What Graeco-Roman Grammar Was About

P. H. Matthews

https://doi.org/10.1093/oso/9780198830115.001.0001
Published: 2019 Online ISBN: 9780191868467 Print ISBN: 9780198830115

Search in this book

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
CHAPTER

5 The eight parts 

P. H. Matthews

https://doi.org/10.1093/oso/9780198830115.003.0005 Pages 71–92

Published: February 2019

Abstract
This chapter identi es the parts of utterances. The ancient ‘parts’ are among the most abiding legacies
of Graeco-Roman grammar. They were rst distinguished in Greek. According to Quintilian, eight
parts were distinguished by Aristarchus in the second century BC ; it can be assumed that they were the
categories of word forms that are later familiar. In the system as set out by Donatus, nouns were rst
distinguished from pronouns, whose roles in syntax are similar. The next were verbs and adverbs; after
them rst participles, then conjunctions, then prepositions. A nal eighth part, corresponding to a
subclass as described in Greek, was the interjection. Ultimately, the parts of utterances formed not
simply a set, but what one would now see as a system, in which categories were ordered rationally, in a
way that re ected the connections between them.

Keywords: utterance, Graeco-Roman grammar, noun, pronoun, verb, adverb, first participle, conjunction,
preposition, interjection
Subject: Historical and Diachronic Linguistics, Grammar, Syntax and Morphology
Collection: Oxford Scholarship Online

The syntactic categories of English, as set out in recent grammars, do not distinguish, at the highest level,
participles from verbs and do at that level distinguish adjectives from nouns. In many ways, however, they
may appear not very di erent from the system of ‘parts of an utterance’ established in Greek and Latin two
millennia before. Nouns are distinguished from verbs; nouns from pronouns; prepositions from both a
wider class of adverbs and a class, itself split or united, of conjunctions or their equivalent. Some
intervening analyses, such as that of Jespersen (1924), have been more radical. Changes, however, have
mostly a ected levels in a hierarchy, either promoting what had earlier been a subclass to the status of a
major category, or consolidating what had earlier been separate categories into one. At whatever level they
have been distinguished, classes such as noun and adjective or verb and adverb ‘have a history’, in the
words of Huddleston and Pullum, ‘going back to the grammar of classical Latin and classical Greek some
 2,000 years ago’. But, they add, ‘they are apparently applicable to almost all human languages’ (2002). In
many accounts a variant of the same scheme is now taken to be universal.

p. 72 The ancient ‘parts’ are thus among the most abiding legacies of Graeco-Roman grammar. They were rst
distinguished in Greek, beginning in a largely discredited history of their development with Plato’s division
of an onoma from a rhēma (above, Box 4.2). According to Quintilian, writing in the rst century AD , eight
parts were distinguished, in the second century BC , by Aristarchus (Inst. 1.4.20); and although he does not
tell us what they were, it can be assumed that they were the categories of word forms that are later familiar.
At the head of the list, as we know it from later sources, nouns are distinguished from verbs, and both from

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
participles. Nouns are distinguished from articles and pronouns, with which they shared in ections for
case; prepositions from adverbs and, at the end of the list, conjunctions. This is the order in which the parts
of the utterance are de ned in, for example, the manual of ‘Dionysius Thrax’, whose ostensible author was a
pupil of Aristarchus.

In another view, still current in Quintilian’s day, the parts were nine, with an additional division, mentioned
brie y near the end of Chapter 1, between names and ‘addressings’. We also have evidence, from the work of
Varro in the rst century BC (Box 4.2), of at least one other analysis that seems to have been airbrushed out
of later doxographies. But eight parts were again distinguished, as Quintilian tells us, by his older
contemporary Remmius Palaemon. Their adaptation to Latin was not slavish: it was clear, in particular, that
there was no equivalent of the words the Greek grammarians distinguished as the article. The languages
were nevertheless more like each other than either is like, for example, English; and the main di erence, as
the categories were distinguished by Roman grammarians, is in the order in which they were presented. In
the system as set out by, among others Donatus, nouns were rst distinguished from pronouns, whose roles
p. 73 in syntax are similar. The next were verbs and adverbs; after them rst participles, then conjunctions,
then prepositions. A nal eighth part, corresponding to a subclass as described in Greek, was the
interjection.

