University of Virginia Library

III. ANTIQUITY: GREECE AND ROME

The Greek study of language may be said to start
with Plato and Aristotle. Both of them approach lan-
guage from a philosophical point of view. They are
concerned with the nature of language and its relation
to thought and reality, rather than with the more
technical matter of providing an exact description of
linguistic forms. Such questions were later to be taken
up by the Hellenistic grammarians of Alexandria, who
were mainly concerned with establishing correct edi-
tions of Homer and other classical poets, and with
preserving the purity of the Greek language as it be-
came the common vehicle of communication through-
out the eastern half of the Roman empire.

The history of Western linguistics ever since has been
characterized by this double heritage: philosophical
and philological. On the whole Western linguists have
continued to be somewhat stronger on the philo-
sophical than on the formal side, until the discovery
of Sanskrit and Indian grammar, which enriched the
Western grammatical tradition in the direction of de-
scriptive accuracy and power.

Plato. Plato's views on language are chiefly put forth
in his dialogue Cratylus. He discusses the relation
between words (rather: names, as Plato consistently
refers to the basic elements as onómata). Plato's view,
put in the mouth of Socrates in the dialogue, seems
to be that words may indeed to a certain extent give
a clue to the nature of reality, but that the guidance
they provide is very uncertain. Even if a name (ónoma)
was given at one time by the wisest of philosophers,
in full conformity with the nature of the thing, it is
later exposed to all the vicissitudes of chance and the
whims of ordinary speakers. Hence no safe conclusions
can be drawn from the etymology of the name to the
nature of the thing the name stands for. On the whole,
therefore, Plato's attitude towards the study of lan-
guage is rather unfavorable. Language does not, to him,
provide the key to the realm of true reality.

Before arriving at this somewhat negative conclu-
sion, however, Plato gives considerable attention to
two ideas which unfortunately have exerted a dis-
astrous influence on linguistic thought ever since. The
first is the idea that there is some inner fitness con-
necting the name and the thing it stands for. The
second concerns the way in which the fitness of the
name should be ascertained. Here Plato applies an
extremely loose and ad hoc method of etymologizing
in which (to use Voltaire's quip made two thousand
years later) the consonants counted for little and the
vowels for nothing at all. The reasoning is often so
ridiculously flippant that later commentators have as-
sumed that Plato really meant to hold up the method
to scorn. There is no doubt an element of playful irony
in some of the wilder flights of etymological fancy in
the dialogue. But though Plato certainly saw that the
etymological approach he exemplified so copiously
might be, and was, misused, it seems quite clear that
in principle he considered it a natural and valid way
of analyzing the meaning of a word. In any case, Plato's
etymologies in Cratylus set more or less the pattern
for Western scholars down to the beginning of the
nineteenth century.

Let us consider Plato's discussion of the name
“Poseidon”:

(fol. 402-03; Plato, Works, trans. of Vol.
IV, H. N. Fowler, London and Cambridge, Mass. [1926],
p. 169).

It will be seen that Plato is by no means dogmatic.
Not infrequently he is quite willing to accept the
possibility that several different derivations of a name
may all be considered as valid. The underlying as-
sumption is that names were consciously invented by
an original name-giver, who may well have had more
than one reason for a certain choice. The nearest
analogy to Plato's name-giver would in fact be a mod-
ern inventor of trade names, who indeed works on the
principle that the name should vaguely suggest those
ideas that he believes the customers ought to associate
with the product.

In his search for the smallest elements making up
the names, Plato also considers the idea of sound sym-
bolism. He finds that r should stand for swift move-
ments, l for softness, and i for smallness. But then, he
asks, how can we justify a word like sklērós (“hard”),
which contains an l, a sign for softness? In the end,
therefore, Plato arrives at the conclusion that it is futile
to try to discover the truth of things by analyzing the
names. That does not mean, however, that he con-
demns his previous argument altogether. It is only
when compared with the high standard of perfect
knowledge that the method of etymology falls short.
Plato would also argue that the kind of knowledge that
we get through our sensory organs is imperfect, com-
pared with the ideal knowledge that purely intellectual
contemplation gives.

