CHAPTER 8

To restrict oneself to the verbs and other syntactic structures discussed so far in an analysis of the lexicon of emotion would be to ignore 30,4% of the EC and 41% of the PC. Apart from being sizeable proportions of the corpora, this is a clear difference between the languages which should not go unnoticed. The examples in the corpora were marked for countability/non-countability, different types of complementation, and their occurrence with certain prepositions and verbs, and it is possible to discern certain syntactic and semantic patterns.

I shall begin by discussing the countability or non-countability of emotion nouns in a way which owes more to the preoccupation of linguistic philosophy with the elements of conceptualisation than to the type of semantically based linguistic theory that has predominated in my analysis of the verbs. It was Oscar Lopes who first drew my attention to the complexity of the nature of the countability of abstract nouns, and although perhaps the theoretical constructs used to analyse them pose as many problems as they attempt to solve, the intellectual discipline imposed by attempting to use them is salutary.

As we have seen, Averill, in his criticism of Ortony et al., brought up the problem of whether the study of emotions was one of natural kinds or logical individuals concluding that, in general, it was the former. When the psychologists or philosophers are talking about Love, Anger or Fear, the chances are they are discussing natural kinds. However, this is not a common use of these nouns outside psychology and philosophy.

If one looks for grammatical information about nouns of emotion in dictionaries and grammars, they will usually provide the information that one should consider them normally as abstract uncountable or non-count nouns, with an occasional countable exception. Quirk et al. (1985: 5.75) go no further than to say:

"Abstract noncount nouns normally have no plural ..... but some can be reclassified as count nouns where they refer to an instance of a given abstract phenomenon......... Many abstract nouns are equally at home in the count and noncount categories."

which is not the most explicit of explanations. However, given the complex nature of the problems involved, perhaps it would be impossible for such a grammar to be more precise without entering areas which go beyond its scope.

Halliday (1984: 161-2) takes a wider view and distinguishes between singular, and plural, and describes mass nouns as either non-singular or non-plural, depending on the system of deictics that are used with them in English. Although this view is fairly suitable for explaining the nature of the patterns that emerge in the corpora, I feel a need to go a little further into what is meant by singular, plural and mass nouns first.

The first type of noun form that is often discussed in an analysis of this sort is that of the noun-as-concept, or what is often referred to as its nominal form. One does not have to be a linguist to talk about the concept of Love, Hate, or Fear, but it helps if one is to understand the full implications of what one is saying. A few examples may help to draw attention to the problems involved:

These examples show that the concepts in the first three examples are (near) personifications and function like names or Proper Nouns. All of the examples show that this form of the noun syntactically requires:

but, despite the counter examples in the sonnet by Camões in which, Amor é um fogo que arde sem se ver, (translatable as Love is a fire which burns unseen), it is normal in modern Portuguese for nouns of this kind, as well as Proper Nouns, to take a definite article - the rules for exceptions to this rule being as relevant for a name like João as for a personification of Amor.

The reason why the verb in these examples is in the singular is because the abstract noun/concept is acting as a proper noun, name or nominal term, and referring, as with Napoleon, Lisbon or the Earth, to something that is singular or unique. This analysis is based on the logic of nouns put forward by philosophers and linguists such as Strawson (1959), Quine (1960), Montague (in Pelletier (Ed.): 1979) and others. Alice ter Meulen in her book on "Substances, Quantities and Individuals" (1980) applies this type of theory to the analysis of mass terms. As part of this analysis she refers to proper nouns as rigid designators. She defines a rigid designator as "a term that refers to the same entity in all possible worlds where that entity exists". She then goes on to demonstrate with examples of substances like gold and water how, if these terms appear in a nominal position unpreceeded by an article (in English), as nominal mass terms, they behave syntactically like proper nouns. This is because "Substances are properties of sets of quantities".

If one applies the same syntactic tests to nominal abstract terms such as love, we obtain very similar results. The main difference between mass and abstract terms is that whereas the former are properties of sets of quantities, the latter are properties of sets of individuals, or, to use Strawson's (1959) term, of individual instances.