For ancient writers the order in which things were listed was important. Even that of the alphabet ought not
to be entirely arbitrary. The parts of utterances formed not simply a set, but what we would now see as a
system, in which categories were ordered rationally, in a way that re ected the connections between them.
An adverb, for example, was de ned by its relation to a verb, and in the list set out by Roman grammarians
they were adjacent. Adverbs, however, are among the parts that were not in ected for case or tense; and, in
the list set out in grammars of Greek, these were grouped together at the end. It is signi cant that, in
following the detailed arguments of Apollonius Dyscolus, Priscian was to reject what we may call the Roman
scheme, which had been standard in the western empire.

Why words were central

It is at best very di cult for a linguist trained in the twentieth or twenty- rst century, and speaking a
language whose type is formally di erent, to look at Greek in the way it may have presented itself to, for
example, a Stoic philosopher of the third century BC , when a system of this type may rst have been
envisaged. To appreciate, however, why it became so central to Graeco-Roman grammar, and the criteria by
which the parts of utterances came to be de ned, it may be helpful to bear in mind some of the salient
characteristics of the older Indo-European languages, of the family to which both Greek and Latin belong.

Words, as units, might have stared one in the face. The order of forms in utterances varied: in Latin, which
p. 74 we may take for illustration, one could say, for example, veniunt hominēs ‘(The) people are coming’, but
one could also say, in talking of the same movement of a group of people, hominēs veniunt. In comparison,
neither veniunt ‘are coming’ nor hominēs ‘people’ could be divided into smaller parts whose sequence could
 itself be varied. At the same time either form was complex. The word that in modern practice can be glossed
as:

was both partly like and partly unlike others that in a modern analysis include the same stem. Compare
among others:

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
or:

As forms varied so did meanings, and the further meaning such forms have in common, which we now
distinguish as ‘lexical’, was also easy to recognize.

Di erent patterns of variation, with similar di erences of meaning, could then be associated clearly with
distinct types of, as we would now say, lexeme. A range of in ections semantically like those found for
‘coming’ also distinguished, for example:

p. 75 from:

or:

and corresponding forms for running, swimming, and so on. Another pattern of variation distinguished
forms like hominēs ‘people’ from the accusative singular hominem or the genitive hominum ‘of people’, and
the corresponding forms for women, children, animals, and so on.

Across languages in general, such distinctions are not always so neat. In English, for comparison, forms like
comes or coming illustrate one pattern of variation; boy and boys, for example, another. Splash, however, is
one of many stems that can combine directly in forms parallel to both. In, for example, a big splash it forms
the singular, as we describe it, of a noun. This is distinguished in dictionaries as one lexeme: ‘splash, n.’. In
They are splashing about, the same form combines with the -ing of They are coming. Therefore we distinguish
another lexeme, ‘splash, v.’, whose entry in a dictionary is separate. But the meanings of ‘splash, n.’ and
‘splash, v.’ are plainly connected, as are those of ‘love, n.’ and ‘love, v.’, ‘run, n.’ and ‘run, v.’, and many
others. If they are seen as separate it is historically, at least, because noun and verb are categories we have
inherited from the ancient grammarians, in whose languages patterns of in ection partitioned forms more
p. 76 clearly. At another extreme are languages in which splash-type distributions, as we might describe them,
are the norm, and it is in analysing these that not surprisingly the distinction between nouns and verbs,
which is the foundation of the doctrine of parts of speech, has often been brought into question. In the
ancient languages, however, in which such categories were initially distinguished, di erent patterns of
in ection correlated far more clearly both with meanings each set had in common, as referring to men or
women, coming or walking, and so on, and the connections of forms within utterances.

Finally, we are talking of patterns in the abstract; not of endings, for example, of forms individually. A nal
-nt in venient or ambulant, among the forms glossed, was unusual in Latin in that, in itself, it distinguishes a

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
third plural of a verb. But a form that ended in -ō could be either a verb like veniō ‘I am coming’; or, for
example, a noun like puerō ‘boy-ABL.SG’ in ā puerō ‘by the boy’. Veniam ‘I will come’ is, like veniō, a rst
singular; but it ends in am. So does, for example, the form puellam ‘girl-ACC.SG’ in puellam vīdī ‘I have seen
a/the girl’. A form in -ēs could have a meaning like that of hominēs ‘men’; but also that of verb in the future
such as veniēs ‘you (sg) will come’ or a present such as timēs ‘you are afraid of’. Other distinctions similar in
meaning could be drawn in varying ways. Where veniam ‘I will come’ di ered from veniō ‘I am coming’ in
its ending, ambulābō ‘I will walk’, which was likewise future, was distinguished from ambulō ‘I am walking’
by, on the face of it, an inserted ā and b. In puerī veniunt ‘(The) boys are coming’ the form for boys, puerī, has
a nal ī; in hominēs veniunt ‘(The) people are coming’, the form for ‘people’ again ends in ēs.

