The semantic classifications of verbs described in Chapter 5 were often based on syntactic criteria. As far as the Emotion lexicon is concerned, the interesting point was considered to be that the SFoc verbs do not normally occur with the progressive aspect and the imperative, and that this should be understood to demonstrate the non-voluntary or stative nature of these verbs. There are other aspects of the syntax of emotion verbs which are equally interesting, even if less has been said about them. They include the behaviour of these verbs with the Perfective and with the true Passive and its cousins, which include the Portuguese verb + -SE pronoun.

Let us begin by analysing why the SFoc verbs resist usage with the progressive aspect. These verbs, more often that not, fit into a simple present or past tense scheme such as:

in which the notion of time is defined in Quirk et al.(1985) as the State present/ past and which can be diagrammed thus:

in which no time span is stipulated, but the state exists insofar as it is relevant in the world referred to.

These examples, translated into Portuguese, show that the presente[1] is semantically equivalent under these conditions,

although, as we shall see, it can also be used with other notions of time. The examples in the past demand a further distinction in Portuguese, as will be explained below.

The progressive aspect in English is normally associated with activity and has the following time frames:

In the case of the present, and the past progressive in a) the notion of time is durative and the time adverb is unspecified, or a fairly elastic period, like last night. In b) the progressive aspect acts as a frame to a definite time reference - when John telephoned. This, at least, is true at the level of the immediate sentence. Regarded semantically, however, there is always a definite interval of time involved. However short or punctual, all notions of time from aeons to fractions of a second are, by definition, durative - including now.

Portuguese has a similar construction with estar a + infinitivo or estar + gerúndio, the former being more typical of European Portuguese, the latter of Brazilian. Ir and andar (conveying ideas of on-going action) are also used with this structure to describe durative action. One can construct similar examples to those given for English with it, and the notions of time are those given for English.

Generally speaking, it can be stated that SFoc verbs of Emotion cannot take the progressive aspect. It is unacceptable to say, *I am/was hating/fearing John, or*Estou a detestar/recear o João, or*Estou detestando /receando o João. However, according to Quirk et al. (1985: 4.29), in English it can be used in situations "when temporariness or tentativeness is being emphasized", as in:

The use of the progressive with the more central verbs is most unusual, but there are exceptions. If enjoy oneself can be classified as Emotion, it can also be durative, and as love sometimes has the sense of enjoy, it is possible to find:

The SFoc verbs of Emotion which seem to contain both the emotion and the behaviour associated with it, like gloat, grieve and admire, can be used progressively as in:

One can also find, particularly in a Brazilian context, Estou gostando de você, which can be roughly translated as I am beginning to like you very much..

The corpora do not produce many examples of the progressive. The EC produced 18 examples for the SFoc ones with admire, adore, hate, hope, mourn, look forward to, suffer and worry, and 6 for the PFoc verbs, with bore, bother and threaten. The PC had even fewer - 8 SFoc ones, with admirar, amainar, animar-se, gozar, recear and sofrer, and 2 PFoc ones with atrair. Most of the examples are explainable in terms of those already cited. Others are of a more marked variety, as in the following example where both the progressive form of hate and the nominalised -ING form of love are seen almost as activities - an interpretation made pragmatically possible by the phrase with all that energy:

(7.14) "When people hate with all that energy, it is something in themselves they are hating. Alex is hating all the illusions of boyhood -- innocence, God, hope. Poor Lady Marchmain has to bear all that. A woman has not all these ways of loving." BH

An example in Portuguese combines the idea of progressive with the verb ir (go), instead of estar, with an emotion -NDO form.

(7.15) Agora, ao aproximar-se da Toca, Ega ia receando o primeiro encontro com Maria Eduarda. M 519

Given the right context, therefore, the progressive can be used, but it is definitely unusual.

The same is true of the verbs usually connotated with intellectual activity, like think, believe, or agree, which are related to cognitive processes. To explain the fact that the progressive cannot be used with these verbs because they are static, and indicate states and not activities or performances as, for example, Kenny (1964: 171-8) argues, is a little misleading as to the nature of both emotional and intellectual verbs. The term state is connotated with a lack of mental activity which hardly corresponds to what is now known about the workings of the brain.

Halliday(1984: 5.3) prefers to classify these verbs as belonging to the "mental processes : processes of sensing", or as describing forms of behaviour associated with these processes, and points out that they do not function with substitute do. He argues too (ibid: 109) that the idea that the progressive present in English is somehow the more basic present is not necessarily justifiable simply because it is the unmarked form with most verbs, which are dynamic and describe actions. Instead, he prefers to consider the progressive aspect as "the present in the present", and suggests that, although it may be the unmarked type of present for such material processes, as in they are building a house, the simple present is the unmarked present for mental processes, as in she likes the gift. He points out that marked examples can be found in context to keep the balance.

Before leaving the progressive type construction, let us first account for the -ING and -NDO forms when they appear in non-finite clauses, because these are more numerous than full progressive forms. The EC has 37 SFoc examples which appear with admire, appreciate, despair, fear, hope, jubilate, like, love, object, pity, regret and worry, and 15 PFoc ones with exasperate, frighten, comfort and soothe. It also has 2 SFoc examples of being angry and 4 PFoc ones of being charming,. The PC has 40 SFoc ones with adorar, amar, apreciar, aspirar, desdenhar, desprezar, esperar, gostar, gozar, lamentar, odiar, recear, temer and triunfar, and 16 PFoc ones with alegrar, aliviar, amargurar, consolar, contentar, divertir, humilhar, incomodar and melancolizar .

As we can see in the following PFoc examples, the use of the -ING or -NDO clauses here does not imply the progressive aspect. They seem rather to be used for stylistic economy in long sentences, describing a sequence of events in the first example, and a sporadic state of affairs in the second:

(7.16) The white man had obtained it, I was told, partly by exercise of his wonderful strength and partly by cunning, from the ruler of a distant country, whence he had fled instantly, arriving at Patusan in utmost distress, but frightening the people by his extreme ferocity, which nothing seemed able to subdue. LJ

(7.17) Diante do canapé das senhoras lá se achava também o fiel amigo, o doutor delegado, grave e digno homem, que havia cinco anos andava ponderando e meditando o casamento com a Silveira viúva, sem se decidir - contentando-se em comprar todos os anos meia dúzia de lençóis, ou uma peça mais de bretanha, para arredondar o bragal. M 69

Some examples are marked because the sense of the emotion verb is different from the central meaning, as in:

(7.18) A questão estava simplesmente em que o Cohen o surpreendera amando-lhe a mulher. Logo, podia matá-lo, podia entregá-lo aos tribunais, podia escavacá-lo na sala a pontapés. M

Some of the SFoc examples combine emotion with the behaviour associated with it, allowing a certain idea of activity, as in these examples with admire and admirar:

(7.19) He stood there for some time, as if admiring the purity and the peace of the night. LJ

(7.20) E veio daí a um instante encontrar Teles da Gama admirando as belas faianças holandesas. M

In these cases the non-finite clause could be expanded to full clauses with He was admiring.... and Teles da Gama estava a admirar... (admirando in Brazilian Portuguese), and some kind of continuous contemplation, if not progressive action, is implied Other examples, though, cannot be explained so easily, as with these examples from the corpora with fear and its synonym in Portuguese, recear:

(7.21) Therefore, fearing he should be called upon to depose about this destroyed child, and so be the cause of her death, he hid himself. GE

(7.22) E ela, receando que a influência debilitante de Lisboa não conviesse a Charlie, estava com o vago projecto de lhe fazer ir passar algum tempo ao campo, em Formoselha, a casa da avó. M

Here the non-finite clauses have an adverbial status, and if one were to expand these -ING / -NDO forms to full clauses, one would end up with Because he feared that... and Como ela receava que.... and not with *Because he was fearing that.... or *Como ela estava a recear que..... In these cases the notion of fear or recear precedes and leads to the subsequent action, or projected action.

The choice of this kind of clause would seem to be a question of style, rather than any basically syntactic choice, and even a grammar like Quirk et al. (1984) has nothing specific to say on the subject. Halliday (1984: 209) describes the construction briefly as an "imperfective clause which is often, but not always, introduced by a preposition". This notion of the imperfective helps explain the point which we shall see more clearly below, and which is obscured by the use of the simple past, feared, in the expanded clause in English.

The perfective aspect in Romance languages tends to refer to actions which have been completed. However, in English this is not necessarily the case with the notorious present perfect, as in:

This means that Mary's love of John has lasted for 20 years so far, but unless the Speaker informs us to the contrary, we can presume she still loves him and will continue to love him.

Calling this a perfective is a misnomer if it is expected to coincide with the idea of perfeito in Portuguese, which implies that the period of time referred to by the verb has ended. This example would be translated with the simple present:

because the Portuguese presente accounts for situations in the past which are still valid in the present, whether they refer to a state or habit. To balance this and show its parallel with Halliday's notions of 'markedness', the unmarked form for a material process would be the estar + infinitive, as in Há 20 anos que estão a construir esta casa. The pretérito perfeito, unlike the English past tense, can only refer to actions completed in the past.

