In this chapter I shall look at how the Emotion lexemes behave as verb phrases in the sentence patterns which are recognized in the two languages, and look into the syntactic gradients between these patterns and the implications this has for the semantics of the verbs of Emotion. It is also important to see which syntactic structures coincide with the semantic notions of SENSER and PHENOMENON.

Most theories of syntax agree that there are certain basic sentence structures underlying even the most complex of texts and that the verb phrase plays a key function in determining these structures. The metalanguage used to describe these structures varies occasionally, but I prefer to use the terminology usual in functional-systemic grammar. However, I shall also use the metalanguage of Quirk et al (1985) to describe certain facets of the problem.

Accordingly, it is understood here that all well-formed sentences in English can be described in terms of S, P, Od, Oi, Oprep, Cs, Co, Cp, and Adj, Disj and Conj, although the last two are not essential elements[1]. These elements can be described as Verb, Noun, Adjective and Adverbial phrases which, in their turn, allow for a certain complexity and tend to form sub-patterns. The elements can be combined in a limited variety of ways, and English grammarians usually accept the following combinations as basic sentence patterns: SP, SPOd. SPOiOd, SPOdOprep, SPOprep, SPOdCo, SPOdCp, SPOdA, SPCs, SPA, and SPCp. This list is longer than some, but that is only because other theories see some patterns as variations of others. In this chapter I shall look at those patterns and sub-patterns which are relevant to the Emotion lexicon, and draw comparisons with Portuguese, and with theories on Portuguese grammar, as and when necessary.

The elements that occur in these sentence patterns are considered to be obligatory insofar as each pattern is concerned, although adverbial phrases are usually considered syntactically optional. However, I believe that it is possible and, at times, necessary to argue for an obligatory < - > optional gradient to allow for the semantics underlying apparently simple syntactic structures.

The SENSER will normally be a person, or personified thing, and this means that it will normally appear syntactically as a simple noun phrase. On the other hand, the PHENOMENON, given its varied nature, will appear in all manner of syntactic forms. Although a syntactic description of the way Emotion verbs fit into the basic sentence patterns is not intended to favour the SENSER as Subject, or the PHENOMENON as Object, this is what tends to happen naturally. This can be explained in part because the simpler noun phrase occurs more easily as the Theme of the sentence, as described by Halliday (1984) and Quirk at al. (1985). It is also partly because the psychological centrality of the SENSER, makes it the more normal, and unmarked choice as Theme.

A perennial problem in the linguistic analysis of English, and other languages, is that posed by the prepositions and their complementation. These structures proliferate in those languages which dispense with inflections, or case marking, to define the type of relationship which, as students of Latin, we learnt to call dative, genitive and ablative.

The more reductionist type of sentence structure analysis prefers to classify verbs followed by prepositions as complex verb forms - as in She looked after her son (meaning She cared for her son), giving an SPOd analysis, and She looked at her son would be given an SPA analysis. However, it seems that the creation of the notion of the prepositional Object has been chosen as a way of avoiding argument when the greater or lesser idiomaticness of a phrasal or prepositional verb causes problems.

The SPOO structure has also caused problems. Sentences like John gave Mary the book have always been interpreted as SPOiOd structures, and the analysis SPOdOi was extended to the sentence John gave a book to Mary . However, examples like John got Mary a book , which could give John got a book for Mary caused some doubts, and some people now prefer to consider to/for Mary as prepositional Objects.

Phrasal and, more frequently, prepositional verbs were found in the corpora and, although phrasal-prepositional verbs are rare, one of them, look forward to, fits into the Hope group, and another, look down on, belongs to the same group as despise. The examples were tagged for verb + preposition + complementation, and 150 examples were found for the EC. The PC appears to provide the oddly large number of 436, but the vast majority, 364, are accounted for by gostar de.

Portuguese linguists may fail to understand why English linguists worry so much over prepositions, and this whole affair of the phrasal and prepositional verbs. In the first place, although many verbs in Portuguese do attract certain prepositions, the situation is by no means as complex as in English. Many phrasal or prepositional verbs in English require Portuguese translations which either do not require a preposition, or resort to alternative structures like ter + noun, estar com + noun, or, in some cases, a -SE type verb. As to the type of question test used to analyse the differences between these varieties of verbs in English, Portuguese has no alternative to separating the prepositions from the verb to form a question. Whereas many English speakers would now consider a question like To whom are you talking? to be acceptable only in formal English, and will prefer the more informal Who are you talking to?, Portuguese has only the more formally structured equivalent, A quem (é que) estás a falar?

As far as adjectives are concerned, Quirk et al.(1985: 16.68) are careful to point out that although certain adjectives seem to "form a lexical unit with a following preposition", as with fond of and afraid of, one should not forget that other adjectives can function without prepositional complementation, as with angry, or combine with more than one preposition, as in angry about, angry at and angry with. The functions of adjectives as attributive or predicative will be discussed, as will the SPCs structures in sentences such as James is fond of / afraid of / angry with Jemima, in which Jemima is syntactically classified as a prepositional complement.

The semantics of both SFoc and PFoc verbs, in which a SENSER and an PHENOMENON must exist, would seem to indicate that these verbs are essentially transitive. For example:

(6.1) "Whoever loved, that loved not at first sight?" ( Shakespeare, As You Like It.)

may seem syntactically intransitive but a native speaker will understand that a PHENOMENON is implied. Emotion lexemes, like sulk, gloat, and panic, that act intransitively, include the notion of the SENSER's behaviour as in:

Even with these examples, though, it is not difficult to point to a PHENOMENON which is either psychologically recoverable, or actually in the context.

These examples were classified on the basis of their actual appearance, or surface structure, in the text. However, if, for example, the sentence He despaired appears in isolation after a lengthy explanation of why he might despair, yet with no specific linguistic structure to act as reference for an Object, one has to allow this as an example of intransitivity. Despite their apparent syntactic independence, however, intransitive verbs rarely function without some sort of 'optional' adverbial, or pragmatic back-up in the surrounding text.