The same points can be made for Greek, as will be illustrated, in conclusion, in Chapter 12. Not every
p. 77 language, however, has in ections that contrast similarly. If we set aside, accordingly, the way the
ancient ‘parts of speech’ have been reshaped by later traditions, it is possible to see them as in origin a
natural though a brilliant response, by philosophers and philologists in the Hellenistic period, to the
problems posed by a language of a particular type, in which the partition of word forms into sets with
similar semantic contrasts was the key to its structure. Nor is it surprising that categories came to be
distinguished by all relevant criteria. If we take parts of an utterance for granted, we may speak in detail of
‘semantic’ criteria, which appeal to types of entities, activities, and so on referred to; of ‘morphological’
criteria, which appeal to in ections; of ‘syntactic’ criteria, which appeal to relations of words in larger units.
In the insight, however, that led to the analysis, all these are aspects of meaning that could have been taken
together.

The system of definitions

Our sources, alas, are spotty and centuries later. If we return, however, to the account transmitted by the
grammarians, the list of parts begins, in either order, with the noun. This was the ‘Hauptwort’, as it has
come to be called in the German tradition, from which other parts were distinguished, and its speci c
character, as Priscian put it in an introductory survey, is to signify (in Latin) substantia and qualitas (GL 2:
55, l. 6). These terms were eventually borrowed into English, as technical terms in philosophy. Their
meanings are not, however, those that are normal in present-day English, of physical ‘substance’ and of a
‘quality’ that may be higher or lower. Substantia was instead a term that corresponded to Greek ousia,
literally ‘being’, by which Aristotle had distinguished something essential or unchangeably real. It is by this
p. 78 that nouns were de ned in one account in Greek (GG 1.3: 215, l. 26 and elsewhere). Qualitas had been
coined by Cicero, translating a term of Plato’s, and was derived transparently from qualis ‘of what sort?’ The
sense was, therefore, literally that of ‘what-sort-ness’, and was opposed directly to that of quantitas, as a
term coined similarly with the meaning literally of ‘what-size-ness’. In a formulation then that in the
Middle Ages was to become especially illuminating, a noun signi ed the essential nature of some entity or a
property that varies. Ancient de nitions also make clear that nouns had ‘fallings’ or, as we now say, ‘cases’.
 The term for a noun was the ordinary word for a name: Greek onoma or Latin nomen. One’s onoma, as a
Greek, could be (as in Plato’s example in Box 4.2) Theaítētos; one’s nomina ‘names’ as a Roman could be
Gaius Iulius Caesar. In a scheme with nine parts, which was a live alternative in the rst century, forms such
as these, which were names in a strict sense, were distinguished from ones such as, in Greek, ánthrōpos
‘human being’ or, in Latin, puer ‘boy’. These were again distinguished as we saw in Chapter 1; and in the
Stoic system of parts of an utterance, as it is represented in later grammatical and other sources, an onoma
and a prosēgoria were two of ve parts of an utterance. They had similar roles, however, in the structure of
sentences, and similar patterns of in ections, and the grammarians of a later period mention this view only
to reject it. The rst subdivision of nouns, as we will see in the next chapter, was then between, in Greek, an

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
onoma kurion, or noun of a subclass that was called in Latin proprium or ‘one’s own’, and what was in Latin,
in a term re ecting the sense of prosēgoria in Greek, an appellatio.

