Vista previa parcial del texto
¡Descarga Chapter 2 Saeed´s Semantics.pdf Chapter 2 Saeed´s Semantics y más Apuntes en PDF de Semántica Formal solo en Docsity!
Introducing Linguistics Semantics 1 Andrew Spenucr Pheonology 2 John Saeed Seman Fohn 1. Saeed Trinity College Dublin Bi BLACKWELL - Copyright 3 loha L Saced, 1997 ¡ar of this work a: ho right ol Juan L Saved Lo de identified as a arsemed in accordance with tac Copyright, Dosigns and Puterts Act l Bluclovell Publishers Lró 108 Cowley Road Oxford OX4 1JF, UK Bl kwell Publishers nc 330 Main Street Melden, Massachuicotts 02148, USA All rigl af criticism and soview, no par: of lus publica oval system, or transm:tted, in any form or by any means, electroa.c, ls a ofthe pus isher. Except in the United States of Amer tus dook is sold subject to (he condition that:t shall not, byoway oí trad , hired om, or MES ted without the publisher's prior consent in any lora of bind or cover other th milar condi wise, be an ticarion Da the B itish La of Congress € jon Dato sed, Joan 1 Libri es: 2) the Centre for Language ation Studies, Trinity College Dublin." —Pref < bibliosrapitical references and index. SAN 0-631-20024-7 (hbk: alk. paper) ISBN 0-631-20025 5 (pbk alk. paper) L es. 1. Titie. Il. Series. lar: Jack Messeager : Lisa Eator Fatan ion Coniroil Ty etin 20.5 or 12pt Plancin by Graphucralt Typeseuters Limited, Eong K Printed and bound in Great Britain 1. Tetemarionel Licr ted, Padstow, Cornw by his >0ok :s purted or acid-free paper To Joan DOCOGCO OAMI AU ALA A AAA AAL AADAAAAMAAM080:00000000 MAMA» As» | 24 Pre iminar es and the world, while refer is used for the action of a speaker in picking out entities in (he world. We will adopt this usage, so hal i'I say 4 sperrow fee into the room, T arm using the two noun phrases a sparroc and ¿he room reler to tings in the world, while the nouns sperrow and room denote certain classes of items. In other words, referring is what speaxers do, while denoting is a property of words. Another diflerence which follows trom these delinitions is ráar denotarion is a stable relationship in a languag: which is not dependent on any one use ol a word. Reference, on the other hand is a momerr by moment relarionship: what entity somebody refers to by using the word sparrow depends un (he context. As we shall see, there are different views of how semanricists should approach (his abiliyy to talk about the world. Tivo of these are particular: important in current semendo theories: we can call tiem the referential tor denotational) approach and :he representational approach. For semanticists adopting the first epproach this accion of putting words into relationship with Ue world is meening. so that to provide a semuntic description tor a language wwe need to show how the expressions of the language can “hovk onto” the world. “Thus theories of meaning can be called referential (or denotational) when their basic promise is that we can give the meaning of words and sentences by showing how they relale Lo situations. Nouns, for example, are meaningíul because they denote enrities in the world and sentences hecause they denote situations and events.! In this approach, the difference in meaning bebween the sentences: 2.3 — There isa casino in Grafton Stre 2.4 There iso't a casino la Grafton Streer. ariscs from the facr thar the two sentences describe different simations. 1£ we assume the sentences were spoxen ar the same time about the same street, then they can be said ro be incompatible, e. one el them is a false description of the situation. For semanticists adopring the second approach vur ability to talk about the world depends on our mental models of ir. In this view a language represents « Tacory about rea.ity: about the types of things and simations in the world. Thus, as we shall sec in later chapters. a speaker can choose to view the same situation in different ways. Example 2.5 below shows us that in English we can view the sane situation as eirher an activity (2.54) « a state (2,5b): 25 a. Joan is slccping. b. Joan is asleep. Such decisions are induenced by cach language's conventional weys of vicw- ing situations, We can compare the three ways of saying thar someone has a cold in 2.6 5 below vicaning, Thought and Reality 235 2 English You nave a cold. 2.7 Somali largab ban ku haya, 4.cold FOGUS you has “A cold has you” ie. “You have a cold? 2.8 Trish Tá slaghdán orr, is acold on.you “A cold is on you? “You have a cold” In English and Somali, 2.6 8: 2.7, we see the situation viewed as posses- sion: n English the person possesses the discaso; in Somal te dis possesses the person. In Irish, 2.8, the situarion is viewed as location: the person is the location for the disease. We shall look ar such differences in later chaprers. The point here is that different concepiualizations influence the description of the real world situations. Theorics of meaning can be called representational when their cuphasis is on the way that our reporls bout reality are infiuenced by the conceptual structures conventionalized our language We cun see these wo approaches as focusing on different aspects of the same process: telling about the world. la referen from language being arrached to, or grounded ia, reality. In representational approaches meaning derivos from language being a reflecion