Portuguese makes a distinction between the pretérito imperfeito and the pretérito perfeito, which is not possible at the level of the English verb. This may seem a fine distinction to make, but it draws attention to a different facet of the situation which is not explicit in English - that the time span can be explicitly undefined, using the pretérito imperfeito, or explicitly defined, by using the pretérito perfeito and a suitable adverb. Quality and state examples which use the simple past in English will use the pretérito imperfeito in Portuguese, as in:

The pretérito imperfeito can also function in a way that would be translated by the past progressive in English as in the following example and its respective translation:

(7.25) Olhou para a esquerda e reparou pela primeira vez numa mulher alta, com cabelos compridos, que, sentada numa mesa a seu lado, o olhava de frente. A

(7.25) He looked to his left and noticed for the first time a tall woman with long hair who was sitting at an adjacent table and looking at him. A:Tr.

(7.26) E Afonso não se atrevia já a contrariar a pobre doente, tão virtuosa, e que o amava tanto! M

(7.26) And Afonso no longer dared to thwart the poor invalid, who was so virtuous and loved him so! M: Tr.

These examples do not make amava and olhava syntactically equivalent in Portuguese, for if the pretérito Imperfeito is substituted in each case by the estar a + infinitive construction - olhava can be changed to estava a olhar, because it is an activity or material process, but amava cannot be substituted by *estava a amar. The semantic difference in the notions of time expressed through the syntactic criteria between English and Portuguese is worth noticing, and support Halliday's perspective that the verbs others label stative are not merely defective dynamic verbs, but have a status of their own.

It is also considered impossible to use SFoc verbs with the imperative. As Kenny (1964: 183) says, "no static verb has an imperative". Although he is careful not to actually state that emotional and intellectual processes are involuntary, he claims that "voluntary action is action which can be commanded" and "only what can be commanded can be decided upon or form the immediate object of an intention". If one follows a strictly Jamesian or Behaviourist line of reasoning, not to mention certain ideas inherent in folk wisdom, this all goes to prove that emotions are beyond our volition, or at least are controlled only in the sub- or un-conscious.

The disallowance of the imperative would seem to be a valid syntactic rule for both English and Portuguese. We do not normally say things like:

but to call these phrases commands is strong. As Bolinger (1977) demonstrated, using context and intonation, what is syntactically an imperative can fulfil a variety of functions semantically. Quirk et al. (1972) discusses commands with the imperative structure, and concedes a persuasive imperative with do, but Quirk et al. (1985) provides fifteen different ways in which the imperative can be interpreted. Thus, one can describe the Biblical exhortation to Love thy neighbour as either a command, a recommendation or a special lexical variation of love; saying Enjoy yourself! functions as Good Wishes; and the Have pity! is a plea, and certainly not a command.

The situation with the negative imperative is different and more complex. Normally the imperative precedes the obeying, or otherwise, of the command. With the negative form one can be asking or persuading someone to stop something they are doing, or change the state they are already in, as in:

Admittedly, this is not a popular construction, but the function is persuasive, and, it is perfectly acceptable, if not very frequent, with the adjectives/past participles of Emotion in English, and other structures in Portuguese, as in:

Of the 40 examples of the Imperative in the EC, 29 were negatives, and 22 of these were Don't worry! Of the others, some were exhortative - like Cheer up! and Be grateful! - or were commanding not the emotion but the behaviour associated with it, as in Look angry! The only example of commanding love, repeated several times, was by the eccentric Miss Havisham in Great Expectations in her attempts to make poor Pip even more in love with Estella - hardly a normal example, but worth noting as an example of how the context can make such as structure acceptable.

The PC produced even fewer examples - 26 in all - and none of them were negative imperatives, although the structure is possible, Don't worry! being translatable by Não se aflige! Interestingly, the pragmatic equivalent to Don't worry! seemed to be more normally expressed in the PC original texts as Calma! and Sossega! and these forms accounted for 20 of the examples. Also whereas the EC examples had 9 examples of be + adjective, there was only one example, with an PFoc adjective, in the PC, when, in Os Maias, Ega, off to a party, is exhorted teasingly by his friends - Sê fascinante! (Be fascinating!)

If one follows the argument that all this shows that emotion is not voluntary, one could argue here that one cannot persuade someone out of a true emotion. However, by the time someone is being persuaded to abandon a certain behaviour resulting from an emotion, the whole situation is largely in the realm of conscious control anyhow.

Before leaving this subject, in which those supporting the involuntary aspect of emotion, may seem to have gained the upper ground, I should like to make a suggestion. Are we not perhaps being over-hasty in assuming that since a person cannot command emotions in another, they cannot, like activities, performances or material processes, be considered voluntary? Although Halliday (1984) does not explicitly make this point, perhaps one could continue his line of reasoning on the unmarked and marked usage of the progressive a step further, and simply assert that with unmarked mental processes imperatives are not used, whereas with material processes they are. This would maintain the balance already established for the progressive.

If one moves away from syntactic linguistics to a more pragmatic level, however, one could argue that, at least in the type of society in which English and Portuguese are spoken, attempting to command someone else's intellectual or emotional processes is not acceptable culturally, whatever political or religious régime one lives under. Perhaps there is some deep-rooted kind of Politeness Principle, of the type discussed by Leech (1985: Ch. 4), which may explain the situation. As already stated, the few examples quoted are of the persuasive or polite variety, and one could argue that Miss Havisham's Love her! does not go against any possibility principle - she already guesses that Pip loves Estella - but rather that she is deliberately violating his personal psychological integrity and asserting her own authority by commanding such a thing.

Psychiatrists recognize that it is more effective to encourage patients to control their reactions, than to attempt to exercise some sort of outside authority over them. Although most of us are aware of how demagogues, and nowadays the media, attempt to control our thoughts and emotions, there is some deep-rooted objection to this that obliges even the most dictatorial of régimes to attempt to disguise what it is doing. So the existence of some sort of Politeness Principle that disallows the imperative may well be a better way of explaining the problem than by dragging in the traditional philosophical hot potato of Free Will.

All the arguments about the problems with the progressive and the imperative have centred on the SFoc verbs, and no-one seems to have suggested that the PFoc ones are subject to similar restrictions. One can immediately point to the syntactic acceptablility of He is irritating me or Frighten him! However, it should not go unnoticed that, in actual usage, PFoc examples with the progressive and imperative would seem to occur even less frequently than those with SFoc ones. The EC shows that there is only 1 example of a PFoc imperative, and 6 PFoc examples of the progressive. Even if one takes into consideration that the SFoc verbs are more numerous than the PFoc ones, 19,2% against 5,1%, one is still left with the SFoc imperatives vastly outnumbering the PFoc ones and the progressive only having proportionally slightly more PFoc examples. In the PC, where the proportion of SFoc to PFoc verbs is more similar, 14,2% against 9,6%, the larger proportion of SFoc examples is even more marked.

Here again one can demonstrate the internal nature of the mental process and the lack of deliberate action on the part of the PHENOMENON. Examples in the corpora were marked DELIBERATE where the PHENOMENON explicitly tried to provoke a reaction - 13 in the EC and 14 in the PC. Even though we can say John is irritating me without John's being aware of the fact, it would seem that we do not even make this sort of claim very often. We seem to prefer to water it down and turn irritating into an adjective instead, preferably an attributive one. With imperatives, PFoc examples are particularly few, perhaps because they involve a third element, the person issuing the command, which makes the whole situation very complex.

An area of syntax which is not mentioned very frequently in the discussion on emotion verbs is their relationship to those verbs which we designate as auxiliary or modal in English. Although these auxiliary and modal verbs by no means translate easily between English and Portuguese, particularly at a syntactic level, at a more semantic level, certain similarities exist.

Quirk et al. (1985: 3.31-8) present the 'primary' verbs, be, have and do and discuss their functions both as lexical and auxiliary verbs. As far as Emotion is concerned, both SFoc and PFoc Emotion verbs in English use the auxiliary do quite normally to form negatives and questions. However, the use of both be and have with the SFoc verbs to form the progressive and perfective aspects is, as we have already seen, restricted to the very few situations in which they occur. Portuguese does not have an auxiliary like do, but estar a and ter are restricted in the formation of the progressive and perfective aspects with the SFoc verbs for the same reasons as their opposite numbers in English. The use of be and ser to form the passive will be discussed below.

Halliday points out that SFoc verbs reject the pro-verbs do and fazer[3], because one cannot ask the question What did you do? or O que fizeste?, and get the answer *I feared / hated John, or *Receei / detestei o João. Again, this is a question of verb type. Do can only be used like this with dynamic or activity verbs, although those Emotion verbs which can be interpreted as behaviour may occasionally appear with them, as in:

The use of modal verbs with SFoc verbs is also very restricted, because of their specific semantic content. Modern grammars are rather less sure about what constitutes a modal verb than they used to be, a point which Quirk et al.(1985) develops in some detail for English, and which Pontes (1973) develops into a book for Portuguese.

