“Who’s afraid of ... what?” – in English and Portuguese

Belinda Maia, University of Porto
Diana Santos, University of Oslo


Fear is generally accepted as a primary emotion in studies on the relationship between cognition and emotion. In this paper we shall use corpora to explore ways in which the English and Portuguese languages reflect the use of language to express this emotion and its relation to cognitive processes.

The paper starts by positioning this study of emotion as an area that can be observed from a viewpoint between the extremes of linguistic universals and relativism. This is followed by a short description of the aspects studied in both languages, and our attempt to identify meta-patterns of the fear lexicon and its usage in context.

Building upon Maia (1994), which contrasted the emotion lexicon in context using comparable corpora of fiction in the two languages, we investigate the subject as follows:

  1. A brief reference to the insights gathered in 1994;
  2. An examination of the patterns for English in the British National Corpus (BNC) and other publicly available corpora, using the major items of the fear lexicon and part-of-speech clues;
  3. An examination of the patterns for Portuguese in the Linguateca AC/DC corpora, using the same clues, but with the added benefit of automatic annotation of a much wider lexicon of fear.

The results from the corpora are not easily comparable because the corpora and the tools available to analyse them are different. We accept that this somewhat limits the contrastive analysis of the two languages, but we feel that it is still possible to demonstrate some interesting points and indicate where the differences can point to further research.

1. Introduction

The research by Eckman (1982) and others in philosophy, psychology and neurology, indicates that fear is generally accepted as a primary emotion, essential for survival, and the physical reactions, both external – facial expression, and internal – blood pressure, release of adrenalin, and associated physical reactions, are classified as universal in human beings and other animals. Eckman’s research, best known today through the popular television series Lie to me, pioneered the mapping of human emotional facial expression. He drew attention to the universality and involuntary nature of human emotional responses, albeit with certain cultural differences when the subjects were aware of being observed, and the more serious research resulting from this has led to lie-detection technology that observes facial expression and body language. It follows that, if fear is a universal of the human condition, it is an excellent testing ground for the nature-nurture debate from a linguistic point of view.

In the 21st century, the extreme views in the nature-nurture debate in linguistics are becoming more moderate. Pinker (2007) suggests that one should aim for a balanced analysis somewhere between the Extreme Nativism proposed by Fodor (1998), the Linguistic Determinism of Whorf (1956) and the Radical Pragmatics of such as Sperber & Wilson (1986). Coming from the opposite side of the spectrum, Deutscher (2005 and 2010) both draws attention to similarities between languages, and argues convincingly for a more scientific, but moderate, view of the Whorfian hypothesis. Our work here aims, too, for a consideration of both sides of the argument.

Maia’s interest in the 90s was inspired by the nature-nurture arguments of that time, but also by the debate on the relationship between Mind and Brain and the role of emotion in our cognitive processes. The best-known work of this period is perhaps Damásio’s Descartes’ Error (1994), although this book was the synthesis of much of what had gone before in the 80s and 90s. The general idea emerging from this kind of research was that emotion, far from being an irrational aspect of human behaviour, was intimately involved in most of our cognitive processes. This aspect also needs to be considered by those studying the use of the fear lexicon, and it would appear to be similar in both English and Portuguese.

Another aspect that needs to be taken into account is that the fear lexicon is not always used for its primary purpose of expressing fear. It is also used in context in areas that are contingent, rather than central, to fear, such as the relationship between fear and regret, two types of emotion, and fear and courage, where the emotion is related to a moral judgment. This is an aspect of the lexicon that needs to be examined in greater depth than is possible here, but an analysis of the kind applied by Deutscher (2005) to the relationship between time and space implicit in actual language usage could also be applied to the close relationship between emotion, cognition and other semantic areas to be found in many of the examples in corpora.

As we shall see, English and Portuguese have a lot in common in relation to the underlying situational and semantic conditions in which we express emotion, and we shall start by pointing to the aspects that can be considered part of the general human condition. However, we shall also draw attention to the finer points of difference that a corpus-based study can provide. As bilinguals with a strong interest in translation, we are particularly aware of the relativity of language, and the way in which it is a window into the way we see the world through our language and its culture. [1] More pragmatically, an intercultural analysis of the attitudes and “common sense” in different languages may help prevent or solve conflicts or misunderstandings when communicating with people of different cultures or languages.

A wider study would observe the expression and objects of fear in different cultures and different languages. Such an empirical study may uncover the fact that different cultures usually fear different things, or rank differently the same fears, in terms of intensity and social acceptance. For example, it is often mentioned that fear of losing face is a typical Asian (cultural) trait and it would be interesting to see if European languages reflect similar or different values.

2. The Language of fear – emotion and cognition

The classifications of the emotion verbs proposed by linguists following Vendler (1957) and those responsible for English grammars, usually include them under the “state” or “stative” label, which also includes several more “cognitive” verbs. The arguments for “stative” verbs are that what we shall call the Senser-as-subject verbs do not “like” the imperative or progressive forms in English. However, the imperative, especially with negation, of some emotion verbs is possible, or can be expressed, as with Don’t be afraid, Fear not and Não tenhas medo or Coragem!. Besides, as we shall see below, there are two kinds of verbs working with fear, and the Phenomenon-as-subject verbs frighten or assustar can use the progressive in both languages.

One of Maias original objectives was to test how far the aspectual classification was actually verifiable in corpora, and the study of the verbs inevitably led to the analysis of other parts of speech as well. Santos (2004) in independent work has pointed out the English specificity of Vendler’s classes and proposed a novel one for Portuguese, keeping the methodology but applying it to Portuguese grammar. She classifies assustar and aterrorizar as obras, while ter medo or temer are estados temporários.

The language of emotion is considered important enough to justify projects such as the Languages of Emotions at the University of Berlin and journals such as Cognition and Emotion, published by Taylor and Francis. Lakoff and Johnson’s (1980) emphasis on what could be deduced about human emotions based on metaphorical usage and physical description popularized the study of expressions of fear like his hair stood on end, she screamed, they trembled, the colour drained from her face, his heart pounding against his ribs and his mouth suddenly dry..., and this type of research has been more popular than actually looking at the use of the emotion lexicon. It is also an area that would probably render some interesting contrasts between languages, but it is far more difficult to study this area using a corpus than that based on the lexicon.

However, not everyone has ignored the emotion lexicon, and Ortony et al (1990) in The Cognitive structure of the Emotions, created a semantic structure for the emotions that could be supposed to exist, and which were lexicalized in several, if not all, languages. This classification has been the basis of several studies of emotion. The authors explained each area as neutrally as possible, and fear was described as being [displeased about the prospect of an undesirable event], and the variables affecting the intensity of the emotion were ‘the degree to which the event is undesirable’, and ‘the likelihood of the event’.