It should be noted that Alice ter Meulen is careful to restrict her examples of nominal mass terms to such scientific definitions as Gold is an element and Water is H20. The nearest one can get to this with the concepts under discussion is Love/hate/fear is an emotion, for there are no suitable chemical formulas (yet!) with which we would be happy to define these phenomena. However, the fact that each of us may have a different notion of love, - the 1981 edition of The Oxford Book of Quotations cites at least 34 examples of "Love is..." - does not invalidate the status of these nouns as proper nouns or nominal terms. Each speaker will make his/her definition based on his/her experience in his/her world, and will be speaking in the context of that world. One must not forget that this also happens with proper nouns - Winston Churchill, for example, must have been described in fundamentally contradictory terms by the English and the Germans during the last War, and in everyday life each of us has a slightly different view of whoever is the subject of the conversation, despite that person's apparent reality in the physical world.

Although this usage of emotion words has interesting philosophical implications, the fact remains that they are very rarely used in this way. 0,8%(E) of the total examples for the EC, and 0,6%(P) for the PC could be described in this way. No doubt, if books on the psychology or philosophy on emotion had been included in the corpora, the percentage might have been higher, but if one examines such books one will find that their authors rarely commit themselves to a direct Love/fear is..... definition[1].

Emotion nouns often appear as the subject of a sentence, and on these occasions they appear without an article in English but with one in Portuguese. However, one should make a distinction between the true Nominal, which acts as a concept and is independent of any specific emotional situation, and cases where the SENSER, and often the PHENOMENON, of the emotion acting as Subject is defined in the sentence or immediate context, as in:

(8.6) He was afraid - afraid for Sam, afraid for Sarah, afraid for himself - fear poured like an invisible gas from the mouth of the silent telephone. HF

The SENSER and several PHENOMENA are actually specified here and this, and a number of similar examples, were classified as normal nouns qualified by zero in English, and by a determiner in Portuguese. The argument, like that given for predicative uses of the noun, is that, despite the apparent Nominal or concept function of fear in the sentence, it should be seen as related to specific instances and not to the concept. If we created an example with water like:

(8.7) When the flood reached her town, there was water everywhere - water in the streets, in the house - water poured through the door as she opened it.

The same analysis would be made - not of water as a concept, but of those actual quantities of water, resulting from the rain and causing the flood.

However, although people may argue that there are several propositions underlying the sentence that define different instances of fear - fear for Sam, fear for Sarah and fear for himself - (or water, for that matter), I feel, intuitively rather than logically, that the image of fear as an invisible gas still carries a vestige of some more generic concept, or Nominal, in the surface interpretation. This is probably another example of a gradient in natural language which evades the stricter classifications of logic.

As with the verbs and adjectival constructions considered so far, underlying nearly all uses of Emotion nouns other than true Nominals is the notion of a relation between the SENSER and the PHENOMENON. These nouns are often used in expressions like John's love for Mary, Philip's fear of thunder, the child's happiness at seeing its mother and related constructions, although, in actual stretches of language, the relation is not always so explicit, and often has to be retrieved from the context. The corpora show that the noun groups show a high proportion of PHENOMENON 1 types, 26%(E) and 26,8%(P) as against the corpora averages of 12,2%(E) and 17,4%(P)[2]. This is to be expected because the noun does not oblige complementation in anything like the way verbs and even copula + adjective structures do. However, if one maintains the view that all emotions have some form of PHENOMENON, the notion of the noun as a relation between SENSER and PHENOMENON is self-evident.

Although one may, therefore, put forward the notion that all non-Nominal uses of the Emotion nouns must have some underlying relation in their semantics, this does not mean that this relation is always explicit at the level of syntax. Let us look at ways in which these nouns appear syntactically, first from a theoretical point of view, and then with reference to the corpus.

The examples used to describe the Relation at the beginning of the previous paragraph are like those used by Quine (1960) to exemplify his idea of individual instances of the abstract nouns, or general things, he classified as 'quality- or property-names':

"Thus we may speak of 'the wisdom of Socrates' as an instance of wisdom; of the redness of Smith's face as an instance of redness; and we may also speak of Jones' present mental state as an instance of anger".

He chose redness, anger and wisdom as his examples and this led him to claim that these nouns "are derived from adjectives and the general things they name usually enter our talk by ways of the adjectives from which their names are derived".

One can make the analysis he makes of angry > anger, with other copulas + adjectives, as in Philip was sad to hear the news and the child was happy to see its mother. However, perhaps he had not considered terms like love, hate and desire which can more suitably be described as derived from verbs. This type of noun can be shown to be a nominalisation of the verb phrases, as in John loves/hates/desires Mary. This point is interesting in that it contributes to the notion that there is a close semantic relationship between verbs and copula + adjective constructions. The semantics of both the verb and adjective structures, therefore, are relevant in an analysis of the nouns.

