Tài liệu đọc thêm môn ngữ nghĩa học dành cho sinh viên khoa tiếng anh

In many Australian languages, such as Yankunytjatjara, there are several words for different kinds of clouds but no general word like English cloud.. Admittedly, each word in itself make

1 Lý do biên soạn tài liệu đọc thêm môn Ngữ Nghĩa Học tiếng Anh cho sinh viên Khoa Anh

1.1 Tài liệu chính để dạy môn Ngữ Nghĩa Học theo qui định của Khoa là 11 bài trong cuốn Semantics: a coursebook của hai tác giả James R Hurford và Brendan Heasley do nhà xuất bản Đại học Cambridge phát hành Sách gồm 25 bài, và mỗi bài đều có phần lý thuyết, định nghĩa, bài tập, và phần đáp án Nhìn chung cuốn sách này viết khá đơn giản, trình bày ngắn gọn các khái niệm khá trừu tượng của môn Ngữ Nghĩa Học Tuy nhiên do các tác giả viết quá ngắn gọn dẫn đến việc sinh viên thiếu những cách nhìn khác nhau của một vấn đề Kiến thức về môn học không được sâu do cuốn sách được viết như một sách giáo khoa Thêm vào đó, các tác giả đã viết theo ý kiến chủ quan của mình đối với những vấn đề còn nhiều bàn cãi Tóm lại, giáo trình đang được sử dụng ở Khoa có một số hạn chế nêu trên Sinh viên sẽ dùng cuốn sách nêu trên tốt hơn nếu họ có thêm những tài liệu đọc thêm về môn học này

1.2 Môn Ngữ Nghĩa Học là một môn học khá mới đối với sinh viên Khoa tiếng Anh Tài liệu

về môn học này chỉ có vài cuốn trong thư viện của Khoa và thư viện của Trường Vì vậy sinh viên Khoa Anh rất thiếu các tài liệu đọc thêm về môn học này Nhu cầu biên soạn tài liệu đọc thêm về môn học này là rất cần thiết

1.3 Số tiết dành cho môn học này theo qui định của Khoa chỉ có 30 tiết (2 ĐVHT) Số bài qui định là 11 bài Do số tiết lên lớp tương đối ít so với một môn học khá trừu tượng như môn Ngữ Nghĩa Học, sinh viên cần phải có thời gian để nghiên cứu thêm ở nhà mới có thể nắm vững được môn học này Do đó sinh viên cần có tài

1.4 Để nâng cao trình độ tiếng Anh chuyên sâu của mình, sinh viên khoa Anh thường theo học các lớp sau đai học (Diploma hoặc Master) Trong các chương trình thi và học của các lớp sau đại học đều có môn Ngữ Nghĩa Học Đây là môn học bắt buộc của chương trình sau đại học do các Trường Đại Học Sư Phạm hoặc Trường Đại Học Khoa Học Xã Hội và Nhân Văn tổ chức Việc biên soạn tài liệu này cũng góp phần giúp sinh viên tiếng Anh có đủ kiến thúc nền để có thể theo học các lớp sau đại học

2 Cơ sở biên soạn tài liệu đọc thêm môn Ngữ Nghĩa Học

2.1 Cơ sở đầu tiên để tuyển chọn các tài liệu đọc thêm cho sinh viên là dựa vào giáo trình do Khoa qui định Chúng tôi căn cứ vào các khái niệm được dạy trong chương trình để chọn những tài liệu đọc thêm có liên quan, gần gũi với những khái niệm đó để giúp sinh viên vừa nắm sâu về mặt lý thuyết vừa có thêm nhiều ví dụ, bài tập minh họa nhằm giúp họ hiểu rõ hơn các nội dung đã được học

2.2 Cơ sở thứ hai là chúng tôi chỉ chọn tài liệu được viết với văn phong đơn giản và đều nằm trong khả năng có thể hiểu được của sinh viên

Tài liệu được chọn trên cơ sở vừa có lý thuyết vừa có thực hành để giúp sinh viên có thể tự học, tự đọc, tự trả lời các vấn đề có liên quan đến môn học Một số chương có nêu lên phần ứng dụng trong việc dạy tiếng Anh Phần này giúp sinh viên thấy được môn học có tính ứng dụng cao

2.3 Tài liệu được chọn để làm tài liệu đọc thêm là các bài viết do các tác giả Anh , Mỹ chuyên viết về Ngữ Nghĩa Học với những cách nhìn khác nhau Điều này giúp sinh

thức và có thể đưa ra ý kiến riêng Khả năng đánh gía, phê bình này rất cần cho sinh viên khi học đại học cũng như theo học các lớp sau đại học sau này

3 Phương pháp chọn tài liệu đọc thêm

Chúng tôi tiến hành thu thập các sách viết về Ngữ Nghĩa Học ( Semantics) của nhiều tác giả

từ các nguồn tư liệu của cá nhân, Khoa, Trường Sau đó chúng tôi đọc và chọn các sách có nội dung gần và sát với nội dung mà sinh viên đang học Chúng tôi lại chọn các bài viết dễ hiểu nhưng chúng có cùng các nội dung mà sinh viên đang học tại Khoa

Các bài đọc thêm đều được sắp xếp theo thứ tự nội dung của sách giáo khoa do Khoa qui định Điều này giúp sinh viên dễ tra cứu khi tự đọc ở nhà

Sau khi chọn xong các tài liệu, chúng tôi cho đánh máy lại theo đúng nguyên bản của các tác giả Chúng tôi sắp xếp các bải thành chương Các chương lại được phân chia theo từng mục nhỏ và được đánh số trang theo thứ tự sắp xếp mới của chúng tôi và đóng thành tập để giúp người học dễ tra cứu

Chúng tôi đặt phần tóm tắt nội dung ở ngay trang đầu tiên của tập tài liệu đọc thêm để sinh viên dễ tìm các nội dung cần đọc Phần tài liệu tham khảo cũng được xếp ở phần cuối của tập tài liệu đọc thêm Phần này giúp sinh viên có thể tự tìm ra các đầu sách ngốc

Chúng tôi ( chủ đề tài: Ngô Thị Thanh Vân và cộng tác viên: Trần Xuân Bình) cùng đọc và chọn lọc tài liệu đọc thêm môn Ngữ Nghĩa Học trong thời gian một năm học Chúng tôi đã chi tiền để thực hiện đề tài với các khoản chi như sau đây:

- Tiền chi cho phần đọc sách, nghiên cứu, phân tích tài liệu và biên soạn lại tài liệu:

7.000.000 đ (bảy triệu đồng)

- Tiền sao chụp 14 chương sách : 20.000đ

- Tiền đánh máy toàn bộ tài liệu theo đơn giá:

5.000đ/tờ x 210tờ = 1 050.000đ (một triệu năm mươi ngàn đồng)

- Tiền in 11 cuốn tài liệu x 50000đ/cuốn = 550.000đ ( 05 cuốn cho 05 thành viên hội đồng,

02 cuốn cho 02 tác giả đề tài, 02 cuốn cho phòng sau đại học, 02 cuốn cho thư viện của Khoa Anh)

- Tiền chi phí cho buổi bảo vệ đề tài: 5 thành viên x 150.000đ=750.000đ

CONTENS

CHAPTER 1 SEMANTIC - THE STUDY OF MEANING 1

CHAPTER 2 INTRODUCTION TO LEXICAL SEMANTIC 19

CHAPTER 3 ON ENTERING THE REALMS 32

CHAPTER 4 WORDS AND THEIR MEANINGS 45

CHAPTER 5 WORDS AND THE WORLD 66

CHAPTER 6 WORDS AND WORDS 76

CHAPTER 7 ANALYSING WORD MEANINGS 85

CHAPTER 9 DEIXIS 117

CHAPTER 10 REFERENCE AND INDETERMINACY OF SENSE 128

CHAPTER 11 REFEREENCE AND PREDICATION 146

CHAPTER 12 OPPOSITES AND NEGATIVES 161

CHAPTER 13 PARADIGMATIC RELATIONS OF EXCLUSION AND OPPOSITION 176

CHAPTER 14 LOGICAL MATTERS 185

CHAPTER 1

SEMANTIC

THE STUDY OF MEANING

1.1 Language and meaning

Semantics, the study of meaning, stands at the very centre of the linguistic quest to understand the nature of language and human language abilities Why? Because expressing meanings is what language are all about Everything in a language-words, grammatical constructions, intonation patterns-conspires to realise this goal in the fìillest, richest, subtlest way To understand how any particular language works we need to understand how its individual design works to fulfil its function as an intricate device for communicating meanings Equally, semantics is crucial to the Chomskyan goal of describing and accounting for linguistic competence, thát is, the knowledge that people must have in order to speak and unđerstand a language Semantic competence is a crucial part of overall linguistic competence

Another concern of semantics is to shed light on the relationship between language and culture, or, more accurately, between languages and cultures Much of the vocabulary of any language, and even parts of the grammar, will reflect the culture of its speakers Indeed, the culture specific concepts and ways of understandine embedded in a language are an important part of what constitutes a culture Language is one of the main instruments hy which children are socialised into the values, belief systems, and practices of their culture

Meaning variation across languages

It's hard to believe the colossal variation in word-meanings between languages You might assume, for example, that since all human beings have the same kind of bodies all languages would have words with the same meanings as English hand and hair But no In many languages, the word which reíers to a person's hand can apply to the entire arm; the Russian word ruka, for example, is like this In many languages, different words are used to refer to head-hair and to body -hair; for example, in Yankunytjatjara ( Central Australia) mangka refers to head-hair and yuru to body-hair (as well as fur)

You might think that since environmentạl features like the sun, moon, sky, and clouds are found everywhere on earth, all languages would have words for these things Well, in a sense that's true In any language, one can say things about the sun and about clouds, for instance, but not necessarily using words which correspond precisely in meaning to English sun and clouds In the Australian language Nyawaygi, for instance, there are different words for 'sun lovv in the sky' and for 'hot sun' (i.e overhead), bujira and jula respectively ( Dixon 1980:104) In many Australian languages, such as Yankunytjatjara, there are several words for different kinds of clouds but no general word like English cloud

The same applies to words for events and actions, as well It is natural (in English) to think that ' breaking' is a single, simple event But in Malay there are three words which can cover the range of the English word, one( putus) for where the thing is completely

( pecah) which is more like 'smash' (like what happens when you break a glass)

