This article is part of a series entitled Redividing Linguistics. See also:
Morphology and Syntax: What’s In A Name?
- Syntax and Semantics: The Search For Meaning
- Semantics and Pragmatics: The Reality Interface
Semantics and pragmatics, in linguistics, distinguish between two types of meaning. Semantics refers to the meaning which is encapsulated in the language itself. Pragmatics refers to meaning which must be accrued through cultural context.
Certain statements are true by virtue of their words’ meaning alone. These exemplars of semantic meaning include (I take these from Fromkin’s Linguistics): “People are people”, “Either there are witches in the forest, or there aren’t,”, and “The number four is even.”  Philosophers often call these semantic certainties analytic truths.
Pragmatic truth requires going out into the real world. Statements — again from Linguistics — like “Nothing travels faster than the speed of light”, a physical truth, or “The king’s subjects pay taxes”, a legal truth, depend not only on language, but also on matters of fact . These statements are called by philosophers synthetic truths.
This so-called analytic-synthetic distinction has played an important role in modern philosophy. What can it tell us about linguistics?
In a seminal paper , William Van Orman Quine challenged the very notion of analytic truth.
There are two things philosophers might mean when they say that a sentence is “analytic”.
(1) No unmarried man is married. “Logical truths” like (1) are true in virtue of a language’s logical particles alone.
(2) No bachelor is married. Statements like (2) can be transformed into logical truths through an appropriate sequence of replacements of synonyms.
It’s with the second category of analytic truth which Quine takes issue. How exactly do we conclude that two words are synonymous?
Definition. To consult the dictionary would be to “put the cart before the horse,”  writes Quine. Lexicographers are “empirical scientists”, tasked with the recording of already-extant synonymic relationships. “[T]he ‘definition’ which is the lexicographer’s report of an observed synonymy cannot be taken as the ground of the synonymy,” Quine reminds us.
Interchangeability. Maybe interchangeability can help us define synonymy. Let’s consider words which can be replaced in all situations while preserving truth. Can we be sure that they’ll be synonyms?
Beginning with the obviously (logically) true “All and only bachelors are bachelors”, a quick replacement yields “All and only bachelors are unmarried men”. This, by hypothesis, must remain true. The truth of the latter sentence, though, can tell us merely that ‘bachelor’ and ‘unmarried man’ have identical extensions — that they refer to the same class of objects — and this is hardly sufficient for synonymity! Indeed, ‘creature with a heart’ and ‘creature with a kidney’ happen to have identical extensions, but they’re not synonyms. 
Let’s begin instead with the logically true “Necessarily all and only bachelors are bachelors”, and replace to yield “Necessarily all and only bachelors are unmarried men.” This, now, seems sufficient to conclude that ‘bachelor’ and ‘unmarried’ man are synonyms. Not so fast, Quine insists. What does ‘necessarily’ mean? “Necessarily all and only bachelors are unmarried men” is true precisely when “All and only bachelors are unmarried men” is analytic. The adverb “necessarily”, which we’ve snuck in in our attempt to define synonymity, presupposes the notion of analyticity.
Defining synonymy through interchangeability is either too weak or circular.
Semantic Rules. Quine pays tribute to various attempts by logicians to define formal languages, and then to define analytic truths as those produced within the languages by a certain set of formal rules. In the end, Quine concludes, these rules must be arbitrary. “Instead of appealing to an unexplained word ‘analytic’,” Quine insists, “we are now appealing to an unexplained phrase ‘semantical rule’.”  We still don’t know what analyticity means.
Quine ultimately concludes that the concept of analyticity does not exist. Indeed, understanding the statements “No bachelor is married” and “John is married” both require matters of fact in a way that is philosophically indistinguishable. Knowledge of synonymity is not inherent in the language itself, but unavoidably dependent on the world.
Perhaps we can take Quine’s startling argument farther. Though Quine attacked analytic truths of the second type — by challenging the notion of synonymity — logical truths, too, seem in danger.
Let’s modernize our terminology. The “logical particles” to which Quine refers are probably grammatical morphemes, “which the speaker uses to signal the relationship between a word and the context in which it is used.”  His “atomic, or noncompound components” can be considered lexical morphemes, which “refer to items, actions, attributes, and concepts.” 
Grammatical morphemes are used in the construction of meaning. Meaning is assembled compositionally. According to the principle of semantic compositionality, “The meaning of a sentence is determined by the meanings of its parts and the ways in which those parts are assembled.”  Lexical morphemes furnish the meanings of a sentence’s parts. Grammatical morphemes — as well as structural properties of the language itself — determine how the meanings of those parts are assembled. We can deem as logical truths those sentences whose truth can be determined through grammatical morphemes and assembly properties alone.
Are logical truths, like non-logical “analytic” truths, inseparably pragmatic?
Quine urged us to refrain from turning to dictionaries for information regarding lexical meaning. Likewise, we should resist the temptation to consult some mystical “official law of language” to determine the meanings of our grammatical morphemes. Such laws, if they existed, could only describe existing relationships.
Grammatical morphemes — despite forming a “closed class” to which members can’t be adjoined — describe relationships which are ephemeral. Linguistics compares prespositions in German and Dutch, providing concrete evidence that even prepositions transform. Though German’s auf, an, and um and Dutch’s op, aan, and om are historically related and hold similar meanings — on horizontally, on vertically, and on around, respectively — the usage of these pronouns has diverged. “[I]n Dutch, the distinction… has less to do with orientation than with method of attachment,” Linguistics observes .
That grammar transforms should raise questions about its embeddedness. Let’s again consider German and Dutch prepositions. Would we consider these facts — especially, to use some poetic license, right around their time of divergence — facts of language, or facts of usage? The distinction is perhaps unfounded.
Why must we cordon off particular meanings as intrinsic to a language and deem the rest facts of the world? Languages are tools of humans, constructed by humans, to deal with objects in the real world. Languages have meanings. These meanings are inseparable from our environment! All semantics are pragmatics.
Chomsky once claimed that “It is possible that natural language has only syntax and pragmatics.”  I’d like to argue that language has only structure and meaning. Structure dictates the rules by which language is formed. Meaning permits us to interface with the world. These branches, of course, are closely related. Structure’s rules are expressed with reference to meaning’s information.
We’ve seen now that meaning is inescapably pragmatic. Linguistics — the science of structure and meaning — is both redivided and unified.
- Fromkin, V. (2000). Linguistics: An introduction to linguistic theory. Malden, Mass.: Blackwell.
- Quine’s Two Dogmas of Empiricism
- Chomsky, Noam. 1995, “Language and nature”, Mind, 104: 1-61