In the controversy that occupied the Greeks so
much, as to whether language was the product of thésis
(“convention”, another term was nómos “order”) or
phúsis (“nature”), Plato therefore seems to have taken
a middle position. Though he concludes that the
meanings of names are in large part determined by
custom or convention, he seems to look upon this as
due either to corruption or to the ignorance of the
name-givers. Most of his discussion is carried on with
the assumption that, at least in an ideal language, there
is a fundamental fitness connecting the name with the
thing. Such an idea leads to confusing the form and
the content of the linguistic sign, and was to form the
basis of both weak linguistics and bad metaphysics.

More important, however, is the fact that Cratylus
gave its sanction to such a disastrous pattern for analyz-
ing words. It is not only that the majority of the ety-
mologies are wrong, if considered as statements of word
history or derivation. The worst of it is that the method
as such was perverse. By allowing sounds to be
changed, dropped, or added in a perfectly haphazard
way, in order to make a suggested etymology fit, one
gave up in advance the possibility of finding a consis-

tent pattern of word-formation in the language. It may
have been this cavalier attitude towards sound changes
in words that prevented the Greeks—and the Romans—
from making consistent use of such a fundamental
distinction as that between stem and ending, between
root and affix. The Sanskrit grammarians did immensely
better in this respect. It must be admitted, though, that
the Greek language is unusually intractable to ety-
mology. Indeed, it may even be said to favor the kind
of analysis where sounds can be exchanged without
limit. The declension and conjugation systems provide
examples of almost every sort of change. We find the
“addition” of letters in gígās, gen. gígantos “giant”;
ónoma, gen. onómatos “name”. We find “loss” of letters
in kúōn, gen. kunós “dog”; and alteration of vowels
as in hēdús, gen. hēdéos “sweet.” Finally the verbal
system yields a rich harvest: gígnomai “I am born,”
gegénēmai, gégona “I was born.” It was not easy to
discover any organizing principle in such a variable
material.

Aristotle. What Aristotle has to say on linguistic
topics is almost wholly incidental to his concern with
logic. We find it chiefly in his treatises on the Categories
and on Interpretation (Perì hermeneías). Scattered re-
marks are also to be found in the Rhetoric and Poetics.

What he says is brief, to the point, and generally
sound. Aristotle does not dabble in etymology, though
he recognizes that words may be connected with each
other: “Things are said to be named derivatively, which
derive their name from some other name, but differ
from it in termination. Thus the grammarian derives
his name from the word grammar, and the courageous
man from the word courage” (Categories, 1; Works
..., ed. W. D. Ross, Vol. I, trans. E. M. Edghill, Ox-
ford [1928]). Here Aristotle was on the point of discover-
ing the difference between base and derivation mor-
pheme, but he never developed the idea any further.

Aristotle comes down squarely on the side of thésis
in the phúsis-thésis controversy. Words, he also says,
are “significant by convention” (On Interpretation,
16). He has no patience with the Platonic idea that
a word, as such, may be “true”: “Nouns and verbs...
as isolated terms, are not yet either true or false” (ibid.).
Truth or falsity can be predicated of propositions only.

His explanation of the relation between writing,
speech, and meaning is admirably clear: “Spoken words
are the symbols of mental experience and written
words are the symbols of spoken words. Just as all men
have not the same writing, so all men have not the
same speech sounds, but the mental experiences, which
these directly symbolize, are the same for all, as also
are those things of which our experiences are the
images” (ibid.). The later part of this quotation is the
foundation of general grammar.

Aristotle's main contribution to linguistics is his
careful definition of some important syntactic terms.
Thus he distinguishes between a proposition, such as
the man runs, which expresses a fact and hence may
be either true or false, and single expressions, such as
a man, or runs, or a footed animal with two feet. Not
all sentences are propositions, though. A prayer, for
instance, is not a proposition. The word sentence itself
is defined, somewhat weakly, as “a significant portion
of speech, some parts of which have an independent
meaning” (ibid.).

Finally, Aristotle defines the principal parts of
speech, ónoma (noun—or subject) and rhēma (verb—or
predicate). An ónoma is “a sound significant by con-
vention, which has no reference to time” (ibid.). A
rhēma is “that which, in addition to its proper meaning,
carries with it the notion of time... it is a sign of
something said of something else” (ibid.).