If even concrete and seemingly universal meanings like 'hand', 'sun', and 'break' are actually not universal but vary from language to language, just think of the variation that exists in relation to more abstract and culture-related meanings How many languages would have words with the same meanings as English privacy, or apologise, and work? How many language would draw a distinction, as English does, between guilt and shame? Obviously, we can't say precisely, but we can say that the number is much, much smaller than most non-linguists would ever imagine In a similar fashion, every language has its own culture-specific meanings, which don't translate readily into English Admittedly, each word in itself makes only a small contribution to the differences between languages, but when you sum up the meaning variation over 10,000 words, perhaps you can see why linguists sometimes say that every language represents a unique way of seeing and thinking about the world

The role of meaning in grammar

In this book we are concerned primarily with semantics, not with other areas of language description such as morphology and syntax Since many readers will have some familiarity with these other fields of linguistics, however, it is worthwhile mentioning the relevance of semantics to the boarder domain of linguistic theory

One of the main concerns of linguistic theory is to identify the governing principles that account for the regularity and orderliness of languages In other words, to answer questions like: Why does language X have the grammatical rules it has? Why does language Y differ from language X in the way it does? What underlying principles apply to both X and Y? For many years the orthodoxy was that semantics did not have much relevance to question like these, because it was believed that the syntactic workings of language were independent

of meaning In recent years, however, as Thomas Wasow (in Sells 1985: 204-5) points out, 'contemporary syntactic theories seem to be converging on the idea that sentence structure is generally predictable from word meanings the surprising thing (to linguists) has been how little needs to be stipulated beyond lexical meaning' If so, semantics is not just an 'add on' to the study of morphology and syntax, but can provide invaluable keys to understanding why morphology and syntax work as they do

1.2 The nature of meaning

Whether we are interested in exploring the connections between meaning and culture, or between meaning and grammar, or simply in exploring meaning for its own sake, the fírst thing we need is a consistent, reliable, and clear method of stating meaning-a system of semantic representation Not surprisingly, the main theoretical controversies in semantics concern the nature of the optimal system pf semantic represẹntation

The vexed question of the nature of meaning is easiest to approach indirectly, by first asking what meaning is not

that it identifíes or 'picks out' in the world (the so-called REFERENT) This seems sensible enough in relation to names, for instance Margaret Thatcher, the Sydney Harbour Bridge, Mexico, or deíĩnite descriptive noun phrases, such as the President of the United States But

to see the meaning is distinct from reference, we only have to think of words which do not refer to anything at all, such as nothingy empty, unicorn, and, usitally, hullo These words are not meaningless, so whatever the meaning of a word may be, it must be something other than what the word refers to

Another argument against the view the meaning equals reference is that if this view were correct, expressions which referred to the same thing would have the same meaning The most famous counter-examples are the expressions The Morning Star and The Evening Star, which clearly differ in meaning, even though objectively they refer to the same thing, namely the planet Venus A more prosaic, but very nice, example (from Allan 1986) is furnished hy the two expressions the man who inventedparking meters and the man who invented the yo-

yo I'm sure you will agree that these tvvo expressions convey different meanings, and I don't expect your opinion will change if I tell you that as a matter of fact both refer to the very same man A classical example is the expression featherless biped Although this has the same referential range as the expression human being, as an explanation of meaning it clearly leaves a lot to be desired

A final reason for rẹjecting the simplistic view that meaning is reference is that the property

of 'making reference' is not something that belongs to words or expressions in themselves at all, but rather to the use of words on a particular occasion For instance, the words this, here, now, and / can refer to any number of things, places, times, or persons depending on the context, but it would be foolish to say that such words had a different meaning every time they uttered

Of course, just because meaning is different from reference doesn't mean that the two are unrelated Obviousiy, they are related: the reference made by the use of a particular expression on a particular occasion depends, at least in part, on the meaning of the expression Linguists sometimes speak of the sense of a word, when they want to make it clear that they are interested in meaning as opposed to reference

Meaning is not scientific knowledge

It is also wrong to think that meaning can be described in terms of scientiíic knowledge, a position advocated by the early American linguist Leonard Bloomfield As a behaviourist, he believed that meaning should be treated as a stimuỉus-response relationship between a speech-form and objective aspects of the speaker's world Bloomfield ( 1933:139) wrote:

we can define the meaning of a speech-form accurately when this meaning has to do with some matter of which we possess scientific knowledge We can deline the names of minerals, for example, in terms of chemistry and mineralogy, as when we say that the ordinary meaning of the English word salt is 'sodium chloride (NaCl)', and we can define the names of plants and animals by means of the technical terms of botany and zoology, but we have no precise way of defining words like love and hate, which concern situations that have not been accurately classified

Some dictionaries follow Bloomneld's prescription, supplying definitions like the following from the Concise Oxford Dictionary

Water: colourless transparent tasteless odourless compound of oxygen and hydrogen in liquid state

red: of or approaching the colour seen at the least-retracted end of the spectrum

Clearly, however, people use the words liver, water, and red quite happily and correctly without having nay ideas about glands, chemical compounds, or the spectrum Knowledge of everyday word-meanings is part of people's linguistic competence, but scientiíĩc knowledge

Meaning is not 'use'

Some people hold that the meaning of a word is 'its use in the language', a slogan taken from the work of the great twentieth-century philosopher Ludwig Wittgenstein As with the reference theory, this view clearly has something going for it Children learn most word-meanings simply by exposure to them in use, without much verbal explanation There is no doubt that the meaning of a word is related to its use; in fact, it can be argued that the meaning of a word is the main determinant of its use But for the purpose of accounting for linguistic competence, the 'meaning is use' theory must be rejected, or at least heavily modified What we are after is an account of what people know about their language Just to say that they know the use of all the words is not very helful—we would have to go on to describe in each case WHAT IT IS that they know about the use

It is also hard to see hovv the 'meaning is use' view could account for the meaning of whole sentences or utterances

1.3 Linguistic approaches to meaning

In this section, we briefly meet some views of meaning which are more relevant to modern linguistics

Truth-conditỉonal theories Although the crude 'meaning equals reference' theory doesn't stand

up for long, the basic notion that meaning is a relationship between an expression and a state

of affairs in the world has given rise to more sophisticated versions of the same basic idea One influential idea from the study of logic is that the meaning of a senterice is the conditions in the world which would have to be met for the sentence to be true To understand

a sentence like Snow is white, for instance, you have to understand how the world would have to be for that sentence to be true Theories based on this approach are called trUth-conditional They say

The main theoretical difficulty faced hy truth-conditional approaches is raised hy the question: How are the truth conditions themselves to be stated? Inevitably, in some language-

or metalanguage, as a language used to describe another language is usually called But how then are the meanings of the sentences in the me- metalanguage to be described? By stating their truth-conditions in terms of yet another metalanguage? Clearly the buck as stop somewhere, or we will have what is called an infinite regress

Another problem lies with what critics call the 'obịectivist' attitude toward meaning, that is

to say, the very idea that meaning resides in the relationship between a linguistic expression and aspects of an 'objective' world For, clearly, there are innumerable expressions in any language whose meanings are inherently subjective and/or culture-bound and cannot be reduced to a correlation with features of the external world How could moral, aesthetic, religious, or philosophical meanings be tied down to correlations with an objective world? Consider words like unfair, beautiful, Gody and rights It seems obvious that these meanings

do not correspond to anything in external reality, and the same applies to a myriad of culture -specific words like Monday, bar mitzvah, associate professor, secondbase, or fiancé (Lakoff 1988: 135)

Conceptual theories

COCEPTUAL THEORIES OF MEANING are closet to the commonsense view They say that a meaning of a world is a structured idea, or 'concept', in the mind of the person using that expression As children growing up in a culture, we acquire, in the form of word-meanings, a huge number of 'prepackaged concepts shared hy those around us Because these underlying ideas are shared, communication becomes possible This view of meaning is often associated with the seventeenth-century English philosopher John Locke, who wrote:

[W]ords, in their primary or immediate signfication, stand for nothing but the ideas in the mind of him that uses them, When a man speaks to another, it is that he may be understood; and the end of speech is that those sounds, as marks, may make known his ideas to the hearer (Locke 1976 [1690]: 208, emphasis in the original)

Conceptual theories take various forms, especially as regards the following important question: If meanings are concepts or ideas, where do those ideas come from, how are they acquired? Locke believed that the human mind is analogous to a blank slate (tabula rasa) at birth, but that since we all share the same kind of sense organs and therefore experience the world in more or less the same way, we come, in time, to have more or less the same basic ideas about it; our complex ideas are then built up from these basic ideas derived from experience This is called an empiricist position, because it emphasizes the importance of knowledge derived through the senses Modern versions of this position are held by many psychologists, and by some linguists, including George Lakoff and Ronald Langacker

The alternative position is the rationalist one held by contemporaries of Locke's such as Gottfried Leibniz and René Descartes This holds that the very simplest concepts are a natural or innate property of the human mind, activated hy experience but not wholly and solely derived from it One argument in favour of semantrc rationalism is that it is hard to see how concepts Iike time, causation, and identity (sameness) could ever be derived from 'pure' sensory experience Another argument comes from the vast complexity and intricacy of language It seems implausible that children could acquire all this as reliably and

version of this theory

Platonist theories

Contrasting with conceptual theories, according to which meaning is 'all in the mind', there are what are known as platonist theories, after the ancient Greek philosopher Plato Platonist theories view meanings as abstract relationships between expressions, existing independently

of speakers, hearers, and actual speech acts Again, there are several versions Plato's own view was a metaphysical one He believed in the existence of a timeless insensible 'world' of perfect Forms or ideas, to which the soul of a person, being itself immortal, has intuitive access Thus he sought to account for the problem of how people could arrive at a concept like that of a perfect circle, a figure which is never found in the natural world

A modern Platonist theory, without the metaphysical slant, is that of linguist Jerrold Katz

He insists (Katz 1981) that meanings are abstract entities, existing somehow outside the mind That is, they are not necessarily located within people's conscious experience, although they may be accessed and known through intuition If that sounds mysterious and confusing

to you, all I can say is that most people have that reaction (myself included) Perhaps an analogy might help Consider the relationship between the number one, two, and four For sure we can known that 'one plus one is two' and that 'two plus two is four', and wouldn't you agree that the validity of these relationships doesn't depend on you, or me, or anyone, actually knowing it? Apply the same approach to word-meanings, and you have a Platonist attitude to semantics What Katz is saying boils down to the assertion that the relationship between meanings is not a matter of individual psychology One reason he wants to reach this position is, I think, to avoid the endless and largely sterile debates about individual variation in meaning which tend to surround the conceptual theories