The Greek order, as we may call it, in which parts were listed is as in the manual of ‘Dionysius Thrax’; and, if
it is not that of the historical Dionysius, it had behind it the authority of Apollonius Dyscolus (Synt. 1.13–29 =
p. 79 GG 2.2: 15–27) in the second century AD. It is again the one applied by Priscian, in the fth century AD, to
Latin. In either arrangement, however, the noun and the verb had a special status. An onoma and a rhēma
had been distinguished by Plato (Box 4.2), and in the standard story, as the grammarians represented it,
these were the only ‘parts of the utterance’ according to the earliest school of philosophers. For the
grammarians themselves they were, as Priscian puts it, ‘primary and outstanding’ (principales et egregiae),
and the other parts were ‘appendages’ (appendices) subsidiary to them (GL 2: 552, ll. 12–14). They were
likewise primary, in the tradition of Roman grammars, for Donatus (GL 4: 372). Apollonius and Priscian
argue, in particular, that both were essential if an utterance, or in their terms a logos or oratio, were to meet
a criterion (as we will see in Box 9.1 below) of completeness. Other parts were, as we would now say,
optional. To illustrate this, Priscian takes an utterance which in ancient accounts included all parts other
than a conjunction:

The example is adapted from Apollonius (Synt. 1.14 = GG 2.2: 17), as is the argument. Delete any of idem,
lapsus, heu, or hodiē and the utterance ‘will not entirely fall short’ (non omnino de ciet). These are the parts
other than the noun and verb. Delete con-, which was a preposition, and the corresponding simple form,
which was cecidit ‘fell’, would be su cient. Delete, however, either homō or -cidit and the utterance does fall
p. 80 short (GL 3: 116). ‘Will fall short’ is in Greek elleipsei. An utterance which consisted of a verb or noun alone
was implicitly, as in one of the plethora of accounts that are current now, ‘elliptical’.

For the earliest philosophers, as the grammarians interpreted their doctrines, other units ‘consigni ed’, or
had meaning, as we would now put it, in conjunction with them. But, as the example of the man falling over
shows, the other parts were also meaningful. Each had, in Greek, its own sēmasia, what it was a ‘sign’ of, or
something that it ‘made clear’. In Latin, it had its own signi catio or ‘signi cation’. The parts in general
cannot therefore be distinguished one from another ‘unless’, as Priscian puts it at the head of his initial
survey, ‘we pay attention to what is speci c to the signi cations of each’ (nisi uniuscuiusque proprietates
signi cationum attendamus) (GL 2: 55, ll. 4–5). Here too he was simply following Apollonius.

Priscian’s de nitions are translated, with those of Apollonius where they are known, in Box 5.1. Where the
proprietas or speci c character of a noun was to signify substantia and qualitas, that of a verb was to signify
something distinguished in part by what is now called ‘voice’. It was thus an action either performed on
someone or something or one which is instead experienced or undergone; or else, as de nitions are worded,
 neither. Verbs also, in particular, distinguish times. The status of nite verbs was from the beginning
undisputed; ‘ nite’ in antiquity meant ‘distinguishing rst, second, and third person’. Some ancient
grammars, such as those of both Donatus and ‘Dionysius Thrax’ (see below, Box 5.2) also included persons
in their de nition. Apollonius Dyscolus, however, had been at pains to make clear that in nitives, which did
not distinguish persons, were also a ‘mood’, as they are still traditionally described, of verbs (Synt. 3.55–9 =
GG 2.2: 320–5). Priscian again followed him.

Participles, as we have seen in the last chapter, were not verbs. The insight is said by Apollonius to be that of
p. 81 Tryphon, a grammarian of the late rst century BC , whose work is often cited and, if we had it, might

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
throw crucial light on how analyses evolved in that period. We may also remind ourselves of the account
adapted to Latin by Tryphon’s contemporary Varro (Box 4.2), in which the main parts were de ned,
consistently but unusually, by in ectional meanings alone. In Priscian’s survey a participle, unlike an
in nitive, is ‘rightly separated’ from a verb because it has cases, which a verb lacks, and genders, like a
noun, without ‘moods’, which verbs do have (GL 2: 55, ll. 10–12). But fuller de nitions make clear that a
speci c ‘signi cation’, or what we would again distinguish as its lexical meaning, is derived from that of a
verb (Box 5.1). As a part related to both nouns and verbs, a participle is the third in the Greek order, which
Priscian also follows.