My original hunch that this sort of structure was rare in English was borne out by the data from the EC which registers 106 examples, or 5,8%(E) of the SFoc verbs. Of these examples, 30 were of worry, most of them negative imperatives, Don't worry!, 10 of suffer , and 22 of rather marginal examples of the more intellectual-type verbs approve, object and disapprove. Of the remaining 44, most, like rage, mourn and fret behaved like panic, and most can appear + prepositional complementation which shows that this type of complementation is near the intransitive verbs in the intransitive < -- > transitive gradient. There were also 10 examples, or 2%(E), of the PFoc verbs. They involved verbs like hurt and please and were usually used intransitively because the SENSER was understood as undefined or potentially several. The few examples left once the examples quoted have been discounted are truly marked examples.

The PC proved to be richer in these structures, with 168 examples, or 9,4%(P) of the SFoc verbs, and 37 examples, or 3,2%(P) of the PFoc verbs. Sofrer accounted for 38 examples and calmar and sossegar for another 38, over half of which were imperatives which might be translated by Calm down! or Don't worry! Amar yielded the surprisingly high score of 24 and points to an aspect of this verb which is different from love. As with the EC, several of the other examples imply the behaviour resulting from the emotion, as with exultar, triunfar and pasmar.

Despite the fact that many of the PC examples can be explained, or explained away, with similar arguments to those used for English, there is no doubt that a fair proportion of the examples are very close to others which take a final -SE, and most of the verbs that function like this can use both structures. The simple and -SE structures considered together constitute a numerically important group which will be analysed in the section on the -SE pronoun.

Before proceeding to look at the SPCs structures in both languages, it is necessary to examine the different copulas that can function in these sentences.

"a verb is said to have COPULAR complementation when it is followed by a subject complement or a predication adjunct and when this element cannot be dropped without changing the meaning of the verb".

They subdivide copular verbs into current and resultative types, the former referring to a stative situation, and the latter to the process leading to that situation. Downing and Locke (1993: 98-9) make a similar distinction between 'verbs of being' and 'verbs of becoming'. However, let us begin by looking at the most typical copula in English, be, and its equivalents in Portuguese.

In English, the most important copula is be, but it has other functions and can probably be considered the most versatile verb in the language. Syntactically it acts as a copula, functioning in the present and imperfect (in the classical, and Portuguese Imperfeito, sense), has present and past participles, and is used to form the progressive aspect and the passive. It can also be used occasionally as a lexical verb, meaning, roughly, exist. Semantically it has a long history. Be and being, and other members of the family in other languages, have been the subject of much philosophical discussion since the Greeks. More recently, Russell (1903) has discussed the implications of be for logic, others have explored its origins and functions in Indo-European languages, and its salient position in these languages has been compared to its non-existence, or only fragmented existence, in other languages, like Chinese and Arabic, in research geared to both linguistic relativism and universalism.

Kahn (1973) and (1986) has taken a particular interest in the function of be as a copula, which he claims is the most important use of be, although his claim is ostensibly made for the Greek einai and Indo-European *ES- rather than the English be. This does not mean that he ignores its existential uses, as in There is... sentences, or its veridical uses, to defend or refute matters of truth. Where others have seen the different uses of be as proof of its weakness, Kahn (1986: 4) suggests that:

"the absence of a separate verb 'to exist' [in Greek] and the expression of existence and truth (plus reality) by a verb whose primary function is predicative will have provided an unusually favorable and fruitful starting-point for philosophical reflection on the concept of truth and the nature of reality as an object of knowledge".

Although he himself is no defender of strong linguistic relativism, he would agree with those who see the notion of being, peculiar to Indo-European languages, as fundamental to much of Western philosophy.

It is not my purpose to enter into the more philosophical discussion of be, but one point of Kahn's that did attract my attention in his discussion of the copular be is the distinction he makes between its syntactic and semantic functions. His argument is against those who, basing their theories on languages which dispense with the copula, as in:

claim that the introduction of be, historically, was basically a syntactic device to fill what was felt to be a space, and that, as such, be is empty of meaning. He counter-argues this view with examples like:

where the emphatic use of be underlines its veridical, or truth based semantics. Like him, therefore, I shall assume a certain assertion of truth to be implicit in copula constructions with be, particularly when comparing be with other copula-type verbs which have more specified semantics.

English monolinguals who are not philosophers must be forgiven for feeling that, if they say something is the way (all possible information on the subject suggests) that thing is, it must be so. The Portuguese monolinguals have at least an option between the more permanent ser and more temporary or transitory estar, to form a gradient for their beliefs. This point is of particular relevance to the semantics and syntax of Emotion.

The differences between ser and estar, particularly in relation to their use as copulas, have been discussed in considerable detail for both Portuguese - Casteleiro (1981), Carvalho (1986), Ranchod (1990), and Spanish - Ruiz (1977), Luján (1980), Vaño-Cerdá (1982), and Ballesteros (1988), but I can give no more than the general rules here. The existential and veridical functions of be, described above, are usually attributed to the Portuguese ser . As an auxiliary, ser is used to form the type of passive described by Quirk et al. (1985: 3.75) as central. However, whereas be combines with -ING form of the verb to provide the progressive aspect, with its durative and Imperfect semantics, estar combines with either the infinitive in Portuguese, or the -NDO form in Brazilian, for roughly the same effect. The progressive aspect typically refers to a temporary and transitory situation, hence the use of estar.

