14.1 General problems
The Mind/Brain debate of the last few decades and its ancestors, like that on the relationship between Body and Soul, are always fascinating themes with which to work. The emotions are of considerable significance to these discussions, whether they are seen as undesirable manifestations of our closeness to the animals, or welcome signs of our superiority to the machine. They are part of our genetic and cultural heritage, and they are essential to the way we live our lives and contribute to the way our brains work. In an intellectual climate which no longer believes that our thought processes are as linear or as rational as was previously believed, hoped or idealized, the role of the emotions in these thought processes, instead of being relegated to the realm of topics which were so subjective as to be irrelevant, has become the subject of much research and argument. The emotions are now no longer seen as just intuitive, and sometimes illogical, reactions to the world around us. Instead, they are seen to involve complex processes in the brain that defy even the laws of physics as we now understand them.
Another debate that has dominated much of linguistic and philosophical discussion in this century has been the extent to which the phenomenon of human language evolves genetically or experientially. This debate also contemplates the role of language in the way in which our thought processes both condition and are conditioned by the language we use, not just in the here and now, but over periods of evolution and in different cultures. Psychologists have tried to understand the emotions by resorting to examples in language, and linguists and philosophers have turned to psychology to defend their theorization about the language used to describe emotion.
By attempting a multi-level form of linguistic I have tried to show how, even in the limited set of human utterances which form the corpora, certain patterns of linguistic usage emerge. These patterns would seem to both demonstrate how we express whatever reality is felt to exist, as well as suggest the factors, which because of such usage, also point to possible reasons why we conceptualize and, at an academic level, theorize, about the phenomena we call the emotions.
14.2 Linguistic indicators of the Universal / Relativist positions
It may seem to some linguists that I have done little more than go over the well-trodden ground of linguistic universals and language relativism, and come up with the rather obvious conclusion that neither extreme of these two poles of linguistic investigation can be examined independently of the other. However, I hope that, by using corpora of a certain size, whatever their imperfections, and by concentrating on data taken from language which has actually been used, rather than on data prepared with a view to describing how it can be used, I have at least come nearer to understanding the relevance of the Universalist / Relativist debate.
I would certainly make no claims that the data found in the corpora demonstrates any form of genetic universalism, but the cultural universalism to be found could supply indicators in the search for universalism at the level of human experience. Both languages share more similarities than they do differences, but there are differences, although they are often subtle.
One thing that I hope will be apparent from considering the tension between universalism and relativism between the languages, is that the holistic view I proposed to take of the language of Emotion in the Introduction - allowing for lexical, syntactic, and pragmatic analysis - has been justified. Semantic holism is only ever partial, since total holism would require total knowledge of a language, its historical and cultural contexts, and the way all individuals, including oneself, use that language in context. However, since, according to Fodor and Lepore (1992 : 32), “almost everyone [is now] a meaning holist” and “it is a widely-held view that much of the history of the philosophy of language consists of a failed attempt to make semantic atomism work”, perhaps an attempt at holism, even a very limited one like mine, is at least fashionable.
The degrees of relativity discussed here are rarely very strong, and the reasons for their existence can be attributed to a wide variety of factors. The pragmatic level of language would seem to suggest that different conventions exist as to which type of syntactic structure or lexeme is most frequently used in a specific language. This does not mean that, when they are considered in relation to each other, certain patterns cannot be seen to compensate for different patterns in the other language. However, some of these differences do indicate a genuine difference in semantic focus.
First of all, there is the relatively greater interest in Emotion lexemes shown by the PC, despite the chances that my relative inexperience in the Portuguese lexicon would produce a smaller set of examples. This may be due to a variety of factors, and it may merely mean that the authors of the the PC were more interested in describing emotion situations than those of the EC. However, it may mean that the PC authors preferred to use officially lexicalized Emotion words to describe the emotions, whereas those in the EC preferred the type of metaphorical reference to emotion which was not the object of this study, and which people like Lutz (1987: 292) suggest may be a particular feature of especially American English. An alternative hypothesis could be that the EC authors preferred to use the specific BEHAVIOUR vocabulary, for example laugh and cry , rather than lexemes of Joy or Distress. This would seem a little less likely, though, as the PC had rather fewer examples of BEHAVIOUR tagged examples with the Emotion lexemes. To insist on this point, therefore, would be simply to assert a greater preoccupation in observation of the Other for the EC authors.
