Theory of meaning

[ Pobierz całość w formacie PDF ]
//-->Part ITheoryChapter1The Nature of MeaningIf we want to extract meaning automatically from text, we first need to investigatewhat ‘meaning’ actually is. What does it mean for a word to mean something? Howis a particular word able to convey a particular content? And how is this contentbuilt up? What is, in short, the nature of meaning?1.1Theories of meaningIn the course of history, the nature of meaning has been one of the major issues inthe philosophical debate. The issue was first raised in the ancient Greek world, andwas subsequently tackled by numerous philosophers. In the 19th century, meaningalso entered the realm of linguistics – first in the context of diachronic linguistics,1later also as a synchronic study. In the following paragraphs, we briefly discussthe different theories of meaning (and their relation to reality) that have beenproposed both in philosophy and linguistics, and assess their potential to servewithin computationally implemented procedures of meaning extraction.1.1.1Referential theory of meaningIn a referential theory of meaning, the meaning of a particular word is regardedas a pointer to the designated object in the real world. The meaning of a wordChristian Karl Reisig proposed the study of ‘Semasiologie’ in 1825, Michel Bréal coined the term‘sémantique’ in 1883.1781.1Theoriesof meaningis what it refers to. If we utter a word likeapple,we refer to an actual apple (orthe set of all apples) in reality. Intuitively, a referential theory of meaning seemsvery appealing. If parents want to teach their children the meaning of a wordlikeapple,chances are pretty high that they will point to an actual apple – or apicture of one. At first sight words indeed seem no more than references to things(entities, actions or relations) existing in the outside world. There are, however, anumber of problems with such a referential theory of meaning. The theory is ableto account for what is generally called thedenotationorextensionof words, butfails to describe other semantic characteristics, generally referred to asconnotationorintension.The German philosopher Gottlob Frege (1848–1925) illustrated thisdeficiency with a by now well-known example. Compare the following sentences:(1)(2)The morning star is the morning star.The morning star is the evening star.Bothmorning starandevening starrefer to the same entity, viz. the planet Venus,which might be visible either in the morning or in the evening (depending onthe relative position of Venus and the earth). Sentences (1) and (2), however,significantly differ in meaning. Sentence (1) expresses a simple tautology, whereassentence (2) expresses a new and important astronomical truth. Sentences (1)and (2) do not mean the same thing, but a referential theory of meaning does notaccount for the difference between them.Frege’s solution to the morning/evening star paradox was to make a distinctionbetweenSinn(sense) andBedeutung(reference).Bedeutungis the object thatthe word refers to, whereasSinnis the cognitive representation of the object. Bymaking this distinction, it is possible for words to have a different sense but thesame referent (as in the paradox above).The referential theory of meaning has been popular with logicians (e.g. theyoung Wittgenstein and Bertrand Russell). It provides a parsimonious and straight-forward model of meaning, but the previous examples have shown that it isincapable of capturing all aspects of meaning. Moreover, it is unclear how we oughtto proceed in order to extract these ‘meaning references’ in a computational way.The theoretical problems as well as the practical drawbacks make the referentialtheory rather unattractive for the computational extraction of meaning.1.1.2Mentalist theory of meaningAnother solution – one that has been very popular throughout the history ofphilosophy, starting with the Greek philosopher Plato – is to represent meaning1The NatureofMeaning9exclusively as ideas. A mentalist theory of meaning associates the meaning of aparticular word with a particular idea in the human mind. This theory effectivelysolves the morning/evening star paradox: The morning star might be the samething as the evening star in reality, but theideaof the morning star and the eveningstar may very well differ. The question that immediately follows is what thisnotion ofideaactually entails. Surely, it cannot be the mental representationsthat are present in each individual person. These mental representations differ alot among different persons. If one person hears the wordstrawberry,an imageof an appetizing dessert plate – possibly covered with lots of whipped cream –might pop up. Another person might prefer them with powder sugar, and anotherone without any topping at all. Or one might even be disgusted by the idea ofstrawberries, because of a severe allergic reaction in the past. To be practicallyusable, theideasneed to have some generality, exceeding the individual level. Butit is difficult to achieve this generalization without resorting to the notion of ideain the platonic sense, that is somehow mysteriously present in people’s minds. Thisis not the direction we want to venture into, especially if we want to implementsemantics in a computational way. If we want a sound theory of semantics that canbe implemented computationally, we will need a theory that is not dependent onreference or ideas.1.1.3Behavioural theory of meaningThe vagueness and non-generality that inevitably seems to surround the mentalistview has led people to abandon the mentalist theory of meaning in favour of atheory that sticks to ‘observable’ facts. Inspired by the behaviourist movementthat became popular within the field of psychology, the American linguist LeonardBloomfield defines meaning of a linguistic form as ‘the situation in which thespeaker utters it and the response which it calls forth in the hearer’. (Bloom-field, 1933, p. 139). The meaning of a word is thus reduced to the speaker’sstimulus that elicits its use, and/or the hearer’s response to that word.Although the behavioural theory of meaning claims to overcome the vaguenessof ideas in the mentalist view, it seems almost as problematic as the theory itopposes. There is a plethora of different stimuli that elicit the same word, andthe number of different responses evoked by that word is equally high. Take, forexample, a word likejazz.In some situations, a person might utter the word toindicate they would like to hear some jazz tunes. In other situations, they mightutter the word to approve – or disapprove – of the music they are listening to at thatmoment. And one odd person – not particularly familiar with different music styles– might even utterjazzwhen in fact they are listening to hip hop. Similarly, people’s [ Pobierz całość w formacie PDF ]

zanotowane.pl

doc.pisz.pl

pdf.pisz.pl

shinnobi.opx.pl