As a copula, ser is nearly always used when the Complement is a noun, so sentences like:

although examples in which the Complement is a profession will typically drop the article, as in:

When the Complement is an adjective, both ser and estar can be used, depending on the semantics of the adjective. If a quality adjective like alto, azúl or feito de pedra - tall, blue or made of stone - is being used, ser will be chosen. Even if the quality is transitory in reality, and only relevant to to a very limited world view, providing it is seen as a quality, as with lindo, irritante and aterrorizante - beautiful, irritating and terrifying - ser is the appropriate copula. However, when the adjective refers to a non-inherent state rather than a perceived inherent quality, estar will act as copula. As Casteleiro (1981 :208) puts it:

"...os adjectivos seleccionam SER, quando a qualidade ou característica por eles expressa fosse encarada como inerente (intrinseca ou permanente) ao SN sujeito. Pelo contrário, os adjectivos seleccionariam ESTAR, quando a qualidade ou característica por eles veiculada fosse considerada como acidental / extrínseca ou ocasional".

will be used with estar. Naturally, once linguists begin to talk of states it is not long before we come up against Emotion. One can say that the PFoc adjectives, which usually describe an emotion producing quality attributed to the PHENOMENON by the SENSER, will generally co-occur with ser, as in Ele é muito irritante and Ela é admirável, but that SFoc adjectives, and particularly past participles, will tend to appear with estar, as in, A Ana está muito triste., or Ele está muito zangado/aliviado.

However, it is not easy to make a rule about these adjectives. With those cases of adjectives that can describe both the state of the SENSER and the quality of the PHENOMENON, like triste and feliz - sad and happy, it is probably true to say that when the adjective functions in a PFoc way, it will take ser, as in:

However, the corpus suggests that, as SFoc adjectives, triste and feliz can take either ser or estar, as in:

(6.10) E Carlos, beijando-lhe devagar os olhos, o cabelo, dizia-lhe quanto era feliz e quanto a sentia agora mais sua entre estes velhos muros de quinta, que a separavam do resto do mundo... M

(6.11) Ela reconhece: parecia outro homem. Tinha precisão no dizer e era feliz naquele momento. BA

(6.12) Disseram todos: grande golo! Os meus joelhos roçaram no saibro, e ficaram em sangue mas eu estava feliz: grande golo! SU

It is difficult to attribute the choice just to the length of time in question, but perhaps ser refers to something that is felt perhaps more deeply.

This is an area where the grammatical rules do not describe the actual patterns of usage of these verbs in context. The English native speakers intuition that be has some truth-assertive function is understandable when one sees that be is to be found with 82,9% of the PFoc examples and 79,1% of the SFoc ones. The remainder are distributed between other copulas, 7,1% for PFoc examples and 12,6% for SFoc ones, and zero copulas, or where the copula has disappeared in a process of ellipsis, 10% for PFoc examples and 8,3% for SFoc ones.

In the PC the situation is very different, as can be seen in Tables 6.1(a) and 6.1(b) As might be expected of adjectives describing qualities, 53% of the PFoc examples use the copula ser, and only 1,6% use estar. However, 35,1% of the PFoc examples have no copula. With the SFoc examples, 6,1% of the examples use ser, and 8% estar. This is not a significant difference, particularly if one notices that, among the other copulas, ficar stands out as occurring with 7,4%, almost as many as estar. But the most interesting point is that 66,9%(P) of the SFoc example use no copula at all.

There are various reasons that could be suggested for this lack of copulas, and perhaps one of the more important ones is the general difference in behaviour between English and Portuguese adjectives that will be discussed below. It is also possible that a spoken corpus would produce fewer examples of this kind, and that some may be examples of ellipsis prompted by stylistic notions of linguistic economy in written Portuguese. However, it is unlikely that such a large percentage can be attributed entirely to these factors.

I confess that I had hoped that the corpus would give me more substantial evidence for my hunch that ficar competes with estar as a copula with Emotion lexemes in these situations, and it does not help to be reminded that, with the zero copula examples, some sort of copula exists at a deeper level where ellipsis has not taken place. However, I should like to challenge a Portuguese native speaker to go through all the examples and decide where estar or ficar would be more suitable in the expanded version. My attempts at this sort of exercise have indicated that few of the examples would accept clear-cut decisions, and most of them would need to refer to the situational context to obtain a satisfactory result. In any case, Portuguese speakers (or writers?) obviously do

not feel the necessity to be (ser or estar) feliz or triste, that the English speaker does. Since I am inclined to believe, with Kahn, that these verbs are not mere space-fillers, but have some kind of meaning, however small, I find this significant. The fact that the SFoc examples of this type are almost twice as numerous as the PFoc ones encourages me in this belief. However, there are several other factors that have to be considered before the situation can be clarified further. Let us start by looking at some other verbs which are considered to function as copulas.

If such an apparently primary copula as be can be seen to diversify into two copulas in Portuguese, it is hardly surprising that the secondary copulas should also show differences in behaviour. I shall select the more relevant items from Quirk et al. (1985: 16.21) 's list as a starting point, add a few self-explanatory examples within the limits proposed, and suggest suitable translations, before going on to look at other copula type verbs in Portuguese.

Parecer would seem to have to cover the differentiated copulas in English which normally refer to what the Speaker deduces from external clues - facial expression, tone of voice and general behaviour - about someone else's internal feelings, a point which is relevant to Emotion. There is no resultative copula in Portuguese quite like become or get, which focus the process + the end result. All the the resultative copulas in English, however, seem to include a certain notion of process, and do not just focus the end result, or state. Emotional states in English will therefore use be to describe the end result, while Portuguese prefers ficar. Ficar is not easy to compare to any one verb in English, as it has a fairly wide range of meaning. Although one can make ficar refer to the becoming process by using it with the progressive aspect, as in:

ficar essentially refers to the result, not the process. The process emerges semantically from the progressive aspect here. With a place adjunct, ficar, as a lexical verb, has the meaning of stay, as in:

and this stativeness allows it to translate remain, keep and stay as copulas, but prevents it from referring to the process leading to these situations.

The only copula which approaches become or get, in the sense of referring to the process + the result is tornar-se. However, it seems to be restricted to more marked or abnormal situations, as in:

and it is, at best, unusual, to use it with the notion of natural aging, for example. Become does not fit easily with Emotion adjectives, unless the emphasis is on the result, and even then the possibilities are limited, as in:

Tornar-se is even less acceptable, but ficar perfectly acceptable, as it refers essentially to the end result, as in:

(6.19) a) *O João tornou-se furioso quando lhe lembrei que me devia dinheiro.