The relationship between emotion, cognition and the language of fear is complex but, however irrational a particular fear may seem, the fact that it is usually projected on to some future situation, even immediate ones, demonstrates that cognition is involved. As with the other emotions, Ortony et al’s variables account for a gradient from clearly immediate situations, such as terror on meeting a tiger, through a variety of less frightening situations. fear can be a latent tendency and only activated in certain circumstances, such as a fear of snakes or thunder. Wittgenstein (1953: 135e) distinguishes between the object of fear, or a property inherent to an individual as in fear of darkness and medo do escuro, and the cause of fear, which causes a temporary psychological event, such as fear on entering a dark room and medo por entrar no quarto escuro. [2]

fear can be caused by easily understood circumstances, like a fire, and highly complex ones like the financial situation of the Euro today. Lexically weaker expressions such as anxiety, worry, timidity and shyness represent concepts related to fear, as anyone in a difficult situation or suffering from social inadequacy knows too well. At the lowest end of this gradient, but important quantitatively, is the use of the fear lexicon in polite expressions of regret such as I am afraid I can’t meet you today.

The syntax of fear verbs is, generally speaking, typical of the “state” verbs mentioned above and, for the semantics of case, Halliday (1994) provides possibly the most perceptive designation with Senser for the person who feels emotion and Phenomenon for its cause. The Senser is essentially a human being, possibly an animal, or something perceived as such using metonymy, or a collection of such beings, such as the people, the government or the state. The Phenomenon can be anything that leads to an emotional response. In this way psychology is reflected in linguistic theory, because fear is “internal” to the Senser and the Phenomenon has no real “active” role. Spiders do not act as agents to frighten you, and even if someone deliberately tries to frighten you, their success depends on you – not them.

3. The lexicon and syntax of fear

Maia (1994) used Ortony et al’s (1990) classification and confirmed that it was very useful for the study of the lexicon of emotion in a corpus of literary texts in English and Portuguese. An analysis of 25,000 examples using the emotion lexicon at textual, syntactic and lexical level provided an insight into the way emotion is expressed in language, and a methodology that can be used with other corpora and for further research. She found that at a textual level the same gradient of intuitive > reasoned explanation > politeness is very similar in both languages. However, there are interesting differences at a syntactic level, and the lexical level presents the usual problems of translation in context. One should also note that the central lexemes like fear use different syntax to those on the periphery, like worry.

Both languages have central nouns in the fear lexicon, for example fear, dread, foreboding, panic and medo, pavor, receio, temor, pânico. The interesting point is that, even when a verb form of these words exists, the noun form is more commonly used. For fear, for example, the BNC registers 9,006 noun forms and 5,117 verb forms, and the Linguateca corpora record 10,262 examples of receio as a noun form and 4,848 of the verb recear. In most of the noun form cases, the Phenomenon is clearly defined in the context. The Portuguese medo, for reasons we shall discuss later, is also used very frequently, but it has no verb form.

Halliday’s distinction between Senser and Phenomenon proved to be crucial to analysis at all levels, but particularly as the starting point of analysing the lexicon at verb and adjective level. The verbs in the lexicon can be divided into two types. In the first the Subject of the active form is the Senser, like fear and dread, and recear and temer. These verbs occur quite often but, in both Portuguese and English, the usage rarely refers to immediate real fear. The most frequent usage reflects varying degrees of regret and disappointment, as in I fear the result will be boring, defensive cricket (BNC). A further point is that the examples found with fear and receio / recear are usually followed by a cognitively analysed explanation of what is feared. The use of both the central nouns and verbs with the Senser as Subject would suggest that this usage is a more cognitive than emotional usage, as Maia’s (1994) results suggested and is confirmed by analysis of contemporary corpora. This would make sense psychologically because the speaker / writer needs to process the situation mentally in order to be able to name an emotion so clearly.

The verbs with Phenomenon as Subject are those like frighten and terrify and alarmar and assustar. The interesting point about these verbs is that the active use is rare, particularly in English where the vast majority of examples are represented by a past participle after the verb be. However, it is here that the usage in English and Portuguese begins to differ, because Portuguese can use the usual Phenomenon-as-Subject verbs like alarmar with what some call the se passive or reflexive passive. In this situation o João assustou-se is literally translatable as *John frightened himself. This structure is quite common in European languages and it reflects the sense that the emotion is internal to the Senser.

Another obvious difference is shown up by the fact that be afraid is the major expression of this emotion in English, whether it is afraid of tigers / thunder / spiders or the more reasoned afraid that she will not come tonight. Afraid is classified as an adjective but it is never used before a noun and has an odd etymology. [3] And there are other adjectives in the fear lexicon – anxious, nervous, timid, and ansioso, nervoso and tímido, which can be analysed on a gradient from temporary condition – anxious or nervous, to character trait – nervous or timid.

Except for afraid, most of these adjectives, like anxious, or past participles, like frightened, may precede the Senser syntactically in English, or follow be. In Portuguese, words like ansioso and assustado can be part of the noun phrase, but the situation is more complex with copula verbs as the sense changes according to whether these adjectives / past participles follow ser, which will imply a certain permanent characteristic of the Senser, estar, which refers to a temporary state, and ficar, which refers to a resulting state.

For example, see:

(1) Quando a atriz Roberta Indio do Brasil, 21, deparou-se com a sinopse de “A Viagem”, que previa uma gravidez e em seguida um aborto para sua personagem, ficou assustada.
‘When the actress Roberta Indio, 21, from Brazil, noticed that the script of “the  Journey” included a pregnancy and an abortion for her character, she was frightened.’
(2) Os portugueses estão assustados com o desemprego e acham que a economia irá de mal a pior.
‘The Portuguese are frightened and think that economy is going to become worse’

The other Portuguese structure that reflects this internal sense of the emotion is the very frequent use of the verb ter, literally to have, + the nouns of fear, medo, pavor, temor and susto. It is possible to use have with some words in the English lexicon as in I was surprised to hear you have a fear of drowning (BNC), but the examples are rare. Quantitatively, to be afraid of is most normally translated by ter medo de, but to be afraid that prefers the expressions that convey a greater sense of cognitive processing, or ter receio / recear.

Finally, there are the adjectives that describe the Phenomenon, alarming, frightening, terrifying, and alarmante, assustador. Although there are cases of people actively making the Phenomenon cause an emotion, the actual reaction by the Senser is not necessarily predictable, and this evaluation of the Phenomenon is always internal to the Senser.

4. Semantic roles and syntactic clues

In the early 1990s it took several months to manually extract and classify 25,000 examples of emotion words (for fear, 1,303 in English and 1,392 for Portuguese) from two literary corpora in English and Portuguese of about 1 million words each. The Phenomenon in each example was classified according to the following set of possible causes of emotion:

  1. Unknown, or unspecified in the immediate context
  2. Self, or permanent quality of SENSER
  3. State or situation of SENSER
  4. Emotion, perception or cognitive processes of SENSER
  5. Action by SENSER
  6. The Other
  7. State or situation of the Other
  8. Emotion, perception or cognitive processes of the Other
  9. Action by the Other
  10. A non-human object, concrete or abstract.
  11. A complex proposition about the world

This classification reflects a certain gradient from fear of an unidentified Phenomenon, with words like foreboding or uneasiness, through states like timidity, to people, their emotions and actions, to a fully expressed and understood reason for fear.