Another point of difference between the corpora is the wider choice of vocabulary offered by the EC. This is a difficult question to discuss, not least because it touches on the sensitive point of the “richness” of a language. However, the most dedicated of Luso-philes should be able to accept the fact that English has been subjected to the influence of both Germanic and Romance vocabulary in a way that Portuguese has not. It is only natural, therefore, that English might have accumulated a wider choice of lexemes by combining words from both sources, particularly in such a subjective and universally relevant lexical field as that of Emotion. Perhaps the same Luso-philes will take comfort from the fact that the PC authors regularly showed a wider and more imaginative use of the vocabulary at their disposal.
When looking for syntactic relativity, there are two points which have to be considered: the options offered by the syntax of the different languages, and the actual patterns of usage of this syntax in the corpora. As we have seen, although both languages share a lot of similar syntactic rules, there are certain differences, the most significant of which would seem to be the differences in copula usage, with Portuguese offering estar, ser and ficar, where English makes do with be; the SE reflexive / passive pronoun, the ter + noun instead of be + adjective, and less obvious types of focus offered by differences in finite and non-finite clause usage between the languages. These differences would suggest that the prism through which this reality is viewed, such as the positioning of the Senser in relation to the emotional state and processes, allows for a rather different focus.
The tendency of the EC is to favour the SFoc verbs, predicative adjectives and, to a lesser extent, past participles, and that of the PC is to favour the nouns and past participles. The use of SFoc verbs is a clear and conscious affirmation of the SENSER's role in the emotional process and, since the large majority of such verbs are transitive, both SENSER and PHENOMENON are explicit with them. The use of the copula be with adjectives and past participles like afraid and happy would seem to point to an affirmation of the SENSER's emotional state, or even quality.
The PC not only favours fewer SFoc verbs, but a proportionately higher number of them are intransitive or -SE pronoun examples, which would seem to show a certain focus on the emotional processes, rather than on the SENSER / PHENOMENON relationship. This tendency is also reflected in the PC´s use of SFoc nouns. Although the nouns are sometimes complemented, they encourage a focus on the emotional process. The option of Portuguese to use ter + emotion noun rather than be + emotion adjective is just a more explicit separation of the SENSER and the process.
Although the variety of the copulas ser, estar and ficar with Portuguese allows for an interesting theoretical difference of focus between quality, state, and resulting state not easy to achieve simply by using be in English, the most interesting factor was the particularly frequent use of the zero copula in the PC. We can consider the emotional situation thus described as running parallel to and affecting some action by the SENSER, emphasising the close connection between the SENSER and the emotion, in rather the same way that using an attributive adjective next to a noun makes them close.
Another noticeable difference is the function of the -SE pronoun in Portuguese. The tiny number of similar examples in English allow at least for the possible existence of a certain reflexiveness in both languages. Whether this phenomenon should be interpreted as a passive, a reflexive or simply as a lexical oddity, will always depend partly on the immediate context in which it appears. However, I believe that, on certain occasions, it can be considered a reflexive, and that these cases reflect expressed conscious involvement of the Self in the emotional process.
Another peculiarity is the liking of English for 'c' type infinitive clause, or non-finite infinitive clauses in which the subject is different from that of the main clause, and usually performs the semantic role of PHENOMENON to the SENSER in the main clause. This situation is nearly always translated into Portuguese using a full -QUE clause. The English structure tends to emphasize the link between SENSER and PHENOMENON here, but the Portuguese one seems to focus the relation or process rather than the participants.
The PFoc area is larger with the PC and, although there are still more nouns in the PC than the the EC, the most noticeable difference is with the PFoc verbs which are more numerous in the PC. Since these verbs are the most active manifestation of the PHENOMENON's role, even if it is not deliberate, this would show that the increased focus by the PC on the PHENOMENON, already demonstrated by the higher number of PFoc items, is not accidental.
Generally speaking, it would seem, therefore, that the EC attributes rather more interest to the role of the SENSER and the SENSER's state than the PC. The PC, on the other hand, shows rather more interest in the emotional process, and slightly more in the PHENOMENON.