Both parecer and ficar occur quite frequently with Emotion words, as can be seen in Table 6.1 (b).

There are other verbs in Portuguese which can be used as copulas, but one must be careful because this is not their only function, and, apart from being used as lexical verbs, they are considered to be auxiliaries by some authorities. Examples given in Mateus et al. (1983: 140) are andar and continuar, which seem to convey a meaning of be + duration/continuity as in:

but they can only occur in relation to the temporary situations, not permanent qualities, of individuals. They are occasionally associated with states of Emotion because of their temporariness, and are interesting in that they allow a dimension of duration of Emotion not easy to achieve in English. The example, Andam zangados desde a semana passado, is better translated by They haven't been on speaking terms since last week than ??They have been angry since last week, . Continuar is probably more easily translatable by adverbs like still, as in, Continuam zangados, which is probably best translated as They are still angry with each other.

Casteleiro (1981: 136) mentions further 'para-' or 'pseudo-copulativas' like ir, vir, partir, voltar, entrar, sair, cair, viver and morrer (translatable asgo, come, leave, return, enter, go out, fall, live and die). Their behaviour does not seem to be well-documented but, like others noted in Quirk et al., most seem to be severely restricted as to the adjectives they can co-occur with, as can be seen in the following examples:

As can be seen from Table 6.1 (b), I have included rather more verbs as sub- or potential copulas, particularly for the PC. The total percentage of examples using these copulas in the EC amounts to 7,1%(E) for PFoc examples and 12,6%(E) for SFoc ones. In the PC, the PFoc examples amount to 10,4%(P) and the SFoc ones to 18,9%(P). As should be expected, the copulas cited above claim the large majority of these examples. The few that remain are interesting, although I am not presenting a strong claim for their status as copulas, particularly those with -SE complementation.

The adjective and past participle data for the EC contain several references to verbs + Object complementation which are not considered SPCs structures, but rather SPOdCo ones of the He made her happy variety, and the same can be found for the PC with verbs like fazer. However, there are also several verbs with -SE that can hardly be considered SPCs structures. The explicit -SE in Considerou-se feliz can actually translate as He considered himself happy, and Tornou-se feliz as he made himself happy. Just because English usually avoids the -SELF pronoun here does not mean that the use of the -SE one in Portuguese may be explained away as unnecessary. Sentir-se is included in this list because it seems to translate feel but is marked ?? because I have doubts whether this classification is valid. For years I have corrected my Portuguese students' sentences *I feel myself happy to I feel happy. Yet they insist that they feel a need to refer to themselves in a way that transcends mere grammar, and I now believe that there is a strong case here for considering these examples with sentir-se SPOdCo ones with the -SE or SELF acting as Od.

There are still one or two -SE verbs, like erguer-se and separar-se in the list and a certain number of verbs of movement often used intransitively, like ir and vir, and others of stance, like parar, quedar and certain senses of ficar. Similar notions can be found with the EC examples depart, and lie and sit. These examples were all recorded because they are followed by adjectives or participles. I am careful to say 'followed' and not 'complemented'. The type of example to be found is:

(6.22) Os seus olhos contentavam-se em adejar sobre a cabeça dos dois pequerruchos, que vinham para a sua beira traquinar. Acocorados no tapete, brincavam felizes e descuidadosos com as coisas mais inverosímeis. AQ

I am sure many linguists will immediately say that there are two underlying propositions here - os pequerruchos brincavam (intransitive use of brincar - the children were playing) and estavam felizes (or SPCs structure - they were happy). I would agree, and I would further point out that this sentence can only be translated either as the children were playing happily or the children were happy playing, which, if analysed, only serve to corroborate this. In Portuguese one need not even add any words to make this meaning clear. One could add a comma, or read the sentence with a certain intonation, to give Acocorados no tapete, brincavam, felizes e descuidadosos com as coisas mais inverosímeis. Many similar examples were excluded precisely because the comma existed. The fact remains that the examples recorded did not use this comma, and it is perfectly possible to read most of them with the same intonation as one would use with estavam felizes, which nobody would argue about. If one can consider that such examples are acceptable, this allows one to argue that they be considered on some sort of gradient between intransitive, or SP, and the SPCs structure proper.

particularly when these emotions can be understood as moods. However, many SPCs structures are not so simple, and the adjective or past participle functioning at Cs often requires complementation of its own. This type of complementation, as Quirk et al. (1985 : 2.52) claim, "may be either obligatory or optional on the syntactic level" and "also overlaps with other functions such as adverbials and modifiers". However, it should be made clear that the complementation of the adjective is dependent on the adjective and not on the copula which introduces it. Thus, despite the fact that fond of requires complementation, one may still make an SPCs analysis of:

In the case of the adjectives and past participles of Emotion, although it may be perfectly justifiable to claim that this complementation is not obligatory syntactically, as in:

it is difficult to dismiss it as purely optional from a semantic point of view. On other occasions the overlapping described by Quirk et al. is apparent because one can paraphrase what appears syntactically to be adjective complementation with an adverbial phrase, as in:

The fact that it is possible to coordinate adjectival and adverbial phrases of this kind, as in:

Another explanation can be given if one accepts the notion that copula + adjective structures can be classified as verbs; in these situations one can compare the adjectival complementation to the object of a transitive verb, as in:

This form of complementation is common in the EC - 52,6% of the SFoc predicative adjectives and 44,4% of the past participles take complementation - and, in the examples contemplated, SFoc adjectives and SFoc past participles seem to function interchangeably.

Similar structures, if not direct translations, can be found for Portuguese, although they are less frequent - 18,7% of the SFoc predicative adjectives and 25% of the SFoc past participles. Thus one can find complementation by a prepositional phrase, as in:

Casteleiro (1981) discusses the syntax of the adjective from a transformational point of view. He rarely deals with adjectives grouped according to lexical meaning, but one lexical area he does discuss, is described as 'adjectivos emotivos', by which he means the PFoc adjectives of Emotion. Casteleiro does not include SFoc past participles in his analysis, for reasons that will be discussed in relation to the passive, but he classifies the SFoc adjectives as using estar.

