A sign is not only something which stands for something else; it is also something that can and must be interpreted. The criterion of inter-pretability allows us to start from a given sign to cover, step by step, the whole universe of semiosis.
This criterion (as we have shown in the first chapter of this book) held for the classical notion of natural signs, based on an inferential model (p q), but it should hold also for linguistic signs, even though they were based, by a long historical tradition, on the model of equivalence (p q). This latter idea of sign as identity was due to the persuasion that the meaning or the content of a given linguistic expression was either a syn-onymous expression or its definition. Irrespective of whether the defini-tion is provided by genus and differentia (/man/ « rational mortal animal») or by a series of semantic components or markers (/man/ «human + male + adult»), there should be a biconditional link between definiens and definiendum.
One can say that not even this model excludes an interpretive process: undoubtedly «rational mortal animal» says some-thing more than the verbal utterance /man/, especially if ‘mortal’, ‘ra-tional’, and ‘animal’ are interpreted in their turn. However, every deci-sion about this matter concerns the choice between a dictionary and an encyclopedia, a crucial question widely discussed in the course of the last decades (see, for instance, Wilson 1967; Katz 1972; Putnam 1975; Rey-Debove 1971; and, for the most complete and convincing overview of the problem, Haiman 1980). Naturally, one must distinguish between the opposition dictionary/encyclopedia as it is intended in the publishing world and the same opposition as conceived in semiotic or philosophical terms. It suffices to read some of the analyses of the current dictionaries and encyclopedias (see, for instance, Weinreich 1980; Rey-Debove 1971) to realize that, if so-called encyclopedias are in some way encyclopedic, so-called dictionaries are rather impoverished encyclopedias.
If one consults the 1974 edition of the Merriam-Webster dictionary, one finds that /bull/ is defined as an «adult male bovine animal» (a definition that would titillate every fan of a semiotic dictionary); but /tiger/ is defined as «a large tawny black striped Asiatic flesh-eating mammal related to cat», and this definition is as if it were conceived to support Putnam’s idea of ‘stereotypes’.
2.1.2. The idea of a dictionary
The first semiotician to outline the idea of a dictionary was probably Hjelmslev. After having analyzed expressions into minor elements or figurae so that «unrestricted inventories are resolved into restricted» (Ι943·7ΐ)> he tried to do the same for the content-plane. If the analysis of the expression-plane consists «in the resolution of entities that enter unrestricted inventories (e.g., word-expressions) into entities that enter restricted inventories, and this resolution is carried on until only the most restricted inventories remain» (ibid.), the same procedure must be followed for the content plane:
While the inventory of word-contents is unrestricted, in a language of famil-iar structures even the minimal signs will be distributed (on the basis of relational differences) into some (selected) inventories, which are unre-stricted (e.g., inventories of root-contents), and other (selecting) inventories, which are restricted (e.g., inventories embracing contents of derivational and inflexional elements, i.e., derivatives and morphemes). Thus in practice the procedure consists in trying to analyze the entities that enter the unre-stricted inventories purely into entities that enter the restricted inventories. (Ibid.)
The example given by Hjelmslev concerns a series of word-contents corresponding to the common nouns ram, ewe, man, woman, boy, girl, stallion, mare, sheep, human being, child, and horse and to the pronouns he and she. Hjelmslev reduces the inventory in a way that can be repre-sented by the diagram in Figure 2.1. He thought that the word-contents of the expressions horse, sheep, human being, and child belonged to unre-stricted inventories, while she and he belonged to a category «with a re-stricted number of members.»
| SHEEP HUMAN CHILD HORSE
|SHE | ewe woman girl mare
|
HE | ram man boy stallion
FIGURE2.1
Hjelmslev’s proposal seems to account for some linguistic phenomena that, according to the further semantic literature, should be explained by a dictionary. If a dictionary concerns purely our linguistic knowledge without giving instructions as to how to use linguistic terms in order to mention things or states of the world, Hjelmslev’s dictionary undoubt-edly explains why such sentences as a ewe is a female sheep and ifx is a ewe then x is not a stallion are semantically well formed, even though the user of that language never had a direct acquaintance with a sheep or with a stallion. In fact, other dictionary theories, in order to provide instructions for reference, provide another kind of information; see, for instance, the idea of ‘distinguishes’ in Katz and Fodor (1963) and the reformulation of this principle in Katz’s ‘neoclassical theory of reference’ (1979).