Syntactic clues were also used to discover the identity of the Phenomenon by classifying examples according to the complementation following the verb, adjective, past participle or noun, and it was found that the following list would also help identify the cognitive appraisal of the cause of emotion:

a = noun phrase
b = non-finite infinitive clause (S = same as main clause)
c = non-finite infinitive clause (S = different from main clause)
d = non-finite -ING clause (S = same as main clause)
e = non-finite -ING clause (S = different from main clause)
f = finite (THAT) / QUE clause (S = same as main clause)
g = finite (THAT) / QUE clause (S = different from main clause)
h = finite WH-/ O QUE clause.

It is important to understand that the usage of the fear lexicon and related syntax varies considerably according to text type. Unlike Maia’s original literary corpus, the BNC contains transcribed oral texts as well as written texts from a variety of areas. The results, therefore, will be different. Mark Davies’ version of the BNC at Brigham Young University allows one to gauge usage in seven major text types, Spoken, Fiction, Magazine, Newspaper, Non-Academic, Academic and Miscellaneous, and each text type can be divided into more specific areas. If one looks at a corpus of formal texts, like the European Commission corpora, or the texts from the European Central Bank in the OPUS corpora, one will see fear as being well defined cognitively and related to situations affecting the many rather than the individual, and very often used as a formal expression of regret. Oral texts, perhaps surprisingly, do not yield many real examples of fear. Perhaps this is partly because the identification of fear needs cognitive processing, and partly because conversational contexts avoid such identification. These factors need to be taken into consideration when discussing results from corpora, and the results for Portuguese in the Linguateca corpora (see below) demonstrate this clearly

Another factor that needs to be remembered is the psychological reality of fear as deduced from corpus data. Maia’s (1994) original analysis, done “manually”, allowed the possibility of discovering how the examples with the fear word functioned in context, sometimes beyond the sentence. These results are difficult to repeat with a large corpus, but we shall include the more interesting ones here, if for no other reason than to show the need to be wary of comparisons based on corpora.

Maia found that the majority of examples with fear, using all the lexical items studied, showed that 84.4% in English, and 78.7% in Portuguese, focused on the lexical items describing the Senser’s emotion, for example to fear / to be afraid, rather than on those, like frighten/ing that focus on the Phenomenon that caused it. This would seem to demonstrate the psychological reality that emotions like these are more important from the Senser’s point of view, particularly in a literary corpus.

The other point that emerged was that the Phenomenon type 1 referred to above – Unknown, or unspecified in the immediate context – accounted for 32.1% of the examples in English and 39.7% in Portuguese. This might indicate the need, in a literary corpus, for the reader to understand the reason for the emotion from the much wider context of the narrative, or perhaps it is a device for creating suspense. Phenomenon types 9 – action by the Other and 10 – a non-human object, concrete or abstract – accounted for 30.3% of the examples in English and 34.1% in Portuguese. Again, these results are probably valid for a literary corpus, but a corpus of formal texts, like the European Parliament proceedings, will provide far more examples of Phenomenon types 9 and 10, as well as 11 – a complex proposition about the world – and very few of Phenomenon type 1.

5. Who is afraid of what – in the British National Corpus?

Although the BNC is annotated for parts-of-speech and would allow one to find certain syntactic patterns semi-automatically, an exhaustive search with the whole lexicon for the purpose of this paper, would require more time and space than we can afford here. For the analysis of English, we shall therefore concentrate on the main means of expressing fear, fear as a Senser focused verb and noun, afraid and anxious as Senser focusing adjectives, and various forms of frighten and terrify as Phenomenon focusing. These words would seem to be central to fear but, as we shall see, their use in context only partly expresses fear, and sometimes favours interpretation as regret, sympathy and even desire. The more we move from these “central” words to the periphery, the more variable the syntax and mixture of emotions. For example, worry and preocupar-se, cognitively processed forms of fear, use the progressive quite frequently.

Fear/s as a noun produces 9,006 examples, 28% of which appear with of, or fear/s of, followed, in order of frequency, by death, God, crime, failure, unemployment, violence, persecution, flying, attack, loss, rape or general public rather than “individual personal” Phenomena; by something more vague as with the + unknown, dark, consequences; and by some happening such as being, losing, falling. The noun forms are also followed by that (10%) and pronouns (6%), indicating complementation by a finite (that) clause in which the subject is different from main clause. A further 8% appear coordinated with another noun of emotion or a negative situation, a common occurrence with emotion words, which favour combinations, both negative and positive, as a way of describing “mixed emotions”. Behaviour is indicated by the 6% preceded by in or with.

As a verb, the various forms of fear showed the Present tense dominating with 35%, 13% of which were preceded by a pronoun; the infinitive was used in 12% of the cases and the 3rd person singular in 10%; the past tense accounted for 28%, 11% of which were preceded by a pronoun. Feared was considered a past participle in 13% of the examples with a significant number of examples such as it is / was feared, or “general public” type Subjects. The use of pronoun + fear followed by punctuation was found for 5%, as in It will be painful, I fear (BNC), as an expression of polite regret or sympathy.

Interestingly, the BNC also has feared classified as an adjective as in None of the feared Mediterranean gales had sprung up. (BNC). The search with fear* will also bring us fearsome (238) and fearful (693), although the vast majority, like other similar emotion adjectives, are used for emphasis rather than as independent lexical items.

The Phenomena found with fear as a noun cover the whole gradient described above, with a greater or lesser emphasis on cognitive processing depending on the text type. Fear as a verb appears largely in more formal texts and favours cognitively complex Phenomena. As a general rule medo and ter medo will translate all these uses, but receio and recear will be preferred in more formal situations. An examination of the political or governmental texts in the OPUS corpora will confirm this.

Afraid is by far the most popular fear word in the English lexicon, apart from fear itself, with 5,558 examples. It appears with 33% of the examples indicating the Senser’s fear of a wide variety of objects, but also the Senser’s own actions, or reluctance to perform these actions. It can be followed by of + being (+ a wide variety of verbs in the past participle with negative implications), asking, losing, getting, or to + ask, say, be, tell, come, admit. In 34% of the examples afraid appears before a (that) clause, which implies more cognitive processing, particularly if the subject of the (that) clause is different from that of the main clause. As Maia (1994) discovered, a (that) clause will necessarily include some form of explanation of the cause of fear. However, the meaning of afraid here is often diluted and is used more as a politeness routine than to express real fear.

Confirmation of the status of afraid as part of a politeness formula can be found in the 25% of the examples that appear at the end of the sentence or clause, as in He is not coming, I’m afraid, or I am afraid (not/so). As with the examples above with fear, these are polite expressions of regret or sympathy added after an unpleasant fact or opinion has been expressed in conversation, and will rarely be found in texts that are written, except as examples of direct speech in text, or transcribed oral texts, such as those in the BNC.

