Towards a Semantic Prosodies Approach in the Study of Irony

The advent of large corpora such as the LOB, Brown...made it possible for readers to perceive consistencies in the influence of collocation on the behaviour of particular linguistic forms. Irony as a stylistic phenomenon procedure relies heavily and draws its effect from a collocative clash. However, in order to be able to identify a potential collocative clash, there must be a sufficiently consistent background of expected collocation against which the depiction of irony becomes clear. The consistent meaning that a form has is referred to in this paper as semantic prosody. Semantic prosodies proved to be largely unaccessible to human intuition about language and they cannot be deduced reliably through introspection. In this paper irony in Hemingway‟s A Farewell to Arms was depicted using a collocational approach. The BNC corpus will be used to depict the semantic prosody of some words in the novel and how through their collocates they yield an ironical meaning. Indexing terms/


INTRODUCTION
Most studies that have been done on irony looked at this phenomenon from the point of view of flouting the conversational maxim (Kaufer 1981, Sarangi and Slembrouk 1992, Colston and O'Brien 2000, Utsumi 2000. In this paper I adopt a new collocational approach using computational methods to show that irony is a foregrounded phenomenon by virtue of the fact that it runs counter a semantic prosody. Four collocations were studied from Ernest Hemingway"s A Farewell to Arms. The BNC corpus was used to depict the semantic prosody of the nodes and show how the collocates that Hemingway chose for these nodes create a semantic clash and yield an ironical meaning. Thanks to available corpora now, it is possible to empirically determine which pairs of words have a substantial amount of "glue" between them, and perceives consistencies in the influence of collocation on the behaviour of particular linguistic forms (Stubbs 1995, p. 2-3). Irony as a stylistic phenomenon relies heavily and draws its effect from a collocative clash or the deliberate creation of an exception to the trend. So, in order to be able to identify a potential collocative clash, there must be a sufficiently consistent background of expected collocation against which the depiction of irony becomes clear (Louw 1993, p. 157). The consistent meaning that a form has is referred to by Louw (1993) as semantic prosody. The notion of semantic prosody or (pragmatic meaning) is defined as 'a given word or phrase may occur most frequently in the context of other words or phrases which are predominantly positive or negative in their evaluative orientation… As a result, the given word takes on an association with the positive or, more usually, the negative..' (qtd. In Channell 1999, p. 38). Louw (1997) argues that semantic prosodies proved to be largely unaccessible to human intuition about language and they cannot be deduced reliably through introspection. In this article, I will seek to demonstrate the role of semantic prosodies in making certain forms of irony possible.

2-Definition of key terms
I will start with a brief definition of some key terms:

-Collocation
Collocation means a relationship of habitual co-occurrence between words. It is an expression consisting of two or more words that correspond to some conventional way of saying things "collocations of a given word are statements of the habitual or customary places of the word." (qtd. Firth 181). Collocations which occur with a high frequency are powerful indicators of a pattern of meaning in a text. Concordance allows you to view collocation lists for any headword (base , node) in the wordlist view. It shows the collocates of the headword with all words occurring up to 4 words before and 4 words after the headword. It is possible to make a concordance with as many words that you want before and after the node (Sinclair 1991, p. 24). For the selection of meaningful and significant collocations, an adequate collocation measure has to be defined. In the literature, quite a number of different collocation measures can be found (t-score, mutual information value, chi-square..). Early work on collocation used frequency as a measure to identify a particular type of collocation. In this paper, I will rely on the loglikelihood value in order to determine the most significant ones (BNC manual).

Caharacteristics of collocations
Collocations vary tremendously in the number of words involved in the syntactic categories of the words, in the syntactic relations between the words, and in how rigidly the individual words are used together. Collocations can be described in a number of ways. One way of thinking about them is in terms of the degree to which you can vary the basic pattern and still have a collocation. However, we can only define the fixedness or unfixedness of collocations in terms of a continuum. All we can say is that some collocations are more fixed than others but we cannot make a neat dividing line between "fixed' and 'unfixed". In a fixed collocation, there are a few expected variations in its pattern. For instance, "kick the bucket" which is an idiom meaning "to die" is a fixed collocation. Less fixed collocations, however, are often more structural. For instance, let"s + verb + pronoun + noun. You can insert a variety of words in this structural pattern and still have commonly used patterns. (let"s move on to the next point, let"s go back to the last chapter…) (Smadja 1993, p. 148)

Types of collocations
Distinctions are made between grammatical collocations and semantic collocations. Grammatical collocations often contain prepositions, including paired syntactic categories such as verb + preposition (e.g. come to, put on) etc.. In these cases, the open-class word is called the base and determines the words it can collocate with. Semantic collocations are lexically restricted word pairs, where only a subset of the synonyms of the collocation can be used in the same lexical context (smadja 1993, p. 147-8) .