Hjelmslev’s dictionary seems to explain (as usually requested to a dic-tionary) at least the following phenomena:
(a) synonymy and paraphrase (a ewe is a female sheep);
(b) similarity and difference (ram/stallion, ewe/mare, and stallion /mare have a common semantic component, but in a different respect a stallion is different from a mare, and a mare is different from a ewe);
(c) antonymy (girl/boy);
(d) hyperonymy and hyponymy (horse/stallion);
(e) meaningfulness and semantic anomaly (stallions are male makes sense, whereas a female stallion is anomalous); (0 semantic ambiguity (a more. complete dictionary should explain the possible ambiguity between ram as male sheep and ram as a warship);
(g) redundancy (unfortunately in such a limited dictionary redundancy coincides with meaningfulness; a male ram is both meaningful and re-dundant);
(h) analytic truth (for the same reason as above rams are male is at the same time meaningful, redundant, and analytically true, since the meaning of the subject contains the meaning of the predicate);
(i) con-tradictoriness (rams are female);
(j) syntheticity (the dictionary recognizes expressions such as sheep provide wool as depending on one’s world knowledge);
(k) inconsistency (this is a ram and this is a ewe cannot be true at the same time if referred to the same individual);
(1) containment and semantic entailment: these two phenomena are strictly dependent on each other since, by virtue of the dictionary, every term is supposed to ‘contain’ certain properties; on the basis of these semantic rules (and independently of any other logical law) the sentence this is a ram entails this is a sheep, the sentence this is not a sheep entails this is not a ram, and the sentence this is not a ram leaves unprejudiced whether this is a sheep or not.
I have severely limited my list of requirements for a dictionary (other authors introduce more controversial requirements; see, for instance, Katz 1972, 5—6). In any case, Hjelmslev’s proposal for a dictionary leaves unsolved two important questions: how to define the meaning of the components or figurae (in other words, if ram means male sheep, what does sheep mean?) and how to obtain a finite or unrestricted inven-tory.
Let us first consider the second problem. The most rigorous support-ers of a theoretical dictionary maintain that the meaning of linguistic expressions should be represented through a finite number of semantic primitives (components, markers, properties, universal concepts).
It is not strictly necessary to assume that the set of definienda be a finite one. Naturally, the ideal condition for a dictionary is that this dic-tionary, being «the reconstruction of an aspect of the speaker’s semantic competence,» storing «only finitely many bits of information about a particular lexical item,» be «a finite list of entries» so that «each entry consists of a finite number of lexical readings, and that each lexical read-ing contains a finite number of semantic markers» (Katz 1972:59—60). However, it is theoretically possible to conceive of a consistent number of primitives whose combination permits the description of an open number of definienda.
At this point the problems are (a) how to determine the primitives and (b) how to guarantee that their number be a finite one. As far as the determination of the primitives is concerned, the discussion is still open. Haiman (1980) remarks that (according to the current philosophical and semiotic literature) these primitives can be identified only in three ways:
(a) Primitives are ‘simple’ (or the ‘simplest’) concepts. It is, however, very hard to decide what a ‘simple’ concept is. In terms of the speaker’s intuition, ‘human’ is simpler than ‘mammal’ (since every speaker is able to tell whether something or somebody is a human being or not, whereas we have problems in telling whether a dolphin is more a ‘mammal’ than a platypus). It must be clear that in this case ‘simpler’ or ‘simplest’ does not mean more general; therefore, ‘simplest’ concepts risk being more numerous than the ‘complicated’ ones. It is true that it is not necessary that the primitives be less than the definienda: «a phonological features analysis would not be invalidated by the discovery that there are more features than phonemes in some language» (Dean Fodor 1977:147). However, this remark can hold in phonemics, since a language has a finite number of phonemes. But if a language has a potentially open number of expressions to be defined, can one accept that this open series be defined by a series of primitives that is indefinitely open as well? Moreover, the quest for ‘simple’ concepts leads one to the second way, and the strictures holding for the latter also hold for the former.
(b) The primitives are rooted in our