of our concep- tual suructures. This difierence of approach will surface droughour this book and we outiine a specific referential Uicory in chapter LO, and vessions Of representational theories in chapters 9 and 11. These wo approaches ay: influenced by ideas from philosophy and psychology and in rhis chapter we review some ol the most important of these. We begin however with lan- guage: by looking at the diffcrent ways linguisric expressions can be used to refer. We tien go on to asx weheiher reference is indeed all of incaning and examine arguments that reference relies on conceptual knowledge. Here we review some basic rheories about concepts from the philosophical and psy- caological literature, Finally we discuss how these ¡des from philosophy and psychology have influenced Lhe ways that semanricists view the task of describing meaning theories, meaning derives 2.2 Reference 5) ol Types of reference We can begin our discussion by looking brielly at some major differences in the ways that words may be used to refer. For the introductory purposes of > de e e AAC... FECTECCCETECEECECECECTCTELLELLELLULLUCULL SA 26 Preliminaries «his chapter we will for (he most pert confine our discussion to the 1efer- ential possibilities of names and noun phrases, which together we can call nominals, since the nominal is the linguistic unit which most clearly revea!s this function of language. Larer, in chapter 10, we loox ar a more fully Nledged theory of denorarion and discuss the denozatins vÉ other linguistic elements like verbs and sentences. ln this section we discuss some basic distinecions in reference. Referring and non=refering expressions We can apply this distincion in two ways. Firstls there are linguistic expressions which can never be used ta refer, for example the words s hese words do of course contribute meaning to the sentences they occur in and thus help sentences denote, bur they do not themselves identify entities in Tac world. We will say thal lhese are intrinsically non- and 2.33 the same individual is referred to by 2 name, Anwar El Sadat, und by a definite description. the President of Egypt. These two expressions would share lle same refercnt but we probably want to say they have different meanings. If so, Ulieze is more ro meaning iban reference. One mieht object that names de not really have any meaning. This is often so n English, where we commonly use names derived from other languages like Hebrew, Greek etc., bur is nor necessarily true of other cultures. Sall, even if we allow this objecrion, he phenomenon is not restricred to Mames. You might refer to the woman who lives nest door to you by various de- seriprions like my neishbour, Pess mothor, Michaels wife. the Head e Science ar Sí Helen's School, etc. lt seems clear that while these expressions might all refer to the sume individual, hey differ in moaning. Indeed it is possible =o know that some nominal expressions refer to an individual but bo jgnorant of others that do. We might understand expressions like the Presid F Unired States and ¡he Commnander-in-chief of he United States Arme: not now that they both refer to the same man, This has traditionally been an issue in the philosophical literature where we can find similar bur more complicated examples: rhe logician Gorrlob Frege (1980) pointed out that a speaker might understand the expressions the morning star and the evening sar and use them to refer to Lwo apparently diferent celestial bodies without A knowing Uat they both refer to sightines of Venus. For such a speaker, Frego noted, the following sentence would nol be a tautology: 23 1e morning star is the evening s:ar. 2.34 z and might have a vory different meaning from the referentially equivalent sentence (but for our hypothetical speaker, much less informarive): 2.35 'enus is Venus. we can understand and use expressions dha do nor have a real world referent, and we can use different expressions to identify the saimne referenz, and even use bwo expressions without being aware thar they share the same referent, then it seems likely that meaning and reference are noz exactly the same Ling. Or to put it another way; there is more to meaning than ref- erence. How should we characterize this exrra dimension? One answer is ro Jollow Frege in distinguishing rwo aspects of vur semantic knowledge of un expression: its sense (Frege used the German word Sínm) and its reference eses Bedercunz). ln this division, sense is primary in thar it allows ref rente: ir is because we understand the expression ¿he Presidens of Ireland that we cán use ir to refer Lo a particular individual ar any given time. Other uys of describing this same person will differ in sense bul have the same reference, ? Tf we follow this line of argument, then our semantic dhcory is going to be more complicated than the simple referential theory: the meening ol an tapression will arise both from its sense and its reference. In 1h nex: section, we discuss some sugyesrions of what this sense elemen: may be like. 2.4 Mental Representations Iroductien In tbe last section we concluded that although reference is an in ant function of language, the evidence Suggests thar there must be more to meaning than simply denotacion. We adopted Lhe convention of calling (his exta dimension sense. ' In the rest of Uxis chapter we explore the view that sense places a new level between words and the world: a level of mental representacion? 