Quirk et al. (1985: 3.40) define a gradient from 'central' and 'marginal' modals, through 'modal idioms', 'semi-auxiliaries' and 'catenatives', to full main verbs[4]. They describe how the modals behave syntactically, giving various different syntactic tests to show how they function on this gradient, using can, which they consider a full modal, in contrast with hope, which they define as a full verb.

Interestingly, few of these tests could be used in a similar analysis of Portuguese of poder and esperar, either because they depend on the use or otherwise of do as an auxiliary, or on conventions of ellipsis which function differently in the two languages. This is also partly due to the fact that whereas can, and the other central modals, are defective and limited to one or two forms, which cannot even be described as present and past versions of each other, poder can function as a normal verb in Portuguese - although it disallows the imperative, as is pointed out by Cunha and Cintra (1984: 428)[5]. Pontes (1973: Chapter 4) shows how certain uses of poder can be considered as modal, and how dever, ter/haver que, and precisar, verbs which serve to translate the notions expressed by the various types of modals in English, and ir, costumar, começar, continuar, por-se a, tornar a, voltar a and acabar, which cover several of the meanings implicit in Quirk et al.'s semi-auxiliaries and catenatives, can be analysed in similar terms.

Apart from describing those verbs normally considered as auxiliaries or modals, Pontes (1973: Chapter 3) also analyses, and rejects, the proposal by Said Ali (1963) to include as modal verbs certain 'causative' and 'sensitive' verbs. The interesting point is that the sensitive verbs include several of the SFoc verbs considered here, such as querer, desejar and odiar. If one takes this point of view together with Quirk et al.'s choice of hope + infinitive complementation as the end of the modal > main verb gradient, it is not unreasonable to suggest that some of the SFoc verbs do in fact come very close to being modal.

Quirk et al. (1985: 4.49-66) describe the semantic uses of the various modal type verbs and how they tend to vary from one variety of English to another. The modals seem to cover the semantic gradients between:

These notions do not easily agree semantically with emotion, for reasons already discussed. Even a simple future prediction, although possible, is extremely rare. It is difficult to imagine contexts for examples such as:

(7.34) He *can/*may(permission) / *shall(obligation) / *must (obligation) love /hate /fear Mary.

In these examples, however, must and can't do not refer to obligation or ability to love. They express instead the Speaker's conviction about the 'epistemic necessity' of what is being said, as explained by Quirk at al. (1985: 4.54).

There are a few modals which do appear with SFoc verbs. For example, the marginal modal used to, expressing a habit or state in the past, can be used with these verbs provided a state and not a habit interpretation is assumed. A hypothetical use of would is often used with like, love, prefer, "to indicate a tentative desire in polite requests, offers or invitations" ( Quirk at al. (1985: 3.64 n.(c)):

Dare is related to fear when used in the negative form but means rather not to have the courage than to be afraid. Catenatives like appear to and seem to, also appear with SFoc verbs, describing the Speaker's impression of someone else's emotion, which puts the situation at one remove from one's personal attitude. However, these are minor examples, which avoid the more central meanings of the modal verbs.

In Portuguese, although, as we have seen, it is even less easy to make syntactic distinctions between verbs on this modal > main verb gradient, the semantics of these verbs offer similar possibilities to the English verbs, and for that reason their usage with SFoc verbs is also limited. Examples of epistemic necessity can be found:

The hypothetical use of would, however, is a function of the presente, futuro do perfeito, or pretérito imperfeito, depending partly on the degree of politeness involved.

Costumar translates the used to modal but, as it functions normally with tense, it also translates English phrases with the present + usually, which probably accounts for Portuguese students' confusion of the English constructions. However, it is limited to habitual situations and does not appear with SFoc verbs. Besides, in Portuguese, the imperfeito is possible with these verbs, thus making a modal unnecessary. Parecer, like appear or seem, could be classified as catenative.

The stative quality of SFoc verbs would, for various reasons, therefore, seem to be emphasized by their incompatibility with the semantic notions expressed by the modal verbs. The few appearances they make together seem to fit into either the epistemic necessity, or tentatively polite categories.

Before leaving this subject, I should like to speculate a little about this notion of the SFoc verbs as close to modals or quasi-modals, despite their apparent status as normal lexical verbs. The semantic gradients proposed for modals above have been formulated very much on the basis of what English-speaking grammarians, including Chomsky and his followers, say, and they tend to be attracted to the subject by the irregularity of the syntax of these modals in English. Pontes (1973) quotes these authorities in defence of a position which rejects the idea that the sensitive verbs are modals, despite the infinitive construction which is common to both the verbs which she agrees are modals, and the sensitive auxiliaries which she rejects as such. However, as she points out in her first chapter, other Portuguese grammarians have toyed with the idea that there is something special about these verb + infinitive structures, although few risk declaring them real auxiliaries.

Naturally, one of the reasons why modal verbs are of interest semantically is that they seem to represent some sort of basic or primitive mood or attitude at the level of syntax. To enlarge or diminish the number of these moods or attitudes is therefore subject to the wider and deeper notions of philosophy and psychology. However, to limit the notions of ability, capability, volition, intention and others to those which are expressed through the small number of defective modal verbs in English, may seem sensible and economical, but one could call it linguistically chauvinistic and restrictive.

It is not my intention to rush in where experts fear to tread, and I would not argue for a strong hypothesis that verbs of Emotion + infinitive clauses in English and Portuguese are quasi-modals. All I would like to suggest, given the tendency nowadays to accept gradients rather than clear-cut distinctions in language, and given the syntactic arguments which I shall now put forward, is that a weaker hypothesis is acceptable - that in certain circumstances, these verbs should be seen as the continuance of the gradient of modality a little further, until it reaches a position on the gradient marked for behaviour rather than attitude.

As I have already pointed out, these verbs have syntactic features in common with other modals. Apart from do, in English, they cannot combine with the other modals in their central semantic senses. They also occur with infinitive clauses in which the understood Subject in the subordinate clause is the same as that of the main clause. Quirk et al. (1984: 16.38) list the verbs that accept this sort of complementation according to semantic category and most of them are of the type we are discussing, or are from the closely related classes of intellectual verbs, like intend or plan, or intellectual performatives, like ask, offer or promise. Significantly, some of the others are of the begin, continue and cease type whose Portuguese equivalents, like andar and continuar, are considered by Cunha and Cintra (1984: 380-2) as a means of conveying aspect, and which Mateus et al. (1983: 142) describe as auxiliares aspectuais or aspect auxiliaries. The remainder are nearly all synonyms of the above mentioned categories, or of the verb try , which also has a trace of volition in its semantics.

The semantic categories proposed by Quirk et al. are not labelled and they are ordered on a syntactic basis - according to which accept other forms of complementation - rather than in relation to any particular semantic criteria. They could, however, be classified in a different order to show a progression from emotion > mental process > action[6]. As one moves away from the emotion end of the gradient, the possibility or permission modals become increasingly acceptable. Interestingly enough, the begin type verbs, similar to the Portuguese aspect auxiliaries, can accept these modals quite easily.

A few examples of the infinitive structure were found with a number of these verbs, in both the EC and the PC, but the percentages are only really significant with the Desire, Liking and Hope groups, all of which are strongly related to volition, particularly when this syntactic structure is used.

Desire would certainly stand consideration as having a modal status of the volition type as one of its most important functions. In the EC, the main verbs are want and wish, and 47,9% of all the examples recorded take an infinitive construction; in the PC, the verb querer is the most important with 60,1% of the examples taking an infinitive. Quirk et al. (1984: 3.51) actually suggest that want in the present tense, as in I want, is "syntactically different from a main verb" because the meaning changes if the past tense or progressive aspect is used. They choose to discuss it as a 'pragmatic particle', and claim that with I, it "introduces an expression of wish", and implies an attitude of volition.

Hope, which Quirk et al. consider to be at the end of their auxiliary > main verb gradient, and which can be seen to mean some sort of volition towards the future, has 4,6%(E) of examples with infinitives in the EC, and 11,2%(P) for esperar in the PC, which could be seen in a similar light to want. I hope usually expresses a similar attitude of volition.

The Liking verbs with this structure can be seen, if not as volition, at least as some sort of positive, or willing, attitude towards whatever process is implied by the infinitive following them. The examples in the PC were few, 4,2%, but in the EC 18,2% makes these verbs, or at least like, which accounts for 16,4%, a strong contender for quasi-modal, or pragmatic particle status. The use of would like + infinitive is very close pragmatically to the modal idioms would rather/sooner.

One can, therefore, point to a certain syntactic connection between the lexical verbs of Emotion, and their close relative want, and the modals. The verbs in the Desire, Liking and Hope groups usually define personal attitudes, moods or dispositions, and, as we shall see later on, these are among the more consciously processed emotions.