Quine's examples with wisdom and redness are different in one important point from that with anger. Whereas the sentences:

either at the level of the sentence or the context, and can, therefore, be considered semantically transitive. This can also be explained by the fact that, whereas wisdom and redness are qualities or properties that are inherent and permanent, anger is related to Jones' present mental state which is transitory. Jones is angry temporarily, and we assume that this anger is provoked by someone or something.

This is also the case with instances of love, fear, and other emotion nouns, and can be seen in phrases like Jones' anger at Smith's rudeness, although John's love of Mary, and Mary's fear of snakes may last a lifetime. An instance of these nouns can, therefore, be seen to reflect a relation between the SENSER and the PHENOMENON of the emotion. In Alice ter Meulen's terms, this relation is an individual, as distinct from a quantity, and therefore indivisible and unique.

However, Quine's meta-term of 'instance' is possibly more appropriate for an abstract noun, because it lacks the definiteness of individual and allows for the fuzziness which is a characteristic of these nouns. Since there are cases of pluralisation of emotion words as in:

which can be shown to be referring to more than one relation between the SENSER/S and PHENOMENA, it is possible to refer to these nouns in terms of singular and plural. Tests using ellipsis and coordination only serve to emphasize this point[3] and, as we shall see, the corpora produce enough examples to merit consideration of the countability of these nouns.

Further comparison of love with gold and water using the tests of Alice ter Meulen reveal that it can also be used similarly as a bare predicative mass term:

When the noun is in a predicative position, or performing a predicative function, therefore, it can be understood to be similar to mass nouns which do not require a determiner but are quantifiable by much and other expressions. Interestingly enough, the same is true syntactically for Portuguese:

Portuguese also omits articles before mass or mass-type nouns with a predicative function.

It should be noted, however, that, despite the apparent 'mass' usage, there is still an underlying notion of 'relation'. Mary and the children need love and attention from some individual or individuals.

It is not always easy to define a particular use of emotion words as being an instance or a mass term. Let us look at the following examples:

(8.19) a) Fear of spiders, arachnophobia, is believed by some psychologists to be innate.

c) The fear of spiders that Jane had acquired from her mother was something John had little patience with.

f) Peter and Mary's fear of spiders (and snakes) was acquired from their mother.

In these examples the phrase fear of spiders/snakes specifies the emotion and its cause, and it acts as a Nominal in (a), as an unspecified singular noun in (b), as an apparently singular noun specified by a definite article in (c) and by a possessive pronoun in (d). The singularity of these nouns seems to be borne out by the singular verb forms, but perhaps Halliday's term of non-plural might be more appropriate with c) and d) if one wants to avoid a mass definition.

In (e), the ellipsis of her fear of spiders and her fear of snakes does not necessarily give us the plural form:

Similarly, in (f), Peter's fear of snakes and Mary's fear of snakes does not become:

(8.22) ??Peter and Mary's fears of spiders (and snakes) were acquired from their mother.

This suggests that fear of snakes has the mass quality which requires that the ellipsis of John's gold is pure and Mary's gold is pure (unspecified quantities belonging to different individuals), gives us John and Mary's gold is pure, and not *John and Mary's golds are pure. Example g) was included to show that emotions, when 'mixed', convey much the same mass sense as sugar and flour when mixed together in cooking.

This analysis, which works similarly for Portuguese, is based on normal everyday language structure. The fact that a certain ambiguity is involved, and that in most cases there is more than one underlying semantic situation, should not obscure the fact that the way the surface syntax deals with it shows that both English and Portuguese speakers are somewhat ambivalent in their interpretation of these abstract nouns.

The general results for the syntactic groups in the corpora, given in Table 4.2, show that there are 10% more noun forms in the PC, than in the EC, and this point needs to be considered before the more specific details. We have described the semantics of these nouns as relations, which reflect the SENSER and PHENOMENON connection. Although the verbs, adjectives and participles also described the same sort of relation, the focus with verbs and copular structures was usually quite unambiguously on either the SENSER or PHENOMENON, and although they posed more problems, the attributive adjectives and participles described the noun they were tied to.

With verbs, the nature of the SFoc verbal construction made it easy to classify the nature of the PHENOMENON in these examples, and only a small percentage, about 3%, were classifiable as PH. type 1, as against the 12,2%(E) and 17,4%(P) average. The copular structures in the PC were less frequently complemented but the PH. 1 types are fairly near the average. This fact, and that that the PC produced a higher number of intransitive and reflexive structures was interpreted as showing a tendency of the PC to focus the SENSER's internal processes, rather than the nature of the PHENOMENON.