'lhus, u noun is said to gain ¡ts ability to denote hecanse li is associated with something in the speaker/hearer's mind. "his gets us out of the problem of insisting everything we talk abour exists is reality, but it raises the question of what these mental representations ure. One simple and old idea is Uat these mental entities are images. Presume «bly the relarionship between the mentel representacion (tae image) and the real world entity would then be one of resemblance; see Kempson (1977) Meaning ough* and Rea iy for disenssion. This mighr conccivably work for expressions like París or mother; it might also work for imaginary entities like Batman. This theory however runs into serious problems with common nouns. This is because of the variation in Tmages thar different speakers mi ght have of a common noun like car or house depending on tncir experience. One example often cited in the literamure is of the word triangle: vue speaker may have a mental image of an equilateral triangle, another might be isosceles or scalene. Ir is difficult to conccive of an image which would combine the features shared by all triangles, just as it is difficult to have an image which corresponds to all cars or dogs. Uhis is ro ignore the diflicultics of what kind of image on< mighi have for words Eke anal or food; or wozse lore, jusrice or demo. cracy. So even if images are associated with some words, they cannot be the whole story. The most usual modification of the image theory is Lo hypothesize that the sense of some words, while mental, is mot visual bur a more abstracr elemen: oncept. This has the advantage that we can eccepr that a con- Cept might be able ro contain ¡hc non-visual features which make a dog a dog, democracy democracy, etc. We might also feel conáden: about coming Up wirh a propositiona! definition of a triange, something corresponding to “three-sided polygon, classiñable by its angles or sides”. Another advantege for linguists is hat they might be able to pass on some of the labour of Gescribing concepts to psychologists rather than have to do it all themselves. Some concepts might be simple and relared to perceprial stimuli — Jike SUN,” WATER, €tc. Others will be complex COLCEPIS like MARRIAGE Or RETIRE= MENT which involve whole theories or culrural comuplexes. This scoms reasonable !cal members than on less typical members; they come to mind more quickly, > Another result of this and similar work (6.2. Laboy 1973) is that rhe boundaries berween concepts can seem Lo Speakers uncertain, or Suzzy, rarher tran clearly defined. This approaca allows for borderline uncertalaty: an ¡om in the world might bear some resemblance ro two different protolypes. Here we might tecall our Lypothetical example in chapter 1 of an “glisk spenker being ubíe to use rhe word ale yet being unsure about whether a whale isa mnammal or Gsh. ln the prototype theory of con epts, this might be ex Platned by the fact that whales are nor typical of tie e Tegury MAMMAL being far from the central Prototype. At the seme time, whales resembie brototypical Ash in some characreristio features: they live Oceans, have fins, etc. : There arc a number uf uterpretations of these tupica ity effects in the psychology literamnre: some researchers for example have argued that the central prototype is an abstracción. This abstraction might be a ser of char= acteristic features, to which We Compare real items; see Smira and Medin (1981) for discussion. These characteristi features OÍ BIRD anighr describe a kind of avecage bird, simall, perhaps. wish wings, fcarhezs, the ability to Ey, etc. but of no particular specios. Otlier researchers hayo proposed that we organizo our categorios Ly exemplars, memories of actual typical birds, sa sparrows, pigeons und hawss, and we compute the li slihood of something mect being a bird on the basis comparison w:rh these memories of 1eal birds. An overview of this area of investigación is gíven hy Medin and Ross (1992) There is another approach to typiculicy effects from witkia Enguistics, hich is interesting because of the light ir sheds on the relario. nship between liuguistic knowledge and encyclopaedic knowledge, a topic we discussed in nderwater in the APEEPOPOEOeeettttrtrr2. riraries 58 Fr chapter 1. Charles Fillmore (1982b) and Goorge Lakoff (1987) both make Similar claims thar speakers havo folk theories about the world, based on ¿lucir experience and rooted in their culture, These theories are called frames by Filmore and idealized cognitive models (ICMs) by Lako(! "They are nor scientife thsories or logically consisient definitions, bur collecrio cultural views. Fillmore gives an example of how these folk theorics mighr work by using the word hachelor. Tr is clear that some bachelors are more prototypical tan others, wirh thc Pope, for example, being fur from proto- iypical. Iillmore, and Lakoff in his discussion of the same point (198: 65- 71), suggests that there 15 2 division of our knowledge about the word bachelor. partis a dictionary-type definition (perhaps simply “an unmarried man”) and parris an eneyclopaedia-Lype emuy of culcoral «nowledge about bachelorhood and marriage — the frame or ICM, The first we can call