There are obviously expressive vocalizations in both humans and animals: grunts, groans, sighs, moans, shrieks, laughs, barks, purrs, meows. Some of these vocalizations use the same sounds as spoken speech, such as the common utterance of Ach when something goes wrong; and some have a conventional element, as with Boo and Hurrah. But it is a further question whether such expressive vocalizations qualify as language. On the face of it they do not: these sounds don’t combine to form syntactic strings in the way words do—they are just self-standing vocal emissions. If we think of the human language faculty as defined by a lexicon and a grammar, then expressive vocalizations don’t count as part of the human language faculty. Their resemblance to language proper is confined to the fact that they are sounds produced by the vocal organs, but they don’t have the internal organization necessary to count as language (any more than purrs and barks do). Why such vocal acts evolved is an interesting question, but it is clear that they did not evolve as part of the human language faculty; they are not components of a system characterized by recursive syntax and a finite lexicon.  Also their use is limited to outer expression: there is no internal mental counterpart to such sounds, as there is an internal use of language—we don’t moan and groan silently to ourselves (“in the head”). Of course, we may complain to ourselves, but then we are using genuine language (“I hate going to the dentist”). Our ability to express ourselves vocally is really a separate psychological system from our ability to speak a human language—the two merely sound similar. To speak of expressive language in this connection is an oxymoron.
This has a bearing on certain philosophical projects and claims. Take the emotive theory of ethics: a moral statement is claimed to be equivalent to an expressive utterance like “Boo” or “Hurrah”.  But such utterances are not part of language, since they have no syntactic structure or lexical composition, unlike the statements they are said to paraphrase. The sentence “Charity is good” consists of words combined according to rules, but the utterance “Hurrah” is not (understood as an expressive performance). It is no more a piece of language than a sigh or a grunt is. That is indeed the point of the proposal: a moral statement is claimed to be nothing more than an expression of approval that lacks the structure of a typical statement. A so-called moral statement is not a statement at all but a mere vocal ejaculation expressing a positive emotion (analogous to clapping). But then how can the one be equivalent to the other—how can a piece of language be equivalent to a piece of non-language? How can something with the formal structure of a sentence be equivalent to a non-sentence? Sentences can paraphrase other sentences, but a non-sentence cannot paraphrase a sentence; it can at best replace it. What the emotivist has to contend is that a moral utterance is not an exercise of the language faculty: it does not consist of words arranged in syntactic combination—that is merely a superficial appearance. But such a contention is massively implausible: how could an utterance of “Charity is good” not be made up of words? The sentence contains words that can occur in other sentences, so are these not sentences either? Everything will go expressive if we follow this line. No, the ethical sentence is a real sentence, a part of language, while its putative paraphrase is not. But then the emotivist analysis has to be wrong. It only seemed plausible because both types of utterance involve vocalization and hence are language-like; but this is a poor guide to the status of an uttered sound—for clearly not all uttered sounds belong to language. More specifically, the cognitive system that outputs “Charity is good” (the language system) is not the same system as that which outputs “Boo” and “Hurrah” (the affective-expressive system).  A swooning sigh is not a sentence, so it can’t paraphrase something that is a sentence. The two have quite different formal properties. A sigh does not have the kind of lexically segmented digital structure characteristic of language.
The same point applies to attempts to explain sentences such as “I am in pain” expressively, as equivalent to groaning or some such. Groaning is not a linguistic act, though it is a vocal act: it doesn’t have the grammatical structure characteristic of language. We can say that an utterance of “I am in pain” can replace groaning (as Wittgenstein remarks), but it is a replacement by something quite new and unprecedented, since groaning is no kind of linguistic achievement (it predates human language completely). So it is not possible to have an expressive (“non-cognitive”) theory of these kinds of speech acts: they cannot be viewed as expressions in any literal sense. Of course, it is possible to express one’s emotions or sensations by talking about them (“I’m feeling very down today”), but that is not what expressive theories claim—they claim that certain verbal utterances are equivalent to vocalizations that lack linguistic structure. The trouble is that no utterance with sentence structure can be equivalent to an utterance without sentence structure; thus the expressivist is stuck claiming that moral utterances are not the utterance of sentences. Whatever emotions a person may have in uttering a moral sentence it cannot be that the utterance is an expression of those emotions in the intended sense—a mere voicing of an emotion, like sighing or humming. Moral language is part of the language faculty with its characteristic syntactic and semantic structure; it is not some vocal hinterland analogous to laughing or grunting or moaning. Strictly speaking, “expressive language” is an oxymoron (though “expressive vocalization” is not).
 I am alluding here to the kind of view of human language developed by Chomsky: a specific biologically based mental module constituted by finitely many discrete elements that combine into infinitely many hierarchically structured sentences.
 There are several doctrines that are often included under the heading “emotivism”; I am discussing just one extreme thesis, namely that moral utterances are equivalent to expressive vocalizations in the sense explained.
 It would be no use to argue that “Boo” and “Hurrah” just mean, “I disapprove of x” and “I approve of x”, since the whole point of the theory is to deny that ethical utterances are true or false. Such sentences are not expressions at all in the sense intended; they are simply statements of fact. It is essential to the theory that “Boo” and “Hurrah” be taken as not statement-like but as analogous to mere expressions of emotion like laughing or crying. And laughing and crying are not part of language: they are not types of speaking.