It will be seen that Aristotle defines the parts of
speech with reference to their function and meaning,
rather than with reference to their form. The categories
he has in mind should really be called sentence constit-
uents. The second part of the verb definition clearly
refers to the predicate of a sentence rather than to
a word class. It is significant that the oblique cases of
nouns are not looked upon by Aristotle as onómata,
since, as he says, they cannot form propositions to-
gether with a rhēma. More surprisingly Aristotle ac-
cepts only the present tense forms as true verbs. Past
and future forms are “not verbs, but tenses of verbs.”

The analysis of the phrase as consisting of a combi-
nation of ónoma and rhēma is not Aristotle's invention.
We find it previously in Plato, who hints vaguely at
it in Cratylus, but is somewhat more explicit in the
Sophist (fol. 262). Like Aristotle, Plato defines the terms
as sentence constituents rather than as word classes.
The ónoma indicates the performer of an action, the
rhēma the action itself. And every complete sentence
has to contain both an ónoma and a rhēma. Unlike
Aristotle, however, Plato does not count time indica-
tion as essential to the verb.

The Grammarians. Neither Plato nor Aristotle were
primarily concerned with linguistics proper, with
questions having to do with the formal structure of
language. If they analyzed expressions and sentences,
it was in order better to understand what they stood
for—ideas and propositions. Plato discussed language
because of the light it might shed on the nature of
knowledge; Aristotle because of its importance to logic.

During the following centuries more and more em-
phasis was given to the formal side of the discourse.
But the logical and semantic foundation that had been
laid was never abandoned. That might be a weakness
from the point of view of theoretical consistency. On

the other hand, thanks to this double base the linguistic
theory of antiquity was flexible enough to be adapted
to the description not only of Greek, but also of Latin,
and later of the modern languages as well. It is no
coincidence that the ónoma-rhēma (noun-verb) di-
chotomy has provided a starting-point for most at-
tempts that have been made to formulate a general
grammar.

The first to elaborate the parts of speech theory
further seem to have been the Stoics. They recognized
four parts: the noun (ónoma), the verb (rhēma), the
conjunction (súndesmos), and the article (árthron). The
latter two terms are also to be found in Aristotle, but
the Stoics clarified the difference between them. The
conjunctions are undeclinable, while the articles are
declined for case.

The term case (ptōsis) was also redefined by the
Stoics. While Aristotle had used it for both nouns and
verbs, the Stoics restricted it to the noun. They also
gave currency to the case names that are still in use:
onomastikē (“nominative”), genikē (“genitive”), dotikē
(“dative”) and aitiatikē (“accusative,” a mistranslation
for “effective” or “causative”).

The Stoics also tried to make a systematic descrip-
tion of the very complicated Greek verb conjugations.
According to Aristotle's definition of the verb, time
indication was essential to it. The Stoics, when pushing
the analysis further, discovered that the Greek verb
forms also had other functions: to indicate aspect (i.e.,
completed or incompleted action), mood (indicative,
subjunctive, optative, imperative), and voice (active,
medial, or passive). As far as we know—no complete
account of the Stoic philosophy has come down to
us—they did not quite succeed in unravelling the in-
terrelation of these concepts. Perhaps because they
held on to the view that time indication was the funda-
mental verbal function, they did not manage to disen-
tangle the aspect and tense functions. The mood con-
cept was treated as belonging to the sentence type
rather than to the verb form. There is nothing surpris-
ing about the Stoic's failure on these points. After all,
they were not chiefly concerned with linguistic form.
Their interest, like Plato's and Aristotle's, was philo-
sophical rather than grammatical.

While the philosophers analyzed language in order
to understand reality, the grammarians were interested
in language for its own sake. Or rather, they needed
a correct description of the language system in order
to judge and interpret the visible products of that
system, whether classical literary texts that needed
explanation, or spoken language that needed supervi-
sion so as to conform to a standard model. Both objects
were of importance in Hellenistic times. The Homeric
text, which was the foundation of all Greek literary
education, was becoming more and more remote from
ordinary Greek, and the spread of Greek culture and
Greek language all around the Eastern Mediterranean
created a demand for pure hellēnismós, as the phrase
was. The time was ripe for the grammarians.