However, unlike the verbs and copular structures, the noun form quite obviously focuses the process rather than the participants, and to describe it as a relation is fairly appropriate. So, if in the other areas it has been noted that the PC focuses the process rather than the participants, it is perhaps natural that the PC noun forms should exceed the EC ones by a large margin. However, the percentage of PH. 1 types - another indicator of less interest in the PHENOMENON - is only slightly higher in the PC, 26,8%(P) as against 26%(E).

When analysing the examples of nouns in the corpus, they were marked for 'singular' when preceded by the indefinite article, as well as expressions like one or only; 'plural' when appearing in plural form; 'determiner' when some form of defining determiner was used; 'possessive' when preceded by a possessive pronoun; 'zero' when no qualifying or quantifying item were used; and quantifiers like some, much and more were recorded as well. In order to simplify this analysis, all optional features of the noun phrase - like attributive adjectives - had to be ignored, despite the influence they sometimes had on the syntax of the phrase as a whole as, for example, when their role was defining.

As the analysis progressed, it became necessary to create a category, described as 'x-of-0', or 'x-de-0', to account for the fact that a substantial number of the examples, 15,8%(E) and 12,4%(P), used phrases which could either be seen as some sort of quantification of emotion, as in a scream of fear, or a mixture of quantification and qualification, as in a sort of fear or a kind of love. The 'x' stands for a variety of possibilities, the connecting preposition is of in English and de in Portuguese, and most, but not all, of the examples take a bare emotion noun.

The different categories were then put into three groups. The first group contained those with overt countability, or those marked for singular and plural, and the very few examples with many and few. The third group contained those with overt non-countability, or those marked with zero or 'x-of-0', and a few examples of quantifiers like much and a little. The middle category absorbed all those examples which belong to Halliday's 'non-plural' categories, which were more often than not of the definite article or possessive pronoun type, or the non-singular quantitative type, like some and any. These examples were all rather fuzzy in nature and any attempt to separate them into countable and non-countable groups would deserve a book to itself.

The results for the two languages for the PFoc nouns were almost uncannily similar, giving approximately 41%, 39% and 20% for the three groups, with a variation of less than 1% between the two languages in all groups. The results for the SFoc nouns showed about 36% for both languages in Group 2, but a variation of about 8% between the other two, leaving the EC with fewer count examples, or 13%, and more non-count ones, 51%, and the PC with more count examples, or 21% and fewer non-count ones, or 43%.

The explanation for the difference in countability between the examples in the EC and the PC is not easy. It may be partly explained by the influence of the fact that the zero qualifier is used in English when the noun acts as the Subject, as in

(8.6) He was afraid - afraid for Sam, afraid for Sarah, afraid for himself - fear poured like an invisible gas from the mouth of the silent telephone. HF

whereas a similar structure in Portuguese would demand a definite article, as is demonstrated by the published translation:

(8.6) Estava com medo - medo pelo Sam, medo pela Sarah, medo por si mesmo - o medo evolava-se como um gás invisível da boca do telefone silencioso.

However, nouns with definite articles were not classified as countable but as belonging to the ambiguous group 2 of non-plural / non-singular, so these structures should swell the ranks of this group and not that of the first one of countables. So there must be some compensating mechanism at work here which allows the numbers in Group 2 to remain the same for both languages for the SFoc nouns, and for the extraordinary sameness of the data for the PFoc ones.

The fact that the results for the PFoc nouns, which refer to the PHENOMENON, are so similar, indicate that there may be some significance in the differences between the SFoc nouns, which focus the SENSER's internal emotional process. If one consults the examples in context, however, the interpretation that results is that the non-count, or zero qualified nouns, are less precise in reference than the countable ones. The former tend to occur with prepositions and to describe behaviour, as in :

(8.24) He was screwing his face up, he was scowling and frowning, and flushing with anger. VW

(8.25) Afonso da Maia ficou diante do filho, quedo, mudo, como uma figura de pedra; e a sua bela face, onde todo o sangue subira, enchia-se, pouco a pouco, de uma grande cólera. M

The fact that the EC seems to favour the less specific or more behaviour orientated usage at the countable/non-countable level is confirmed by its greater tendency to produce examples preceded by prepositions with an adverbial status. The tendency to countability of the PC therefore demonstrates a certain preoccupation with the specific relation, rather than with the more generic behaviour.