Structuralist theories

Structuralism is a powerful intellectual tradition in modern thought, especially in literary studies and anthropology, as well as in linguistics One of its pioneers was the great French linguist Ferdinand de Saussure, who was fond of drawing an analoey between language and the game of chess Just as the various chess-pieces, and their moves, can only be understood

in terms of how they function together and in contrast with one another, so it is in language also: to understand the meaning of a world, for instance, we have to understand how it functions together with and in contrast to other, related words According to a fully structuralist theory of meanlng, the meaning of any word has with the other words in the language

Taken to its logical extreme a structuralist view holds that meanings (and sounds also, for that matter) cannot be characterised in positive terms at all A famous quotation to this effect comes from Saussure's Course in General Linguistics (Harris 1983: 118):

In langUage 1tself, there are only differences Even more important than this is the fact that, although in general a difference presupposes posrtive terms between which the difference holds, in a language there are differences, and no positive terms language includes neither sounds nor ideas existing prior to the linguistic system, but only conceptual and phonetic differences arising out of that system (emphasis in original)

The semiotic approach

A final approach to meaning may be termed SEMIOTIC or translational This holds that it

is a mistake to become preoccupied with 'what meaning is', as though 'meaning' were a thing

of some kind Since our whole lives are infused with meanings, since we are communicating

in meanings even as we try to speak about them, we can never (on this view) escape from language and find a vantage point.from which we can look back and describe what it is we have been doing

This certainly does not mean, however, that we cannot describe and analyse specific, concrete meanings True, we cannot get outside the system of meaning-expression and reduce meanings to anything else (individual ideas, brainwave patterns, mystical Platonist forms, patters of usage, or what have you) If we want to describe the meaning of one linguistic expression we have no choice but to do it in terms of the meanings of other linguistic expressions But why not? After all, it is a commonsense, pretheoretical idea that one can explain the meaning of an unknown word in terms of the meanings of other, known words-by paraphrasing the original word

This approach to meaning can be called 'translational' because, in this theory, the meaning

of a linguistic expression is regarded as a kind of translation; that is, you give the meaning of one expression hy way of a translation into another, more easily understood expression it is called 'semiotic' because it takes the world of 'signs' as irreducible In case you have not already guessed, this is the approach adopted in this book

1.4 Issues in semantic theory

The 'irreducibility of the sign' was emphasised by Charles Sanders Peirce, the great American logician and philosopher It was Peirce who coined the terms semiotics, which he envisaged as a new science devoted to the study of 'signs' For our purposes a sign is just an expression, such as a word, which has a meaning One of Peirce's (1932: 2.230 f.) main contentions is that it is impossible to reduce a sign to any elements that are themselves not signs For all intents and purposes, it is impossible to analyse or describe meanings except in terms of some other language-perhaps a more or less 'technical' language consisting of special symbols, but a sort of language nonetheless

To make this clearer it is useful to introduce the terms metalanguage In general, a metalanguage simply means a language that is used to describe or analyse another language For instance, in a bilingual Russian-English dictionary, where Russian words are defined or explained in terms of English words, the metalanguage is English; in an English-Russian bilingual dictionary, on the other hand, the metalanguage is Russian because it is the language in which the definitions are given Another example: technical terms like noun, verb, adjective, subject, object, active, and passive are part of the grammatical metalanguage

of linguistics, that is, they are words used to describe or analyse the grammatical relationships in ordinary everyday language Of course, a metalanguage doesn't have to be a different language (or even a different register) from the one being described A normal English dictionary uses English words to describe the meanings of other English words Now, any semantic theory needs some way to state the meaning of an expression An object

of any kind which states the meaning of an expression can be called a SEMANTIC

the best kid of semantic metalanguage

Universality

On the face of it, it would seem only sensible to approach semantics with the aim of finding

a universally applicable system for describing meaning wherever we fĩnd it; just as in phonetics, phonology, and syntax we seek frameworks and principles which are equally applicable to all the world's languages As Noam Chomsky once remarked ( 1965:160):

It is important to determine the universal, language-independent constraints on semantic features- in traditional terms, the system of possible concepts The very notion of "lexical entry" presupposes some son of fixed, universal vocabulary in terms of which these objects are characterised, just as the notion of "phonetic representation" presupposes some sort of universal phonetic inventory

The possibility of translation between languages, and the ability that we all have to learn and understand new languages, also seem to indicate that there is some universal framework for understanding, which is shared hy all human beings The notion that there is a "psychic unity" to humankind has a long history in philosophy, and, more recently, in linguistics and anthropology

Many modern linguists, however, doubt that a universal system for describing meanings is possible For one thing, translatability is not always a simple matter if there are wide differences in culture and grammar to be reckoned with Also, many linguists are daunted by the 'slipperiness, of meaning, and fear it would be impossible to establish a universal inventory of semantic features because it would be too difficult to choose between competing analyses

Primitives

If meaning analysis is possible at all, it would seem only logical that there must be a set of basic terms which cannot be defined or reduced further which we would reach as the endpoint of all the analysis Such elementary meanings are these days often called SEMANTIC PRIMITIVES, though they were known to the rationalist philosophers of the seventeenth century as 'simple ideas’

A conceptualist argument for elementary meanings is that without them, we would be unable to understand anything; because to understand anything, there must be something in terms of which it is understood As Leibniz put it back in 1697:

Whatever is thought of by us is either conceived through itself, or involves the concept of another Whatever is involved in the concept of another is again either conceived through itself, or involves the concept of another and so on So one must either proceed to infinity, or all thoughts are resolved into those which are conceived through themselves If nothing is conceived through itself, nothing could be conceived at all (Leibniz 1973:1)

Modern linguists vvho have taken a positive approach to universal semantic primitives include Manfred Bierwisch, Jerold Katz, Ray Jackendoff, and Anna Wierzbicka

Discreteness

Are the terms of semantic description discrete elements, each one clearly separate form the others (like words, for instance)? Or is it necessary to have a system with continuously

One point upon which almost all semanticists agree these day is that there is a certain vagueness and subjectivity to the meaning of many (perhaps most) words, which makes it impossible to pin down the complete meaning in terms of a list of cut-and-dried objective features Take the words man, fruit, or red It isn't hard to think of situations where one wouldn't be quite sure if they were applicable or not Is a male person aged 16 or 17 a man,

or not? Is a tomato a fruii? When does a reddish-orange become red? Obviously, it is hard to say Some linguists have concluded from facts of this kind that the underlying components of meaning are 'fuzzy' rather than fixed and discrete

The influence of Ludwig Wittgenstein, possibly the twentielh century's most brilliant philosopher, cannot be underestimated in this regard In a famous passage Wittgenstein came

to the conclusion that the word game (or, more precisely, German Spiel) could refer to so many different kinds of thing (board-games, card-games, ball-games, chess, noughts-and-crosses, tennis, ring-a-ring-a-roses, etc.) that it was inconceivable that its meaning could be captured in a specifiable set of common features 'What is common to all of them?', he wrote, 'Don't say: "There must be something common, or they would not be called games"-but look and see whether there is anything common to all For if you look at them you will not see something that is common to all, but similarities, relationships and a whole series of them at that' (Wittgenstein 1953:31-2)

In place of the traditional notion of meaning Wittgenstein introduced the notion of family resemblance: 'a complicated network of similarities overlapping and criss-crossing.' The various uses of a world cannot be summed up in a single statement of meaning, he insisted, but overlap and criss-cross in much the same way as the various resemblances (build, features, colour of eyes, gait, temperament, etc.) between members of a family do

Although many have tried, however, no one has been able to adapt the family resemblances idea indo a practical analytical system We will see at many places in this book that vagueness are not actually incompatible with the idea of discrete and specifiable meanings

Linguistic vs encyclopedic knowledge

Most linguists accept that in principle it is possible to draw some son of line between our knowledge of a language and our other knowledge, that is, between linguist knowledge and 'real-world' knowledge (encyclopedic knowledge, as it is usually called) Thus, what we know about the meaning of the words dog, hair, and light bulb (for example) should be distinguishable in principle from our other factual knowledge about dogs, hair, and light bulbs One argument in favour of this view is that linguistic knowledge is essentially shared between all the speakers of a language, whereas real-world knowledge is not Dog-breeders, hair stylists, and electrical engineers, for instance, know a lot of specialised things about dogs, hair, and light bulbs which are not part of the shared meanings of the words

So far so good But what about 'common knowledge'-the kind of thing which almost everyone knows, for instance, that dogs have four legs, bark and wag their tails? It is not always easy to decide whether such information should be regarded as a part of die meaning

of a word or not Geoffrey Leech (1981) expressed the minimalist view when he said that it

is a matter of real-word contingency, and nothing to do with word-meaning, that an elephant has four legs rather than eighty Jurij Apresjan and Anna Wierzbicka are associated with the maximalist view Apresjan (1992: 32-3) says that there is nothing less

material and spiritual experience of a people' Wierzbicka (1985) argues that the full meanings of words like dog, cat, and mouse incorporate a great deal of folk knovvledge; that the linguistic concept of dog, for instance, includes barking, tail-wagging, and much more besides

A minority view, but an important (and perhaps ascendant) one, denies the existence of any boundary between real-world and linguistic knowledge (cf Haiman 1980) Theorists like Charles Fillmore and Ronald Langacker believe that knowledge of all kinds is integrated in the mind to such an extent that it doesn't make sense to partition it into two distinct realms

Semantics vs pragmatics

Linguists generally agree that, when people are communicating, they do more than just attend to what is actually being said hy way of words, grammatical constructions, and intonation patterns They are also alert to what is implied-for example, by the speaker's choice of certain words rather than others, and by the extralinguistic context in which the communỉcation is taking place

A distinction is often drawn between meanings which are encoded in the structure of the language as opposed to meanings which can be derived or deduced from how language is used on a particular occasion, in a particular context Semantics is usually understood to be the study of meaning in the first of these senses, that is, the study of the more-or-less stable, conventionalised meanings of linguistic signs The study of how speakers and hearers interpret meanings in particular contexts-taking account of the physical and social situation, knowledge of each other's backgrounds, and cultural conventions, among other factors-is usually termed pragmatics Roughly speaking, semantics deals with meanings which are coded into linguistic forms, while pragmatics deals with the inferences people make from how linguistic meanings are used in particular situations Even more roughly, semantics deals with words and sentences, while pragmatics deals with words and sentences as used on particular occasions, i.e with utterances