Next, in that order, came the pronoun. The term in Greek (antōnomia) was a compound formed with a
preposition, ant(i) ‘instead of’, and the term for a noun, and made clear how these parts were related. In the
‘Roman’ order of Donatus and others pronomina were listed directly after nouns, but in all accounts they
signi ed in place of nouns in general or some nouns in particular. For Priscian what was speci c to a
pronoun was to be used in place of some noun in the ‘proper’ sense (pro aliquo nomine proprio poni), and to
‘signify de nite persons’ (certas signi care personas) (GL 2: 55, ll. 13–14). This is one part for which the
reasoning of Apollonius survives in detail. The distinction in Greek between a pronoun and an article, as he
and Priscian put it in another context, is that pronouns are admitted ‘in place of’ (pro) nouns. (Note in this
context that ‘nouns’ include, as we will see, what we distinguish as adjectives.) Articles are instead admitted
‘with’ nouns (Apollonius Synt. 2.1 = GG 2.2: 267; Priscian GL 3: 139, ll. 25–6). A neuter article could also
combine with an in nitive:

p. 82

and forms which for us are relative pronouns, beginning what we now call relative clauses, were described,
as we will see in Chapter 10, as articles postposed to nouns. To say that Latin had no article, as Priscian’s
survey states at the outset, is to acknowledge that no forms entered into a similar range of uses.

A noun, a participle, a pronoun, and an article all had in common, as grammarians remark, that they
distinguished cases. They were therefore identi ed collectively, in distinction in particular from verbs, as
parts which had that property: in Greek ptōtika (having a property of ptōsis, literally ‘falling’); in Latin
casuales, from casus. As such they are distinct from verbs, which were regularly de ned as, among other
things, ‘without case’; also from the parts which remain, at the end in the Greek order, which are all
‘unin ected’.

The rst of these is the preposition. In some uses it was, like other parts, a dictio or word form: thus, in
Latin, in a form that grammarians distinguished as accented, the in of in Italiā ‘in Italy’. But similar forms,
which they distinguished as unaccented, had another use, as we have seen in the last chapter, in a
compound such as advenit ‘arrives’ or in uit ‘ ows in(to)’. The notion of ‘composition’ is ancient (Greek
sunthesis ‘placing together’) and could in principle be distinguished from suntaxis or ‘arranging together’,
which is the source of the modern ‘syntax’. In either relation units like in had a xed position: before uit
and, unusually in a language in which the sequence of most combinations of words could vary, before a
 noun such as Italiā. It was therefore natural to see them as, in either case, the same part of an utterance, in
p. 83 Greek a prothesis or in Latin a praepositio ‘placing before’. What was speci c to them was accordingly just
that position, in relation variously to parts which, like Italiā, were in ected for case, or in composition with
other parts generally.

An adverb (Latin adverbium, Greek epirrhēma), was both named and de ned by its relation to a rhēma or
verb. It followed immediately in the Roman order, as the pronoun came after the noun; and what was
speci c to it, in the words of Priscian’s survey, was ‘to be placed with a verb and, without it, to be unable to
have a complete signi cation’ (cum verbo poni nec sine eo perfectam signi cationem posse habere). In this

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
case, again, his source in the work of Apollonius is preserved in detail. A part that, as distinguished by
Roman grammarians, was an interjection corresponded to a subclass of forms, as distinguished by
grammarians of Greek, which they continued to describe as adverbs. One reason for treating them as
separate is implied by the term interiectio, literally ‘something placed between’. They were not added, that
is, to another part speci cally.

The nal part, in the Greek order, is the conjunction. The term too was transparent: in Greek a sundesmos or
‘binding together’; in Latin a conjunctio or ‘conjoining’. What is speci c to it, in Priscian’s account, is ‘to
join together di erent nouns or any other words that distinguish cases, or di erent verbs or adverbs’
(diversa nomina vel quascumque dictiones casuales vel diversa verba vel adverbia conjungere) (GL 2: 56, ll. 11–
12). But in other accounts, for which one source is in the later commentaries on ‘Dionysius Thrax’, a
conjunction, which included what in modern accounts are classed as particles, was a part of an utterance
that still ‘consigni es’ (sussēmainei) or ‘signi es jointly’ (compare GG 1.3: 284, ll. 6–10). It did not signify,
that is, on its own; but only in linkage with others.

