Semantic primitives

Considered as a cognitive definition, the definition of balance as ‘keep in equilibrium’ poses an obvious problem: if someone does not know the meaning of balance, they are unlikely to know the meaning of equilibrium. And the most obvious way to defi ne equilibrium would seem, in fact, to be by way of the term balance: to keep something in equilibrium is, quite sim ply, to balance it. Defining keep in equilibrium by balance, and balance by keep in equilibrium is a simple example of definitional circularity, which was introduced in Chapter 1. As discussed there, it is impossible to give a definition of every word in a language using other words of the same language: at some point the chain of definition must come to an end. Since the vocabulary of any language is limited, the metalanguage will eventually have to include object language definienda, thereby leaving some of the latter without independent definition. This is a problem for any attempt, such as that made in linguistic semantic theories, to specify the meaning of every lexeme in a language. Leibniz likened this problem to the situation of someone who is promised money and continually strung along from one alleged payer to the next, without ever actually receiving anything:

 I give you a hundred crowns, to be received from Titus; Titus will send you to Caius, Caius to Maevius; but if you are perpetually sent on in this way you will never be said to have received anything. (Parkinson (ed.) 1973: 1–2)

 In the same way, a metalanguage which incorporates elements of the object language can also be said to ‘defer full payment’. Only a metalanguage which is completely independent of the object language is in a position to offer a complete, non-circular explanation in which every definiendum receives its own semantic analysis independently of the analysis of the others. Without such a metalanguage, there will always be a residue of unexplained terms which escape definition.

In Chapter 1 we saw some proposals about how to escape from this chain of definitional circularity by grounding the study of meaning in various extra-linguistic realities. One of these was the conceptual theory of meaning, which identified meanings with concepts. Given such an identification, we can imagine several different possibilities for the relation between lexemes and concepts. One is that each lexeme corresponds to an entirely different concept. Thus, the English lexemes cup and mug would each correspond to the separate concepts CUP and MUG, each of which is unitary and undecomposable. The fact that cup and mug seem to share certain properties – they both refer to drinking vessels usually reserved for hot liquids – is not reflected on the conceptual level by any shared conceptual content. The two concepts are, that is, semantic primitives: in spite of appearances, they cannot be completely broken down into anything conceptually simpler. Something approaching this view has been advocated by Fodor, who argues for a lexicon where each lexical item is a semantic primitive or atom with no internal definitional structure:

 . . . I take semantic facts with full ontological seriousness, and I can’t think of a better way to say what ‘keep’ means than to say that it means keep. If, as I suppose, the concept KEEP is an atom, it’s hardly surprising that there’s no better way to say what ‘keep’ means than to say it means keep. I know of no reason, empirical or a priori, to suppose that the expressive power of English can be captured in a language whose stock of morphologically primitive expressions is interestingly smaller than the lexicon of English. (Fodor 1998: 55)

 Typically, however, proponents of the conceptual theory of meaning in linguistics believe that most word meanings are not themselves primitive, but are composites of a finite stock of primitive concepts. These semantic primitives are the basic building blocks of meaning out of which all other meanings can be constructed (see Chapter 8).

The belief that responsible semantic analysis must be grounded in a level of elementary, primitive units is implicitly or explicitly held by many semanticists (Fillmore 1971, Jackendoff 1983, Allan 2001: 281; for some criticisms of primitives see Aitchison 1994). The most thorough-going example of a theory of semantic primitives in modern linguistics is the Natural Semantic Metalanguage (NSM) theory of Wierzbicka and Goddard. Painstaking cross-linguistic research in this framework has led to the development of the following list of semantic primitives which the NSM approach uses for the definition of meaning:

 I, you, someone, people, something/thing, body; this, the same, other; one, two, some, all, much/many; good, bad; big, small; think, know, want, feel, see, hear; say, words, true; do, happen, move; there is, have; live, die; when/time, now, before, after a long time, a short time, for some time; where/place, here, above, below, far, near, side, inside; not, maybe, can, because, if; very, more; kind of, part of; like. (Goddard 2002: 14)

These 58 elements represent the ‘atoms of meaning’ which are claimed to be impossible to defi ne in a non-circular manner, and which can be used to fashion definitions for a large range of words.

 A definition in NSM is a reductive paraphrase of the definiendum’s meaning. Reductive, in that it reduces this meaning into a set of primitive components, and a paraphrase in that it consists of a textual explanation of the meaning, which is not simply a list of synonyms or a GD definition (although it may contain this structure) but a collection of natural language sentences in which the meaning is expressed. Among the many definitions proposed in NSM are those of sun and watch:

(Note that sky is not a semantic primitive, but a ‘semantic molecule’ which can be independently defined with the 58 primitives, perhaps as ‘a place above all other places’, and which is used here as shorthand for pure space-saving reasons.)