This tendency to concentrate on the specific relation in the PC reappears when one examines the degree to which these nouns are followed by complementation. The PFoc nouns, with 12,5%(E) and 11,9%(P), show slightly fewer examples for complementation for the PC, but the SFoc ones with 13,2%(E) and 15,9%(P), show a slight tendency for the PC to favour complementation here. The difference is small, but is probably related to this attitude to specificity.

The majority of complements are of the simple noun phrase variety and are predominately introduced by of in the EC and de in the PC. These prepositions also tend to appear with possessive constructions, and confirm the idea that the explicit relations expressed by John's love of Mary and Philip's fear of thunder are used, but are not often so fully developed at the level of syntax.

When the Emotion nouns appear in adverbial phrases, they assume a more explicitly uncountable function. These phrases are rare with the PFoc nouns in the corpora, 3,8%(E) and 5,6%(P) of the totals, but account for significant proportions of the usage of SFoc nouns, with 29%(E) and 24,2%(P). The lower percentage for the PC has already been referred to as a supportive factor for the suggestion that the PC shows a tendency to focus the relation rather than the behaviour.

The classification of these examples sometimes posed the problem of whether the preposition belonged to the noun following it, or to some other structure, like a phrasal verb, before it. In the first of the following examples the phrase with admiration can be removed without creating a problem of understanding. In the second there is a definite connection between bristled and with which makes it less easy to remove with anger.

(8.27) 'I don't know that I want to do anything with him.' 'Don't you?' he spluttered; his grey moustache bristled with anger. LJ

Since bristled with requires the complementation of anger, however, there is a connection between with and anger which allows this example to be classified as 'with + 0 noun'. It is difficult to make rules about this sort of phrase and these examples had to be analysed individually in context.

The main prepositions used in the EC with SFoc nouns are with/ without which account for 54,3% and in/into/out of which account for 34,5%. The phrases with with /without tend to describe the emotion which accompanies, and is sometimes exemplified in a certain behaviour, as can be seen in the examples above. A similar effect is achieved with the Portuguese com/sem pair which account for 41,5% of the PC examples. In /into /out of tends to give the idea of being in a state of, and refers rather to emotional process or state than to the behaviour resultant from it. The PC examples with em, which account for 24,3%, convey a similar idea. The Portuguese preposition de, standing alone, can often be translated by with. It is often preceded by an adjective or past participle - the most significant of which is cheio - and on these occasions, as with most of the other less quantitatively used prepositions, the translation will vary according to the context. A comparison of the languages here would not prove very fruitful or accurate because of the fluidity caused by the possibility of more than one translation or interpretation in a number of examples.

The two main notions conveyed by the prepositional phrases, however, seem to allow for the idea of behaviour and of emotional state or process. These phrases are important in a consideration of the Emotion nouns, as they account for over a quarter of the noun examples in both corpora.

One of the elements that qualify the nouns in the non-singular and non-plural group is the possessive pronoun. Whereas determiners like the definite article account for 12,4%(E) and 24,6%(P) of all the SFoc noun examples, the possessive element accounts for 17,8%(E) and 10,7%(P). This shows an interesting difference of focus between the corpora, the EC showing a decided preference for possessives and the the PC for determiners. With the PFoc nouns the EC shows a marked reversal of this situation with 26,6% for the determiners and only 10,7% for the possessives. The PC merely shows an added preference for determiners, 33,2%, and even less interest in possessives, 5,2%.

The difference between the languages probably owes a certain amount to different syntactic usage. English uses possessives rather more than Portuguese, particularly when referring to parts of the body[5], and the Portuguese possessive pronoun is preceded by a definite article anyhow. The examples recorded in the corpus, however, included not just those examples using possessive pronoun, but also a few which used the OF or DE genitive.

The possessive notion with the SFoc nouns would seem to show a stronger connection between the SENSER and the emotional process than the more impersonal determiner. Patterns of usage seem a little ambiguous though, and can only be analysed in relation to the different groups. Perhaps another reason why the PC produced fewer examples than the EC is because the TER + noun construction, which is used quite frequently, also contains this possessive notion.