A survey such as this, across two languages and grammars over a long period of time, may seem perilously
p. 84 synthetic. If we compare, however, the de nitions of Priscian with those of Apollonius in

p. 86
Box 5.1 and these in turn with those of other ancient authorities, the similarities are more striking than the
variation. The de nitions of Donatus and ‘Dionysius Thrax’, which have both had a great in uence beyond
the time when they were written, are set out for comparison in Box 5.2. One di erence, for example, is that
verbs are de ned as words that distinguish persons. That was strictly true of some verbs only, since in both
accounts the category included in nitives. The formula may, of course, have been inherited from a period
before their status had been made clear. Other verbs, however, did distinguish a person, and a reminder that
it was so might also have been helpful for a teacher.
 Box 5.1 The parts as defined by Apollonius Dyscolus or Priscian

Sources are partly in the work of Apollonius that survives. For the rest we rely on Priscian, who
explicitly admired him; but equivalents in Greek are often cited, from sources that in many instances
remain anonymous, in successive commentaries on the manual of ‘Dionysius Thrax’.

A noun, for Apollonius as for Priscian, is in ected for case and ‘assigns to every material or non-
1
material entity that is the subject of predication a shared or individual what-sort-ness’.

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
A verb, as de ned by Priscian, is a part of an utterance ‘with times and moods, without case, signifying
2
what is done or experienced’.

A participle, again following Priscian, is a part ‘which is admitted in place of a verb, from which it is
also by nature derived, which has gender and case like a noun, and properties applying to a verb
3
without distinction of persons and moods’.

A pronoun is distinguished by Apollonius, in a careful study which survives, as ‘a word indicative, in

place of a noun, of de nite persons, distinguished according to case and number’. Forms do not
4
always, he adds, distinguish gender. As distinguished by Priscian, it ‘is admitted in place’ of a proper
5
noun of an individual, and ‘receives de nite persons’.

For the article in Greek, which had no equivalent in Latin, no formula of Apollonius survives directly.
Of the remaining parts, a preposition is in Priscian’s de nition ‘unin ected’ and ‘is placed before
6
other parts either in juxtaposition or in composition’.

p. 85 An adverb, for which we have again a detailed study by Apollonius, is likewise ‘unin ected’ and ‘is
7
predicated of the moods of verbs, wholly or partly, without which it does not conclude a thought’. In
8
Priscian’s more summary de nition, it is an unin ected part ‘whose signi cation is added to verbs’.

A conjunction, in Priscian’s de nition, is an unin ected part ‘conjoining other parts of an utterance,
9
together with which it signi es, making force or order clear’.

1. onomá esti méros lógou ptōtikón, hekástōi tôn hupokeiménōn sōmátōn ē pragmátōn koinē´n ē´ idían poiótēta
aponémon (ascribed to the school of Apollonius and Herodian, GG 1.3: 524, ll. 9–10); quae unicuique subiectorum
corporum seu rerum communem vel propriam qualitatem distribuit (GL 2: 56–7).
2. cum temporibus et modis, sine casu, agendi vel patiendi significativum (GL 2: 369).
3. quae pro verbo accipitur, ex quo et derivatur naturaliter, genus et casum habens ad similitudinem nominis et
accidentia verbo absque discretione personarum et modorum (GL 2: 552, ll. 18–20).
4. léxin antʼ onómatos prosō´pōn hōrisménōn parastatikē´n, diáphoron katà tē´n ptôsin kaì arithmón, hóte kaì
génous estì katà tē`n phonē`n aparémphatos (GG 2.1: 9, ll. 11–13).
5. quae pro nomine proprio accipitur uniuscuiusque personasque finitas recipit (GL 2: 577).
6. indeclinabilis, quae proponitur aliis partibus vel appositione vel compositione (GL 3: 24, ll. 13–14).
7. léxis áklitos, katēgoroûsa tôn en toîs rhē´masin enklíseōn kathólou ē` merikôs, hôn áneu ou katakleísei diánoian
(GG 2.1: 119, ll. 6–7).
8. pars orationis indeclinabilis, cuius significatio verbis adicitur (GL 3: 60).
9. indeclinabilis, coniunctiva aliarum partium orationis, quibus consignificat, vim vel ordinationem demonstrans (GL 3:
93).
 Box 5.2 The parts as defined by Donatus and by ʻDionysius Thraxʼ

A noun is de ned by ‘Dionysius’ as a part of an utterance ‘subject to case which signi es a material or
1
non-material entity both in a strict sense and a common’. The name Sōkrátēs, for example, signi es
2
in one way; ánthrōpos ‘man’ in the other. The de nition of Donatus corresponds exactly.

p. 87 A verb is de ned by ‘Dionysius’ as a part of an utterance ‘without case, admitting times and persons
3
and numbers, representing an activity or an experience’. As de ned in Latin by Donatus, a verbum was

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
a part ‘with time and person, without case, signifying either doing something or experiencing it or
4
neither’.

