of underdetermination is at issue rather than conflating ordinary information gaps with a specific kind of missing semantic argument.

Knowledge Attributions and Epistemic Contextualism

Epistemic contextualism is a restricted thesis about uses of terms such as “knows.” On a prominent semantic version, the standards governing whether a knowledge attribution is true vary with features of the attributor's conversational context. In a routine setting one may accurately say that a commuter knows the train departs at eight; after skeptical alternatives become salient, a similar attribution may be challenged. The relevant change is commonly modeled in the standards that must be met for the attribution to count as true (Cohen, 1988; DeRose, 1992).

The attractiveness of the view is its potential to explain the contrast between ordinary knowledge judgments and demanding skeptical discussions without attributing a simple logical error to one party. If “knows” is context-sensitive, different uses may express different conditions while employing the same word. The philosophical price is the need to explain how that variability is linguistically realized, what contextual parameters determine it, and why speakers often regard themselves as directly disagreeing across contexts.

This domain also clarifies the difference between semantic and subject-sensitive invariantist approaches. A contextualist places variation in the content or truth conditions of knowledge-ascribing sentences as used by attributors. An invariantist can hold the semantic content constant while arguing that practical stakes affect whether a subject knows, or while explaining shifts in judgments through pragmatic salience. These approaches require different accounts of what changes: the meaning of the attribution, the epistemic situation, the standards of evaluation, or the appropriateness of making the claim.

The epistemic example belongs to this dossier because it tests the explanatory reach of linguistic contextualism. It does not collapse the general theory of knowledge into language. The metaphysics of knowledge, the norms of inquiry, and the semantics of “knows” remain analytically distinguishable. The strongest contextualist claim is a linguistic hypothesis with epistemological implications; its success cannot be established by skeptical intuitions alone (Stanford Encyclopedia of Philosophy, “Epistemic Contextualism,” revised 2026).

Context of Use, Context of Assessment, and Relativism

Contextualism commonly permits the same sentence type to express different propositions across contexts of use. Assessment relativism introduces a further possibility: a proposition already fixed by one context may vary in truth value with a context of assessment. MacFarlane's Assessment Sensitivity (2014) develops this approach and applies it to domains including taste, knowledge attributions, future contingents, and epistemic modals. The crucial distinction is between a contribution made when a sentence is used and a parameter involved when the resulting claim is later assessed (MacFarlane, 2014).

Suppose one speaker says “This soup is delicious” and another says “It is not delicious.” A conventional contextualist may interpret their utterances as expressing different propositions relative to different taste standards. A relativist may instead propose that the same proposition is appropriately assessed from different standards and that the speakers engage in genuine disagreement despite the possibility of faultless assessment. How much disagreement each theory preserves depends on the adopted account of proposition identity, retraction, and norms of assertion.

A general defense of assessment relativism must explain why it is preferable to ordinary context-sensitive semantics plus pragmatic accounts of disagreement. The phenomenon of apparently faultless disagreement motivates the theory but does not uniquely establish it. A contextualist can describe conflicting attitudes, rival standards, or a debate over which standard to adopt. A relativist gains a distinctive account if assessment-sensitive truth helps explain the rationality of retraction, persistence of conflict, and the evaluation of past assertions without multiplying use-relative propositions unnecessarily.

The formal difference matters for this monograph's central taxonomy. Context of use determines which content is expressed. Context of assessment, in a relativist model, helps determine how an already identified content is evaluated. Conflating these roles makes any perspective-dependent judgment look like contextualism. A precise theory identifies whether the variation lies in lexical interpretation, propositional composition, assertion, or the circumstances governing an assessment.

Truth Conditions, Assertions, and the Phenomenology of Disagreement

The test of truth conditions is influential because a change in conversationally suitable interpretation often seems to change what must obtain for an utterance to be true. Yet truth judgments are elicited from agents performing multiple tasks at once. They assess lexical meaning, communicated intention, conversational appropriateness, background presuppositions, and sometimes the moral or practical acceptability of what was said. A single intuition about whether “What she said was true” establishes less than a theory of semantics might require.

A methodological distinction is therefore needed between truth of the proposition conventionally supplied, truth of the primary asserted proposition, and correctness of the communicative act. Consider a speaker who says “There is no food” while a jar of mustard remains in the cupboard. The speaker can accurately communicate a lack of suitable dinner despite the literal presence of something edible. Whether the utterance is false under a minimal semantics and true under a pragmatic assertion standard depends on which semantic object the theory identifies as the bearer of truth. The example demonstrates interpretive divergence; the disputed classification requires further argument.

Reporting tests may also matter. “She said that everyone had passed” can be an accurate description even when the reporter and speaker have different salient groups. The degree of flexibility in indirect report supports Cappelen and Lepore's speech-act pluralism, but it may also be explained by tolerant paraphrase or anaphoric context reconstruction. Reports are evidence about ordinary practices of content attribution; they must not be promoted automatically into a unique criterion of semantic proposition identity.