inguistic or semantic knowledge and the second real world or general k:ow ledge. Their point is we only apply the word dachelor within a typical mar- riage ICM: a monogumous union berween eligible people, typicallv involving romantic Jove, etc. Ir is this idealized model, a form of general kuowledge, which governs our use of the word bachelor and restrains us from applying it to celivate priests, or people living in isolarion like Robinson Crusoe on his island or Tarzan living among apes in the jungle. La this view then using a word involves combining semantic Ínowledge and encyclopeedio know- ledge, and this interaction may result in typicality effects Prototype theory. frames and ICMs are just a few of ¿he large number of proposals for conceptual structure. We will look at some suggestions from the speciically linguistics literature in later chapters. o 24.5 Relalions between concep:s One important issue thar our discussion has Dypassed so far is the rela- tional nature of conceptual knowledge, We will sce in chapter 3 that words are in a nerwork of semantic links with other words and it is reasonable Lo assumo that conceptual struerares are similariy linked, Thus if all gou know about peccary is Thar ir is a kind of wild pig end of pecorino that it is a kind of lralian cheese, then your kaowledge of these concepts “iaherits” know- ledoe you have about pigs and cheese. This £as implications for our earlier ¡much knowledge a speaker has Lo have in order 10 use 2 discussion of ho word. It suggests that the crucial element is not the amount of knowledge but its integration into existing knowledge. Thus, knowing that a peccary is a kind of pig, rogerher with whai you koow about pies, is perhaps enough to begin to understand the meaning of sentences containing the word, and thereby Lo start to gain extra kuowledge abour rhe concept. Such relations between concepts have becn used to motivate models of conceptual hicrarchics in lhe cognitive psychology literature. A model bascd on defining aturibures was proposed by Collins and Quillian (1969). Tn this model, concepts are represented by nodes in a network, to which Meanine, Thought and Re 39 Tigure 2.1 Conceptual nenvorks y Has skin É Caimanes — Das S Breaties ES SA OL Can sima mise PE Has gulls / So smanx — Can bite , > Ta dangerous Ban 7 E pins saLmon “a Is ediblo :g thin legs Swiras upstrecn OSTRICIL Ís call > Candy Colíins and Quillian (1969, artributes can be attached and between which there are links, One such link is inclusion so thar a subordinate node inherits nttributes from a super- ordinate node. An esample of such a neuwork is in gure 2.1. Here we can see “hat CANARY inherits (he attributes Of BIRD and ANIMAL and thus inherits the arrributes breathes, eats, has skin, ha ngs, can iy, has feathers. We can 3ee too that the Collins and Quillian model has the abiliry ro block inher- irance, so thal for example ostx1cH docs not inherit can fly from BIRD, If the attribures in this model are taken to be the equivalent of the necessary and sufficient conditions we discussed carlier (hen it suTers from the disadventages of that approach. Proponents of protorype theory, for example Rosch et al. (1976), have also investigated conceptual hierarchics and have proposcd that such hierarchies contain three levels of gencrality: a superordinate level, a basic level, and a sudordinate level. The idea is thar the levels diffcr in their balance between informativeness and uscfulness. s (1976) examples, ¿hal 07 furniture, the superordinare level is FURNITURE, which has relatively few characteristic fea tures; the basic level would include concepts like CHAIR, ich has more features, and the subordinate level would include concepts like ARMCHAIR, DINING Crrarr, ecc. which have still more features and are thus more specific again. The basic level is identified as cognitively important: it is the level that ds most used in everyday life; it is acquired first by children; in experi. meets ir is the level al whick adults spontansously name objects; such objects are recognized more quickly in tests, und so on. “This model has proved to be very robust in the psychological literature, though the simple picrure we have presented here needs some modificatións. Ir seems tral Uhe relationship berween the basic level and the intermediate term might vary somewhat [rot domaia to domain: man=mede categon CLEAR AA ADA A A A Pel mineras ES 5) le seems faisiy evident Uat the seleccion ol such simple terms st ro a certain exient depend upon tho chief interests of a peo- e o e ln it is necessary to distinguish a certain phenom- e many aspcuts, which in the life of the people play each a eno%-ly independent role, 1uany independent words may develop. mer cases moditicarions of a single Lerim may suffico. aguage, from the point of view of classificalions; that whal charac- while in Thus ie happens that sach laz se another language, may be arbitrary in ls A appears us a singie simple idca in one language May ee Smued by a series of distinct phonetic groups in another. (Boas teriz: 1966: 22) wed that the effect of this was largely unconscious because the E 5 which we do not normally use of language is largely an automaric proces pause to reflect on . Dies observations open the debate in this literamre about the relation ship between