The Language of Emotion
Proponents of the language of thought typically don’t have much to say about emotion. We are said to deploy an internal language when we think, but it is not suggested that we do so when we feel. Internal speech is characteristic of thought but not of emotion—we don’t “feel in words”. And the same might be said of desire: the idea of a “language of desire” has not met with enthusiastic acceptance (or even formulation). Language has to do with the cognitive part of the mind not the affective. Perhaps theorists think that the affective part of the mind is what we have in common with non-linguistic animals and so is not an appropriate object for linguistic explanation; only thought calls for linguistic representation. Emotion and desire are like bodily sensations: no one thinks that pain and pleasure should be analyzed linguistically—to be in pain is not to say to oneself “That hurts!” and the taste of pineapple is not an inward utterance of “Lo, pineapple”. Emotion just doesn’t have this kind of intellectual sophistication: it has no grammar or logic, no internal discursive structure. Emotions, like sensations, don’t entail each other or have subject-predicate structure. So it may be supposed.
But is that true? Take fear: we can fear that p as well as fearing x. For example, yesterday I feared that I would collide with a car that pulled out in front of me. Fear has propositional content: at the moment I slammed on the brakes I was afraid that I was about to have an accident. This is the same proposition that I believed to be true—in fact, I feared its truth because I believed it to be true. I thought that a collision was imminent and so I feared that a collision was imminent. We can recognize this connection between mental states without committing ourselves to a cognitive theory of emotion: it is simply a fact about our psychological economy. Of course, if emotions arethoughts (or essentially incorporate thoughts), then we can derive a language of emotion directly from a language of thought, but even without that assumption it is evident that emotions are (or can be) propositional. If emotions of fear have propositional content, then they have logical form, in virtue of the propositional object of the emotion. And so they have logical entailments—the content of my fear entailed, for example, that someone was about to have an accident. But then the case for a language of emotion is exactly as strong as the case for a language of thought, insofar as the latter case rests on the propositional content of thoughts. One of the main arguments for LOT is the productivity of thought, but emotions are also productive in this sense, since they invoke conceptually structured propositions—so we have the same argument for LOE. I can fear that I will not be selected for clemency just as I can believe that I will not be selected for clemency, and I can fear that I will be captured by the enemy and then tortured just as I can believe that conjunctive proposition. I can fear the same propositions that I can believe, including those built by logical operations like negation and conjunction. Thus emotions are logically structured, combinatorial, finitely based, and potentially infinite—just like beliefs and thoughts. If there is a LOT, then there must be a LOE.
It might be wondered whether emotion verbs accept every complement clause that cognitive verbs accept. Can we fear everything we can think? Can we feel sad about every state of affairs that we can believe to obtain? Can we be disgusted by everything to which we can assent? For example, I can believe that necessarily 2 + 2 = 4, but can I fear that necessarily 2 + 2 = 4? Can I feel sad that gravity obeys an inverse square law? Can I be disgusted that Hesperus is Phosphorous or that modus ponens is a valid rule of inference? With sufficient ingenuity we could probably contrive situations in which each of these peculiar emotions could be felt, though they are certainly not part of the normal run of things. But we don’t need to establish full correspondence between thought and emotion in order to recognize that emotions have an extraordinary variety of complex propositional objects, and that they therefore qualify for linguistic analysis given that thoughts do. Just as we think in a language, so we feel in a language—the content of our emotions has a linguistic underpinning. Other animals may not, just as other animals may think without deploying an internal language (possibly in images). But human emotions, like human thoughts, have a degree of conceptual sophistication that invites the idea of a LOE. Indeed, if we call the human LOT “Mentalese”, we can say that the LOE is also Mentalese: we feel in the same language in which we think. Why would we (or our genes) deploy two distinct languages for these two tasks? And if the propositional character of emotions derives from their cognitive component, we would expect that Mentalese would simply carry over to LOE. Thought and emotion would then share a common underlying symbolic system, with the same grammar and lexicon. 
The picture that results regards Mentalese as an internal language suitable for deployment in both thought and emotion (as well as desire, since we have complex logically related desires too). We might take it to be neutralbetween cognitive and affective uses, not privileging thought over emotion. It is not that we first have a language specifically of thought and then co-opt it to serve our emotions; rather, we have a neutral language that can be deployed for both thought and emotion. The Mentalese language faculty is a psychological module ready to be exploited by different parts of the mind—a general machine that can be used for different purposes. It doesn’t have thought built into it any more than it has emotion built into it; it’s more abstract than that. It is a language of mindgenerally (LOM). Thus LOM can be employed as an LOT or as an LOE. Some theorists might wish to go even further in divorcing LOM from thought specifically by suggesting that emotion and desire are primary in the mind. These theorists might maintain that desire and emotion precede thought in evolution, and that they require a symbolic medium in order to achieve their purposes optimally. Thus there was an LOE before there was an LOT: LOT is a later adaptation grounded in LOE. Maybe LOE evolved in fish long before anything deserving the name of thought arrived; then thought came along and recruited LOE for its purposes. There is no need to privilege the cognitive just because one adopts an internal language theory of mental operations. To put it differently, a computational model of mind is not committed to taking thought to be primary in the mind. Conceptually structured emotions (or desires) might be more basic than conceptually structured thoughts. Emotions are clearly important biologically, as well as being ancient, and having a sophisticated structure clearly aids their effectiveness. The affective is discursive.
When it was believed that thoughts consist of mental images the idea of a language of thought held little appeal; similarly for the theory that thoughts are behavioral dispositions. It took appreciation of the propositional nature of thoughts for LOT to gain traction—theorists had to accept that a thought is always a thought that p. Likewise, if we think of emotions as bodily sensations (as with many traditional theories), or as dispositions to behavior, then we will not appreciate their propositional nature. But once we accept that fear and sadness are fear and sadness that p, we are prepared to accept that emotions are underwritten by an internal symbolic system. The important move in both cases is accepting the correct logical analysis of ascriptions of thought and emotion. Once philosophers had grasped how reports of thought worked they were ready to take the plunge into LOT, but they don’t seem to have appreciated that emotion reports are much the same, so that a dip into LOE might be indicated too.
 There are also such attitudes as hope and trust: these are clearly propositional and close to belief and thought. If thought comes with an internal language, surely hope and trust do. But these attitudes have an emotional dimension, so we are already close to a language of emotion. In fact, the whole distinction between thought and emotion is quite artificial, so we should expect a general theory that subsumes both.
 I mean such things as referential opacity, the de re/de dicto distinction, the connection between entailment and logical form, the notions of sense and reference, semantic externalism, and so on. These are the things that encouraged philosophers to postulate a language underlying thought (Fodor needed Frege and Quine), but the same points apply to emotion and desire. The mind is thoroughly propositional, a subject of that-clauses.