Another point that can be made about the syntax of SFoc verbs is that they do not like passivization. This point comes across when one observes the tiny or non-existent percentages of passives involved with these verbs in both the EC and the PC. Hope and fear produce a small but noticeable number of passives in the Birmingham Corpus, owing to impersonal constructions like It is hoped/feared that. These structures are found in official letters or informative documents and so they do not appear in my corpus of literary texts. The passive of love appeared only about 15 times in the whole Birmingham Corpus (BC), which had 1420 examples of the form loved alone, and a total of 8642 examples for love. These examples are rather unusual. Apart from examples like:

where the SENSERS are several in relation to a PHENOMENON which is also the Theme, they refer to love in general, rather than in particular, as in:

Others have noticed this point and yet, influenced perhaps by early training in reciting verbs based on the Latin amare, Quirk et al. (1985: 3.70) use the example Mary was loved by John as an example. I have yet to find an example like this outside books on grammar and linguistics, where, nevertheless, it continues to be a favourite. Bresnan (1982: 6) uses Fred is loved by Mary as an opening example in the 'The Passive in Lexical theory', and in Bresnan (1976: 41) she has the complicated John wants to appear to be loved by Mary. Even Halliday (1984), when arguing the problem of whether The gift pleased Mary is the passive of Mary liked the gift, says that the passive sentence The gift was liked by Mary, "although a much less common type, undoubtedly exists also".

I am not claiming this type of structure does not, or cannot, exist. The BC also produced about 17 examples of the passive among 1746 occurrences of liked. Besides, so many eminent linguists, speaking as ideal native speakers, feel that the structure is acceptable, so it must be. I merely want to stress the fact that it is much rarer than they seem to realise. Halliday was arguing against attempts by Lakoff, Chomsky and others to make like a sort of passive form of please, a point on which I agree with him, but why, one asks, could those discussing the passive in other circumstances not have used some more obviously agentive verb, like kiss or kick, instead? For the purposes of the examples quoted, it would not have made any difference.

If the rejection of passivization were confined to just one lexeme or another, it might be a problem of the lexeme, but in the analysis made of the BC and the other corpora, whole groups of Emotion lexemes - Satisfaction, Relief, Pride, Self-Reproach, Gratitude, Gratification, Remorse, and Surprise - produce not one example of the passive, and two other quantitatively important groups, like Joy and Distress, only manage to produce a tiny number of rather doubtful examples. To quote examples of the passive using these verbs, therefore, is to be unforgivably unrepresentative

In my own English and Portuguese corpora, where not only more obvious passive structures, but also every possible example of an PFoc past participle form in this lexical area was analysed, these examples only accounted for 0,6%(E) and 1,3%(P) of the total examples in the corpora. Although such a tiny number, it is noticeable that this still leaves the PC having proportionally more than twice as many examples as the EC, but, as we shall see, this use of the past participle in Portuguese is sometimes different. There are 12 attributive and 40 predicative examples in the EC, and over three quarters come from the Appreciation and Liking groups, with examples like admired, loved, respected and favoured, or from groups allowing a negative appreciation, with examples like detested, dreaded and despised. The only example which did not imply either multiple or unknown SENSERS was the following one with despise:

(7.45) I began to consider whether I was not more naturally and wholesomely situated, after all, in these circumstances, than playing beggar my neighbour by candlelight in the room with the stopped clocks, and being despised by Estella. GE

and this is part of an -ING clause which, as we have seen, favour unusual structures.

The situation with past participles formed from SFoc verbs is by no means as straightforward in the PC. Although the majority of examples come from much the same lexical groups, and follow a similar pattern to the EC, as in:

(7.46) Às vezes pergunto-me se todos temos mesmo que fazer esta via-sacra porque a verdade é que o dinheiro compensa muito os que não amam nem são amados, os que não acreditam em si... SU

there were more examples in which an individual SENSER was explicitly described, as in:

There are also several examples in which o meu amado/a minha amada have a Vocative function, as in:

(7.48) Tu, minha amada, és o inconsciente; amas as coisas pelas coisas; eu amo-as por mim. AQ

In English, this could be translated as my beloved, which is used very rarely, and the attributive use of amado, as in:

(7.49) Raras vezes se pronuncia a palavra revolução como raras vezes se pronuncia o nome da mulher amada, no café, em frente de estranhos. A

would also be rendered by beloved, of which there are 4 examples in the EC. However, use of the participle form of these verbs attributively is rare in the EC, the very few examples in the corpus coming in phrases like a wealthy and respected merchant or a dreaded enemy and all but one - the cherished glamour - refer to human, or human type, PHENOMENA, which are the subjects of the emotion described. In this way, the PFoc past participle typically refers to the accepted fact that the PHENOMENON is detested, respected, or beloved, rather than to any emotion-causing quality it possesses. For this reason, these participles are quite distinct from the PFoc adjectives which typically describe just these qualities.

In the PC, however, there are about seven times as many examples. Several are of the attributive type which projects the SENSER's feeling onto something else, as in:

(7.50) O povo e nós próprios já considerávamos o carbúnculo um acidente desprezível; mas alguns haviam-me custado noites atormentadas. N

but there are examples of past participles like desconsolado and divertido, which appear describing the quality of the PHENOMENON, as in:

(7.51) Não era divertido assistir em silêncio, do fundo de uma poltrona, às infindáveis discussõs de Carlos e de Craft sobre arte e sobre ciência. M

These examples are nearer the PFoc adjectives than past participles, and this example could only be translated by amusing and never amused.. However, it is interesting that they can also be used as SFoc past participles, translatable by amused, as in:

(7.52) Carlos pulava nos joelhos do avô, muito divertido com aqueles longos abraços que juntavam as duas cabeças dos velhos. M

The distinction between adjective and past participle in Portuguese would therefore seem to rather fuzzier than in English, but the more passive-like examples of the PFoc participles would seem to follow similar pragmatic rules in both languages. The copula used in the EC is be, and the vast majority of the PC ones used either ser or a zero copula. The unknown nature of the SENSER(s) is underlined by the fact that complementation by phrases such as by X are only found with a tiny minority in both languages.

The semantic reasons why the SFoc verbs reject passivization are not difficult to find. It would seem fairly obvious that, if the Subject of these verbs is not an Agent, but a SENSER or EXPERIENCER, they are hardly likely to accept passivization, which usually requires some kind of AGENT even if it is neither known nor mentioned. It also should not be forgotten that the status of the PHENOMENON depends on the SENSER, as explained above. If the nature of the PHENOMENON is essentially internal to the SENSER, it is psychologically difficult to make the PHENOMENON an independent reality which acts as a Theme[7], or in the normal Subject position in the sentence. This is emphasized by the fact that the avoidance of the passive by the SFoc verbs is balanced by a high proportion of passive-type examples with the PFoc ones, as we shall see below.

Perhaps it is precisely the fact that so many well-known linguists have argued the please/like dichotomy backwards and forwards for so long that has made people shy away from this problem. Passivization has raised all kinds of problems for the different grammatical theories, but more has been said about the syntax than the pragmatic or psychological reasons for the passive.

Although the passive is conspicuous by its absence with emotion verbs, similar structures with adjectives and past participles of emotion, and which some see as being on a Passive > Active gradient, are quite important. One needs to consider the morphology and syntax of adjectives in order to understand this better.

Although certain adjectives in both languages are recognizable from the suffixes, like -able or -ável, -ive and -ivo, which are added to verb or noun stems, many, like happy, old or good are not immediately identifiable from their morphology. One of the problems with adjectives is whether they derive from verbs or nouns, or whether verbs and nouns derive from them. Another linguistic problem is whether adjectives are simply surface manifestations of deeper verbs or nouns. Of course, the validity of these arguments will depend very much on the adjectives one chooses to prove one's point. With Emotion there are plenty of adjective/noun pairs with no verb form, like happy /happiness, furious/fury, alegre/alegria and fúria/furioso which would favour an adjective/noun relationship. With some SFoc verbs like love and fear in English, it is not apparent from the form which comes first, the verb or the noun, although adjectives like lovable seem to derive from them. PFoc verbs often seem to be the nearest to a stem, as with irritate > irritation, irritated, irritating and irritable, and its cognate in Portuguese, irritar > irritação, irritado, irritante and irritável.

In the case of Emotion, a large number of adjectives seem, at least morphologically, to resemble participles, as with annoyed/annoying, and and the verb annoy. There are some past participle forms, like anguished, aggrieved, disgruntled and dissatisfied, which seem to function without any real corresponding verb form. Used predicatively, participles can appear to be ambiguous, it being unclear syntactically if they are being used as the adjectival form or with the progressive aspect, unless an Object is present, as in:

although the agreement between the subject and the past participle points to an adjectival interpretation, However, a present participle like alarmando is easily distinguishable from its adjective form, alarmante.