One of the outcomes of the search for a standard
was the struggle between anomalists and analogists.
The analogists stressed the regularity of language. They
tried to reduce its apparent chaos to order by estab-
lishing analogies: paradigms and schemas for the de-
clension of words. The anomalists insisted that lan-
guage was not determined by rules, but by custom.
While naturally not denying that analogies could be
established, they maintained that the assignment of
words to different declension classes or conjugations
was largely haphazard.

When faced with textual obscurities and difficulties,
the analogists tended to emend the text by analogy
with more common forms. The anomalists were more
prone to accept the text as it stood. In their attitude
to the standard language analogists took a normative
stand, while anomalists would bow to custom and good
usage. Analogists were authoritarian (Caesar published
a pamphlet on the analogist side); anomalists were, if
not democrats, at least liberal conservatives. The de-
bate continued for several centuries. Out of it classical
grammar developed. The only way for the analogists
to prove their case was to show that it was indeed
possible to reduce the apparent chaos of language to
some sort of order. And all the time the anomalists
compelled their adversaries to improve their descrip-
tions by pointing to cases which did not fit the rules.
In this way it was the language as such, not the philo-
sophical uses of it, that occupied the center of the
stage.

The work of the Greek grammarians, of whom
Aristarchus of Alexandria (fl. ca. 160 B.C.) was the most
famous, has come down to us in two versions, the
grammars of Dionysius Thrax (fl. ca. 100 B.C.) and of
Apollonius Dyscolus (fl. ca. A.D. 180). The former is
a brief compendium of some twenty pages, which
contains little more than definitions and explanations
of the chief grammatical terms. The latter, which uses
essentially the same terminology, is considerably longer
and has a large separate section on syntax, which is
entirely absent in Dionysius' book.

Dionysius defines grammar as having to do with
pronunciation, explanation of textual difficulties and
stylistic features, etymology, and, as he puts it, “the
discovery of analogies.” His approach is entirely philo-
logical and literary. This is a reflection of the fact that
grammar had arisen out of the study of the old texts,
and especially Homer. The wider, philosophical ques-
tions of the nature of language and the relation be-

tween language and thought were not touched upon
at all by Dionysius.

Dionysius' “discovery of analogies,” i.e., his mor-
phology, introduces practically the whole of the con-
ceptual apparatus of what was later to become tradi-
tional grammar. Like Aristotle, Dionysius recognizes
two main parts of speech, noun and verb. His definition
of them vaguely recalls Aristotle's, but it is more
formal. He definitely seeks to define a word as a mem-
ber of a class, not as a constituent of a sentence. The
noun is defined formally as having case inflection,
semantically as signifying “a person or thing.” The
different cases are enumerated and named. This is one
of the great achievements of the Greek grammarians
(and philosophers), since it requires quite a high degree
of linguistic abstraction, in view of the fact that the
declension morphemes in Greek and Latin express not
only case but also number and gender. They also differ
from one declension to another, and—especially in
Greek—undergo extensive changes in different envi-
ronments. As the classical grammarians never recog-
nized any smaller semantic unit than the word—they
never spoke about morphemes or suffixes, only about
word endings—their task was made all the more diffi-
cult. All through antiquity the noun class included the
adjectives. The adjective was not even recognized as
one of the major subdivisions of the noun class.
Dionysius' major subdivision is a formal one: original
words (like earth, white) as against derived ones (like
earthly, whiteness). The failure to separate the adjec-
tive class was obviously due to the ancient grammari-
ans' weakness in the field of syntax.

The verb is defined by Dionysius as being devoid
of case, but having tense, mood, person, number, and
“kind” (active or passive). This is the system of tradi-
tional grammar almost full-fledged. Where Dionysius
was weak, no improvement was to be made until mod-
ern times. That concerns above all the tense system,
which he mixed up with the aspect system.

The other word classes recognized by Dionysius are:
metochē (participle; thus called because it had both
case and tense inflection, and thus participated in both
the noun and the verb class), árthron (article; a case-
forming part preceding a noun), antōnymía (pronoun;
used in place of a noun), próthesis (preposition; could
occur before all parts of speech—prefixes like ad- in
adapt were considered as prepositions), epírrhēma (ad-
verb; says something about a verb), and súndesmos
(conjunction; “links together our thoughts in a deter-
mined order”).

