001. We understand sentences because we know (in the sense of, are able to respond to) (1) structure signals and the way they function, and (2) the meaning of words. Words fall for the most part into these two categories: (1) structure-signaling words, (2) major vocabulary.
A sharp division cannot be drawn between structure-signaling words and major vocabulary. Words in the lexical stock fall on a spectrum, some veering more towards mere signaling function and therefore lexically empty, and some specializing in meaning.
A look at the Greek lexicon is instructive in this regard. Bauer’s entry s.v. ἄγγελος indicates that this word is lexically full. ἄγγελος means messenger, of which there are two types: (1) human, (2) supernatural (= angels, an English loan-word). There is virtually no discussion of the grammar of ἄγγελος By contrast, ἵνα is primarily a structure-signaling word. Bauer stipulates that ἵνα has two “meanings”: (1) to denote purpose, aim, or goal, (2) to indicate something other than purpose. The vagueness of the word definition is a clue to the lexical problem. Bauer is forced to organize the long entry as a grammar of ἵνα, in relation to the two “meanings” mentioned above. Under the first he treats ἵνα as a conjunction used to introduce purpose (final) clauses; these are divided up in accordance with the tense and mood of the verb which precedes and follows. Under the second “meaning” he treats other constructions for which a ἵνα-clause may substitute. It is thus clear that ἵνα does not “mean” in the lexical sense, but in the grammatical sense. What it “means” are the structures in Greek that it signals. In some cases it may rightly be translated that, in order that, but it “means” — in those same cases — a grammatical structure (a final clause), the corresponding form of which in English is introduced by in order that.
Every word participates in structure in some way, and most structure—signaling words have some minimal lexical meaning. For practical purposes, however, vocabulary can be thought of in terms of the two basic categories (Gleason: 97 and n. 8).
002. It must not be supposed that structure signals are confined to words. Some structure signals are less than words (in English the plural ending -s [-es] for example, is a structure signal), some are greater than a single word (e.g. either … or). Only in the case of function words is the structure signal coextensive with the word, and even then there are composite function words and there is often correlation with some other signal (e.g. both … and; cf. the discussion of ἵνα in §001).
The structure signals or grammatical devices of English and Greek may be conveniently divided into five categories (Schwab: 3):
(1) function words, such as the, may, and, which signal word classes, phrase structure, sentence structure;
(2) inflections (morphological variables), such as the plural ending -s and the verbal suffix -ed, which signal number, tense, etc.;
(3) derivational forms, such as -ly, -ness, which often, but not always, signal word classes (e.g. glad, gladly, gladness);
(4) word order (word constellation), which serves to mark off the limits and internal relations of word groups;
(5) intonation (e.g. a falling pitch contour marks You don’t believe me.↘ as an assertion in spoken English, as distinguished from a rising pitch contour, You don’t believe me?↗ which signals a question). Intonation is represented in the written language partially and imperfectly as punctuation (cf. the period and question mark).
It should not be supposed, either, that major vocabulary is confined to words. Put up, for example, is a different lexical item than simple put, as the sentence, She put up preserves, indicates. Put up with is still another lexical item, as Winston Churchill observed when he protested: ‘The rule, a preposition should not be used to end a sentence, is a piece of nonsense, up with which I will not put.’