Anxious, with 2,951 occurrences, appears to be a combination of fear and desire when used predicatively, as can be seen from the way anxious to (44%) combines largely with get, see, avoid, be, make, find, have, keep, know, please, do. However, fear is implicit when it is used attributively, as in anxious + moments, face, parents, eyes, wait, state, expression (10%).

Searches with frighten* and terrify* will produce 4,147 and 1,754 examples, respectively. The active verb forms in the present, infinitive, past and “gerund” form account for 18% / 15% of the examples, but a considerable proportion favour x frightens /terrifies + pronoun, which focuses the Senser’s reaction rather than any agentive action on the part of the Phenomenon. Frightening and terrifying as adjectives account for 25% / 28% of the examples, frightened / terrified as adjectives 47% / 46%, and frightened / terrified as a past participle, 10% / 11%. The noun fright is not used that frequently and tends to belong to phrases like take fright or give x a fright, while combining elements of fear with shock or surprise. However, the overall result is for the focus to be on either the Senser’s resulting state or his/her perception of the Phenomenon.

This quick analysis of the principal fear words in the BNC is based on Maia’s approach to analyzing the syntactic patterns that might indicate the psychological reality expressed by Halliday’s semantic roles of Senser and Phenomenon. The results would seem to confirm the work done almost 20 years ago, but a thorough analysis would need time and the processing and annotation of English corpora with the fear lexicon such as that described in the following section on Portuguese.

6. Who’s afraid of what – in Portuguese

This part of our article is able to look at the same aspects of fear as those described above with the BNC, but we can use a more varied and morphologically detailed approach because the Linguateca corpora are much larger, some individually and certainly on aggregate, and the linguistic analysis of these corpora is finer-grained. In order to be able to provide the results described below, one must explain that the AC/DC corpora (Santos 2011, 2012) are not only annotated for parts of speech as in the BNC, but have also been parsed automatically by PALAVRAS (Bick 2000), turned into the Linguateca format, and then automatically tagged with the fear domain, grouped under the general concept of medo. As with other semantic domains, this annotation will be humanly revised at a later stage (see e.g. Santos et al., this volume), but has not yet taken place for fear. Until this revision has taken place, it is only natural that certain lexical items that have been included may turn out, on further observation, to be peripheral to fear, or simply be homonymous forms (as with aterrar, also meaning “to land”).

Maia (1994) followed Ortony et al’s example to exclude words related to behaviour, like shiver or arrepiar because such physical reactions were not restricted to fear. Similarly, concepts like cowardice or courage were excluded on the basis that they were felt to describe moral judgment on behaviour, rather than the central emotion itself. Besides, there is no comparable English verb for the Portuguese acovardar (cf. cowardice), and, although encourage or encorajar may derive from the same source as courage or coragem, the relationship to fear in context is tenuous.

However, the annotation of the Linguateca corpora to allow for analysis using these areas of the lexicon suggests that there is plenty of room for expansion from the central concept of fear.

6.1 General results for fear in the corpora and according to genre

The table below describes the 88,605 occurrences of expressions using the fear lexicon in all the AC/DC corpora, both in terms of absolute number of occurrences and relative to the size in words of the respective sub-corpora. Although the reader will need to consult the AC/DC site of Linguateca to learn more about each individual corpus, we present below a short description of the genres and variety of the material as included in the “complete corpus”.

Corpus Fear Words Genre Ratio per 10,000 words
CETEMPúblico 37,297 141,643,293 newspaper 2.633
CHAVE 27,876 104,668,900 newspaper 2.663
VERCIAL 13,021 15,736,845 fiction 8.274
NILC/SãoCarlos 2,262 7,176,448 balanced 3.152
FLORESTA 1,824 6,367,935 blogs 2.864
DIACLAV 1,790 6,779,705 newspaper 2.640
CONDIVport 1,789 5,693,970 newspaper 3.142
AVANTE 1,625 6,997,498 newspaper 2.322
ECI-EBR 427 776,928 balanced 5.496
Diário do Minho 250 1,816,653 newspaper 1.376
Museu da Pessoa 134 406,492 interviews 3.296
ANCIB 129 1,364,356 mail 0.946
CONE 64 736,216 mail 0.869
AmostRA 58 105,015 balanced 5.523
CDHAREM 47 240,921 div. 1.951
ENPC 38 77,737 fiction 4.888
FrasesPB 8 19,893 div. 4.022
FrasesPP 4 16,976 div. 2.356
ECI-EE 0 28,263 agreement -

Table 1. AC/DC corpora in a nutshell.

The picture that emerges shows clearly that there is a gradient from original fiction > translated fiction > oral interviews > blogs > newspaper text > informative prose and mail.

In Table 2 we describe the distribution in terms of the different lexical items and constructions using a cutoff frequency of 7. [4]

Word Frequency Word Frequency
medo 21,727 temerário 534
receio 10,262 sobressaltar 534
terror 5,790 aterrador 459
assustar 5,087 amedrontar 383
recear 4,848 apavorar 368
pânico 4,094 medroso 353
susto 3,200 apavorado 215
aterrar 2,993 arrepiado 181
temor 2,834 amedrontado 154
sobressalto 2,113 co[bv]ardemente 153
assustador 1,744 pusilânime 111
alarmante 1,272 timorato 110
co[bv]arde 1,256 arrecear 54
receoso 1,123 aco[bv]ardar 97
assustado 1,062 temente 43
temível 1,004 mêdo 29
alarmar 963 antiterror 17
apreensivo 918 terrificar 11
aterrorizar 789 cobardola 8
temeroso 771 temerosamente 8
arrepiar 627 arreceio 7

Table 2. fear words (lemmas) in Portuguese in decreasing frequency order

In order to investigate more thoroughly the importance of genre in the fear domain, we also looked at its distribution in two balanced corpora, namely NILC/São Carlos and ECI-EBR, both containing only Brazilian material. We provide the ratio of fear words per 10,000 running words, and the number of different fear lemmas in each section.

Genre in NILC Fear Words Ratio (x 10,000) Lexical variety
newspaper 8,150 29,821,714 2.733 61
fiction 834 921,368 9.512 44
essay 429 2,193,638 1.956 32
magazine 57 153,786 3.706 20
textbook 52 426,766 1.218 14
legal 34 1,111,864 0.306 4
encyclopedia 32 286,559 1.192 17

Table 3. Distribution of fear words according to genre in Corpus NILC/São Carlos

It is quite obvious that legal text is different, as it not only shows far fewer cases, but 11 of the fear occurrences in legal text corresponded to a technical meaning of temerário. Also, as already noted, fiction is at the opposite extreme, with almost one word per thousand words denoting fear.

