Four Ways of Studying Language

Four Ways of Studying Language

What is the linguist or philosopher of language studying? It will be useful to distinguish four different (but overlapping) areas of study: ontological, epistemological, behavioral, and phenomenological. By “ontological” I mean the study of language as a formal object: its nature, structure, and inner workings. What is it composed of? How is it structured? How does it work? Here we will find discussions of the nature of propositions, grammar, logical form, sense and reference, objects and facts, Platonism and anti-Platonism. We will find proposals about what meaning is and what words refer to (if anything). The subject is language itself not our psychological relation to it. The case is like the ontological study of mathematics—what it is in itself not how it relates to the human mind. By “epistemological” I mean (predictably) our knowledge of language—what is variously called linguistic competence, mastery, understanding, grasp. This is analogous to our knowledge of mathematics. This kind of study is clearly dependent on the first kind, since how something is known depends on what it is. What kind of knowledge is this and how is it possible? Third, we have behavioral studies—what is called linguistic performance, use, action, utterance. This will involve how linguistic knowledge is related to actual speech—the cognition-action nexus. Fourth, we have the phenomenological investigation of language—its expression in consciousness. This takes in all the ways in which language finds its way into our subjective experience: the felt character of language, its mode of seeming. Thus, we have what language constitutively is, our mode of knowing it, how it manifests itself in behavior, and what it is like to have it. These should not be confused or run together, though there will no doubt be plenty of overlap. The totality of these ways forms the subject matter of the study of language. We might also add the social psychology of language, or the politics of language, or its relation to human emotion; but these may be subsumed under the categories already listed, being concerned with the psychology of language generally. Epistemology is part of psychology, broadly construed—one department of the human mind, the knowing part. In practice, such non-epistemological psychological studies are seldom encountered in theoretical linguistics treatises.

I will make some remarks about the nature of linguistic knowledge, which seems to me in need of fresh insights. There is the much-debated question of whether linguistic knowledge is similar to other kinds of knowledge, in particular, knowledge-how and knowledge-that. When a speaker knows what a sentence means, does he or she have a true justified belief about that sentence, or is it more akin to an ability?  Is unconscious knowledge of the syntactic rules of a language a type of true justified belief? Is knowledge of language simply the application of a general faculty for knowledge to a particular subject matter, viz. language, or is it a special type of knowledge? Do we know language in the same sense in which we know geography and history? Then there is the question of how innate our knowledge of language is—completely, partially, or not at all. The question I want to focus on is whether linguistic knowledge is a priori, and if so how. The other questions have been widely debated, if not resolved, but this question has received little or no attention. It also strikes me as hard. I incline to the view that knowledge of language is special and sui generis; it is not just one application of a general knowledge faculty. That idea has already been cast into doubt by the existence of a priori knowledge, but linguistic knowledge puts a new wrinkle on it. Suppose I know what “Tom is bald” means: I know that the sentence in question means that Tom is bald. Is this knowledge a priori or a posteriori? My knowledge that Tom is bald is clearly a posteriori (I saw his bald head yesterday), but what about my knowledge of the meaning of the sentence? Well, how do I know its meaning? You might say I know it by learning certain empirical facts about the sounds or marks composing the sentence—facts of a conventional nature. That is no doubt true and to that extent my knowledge is a posteriori: but is it all there is to my knowledge of the sentence’s meaning? No, because I have to grasp how the sentence is put together, its grammatical structure. This involves knowledge of how nouns and verbs combine to produce sentences, i.e., form predicative propositions. So, do I know a posteriori how nouns and verbs work together to produce sentences? Have I seen this happen a number of times in the past and thereby infer that it will go on happening in the future—is it a case of empirical inductive knowledge? Apparently not: I know it without needing to undertake empirical investigations of this type. I know it a priori, just by knowing what nouns and verbs are and how their combination produces whole sentences; arguably, this depends on my knowing what reference and predication are and how they generate true propositions. That is, my knowledge of the meaning of the sentence “Tom is bald” incorporates an a priori component in addition to an a posteriori component. We could put this by saying that our knowledge of grammar is a priori, i.e., the combinatorial principles of sentence formation are known a priori.[1] When I know (empirically) the meaning of individual words, I know without further empirical input the meaning of phrases and sentences composed of those words. I know a priori that “Tom is bald” is grammatically correct and “Bald Tom is” is not, given knowledge of the conventional meaning of those words. Thus, linguistic knowledge has an a priori component, like geometrical and arithmetical knowledge. It isn’t entirely a posteriori. It may also be that this kind of knowledge is innate, as a matter of fact, but it is in any case an example of a priori knowledge. Is it knowledge of analytic truths? I won’t pronounce on this question, but that may also be so, pending an account of the scope of analytic truth and its epistemology. If we know what “noun” means and what “verb” means, then we know that noun-verb conjunctions produce sentences; there is an analytic necessity at work here.