Although English adjectives seem to include prepositions in their lexical unit when complemented by a noun phrase, a WH- nominal clause or an -ING clause, the same preposition is absent when the adjective is complemented by a (THAT) clause. For example:

Portuguese usage is different in this respect. First of all, there seems to be no real Portuguese equivalent of the non-finite -ING clause, at least in the sense we are using them here, most such structures being translatable with non-finite infinitive clauses or by expanded versions in finite clauses. Also, although the equivalent of WH- nominal clauses exists in Portuguese, it seems impossible to use this construction with adjectives, with or without prepositions, in the lexical field under consideration. The construction seems to have been absorbed by the ter + noun one, as in::

Adjectives + prepositions are used before noun phrases and infinitive clauses, but, unlike English, the prepositions also seem to be necessary with some QUE/(THAT type) clauses.

The biggest group of SFoc Emotion adjectives, to be found in Casteleiro (1981: Table 3: 496-501), which contemplates the adjectives given with de + non-finite infinitive or finite QUE (THAT) clauses in which the verb is in the 'conjuntivo'. Not all of the following adjectives can be used with both contructions, but all of them, he claims, can be used with at least one of them. Therefore alegre, ansioso, apreensivo, atónito, ávido, cioso, cobiçoso, contrito, desejoso, desgostoso, feliz, impaciente, medroso, radiante, raivoso, receoso, sôfrego and úfano, can be used in examples like:

(6.41) Algumas crianças estão felizes de (ver / verem) os brinquedos funcionar.

Similarly, (ibid: Table 4: 502-506) describes constructions with em with ansioso, compassivo, confiante, contente, esperançoso, paciente, relutante, tímido and vaidoso, as in:

His Table 5: 508-9, with the preposition para, only seems to contemplate ansioso, as in:

His Table 6: 510 lists ansioso, ávido, curioso, furioso, impaciente, radiante, raivoso and sôfrego in constructions with por, as in

Finally, in (ibid: Table 7: 512-3), alegre, apreensivo, desgostoso, estupefacto, furioso, perplexo and triste are seen to take the constructions com o facto de que, as in:

(6.49) Os bombeiros estão tristes com o facto de que ainda não conseguem dominar este tipo de incêndios.

Casteleiro's work describes adjectives from a syntactic point of view and is obviously based on what can be done, rather than on a large corpus. I cannot enter into any discussion of the points he discusses, however. Consultation of the corpora will reveal that there is only one example of an adjective + preposition + QUE clause, and only 2 with an adjective + QUE clause, on which I could begin to argue[2].

PFoc adjectives describe the quality ascribed to the PHENOMENON by the SENSER. Both languages seem to prefer the attributive type adjective to predicative ones here in a ratio of 6,9%(E) to 3,1%(E) of the total, and 7,1%(P) to 2,8%(P). Most of the predicative examples in the EC appear in simple SPCs patterns, with about 11,2% of these adjectives taking some form of the complementation described above.

(6.50) She/the film/the actor in the play we saw last night is very annoying/depressing/boring.

(6.50) Ela/o filme/o actor na peça que vimos ontem à noite é irritante/deprimente/aborrecido.

In English, we would describe this as a case of extraposition, or that we have an anticipatory, empty or dummy subject in it which stands in for the real Subject which is normally a (THAT) clause. The reasons for creating this type of structure are largely stylistic, or related to thematisation, to avoid a clausal Subject that would make the sentence top-heavy, or to make the process the Theme of the sentence. It is not a very usual structure and there are only 6 examples in the corpus.

To achieve the same effect in Portuguese there is inversion of Subject and Predicate, to Predicate coming at the beginning of the sentence, without the need for any Subject substitute. Most of the PFoc adjectives of Emotion analysed by Casteleiro can take this construction. However, given the generally complex and fluid system of word order in Portuguese sentences, it was impractical to mark examples for this sort of structure.

Although (THAT) clauses frequently appear as the subject of sentences in books on linguistics, as Quirk et al. (1985: 18.33) say, "it is worth emphasizing that for clausal subjects the postponed position is more normal than the canonical position before the verb". Although I was prepared to mark examples of this type, I did not find any.

There are certain verbs which cannot function alone and need complementation, even though they cannot be described as transitive. Quirk et al (1985) would classify this structure as SVA, as in He (S) lives (V)in Paris (A), or as in He(S) went (V) to Paris (A), and these verbs as belonging to the 'stance' and 'movement' classes. Functional-systemic grammar prefers to consider the A element as a Cp, or predicator complement. The use of this structure with Emotion lexemes occurs in both languages, with the noun form of the emotion in the prepositional phrase, as in:

The interpretation of these sentences sees the in / em + noun phrase as either referring to a metaphorical space concept, or to a state. Portuguese allows a large number of emotion words to combine with em / com + noun of Emotion, like medo and vergonha. Ranchod (1990) has analysed this and related structures in detail. This study covers a wide variety of lexical areas but, being another disciple of M. Gross, she is more interested in syntactic than semantic data. Although the theme of her thesis is estar + nominal predicates, she also shows the relationships between these constructions and others, like ter and the copula type verbs described above, + nominal predicates. From her point of view, these verbs have little intrinsic meaning of their own and are there to carry tense, aspect and mood for the noun which gives the meaning to the phrase. Her reason for choosing estar as the focus of interest seems to be that it is the most meaning-free verb of those considered, rather than any criteria like frequency of usage. On the other hand, it is as well to point out that it is the state type estar, and not ser, that is used here, usually with em or com (in or with), which would seem to indicate an idea of be in (a state of) X or be with X . Several of the nominal predicates she cites are related to Emotion lexemes, and most of them appear in the same syntactic group. This syntactic group shows the close relationship that exists between the ser, estar, ter + noun and other, copula type, verbs, nominal predicates, and, on occasion, the PFoc and -SE type verbs. Ranchod and others would consider ter as having little semantic content.