language, culture and thougat. To whal extent does the pe ticular language we speak determine the way that we think about the ? Perhaps Boas's mos" famous student is the anthropologzist and. asus. S de Sapirs in the following quoration, we see him proposing the: vie anne particular language we speak conditions vur conceptuslizatios ofthe word 2.44 Lansuage is a guide to “social reality”... Human beings do not live in the objective world alone, nor alone in vie world of social activity as ordinarily understood, out are very much at the mercy of the particular language which has become the medium of ex- pression for 1heir society... . the “real world” is ro a lumge exton unconsciously built up on the language habits of the group. No two Janguages aro ever sufficientiy similar to be considered as tep- resenting (he same social realivy. The worids in which different societies live arc distinct worlds. not merely the same world with different labels attached We see and hear und otherwise experience very larg 1 because the language habits of vur community predispose certain choices of interpretation . . . rom this standpoint we may Link of language as the sywmbolc guide to culture. (Sapir 1949b: 102) as we do Tr seems fair to say that Supir had a stronger view of the derermining role of language than Boas. Srronger still are he views of Benjamin Lee Whor, « linguist well-xnown for his work on native American languages, especially the Uto-Aztecan languages of the south west United Slutes and Mexico. Wihorf strengthened this idea of the link between languago and hought into the notion he called linguistic relulivity. 1rs basic premise is that the way we think about the world is determined by our cultural and linguistic background: Meaning, Tha: 2.45 We ent nature up, organize it into concepts, and ascribe significances as we do, largely because we are parties to an agreement to organ- ize it in ¿his way — an agreement thar holds through our speech community and is codiñed in (he patterns of our language. The agreement is, of course, an implicit and unstated one, BUT ITS TERMS ARE ABSOLUTELY OBLIGATORY; we cannot telk at all except by subscribing to the organization and classificarion of data which the agrecment decrees. (Whorf 1956: 213-14) WhorP's observation is not restricted to word meaning; indced, he believed that meanings derived from grammatical systems (e.g. notions of number and space in nouns. or aspect and tense in veros)'* were even stronger deter- minants of thought. he idea is that speakers can reflecz on word mear- ings but gramatical sysrems are largely unavailable to conscious reflection. Tf this view is correct then our own language predisposes us to see both reality and other languages through its own filter. This would have serious implications for the prospects of a universal semantic theory. L: might mean tal we could always, wirh some difliculty and inexactitud, translate from one language to another. But if speaking different languages means that we think in different ways, how could we ever step outside our own language to set up a neutral metslanguage which does nor privilege any particular other areas of lingnistics like synrax or phonology. anguage of thought hypotnesis The idea of linguistic relativity is rejected by many linguists and researchers in cognitive science, the interdisciplinary study of intelligence which drews on cognitive psychology, computer science and linguistics. A typical re- sponse is to dismiss as a fallacy such a strict identiócation of tiought and language. We can identify wo main types of argument used to support this view. The first is that there is evidence of thivking withou: language; «nd ¿he second is that linguistic analysis has shown us thar language underspecifies meaning. We can look briefiy at these two types of argument. Á succinct presentation of the first typc of argument is given by Pinker (1994), who presents various kinds of evidence that thinking and language gre not (ie same thing. He gives examples of evidence of thought processes, such as remembering and reasoning, which have been identified in psvehological studies of human babies and of primares, both providina examples of crea- tures without language. He also recounts the various reports oí artists and scientists who claim rhar their creariviry somerimes derives from ideas which are non-linguistic images. There is also evidence from psychological ments of speri- visual thinking: subjects seem able to manipulate images mentally, rotating thera. scanning them, zoomine in and out, etc., exbibiting a variety of mental processes which de na seem to involve language. lónally Pinzer 44 Preliminaries casts doubt on the varivus actempls in psychological experiments to suggest that people from different linguistic communities perform reasonine or other cognitive tasks in any very different ways.' Such evidence for mental processes not involving language is often used lo urgue that cognitive processes de nor employ a spoken language like Unglish or Arabic but make use of a separate compurational system in the mind: a language of thought. For a philosophical delense of this position ses for example J. A. Fodor (1975). Stillings er al. (1995) provide a range of evidence from psychological experiments to support the same view, The basic idea