If we look now at ECI-EBR, a much smaller corpus with a very different choice of texts and genres (see Table 4), we see that, again, and rather surprisingly, the oral part does not bring as large a quota of fear as expected, probably because it is formal oral text and not colloquial. Although one should not put too much emphasis on data from such a small corpus, one should note that essays are also low in fear words, while drama is the highest fear-containing material.

Genre in ECI-EBR Fear Words Ratio per 10,000 words
fiction 219 241,171 9.080
drama 123 129,640 9.488
textbook 42 203,436 2.064
news 20 88,946 2.248
unknown 8 16,909 -
political speeches 7 47,588 1.471
parliament debates 4 15,277 2.618
literary criticism 2 15,162 1.319
essay 1 14,150 0.707
interviews 1 3,226 3.100

Table 4. Distribution of fear words according to genre in Corpus ECI-EBR

6.2. What is it that people are afraid of in Portuguese?

The Linguateca corpora allow one to count the distribution of the fear lexicon in terms of frequency of parts of speech and syntactic patterns. Just like for English, the noun forms of fear are more frequent, 50,623 examples, compared with the 27,179 verbs, and 10,763 adjectives. And there was only a very small number of adverbial forms, 16.

In the case of nouns, we can also provide partial counts in terms of the Phenomenon: fear of + N or PROP (9,097), fear of + infinitive (3,891), and fear of que-clause (that-clause) (1,976). This would confirm a point made earlier that, by using the fear lexicon, the speaker or writer has cognitively processed the emotion on a gradient from a specific object, to an action by the Senser, an action by the Phenomenon, or a cognitively analysed situation.

The Phenomena found with temer, (ar)recear would seem to be similar to those described as “public” for the English fear of, as can be seen from the list in Table 5:

What is feared Gloss Frequency What is feared Gloss Frequency
mau bad/evil 248 agravamento worsening 37
represália reprisal 170 perigo danger 36
efeito effect 137 fim end 35
consequência consequence 110 problema problem 35
ataque attack 83 nada nothing 31
concorrência competition 81 intervenção intervention 30
aumento increase 69 repercussão bad consequence 29
perda loss 63 vitória victory 29
futuro future 63 reação reaction 28
invasão invasion 58 impacto impact 26
morte death 57 resultado result 26
confronto confrontation 55 atentado attack 25
ameaça threat 50 comparação comparison 25
reacção reaction 48 poder power 25
repetição repetition 48 ocorrência occurrence 24
coacção coercion 44 pressão pressure 24
regresso return 43 vaga wave 24
subida raise 40 guerra war 24
explosão explosion 40 mudança change 24
risco risk 40 violência violence 23
retaliação retaliation 39 falta lack 23
possibilidade possibility 39

Table 5. Distribution of objects of fear verbs in Portuguese

Similarly, Phenomena following a support verb with medo or receio are displayed in Table 6.

What is feared Gloss Frequency What is feared Gloss Frequency
represália reprisal 253 doença illness 30
morte death 204 autoridade authority 30
intervenção intervention 126 ataque attack 29
subida raise/increase 66 coisa thing 29
população population 61 sida aids 28
polícia police 60 inflação inflation 28
retaliação retaliation 58 mulher woman/wife 28
guerra war 52 pressão pressure 27
consequência consequence 49 verdade truth 26
reacção reaction 48 perseguição persecution/chase/harassment 26
violência violence 47 assalto assault 26
aumento raise/increase 47 eleição election 26
concorrência competition 42 água water 25
perda loss 42 explosão explosion 23
união union 39 invasão invasion 23
crise crisis 39 falar talk 22
mudança change 39 confronto confrontation 20
atentado attack 38 derrota defeat 20
palavra word 38 corte cut 20
pessoa person 37 agravamento worsening 19
vida life 34 vingança revenge 19
ameaça threat 31

Table 6. Distribution of objects of nouns of fear in Portuguese

Note that these numbers were not subject to human screening, and the actual pattern in structures such as noun phrases with de may produce both reference to the Senser or the Phenomenon as in (3) opposed to (4):

(3) Um diplomata ocidental comentou que os consumidores têm razão para ter medo da inflação, porque ainda guardam na memória os tempos em que passaram fome.
‘A Western diplomat commented that consumers are right to be afraid of inflation (have “fear of inflation”), because they can still remember times when they went hungry.’
(4) A frase, proferida ontem por Ieltsin no penúltimo dia do Congresso dos Deputados do Povo, marcou um dos momentos altos da sessão, ao tranquilizar muitos receios das vizinhas repúblicas.
‘Yeltsin’s statement, during one of the last days of the Congress of the People’s Representatives, was one of the high points of the session, and calmed the fears of neighbouring republics.’

Interestingly, receio shows a much higher proportion of cognitively processed Phenomena than medo, and susto de behaves similarly to fright (see above). Also, de, like of in similar circumstances, may signal still other constituents, such as temporal qualification, as illustrated by examples (5) and (6). De can also indicate the kind of fear, as in examples (7) and (8).

(5) Na passada quinta-feira, dia 24, apanhou, com toda a certeza, um dos maiores sustos da sua vida.
‘Last Thursday, on the 24th, he got what was certainly the greatest fright of his life’
(6) O escocês, recomposto do susto da véspera – foi desclassificado.
‘The Scot, having recovered from the fright of the day before, was ...’
(7) Pacheco Pereira apanhou um susto de morte quando soube do naufrágio do Estónia, ao largo da Finlândia.
‘Pacheco Pereira had a (deathly fear) terrible fright when he heard of the shipwreck of the Estonia, off the coast of Finland.’
(8) Mas foi susto de pouco fôlego.
‘But the fear was unfounded. (*little breath fear)’

We can also find examples following medo similar to those with the English fear of and afraid of, that indicate fear of an action, expressed by a verb, as shown in Table 7.

What is feared Gloss Frequency What is feared Gloss Frequency
ser be 657 falhar fail 34
perder lose 414 poder be able, unadvertently 33
ver see 157 cair fall 32
morrer die 140 tomar take 32
ficar stay/become 121 mostrar show 31
ter have 117 arriscar risk 28
ir go 111 passar go through/spend/cross 28
fazer do 90 jogar play 27
falar speak 85 levar take 25
sair go out/resign 79 mudar change 24
vir come 74 sofrer suffer 22
dizer say 64 encontrar find 21
estar be 64 ofender offend 18
errar err 61 contrair incur/develop (disease) 18
enfrentar face 60 discutir discuss/argue 18
dar give 59 comprometer compromise 17
entrar enter 52 ouvir hear 17
conseguir manage/succeed 48 tocar play (music)/touch/raise (issue) 16
voltar return 46 ferir hurt 15
assumir assume/accept responsibility 39 voar fly 15
viver live 39 ganhar win 15
andar go/walk 38 chegar arrive 15
deixar leave/stop 37

Table 7. Distribution of verbal objects of nouns of fear in Portuguese

As mentioned above, ser, estar and ficar act as copulas or passive auxiliaries, and allow for a finer distinction than the English be. The verbs in the passive following a fear verb and these auxiliaries, as in tenho medo de ser preso (“I fear being arrested”) are shown in Table 8 below. [5]