14.3 The Fragility of Linguistic metalanguage
This study also aimed at looking at the lingistic concepts with which we attempt to analyse language and to show how these concepts, too, are only useful insofar as they are relevant. After all, the way they carve up the linguistic world is as arbitrary as the way in which any other set of concepts carves up the world it refers to and, as theoretical constructs, they are only as good as the theory they represent.
The analysis of deep cases and semantic classification of verbs in relation to emotion may be very useful for language analysis, but both traditional grammar and contemporary psychology have influenced not only the names but also the semantic interpretation given to these classifications. Similarly, consideration of the syntactic behaviour of the Emotion lexicon suggests that the notions of stative / dynamic, the Progressive and the Imperative, the Passive - Active gradient, what constitutes a modal verb, the function of copulas, and other ‘dummy’ verbs, are just a few of the more obvious cases of the fragility of the meta-language we use to describe what we see, or want to see.
14.4 Language and the psychology of the emotions
As we have seen psychologists not only use language to discuss emotion, like everyone else, but they also try and use linguistic tests to try and prove something about the psychology of emotion. In the last few years they have come a long way from viewing emotion as irrational and involuntary, and from analysing it simply in terms of facial expression and physical reaction. There is no longer such a strong distinction made between emotion and mood, and neither are studied as separate from cognition, but rather as essential parts of the cognitive process. Thus we get articles with titles like 'Mood Affects Memory Because Feeling Are Cognitions' (Laird: 1991) and 'Neuropsychology and the Cognitive Nature of the Emotions' (Parrott & Schilkin: 1993) arguing that emotion and cognition are inseparable. Ledoux (1993) counters with the separatist argument, but when a heavyweight emotion theorist like Fridja (1993: 381) weighs in with a strong statement like "emotions in all instances involve a process of appraisal, and cognitive appraisal at that", one realizes that the defenders of the 'inseparable' point of view are definitely gaining the higher ground.
I hope I have shown how a re-evaluation of the linguistic meta-language used to describe the behaviour of the language of emotion is needed to accompany the changing concepts in psychology. Using the SENSER / PHENOMENON distinction, I have tried to show how, both theoretically and at the level of the evidence from the corpora, the PHENOMENON must be seen as dependent on the SENSER, and very rarely as acting independently. If one views the PHENOMENON in this way, it is difficult not to come to the conclusion that it results from some sort of cognition or mental process. The distinction made between verbs of perception, emotion and cognition is perfectly legitimate, and one can demonstrate degrees of difference between them, but the most interesting point that should be made is the extent to which they are similar, and how the the borders between the linguistic areas are decidedly fuzzy.
Perception focuses on the exterior object as it is perceived sensorially, but emotion and cognition combine in the appraisal process of this object in a way which then leads to some kind of action or statement. The most obvious indicator of this to be found in the corpora is the way the PHENOMENON is more often than not explicit at a linguistic level. However, as we have seen, the different emotions vary considerably as to the type of PHENOMENON they attract, and this, and the fact that they also vary in the extent to which they are SFoc or PFoc, draws further attention to the fact that some emotions are more “emotional” than others.
If one is looking for central type emotions which focus largely on the SENSER's state rather than cognitive processes, Joy and Distress would be one's first choices, followed by Fear and Anger. These four emotion groups are accepted by all emotion theorists, yet the linguistic evidence is that the strong emotions are often more thoroughly processed cognitively than the weaker moods. Simply to argue, Jamesian fashion, that the emotion is only identified in words after the event, is, as we have seen, no longer considered an acceptable explanation.
The Liking and Dislike groups pose several problems. Everyman tends to regard them as the most typical of emotions, whereas the theorists tend to classify them as less emotional and more cognitive. They are decidedly SFoc, but they nearly always demand an expressed PHENOMENON. Although they may be seen as the result of complex cognitive processes, these factors are often buried in the unconscious. Besides this, they seem to condition several of the other emotions, and must work in conjunction with or even precede them.
Hope is seen to be closely linked to the more intellectual side of the spectrum, and yet the status of Hope as an emotion is justified in part by the general rather undirected state of hoping. However, it would seem that some pragmatic notion of correctness or politeness interferes in the degrees to which English and Portuguese express a similar future-directed wish by means of hoping, expecting, wanting and liking, and the Portuguese words that can be used to translate them.