Although there is no doubt that this structure is perfectly possible with Emotion lexemes, and should be analysed in any reference to the subject, the actual number of occurrences of the structure would seem to be fairly rare. In the PC, for example, only 9 examples of estar + preposition + noun were recorded. I suppose most believers in underlying or deep structures in languages would point to adverbials consisting of a preposition + noun phrase as being abbreviated versions of full propositions, and that He stared at her in terror is a condensed version of He stared at her + he was in a state of terror.

The SPOd construction is the most common in the corpora, but the EC contains more examples, 83,7% for PFoc verbs and 54,8% for SFoc ones, as against the PC, 60,7% for PFoc verbs and 43,8% for SFoc ones. The extra dimension of the -SE verbs, not included in these percentages, may have something to do with the difference between the languages. However, the SPOd pattern does not always appear as the simple noun + verb + noun type of John loves Mary. The direct object, Od, may also consist of a complex noun phrase, a non-finite clause or a finite clause. With PFoc verbs it is unlikely that clausal complementation will be involved since the SENSER Object is normally a person or personalised noun. These verbs will occur in sentences like:

(6.53) John frightened/annoyed Mary/the monster/the man selling peanuts at the cinema last Saturday night.

SFoc verbs, however, are likely to occur with most forms of a clausal nature acting as Object, given the nature of the PHENOMENON.

Interestingly enough, Quirk et al. (1985) classify the verbs which take noun phrase Ods according to whether they take the passive or not, and this is a common method of distinguishing between different types of transitive verbs. Love and like are both in the list of "common examples of monotransitive verbs allowing the passive", and like is used as an example of a sub-class taking a "typically animate subject + typically animate object". Another sub-class, however, is also made for those taking a "typically concrete or abstract subject + animate object", which largely consists of PFoc verbs of Emotion, like shock, appal and please. As I hope to show later, passives with SFoc verbs like love and like are extremely rare, but the PFoc ones are unusually frequent in their passive form

Prepositional and phrasal-prepositional verbs are usually included in SPOd constructions because they behave more like mono-transitive than intransitive verbs. This is partly because of the resemblance of the prepositional object to a direct object, in that it can accept a passive voice, albeit with some awkwardness of style. When the prepositional verb is followed by a (THAT) or TO-infinitive clause, the preposition disappears, although it may reappear in the corresponding passive, or in extraposition, as Quirk at al. (1985: Ch.16.35) explain. The resulting passive and extraposition type sentences are of the kind that linguists stretch to suit their theories, rather than naturally occurring ones. Perhaps this fact could also be attributed to the idea of gradience which would leave these examples somewhere on the SPOp to SPOd gradient, making their status as direct or prepositional objects unclear. Examples of Emotion verbs in this class are hope for, long for and rejoice at, and the syntactic rules can be demonstrated as in:

As already stated above, although there seem to be a certain number of English verbs+ prepositions functioning like this, about 8,3%(E) of the total EC verbs, except for the obvious exception of gostar de, there are few examples in the PC.

The verbs which take (THAT) clauses as Od in SPOd sentences are carefully classified by Quirk et al. (1985 : 16.30-4) into factual, suasive, emotive and hypothesis verbs. Interestingly, hope is classified as factual and prefer as suasive. The emotive verbs given include regret, marvel, rejoice and wonder, and the adjectival constructions be sorry / anxious are felt to belong to this type semantically.

Emotion verbs are not really involved in the semantics which produce a WH-interrogative clause as Object. It is interesting to notice, though, that verbs of perception, like discover, notice and see, come into this class, as do verbs of intellectual process or action, like doubt, know, say, tell and think.. These verbs would seem to be involved in the rational receiving, processing, and discussion and explanation of information in a way that Emotion verbs are not. The only way in which emotion can be expressed this way, and only negatively, is with the negative constructions like not care and not mind. However, nominal WH- clauses - and O QUE clauses in Portuguese - can occur quite easily with these verbs, as in:

There are a few examples of these structures in the corpora, 22 in the EC and 6 in the PC.

Complementation by non-finite clauses is fairly common with certain SFoc Emotion verbs in English, whether the subject of the non-finite clause is the same as that of the main clause or not, as in:

The EC contains 237 examples of the first structure and 18 of the second, a total of 14% of all the verbs. The frequency of infinitive clauses in this this area is lower in the PC, as only 117 examples of the first structure were found in the corpus, or about 6,6% of all verbs. This is an area of syntax which has always provided trouble for the English language teacher with Portuguese students, as the following explanation using the Emotion verbs will show.

When the subject of a nonfinite clause is the same as that of the superordinate clause we get examples like:

The difference between the two sentences is minimal, although examples exist which show more clearly that the infinitive favours a potential or hypothetical interpretation, as in:

whereas the -ING clause is more likely to refer to something which definitely happens or has happened, as in:

This difference is not always easy to explain, even to English speakers, but it is particularly difficult for Portuguese speakers as only one translation is normal for the pair of examples:

and the infinitive construction could be used to translate the others. This would not matter too much if the -ING clause were less important - there are about 40 examples in the corpus - but it is the normal form for things done habitually, as in:

However, when the subject of the nonfinite clause is different from that of the superordinate clause, the problem is even more complex. Here it is not just a question of explaining the difference between two perhaps stylistically different alternatives. If we say:

we cannot resort to a nonfinite verb in the subordinate clause in Portuguese at all with these verbs. Instead we have to use a finite QUE clause with some form of the 'conjuntivo', or subjunctive :

The interesting semantic point here is the more or less hypothetical nuance that can be given by the type of complementation used.

Although all sentences in English which are classified as SPO are presumed to be SPOd patterns, one could argue for an SPOi pattern in a sentence like:

but I think most people would agree that the SPOi sentence is a little forced and that the Od could only be omitted by some process of ellipsis which allowed it to be retrieved from the immediate context.