What is feared* Gloss Frequency What is feared Gloss Frequency
preso arrested 30 isolado isolated 6
acusado accused 29 esmagado crushed 6
morto killed 25 punido punished 6
atacado attacked 21 roubado stolen 6
descoberto discovered 18 excluído excluded 6
assassinado murdered 17 esquecido forgotten 6
visto seen 16 contagiado contaminated 6
apanhado caught 15 responsabilizado made responsible 5
chamado called 12 espancado beaten up 5
assaltado assaulted 11 sequestrado kidnapped 5
atingido hit 11 envenenado poisoned 5
obrigado forced 10 julgado judged 5
expulso expelled 9 confrontado confronted 5
contaminado contaminated 9 confundido confused 5
considerado considered 9 criticado criticized 5
despedido fired 8 conhecido known 5
prejudicado damaged/hurt 8 reconhecido recognized 5
ultrapassado overcome 8 estigmatizado stigmatized 4
posto put 8 conotado associated 4
surpreendido surprised 7 deportado deported 4
agredido hit/beaten 7 engolido swallowed 4
interpretado interpreted 7 enganado cheated 4
identificado identified 7

Table 8. Distribution of verbal (passive) objects of nouns or verbs of fear in Portuguese
* The past participles are given in their masculine singular form.

This list is longer and the structure appears to be far more common in Portuguese than similar situations with be in the BNC (see section 5). This may be due to the larger corpora, but the extra auxiliaries and their related meanings may contribute to more explicit descriptions of the situations.

The parsed AC/DC corpora allow one to search for nouns acting as both the Subject of the sentence and the Phenomenon, as with verbs such as assustar and sobressaltar, and this produced a small number of examples such as those in Table 9.

What frightens Gloss Frequency What frightens Gloss Frequency
fogo fire 7 voz voice 3
explosão explosion 6 gente people 2
tiro shot 5 declaração declaration 2
preço price 5 notícia news 2
sismo earthquake 4 estatística statistics 2
incidente incident 4 ideia idea 2
incêndio fire 4 população population 2
violência violence 4 conservador conservative 2
número number 4 resultado result 2
futuro future 4 perspectiva perspective 2
propano propane 3 possibilidade possibility 2
tiroteio shooting 3 situação situation 2
chuva rain 3 criança child 2
exame exam 3 pessoa person 2

Table 9. Distribution of subjects of verbs of active fear in Portuguese

If we enlarge the possible search candidates (and therefore accept some noise in the 2,133 occurrences, as opposed to the previous 140) we obtain the set given in Table 10. It is interesting to see that Portuguese displays a considerable number of abstract nouns as fear-inspiring Phenomena, just as would be predicted by the Romance languages’ preference for abstract as opposed to concrete descriptions, as pointed out by Vinay and Darbelnet (1958) for French vs. English, see also Santos (2004).

What is feared Gloss frequency What is feared Gloss frequency
pessoa people/person 48 crescimento growth 12
fa(c)to fact 34 pai father 12
população population 30 mulher wife/woman 12
número number 28 morte death 12
presidente president 22 habitante inhabitant 11
governo government 21 ideia idea 11
notícia news 19 violência violence 11
ministro minister 19 resultado result 11
incêndio fire 18 perspectiva perspective 11
coisa thing 18 português Portuguese 11
situação situation 17 preço price 11
homem man 16 líder leader 10
ameaça threat 15 verdade truth 10
gente people 15 fogo fire 9
hipótese hypothesis 15 tiro shot 9
crise crisis 15 aumento raise/increase 9
possibilidade possibility 15 trabalho work 9
questão question 14 caso case 9
problema problem 14 criança child 9
nome name 13 fantasma ghost 9
explosão explosion 12 fuga escape 8

Table 10. Distribution of subjects of verbs of active fear in Portuguese: what inspires fear

Although one can understand the danger implicit in most of these words, none of them can be seen as seriously agentive and, in context, it is natural that the Senser will take the focus of the fear. However, the numbers also indicate a greater possibility in Portuguese of allowing the Phenomenon to be the Subject of the sentence, which confirms a cultural tendency to place responsibility on something external rather than on the individual (see Maia 1994). This aspect can also be noticed in other linguistic structures that we cannot discuss here.

When a search is made for who is frightened (the Senser as Object), the results are quantitatively higher, as Table 11 shows. Note that subjects like avião (airplane), helicóptero, piloto and aparelho (engine) are obviously traces of the other meaning of aterrar (land), and testify the need for human revision of the data presented here.

Who is frightened Gloss Frequency Who is frightened Gloss Frequency
pessoa people/person 93 comunidade community 9
gente people 52 classe class 9
população population 43 empresário businessman 9
investidor investor 35 presidente president 9
adversário opponent 27 técnico technician 9
morador householder/resident 24 turista tourist 9
país country 23 aparelho engine 8
avião airplane 22 menino boy 8
governo government 22 candidato candidate 8
mercado market 19 mulher woman 8
criança child 19 cliente customer 8
inimigo enemy 16 consumidor consumer 7
vizinho neighbour 15 torcida supporter team 7
helicóptero helicopter 14 piloto pilot 7
povo people 13 família family 7
homem man 12 brasileiro Brazilian 7
eleitorado electorate 11 responsável responsible 7
eleitor elector 11 grupo group 7
opinião opinion 10 português Portuguese 7
equipa team 9 mundo world 7
elite elite 9 comprador buyer 7
dirigente leader 9 direita right (parties) 7

Table 11. Distribution of patients of verbs of active fear in Portuguese; who is frightened

The last case we will address is interesting in that assustar o medo (frightening fear) is possible in Portuguese, meaning avoiding or preventing it. However, when the actual examples were inspected in Portuguese, we observed a different phenomenon. They were wrongly caught by a wider query, but showed a different situation, probably even more interesting in itself: the common repetition or reinstatement of fear in adjacent clauses, or the description of several fears and counter-fears, such as the following examples (9) and (10) (out of 23) illustrate:

(9) Receando, porém, assustar Teresa e privar-se da entrevista, escreveu nova carta, em que não transluzia medo de ser atacado, nem sequer receio de marear-lhe a fama.
‘Therefore, afraid of frightening Teresa and missing the interview, he wrote another letter in which there was no sign of any fear of being attacked, or even of being worried about affecting her fame’
(10) Eu me assustei, fiquei com medo, mas ele foi tão veemente, que acabei fazendo.
‘I was frightened and afraid, but he was so insistent that I ended up doing what I was told.’

The BNC will also render similar examples as in (11).

(11) All her little fears were burned away in the great fear, the fear of ceasing to be. (BNC)

These examples draw attention to the point sometimes made that clusters of synonyms tend to occur in text for reasons related to context and emphasis, or, to use Halliday and Hasan’s (1976) term, “lexical cohesion”. For example, the BNC shows that fear collocates frequently with anger, pain and anxiety. If the Linguateca corpora are later annotated with other emotions, such as regret, sympathy or desire, we may be able to follow up the point made above of combinations of emotion words to suggest “mixed emotions” or even, using more sophisticated searches, explore Hoey’s (2005) notion of lexical priming. For the moment we can say that there are 3,304 sentences in the Portuguese resources where more than one fear word is present.