These head the list in the Greek order. A participle, which followed, is de ned by ‘Dionysius’ simply as
5
a word which ‘shares the speci c character of verbs and of nouns’. It is ‘so called,’ as Donatus
explains, ‘because it takes a part of the noun and a part of the verb’. From the verb alone it ‘receives
6
genders and cases,’ from the verb alone, ‘times and signi cations’.

In the Roman order the noun was followed by the pronoun. In the de nition of ‘Dionysius’, which is
7
one of the simplest, it is a word ‘employed in place of a noun, making clear de nite persons’. The
Latin term, pronomen, was likewise transparent and applied to some words that did not distinguish
persons. As de ned by Donatus, it is a part ‘placed instead of a noun’, which ‘signi es barely as much’;
8
and, he adds, ‘sometimes admits person’.

The article, which follows the participle in the Greek order, is de ned by Dionysius as ‘subject to case’
9
and ‘ordered before and after the in ection of nouns’. Articles ‘ordered after’ are again ones now
described as relative pronouns.

The remaining parts are those not distinguished by in ections. The adverb follows the verb in the
Roman order and is de ned by Donatus as a part which ‘added to a verb makes clear and lls in its
10 11
signi cation’. In the de nition of ‘Dionysius’ it is simply ‘said of or in addition to a verb’.

p. 88 A preposition, according to ‘Dionysius’, is ‘placed before all other parts of the utterance in both
12
composition and syntax’. As de ned by Donatus, it is ‘placed before other parts of an utterance and
13
either lls out or changes or reduces their signi cation’.

A conjunction, which is the last in the Greek order, is a part that for ‘Dionysius’ ‘binds a thought in an
14
ordered way and makes clear a gap in interpretation’; in the formula of Donatus, it ‘ties together and
15
orders a thought’.

An interjection, nally, is added in the Roman order and de ned by Donatus both by its syntax and an
emotive meaning, as a part ‘added among other parts of an utterance to express feelings of the
16
mind’.

1. méros lógou ptōtikón, sôma ē prágma sēmaînon … koinôs te kaì idíōs legómenon (GG 1.1: 24).
2. pars orationis cum casu corpus aut rem proprie communiterve significans (GL 4: 373).
3. léxis áptōtos, eidektikē` khrónōn te kaì prosō´pōn kaì arithmôn, enérgeian ē` páthos paristâsa (GG 1.1: 46).
4. pars orationis cum tempore et persona sine casu aut agere aliquid aut pati aut neutrum significans (GL 4: 381).
5. léxis metékhousa tês tôn rhēmátōn kaì tês tôn onomátōn idíotētos (GG 1.1: 60).
6. dicta quod partem capiat nominis partemque verbi. recipit enim a nomine genera et casus, a verbo tempora et
significationes (GL 4: 387).
7. léxis antì onómatos paralambanoménē, prosō´pōn hōrisménōn dēlōtikē´ (GG 1.1: 63).
8. quae pro nomine positum tantundem paene significat personamque interdum recipit (GL 4: 379).
9. méros lógou ptōtikón, protassómenon kaì hupotassómenon tês klíseōs tôn onomátōn (GG 1.1: 61); for the text
compare commentators (GG 1.3: 256 and elsewhere).
p. 89 10. quae adiecta verbo significationem eius explanat atque implet (GL 4: 385).
11. katà rhēmatos legómenon ē` epilegómenon rhē´mati (GG 1.1: 72).
12. protitheménē pántōn tôn toû lógou merôn én te sunthései kaì suntáxei (GG 1.1: 70).
13. proposita aliis partibus orationis significationem earum aut conplet aut mutat aut minuit (GL 4: 389).
14. sundéousa diánoian metà táxeōs kaì tò tês hermeneías kekhēnós dēloûsa (GG 1.1: 86).
15. adnectens ordinansque sententiam (GL 4: 388).
16. interiecta aliis partibus orationis ad exprimendos animi adfectus (GL 4: 391).