Distinctions like this may sound clear enough in theory, but when it comes down to practical applications there is often disagreement about where the line should be drawn Much of it stems from disagreement about the criteria for stating and testing semantic meaning Some linguists, such as Fillmore, Langacker, and Lakoff, argue that it makes no sense to draw a hard and fast distinction between semantics and pragmatics, just as it makes

no sense (they say) to draw a rigid distinction between linguistic and real-world knowledge

In any case, even assuming that semantics and pragmatics are conceptually distinct fields of study it is obvious that they must be very closely integrated if we are ever to achieve a satisfying theory of linguistic communication There is also plenty of evidence from the study of language change and language acquisition which leads to the same conclusion

In this book we stay largely within the realm of semantics, in the sense just described ít should be noted, however, that there is another use of the term pragmatics, employed hy linguists whose approach to meaning is strongly influenced hy the study of logic As mentioned earlier, in the logical tradition meaning is defined in terms of relationship (truth-conditions) between sentences and an objective outside world In this framework, all aspects

of meaning which cannot be stated in terms of truth-conditions are regarded as part

1.5 Semantic phenomena

As you can probably appreciate hy now, semantics is such a controversial field that there isn't even consensus on what it should cover To conclude this chapter, we take a look at some phenomena which are widely accepted as semantic in nature-that is, at the kind of phenomena any reasonable semantic theory could be expected to cover

Meaning relatíons

There is general agreement that semantic analysis should give an illuminating account of meaning relations, i.e the systematic relations between word meanings The meaning relations which have attracted most attention by semantic theorists are:

HYPONOMY-where the meaning of one word is included in another; for instance the meaning of the word human is included in woman and man; the meaning of say is included

in tell promise, and denounce

INCOMPATIBILITY of various kinds-where meanings conílict; for instance, by being opposites like up and down (complementary antonyms) or hot and cold (gradable antonyms),

or by being in a converse relationship with one another like huy and sell (RELATIOLNAL antonyms), or hy belonging to a set of mutually exclusive words like red, blue, green, etc Most textbooks also say that another important semantic relationship is synonymy or sameness of meaning, but as soon as the term is introduced they have to add that true synonymy is so rare as to be almost non-existent The usual examples are pairs like father and dad, or couch and sofay but these are better termed 'near-synonyms because most people wouỉd say that they don't convey exactly the same meanings

Semanticists in the sừucturalist tradition have spent a great deal of time devising schemes for classifying different kinds of meaning relation Why? Because the basic idea behind structuralism is that meanings is not exist 'in themselves', but only in relation to one another

as part of an overall system of cross-cutting contrasts and similarities Whether or not one accepts the structuralist approach, however, it is obvious that an intuitive knowledge of such meaning relations is part of ordinary linguistic competence You can learn more about meaning relations by doing the Exercises at the end of this chapter

Entailment, contradiction, and paraphrase

Strictly speaking, the term 'meaning relations, is reserved for relations between individual words, but speakers also have an intuitive knowledge of certain relationships between single words and word combinations ( phases and sentences), and these ought to be accounted for

by an adequate theory of semantics ENTAILMENT is a relationship that applies betvveen two sentences, where the truth of one implies the ưuth of the other because of the meanings

of the words involved CONTRADICTION is where a sentence must be false because of the meanings involved

Many obvious entailments comes about because of hyponymy relations between words For instance, because the meaning 'do something to' is included in eat, the sentence MAX ate the pizza entails Max did something to the pizza Similarly, many obvious

contradictions which are subtler than these The contradictoriness of I'm my own mother, for instance, does not come from any special relation between I and mother And if I love her entails I want the best for her, this isn't on account of any obvious relation between the individual words either

Perhaps the most important relationship for the whole enterprise of linguistic semantics is PARAPHRASE - the relationship between a word and a combination of other words with the same meaning For instance, most people would agree that loud means something like "can

be heard from far away" Ultimately, the whole prọịect of describing or explaining meanings depends on paraphrase because we must use words - or other equivalent symbols -

word-to explain other words

Homonymy, polysemy, and generality

HOMONYMY designates a situation in which different words (homonyms) happen accidentally to have the same form; as for instance, bank (as in She robbed the bank) and bank (as in We walked (along the bank), left (as in Turn left) and left (as in He left\ port (as

in The ship left port) and port (as in He drank port) Generally speaking, sameness of form is taken to refer to phonetic form If it is necessary to distinguish terminologically between sameness of phonetic form and sameness of graphic form, one can use the terms homophone and homograph, respectively Because English spelling is not phonetically consistent, there are many English homophones which are not homographs ( e.g two, to, and too; sight and site, or sun and son ); and vice versa, there are some homographs which are not homophones (e.g the written form live can be pronounced to rhyme with grve, or with strive)

Defining homonyms as "different words with the same form", as I have just done, does not take into account the fact that the term "word" can be used in several different ways Linguistics often distinguish between "word as lexeme" and "word as word-form" Lexemes are words through of as items in the vocabulary of a language (roughly, the expressions one would expect to find listed in a dictionary); for example, talk, think, go Word-forms are variant forms which lexemes adopt due to the grammatical rules of the language; for example, talk, talks, talked Word-forms of the same lexeme can be quite dissimilar from one another (e.g think, thought), or even completely unrelated in form (e.g goes, went) When

we look at things from this point of view, we can see that it is possible to have partial homonymy: a situation in which some word-forms of tvvo different lexemes are identical, but others are not (cf Lyons 1995: 55-8) For example, the same word-form found can belong either to the lexemefìnd (as in I can 7 fmd my pen) or to the lexeme found (as in Hefounded the bank in 1922) Since homonymy is not a meaning-based phenomenon we will not pursue these complications here

POLYSEMY designates a situation in which a single word has a set of related meanings Many- perhaps most -words are polysemous For example, the noun chip can mean (i) a small piece of some hard substance which has been broken off from something larger, e.g a chip of wood/ glass; (ii) a small cut piece of potato which is fried for eating, e.g Can i try

om ofyour chips ?; (iii) a small but vital piece of a computer, e.g It's got a faster chip than the old one The meanings are related because they all contain the component "small piece"

a single general meaning which can be used in different contexts Consider, for example, the word wrong as used in these two sentences: We thought that the war was wrong and It was wrong not to thanh your host It would be easy to jump to the conclusion that there are two different meanings involved ( roughly, 'immoral" vs "improper"), but closer thought will show that it is possible to íbrmulatẹ a single meaning which is applicable to both contexts; roughly, saying it is wrong (to do such-and-such ) means that "if one thinks about it, one can know that it is bad (to do such-and-such)" Admittedly, it is not always easy to tell the difference between polysemy and generality

A useful indicator that we are dealing with polysemy, rather than with generality, is the presence of different grammatical properties associated with the (proposed) different meanings Consider the verb to skip, for instance: How many meanings are involved in the following sentences?

a The children skipped happily down the street

b We skipped the first chapter

Even without our intuitions as native speakers of English, we can tell that two distinct meanings are involved because skip is an intransitive verb in (a), but a transitive verb in (b) The existence of different meaning relations can also be diagnostic of polysemy For example, one piece of evidence that the word faithful is polysemous is the existence of two corresponding nouns, faithfulness and fidelity To describe the quality exhibited by a faithful friend, a faithful servant, or a faithful disciple we would speak of this person's faỉthfulness, but to describe the quality designated hy a husband or lover who is faithful we would speak

of his or her fidelity This suggests that faithful has two ( related ) meanings, one of which is more specialised than the other Further support for this idea is that the word unfaithful relates only to the more specialised, "love-related,, meaning; one could have an unfaithful husband bui not an *unfaithful friend

Even when it is clear that two distinct meanings are involved, it can sometimes be difficult

to decide whether these meanings are related closely enough to warrant being considered polysemic This problem arises particularly when working with a language which is not one's mother tongue Ultimately the decision depends on being abỉe to state the common component of meaning, in order to assess how important it is to the semantic structure of the two meanings involved

What would we want to say about to skip for example? Is the meaning in (a) closely related

to the meaning in (b)? For someone who is not a native speaker of English they could seem quite unrelated Let's analyse each meaning more closely The (a) meaning (as in skipping down the street) involves a certain sequence of moving the feet: each foot touches the ground more than once before the other foot does, unlike the normal sequence in walking or running

in which one foot touches the ground followed immediately by the other That is, physically skipping involves missing out one of the usual 'steps' The (b) meaning (as in skipping a chapter or skipping a class) also involves missing out one element in the normal or expected sequence Now that we have analysed each meaning, we can see the connection between them more clearly and would have little hesitation in recognising polysemy

that even through table1 and table2 are identical in form and closely related in meaning, they must be different words because they belong to different part-of-speech categories Strictly speaking, therefore, such a situation doesn't quality as polysemy but you will sometimes see

it designated as such; some linguists use the term heterosemy instead Other examples are behind1 ( preposition, as in behind the couch ) and behind2 (noun, as in He kicked me in the behind)

There are some Exercises on polysemy and homonymy at the end of this chapter

Connotations

What are 'connotations' ? Unlike most of the technical terrms of linguistics, connotation exists as a part of everyday English, where it means something like 'a subtle aspect of meaning' In the strict linguistic sense, CONNOTATIONS are shifting and idiosyncratic associations which a word may have for some speakers but not for others (as opposed to the fĩxed and generally accepted meaning of a word) For example, the words feminist and environment tend to have very different connotations depending on a person's attitudes and experience

Sometimes, however, even linguists slip up and use the term connotation to refer to an aspect of word-meaning which is, so to speak, subjective and value-laden For instance, one textbook says that stingy sees the attitude toward money as bad and thrifty sees it as good Bút- and this is the crucial point- these evaluational components are integral to the meanings involved and are therefore not connotations People are judgemental creatures ít stands to reason that many word-meanings incorporate evaluational components Connotations in the strict sense of the word have only a minor place in semantic analysis, since they are idiosyntactic and not part of word-meaning They are important in the study of semantic change and variation in the speech community, but we will not touch on these matters in this book

Collocations and fixed expressions

The term COLLOCATION refers to a combination of words - usually, to a frequently found combination A fĩxed expression, as ứie term suggests, refers to a collocation which is fíxed to the extent that it could be listed in a comprehensive dictionary Some semantic theorists believe a complete semantic analysis of a word should provide a basis for understanding its appearance in at least some kinds of fixed expression For instance, they would say that the meaning description of snow and mouse should help account for the expressions white as snow and quiet as a mouse Most people find it easy to accept that the meaning of snow includes the component 'it is white', but many wouldn't go as far as saying that the meaning of mice includes reference to the fact that mice are quiet Similarly, some semantic theorists believe that a complete explication of rage and blue, for instance, would help us see the sense in expressions like a ragingstorm or having the blues; but others disagree