 QUESTION Are these definitions accurate? What is their structure?

But while the semantic primitives can be used to defi ne other members of the vocabulary of a language, the primitives themselves are impossible to defi ne in terms of anything simpler.

Semantic analysis in NSM thus consists in explaining a definiendum in simpler and more comprehensible terms than the definiendum itself. The 58 semantic primitives are supposed to represent the simplest possible explanatory terms, which cannot themselves be explained by anything simpler. It is not, for NSM, simply a matter of chance that it is these particular terms, and not some others, that represent the simplest metalinguistic definientia. NSM theory claims that the indefinable nature of its primitives derives from their status as conceptual primitives: the primitives are hypothesized, in other words, to express the set of ‘fundamental human concepts’ (Wierzbicka 1996: 13), considered to be both innate and universal. What this means is that every natural language possesses an identical semantic core of primitive concepts from which all the other lexicalized concepts of the language can be built up. Since this common core is absolutely universal, it can be stated in any language, and NSM scholars have devoted considerable energy to testing the list of primitives reproduced above in order to confirm that every language does, indeed, have an ‘exponent’ of each suggested primitive.

This commitment to the universality of the metalanguage has involved NSM theory in more cross-linguistic investigation than is common in many semantic theories, which still largely focus on familiar Indo European languages as the empirical base for the assessment of theoretical semantic adequacy. This rigorous cross-linguistic orientation is one of the main advantages of the NSM approach to meaning. Not least among its benefits is NSM’s utility as a practical tool. Because only defining terms have been chosen for which well-established cross-linguistic equivalents exist, the NSM set of primitives provides a convenient tool for cross linguistic explanation, and NSM scholars have been able to defi ne a surprising range of vocabulary, including words from domains in which strikingly delicate and culture-dependent information is contained. An example would be Wierzbicka’s preliminary definition of the Japanese emotion term amae:

The reader of this definition is arguably in a better position to appreciate the intricacies of the concept amae than is the reader of any of the numerous other English explanations of this term which have been proposed in the literature, and which include ‘lean on a person’s goodwill’, ‘depend on another’s affection’, ‘act lovingly towards (as a much fondled child towards its parents)’, ‘to presume upon’, ‘to take advantage of’, ‘be coquettish’, ‘coax’, ‘take advantage of’, and ‘play baby’! As a cognitive definition, in other words, the NSM paraphrase may be an attractive way of conveying the meaning of definienda.

 In spite of the fact that some linguists find NSM a useful tool for the analysis and representation of certain aspects of meaning, the method’s theoretical soundness, as opposed to its practical utility, has been the subject of considerable criticism (Riemer 2006). As for other definitional theories of semantics, many of these critiques bear on the alleged inadequacy of NSM’s proposed definitions: we will look at this type of criticism more generally in 2.6.

 Aside from these more general problems, we may distinguish two particular challenges faced by NSM, each of which tells us a lot about the problems of cross-linguistic definition. The first concerns the universality of the NSM primitives. Some languages simply lack some of the allegedly universal primitive terms. As discussed by Bohnemeyer (1998, 2002, 2003), Yukatek Maya (Mayan, Mexico) lacks lexical exponents of the primitives ‘after’ and ‘before’: the expression of these temporal relations is achieved primarily through particular combinations of aspectual and/or modal information (Bohnemeyer 2003: 216). Another particularly common case is where languages merge putative primitives, as when Japanese ba translates both the primitives ‘if’ and ‘when’ (Goddard 1998: 138), or when Pitjantjatjara (Pama-Nyungan, Australia) kulini has the meanings ‘think’ and ‘hear’ (Goddard 1996: 44). If each primitive represents a genuinely distinct semantic atom, these mergers should not be possible.

Even if a language does have unproblematic equivalents for all the NSM primes, it is not necessarily the case that the types of paraphrase that can be written in English using these primes will also be possible in other languages. Diller (1994), for example, discusses some of the problems raised by the attempt to write NSM definitions using a set of possible Thai (Tai, Thailand) exponents of the primitives. Thai typically omits overt pro nominal reference forms – the equivalents of the primitives ‘I’ and ‘you’ – unless speakers want to achieve specific communicative goals. When explicit equivalents of the pronouns are present, Thai offers a number of possible translations of ‘I’ and ‘You’, which differ in the sociolinguistic roles (gender, politeness, formality, deference, urbanity) which they impute to speaker and addressee, as the following list shows:

To illustrate the effect of these differing dynamics, consider a possible Thai translation of the sentence ‘I want to go with you’, which only uses primitives:

From a sociolinguistic point of view, English ‘I want to go with you’ is relatively neutral, able to be uttered by a large range of speakers of differing genders, habitual politeness relations and social statuses. The Thai equivalent in (27), however, is much more restricted: it is the type of sentence which a mother might say to her daughter, but which a daughter would never say to her mother. None of the possible translations into Thai of ‘I’ and ‘You’, in fact, is sociolinguistically neutral, but carries with it suggestions about the identities of the speakers and the relationships between them. As a result, a semantic explication which uses any of them will have nothing of the neutral force of the English NSM paraphrases quoted above, but will carry with it all sorts of subtle indications about sociolinguistic register – indications which could be misleading if they are meant to embody entirely abstract, general statements about the meanings of words throughout an entire language community.

An NSM supporter might reply that this is just a fact about the way these forms are used, and not a fact about what they mean: we will discuss this distinction in the next two chapters. But even if we decide that this type of issue is not relevant to the analysis of the meaning of the forms, the bias introduced by the particular meanings of the Thai pronouns looks like a significant obstacle to the efficacy of NSM as a real tool of semantic analysis. NSM was designed not primarily as a method of extensional definition, but as a tool of cognitive definition. It is intended to make hard to under stand concepts in one language comprehensible by translating them into primitive, allegedly universally understood terms. To the extent that the sociolinguistic values of the Thai forms confuse this task, NSM’s claim to furnish cognitive definitions must be considered as compromised.

The second problem with NSM is the fact that many – perhaps all – of the English exponents of the NSM primes have several senses, with only one of these senses being identified as universal. (This is not just a problem with the English exponents of the primitives; the same problem would exist regardless of the language in which the primitives were expressed.) For example, in testing for the presence of an exponent of a primitive meaning in some language, it is not enough simply to ask whether the language in question has words for ‘I, you, someone, etc.’; instead, we have to distinguish the sense claimed as universal from the others: is the primitive TRUE, for instance, better represented by the meaning present in (28) or (29)?

The most obvious way to distinguish the intended sense would be simply to defi ne it verbally. But since, ex hypothesi, the semantic primitives are indefinable, this option is unavailable. As a result, the project of testing the primitives cross-linguistically is seriously compromised, since you can never be sure that a claimed exponent of a primitive in a given language does in fact correspond to the required primitive meaning. (See Goddard (2002) for some suggested solutions to this problem.)

These problems disappear if NSM’s claim to universality is abandoned, since only universal theories of semantic primitives are open to the sorts of criticism just advanced. If your theory only commits you to the existence of indefinable primitives within a single language, problems about cross linguistic equivalence do not arise. This is the approach of, for example, Apresjan (2000: 228), for whom ‘[s]emantic primitiveness is determined . . . by the structure of the lexicon of the language being described: lexeme L is considered a primitive if the given language has no L1 , L2 . . . Ln via which it might be explicated’. In the next section, however, we will see that even this type of practice is threatened by quite general problems with the notion of definition.

اخر الاخبار

اخبار العتبة العباسية المقدسة

واحات الحزام الأخضر الجنوبي ترفد الأسواق المحلية بنحو 35 طنًّا من المحاصيل الزراعية شهريًّا

وفد تدريسي من جامعة واسط يطلع على مقتنيات متحف الكفيل

مؤسّسة الوافي تُهدي مكتبةَ مسجد الكوفة نسخةً من موسوعة فتوى الدفاع الكفائي

المجمع العلمي يقيم برنامج (سراج المنتظرين) في محافظة المثنى

المزيد

الآخبار الصحية

دواء جديد يحمي القلب ويعزز علاج السرطان في آن واحد

سر غذائي بسيط لحماية القلب مع التقدم في العمر

علامات تشير إلى اختلال عمل الغدة الدرقية

السمنة واللقاحات.. دراسة تكشف تأثيراً سلبياً

المزيد

الآخبار التكنلوجية

البعوض في آيسلندا إشارة مبكرة على تحولات بيئية في القطب الشمالي

صاروخ يطلق قمرا اصطناعيا في مدار خاطئ

"أوبن إيه آي" تطلق نموذجا للعلوم الحيوية واكتشاف الأدوية

تفوق على البشر.. روبوت يحطم الرقم القياسي لسباق نصف ماراثون !

المزيد

مواضيع ذات صلة

Definition by context or typical exemplar

Definition by synonymy

Definition by ostension

Real and nominal definition

Contextual modulation of meaning

Mismatches between grammatical and phonological words

Words and morphemes

The units of meaning

Word-based and meaning-based approaches to definition

The history of the dictionary

Meaning and use

Meanings as brain states