The accidence expounded by Dionysius Thrax is
found practically unchanged in Apollonius Dyscolus,
and was taken over almost completely by such Roman
grammarians as Donatus and Priscianus. A somewhat
more independent position was taken by the Latin
writer Varro in his book De lingua latina, of which
only six of the twenty-five chapters have been pre-
served.

The main change introduced by the Latin authors
was to drop the article as a special word class (as Latin
has no articles), and to introduce the interjection in
its stead. Minor changes were due to the obvious differ-
ences between the Greek and Latin accidence, for
instance, the ablative case. But the rather fundamental
differences between the Latin and Greek verb systems
were not clearly recognized. Priscianus tried, for in-
stance, to carry over the Greek distinction between
the optative and the subjunctive to Latin, thus indi-
cating that he had not firmly grasped the fundamental
principle of basing the morphology on the formal
distinctions made in the language. The same weakness
can be found, down to the present time, in a host of
traditional grammars of modern languages.

The ancient grammarians' work on syntax was far
inferior to that on accidence. This is due to the fact
that they did not develop any theory of the sentence
and sentence constituency. To the philosophers, ónoma
and rhēma had been sentence constituents rather than
word classes. The grammarians, while retaining the
terms, had changed their function. As a result they did
not know how to start analyzing the constituents of
sentences. To Apollonius Dyscolus, as to Priscianus who
followed him, syntax was therefore a question of finding
out how word classes and word forms could be com-
bined with each other on the basis of their intrinsic
characteristics. This approach can achieve a limited
success—for instance, to explain the concord of adjec-
tives and nouns in the noun phrase, and the government
of prepositions. But it cannot succeed at all in explain-
ing the internal constituency of the verb or noun phrase
in general, nor the constituency of the simple sentence,
nor the interconnection of sentences and clauses, or
infinitival or participial clause-like construction.

Every classical grammar started out with an account
of letters and their pronunciation. That the point of
departure was the letter, not the speech sound
(phoneme) was natural, as grammar had arisen out of
the study of literary texts, not the spoken language.
The consequence of this approach, however, was that
the letter and not the sound continued to be the funda-
mental unit in terms of which the form of words was
discussed. Moreover, writers were constantly mixing
up the two concepts.

The phonetic theory of the ancients was very defi-
cient. No even remotely exact articulatory phonetics
was developed. (On the Indian achievement, see be-
low.) It is true that a consistent distinction was made
between vowels and non-vowels. But the consonants

were not described very efficiently. The distinction
between voiced and unvoiced consonants, such as b
and p, was not stated with any accuracy. Great impor-
tance was attached to the distinction between stops
and continuants. The latter were generally called semi-
vowels, a group which accordingly contained not only
l, m, n, r, but also s and f.

As regards etymology no real improvement was
made on the state of things illustrated in Cratylus.
Though the grammarians recognized several types of
derivation of verbs and nouns, and accurately described
the different declension and conjugation classes, they
never arrived at a clear view of the concept of the
morpheme, whether word-base, derivational affix, or
case affix. As long as this was so, they could not estab-
lish the main connections between the words in their
own language, let alone those between Latin and
Greek. The ancients were of course aware that these
two languages were related, but could not make the
proper distinction between similarities due to common
origin and regular change, and similarities due to word
loans. They had no realistic idea of the mechanism of
language changes. They tried to explain them as due
either to chance or to conscious manipulation by the
speakers. Words were changed, they thought, or ap-
peared in irregular shapes, for reasons of euphony, or
to avoid ambiguity, or for some metaphysical reason
or other. For instance, an e might be changed into a
because a is more “dignified,” as it comes first in the
alphabet.

Against such a background one should not be too
surprised to find even the most ridiculous etymologies
advanced quite seriously. Lapis “stone” was derived
from laedens pedem “hurting the foot.” Even more
remarkably, lucus “forest” was derived from lucere “be
bright” because of the lack of light in the forest. In
the same vein, bellum “war” was derived from bellus
“beautiful” because war is the opposite of beautiful.
These and other similar “etymologies” kept reappear-
ing all through the Middle Ages and later, until the
Europeans learned better from the Indians.