How far collocational possibilities foIlow from meaning is a contentious point, and, ỉike most interesting questions about meaning, one which cannot be resolved vvithout a workable method for deciding on and stating meanings in the first place In modern linguistics, the most detailed theory of collocation is that of the Russian linguists Jurij Apresjan, Igor Mel'cuk, and Alexsandr Zolkovskij (Zholkovsky)

collocation paraphrase

encyclopedic knowledge pemantic metalanguage

family resemblance semantic representation

hyponomy theories of meaning: conceptual

incompatibility platonist, truth – conditional

meaning relations semiotic, structuralist

EXERCISES AND DISCUSSION QUESTIONS

+ next to a problem means that a solution or some commentary can be found at the end of the book

1+ In daily life, people use the words mean, meaning, etc in a wide variety of ways, including sometimes to indicate reference In the following examples, mark whether what is intended is sense (S) or reference (R)

a Extinguish has the same meaning as put out S/R

b When he 'my ex\ he meant Helen S/R

c if you look out the window, you'll see who I mean S/R

d What do you mean, you've been 'seeing' my sister? S/R

2+ Decide whether the following are examples of polysemy or of homonymy

a bark {of a dog vs of a tree)

b fork (in a road vs of instrument for eating)

c tail (of a coat vs of an animal)

d steer (to guide vs young desexed bull)

3+ The distinction between polysemy and homonymy is not alvvays easy to draw The example below contain sentences in which a single vvritten word from is used in different ways In each case, decide whether we have polysemy or homonymy Explain your reasons (For polysemy, say what the common element of meaning is or how one meaning is an extension of the other.)

a You've tried the rest Now try the best

My aching limbs cried out for rest

b I can't see the reason for it

What did you see?

d Don't patronise me!

I always patronise this place

4 Consider the following pairs of sentences and decide in each case whether it is an example

of homonymy or polysemy In case of polysemy, state clearly what the relationship between the meaning is

a Lend me your pen

They put the pigs in the pen

b I got a cut lip

The lip of the cup was chipped

c She landed the first punch

We all got stuck into the punch

d When I started, I couldn't even drive in a nail

Suzie decide she'd better drive

5 The following expressions all have something odd about them Say what it is in each case, and what it tells us about the meanings of the words involved in each case:

intentionally murder, accidentally chase, scrutinise carefully, wailed silently, circumnavigate around, male uncle

6+ As mentioned in the chapter, it is usual to distinguish at Ieast three kinds of antonyms Complementary antonyms are like dead-alive: within the relevant domain of meaning you have to be either one or the other Gradable antonyms, on the other hand, allow an indefinitely large number of gradations between one pole and the other; for example hot' cold Relational antonyms (or converses) are pairs in which the relationship between the two words is symmetrical; for example buy-sell and husband-wife Classify the following pairs of antonyms:

(a) cruel-kind (b) win-lose (c) grandparent-grandchild (d) present-absent (e) quictty-shwly (f) serve-receive (in tennis) (g) soft-loud

7+ For each of the following pairs, say whether the relationship between the two words is one of hyponymy If it is, identiíy the hyponymy and the superordinate, i.e the broader or more inclusive word; for example, terrier is a hyponymy of the superordinate dog

(a) cottage-house (b) team-club (c) run-sprint (d) king-queen (e) chess-game (f) hot-warm

8 Consider the pair of words given in (a)-a) below, and decide which of the following terms best characterises the relationship: gradable antonyms, complementary antonyms, relational antonyms, homonyms, hyponyms, synonyms

(a) landlord-tenant (b) high-low (c) two-too (d) teach-learn (e) death adder-snake (f) book (g) give-take (h) hot-warm (i) fish-perch (j) someone-somebody (k) cricket-sport (l) left- right

novel-9+ The following data about Ewe, a language of West Africa, comes from Adzomanda,s (1969) Dictionary of Ewe Homonyms Carefully consider each set of forms and put forward

there is polysemy#some are separate words because they belong to different parts-of-speech, but are closely related in meaning •for some forms, there is neither homonymy nor polysemy because the 'senses' given are not distinct; instead, there is a general meaning which is applicable to the various uses To give you a clear idea of what is involved, try your hand with the set of four forms dzo shown below

dzo fire

dzo 'juju' (i.e sorcerous powers)

dzo to fly (e.g as a bird does)

dzo to jump

Answer: This set of forms contains two unrelated worđs The word dzo, is polysemous, with two meanings: (i) fire, (ii) sorcerous powers The meanings of dzo1 are related through the notions of power and danger The word dzo2 has a single general meaning, namely, 'move above the ground', which requires the English translations 'fly' or 'jump' according to context Now consider the data below Take your time This will need careful thought There is some further information and advice after the data

zo day container set in the ground, used for storing palm wine and water

zo heel

zo - butt end

zo to walk

nya word

nya to chase (away)

nya to wash (clothes), to purge of dirt

CRUISE, D.A (1986), Lexical Semantics Cambridge: Cambridge University Press: ch 4 'Introducing Lexical Relations' (pp 84-111)

Eco, UMBERTO (1984), Semiotics and the Philosophy of Language London: The Macmillan Press Ltd: ch.2 'Dictionary vs Encyclopedia' (pp 46-86)

CHAPTER 2

INTRODUCTION TO LEXICAL SEMANTIC

2.1 The nature of words meaning

In a descriptive introduction to meaning such as this, it is inevitable that the meanings of words should loom large, even though in more formally oriented accounts, words meanings are left largely unanalysed, or are in reduced to mere skeletons of their true selves There are,

of course, more or less reputable justifications for such neglect However, most (linguistically innocent) people have an intuition thát meaning is intimately bound up with individual words; indeed, this, par excellence, is what words are for While such an intuition seriously underestimates other aspects of meaning, it is not, in itself, wrong, and an adequate introduction to meaning should nót shrink from the slipperiness and complexity of word meaning simply because it can not be neatly corralled into a favoured formalization Hence, the present and the following eight chapters will be devoted to various aspects of lexical semantics

2.1.1 What is a word?

There has been a great deal of discussion of the nature of the word as grammatical unit, too much even to summarize here Most of it, anyway, is not relevant to our concerns But it is as well to have some idea of what we are dealing with The notion has notoriously resisted precise definition Probably the best approach is a prototypic one: what is a prototypical word like? Well for our purposes, the classical characterization as "a minimal permutable element" will serve This attributes two features to a prototypical word:

(i)It can be moved about in the sentence, or at least its position relative to other constituents can be altered by inserting new material

(ii)It cannot be interrupted or its parts reordered

In other words, in making changes to a sentence, we are by and large obliged to treat its words as structurally inviolable wholes Let's see briefly how this works Take a sentence like(l):

(1)The government is strongly opposed to denationalization

Reordering appears in such examples as (2)-(4):

(2)The government is opposed to denationalization- strongly

(3)What the government is strongly opposed is denationalization

(4)It is denationalization thát the government is opposed to

And the possibilities for the insertion of new material are as follows:

(5)The (present) government, (apparently), is (very) strongly (and implacably) opposed (not only) to (creeping) denationa!ization, but etc

(6)

Notice that the only possible insertion points between words Words, of course, are separated

by spaces in writing, although not usually hy silences in speech They also have a

characteristic internal structure, in that they prototypically have no more than one lexical root.(This notion will become clearer below, but, for instance, the lexical roots of the following words are shown in capitals:

GOVERNment reORDERing STRONGly deNATIONalization OPPOSed TYPically CLEARer LEXical)

Some words, such as HEDGE-HOG, BUTTER-FLY, and BLACK-BOARD seem to have more than one lexical root These, however, are atypical, and for many of them it is possible

to argue that the apparent roots are not fully autonomous, semantically, but form a fused root Other words have no lexical roots at all: these are the so-called grammatỉcal words like the, and, and of There will be more on the 'lexicaiv 'non-lexical' distinction below

At this point it is necessary to be somewhat more precise about what we mean hy a word

In one sense, obey, obeys.obeying, and obeyed are different words (e.g for crossword purposes); in another sense, they are merely different forms of the sarae word (and one would not, generally speaking, expect them to have separate entries in a dictionary) On the other hand, obey and disobey are different words in both senses, whereas banh (river) and banh (money) are the same word for crossword purposes, but we would expect them to have separate dictionary entries and they are therefore different words in the second sense Finer distinctions are possible, but for our purposes it will be sufficient to distinguish word forms and lexemes Word forms, as the name suggests, are individuated by their form, whether phonological or graphic (most of our examples will be both); lexemes can be regarded as groupings of one or more word forms, which are individuated hy their roots and/or derivational affixes So, run, runs, running, and ran are the word forms belonging to the same lexeme run, while walk, walks, yvaỉking, and walked belong to a different lexeme, walk, distinguished from the former hy its roots; likewise, obey, obeys, obeying, and obeyed belong to a single lexeme and disobey, disobeys, disobeying, and dỉsobeyed, despite having the same root as the fưst set, belong to a different lexeme, distinguished this time by the possession of the derivational affix dis-s A simple test for the derivational affixes (the matter

is in reality, however, complex and controversial) is that they are never grammatically obligatory For instance, in John is disobeying me, disobey can be substituted by watch, vvithout giving an ungrammatical sentence.This is true of all occuưences of dis- On the other hand, any verb which will fit grammatically into the frame John is- me must bear the affix 'ing, showing that it is not a derivational, but an inflectional affix: word forms that differ only in respect of inílectional affixes belong to the same lexeme ít is the word-as-lexeme which is the significant unit for lexical semantics

2.1.2 Lexical and grammatiacal meaning

A distinction is often made between lexical and grammatical meaning (sometimes only the latter is allovved as being properly linguistic) There are dangers in all dichotomies; this one

is harmless provided it is borne in mind that in reality there is a continuously varying scale,

of what might be termed lexicality and grammaticality A convenient way of presenting the distinction is in terms of the sorts of element which carry the meaning in question We can divide grammatical units into closed-set items and open-set items (another dichotomy which disguises a graded scale) Central examples of closed-set items have the following characteristics:

(i) They belong to small substitution sets (perhaps as small as one)

(ii) Their principal function is to articulate the grammatical structure of sentences (iii) They change at a relatively slow rate through time, so that a single speaker is unlikely to see loss or gain of items in their lifetime (No new tense markers or determiners have appeared in English for a long time.) In other words, the category is effectively fixed (i.e 'closed', hence the name)

These may be contrasted with open-set items, which have the following characteristics: (i) They belong to relatively large substitution sets (especially if semantic plausibility

is ignored)

(ii) There is a relatively rapid turnover in membership of substitution classes, and

a single speaker is likely to encounter many losses and gains in a single Iifetime (Think of the proliferation of words relating to space travel, or computing, in recent years.)