Disagreement is similarly ambiguous. Two speakers can dispute a fact, disagree about a standard, express conflicting preferences, or attempt to negotiate the conversational framework. Cross-context disagreement may survive differences in truth-conditional content because participants contest how a claim should be understood or used. A philosophical account should classify the disagreement before inferring that both speakers expressed the same proposition or that they expressed different ones.

Central Arguments and Their Critical Evaluation

A strong argument for contextualism begins from cases where ordinary truth assessment varies with circumstances while overt syntax remains constant. The same sentence appears appropriately true in one situation and false in another. If this variation is not adequately modeled as a shift in what the speaker merely implies or as a change in a linguistically licensed parameter, the proposition asserted requires pragmatic contextual determination. This argument gains strength from cases in which the disputed material contributes to what participants directly reject, deny, or infer from a claim.

The minimalist reply begins from the division between sentence semantics and utterance interpretation. Different communicative judgments need not imply a plurality of conventionally expressed semantic propositions. A thin semantic contribution can be a stable object of linguistic explanation, while pragmatics identifies what a speaker asserts using that contribution. The reply gains strength from syntactic productivity, compositional relations, and cases where the same sentence can be reused under divergent purposes without linguistic reanalysis.

The indexicalist response accepts context-dependent truth conditions but seeks a structural source for every such dependence. Its strength is predictive when hidden parameters interact with binding, quantification, or other compositional operations. Its weakness arises when the proposed structure is justified solely by the contextual reading it was designed to explain. To avoid circularity, covert positions should be constrained by independent evidence. Conversely, contextualists must not treat a failure to identify overt structure as proof that no relevant syntactic structure exists.

The occasion-sensitive response insists that even complete compositional representations can leave unsettled how predicates apply to particular objects in particular activities. This challenges the assumption that every correct application must have been fixed by a context-independent extension. The difficulty is identifying a principled account of stability and rational adjudication. If a competent speaker can always alter what counts as truth by selecting a novel occasion, the account needs constraints explaining why some uses are mistaken, unreasonable, or unavailable.

No one argument by itself establishes that all expressions are semantically context-sensitive or that only a short list is. The philosophical gain comes from disaggregating cases by the kinds of evidence they provide. Linguistic productivity bears on compositionality; ordinary truth judgments bear on assertoric practice; binding and embedding bear on structural hypotheses; accommodation bears on discourse dynamics; retraction bears on assessment norms. Competing theories often succeed at explaining different subsets of these phenomena. An adequate general account must say when the explanations should be integrated and when they must remain distinct.

Linguistic, Cognitive, and Computational Interfaces

Experimental pragmatics investigates how listeners choose interpretations under changes in available information, task constraints, and expectations. Studies of scalar implicature, reference resolution, and gradable predicates bear on the timing and robustness of contextual inference. The conceptual significance of these studies is conditional: observed interpretive behavior can constrain models of processing and acquisition, but the semantics/pragmatics boundary is a theoretical distinction that cannot be read directly from reaction times or preference frequencies. One must specify a hypothesis whose competing interpretations generate different experimental predictions.

Corpus linguistics reveals how context-sensitive expressions occur in discourse rather than in philosopher-designed isolated sentences. Demonstratives co-occur with gestures and shared attention; domain-restricted quantifiers occur in topical sequences; gradable adjectives cluster around practices of comparison. Such data can improve the empirical coverage of theoretical examples. However, corpus distributions cannot by themselves determine whether a pragmatic effect is obligatory in the semantic representation, conventionalized through repeated interaction, or inferred afresh. Annotation schemes necessarily encode some theory of what counts as the relevant context.

Computational language systems introduce an additional level. A model can use a textual window, retrieved documents, tool outputs, and conversational history to infer referents, resolve ambiguity, and generate contextually appropriate responses. These artifacts make interpretive dependence visible in a form that can be experimentally varied and archived. Their success at contextual resolution does not settle the philosophical question of whether the system's outputs bear identical kinds of intentions, commitments, or pragmatic agency to human utterances. The computational finding concerns performance under supplied information; a theory of assertion and social responsibility requires further premises.

The editorial horizon Artificial Era · From Homo to Artificial provides a further question about the persistence of communicative context across nonbiological public records. A digital text can be quoted, reissued, excerpted, and reinterpreted after the original occasion of production. Accurate interpretation may therefore require distinguishable records of provenance, version, intended audience, and later uptake. This is an original application of the distinctions developed above, not an assertion that any classical theorist anticipated Aisentica's ontology or that the canonical definitions of that system have been revised. The relevant philosophical contribution is the claim that context reconstruction is a problem of publicly recoverable relations among utterance, record, interpreter, and assessment.

A Layered Synthesis of Context Determination

The examined theories support a structured synthesis. The first layer is conventional linguistic character: lexical meaning, syntax, compositional constraints, and rules for overt indexicals. The second is parameter assignment, where discourse and situation supply values for positions whose status a semantic theory can independently motivate. The third is primary utterance interpretation, where speakers and audiences use background assumptions and communicative purposes to identify the content of a particular speech act. The fourth is conversational consequence, where presuppositions, implications, commitments, and common ground are updated. A fifth layer, when the theoretical framework calls for it, governs assessment from a perspective not identical with the original context of use.