Like the verbs, the adjectives related to Emotion can be classified as SFoc or PFoc. SFoc verbs like love are related to SFoc adjectives like loving and to PFoc adjectives/participles like lovable and loved. PFoc verbs like irritate are connected with SFoc adjectives like irritated and irritable, and with PFoc adjectives such as irritating. The participle type adjectives would seem to follow a distinct pattern according to which verb type they derive from. Thus, -ing and -nte adjectives deriving from SFoc verbs will be SFoc, and -ed/-do and -ble/-vel adjectives will be PFoc. -Ing/-nte adjectives deriving from PFoc verbs will be PFoc, and -ed/-do and -ble/ -vel adjectives will be SFoc. Other verb-related adjectives, like envious and resentful are usually SFoc, whereas delightful and offensive are PFoc, but it is not so easy to predict how the suffix will influence the focus, as a study of hateful, usually PFoc, will show.

Those adjectives which are not related to verbs are less easy to analyse, particularly if they are used attributively. They can be quite clearly SFoc, as with jealous, anxious and furious, or ciumento, ansioso and furioso. Some English adjectives, like afraid and aghast, can only be used predicatively, and are always SFoc. However, there are others, like happy and sad, or feliz and triste, which can be both SFoc and PFoc.

There would seem to be a gradient for adjectives, therefore, which would stretch from those related to nouns only, like happy or sad, through those which can be related to both nouns and verbs like irritable and lovable, to those like alarmed and disgusting, whose relationship to the verb is self-evident. A similar gradient can be shown in Portuguese, with the exception of the present participle type adjective. As we shall see, this gradient forms part of another which Quirk et al. (1985) call the passive gradient. The more sophisticated linguistic arguments, however, go well beyond morphology.

The adjective is traditionally regarded as modifying the noun in some way, and in Portuguese the bond between noun and adjective is strengthened by the fact that they agree in number and gender, whether used attributively or predicatively. The position of adjectives is what usually identifies them, and in English one can distinguish between attributive and predicative adjectives, those that usually occur before the noun described and those that act as complements to a copula, usually be. The relationship between attributive and predicative uses of adjectives is close and it can usually be assumed that if we talk about the beautiful girl, we are usually prepared to state that the girl is beautiful. For this reason transformational grammar, as Wardhaugh (1977: 120-1) explains, liked to show how the beautiful girl derives from the deeper structure the girl is beautiful, using a BE-deletion transformation to give us *the girl beautiful, followed by an adjective movement one to make the adjective precede the noun - a transformation that was not necessary for languages like Portuguese. Nowadays more attention is given to the pragmatic reasons why one structure is used in preference to the other.

Halliday (1984) prefers to call attributive adjectives 'modifiers' within and at the level of the nominal group in the clause, and to classify the copula + predicative adjective as a process of being, one of the relational processes to be found at the level of the clause. The modifiers are of various kinds, which I need not describe here, and when, as often happens in English, there is more than one of them, there are certain definable rules as to the order in which they appear.

Halliday's classification of the predicative function as a relational process is in line with the assumption by many linguists that adjectives occurring in a predicative position are usually the equivalent of stative verbs. It follows from this, therefore, that stative verbs, or mental processes, and copulas + adjectives, or certain relational processes, must be semantically close.

In Portuguese the same distinction can be made between attributive and predicative adjectives, and the rules for the copula + predicative adjective constructions, and their relationship with verbs, are very similar to those for English. However, it is not so easy to compare the usage of attributive adjectives, as Portuguese normally places them after the noun, and, unlike English, dislikes accumulating a large number of adjectives to describe one noun. This means that certain adjectives, particularly those which look like nouns in English, like stone wall, are expressed by other structures, like prepositional phrases, as in parede de pedra.

As has already been mentioned in Chapter 6, although the grammatical rules for the behaviour of Portuguese predicative adjectives are very similar to those for English, pragmatically, and particularly with the SFoc adjectives, there is a significant difference between actual usage in the two languages. When one examines the behaviour of adjectives in actual texts in both languages, one notices other differences in their behaviour.

In English an adjective is usually securely tied to the noun it modifies, either by proximity, usually immediately preceding it as when it is attributive, or by a connecting copula. The rules of agreement of adjectives with nouns in Portuguese, on the other hand, allow for far greater flexibility of sentence structure. This can be seen if one examines the following sentence:

(7.56) Tirei-a depois com ostentação dos dedos engelhados da comadre, lavei-a com carinho, feliz, alvoroçado. N

which describes the doctor, the I or eu implicit in the first person use of the verb, who has just delivered a baby or criança - understandable from the feminine pronoun a after tirei - with the help of the comadre or village midwife. Despite the fact that three people are involved, - two defined by feminine gender - the apparent isolation of the adjective feliz, and the fact that it is an adjective which does not have masculine/feminine agreement, one can deduce that it describes the (male) doctor. This is partly because it appears to be related to the adjectival past participle alvoroçado, which shows masculine agreement, but also through conventions of conjunction and ellipsis in Portuguese which allows a conjunction and a copula like estava, to disappear. An extended version of this sentence would give something like" ..... carinho, e eu estava feliz...". Any attempt to reproduce the same structure in English without a copula would give, at best, a very

The flexibility of the position of the adjective in the sentence in Portuguese sometimes makes it difficult to decide if it is acting attributively or predicatively. The following examples from the PC give an idea of the problem:

(7.57) A mamã chamava a Mac Gren o 'bebé'. Era com efeito uma criança estouvada e feliz. M

(7.58) E aquela aldeia de que nunca soubera o nome, tão quieta e feliz na luz, deu a Carlos um desejo repentino de sossego e de obscuridade, num canto assim do mundo, M

(7.59) Ele mesmo descerrou os estojos e a rapariga, enlevada e feliz, mergulhou os olhos em tais lindezas. AQ

In (7.57) both estouvada and feliz can be described fairly easily as attributive adjectives, as they immediately follow the noun with no punctuation or intonation breaks. In (7.58) the agreement of quieta indicates that quieta e feliz describe the aldeia but the adjectives are separated from the noun by another clause, and the possibility of extending the sentence would allow a copular structure to emerge. Therefore this is best described as an ellipted predicative example. (7.59) is more difficult to classify. Although the adjectives enlevada e feliz appear in the normal attributive position after the noun, they are separated from it by a comma in the written text, and this would indicate a slight break in the intonation when reading aloud. One can assume from this that the girl is not being described as essentially enlevada e feliz, but that she is like this temporarily as a result of what ele - he - has just done. The apparently marked English translation and the girl, excited and happy, gazed at the beautiful jewellery, would be definitely preferable, in this context, to ?*and the excited and happy girl gazed at the jewellery, for similar pragmatic reasons. (7.58) is therefore best classified like (7.59), as an ellipted predicative one.

The difference in use of copulas between the two languages described in Chapter 6 can be largely explained by the type of flexibility of the Portuguese adjective described above. Another factor which probably contributes slightly to the difference is the fact that the EC uses -ly adverbs rather more than the PC uses -mente ones. Although these adverbs account for only a small number of the total examples, Table 4.2 records over twice as many examples in both the PFoc and SFoc groups for the EC. This type of difference can be seen in examples such as the following, using angrily:

which was translated by two different translators as an adjective separated from the preceding noun by a comma as: ...tornou o Coelho, furioso, and ...disse o Coelho, irritado.

I was particularly careful with these examples but, if the results had shown a bias in favour of predicative over attributive examples in the PC, I might have doubted my judgement. As it is, however, the data actually show a slightly higher percentage of attributive adjectives and past participles for the PC. In any case, with or without copulas, the (copula) + adjective/ past participle structure needs to be considered in relation to the passive.

The passive voice is traditionally considered to be formed by "putting the verb to be into the same tense as the active verb and adding the past participle of the active verb", and "is used in English when it is more convenient or interesting to stress the thing done than the doer of it, or when the doer is unknown" (Thomson and Martinet: 1969). In more formal grammatical theory, the passive as a transformation of an underlying active sentence is one of the few transformations that is still taken seriously. This means that the passive form of John kicked Mary is Mary was kicked (by John). In the surface structure the S and O switch roles, but the semantic roles remain the same, although the Agent function, obligatory in the active sentence becomes optional in the passive construction.

Portuguese grammars such as Cuesta & Luz (1971) and Cunha e Cintra (1984) describe the form of the passive as being formed by combining the verb ser with the past participles of active verbs. Mateus et al. (1983) go a little further than the ser passive and include a -SE passive construction which in some ways resembles a reflexive and in other ways an impersonal verb structure.

As the notions of transformation and deep case with the passive were examined further, it emerged that a finer definition was needed and, for English, Quirk et al. (1985: 159-171) describe a 'passive gradient' from 'central passives', through 'semi-passives' to 'pseudo-passives'. This proposal is a synthesis of several individuals' work, notably Svartvik (1966) and Stein (1979).

The classes called semi-passives and pseudo-passives are of particular interest to anyone studying the language of Emotion. The first of these classes is represented by examples that "have both verbal and adjectival properties". According to Quirk et al., "they are verb-like in having active analogues":

They add that these examples are "stative rather than dynamic" which "tilt(s) the scales in favour of an adjectival analysis, since all participial adjectives have stative meaning, whereas the corresponding verbs do not". This class of passives does not often use BY-phrases but some agent-like phrases are introduced by prepositions such as about, at, over, to and with.