We can now say something about the special character of linguistic knowledge (knowledge of meaning, in particular): it is a type of mixed knowledge. It has a base component that ranks as a priori, and it has a surface component that is a posteriori. When I speak and know that my words are meaningful, and know also whatthey mean, my knowledge has two components: an empirical component corresponding to certain contingent conventions of use, and an a priori component corresponding to the underlying grammar of language (a linguistic universal). As it were, I know the logic of my language a priori and I know its manifestation in speech a posteriori; my knowledge is a hybrid of disparate elements. Moreover, the a priori component is peculiar to language, not shared by other domains of a priori knowledge (geometry, arithmetic, ethics). It is knowledge of (universal) grammar. So, there is something unique about our knowledge of language; it has a sui generisinner complexity. In speaking and understanding, the mind is combining its a priori faculty and its a posteriorifaculty, to produce a special species of knowledge. Linguistic knowledge is a synthesis of these two types of knowledge. Thus, what is called linguistic competence is a synthesis of the a priori and the a posteriori. The work of Frege and the early Wittgenstein illustrates nicely how the element of a priori knowledge enters the picture (they didn’t think they were dealing in contingent empirical facts but rather in a priori necessary truths). Knowing a language isn’t just knowing empirical truths about that particular language.

Circling back to the four ways, we can draw some conclusions about the other three. First, the ontology of language: it must have an a priori structure, indeed a necessary structure, given that this structure is known a priori (just as Frege and early Wittgenstein maintained). Second, linguistic performance must stem from a dual competence: speech and understanding must be (partially) controlled by a priori and a posteriori knowledge working together. There is a kind of double causation at work (no doubt reflected in the relevant brain mechanisms). This is a far cry from old-fashioned stimulus-response psychology. Third, the phenomenology of linguistic activity will surely reflect its epistemic underpinnings; in particular, the phenomenology of a prioricognition will show up in consciousness. What it is like to speak and understand will bear the marks of this type of knowledge, as well as the more humdrum type of knowledge deployed in knowing sound-meaning associations; we will be conscious of the deep a priori knowledge involved in mastering a natural language grammar. In this respect, linguistic phenomenology belongs in the same camp as mathematical phenomenology (as well as ethical phenomenology, if we adopt a rationalist view of ethical knowledge). Perhaps this helps to explain the rather enigmatic and tantalizing nature of linguistic knowledge: it partakes of the puzzle of the a priori, going back to Plato and shaping the rationalist tradition in philosophy. Knowledge of language is not the most pellucid of things, which is why not much is said about it beyond the superficial.[2]

[1] I mean here the basic principles of universal grammar, not the idiosyncratic rules of particular languages, e.g., adjective placement or whether it is “good grammar” to split one’s infinitives. These latter are a posteriori.

[2] Knowledge of mathematics is notoriously problematic, as is knowledge of ethics. Knowledge of language, likewise, incorporates a problematic a priori component, and so inherits that source of puzzlement. How exactly do we come to know that nouns and verbs combine to form whole sentences? It isn’t like knowing that fish and chips go nicely together. It certainly doesn’t seem to be based on sensory experience, or to have a causal basis. It is, in some sense, “intuitive”. Rationalist epistemology may be true (I think it is) but it isn’t free of mystery.