However, in Portuguese an SPOi pattern can be specified, as in these examples from Vilela (1992: 34):

where, although no Od is contemplated, the object is preceded by a preposition and, in the first case at least, the Od becomes a dative when made into a pronoun, as in Ele obedece-lhe.

Examination of the PC data shows that about 60 examples of indirect objects have to be accounted for among the PFoc verbs, and predicative adjectives and participles, and a further 24 among the SFoc ones, not to mention a further 14 in the more ambiguous territory of the -SE verbs. Agradar accounts for 25, desagradar for 15, repugnar and doer with 11 each, and the rest from a variety of synonyms of these lexemes. Doer, as Vilela (1992: 122) explains, usually relates the part of the body which hurts to the person who feels the pain. Like agradar and desagradar several of the lexemes belong to the Appreciation or Liking and Dislike groups, and many if not most of the examples use the dative pronoun, as in:

(6.63) Agradava-lhe também muito a vizinhança, aquela doce quietação de subúrbio adormecido ao sol. M

In English the existence or otherwise of such a structure is obscured by the fact that no dative pronouns exist with which to perform tests, and by the controversial subject of prepositional verbs. While in no way challenging the Portuguese interpretation of this structure when a dative test can be performed successfully, I would suggest that some form of gradient could possibly be postulated for Portuguese between these examples, and those where a preposition appears between a verb and its direct/indirect object or prepositional complement.

Extraposition can occur with PFoc verbs for the same reason that it does in SPCs patterns with adjectives and there are 30 examples in the corpus. Examples like:

However, it is is rare to find extraposition with SFoc verbs - only 3 examples were recorded in the EC. They are usually found with hope or fear, as in:

but this is due as much, if not more, to the inadmissibility of the structure with a non-passivizing verb like esperar than the problem of extraposition.

The reason I have drawn attention to the gradient between the indirect object, the prepositional object and the adverb type of prepositional phrase is because it begs a series of questions for Portuguese analyses of the SPOO type of structure. In the first place, although Portuguese linguists happily use the term indirect object or 'complemento indirecto', it must be pointed out that, except when the dative form of the pronoun is used, as in Dei-lhe a carta, the 'complemento indirecto' requires a preposition like a as in Dei a carta à Maria. To prove this is an indirect object, they can resort to tests with dative pronouns but, according to the classification used here, the phrase is a prepositional object, albeit within a SPOO structure. The nature of this dative or SPOO structure is of little interest to English for the syntax of the Emotion lexicon, but the dative structure appears in over a hundred examples in the PC, the majority for PFoc lexemes.

It is only natural that this area of syntax should provide food for thought, given the problems already discussed as to the relation between indirect and prepositional objects, and complements. Quirk et al. (1985 : 16. 59 -67) are at pains to demonstrate that this an area of gradience rather than of clear-cut distinctions. In (ibid : 55-63) they go a little further and describe a third type of SPOO structure - in which the indirect object takes no preposition but the direct object does, as in:

SPOO constructions are rare with Emotion verbs, the only really important one being envy in the EC and invejar in the PC. Envy would seem to use the first and third types of SPOO structure, as in:

It would seem that Portuguese would prefer the following interpretation of the equivalent structure:

(6.68) a) Ela (S) invejou (P) o éxito (Od) do João (prepositional phrase qualifying 'éxito').

The groups of emotions which have this element of ditransitivity are those which Ortony et al. describe as related to the Fortunes of Others, the Resentment and Sorry For groups. From a syntactic point of view, only envy and invejar behave in a true ditransitive way, but from a more semantic point of view, there is an element of ditransitivity in the other lexemes of these groups which can be recovered from a wider or more psychological view of the text. Only two of them are accounted for by invejar, admirar has 4, and there is another from respeitar, and one for amar which is quoted below. An example in context is:

(6.69) O administrador ergueu o copo, depois de cheio, admirou-lhe à luz a cor rica, provou-o com a ponta do lábio, e piscando o olho para Afonso: M

If we were to translate an example like this, the dative structure would be turned into a possessive form as in he admired its rich colour against the light.

The SPOCo structures occur with certain groups of verbs specified by Quirk at al. (1985 16.47), but few have direct connections with Emotion, the nearest being the intellectual stative verbs like believe and imagine. They presume that the noun or adjective which acts as the C is Co, or an Object complement. Certain verbs seem to like the prepositions as or for before the Co, but they are largely optional. The type of example usually given is:

(6.71) Toda a gente (S) considera (P) Mario Soares (como) (Od) bom presidente (Co)[4].

These structures can be passivized, with the Od becoming the S of the passive sentence. They are of indirect interest to Emotion with examples in which the verb has a causative function and the Co is a resulting attribute, such as She made him happy/angry, or Ela fez a família feliz, or when an opinion is given as with verbs like consider and considerar. Several examples can be found in both corpora - about 75 for the EC and 25 for the PC.

One of the more interesting points of discussion in this analysis, from the point of view of comparing English and Portuguese, are the SPOdCo and SPOdA patterns which cover an area of syntax that causes confusion between the two languages, because they can seem similar with certain verbs, as in:

Also, given the normal post-noun position of Portuguese adjectives, one might argue for ambiguity of structure in:

The examples given so far have nothing to do with emotion and it is fair to say that any further discussion of the SPOdA structure would be irrelevant here. However, Emotion verbs in English, like like, prefer, want and wish, can appear with the SPOdCo structure, as in:

although intonation alone will disambiguate the first example, so that it either translates the English example given, or means:

The analysis made here of these structures is made to draw attention to the fact that attributive adjectives in English and Portuguese can cause problems of interpretation. The post-noun position of the adjective in Portuguese, as with café quente, by no means constitutes a proof that the noun+adjective order in Portuguese is always an Object+Complement phrase. However, it does draw attention to its position on the gradient which begins in what, in English, is an adjective premodifying a noun phrase in a simple SPOd structure. This may not seem particularly important here but it starts a train of thought that, after the functions of adjectives, past participles and passives have been considered, may help to explain certain features of the PC.