The pseudo-passive has "neither an active transform nor a possibility of agent addition" and is known by some as a statal passive. Although many examples of this type of passive use be + past participle forms, the fact that be can be substituted by other copulas and the past participle by adjectives, as in:

(7.63) A Maria estava/sentia-se/parecia + excitada/disillusionada/ feliz/zangada.

shows that we have reached the end of the gradient between central passives and copula + complement, or SPCs constructions. At this stage the current and resulting copulas are all possible with adjectives and adjectival past participles.

Most of the cases when Emotion is expressed using the SFoc past participle of an PFoc verb can be considered semi-passives or pseudo-passives, and about 47% of these examples in the EC had some form of complementation indicating the PHENOMENON. 11,6% were accounted for by clausal complementation, but the remaining 35,4% consisted of a preposition + noun phrase as complement. However, by phrases accounted for only 14,2%, with, of, at and about being used with most of the rest. The phrases with these latter prepositions emphasize the Senser's attitude, rather than any Agentiveness on the part of the PHENOMENON. The PC past participles showed a far lower percentage of examples + complementation, only 26,8%, clausal complementation being virtually non-existent, and only 6,3% appearing with por, the usual equivalent to by with passive constructions. Com and de, roughly the same as with and of, both exceeded the number of por phrases, for similar semantic reasons.

The SFoc adjectives show a similar pattern to the past participles in copular structures. In the EC these adjectives are complemented in 49,2% of the examples, slightly more than for the participles, and in the PC, 21% are complemented, rather less than with the participles. The difference between the languages shows the same similarity as regards complementation in general. However, when an analysis is made of the type of complementation, the PC shows clausal complementation again to be minimal, whereas 21,7% of the EC adjectives take this type of complementation. Of the remaining 27,5%, about two thirds use of, and for, with and about account for most of the rest. None of the EC adjectives use by phrases, and only a small number of the PC ones use por, the most popular Portuguese preposition here being de, similar to of, which accounts for two thirds of the examples. Despite the fact that these prepositions can be considered of little semantic value, and often apparently tied by linguistic convention to the preceding adjective, as with afraid of, the absence of by/por phrases in favour of of/de ones, indicates again that the focus is on the SENSER rather than on any Agentive PHENOMENON.

The problem of the choice of ser and estar arises again with the passive. The central passive uses ser, but the semi-passives and pseudo- passives will very often use estar, particularly with the past participles of PFoc verbs of Emotion. Casteleiro (1981: 69-102) describes the debate between grammarians as to whether past participles can, under certain circumstances, be considered adjectives. Although recognising the fact that past participles in examples like:

agree with the Subject, which does not happen with a regular passive type with ser, as in:

he argues from a transformational point of view which naturally favours the explanation of the SFoc past participles being essentially passives. With examples like:

he argues that the second sentence is a reduced form of the first, and that the disappearance of the perfective ter and the -SE must contain a rule obliging the resulting past participle to agree with the noun. Few would quarrel with the view that most of the PFoc verbs, and their SFoc past participles, are connected syntactically in what would be described by traditional grammarians as an Active <---> Passive relationship, and transformational grammar will naturally accentuate the importance of this fact. However, the passive gradient described by Quirk et al. allows for a more flexible interpretation

This passive gradient is of considerable significance when describing the language of Emotion. Given the relatively large quantity of copula + participle /adjective structures with the PFoc verbs already discussed, the interpretation of this phenomenon is important. We have already seen that the SFoc verbs do not like passivization, probably because they already refer to a state in which the agentivity of the PHENOMENON is minimal. It should follow that PFoc verbs and adjectives are unlikely to be strongly agentive. Besides, since we prefer to focus the more psychologically relevant SENSER of the emotion in actual texts, it is only natural that normally PFoc verbs should favour the passive voice syntactically as a way of expressing this. However, given that the PHENOMENON is only very occasionally agentive, it is unlikely that an unambiguous passive, with an obvious Agent, should emerge, in contrast with strongly agentive verbs, like make, which need an expressed Agent.

As we have seen, the fairly high level of complementation of the EC past participles and adjectives does not indicate Agentivity on the part of the PHENOMENON, and the smaller percentages for the PC also lacked Agentivity. The complementation that does exist can be better understood as some form of explicit recognition of the nature of the PHENOMENON by the SENSER, or by the Speaker. This still leaves us, however, with over half of the EC examples, and three quarters of the PC ones, with no complementation at all, which would seem to indicate a definite tendency to concentrate on the Senser's emotional state or process, particularly in Portuguese.

7.7.4 The semantic importance of the Copula + adjective /past participle structure

Copulas + adjective / past participles indicate either qualities or states, which in their turn can be seen as current or resulting. However, a more cognitive classification of Emotion lexemes does not fit easily with the classification of these SPCs structures as states, unless some form of process leading up to this state, and the nature of its inherent temporariness is somehow taken into consideration. To become this specific in relation to the SFoc adjectives/past participles in English requires reflections on the nature of be which go beyond the evidence of syntax. The other copulas with SFoc words, 12,6%, give a few clues to the semantic possibilities involved. Table 6.1 showed that feel accounts for most of the self-reports on these situations, 4%, and look and seem for most of the reports on others, 2,2% and 1,8%. Only a tiny percentage of examples, 1,2% explicitly use become. The zero copula examples, which may indicate some type of inherent temporariness in the situation, do not seem to be particularly significant in this area, the SFoc ones being actually proportionally fewer, 8,3%, than the PFoc ones, 10%.

The evidence from the PC would suggest, however, that language usage in this area can allow for a more finely discriminated perspective. Ser, which is predicted to co-occur with adjectives or participles describing 'quality', is, as would be expected, the major copula with PFoc words, accounting for 53% of these examples. However, despite predictions to the contrary, ser still appears with 6,1% of the SFoc ones, which allows for a degree of permanence to be expressed with certain lexemes. Estar, which is predicted as a major copula here, and appears regularly in any analysis done by linguists orientated towards syntax, in fact accounts for only 8% of the examples. Its significance lies in its specific implication of a 'temporary state'. As with the EC, a small number of examples can be found with other copular type verbs like sentir-se and parecer, for self-reports and reports on others. However, ficar, which describes a state, but implies a process leading up to it, has 7,4%, and here we have evidence of the process, probably mental, leading to the state. If one can then presume that the majority of examples, 66,9%, which use a zero copula, imply some sort of temporariness, and can usually be expanded to include an estar or ficar copula[8], thereby underlining both the temporary and process aspects of emotion lexemes in copular structures, there is no doubt that the intuitions based on the mother tongue of the Portuguese speaker can contribute more positively to the analysis of the mental processes involved, than those of the English speaker.

We saw in Chapter 6 that the corpora suggest that the Portuguese speaker feels less need to specify the PHENOMENON than the English speaker, and more interest in the process itself. This may be due to this greater range of possibilities offered by the Portuguese language for describing the process. Whether we can deduce from all this that the Portuguese speaker is generally more interested than the English speaker in the process than the PHENOMENON, is a matter for speculation.

More immediately, the different lexical groups vary considerably as to the degree in which they favour the various possibilities offered by all aspects of the SFoc and PFoc verbs, adjectives and past participles, and the way they behave on some sort of Active < - > Passive gradient, and this gradient will be important in assessing the psychological nature of these groups later on.

One difference between Portuguese and English, which is interesting as far as Emotion is concerned, is the Verb + -SE construction. There is a small percentage of occurrences in most Emotion groups. -SE can act as both a reflexive and a reciprocal pronoun. Like other clitics, it can appear separated from the verb, especially in negative and interrogative constructions. The question that arises with Emotion verbs is whether -SE is acting as a reciprocal or a reflexive pronoun, or whether the verb + -SE should be considered a compound verb form, rather like an English phrasal verb, where the -SE, like the adverbial particle with the phrasal verb, is just an extra bit of the verb. Another aspect of the construction is its function as a sort of passive. All of these positions are defensible - depending on which verb and example one chooses to support one's point of view.

Apart from the examples already quoted, when -SE appears with a SFoc verb, it is usually reciprocal, as in:

(7.67) E nada os podia embaraçar; amavam-se; confiavam absolutamente um no outro; ele era rico, o mundo era largo ...... M.

(7.68) And they had nothing to hinder them; they loved each other, they trusted in each other completely; he was rich and the world was large ....Tr. M.

When discussing the transitivity of SFoc verbs in English above, the fact that apparently intransitive uses of these verbs could be translated in Portuguese by the verb + -SE was mentioned. On the one hand there are examples like divertir-se which correspond to similar expressions in English, enjoy oneself; and on the other hand there are those that can act intransitively, like amuar-se, to sulk, and arrepender-se, to repent. With these verbs, as Cunha & Cintra (1984: 308) and Vilela (1992: 77-8) point out, the -SE can be considered an integral part of the verb. The argument put forward by Vilela is that these are inherent reflexives which are merely variants of other non-reflexive structures, and, since the reflexive particle cannot reappear in tests using questions, substitution, coordination or modification, they are not true reflexives.

