Distinguishing Semantics and Pragmatics

What’s the difference and how do they relate?

Pragmatics is a sort of funny thing. On the one hand, pragmatics is an important subfield in linguistics that produces a valuable research and contributes greatly to our understanding of language. On the other hand, pragmatics probably wouldn’t even exist at all as a field if Chomsky had not decided to say that language was this autonomous thing that exists separately from how we conceptualize & interact with the world, not to mention how we socialize & communicate with other human beings.

If there’s a pattern in the data,
it probably isn’t pragmatics.

You could easily make the case that all meaning is pragmatic because it always requires context-sensitive interpretation of not only language, but also interaction with others. In Part I of my review of Porter (2015), there was one fairly long quote that I included that is also relevant here. It is one of my favorite quotes on the topic of pragmatics and semantics and it comes from John Taylor’s (2004) book, Linguistic categorization. This book is one of a handful of linguistics books that is so useful and generally accessible that I think everyone should read it, regardless of whether you are a New Testament scholar/student, Old Testament scholar student, or grammarian/linguist.

The larger context of the quote is about the status of metaphor in language and grammar. Traditionally metaphor has had a low status and anything that was metaphorical was viewed as somehow deviant or non-normal/standard. Metaphor just isn’t as well-behaved in terms of being easily squeezed into boxes with +/- features. It’s harder to control with formal logic and mathematical operators. With that background, here’s a larger version of the same quote:

The view that metaphor lies outside the study of linguistic competence proper underlies Searle’s (1979) well-known account. The sentence in (22) is, if taken literally, semantically anomalous.

(22) Sally is a block of ice.

Ice (and block of ice) possesses the feature [-ANIMATE]; one cannot therefore predicate ‘be a block of ice’ of an entity (i.e. Sally) which is [+ANIMATE]. The sentence is only acceptable to the extent that a listener can go beyond the literal meaning and construe the speaker intended meaning. To perform this task, the listener needs to supplement linguistic competence with proficiency in pragmatics. Searle’s account thus presupposes a distinction between semantics and pragmatics [his emphasis], the former having to do with literal, or purely linguistic meaning, the latter with the context-dependent construal of intended meaning. Over the past decade or so, pragmatics has emerged as an important subdiscipline of linguistics, taking its place alongside the more traditional components of linguistic study, such as phonology, syntax, and semantics. Given the basic assumptions of the generative paradigm, the emergence of pragmatics as an independent object of study was perhaps inevitable. If language constitutes an autonomous cognitive system, then, given the self-evident fact that language is an instrument for conceptualizing and interacting with the world, the need arises for an interface that links these otherwise independent systems. Pragmatics functions as precisely such an interface. In rejecting the notion of an autonomous linguistic faculty, cognitive linguistics necessarily removes the need for pragmatics as a separate branch of study. All meaning is, in a sense, pragmatic, as it involves the conceptualizations of human beings in a physical and social environment. As Bosch (1985) has argued, the understanding of any utterance requires an act of context-sensitive interpretation by the listener/hearer; metaphorical utterances, on this view, do not form a special set.

(Taylor 2003, 132-3).

That last sentence is the crux of the matter: “the understanding of any utterance requires an act of context-sensitive interpretation by the listener/hearer; metaphorical utterances, on this view, do not form a special set.” So then, when we talk about pragmatics vs. semantics, how do we go about distinguishing them (or should we)?

That’s the harder question. In terms of a continuum, the defining feature of the continuum is conventionalization. The more a ‘meaning’ or ‘function’ is closely paired to a linguistic form, the more semantic it is. Conversely, the more a meaning or function is only evoked because of social or cultural context, the more pragmatic it is.

If we’re sitting in a hot room in the summer and I ask you, “Is the air on?” The contextual cues are probably enough for you to know that I would like you to turn on the air. That’s extremely high on the social context side and very low on the form-meaning pairing side.

All meaning is, in a sense, pragmatic, as it involves the conceptualizations of human beings in a physical and social environment.

—John Taylor

On the other side of the spectrum, individual lexical items and syntactic structures are highly conventionalized. They’re super-well-established form-meaning pairings. You know that “give” implies transference of something from an agent to a recipient that will have the structure:

subj – verb – indobj – dobj
John gave Mary a pizza.

None of the nouns are inherently paired with the pattern, but once you’ve heard “give” you know the pattern and know how to interpret the structure. Those are sort of the obvious instances. But the transference structure itself has been conventionalized sufficiently that it can function as a template and give can be removed and other verbs can be inserted.

subj – verb – indobj – dobj
John baked Mary a pizza.
John slid Mary a pizza.
John tossed Mary a pizza.

Baked is not inherently ditransitive (two objects), so there’s more here than merely a syntax tree and inserted lexical items. But it is also not merely context and culture either. The structure is so conventionalized to the pairing that it takes on the meaning, too (for more on these sorts of constructional generalizations and many others, see Goldberg 2006).

So the question ends up being: “Does this form meaning X or does the context merely imply X?” That’s usually a pretty easy question for your native language. But it’s going to be harder for Greek.

Chris Fresch (2016), in his article on non-past referring aorists, talks about the criteria for this in his chapter in the Greek Verb Revisited and does it really well. Cancellability is a part of it. But cancellability is a question of an individual instance and there are a number of complicating issues beyond that. It certainly is not a statistical standard in the way that some want it to be for the ἐ- prefix/augment and past tense.

Because of that, when we’re dealing with a language like Greek, we can be more flexible with our statistics. How often does a form and a meaning need to be paired together to become a convention? Not much. If a lexeme pairs with a meaning 25% of the time and pairs with another meaning 75% of the time, any reasonable lexicographer would say that lexeme has two senses. They wouldn’t say that the sense that appears 25% of the time is merely pragmatics. And grammatical and syntactic patterns should be treated similarly.

Basically: If there’s a pattern in the language data, it probably isn’t pragmatics.

A good resource for learning more about pragmatics and grammar is Ariel’s (2008) Pragmatics and grammar.