is that memory and processes such as reasoning seem to make use of a kind of propositional representation that does not have the surface syntax of a spoken language like English. Turning to the second type of argument — thar language underspecifies meaning — some indirect support for this position emerges from the char- acteristic view oí the communication of meaning that hes emerged from research in semantics and pragmatics, as we shall sec in the course of ¿his bcox. IL has become clear that meaning is richer than language er borh ends, so to speak, of the communication process. Speakers compress (cir thoughts, and often imply rather han state explicity what dey mean, while Acarers fill out their owa version of the intended meaning from the language presented Lo thom. This idea, thal language underspeciles imeanino and has to be enriched by heurers, would seem to fil narnrally with the idea thar speakers are purting their thoughts into language, ¡.e. translarine into the spoken language, catlies thax simply voicing Uncir thoughts directly. This does not of course provide direct evidence for this view: we could equally imag- inc English speakers hinking in English and still compressing their thoughts when speaking, on some grounds of economy aud sucial cooperation, Nonetheless these different rypes of argument are often taken, especially in cognitive science, to support the view that we ¿hink in a language of thought, sometimes called mentalese, When we want to speak, we translate from mentalese into our spoken language, be it Mohawk or Russian One natural extension of this view is The proposal thar everybody's mentalese ja roughly the same, that is that the language of thought is universal. Thus arrive at a position diamerrically opposed to linguistic relativiry: human beings have cssentially the seme coenitivo architecture and mental pro- cesses, even though ¿hey speak different languages. 2 25.3 |nougat and reality If we leave this question of die relarion between words and thinking for the time being, we night ask whether sernanticisrs must also consider questios: of the relationship between (houghr and reality. We can ask: must we as aspiring semanticists adopr for ourselves a position on traditional questions of ontology, the branch of philosophy that deals with rhe nature of being and the structure of realiry, and epistemology, the branch of philosop. ught anc Acality 18 concerned with the nature of knowledge? For example, do we believe rhar reality exists independently of the workings of human minds? If not, we are «udherents of idealism. Tf we do believe in an independezt reality, can we peresive the world as it really is? One response is to say yes. We might assert Uat knowledge of rcaliry is attainable and comes from correcrly conceptu- alizing and categorizing the world. We could call this position objectivism. On the other hand we might believe that we can never perceive the world as it reelly is: that reality is only graspable through the conceptual filters derived from ouz biological and cultural evolution. W: could explain the fact that we successfully interact with reality (run away from Lions, shrink from. ire, etc.) because ofa notion of ecological viabiliry. Crudely: that those with very inclficien: conceptual systems (nor efraid of lions or fire) died out and weren't our ancesrors. We could call this position mental construetivism: we can't get to a God's eye view of reality becanse of the way we are made. These axe of course very crude characterizations of difficult philosophical issues. By now any philosophers chancing on this text will have thrown into the back of their own fire. But the relevance of these issues to semantios is that, as we shall see in later chapters, different Theories of semantes often presuppose different answers to these very basic questions Srill. for the linguist keen to describe thc semantics of Swahili or English these are a heavy set of issues Lo deal wirh before gerring on with the job, eciallv when added to the complex issues of conceptual repz that we discussed a little earlier. One understandable 'esponse is to decide that only language is the propes object of stady for linguists and issues of inental representatións and Ue existence of reality are best lelt to psycholo s and plilosophers. See for example the following com: Hockerr: entalion nt by Charles 2.46 We can leave to philosophers the argument whether the abstract relationships themselves have any sort of existence in he world side of spesch. Whatever they may decide, il is clear that the “meaning of a word like and or the... is a very different thing trom the meaning of a word like morktmg or sumbeam. (Hockett 1958: 203) and we can sec a similar sentiment in John Lyons's (1968) discussion of semantics: 2.47 tae view that semantics is, or ought to be, an empirical science, which as tar as possible avoids comunitment with respect ro such philosophical and psychological disputes as (he distinction of “body” and “mind” and the starus of “concepts”. This view will be accepted in the discussion of semantics given in this chapter. lt should be stressed, however, that the methodological renunciation of “men- lienv docs not imply the acceptance of mechanisn”, as some linguísts have suggested . ... The position that should be maintained 