The notion that the -SE construction is a passive can be found in Mateus et al. (1983) and Busse & Vilela (1986). The examples favour the verbs comprar and vender, buy and sell, as in:

Among the arguments put forward in favour of this interpretation are a) the agreement of the verb with the S of the sentence, and b) the impossibility of providing another, agentive, subject for the sentence without removing the -SE. The English translations do in fact favour a passive interpretation:

and books and houses are usually sold by some kind of agent. However, if one wishes to create an example to prove a point, one could suggest that the following example is ambiguous:

Who sells the prostitutes - their protectors, or themselves - and is this a passive or a reflexive situation? A literal, rather than a more colloquial, translation, would give us The prostitutes sell themselves on the Via Norte, even if their protectors are more active in this respect than they are.

Other examples that are sometimes described as passive are those in which the subject is undetermined, or can be substituted by the third person plural (see Vilela (1986 : 72)[9] as in:

which in English give us the empty It subject or the impersonal they equivalent with an existential clause or THAT clause:

These structures are more typical of official letters and documents and rarely appear in the type of literary texts that form the corpora.

Both in Portuguese grammars and in conversation with Portuguese colleagues, the explanations which are given for the Emotion verbs + -SE are those which favour either -SE as an integral part of the verb, or the passive explanation. However, in many cases, the Emotion verbs that take -SE can also function as normal PFoc verbs which, as such, combine with normal Objects. Macedo de Oliveira (1979) describes this facet of the PFoc verbs which are the subject of her thesis. She quotes Rouveret and Vergnaud (1978) and Gross (1975) on reflexive clitics in Romance languages. The former consider that "the reflexive clitic -SE is base-generated as an affix on the verb". Apparently Gross has suggested another explanation for the reflexive forms as follows:

"l'óbjet de No V N1 doit être um nom 'approprié' dont le déterminant est un adjectif possessif obligatoirement coréférent du sujet. Cette formulation de la dérivation de -SE reflexif a une géneralité très souhaitable parce qu'elle permet d'amorcer les cas où le verbe, dans son emploi transitif, n'accepte pas un complement d'óbjet 'humain'".

Macedo de Oliveira's analysis shows the influence of her supervisor, Maurice Gross. According to her analysis

and that this shows that the -SE is an affix of angustiar when it is acting as a -SE type passive. She supports this claim by saying that, although one can say:

This, she argues, is because in the first examples o comportamento do Pedro is non actif, whereas o Pedro, in the second set, is actif. She reinforces this argument by exchanging the -se a si própria for o João and assuming an actif role for A Maria, on the understanding that one cannot have two actif or agentive roles in the same sentence. She explains the correctness of the following example by suggesting that Pedro, as actif, upsets Maria by telling her about João's behaviour:

b) O Pedro angustia a Maria com o facto de que o João tenha procedido duma forma ignóbil.

If one assumes, as she does, that there is an actif or Agentive role somewhere in the sentence, it is difficult to argue with these syntactic arguments, especially as certain semantic and pragmatic arguments are used to defend them. She does, however, say that with the pair:

Pedro must be considered as non actif to be acceptable, thus making Pedro and o comportamento do Pedro semantically equal. She suggests that these sentences are syntaxiquement ambigües as both must be interpreted actively as:

It is on this point that she, and her mentor, draw attention to a certain weakness in their argument. They have presumed all along that there is an equivalence between the verb + -SE and a passive construction with estar. They have also assumed that a reflexive reading of A Maria angustia-se and their understanding of O Pedro angustia a Maria to have the same case roles, of an AGENT Subject + PATIENT Object variety. As we have already seen, using the concepts of another school of linguistic analysis, it is more appropriate to argue for SENSER Subject + PHENOMENON Object roles for both sentences, in which case o comportamento do Pedro is merely a more explicit form of Pedro and both act as PHENOMENA. One can argue for the AGENT/PATIENT roles for o Pedro angustia a Maria, because one can imagine Pedro deliberately doing something to upset Maria. This, I agree, is not the case with A Maria angustia-se, but, as I have pointed out before, an AGENT understanding of o Pedro angustia a Maria is extremely rare, and a PHENOMENON reading is more normal. If, therefore, we assume the same notion of mental process, with SENSER and PHENOMENON roles, that has been discussed in previous sections, it becomes clear that to simply write this structure off as a passive does not do justice to its overt reflexiveness.

As we have seen, there is a resistance to the idea of reflexiveness attached to Emotion which may have something to do with Behavioural psychology, or with the older ideas that Passions are caused from without. If, as Cognitive psychologists would argue, one can attribute mental processes, even of a sub-conscious kind, to Emotion, it becomes possible to explain A Maria angustia-se on a PHENOMENON / SENSER basis, in which the subconscious mental processes of Maria affect her conscious self. This would add a dimension of conscious appraisal to Emotion in Portuguese which is not easy to achieve in English.

So far I have used the arguments of others to discuss this problem, together with a lot of linguists' sentences as examples. One important point on which I would disagree with Ranchod is that, however much she may argue for syntactic equivalence of the verb + -SE and the estar + past participle, and although, according to the school of syntax to which she subscribes, there is this idea that the passive is just a transformation of the active, one still has to find a semantic and pragmatic justification for having two different ways of saying (approximately) the same thing in a language. Although one can trace a deeper common structure if one tries, the fact is that, in everyday usage, people continue to choose between the two possibilities and may feel that there is a different explanation from the one given by the experts.

I shall now approach the problem using more naturally occurring sentences, and compare an acceptable reflexive verb, like lavar-se or wash oneself, which can also be used non-reflexively, and compare it with assustar-se, which, translated by frighten, gives a similar pattern in English, and with irritar-se, which does not. One can construct the following sets of examples:

These examples can accept tests with other tenses, and with the negative or interrogative, without showing any difference in behaviour. The -SE pronoun acts as a direct object, and the additional information, com + noun can be described as prepositional object, with an Instrument-type semantic role, and the quando + clause as an adverbial. Furthermore, if one prefers to construct examples with the passive, with the non-reflexive verbs one gets:

but none of the -SE versions can be passivized, as reflexives cannot passivize directly. This point could be used to demonstrate that reflexives are not true SPO constructions, but, as Klein (1979: 63-4) points out, in English one can make a construction with get, although the reflexive feature is dropped. For example, the examples given here, using a get passive, would give He got washed / frightened / irritated.

Although the reflexive pronoun allows one to distinguish actions done to oneself, or some sort of process between the unconscious and conscious self, it does not seem strong enough to create a passive. This would indicate that there is a gradient of passivization which functions at a semantic and pragmatic level for all verbs.

As we have already seen, one argument that is put forward in defence of -SE as a verb particle is that, whereas English need only add the -SELF reflexive pronoun to the verb, and nothing else, in Portuguese one can also add a si próprio, as in:

Despite the syntactic arguments put forward by Macedo de Oliveria that one has to view a si própria as an emphatic interpretation of a si as in:

one could equally well argue that the a si próprio after -SE is an emphatic additive - rather like emphatic do accompanying the verb in an English statement, adding force, but not essential - and say that the reflexive meaning in Portuguese is perfectly understandable without it, whereas in English the reflexive pronoun is essential, except with wash, because wash is marked to mean wash oneself unless some other object is specified. The additive nature of a si próprio after -SE is in fact easier to defend than the emphatic do - which usually has to come before the verb it is emphasizing. In spoken Portuguese, it will be distinguished as an additional factor by marked intonation. Instead of saying that Portuguese allows for excess baggage here, one could sympathise with English for lacking the emphatic option.

I hope I have demonstrated that there is a case for considering at least the PFoc type verbs of Emotion as reflexive constructions when used with -SE.

Although I understand the position that with verbs + -SE, which do not have a normal PFoc function, like arrepender-se, the -SE can be considered an integral part of the verb, I wonder if it is really defensible. Vilela (1986: 69) considers a case like this as a lexema complexo, or complex lexeme, in which the -SE has no syntactic function, but in Vilela (1992: 77) he prefers to describe it as an inherent reflexive in which the reflexive forms part of the verb. I do not think there is anything wrong with considering it a reflexive, even if it is inherent rather than true and is syntactically limited, particularly since it is rather an intellectual lexeme. After all, English linguists see no harm in considering pride oneself reflexive and, not so long ago, according to the S.O.E.D., repent was also reflexive. If this is acceptable, one can argue that these verbs are SFoc reflexives, and perhaps there is a reflexive notion half-buried somewhere in these languages which may appear in other languages as well. Sentir-se, which will be discussed separately in Chapter 13, is very interesting because it can act as a copula in a SPCs structure, can be considered as a verb in what could be called a SPOCs structure, would appear to be a passive as in sentiu-se o ar do rio, and also has some notion of reflexiveness inherent in its meaning.