The remaining Emotion groups were all rather smaller, and most of them turned out to be cognitively complex, with fairly complex scenarios which included influence from social as well as more personal factors. The two lexical groups which Ortony et al. preferred to consider as variables rather than emotions, Surprise and Desire, should also be considered in any approach to emotion. Ortony et al.’s classification, although attractive, is not foolproof, and could probably be improved.
14.5 Will Artificial Intelligence ever experience emotion?
Eça de Queiroz, through Carlos da Maia, voices the belief that the curse of Satan is that he is unable to love, when he says Sou um impotente de sentimento, como Satanás ... Segundo os padres da Igreja, a grande tortura de Satanás é que não pode amar. There is also the age-old myth in our collective unconscious that shows lack of love as a curse, as when the monster, beast or frog in the fairy tale needs the love of the princess to regain his human status. Some would now argue that love is what distinguishes us, not from Satan, but from our robot. The question is - do we want our robot to be kissed by the fairy princess, or AI experts, and become human?
If there is now little doubt that emotion is inextricably bound up with cognition, it follows that if we aim to produce intelligent robots, whether or not they can be given consciousness, some attempt must be made to teach them something about the emotions. Although I believe it will be extremely difficult to produce a human-like robot, I hesitate to say such a thing will never exist. I remember reading Huxley's (1932) Brave New World thirty years ago and I was not alone in believing then that, although imaginative, it was still safely in the realms of science fiction. Nowadays, as one of the burning questions in the media is whether legislation should be brought in to prevent some entrepreneurial genetic scientist starting a Design-a-Child agency, I hesitate to say that future generations will never see a version of Lieutenant Data.
Therefore I prefer to say that once - rather than if - artificial intelligence can use language, there is no reason why it cannot be taught how to use emotion words correctly in the appropriate context. At the level of Searle's Chinese room type of non-consciousness, there is reason to believe that a robot will eventually be able to master the rules and regulations of language enough to be able to convince many people that it understands what it is saying. At this stage, the use of the words of emotion will constitute just another semantic field in the robot's vocabulary but, unless some bright programmer deliberately sets out to make the robot able to convince people that it has emotions, it is unlikely that its application of emotional language will go much beyond stating programmed information about people, and the social formulae which often employ this lexicon. However, whether the robot can ever become conscious depends very much on what we manage to find out about consciousness, and that problem is very much more complex than the linguistic one.
Let us, however, take a quantum leap and imagine that consciousness can be produced in our robots. What purpose will it serve to produce more than the elementary emotions needed to aid cognition in our robot? Presumably a suitable form of fear might aid its self-preservation. Perhaps liking for its controller and dislike for the controller's enemies might be an advantage, but surely, we could provide for it to build another version of itself without getting involved with love. It might also be flattering if it had pride in its controller's achievements, but not very desirable for it to be too proud of its own. However, if the emotions are the essence of consciousness, as many now believe, how conscious should we make our robot?
I use the word controller for a simple reason. Since it is highly unlikely that even the strongest AI enthusiast will want to produce a machine which will dominate its human creator, the obvious role for the future robot would seem to be that of servant to its human master / mistress. Now, since history has proved that the big problem of the master-slave relationship is that, to the chagrin of the dominant element, the subordinate element has opinions and feelings of its own, it would seem foolhardy, at the moment when the possibility of creating an ideal, non-human, conveniently intelligent, all-purpose, but docile and obedient slave finally exists, to then go and spoil it all by giving it those troublesome human characteristics of complex emotions, opinions and Self-hood.
Perhaps I am cynical, but I suggest that although we might teach our robot how to use emotion language when appropriate, and supply it with some form of elementary consciousness and emotions, for its own peace of mind we should limit its ability to ask too many questions about the meaning of words, life or whatever. If we decide to go any further along the road to consciousness, Design-a-Child would need to join forces with Design-a-Robot and become Design-a-Personality, an area where distinctions between human and robot would become difficult to establish. However, I am not over-worried we shall ever have to take these decisions - unless, in the meantime, we evolve into God.