Can the linguistic procedures applied to modern languages be used in the analysis of Latin? Looking first at the smallest grammatical units, we can see that most Latin words are inflected and that the inflections are bound morphemes. The endings of the present indicative of first conjugation verbs – -o, -as, -at, -amus, -atis, -ant – show number and person. The endings of first declension nouns – -a,- a, -am, -ae, -ae, -a in the singular and, -ae, ae-, -as, -arum, -is, -is in the plural – show number and case. Bound morphemes all, but Latin differs from English in that the first element in most inflected words is also a bound morpheme. In English, the regular verb walk has four inflections: zero, -s, -ed and –ing. The last three are bound morphemes, but the first is not, walk being a free morpheme. This is not the case in Latin. The root of the Latin verb is ambul- and, since it cannot occur alone, is as much a bound morpheme as its inflections. It thus seems as if there may be a far higher incidence of bound morphemes in Latin than there is in English, but with that qualification, the concept of the morpheme as the smallest grammatical unit in Latin and of the word as the next smallest looks valid.
What of larger structures? An English sentence is on one of these seven patterns:
(1) Subject – Verb
(2) Subject – Verb – Adverbial
(3) Subject – Verb – Complement
(4) Subject – Verb – Object
(5) Subject – Verb – Object – Object
(6) Subject – Verb – Object – Complement
(7) Subject – Verb – Object – Adverbial
Can we find the same patterns in Latin?
Because Latin verbs incorporate the subject, there is no difficulty in matching (1). Any finite Latin verb used intranstively will do and the addition of any adverb will give us (2). A Latin copulative verb such as sum will give us (3), as in homo sum (‘I am a man’) and a non-copulative verb will give us (4), as in Catullus Lesbiam amat (‘Catullus loves Lesbia’). Latin, like English, has indirect as well as direct objects, so we readily find (5) in sentences having verbs that can be used with the accustive and dative cases, as in pecuniam filio meo dedi (‘I gave money to my son’). We can see (6) in Terence’s Humani nihil me alienum puto (‘I think nothing human (to be) unrelated to me’) where the subject and verb occur in puto, the object occurs in (humani) nihil and the complement in alienum (although I’m prepared to be challenged on this analysis). Add an adverb to Catullus Lesbiam amat and we have (7): Catullus Lesbiam ardenter amat.
So far, so good, but is Latin susceptible to Immediate Constituent Analysis? Let’s try it on this simple sentence from Seneca: immodica ira gignit insaniam (‘Excessive anger causes madness’), using the conventional abbreviations (S)entence, (N)oun (P)hrase, (V)erb (P)hrase, (Adj)ective. So, S breaks down into NP immodica ira and VP gignit insaniam. NP breaks down into N ira and Adj immodica, while VP breaks down into V gignit insaniam, which in turn breaks down into V gignit and N insaniam.
These elementary analyses suggest that the tools of modern linguistics can be used in the investigation of Latin, but perhaps only up to a point. There is, for example, no Latin speech which can be examined using the techniques of discourse analysis. Whether it can be analysed in terms of functional linguistics would have to be determined by classicists with a knowledge of the subject, if such there be. But it does seem clear that Latin still has much to yield in addition to the literature and history which have traditionally be seen as its major prizes.