(iii) Their principal function is to carry the meaning of a sentence

Both closed- and open-set items carry meaning, but their đifferent functions mean that there are differences in the characteristics of the meanings that they typically cany

A closed-set items, in order to be able to function properly as a grammatical element, has to

be able to combine without anomaly with a wide range of roots, and for this to be possible, it must have a meaning which is flexible, or broad enough, or sufficiently 'attenuated' not to generate clashes too easily, and it must signal contrasts which recur frequently Hence, meanings such as "past", "present", and "future, which can co-occur with vast numbers of nominal notions, are prototypical grammatical meanings

In contrast, there is no limit to the particularity or richness of the meaning an open-set element may carry, as there are no requirements for recurrent meanings or wide co-occurrence possibilities Hence, open-set items typically cany the burden of the semantic content of utterances Because of the richness of their meanings and their unrestricted numbers, they participate in complex paradigmatic and syntagmatic structures

What are called content words (basically nouns, verbs, adjectives, and adverbs) prototypically have one open-set morpheme (usually called the root morpheme) and may also have one or more closed-set items in the form of afixes Lexical semantics is by and large the study of the meanings of content words, and is oriented principally to the contribution that open-set items make to these Grammatical semantics concentrates on the meanings of closed-set items However, a strict separation between grammatical and lexical semantics is not possible because the meanings of the two kinds of element interact in complex ways

2.1.3 Word meaning and sentence meaning

In general, word meanings are not the sort of semantic units that one can communicate with

on an individual basis, unless other meaning components are implicit A word, on its ovvn, does not actually say anything, does not convey 'a whole thought': for that purpose, more complex semantic entities are necessary -built out of words, certainly -having at least the complexity of propositions (argument + predicate) Words (and at a more basic level, morphemes) form the building blocks for these more complex structures

2.1.4 The notion "possible word meaning"

It is worth while to pose the question of whether there are any restrictions on possible meanings for ưords We may approach this in two stages We can first ask whether there are any universal restrictions; and we can then enquire as to the existence of language-specific restrictions

Let us take the first question first Is there anything conceivable that could never be the meaning of a word? It will be as well to restrict ourselves to notions that can be expressed by

a combination of words, otherưise we shall be in really deep water One line of thinking can,

be disposed of relatively quickly It may be thought that no language could possibly have a word meaning, for instance, "to face west on a sunny morning while doing something quickly" I confess that I would be astonished to find such a word But the reason is not that

it is theoretically impossible, but that it would be of such limited utility Languages have words, at least partly, becạuse in the cultures they serve, the meanings such words carry need

to be communicated (Of course, cultural evolution can leave words stranded, as it were, but this does not invalidate the basic point that words at some stage must be motivated in terms

of possible use.) This means that if some culture had a use for the notion expressed, then it would not be surprising if there were a word for it In the case in question, for instance, maybe the word could designate a specific son of art of disrespect toward the Sun God, which carried specific penalties If we take into account the possibility of outlandish (to us) religious beliefs, ít is clear that the scope for improbable word meanings of this sort is (almost) unlimited

Now let us look at a difierent son of case Take the sentence (6):

(7) The woman drank the wine slowly

The notion "drank sloưly" could easily be lexicalized (i.e expressed by a single word): we have in English, after all, verbs such as quqff, and Sip, which combine the meaning of

"drink" with some adverbial manner component Similarly, a verb meaning "drink wine" is not at all implausible, as one of the senses of drink in English is specifically "drink alcoholic beverage" (as in Mary doesn't drink, she'll just have an orơnge juice) In contrast to these more-or-less plausible word meạnings, consider next the possibility of having a word meaning "The woman drank" (blisk), or "the wine slowly" (blenk) On this system, Blisk wine would mean "the woman drank wine", and The woman drank blenk would mean "The woman drank the wine slowly" It seems clear that here we are in the realms not of implausibility, but of impossibility As a further example, consider the phrase very sweet coffee It is perfectly within the bounds of possibility that there should be a single word meaning "sweet coffee", or 'Very sweet", even "very sweet coffee", but it is not conceivable that there should be a word meaning "very coffee" (i.e any adjective applied to coffee would be automatically intensified) What is the difference between the possible and impossible cases? There seem to be two parts to the answer First, a word meaning is not allowed to straddle the vital subject-predicate divide Second, possible word meanings are constrained in a strange way by semantic dependencies It is first necessary to distinguish dependent and independent components of a semantic combination The independent component is the one which determines the semantic relations of the combination as a whole with extemal items So, for instance, in very large, it is large which governs the combinability of the phrase very large with other items Thus the oddness of, say, ?a very large wind is attributable to a semantic incompatibility between large and wind- there is no inherent clash between very and wind, as the normality of a very hot wind demonstrates By similar reasoning, the independent item in warm milk is milk, and in drink

warm milk is drink By following this line of reasoning, we can establish chains of semantic dependencies For instance, the chain for very young boy is:

"veiy"→ "young"→ "boy"

and that for drink warm milk is:

"warm"→ "milk"→ "drink"

The constraint that we are looking at says that the elements that constitute the meaning of a word must form a continuous dependency chain This means, meaning of a word must form a continuous dependency chain This means, first, that there must be a relation of dependency between elements This rules out "wine slowly" as a possible word meaning, because there is

no dependency between "wine" and "slowly" in "Drink wine slowly" Second, there must be

no gaps in the chain which need to be filled hy semantic elements from outside the word This rules out cases like "very—milk", where the dependency chain would have to be completed by an external kem such as "hot"

Another way of approaching the question under discussion is to take an extensional viewpoint and ask what are the characteristics of a 'possible nameable' A full answer to this question would deal with possible nameable objects, events, states, and so on.The present discussion will deal only with the first of these, and will follow Pulman (1983) Nameables,

in general, are distinguished linguistically hy the ability to take proper naraes (obviously), but also hy the ability to be referred to by means of singular, non-collective count nouns (at least in English) (The characterization of nameability in states, processes, events, and so on, would require reference to adjectives and verbs, as well as nouns.)

We shall illustrate the sort of argument involved in saying what is nameable hy reference to

a particular son of nameable, namebly, physical objects.The concept of physical object seems

to involve at least the notions of perceivability, relative continuity in space and time, and potential detachability from surroundings However, not all physical objects, by these criteria, are naturally nameable A couple of quotations from Chomsky are relevant here: (i) the most elementary notion we have, the notion 'physical object' seems to be quite complex One wings of an aeroplane is an object, but its left, though equally continuous, is not (1976:203)

(Since, in an obvious sense, the left half of an aeroplane wing is a physical thing, we may interpret Chomsky as talking about nameability.)

(ii) there are no logical grounds for the apparent non-existence of words such as LIMB, similar to limb except that it designates the single object consisting of a dog's four legs so that its LIMB is brown would mean that the object consisting of its four legs is brown Similarly, there is no a priori reason why a natural language could not contain a word HERD, like the collective herd except that it denotes a single scattered object with cows as parts, so that a cow lost a leg implies the HERD lost a leg, etc (1965:201)

The existence of physical things which intuitively do not seem to be maneable implies the existence of principles of nameability, which may not be rigid, but which will at least render some 'things' more readily nameable than others ( The notion of 'prototypicar, is relevant here: what we are looking for are the prototypic features of nameability.) A fairly basic suggestion is that to be nameable, a physical thing must be bounded, that is, it must have boundaries sét on the basis of either physical detachability, or characteristic function, appearance, or behaviour This is, of course, pretty vague, and a full treatment would critically examine all these terms, but take, for instance, Chomsky's left half of an earoplane wing This is physical, but while the whole wing is bounded hy distinctive

function and appearance, the left half is not separated from the rest of the wing hy any salient tunction or visual discontinuity, nor does it behave in a characteristic way In this way, the left part of an aeroplane wing differs from an earlobe, which is visually separated from the rest of the ear, and the tip of the tongue, which has no visual separateness, but has its own characteristic functions and a special place in our experience of our bodies (Notice that some nameables, by the definition given above, will count as whole things and others as parts of those wholes; that is a separate question.)

Chomsky's examples of LIMB and HERD are more diffícult and controversial First it must

be recognized that some 'scattered' individuals are nameable, such as fences, constellations, villages, forests, and so on To adapt Pulman slightly, we can say that nameable collections

of othenvise independently nameable entities generally show one (or more) of the following features:

(i) The collection is relatively spatio-temporally contiguous (fence, forest, village) (ii) It is the product of human agency (fence, village, artistic installation)

(iii) The members of the eollection jointly fulfill a function not fulfilled hy any of them separately (fence, bikini)

Notice that both Chomsky and Pulman insist on a distinction between singular scattered objects and collectives But the criteria are not clear Pulman refers to "things which are designated hy sinsular count nouns or proper names but nevertheless regarded as plural: collective words like herd, pile and flock, and proper names like the United States or the Commomvealth" (Notice that the possession of one of the features mentioned above seems

to be necessary for these collective words.) But what is meant hy "are regarded as plural"? A word like committee can take plural concord with a verb: The committee have decided, but this is not the case with, for instance, pile: *The pileofstones are black

Chomsky is not much more explicit for LIMB, although he is for HERD In the case of LIMB, he gives as the son of sentence which would prove that there was a genuine word LIMB, something like The LIMB of the dog is brown Actually, such cases are not rare: The foliage of this tree is light green means simply that the leaves of the tree are light green Chomsky's requirements for HERD are perhaps more strict It seems that for HERD to be a bona fide example, a part of a cow must count as a part of a HERD (which it clearly does nót for the 'normal' word herd) Notice that this criterion would rule outfoliage: one would not say The foliage of this tree has prominent veins, bút The leaves of this tree have prominent veins (Similarly: *John's priceless library of first editions has lost several pages) But it is not clear that it holds for fence, either (and others discussed hy Pulman as bona fĩde singular non-collectives) If the separate (and separated) posts which constituted a fence each had a hole in it, would one say The fence has holes in it or The fence poles have holes in them? I would be happier with the latter On the other hand, I would be happy with You can 7 wear íhis bikini because it has holes in it (cf also This bikini has a reinforced gusset)

