[189]
One important subject has not yet been touched upon. It is one which shows, precisely, how necessary it is to examine every grammatical question from the two points of view distinguished above. It concerns abstract entities in grammar. We shall take first of all the associative aspect.
To associate two forms is not only to feel that they have something in common, but also to distinguish the nature of the relations which govern these connexions. Speakers are aware that the relation between enseigner (‘to teach’) and enseignement (‘teaching’), or between juger (‘to judge’) and jugement (‘judgment’), is not the same as that between enseignement (‘teaching’) and jugement (‘judgment’) (cf. p. [173] ff.). This is where the system of associations links up with grammar. One may say that the sum total of deliberate, systematic classifications set up by a grammarian studying a given linguistic state a-historically must coincide with the sum total of associations, conscious or unconscious, operative in speech. These are the associations which establish in the mind the various word families, flexional paradigms, formative elements (stems, suffixes, endings, etc. (cf. p. [253] ff.).
But does not association operate only upon material elements? Not at all. As noted previously, association may connect words linked by meaning only: e.g. enseignement (‘teaching’), apprentissage (‘apprenticeship’), éducation (‘education’), etc. The same must apply in grammar. The three Latin genitives dominī (‘of a master’), rēgis (‘of a king’) and rosārum (‘of roses’) have three different endings, -ī, -is, and -ārum, which afford no phonetic basis for association. None the less, these endings are linked by one’s awareness of their common value, which prescribes identical uses for them. That is sufficient to set up an association, in the absence of any material support. And that is how the notion of ‘genitive’, as such, takes its place in the language. A very similar process makes us aware of connexions between the Latin flexional endings -us, -ī, -ō etc. in a series like dominus, dominī, dominō, etc., and thus emerge our more general notions of case and case flexion. Associations of the same order, but of still wider scope, link all nouns, all adjectives, etc., and thus establish the notion of parts of speech.
[190]
All these things exist in the language, but as abstract entities. It is difficult to study them, because one can never be sure whether the awareness of speakers of the language always goes as far as the grammarians’ analyses. But the essential point is that abstract entities are based ultimately upon concrete entities. No grammatical abstraction is possible unless it has a foundation in the form of some series of material elements, and these are the elements one must always come back to finally.
Let us now look at it from a syntagmatic point of view. The value of a group is often connected with the order of its elements. In analysing a syntagma, the speaker does not stop at distinguishing the various parts: he also recognises their order of occurrence. The meanings of the French word désir-eux (‘desir-ous’) and of Latin signi-fer (‘standard-bearer’) depend on the relative positions of the smaller units. There is no word eux-désir, or fer-signum. A value may even have no concrete element of its own, unlike -eux, -fer, etc. It may result simply from the order of elements. In French, for example, je dois (‘I should’) and dois-je? (‘should I’?) have different meanings, which depend entirely on the word order. A language sometimes uses the order of elements to express an idea which another language will express by means of one or more concrete terms. English, in types of syntagma like gooseberry wine and gold watch, uses word order to express relations which French marks by prepositions (vin de groseilles ‘wine of gooseberries’; montre en or ‘watch in gold’). French, on the other hand, renders the notion of direct object simply by placing the noun after the transitive verb (je cueille une fleur ‘I pick a flower’); whereas Latin and other languages do this by using an accusative case form with special endings.
Word order is undeniably an abstract entity. But it is none the less true that it owes its existence to the concrete units involved, which are aligned in a single dimension. It would be a mistake to believe in the existence of an incorporeal syntax apart from these material units distributed in space. In English, the man I have seen illustrates a syntactic feature apparently represented by zero, whereas French marks it by que (‘that’), as in l’homme que j’ai vu (‘the man that I have seen’). But it is just this comparison with French syntax which produces the illusion that nothing can express something. In reality, the material units in the English example, aligned in a certain order, create this value themselves. Without a set of concrete terms, no discussion of any point of syntax is possible. Furthermore, the very fact that we do understand a complex of forms such as this English example demonstrates its adequacy to express the thought in question.
A material unit exists only in virtue of what it means, what function it has. This principle is particularly important to bear in mind where smaller units are concerned, because there is a temptation to think that they exist simply as material elements. We may be tempted to think, for instance, that the word aimer (‘to love’) owes its existence entirely to the sounds which go to make it up. On the other hand, as we have just seen, a meaning or function exists only through the support of some material form. This principle was introduced above in connexion with larger syntagmas and syntactic types, simply because in their case the temptation is particularly strong to imagine immaterial abstractions hovering above the actual words in the sentence. The two principles just mentioned are complementary, and conform to our earlier remarks about the delimitation of linguistic units (cf. p. [145]).
[192]