When the SPCp patterns were discussed in 6.4, it was seen that Ranchod (1979) considered the ter + noun structure as similar to the SPCp structures with estar com + noun. Ter + noun falls instead into the SPOd class. Ranchod and others see estar and ter in these sentences as merely carriers of tense, aspect and mood for the noun, but I believe that the use of a have type verb here rather than be is significant. In common with other Romance languages, Portuguese frequently uses the verb ter, or have, + a noun of Emotion where English would use a verbal or be + adjective construction. Also, whereas the estar com + noun structure was represented by a tiny number of examples in the PC, the ter + noun is responsible for 286, or a significant 6,9%(P) of all the noun examples. This percentage goes some way to explaining the higher quantity of noun examples in the PC in relation to the EC.

The use of ter has, no doubt, an interesting history, but I shall not go into that here. As an English > Portuguese bilingual, however, I used to find the use of a have verb with Emotion slightly amusing, although, now that I have examined the problem more closely, I can suggest that the implications can be seen as far from frivolous. In English we can have most kinds of illness - flu, appendicitis, measles, mumps and a nervous breakdown - as we can with ter in Portuguese. One of the lexical areas Ranchod (1990) deals with exhaustively is that of illness, showing how these lexemes function with both the estar com and ter structures. When it comes to Emotion, though, the Portuguese can also ter (have) an Emotion, and estar com + a noun like medo/vergonha etc. (*be with fear/shame etc.). Apart from rare have constructions, like have pity, English either translates the ter/estar com + Emotion noun constructions with a SFoc verb, like fear or hope, or, more frequently, with be + adjective, as with be afraid/ashamed/jealous.

To minimise the semantics of ter and be by saying they are only space fillers between the SENSER and the Emotion is to dodge the issue. Be is felt by many English speakers to have a very fundamental meaning, and English-speaking psychologists assume a certain identification of the Self with a state of Emotion. I just wonder if Portuguese psychologists or doctors feel the same. If they are speaking a language which uses the same syntax for illnesses as it does for emotions, do they tend to feel that emotions, like illnesses, are temporary nuisances that will go away if properly treated? Or have I got it wrong, has ter here lost the force it has with more material things, like money and sports cars - although the syntax is the same, and do they somehow see illnesses as associated, like Emotion, with the state of being ? I have no statistics to show the relative frequency of the ter and estar com + nouns with examples describing illness, but it would be interesting to know which prevails.

In any case, I think that one could say that the ter + noun construction does provide for an idea of distinction between the SENSER and the emotional process at a syntactic level which is not evident with the be + adjective one. Ter and have may only be a little way further down the Stative > Dynamic gradient than be, but it is far enough to make the difference between the notion of identity claimed for be/ser/estar, and the distinction between possessor and possessed claimed for ter/have.

In Portuguese, the verb ter combines with nouns of Emotion like medo, vergonha, ciúmes or pena, and the PC show that most of these nouns, 78% of them, are preceded by zero qualifiers. There are a few examples of ter with the plural, particularly with examples like ciúmes as a plural form of ciúme, and esperanças, and some are quantified with mais and , depending on the possibilities of the context. Some of these nouns can take an indefinite article, particularly in rather emphatic usage, as in:

(8.28) Além do mais, Diamantino tinha um medo pavoroso à trombose das coronárias com que o ameaçara, a certa altura .da vida dissipada, o famoso lente da Faculdade de Medicina do Rio Dr. Algorindo Ariosperis. AQ

This structure is an interesting facet of Portuguese usage with this lexicon, and it again highlights a certain preoccupation with the idea of process, or a distinction between SENSER and process, which seems to be a feature of the PC.

[1] For example, Fridja (1986) makes statements like:

"Love and affection refer... to that urge toward..." (p.83)

"Hatred is an emotion that contains the component of object evaluation".

which show that he is conscious that he is talking about the term rather than any reality, although he only makes this clear, by using italics, in the second example.

[2] Since the number of noun examples overall for Portuguese is about 10% higher than English, it is possible that the higher percentage of Phenomenon 1 types overall for Portuguese may well be influenced by this factor.

[3] In the case of the examples given we can analyse the first example as referring to:

His (love for Mary + love for Susan + ....) and (hate for John + hate for Jane + ...) are no affair of mine.

(Mary's fear for my safety + John's fear for my safety + .... ) were unjustified.

[4] One could, of course, say her book on spiders and snakes, meaning a book which dealt with both topics, but this would not be a case of ellision. One must also accept that her books on spiders and snakes has other interpretations than the type of ellision proposed here - she may have a whole collection of books on these topics - but that does not affect the type of ellipsis proposed here.

[5] For example, in English one says: "He hurt his arm", and in Portuguese: "Magoou o braço".