6.3 The negation of fear and courage – an interesting subject for future development

The AC/DC corpora were searched for negation of fear. There are only 2,951 cases of negation, compared to the 27,113 cases of verbal fear, and it is interesting to check whether the negated proportion was higher with any of the verbs. Table 12 presents, for all verbal lemmata, the relevant frequencies, and the percentage of the cases they were negated in the material.

Verb negated total %
temer 1,268 9,785 12.9
assustar 842 4,514 18.6
recear 562 4,634 12.1
aterrar 56 2,905 1.9
alarmar 51 921 5.5
amedrontar 48 355 13.5
arrepiar 28 555 4.04
sobressaltar 21 517 4.06
arrecear 13 54 24.1
apavorar 13 341 3.8
aterrorizar 11 770 1.42
acobardar 19 77 24.7

Table 12. Negation of verbal fear

It appears that there are wide differences between the verbs, with acobardar, assustar and arrecear much more prone to negation than apavorar or aterrorizar.

As far as fear in the noun form is concerned, 2,841 negated cases were found as opposed to 14,664 of support expressions with a fear noun. There are also 2,274 cases of sem (without) followed by a fear word in the material, as presented in Table 13.

After sem Frequency
medo 788
receio 482
sobressalto 475
temor 124
temer 111
susto 87
recear 28
terror 18
pânico 13
assustar 8

Table 13. fear words after the preposition sem (without)

These examples “without fear” bring up the question of the status of courage in relation to the fear lexicon. There is no doubt among psychologists or cognitive scientists that courage is a value and not a feeling or sensation, and we do not want to quarrel with that. However, as noted above, the fear lexicon is sometimes used to express different emotions such as regret and shock, and overlaps with meanings like politeness that are outside the actual fear domain. Few lexical or conceptual domains are “watertight” and, in certain cultures, it is clear that fear and courage will be linked on a moral level. This makes it all the more interesting to note that in the Portuguese language courage and fear are similar, in the sense that they are expressed with the same grammatical operators and using the same metaphors, as the parallel expressions in Table 14 show.

Involving courage Involving fear English gloss
Enchi-me de coragem Enchi-me de medo. fill oneself with courage/ fear
Cheia de coragemCheia de medo full of courage/ fear
Não tem coragem = tem medoNão tem medo = tem coragem NEG has courage/ fear = has courage/ fear
Destemida (=corajosa) Desencorajada (=sem coragem) with/ without courage/ fear
Infundir coragemInfundir medo inflict courage/ fear

Table 14. Parallel expressions for fear and courage in Portuguese

It would be interesting to pursue this matter with expressions in other languages, particularly from the moral point of view.

The English courage, for example, is decidedly positive in context, whereas fearless, although generally positive, can co-occur with less positive notions like aggressive. Also temerário is usually rash or reckless which is not even courage, but more of a judgment on an action as unnecessarily dangerous and stupid. While this is the same in Portuguese, it is undeniable that it is a derived form from temer.

Temente is also an interesting case of a fear-denoting adjective because it is mostly positive, and mainly used in the collocation temente a Deus (“God fearing”). It includes respect in its meaning, so it is interesting also to see that it can also be used negatively (and creatively) as in temente à PIDE (political police) to convey the impression of oppression equated with (the official) religion, as in example (12).

(12) Entre o fogão e o telefone descreve o Portugal dos Pequeninos – rural, atrasado, obediente e temente à PIDE --, onde o “modernismo” e a “eficiência” eram palavrões vindos da América que, tal como a Coca-Cola, deveriam ser proibidos .
‘Between the kitchen and the telephone, she describes Portugal for Children – rural, backward, and afraid of PIDE – where “modernism” and “efficiency” are dirty words from America that, like Coca-Cola, should be forbidden.’

There is also a negative version of courage expressed in Portuguese by colloquial lata, impertinência, audácia, and also in English by the colloquial cheek or nerve, dare and audacity, which shows that this would be an interesting area to explore further. [6]

7. Conclusions and suggestions

We have only started to scratch the surface of this fascinating area. Although we know that a lot has been written about the language of emotion over the last two decades (not least in the field of polarity detection, sentiment analysis and so on, see Pang & Lee (2008) for an overview), we believe that using corpora annotated with all the information described in this article can only help to deepen our knowledge of languages in general and individual languages as well.

The data we have examined here draws attention to the way the human condition contributes to some sort of universal semantic structure in the language of emotion. It would make psychological sense if at least most languages in the Western world shared the tendency to focus on the Senser, or individual, that is common to English and Portuguese. One wonders whether the languages of cultures that give primacy to the social group do the same, and whether they also use fear words in politeness strategies.

We are dealing with two European languages with similar cultures so, even given the human condition underlying emotion, we cannot presume to use these data towards proving any language universals. However, perhaps we can use them to speculate about a certain European mindset, or to use this as a basis in which to study European languages in the past, to see if these tendencies have developed over time, in the way Deutscher (2010) describes in relation to colour terms.

Whatever differences exist will be in the details, and we have seen a few. Do we feel it is different to be afraid and or to ter medo (have fear)? Is there a difference of recognition of “selfhood” in O João assustou-se, or of temporary and imperfective state in O João estava assustado, or of the resulting aspect of O João ficou assustado? All of these can be translated into English as John was frightened, with the extra semantic traces only being retrievable from a larger context, if at all. We could discuss the relationship between be and the Portuguese ser, estar and ficar, and whether they are copulas or passive auxiliaries with emotion adjectives or past participles. We could create carefully controlled corpora in specific genres to see whether fear is more focused on the Senser or the Phenomenon. Maia (1994) showed, based on comparable corpora, that English was slightly more Senser focused than Portuguese, and results in the last section of this paper suggest the same. And we have not even started on the problem of the different lexical realizations of fear in both languages.

Further study is needed and could, no doubt, benefit from going from monolingual corpora to parallel ones in different languages as well. Linguateca provides the COMPARA (Frankenberg-Garcia and Santos 2003) and CorTRad (Tagnin et al. 2010) parallel corpora for this type of work, and the OPUS corpora (Tiedemann 2009) could provide interesting material for these purposes too. We hope to continue this study of fear and also identify related words that implicitly convey fear, such as fugir de in Portuguese or flee in English.

There are other emotions to study and they will help to remind us how inventive we are with language. Philosophical and linguistic knowledge suggest we cannot give orders on emotion, so why is Bobby McFerrin’s Be happy enduringly popular? And do we translate it with ser or estar in Portuguese?