I suspect that there is, in fact, no sharp distinction between the HERD type of example and the herd type I am inclined to agree with Chomsky, however, to the extent that the HERD type are somewhat rare (It may be that more relevant factors remain to be discovered.) Obviously a similar investigation needs to be carried out on staíes, actions, processes and events, and so on, to see what factors determine nameability hy a single lexical item (notice that proper names are largely (?totally) confined to nouns)

Of a more controversial status are like the putative 'impossible' words benter and succeive (Jackendoff 1990:261) Let us consider benter first This is proposed as a logicalỉy coherent

converse of enter which cannot be lexically realized Sentences such as (7) are fully normal: (8) Mary enter the room

The proposed converse of this would be (8):

(9) The room bentered Mary

(On the pattern of: Mary followed John and its converse John preceded Mary.) The other example sometimes cited is succeive, which is intended to denote the true converse of receive:

(10) John received the parcel

(11) The parcel succeived John

The explanation given for these constraints is expressed in terms of semantic roles such as agent, patient, etc There is a certain plausibility about the claims However, although there may be some resistance to words having such meanings, and in the cases cited there are no obvious candidates, the prohibition is perhaps not absolute, as the following observations suggest First, the meaning of benter is not all that far removed from one reading of receive

We also have words such as envelop, incorporate, which seem to have the right sort of meaning In the case of succeive, the word reach appears to encode approximately the right sort of meaning:

(12) I sent John a parcel; he received it yesterday

(13) I sent John a parcel; it reached him yesterday

The constraints on word meaning discussed above would seem to be universal in nature However, there also exist constraints of a more language-specific type Some languages seem

to proscribe the packaging together of certain sorts of meaning in a single word A single example will suffice Consider sentence (13):

(14) John ran up the stairs

Here, the word ran encapsulates two notions, that of movement, and that of manner This is

a common pattern in English:

(15) John crawled across the road

staggered into the room waltzed through the office etc

However, this pattern is not possible in many languages including French In French, such sentences must be rendered as in (15):

(16) Jean monta Pescalier en courant

Here, the notions of motion and direction are jointly packased into monta, but manner has

to be expressed separately (Notice that the French pattern is not prohibited in English: John mounted the stairs running, but is márkedly less natural.)

2.2 The major problems of lexical semantics

Linguistic with diffrerent theoretical commitments will give different accounts of what the core tasks of lexical semantics are; the following is an attempt at a relatively theoretically neutral summary

2.2.1 Description of content

Describing content is in a sense the most obvious task: how do we say what a word means? Unfortunately, even at this level of generality ít is impossible to escape the tentacles of theory, because there are scholars who maintain that the notion 'the meaning of a word' is not

a coherent one; and for those who believe there is such a thing, the nature of the description

of it will hang crucially on what sort of thing it is believed to be We shall look briefly below

at some of the options

2.2.2 Contextual variation

However one characterizes the notion of the meaning of a word, one is forced to confront the fact that the semantic import of a single word form can vary greatly from one context to another There are various theory-dependent strategies for attacking this problem, but the facts will not go away: the variation must be accounted for Variation is not random: part of satisfactory account will identify and explain patterns of variation

2.2.3 Sense relations and structures in the lexicon

Regular patterns appear not only in the nature and distribution of the meanings of a single word in difierent contexts, but also between difierent words in the same context This result

in structured groupings of words in the vocabulary on the basis of recurrent meaning relations

2.2.4 Word meaning and syntactic properties

An important question is whether and to what extent the syntactic properties of words are independent of, or are controlled hy, their meanings There are still many difierent views on this topic

2.3 Approaches to lexical semantics

2.3.1 One-level vs two approaches

A major dividing line which separates semanticists is the question of whether a distinction can be made between semantics and encyclopaedic knowledge Those who believe such a division can be made often draw an analogy with phonetics and phonology Human beings can make and learn to recognize an almost infinite variety of speech sounds, but in any particular language, only a handful of these function distinctively to convey meanings, or enter into systematic relations of any complexity These are true linguistic elements on the 'sound' side of language (Saussure's cxpression plane) In a similar way, the variety of 4raw' meanings is virtually infinite, but only a limited number of these are truly linguistic and interact systematically with other aspects of the linguistic system The vast detailed knowledge of the world, which speakers undoubtedly possess, is, according to dual-level view, a property, not of language elements, but of concepts, which are strictly extralinguistic Truly linguistic meaning elements are of a much Mearner' son, and are (typically) thought of

as (more) amenable to formalization One criterion suggested for recognizing Minguistic' meaning is involvement with syntax, whether by virtue of being the

meaning carried hy some grammatical element, or because it coưelates with such factors as agreement patterns or sub-categorization of mạjor syntactic categories

Partisans of the single-level view claim that no non-arbitrary basis for assigning aspects of meaning (or knowledge) to the 'semantic' or 'encyclopaedic' side of a purported dichotomy has been put forward which survives even a cursory scrutiny Most cognitive linguists would take the view that all meaning is conceptual, and that the 'extra' level of structure proposed

hy the two-level camp does not actually do any theoretical work The distinction between grammatical and ỉexical/encyclopaedic meaning is not necessarily denied, but it is likely to

be seen as a continuum, rather than a dichotomy, and entirely conceptual in nature

2.3.2 Monosemic vs polysemic approaches

The point at issue in relation to the distinction between the monosemic and the polysemic approach is how many meanings ought to be attributed to a word There is no dispute about clear-cut cases of homonymy, like that of banh, where there is no conceivable way deriving one meaning from the other The dispute centres on clusters of related senses characteristic of polysemy The monosemic view is that as few senses as possible should be given separate recognition in the (ideal) lexicon of a language, and as many as possible derived from these The argument usually goes like this: if one reading of a word is in any way a motivated extension of another one, then onỉy one should be recorded, and the other should be left to the operation of lexical ruỉes, which in general apply to more than one instance and hence represent systematicity in the lexicon

The polysemic approach rẹjects the assumption that a motivated extension of a word sense does not need to be recorded in the lexicon The basic reason for this is that lexical rules only specily potential extensions of meaning, only some of which become conventionalized and incorporated in the lexicon: others are possible, and may appear as nonce forms, but there is none the less a clear distinction between these and those which are established (in principle, anyway: actually there is a continuous scale of establishment) Take the case of drink In many contexts, it is clear what is being drunk, but obviously one would nor wish to create a different lexical entry for drink corresponding to every possible drinkable liquid To this extent, the monosemists and the polysemists would agree However, it is possible for some particular drinkable items to be incorporated into a specific reading for drink In principle, any class of beverage could be incorporated in this way, but in fact, in English, only

"alcoholic beveraged" can be encoded thus: I'm afiraid John has started drinking again Now

in principle, this could have happened with frruit juice instead of alcohol, but it is a fact about the English lexicon that drink has one of these possibilities, but not the other The majority view nowađays is probably monosemic, but the position adopted in this book is polysemic

2.2.3 The componential approach

One of the earliest and still most persistent and widespread ways of approaching word meaning is to think of the meaning of a word as being constructed out of smaller, more elementary, invariant units of meaning, Somewhat on the analogy of the atomic structure of matter (although the immediate inspiration for the first proposals on these lines was not physics, but phonology) These 'semantic atoms, are variously known as semes, semantic features, semantic components, semantĩc markers, semantic primes (to cite a few of the

terms) Here, the merest outline of the approach is presented; componential semantics is treated in greater detail in Chapter

Probably the first statement of a componential programme for semantics within modern linguistics ws due to Hjelmslev (1961) He believed as a matter of principle that the meaning side of the linguistic sign should show the same structuring principles as the sound side For him the notion of reduction was of mạjor importance The phonological structure of hundreds

of thousands of different signs in a language can be analysed as combinations of syllables drawn from a list of a few hundred, and these, in turn, can be shown to be built out of phonemes belonging to an inventory of fifty or so, thus arriving át the ultimate phonological building blocks, the distinctive features, whose number is of the order of a dozen In the same way, the meaning side of signs should be reducible to combinations drawn from an inventory significantly less numerous than the stock of signs being analysed Hjeimslev did not have any universalist pretensions, each language being unique and needing an analysis in its own terms, nor were his 'figurae' (his term for the basic elements) in any way abstract: they were the meanings of words in the language What he seemed to have in mind, therefore, was the discovery of a set of basic words, out of whose meanings all other word meanings could be constructed Hjelmslev was the first structural semanticist: the approach was developed considerably by European linguists, with a German variety and a French variety

A componential approach developed in America, seemingly independently (and largely in ignorance) of the movement in Europe It first appeared amongst anthropological linguists, and scored a significant success in reducing the apparent impenetrable complexity of kinship systems to combinations from a limited set of features A new version, proposed hy Katz and Fodor (1963), appeared in the context of early Chomskyan generative grammar This was much more ambitious than anything which had appeared previously: íỉrst, Ít formed an integral part of a complete theory of language; second, it made claims of universality and psychological reality; and third, the features were nót confined to the meanings of existing words, but were of an abstract nature This approach did not take hold in mainstream generative linguistics, and among current generativists a thoroughgoing componential approach is found only in the work of Jackendoff (1983,1990,1996)

An extreme version of componential semantics is found in the work of Wierzbicka (1996) This is a highly original approach, which is not offshoot of any of the approaches described above, but takes its inspiration from much earlier philosophical work, notably by Leibniz (1903) Wierzbicka's view is that there exists a very restricted set of universal semantic atoms

in terms of which all conceivable meanings can be expressed Her inventory of primes is astonishingly small (she started out with eleven, but the list has now grown to fifty or so), and they are not abstract, and hence unverifiable by direct intuition, like those of Katz and Fodor, but are concrete, and any analysis should satisíy the intuitions of native speakers

2.3.4 'Holist' approaches"

It is a belief of all componentialists that the meaning of a word can, in some useflil sense,

be specified, in isolation from the meanings of other words in the language Among philosophers of language, this is knowrras the localist view For a localist, contextual variation can be accounted for by rules of interaction with contexts The contrary position is the holistic view, according to which the meaning of a \vord cannot be known without

taking into account the meanings of all the other words in a language There are various versions of holism: two will be outlined here