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
A de nition was nevertheless a de nition, and where distinctions are exact and thorough the contribution
of Apollonius Dyscolus seems to have been crucial. His style is never easy, and while studies of some parts
survive, much has been lost. But it was Apollonius who established de nitively what was, in particular, a
verb. Predecessors, unnamed, had classed in nitives as adverbs; so to, as we will note in Chapter 10, a pair
of verbs with the meaning ‘should’ or ‘ought to’, which again did not distinguish persons. The problem,
however, was not simply to distinguish categories, as we would now say, extensionally. Which forms were
assigned to which, in which uses, should be determined, in the concept of science implicit in antiquity, by
criteria as clear as possible, which individual forms either met or did not meet. In de ning adverbs, for
example, Apollonius begins by saying that they are indeclinable. This is not a mere descriptive observation;
without it their intension, if we may use twentieth-century terminology, was not clear. For the same form,
as we have seen in the last chapter, could be an adverb in one utterance, but a noun, for example, with the
in ection of a noun, in others.

Adjectives

One major di erence between the ancient system and modern concepts of the ‘parts of speech’ follows, as
we have seen, from the modern focus on lexemes. Another particular di erence, in addition to those
discussed already, is that adjectives were seen in antiquity as no more than a subclass of nouns. In a modern
analysis of Greek or Latin, as of most other languages, nouns and adjectives are instead assigned to di erent
major categories.

p. 90

was a noun of a type distinguished in Greek as epithetikon: as having the property of being ‘placed next to’.
This is one of twenty-four subtypes of prosēgoriai, or what are later ‘common nouns’, listed in the manual
of ‘Dionysius Thrax’. In Latin, bonus in:

was correspondingly a noun ‘added to’ (adiectivum), distinguished at the same level as many other
subclasses by, for example, Donatus. A syntactic relation like that of bonus to homō was obvious and invoked
for comparison in accounts of adverbs, to clarify their relation to verbs. Priscian, following Apollonius,
points to the parallel with:
 (GL 3: 60). The property, however, speci c to nouns was for Priscian that of signifying substantia ‘being’ and
qualitas ‘what-sort-ness’, and both bonus ‘good’ and homō ‘person’ had it.

To a modern typologist it is clear that Latin ‘had’, as we may put it, a major category of adjectives. For many
linguists, this is now a term in a universal system that constrains all languages that people can speak. In
arguing, however, from that assumption we introduce preoccupations that belong to later periods in the
p. 91 history of our subject. It was a mediaeval Englishman, not an ancient Greek philosopher, who made the
much-trumpeted pronouncement that ‘grammar is substantially the same in all languages, even though it
may vary accidentally’ (trans. Lyons 1968: 15–16). Other languages were spoken within the Roman empire

Downloaded from https://academic.oup.com/book/36516/chapter/321256238 by University of Ghana. Balme Library user on 08 May 2024
as, before they were conquered, in the Greek-speaking states in the east. Punic, Egyptian, and others were
not Indo-European. But they were of interest to ancient scholars mainly in that etymologists, in the sense of
Box 1.1, might appeal to them.

In both Greek and Latin, the forms called adjectives entered into patterns of in ection similar to those of
nouns in general. Bonus, for example, is distinguished from bonum in:

as dominu-s, a nominative singular meaning ‘master of a household’ is from a corresponding accusative

dominu-m. Bon-u-m is in turn distinguished from bon-a-m, as in:

as domin-a-m, an accusative meaning ‘mistress of a household’, is from domin-u-m, and so on. Other
adjectives had endings like the em of hominem or mulierem, but their meanings and their relations to other
units are the same.

The words called adjectives included simple forms like bonus plus comparatives and superlatives. While this,
however, distinguished them from nouns of other subtypes, the di erence in their range of in ections is
p. 92 less than among subtypes of verb, for example, that we may distinguish lexically as having or not having
passives. Apart, too, from the uses illustrated, their relations to other forms within an utterance were like
those of other forms distinguished as ‘with case’. Thus, in particular, an adjective in Greek could combine
with an article:

as straightforwardly as a noun such as ánthrōpoi ‘human beings’. In an analysis restricted to two of the older
Indo-European languages, by the criteria implicit in ancient grammars, it would have been wrong to see
this as a category separate at the highest level.

What Graeco-Roman Grammar Was About. First edition. P. H. Matthews. © P. H. Matthews 2019. First published 2019 by Oxford
University Press.