Although all the examples in the EC of verbs in this lexicon with a SELF reflexive pronoun only account for the tiny percentage of 0,4% of the total lexemes, in the PC the verbs + -SE account for 4,2%. As percentages of the syntactic groups the -SE examples amount to 28,8%of the PFoc verbs, and 11,6% of the SFoc ones. The PC examples are, therefore, quantitatively significant.

One of the problems with these structures was how to classify them. First of all they had to fit into the classification of SFoc/PFoc when, in practice, they could be seen as somewhere between these two poles. However, the distinction was made by taking the following factors into consideration.

Some verbs, such as irritar and afligir, could be used as ordinary PFoc verbs, as in:

but they also could appear with -SE alone, although the PHENOMENON is retrievable from the context, as in:

(7.86) Havia um comerciante de madeiras, de nenhumas letras, também viciado nas cartas, que ali vinha todas as noites mas demorava a jogar. O João irritava-se: SU

and also with -SE, together with the PHENOMENON, usually introduced by a preposition like com, as in:

(7.87) Abel, que magicava adular Quina por intermédio do rapaz, mostrando-se afável e mesmo sugerindo uma disposição testamentária que o resguardasse de privações futuras, irritou-se muito com aquela selvagem atitude. Si

These examples were classified as PFoc, although some might say that only the first example is truly PFoc. However, whether the PHENOMENON is implicit or explicit, there is a certain degree of overt reflexiveness in these examples which allows for an PFoc classification.

The SFoc examples were those which could not be used without -SE as normal PFoc verbs, although they could take prepositional objects. So although one cannot find *O João zangou a Maria, one could find them alone with -SE, as in:

With these examples the verb refers more directly to the SENSER's feeling or behaviour, and the syntactic arguments in favour of considering

The difference between irritar as a full transitive and zangar-se as an intransitive verb is one of a gradient between the two, with the other examples of irritar-se, and that of zangar-se + prepositional object falling somewhere in between, as either true or inherent reflexives. The idea of reflexiveness is perhaps more obvious with irritar-se than with zangar-se, given the possibility of forming a normal transitive sentence with irritar. However, the distinction between the irritar-se type verbs as PFoc, and the zangar-se ones as SFoc is very fine, and was made more to give some idea of the relative distribution of the different types within the corpus, than as a rigid classification. Both the PFoc and SFoc examples thus classified take prepositional complements in much the same proportions, and, if one considers all the verb + -SE alone examples together they account for 2,1% of the total examples, whereas those with a prepositional object amount to 2,5%.

The corpus would seem to show that, with these structures, transitivity and reflexiveness are a matter of degree rather than a fixed norm, but that some sort of reflexiveness, of the SENSER explicitly formulating the emotion internally, even when an explicit PHENOMENON is given, is a feature of the Emotion lexical verbs. Here again, we come up again the tendency of Portuguese to focus on the internal emotional process, rather than the state. If the -SE verbs were the only feature of Portuguese suggesting this tendency, the syntactic arguments against reflexiveness would be more valid but, when considered as part of a bigger picture, the semantic points in favour of this reflexiveness become more acceptable.

Most of the problems of adverbial uses of Emotion lexemes have already been discussed. There seems to be a slight preference in Portuguese for zero copula adjectives over -mente adverbs, and preposition + noun phrases.

Adverbs of the type formed by adding -ly, in English, and -mente, in Portuguese, to the adjectival lexeme are not a common feature of the Emotion lexicon, the SFoc and PFoc examples accounting for a total of 6,9%(E) and 2,6%(P). Of those that do appear, the SFoc type are more frequent - 4,7%(E) and 2,2%(P). These describe the emotional behaviour or physical feelings of the SENSER, as in:

(7.90) The baby grunted again, and Alice looked very anxiously into its face to see what was the matter with it. W

(7.91) O seu pobre coração pôs-se a bater ansiosamente, no terror daquela decisão. M 628

and were all classified as examples of BEHAVIOUR, although a few examples were not explicitly so orientated.

Classifying the type of PHENOMENON was not always an easy task as, since the adverb is typically used to describe the behaviour of the SENSER by an outside observer, one has to assume that the observer is interpreting this behaviour correctly. Thus, in an example like:

one has to take her behaviour on its face value and presume that she loves the dog, and is not just pretending affection in order to curry favour with its owner. Besides, description of emotional behaviour is often made when the observer is actually in doubt as to the true nature of the SENSER's feelings. In any case, doubtful examples were always classifiable as PHENOMENON 1 types.

PFoc adverbs of this kind which actually focus the PHENOMENON are rare, but they occur in examples like:

where there is an implication that Yvette is being irritating. This rarity is quite natural if one considers the non-active function of the PHENOMENON, or PHENOMENON. After all, this type of adverb is essentially dynamic in nature.

There are a number of adverbs of this kind which may seem to refer to the causative or PHENOMENON aspects of Emotion, but which, in practice, have become more normal as Emphasizers, as in:

(7.94) She looked up -- what demon possessed him, her youngest, her cherished? -- and saw the room, saw the chairs, thought them fearfully shabby. VW

(7.95) Era apenas introduzir um burguês mais num segredo, tão terrìvelmente delicado, que ele mesmo se assustava de o saber. M

Another use of Emotion lexicon adverbs is as disjuncts, or adverbs with which the Speaker inserts his/her own evaluation of the situation, as in:

(7.96) Above all, she was a blessing to Joe, for the dear old fellow was sadly cut up by the constant contemplation of the wreck of his wife. GE

(7.97) - Não se sabia amar... Ninguém nos ensinou a amar e a gente amava aquilo que era mais fácil: as flores, os gatos, e, felizmente, as crianças... Su

These examples, which focus neither the SENSER nor the PHENOMENON, but are an exterior comment on the whole emotional process or relation, were not included in the corpora. This decision was taken partly because of the unrelatedness of the items to the process, but also because to include them would have caused certain anomalies in the balance of particular lexical items. For example, happily is rarely used as a disjunct, but its apparent Portuguese equivalent felizmente is used frequently in this way. However, felizmente would usually be translated as fortunately, and to include these examples would have definitely upset the balance of these items in the Joy groups without contributing significantly to the general understanding of the lexicon of Emotion.

[1] The reason I prefer to use the Portuguese grammatical term - taken from Cunha e Cintra (1984) - is to avoid the confusion caused by rough translation. These terms are traditionally used to describe Portuguese verbs, not English verbs, and to assume that an equivalent exists in English is to start off with the wrong presuppositions.

[2] This is a good example of the Portuguese verb with a dative - a P + Oi + Od pattern - which does not have an equivalent in English.

[3] See Halliday (1976: 3.3.3.3) and Maia (1976: 4.4)

[4] The central modals are - can, could, may , might, shall, should, will, would, must..

The marginal modals are - dare, need, ought to, used to.

The modal idioms are - had better, would rather / sooner, be to, have to, have got to, etc

The semi-auxiliaries are - have to, be about to, be able to, be bound to, be going to, be obliged to, be supposed to, be willing to, etc.

The semi-auxiliaries are - have to, be about to, be able to, be bound to, be going to, be obliged to, etc

Hope is used as an example of a 'full main verb' here specifically because it can be complemented by a non-finite infinitive clause, and also begin because it can also be complemented by both infinitive and -ING clauses.

[5] The use of poder with the 'tempos compostos' with ter, tenho podido, terei podido etc., and the structure with estar - estou a poder, etc., although not impossible, must be extremely rare.

[6] The present order could be classified roughly as follows:- i) emotion verbs; ii) aspect verbs; iii), iv) and v) intellectual/mental processes and attitudes; vi) and vii) intellectual performatives, and viii) verbs of trying etc. If they were reordered to allow for the emotion > mental process > action gradient, ii) would come at the end.

[7] I am using here the notion of Theme/Rheme, put forward by Halliday on several occasions, which is used to explain the Passive as a way of 'thematizing' the Object of an Active sentence, i.e. of putting the 'theme', or 'topic', of the communication at the beginning of a sentence.

[8] It is possible that ser could be used in some of these expanded examples, but it is more likely that if ser were possible on these occasions, it would be elided in a way which produced the normal attributive adjective.

[9] For non-Portuguese speaking readers, I should point out that Portuguese verbs can, and very often do, function without a subject in the 'surface' structure, and rely on the co-text to provide the necessary information .

[10] Tr. (7.80) a) He washed me.

b) He washed me with the sponge.

c) He washed himself.

d) He washed himself with the sponge.

e) He washed himself when his mother got angry with him.

f) He washed himself when he dirtied his fingers.

(7.81) a) He frightened me

b) He frightened me with the fireworks.

c) He frightened himself.

d) ?He frightened himself with the fireworks.

e) He frightened himself when he burnt his fingers.

(7.82) a) He irritated me.

b) He irritated me with his behaviour.

c) ?He irritated himself.

d) *He irritated himself with my behaviour.

e) *He irritated himself when his mother became angry with him.

f) *He irritated himself when he burnt his fingers.

[11] I was washed by John.

I was frightened by John.

I was irritated by John.