6.8.2 The SPOd, SPOdCo, and SPOO patterns and syntactic and semantic gradients

Quirk et al. (1985: 16.63-4) describe a syntactic gradient for interpreting patterns with infinitive complementation which is of interest here because it involves Emotion verbs which function in the more consciously perceived part of the emotional > intellectual spectrum. The problem is dealt with in detail in Quirk et al. (1985: 16.63 - 67), but here I shall just use some of their examples, modify others and translate them into Portuguese to allow for comparison. The problem is posed when a TO-infinitive clause is used after certain verbs, as in:

The distinction between the different patterns underlying the apparently similar sentences is achieved by a series of tests using the passive, as in;

As can be seen, the nearer the verbs are to the Emotion spectrum, the more easily the complementation of the verb can be seen as a direct object. Expect, which is in the more conscious part of the spectrum, allows two types of passive, which allows for an SPOd and an SPOdCo interpretation of the pattern. The syntactic independence of the students as Ois with the behaviour type verbs, ask and invite, allows this phrase to become the subject of the passive sentence.

we can see that, despite the fact that only a) and g) allow for an infinitive structure, that the SPOdCo interpretation is not a feature that can be applied to Portuguese, and that the passivizations of e) and f) are a little clumsy, a similar pattern emerges. One can therefore suggest that the more emotional words only favour an SPOd interpretation.

In describing this gradient, Quirk et al. are largely interested in the syntactic point of view and in the specific problem of infinitive clauses and their interpretation. Before leaving this subject, however, I should like to draw attention to the similarity between They expected James to win as an SPOdCo pattern and the SPOd pattern also possible with expect:

Although like, want and other verbs, like love and hate, use the infinitive construction within an SPOd interpretation, expect is unusual in taking both this and the (THAT) clause complementation. When arguing for a semantic gradient between more or less conscious processes of emotion, it is as well to remember both types of clause and to consider the following SPOd sentences:

These Clausal Objects or PHENOMENA involve some sort of proposition about the world, so we are dealing with the more consciously analysed emotional situations in all these examples. Although I shall argue for a certain degree of conscious evaluation in most cognitive processes related to emotion, it is at this point that the conscious processing of emotion is most obvious.

Earlier in this chapter the nature of the PHENOMENON was discussed from a semantic point of view and the classification used in the analysis of the corpora was given. After analysing the more syntactic aspects of the problem, it should now be clear that the syntactic forms of the PHENOMENON can give us several clues as to the degree to which conscious mental processes are involved in the emotional situation. For this reason an examination of the PHENOMENA which occur with the different lexemes and lexical groups can help to establish criteria for discussing this aspect of emotions.

When there is no expressed PHENOMENON, as in SP and simple SPCs and SPA sentence patterns like:

(6.78) It is better to have loved and lost, than never to have loved at all. (Tennyson)

it either has to be retrieved from the text, or reconstructed from the constellation of stimuli implicit in the context. In these cases we are probably dealing with the least consciously processed types of emotion - the objectless emotions or moods - for which the SENSER finds it difficult to provide a rational explanation.

As we have seen, SPOd and more complex SPCs patterns are not always as simple as:

In these examples the PHENOMENON is identified and localised in a person, a group of people, or personal behaviour. One can also identify objects in our world as causing emotions:

However, the non-finite and finite clauses which frequently occur as PHENOMENA do demonstrate a fairly high degree of conscious mental processes. When the Subject of both the main and subordinate clauses is the same, the phenomena may be situations or activities in which the SENSER is involved, as in:

In these cases the PHENOMENA are more specific and show a higher degree of conscious assessment by the SENSER.

As the clausal PHENOMENON becomes more complex, so too does the level of conscious appraisal involved. When the subject of both clauses is the same, as in:

the syntactic structure of the PHENOMENON already shows a high degree of conscious mental process in relation to events affecting the SENSER. When the subjects of the two clauses are different, as in:

the situation is understood intellectually to the degree that the emotion is recognised as not even involving primarily the SENSER, but some Other whose fortunes are affected.

carries the gradient of conscious appraisal to the point where the PHENOMENON is the real focus of interest and the emotion, if such it can be called, is assessed as being common to several SENSERS.

Whenever a verb, adjective, past participle or noun was complemented in a way that gave a syntactic clue to the identity of the PHENOMENON, the example was marked accordingly, as follows:

The classifications 'c', 'd' and 'e' do not apply to Portuguese as, at least in this area, no examples were found. The analysis of the individual lexemes and lexical groups in later chapters will explore this syntactic gradient in order to show why certain emotions can be classified as more sophisticated processes of cognition than others.

[1] This analysis assumes that:

S = subject of the verb

P = predicator

Od = direct object

Oi = indirect object

Oprep = prepositional object

Cs = subject complement

Co = object complement

Cp = predicator complement

Adj = adjunct

Disj = Disjunct

Conj. = Conjunct

[2] Casteleiro's work covers a very wide selection of adjectives, and his analysis of Portuguese adjectives is exhaustive. No doubt there are similar works on English adjectives, but I have not yet found anything on quite the same scale.

[3] Quirk at al. (1985:16.56) analyse the indirect object nature of John by referring to the underlying RECIPIENT or BENEFICIARY semantic role, and by demonstrating how the prepositional object cannot be the subject of the passive.

[4] This and other examples are from Vilela (1992 : 36-7). He analyses them as " ... um grupo mais ou menos numeroso de verbos, que prevêm, além de um complemento sem preposição (=cd [complemento directo]), um outro complemento (= designado comummente <<nome predicativo de complemento directo)>>, podendo pertencer à categoria nome ou adjectivo, e que pode vir ou não acompanhado por como / por.

[5] The WH or O QUE clause is when the clause in English begins with WHAT or another WH- form, as in: I liked what I saw.

Gostei do que vi.

It is often found with questions.