Node
A Node or base is the word that one chooses to study to see its co-occurrence with various collocates within a certain span or window, say four words to the left or right.

The need for the quantitative study of collocations
McEnery and Wilson (1996, p. 71) argue that the idea of collocation is very important in many areas of linguistics. They report Kjellmer (1991) who assures that "our mental lexicon is made up not only of single words but also of larger phraseological units, both fixed and more variable" (p. 71). It is well-known that some words habitually collocate with other words, and that attested data are required in collocational studies, since native speaker intuitions are not a reliable source of evidence. Native speakers can often give a few examples of the collocates of a word (sometimes accurately) and they may be able to guess, the likelihood of collocations they are presented with. But they certainly cannot document collocations with any thoroughness, and they cannot give accurate estimates of the frequency and distribution of different collocations (Louw 1993, p. 173).
For instance, most native speakers that I have informally tested provided the right collocations for beautiful, handsome, devoted and disgraced, (these are the nodes I chose to study) but such native speaker data are very sparse and unreliable indeed. If we look at dictionaries, they give examples but do not say explicitly whether these examples are frequent or typical. Here comes the importance of studying collocations and using them. In combination with native speakers" intuitions, a corpus allows us to get the facts right about frequency and typicality which are not open to introspection and which are not well-described in current dictionaries and allows extensive documentation of collocates (Kennedy 1991, p. 110).
It is becoming increasingly well documented that words may habitually collocate with other words from a definable semantic set. Sinclair (1991) gives several examples of words which have a "negative" semantic prosody. He shows (1991, p. 70) that the phrasal verb "set in" occurs primarily with subjects which refer to unpleasant states of affair. Sinclair (1991) also points out that it is bad things which "break out" ( violence, riots, protests broke out). Louw (1997, p. 165) uses Sinclair"s term "semantic prosody" for this collocational phenomenon, and shows, for instance, that "utterly", "bent on", and "symptomatic of" have predominantly unpleasant collocates. Similarly, positive prosodies exist. "Provide" is usually used in positive context (provide facilities, information, services, aid, assistance, help, support, care, food, money, nourishment, protection, security…) Sinclair (1991) suggests that many researchers still consider these words as neutral but he maintains that these words acquire positive or negative meaning by association. At some point, the word itself acquires unpleasant connotations and parole affects langue. The habitual collocates of a given word (i.e. set in) are capable of colouring it, so it can no longer be seen in isolation from its semantic prosody, which is established through the semantic consistency of its subjects. Louw (1997) suggests that inferring semantic prosodies, their potency, and proportions, cannot be accessible to intuition, and holds the view that they are essentially a phenomenon that has been only revealed computationally and whose extent and development can only be properly traced by computational methods. I disagree with Louw's point of view that semantic prosodies are not accessible to intuition. He exaggerates the need for computational methods. People, (especially native speakers) always have intuitions about the use of certain words. These intuitions are tested through computational methods to check if they are well based or not. For in stance, when collecting data for this paper I asked native speakers about their intuitions about the words, 'beautiful', 'handsome, 'devotion' and 'disgrace' and their guesses about the use of these words were the same as those given by computational tools. But the problem is that native speakers cannot provide an exhaustive list of all the uses of these words which computational tools do very accurately and rapidly.
Louw criticised Grice (1978) who names everything from pragmatics to tone of voice in search of the perceived irony involved in "He"s a fine friend". He assumes far too readily that the expression has no anchor in the conventions of the language apart from tone of voice. When Louw (1997) studied the collocations of "fine" he found out that it has, indeed, an ironic meaning anchored in language.

4.1BNC corpus:
The corpus is designed to represent as wide a range of modern British English as possible. The written part (90%) includes, for example, extracts from regional and national newspapers, specialist periodicals and journals for all ages and interests, academic books and popular fiction, published and unpublished letters and memoranda, school and university essays, among many other kinds of text. The spoken part (10%) includes a large amount of unscripted informal conversation, recorded by volunteers selected from different age, region and social classes in a demographically balanced way, together with spoken language collected in all kinds of different contexts, ranging from formal business or government meetings to radio shows and phone-ins. The corpus comprises 100,106,008 words (BNC manual)

Instruments
The SARA program is the concordance software used in this paper to retrieve the collocations of the following words; "beautiful", "handsomly", "disgraced" and "devotion" from the BNC. SARA works by linking your computer to the one where the corpus is stored, which function as a server (Aston and Burnard, 1998). You can search the entire BNC or limit yourself to one part of it. In this paper the whole BNC corpus was searched. As soon as you log on to BNC web you will have a dialogue box which includes a list of tasks, you select "quick query", you type your word query, then you click "start query". You will get a whole page of hits. O c t o b e r 9 , 2 0 1 4 In the BNC query result window, select "collocations…" from the drop-down menue and press "go". You will have a screen with options… You should choose some options before the collocational search is set in motion. The choices made in this stage will determine the results you get (BNC web manual)