The work by the second author has been partially funded by the Linguateca project, which has throughout the years been jointly funded by the Portuguese Government, the European Union (FEDER and FSE), UMIC, FCCN and FCT. The annotation of the corpora with the fear domain has been accomplished at the Titan cluster at the University of Oslo, and we are therefore grateful to its Research Computing Services Group.


[1] It is perhaps relevant to indicate that the two authors do not necessarily share every opinions – and especially not research history. Maia was interested in the relation of emotion to cognition. Santos’s interest in emotions was basically aroused by her interest in evaluation in language and her realization of its prime importance in natural language, as argued by Ellis (1993). If good is one of the most basic adjectives in language, emotional aspects of human evaluation are also essential for the understanding of how language works.

[2] 476. We should distinguish between the object of fear and the cause of fear. Thus a face which inspires fear or delight (the object of fear or delight), is not on that account its cause, but – one might say – its target.

477. “Why do you believe that you will burn yourself on the hot-plate?” – Have you reasons for this belief; and do you need reasons? (Wittgenstein 1953)

[3] Afraid – Online Etymology Dictionary early 14c., originally pp. of afray “frighten,” from Anglo-Fr. afrayer, from O.Fr. esfreer (see affray (n)). A rare case of an English adjective that never stands before a noun. Because it was used in A.V Bible, it acquired independent standing and thrived while affray faded, chasing out the once more common afeared.

[4] We have not attempted to translate these words into English because their meaning will often change considerably according to context.

[5] While in both languages in the passive the main verb is in the past participle (and thus the past participle in the English translation), in Portuguese there are four forms of the past participle, since it agrees in gender and number with the subject, so we chose to display the infinitive of the verb instead.

[6] Interested readers can do it in the AC/DC corpora by looking for [sema="coragem"], or, compare both domains with [sema="medo|coragem"] (and select “Distribuição por campo semântico (sema)” in the Resultado field).


AC/DC. http://www.linguateca.pt/ACDC/

BYU-BNC. http://corpus.byu.edu/bnc/

Cognition and Emotion. http://www.tandfonline.com/toc/pcem20/current

Languages of Emotion. http://www.loe.fu-berlin.de/en/index.html


Bick, Eckhard. 2000. The Parsing System “Palavras”: Automatic Grammatical Analysis of Portuguese in a Constraint Grammar Framework. Aarhus: Aarhus University Press.

Damásio, António. 1994. Descartes’ Error: Emotion, Reason, and the Human Brain. New York: Putnam.

Deutscher, Guy. 2005. The Unfolding of Language. William Heinemann.

Deutscher, Guy. 2010. Through the Language Glass – why the World looks Different in Other Languages. William Heinemann.

Eckman, Paul. 1982. Emotions in the Human Face. 2nd edition. Cambridge: Cambridge University Press.

Ellis, John M. 1993. Language, Thought and Logic. Evanston, IL: Northwestern University Press.

Fodor, Jerry. 1998. Where Cognitive Science Went Wrong. New York: Oxford University Press.

Frankenberg-Garcia, Ana & Diana Santos. 2003. “Introducing COMPARA, the Portuguese-English parallel translation corpus”. Corpora in Translation Education, ed. by Federico Zanettin, Silvia Bernardini & Dominic Stewart, 71–87. Manchester: St. Jerome Publishing.

Halliday, M.A.K. 1994. An Introduction to Functional Grammar. 2nd Edition. London & New York: Arnold.

Halliday, M.A.K. & Ruqaija Hasan. 1976. Cohesion in English. London: Longman.

Hoey, Michael. 2005. Lexical Priming. London & New York: Routledge.

Lakoff, George & Mark Johnson. 1980. Metaphors We Live By. Chicago: University of Chicago Press.

Maia, Belinda. 1994/1996. A Contribution to the Study of the Language of Emotion in English and Portuguese. PhD dissertation, Faculdade de Letras da Universidade do Porto. Revised version: http://web.letras.up.pt/bhsmaia/belinda/pubs/thesis.htm

Ortony, Andrew, Gerald Clore & Allan Collins. 1990. The Cognitive Structure of the Emotions. Cambridge: Cambridge University Press.

Pang, Bo & Lillian Lee. 2008. “Opinion mining and sentiment analysis”. Foundations and Trends in Information Retrieval Vol. 2, No 1–2: 1–135. http://www.cs.cornell.edu/home/llee/opinion-mining-sentiment-analysis-survey.html

Pinker, Steven. 2007. The Stuff of Thought: Language as a Window into Human Nature. London: Allen Lane.

Santos, Diana. 2004. Translation-based Corpus Studies: Contrasting English and Portuguese Tense and Aspect Systems. Amsterdam & New York: Rodopi.

Santos, Diana. 2011. “Linguateca’s infrastructure for Portuguese and how it allows the detailed study of language varieties”. OSLa: Oslo Studies in Language Vol. 3, No. 2, 113–128.

Santos, Diana. Forthcoming 2012. “Corpora at Linguateca: vision and roads taken”. Working with Portuguese Corpora, ed. by Tony Berber Sardinha & Telma São Bento Ferreira. London: Continuum.

Santos, Diana, Stella Tagnin & Elisa Duarte Teixeira. “CorTrad and Portuguese-English translation studies: investigating colours”. Aspects of corpus linguistics: compilation, annotation, analysis. (Studies in Variation, Contacts and Change in English 12). Helsinki: Research Unit for Variation, Contacts, and Change in English. (This volume). http://www.helsinki.fi/varieng/journal/volumes/12/aclcaa/santos_tagnin_teixeira/.

Sperber, Dan & Deirdre Wilson. 1986. Relevance: Communication and Cognition. Oxford: Basil Blackwell.

Tagnin, Stella E.O., Elisa Duarte Teixeira & Diana Santos. 2009. “CorTrad: a multiversion translation corpus for the Portuguese-English pair”. Arena Romanistica 4: 314–323.

Tiedemann, Jørg. 2009. “News from OPUS – A Collection of Multilingual Parallel Corpora with Tools and Interfaces”. Recent Advances in Natural Language Processing, ed. by N. Nicolov, K. Bontcheva, G. Angelova & R. Mitkov, vol V, 237–248. Amsterdam & Philadelphia: John Benjamins.

Vendler, Zeno. 1957/1967. “Verbs and Times”. The Philosophical Review LXVI: 143–60. Reprinted with minor changes as Chapter 4 of Vendler, Zeno. Linguistics in Philosophy, 97–121. Ithaca: Cornell University Press, 1967.

Vinay, J.-P. & J. Darbelnet. 1958/1977. Stylistique Comparée du Français et de l’Anglais: Méthode de traduction, Paris: Didier, 1977, Nouvelle édition révue et corrigée. First edition: 1958.

Whorf, Benjamin Lee. 1940/1956. Language, Thought and Reality: Selected Writings of Benjamin Lee Whorf. Cambridge, Mass.: MIT Press.

Wittgenstein, Ludwig. 1953. Philosophical Investigations. Translated by G.E.M. Anscombe. Oxford: Basil Blackwell.