These layers describe explanatory functions, not a mandatory psychological sequence. Human comprehension can be rapid and interactive; contextual constraints may guide lexical access before an explicit composition is consciously available. The point is that a philosopher assessing a context-dependent interpretation must specify which explanatory function it performs. The five-layer distinction prevents an inferential enrichment from being assumed to be a lexical meaning, and prevents an evaluation parameter from being treated as an element of the proposition merely because both depend on circumstances.

The synthesis yields a three-part adequacy condition for appeals to context. A theory must identify a variable whose value matters to interpretation; provide a selection rule or publicly intelligible practice for determining that value; and explain the representational location at which the value contributes—character, content, asserted proposition, implicated proposition, conversational commitment, or assessment. Without these requirements, “the context changes the meaning” remains an evocative description with limited argumentative force.

This analysis also preserves genuine disagreement among major theories. A radical contextualist can argue that ordinary predication remains open to occasion-specific specification. A minimalist can argue that systematic sentence meaning remains relatively invariant. An indexicalist can search for structurally realized contextual parameters, and a relativist can vary assessment instead of use-relative content. The synthesis does not turn them into verbal variants. It establishes a common set of diagnostic questions against which their competing commitments can be compared without assuming any one theory's preferred level of description.

A further conclusion concerns the identity of linguistic content across time. A recorded sentence can preserve a stable sequence of signs while losing essential interpretive conditions. The recovery of a historical utterance requires more than a lexical dictionary; one may need the surrounding discourse, referential anchor, social role of the speaker, and prevailing standards of evaluation. Yet recoverability is often partial. Scholarship should indicate which contextual elements are documented, which are inferred, and which remain indeterminate. This makes context a question of evidence and philosophical interpretation as well as of semantics.

Open Questions for Contemporary Philosophy

A fundamental open question is how to identify the boundary between the linguistic and the pragmatically supplied without making it depend entirely on the theory one has already adopted. The evidence supplied by grammatical binding, interpretation under embedding, cross-linguistic variation, and explicit contradiction can be combined, but the combination needs a principled method of weighting. A disputed interpretation may pass one diagnostic test and fail another, leaving the explanatory level unsettled.

Another question concerns the individuation of propositions. If two utterances have different relevant comparison classes yet speakers treat their claims as directly opposed, is the opposition a relation between distinct propositions, between attitudes toward one proposition, or between incompatible standards? The answer affects theories of assertion, belief, truth, translation, and rational disagreement. A successful theory must explain how proposition identity interacts with the pragmatic tolerance of ordinary reporting.

Context reconstruction in digital communication provides a third domain. Messages survive the situations that initially made them intelligible. Search snippets, quotations, and model-generated summaries may move text into new settings with different questions and audiences. A semantics of archived communication should distinguish the original context of production, the context preserved in the record, the later context of retrieval, and the current context of assessment. This problem connects classical indexicality and common ground to questions of provenance, reproducibility, and interpretive responsibility.

Finally, a philosophy of context must explain its own standards for error. Interpretation is constrained by linguistic form, facts, and recognizable public practices. Contextual variation does not authorize unlimited reinterpretation. One can misidentify a demonstratum, impose an irrelevant domain restriction, attribute a false presupposition, or misconstrue an institutional utterance. A theory with no account of these errors cannot distinguish flexible understanding from arbitrary projection.

Conclusion

Linguistic context is a structure of interpretive conditions that enables language to function as determinate communication across changing circumstances. Its philosophical significance lies in the fact that the same recurring linguistic forms can enter different acts, contribute different contents, and be evaluated under different standards. The history from Frege and the ordinary-language tradition to formal semantics and contemporary contextualism reveals a durable problem about the allocation of explanatory responsibility among linguistic convention, compositional form, public discourse, and pragmatic inference.

The principal results of the investigation are methodological and substantive. Indexicals exhibit rule-governed sensitivity; hidden-structure theories give contextual variation a compositional location; moderate contextualism and relevance theory identify forms of directly communicated enrichment; truth-conditional contextualism and occasion sensitivity challenge the stability of literal truth conditions; minimalism preserves invariant sentence content alongside plural speech acts; and assessment relativism identifies a distinct source of variability in the evaluation of already expressed contents. These positions offer competing explanations of specific phenomena rather than a single disagreement over whether circumstances matter.

The resulting philosophical criterion is precise: any explanation by context must identify what varies, which feature selects the variation, and where that selection enters the theory of meaning, content, assertion, or truth. This criterion allows historical and contemporary approaches to be compared on explanatory grounds. It also extends the study of context to durable digital discourse, where the original occasion of use, the surviving textual record, and later acts of interpretation are distinct yet related parts of one public history of meaning.