3 Statistical methods
There are many statistical methods that can be used to identify collocations. The method I used in this paper is the loglikelihood because this statistical measure is suited to low frequency items. The loglikelihood ratio measures the divergence of observed and expected sample counts (I = 1, 2, j = 1,2). The expected values are for the model that assumes independence (assumes that the null-hypothesis is true). For each cell in the contingency table, the expected counts are:

1 'Disgraced'
Frederick Henry, the hero in this novel, describes the soldiers repairing an ambulance and then he describes the car itself saying: " …The car looked disgraced and empty." (Hemingway 1929, p. 15) Here, if you check the meaning of this word in the dictionary you would find the following definition: Disgrace; state in which others think that one has behaved badly and no longer deserves respect.
Disgrace: cause somebody to lose a position of power, honour or favour. (Cowie 344) When I ran the concordance for this word I found out that it collocates most of the time with: tycoon, communist, former leader, star, chief, a party"s general secretary, minister, politician, champion, coach, schoolboy, schoolmaster, aristocrat, lover, official, boss, soldier, mayor… We notice that "Disgraced" is always associated with people who have good ranks and status and who no longer hold this good position. Most of the time the word is associated with politics also. For Hemingway to associate it with an ambulance he does not only create a semantic clash but uses also the word in a totally different context. Perhaps, subcounsciously, Hemingway refers to his own disgrace when he was wounded with shellfire while distributing supplies to Italian soldiers. Although he was decorated because of the injury like Frederick Henry who was wounded while eating spaghetti, Hemingway was not proud of this experience because he was not doing any heroic act. This direct experience of physical wounding affected not only Hemingway"s life but all that he was to write. That"s why ironic strokes about war pervade Farewell to Arms and many of his writings.

2 'Beautiful'
If we look at the third example, "beautiful", Hemingway used this word ironically to describe the retreat from Caporetto. Hemingway"s famous description of this debacle is a stringent comment on the bewildering stupidity and chaos of war, but he takes the opportunity to inject again a shot of special irony. He comments on the stupidity of soldiers who execute deserters by saying: "I saw how their minds worked; if they had minds and if they worked. They were all young men and they were saving their country…The questioners had that beautiful detachment and devotion to stern justice of men dealing in death without being in any danger of it." (Hemingway 1929, p. 201) O c t o b e r 9 , 2 0 1 4 Again here the use of "beautiful" is striking. 'Beautiful' means having beauty, giving pleasure to the senses or the mind, very satisfactory (Cowie 1989, p. 92) But if we look at the context in which this word was used, there is nothing satisfactory. In the BNC "beautiful" collocates with things that bring pleasure to the eye. However, Hemingway associates this word with a negative concept; detachment; lack of emotion, indifference. How can this indifference be described as "beautiful".

'Handsomly'
In Farewell to Arms an American soldier, Frederick Henry (who is actually Hemingway"s alter ego) volunteered as an ambulance driver in the Italian front and talks about his experience in the war. At the beginning of the novel Frederick Henry describes the victories of the army and how towns were captured and he says: "The river ran beside us and the town had been captured very handsomly." (Hemingway 1929, p. 5) When I first read this sentence I was struck by the use of the word "handsomly" in this context. Handsome means goodlooking and of gifts or behaviour it means very generous. I felt some kind of irony because if you look the word "capture" in the dictionary you would find the following definition: Capture: take or win something by force. (Cowie 1989, p. 168) So how is it possible to do such a violent act "handsomly" or in a handsome way. In addition to the use of the intensifier "very" accentuates the irony. So I had the idea to look at the collocates of "handsome" to see the nouns that normally occur with it. What is striking also is that when I ran the concordance for the word "handsomly" (in the BNC corpus which is made up of over one hundred million words) I did not find any match for this word. That is the result of my query was zero. Normally this word is not used frequently in English. When I ran the concordance for "handsome" I found out that it collocates with th e following words; plumage, dividends, trophy, chap, girl, husband, father, year…So the word "capturing" does not collocate with "handsomly". In addition, by putting the word "handsome" in this negative context Hemingway uses an exception to an established semantic prosody. The irony here is not only textual but situational as well. The same thing can be applied to the other collocations.

Conclusion
To conclude, it is worth noting that for thousands of years semantic prosodies remained hidden from our perceptions. Thanks to the advent of corpora it is possible now to extract profiles of semantic prosodies. From these prosodies we may extract the narrow band of irony they contain. In addition, since our mental lexicon is made up of phraseological units, the study of collocations is important for natural language processing, language teaching and dictionary writing.