2.3.4.1 Haas

I first learnt semantics from W.Haas (1962,1964), whose highly idiosyncratic view of meaning derives from an aspect of Wittgenstein's work, namely, his 'use' theory of meaning, which is encapsulated in the dictum: "Don't look for the meaning-look for the use" In other words, the meaning of an expression is the use to which it is put As it stands, this is not very helpful, merely suggestive Haas gave it a personal twist, inspired by J.R.Firth's dictum:

"Words shall be known hy the company they keep" This interprets 'use' as the contexts, actual and potential, in which the expression occurs normally (i.e without anomaly) Haas went further than this He said that the meaning of a word was a semantic fíeld (not the usual semantic field) which had two dimensions: a syntagmatic dimension, in which all possible (grammatically well formed) contexts of the word were arranged in order of normality Relative normality was for Haas a primitive In principle, 'context' includes extralinguistic context; but Haas argued that since every relevant aspect of extralinguistic context can be coded linguistic contexts The word's semantic field, as understood hy Haas, constitutes its meaning Notice that every word therefore participates in the meaning of every other word (he was inspired here hy Leibniz's monads); there is therefore no distinction between word meaning and encyclopaedic knowledge Haas's view was that the semantic field of a word (as

he defined it) actually constituted the meaning of the word; here, the view will be taken that the semantic field of a word reflects its meaning

2.3.4.2 Lyons

A second variety of holism is represented by Lyons (1977) The essence of this approach is the quintessentially Saussurean belief that meanings are not substantive, but relational, and are constituted by contrasts within the same system Lyons states that the sense of a lexical item consists of the set of sense relations which the item contracts with other items which participate in the same field Sense relations, he insists,- are not relations between independently established senses; one should rather say that senses are constituted out of sense relations So, for instance, the meaning of horse should be portrayed along the lines shown in Fig 2.1

In this system, the links are of specific sorts, such as "is a kind of' (e.g horse.animal), "is not a kind or (e.g horse:cow\ "is a part of * (e.g mane-:horse), "is characteristic noise produced hy" (e.g neigh:horse\ "is a dvvelling place for" (e.g stable:horse), and so on Since the words illustrated also enter into relations with other words than horse, the full meaning of horse is a complex network bf relations potentially encompassing the whole lexicon

2.3.5 Conceptual approaches

Conceptual approaches (at least as the term is used here) are single-level approaches and identity the meaning of a word (or at least a major part of it) with the concept or concepts it gives access to in the cognitive system Among cognitive Hnguists, the prototype model of concept structure holds sway

The origins of the prototype approach can be traced to Wittgenstein (1972) (who initiated more than one line of thinking that was to iníluence linguistics) He is usually credited with being the first to challenge the classical Aristotelian notion of natural categories as being definable in terms of necessary and sufficient criteria He put forward the well-known example of GAME, challenging his readers to come up with the necessary and sufficient criteria for something being a game None of the obvious suggestions is criterial: involves physical activity has winners and losers is played for amusement has rules,etc

None of these is either exclusive to games or necessary for something to be a game Wittgenstein proposed the notion of family resemblance: the members of a large family typically resemble one another in a variety of ways, bút there are no features which they all have, and there may be members who share no features, but these will none the less be linked

to the others by a chain of resemblance Although important in breaking the stranglehold of the Aristotelian theory, this notion is not very helpful for semantic analysis

Wittgenstein did not say what family resemblance consisted of, in particular, how in-family resemblances differ from out-family resemblances In other words, there was nothing other than arbitrary stipulation to stop everyone in the world from belonging to the same family (However, a similar problem still bedevils modern descendants of Wittgenstein's family resemblance.)

The notion of non-Aristotelian categories was taken up and fuirther refined hy cognitive psychologists, especially Rosch (1973,1978) and her co-workers, who established what is now known as prototype theory as an account of natural categories On this account, members of a category are not equal- they vary in how good they are, or how representative,

of the category The very best are the prototypical members, and the category is essentially built round these: other examples are assimilated to the category or not, according to how closely they resemble the prototype

Jackendoff (1983, 1990,1996) is another linguist who locates word meaning in conceptual structure (his picture of conceptual structure bears strong resemblances to that of the cognitive linguists) Like the cognitive linguists, he sees no need for an intermediate 'linguistic semantics' Unlike many cognitive linguists, however, he is strongly componentialist, and believes that intuitively perceived relationships should (must) be accounted for in terms of shared semantic building blocks He also has a strong predilection for precisely formalized representations Perhaps the most important characteristic separating Jackendoff from the cognitive linguists is his continued espousal of the Chomskyan precepts

of strong innateness, the insufficiency of general cognitive abilities to explain all linguistic behaviour, and the autonomy of syntax

2.3.6 Formal approaches

Formal approaches to semantics attempt to express the facts of meaning through a strict formalism, preferably closely related to one of the standard logics The hoped-for pay-off from adopting this sort of approach includes greater explicitness, testability of hypotheses, easier link-up with syntax, and machine implementability Those who are less sympathetic to this kind of approach point to the existence of signifícant aspects of semantics which are continuously variable, and to the somewhat meagre descriptive results so far achieved Formalist approaches will not be given any prominence in the present work, vvhich aims rather at a certain descriptive richness

CHAPTER 3

ON ENTERING THE REALMS

How is it that I get ideas in my head when I hear sounds of my language, but not when I hear sounds of another language?

Humpty Dumpty said to Alice that he could make a word mean exactly what he wanted

it to mean; that it was only a question of who was to be the master Can you? Can he really?

It is no doubt more interesting to talk about the meanings of words and sentences than to try to talk about what meaning is Nevertheless, it is worth a few words at least, to know where we are going and how to fit things together

The English verb to mean can be used for many things: in just about any case where you can learn something important, F, from something else, X, we can say 'X means F (to you)' For example, if we can know that it is going to rain from some dark clouds blowing up, it is perfectly natural to say 'Those clouds mean rain.' The optional '(to you)' is used in case you would conclude it but others might not

what did she mean by winking like that?

that nasal accent means he comes from Chicago

how she feels means a lot to me

who do you mean to refer to by that?

who do you mean by 'that guy with an earring'?

what does the word teacher mean?

what do you mean by freedom?

Really, however, we are not so interested in this particular word of English, even if philosophers worry a lot about Ít As would-be scientists, we are interested rather in what the world of meaning in language is like, and freely accept that English words might not match that world too closely The noun meaning is closer to our needs, for it cannot be used in all the ways the verb can

Q Which of the ways that to mean was used above can be paraphrased (i.e said in a different way) with the word meaning? The last, for example, can be paraphrased as 'The meaning of (the word) freedom for me is '1

On a most basic level, the meaning that we are interested in is cognitive or descriptive meaning: what is communicated when one person tells another something That is, we are interested in the meaning that can be expressed in language, meaning that describes something There are also some other things that we can léarn by listening to a person, like his social or geographical origins or his emotional state, that we should not want to call meaning If we heard 'Gimme a cup o' wa'er', we can guess a lot about its speaker-where

he comes from as well as the fact that he is thirsty-but all that is hardly his meaning He did not say he was from London, only that he wanted some water - and he might be an actor putting on a local style of speech Happily, these other things are seldom subject to much control, so descriptive meaning can usually be distinguished as being easily controlled hy all competent speakers Of course, some people control it less well, but by learning more about

it we can all control it better

Q Is there any difference between saying 'She gave me, a man from York, a call' and 'She gave us a cair in a male Yorkshire voice? If I were a New Yorker, would I be lying if I said these, pretending, or what?2

3.1 Language as a tool

Although we have excluded a lot of things from meaning, there is plenty left Enough, in fact, to make this one of the most exciting areas in language to study; it is, I believe, much of what makes human beings human, and what allowed us to dominate all other animals so that now they work for us or entertain us Each type of animal has some special weapon or defense, like running fast or sharp teeth, sharp ears, long claws or a long neck We humans excel in none of these ways, but we have language

How is language stronger than claws and teeth and speed? Simple: ten men armed with only stones or clubs can take a tiger or an elephant by surrounding it and coordinating their actions so that several attack its weak spots whenever it attacks one of them Some animals like dogs or sheep use numbers as a weapon, but in coordinating ourselves we become like a single animal spread out in many different places! This was so effective that even thousands

of years ago, the only wild animals that didn't avoid human beings were dead animals

It used to be said that what distinguished human beings was that they make and use tools Unfortunately for our pride, however, some types of great apes have seen to use tools, and even to invent simple ones, and some birds will also do so Chimpanzee mothers have even been seen in the wild showing their offspring how to use things as tools - bút how much can you learn hy imitation? Some animals have communication systems, of course, but they are

so limited in what they can communicate thát linguists reíuse to call them languages The most extensive one known, thát of some honey bees, can (apparently) communicate only a location - where some nectar is to be found Our languages hy contrast can communicate apparently anything - locations, emotions, facts, procedures, possibilities, fantasies, lies and many other things

Q: What is communicated in:

a dog's bark, whine, snarl?

a cat s meow, purr, ?

There is no doubt that human beings are superior to most or a!l animals in being able to think, but what is thinking? A lot of our thinking is done in our native languages, son of like talking to ourselves If we can understand what meanings a language can express, we will be much closer to knowing what thinking is In any case, it seems doubtful whether we might be any better than other primates in thinking without language

Q Might there be something that human language cannot communicate? What? Please tell

me about it, if you can.4

I am not even convinced that we can think much better than the great apes - you or I, that is What tools have you invented recently? Our languages give us, almost from birth, the refined wisdom of our ancestors Every useful procedure and tool has a name, and when we learn our first language we learn these names and what to apply them to The words for concepts that

do not prove so useful are normally forgotten in a generation or so.5 By going on from what our ancestors have learned we have left our prirriate cousins far behind, for each one of them has to start once again from the beginning No doubt the chimps have had their geniuses, but their discoveries are always lost when they die

Q When you take an elementary course in chemistry, or poetry, or any subịect, a lot of your effort goes into learning the specialized vocabulary What are you really learning?6

Even that is not enough; they must get the right ideas If I pronounced the sound [hai] in the right way, they might think I was agreeing with them if they thought I was speaking Japanese or Cantonese Chinese, while I might only be greeting them in English Or I might

be saying ashes in Japanese or high in English Unless they get the idea I want them to, we don't have good communication A simple diagram of this is in Fig 1.1

Figure 1.1 Diagram of ‘good communicarion’

Communication is successful if the idea they get (the impact on them) is the same as what I intended them to get (my intent) If they don't match, then we have poor or no

Figure 1.1 Diagram of'good communicarion,