Вы находитесь на странице: 1из 27

Histoire Épistémologie Langage

Latin declensions and conjugations: from Varro to Priscian


Daniel J. Taylor

Citer ce document / Cite this document :

Taylor Daniel J. Latin declensions and conjugations: from Varro to Priscian. In: Histoire Épistémologie Langage, tome 13,
fascicule 2, 1991. Théories et données. pp. 85-109;

doi : https://doi.org/10.3406/hel.1991.2334

https://www.persee.fr/doc/hel_0750-8069_1991_num_13_2_2334

Fichier pdf généré le 17/01/2019


Abstract
ABSTRACT : Declensions and conjugations are easily the most recognizable components of Latin
grammar, but such was not always the case. Moreover, their evolution in Roman language science
contains a number of significant discrepancies in enumeration, linguistic analysis, and terminology. In
other words, declensions and conjugations do have a history, and this paper surveys that history.
Thanks to the efforts of the many grammarians and scholars who collectively comprise the Roman ars
grammatica, which culminates in Priscian's monumental tome, the embryonic declensions and
conjugations first formulated by Varro are transformed into the declensions and conjugations with
which we are so familiar today.

Résumé
RESUME : Les déclinaisons et les conjugaisons sont les composantes les plus aisément
reconnaissables de la grammaire latine, mais ce ne fut pas toujours le cas. En outre, leur évolution,
dans la science linguistique romaine, montre un certain nombre de divergences, concernant
l'énumération qui en est faite, l'analyse linguistique proposée, ou la terminologie. En d'autres termes,
les déclinaisons et conjugaisons ont bien une histoire, et cet article se propose de la décrire en ses
grandes lignes. Grâce aux efforts des nombreux grammairiens et chercheurs qui constituent Yars
grammatica prise comme un tout, et qui culmine dans l'ouvrage monumental de Priscien, la forme
embryonaire des déclinaisons et conjugaisons d'abord proposée par Varron se trouvera transformée
en celle avec laquelle nous sommes si familiers aujourd'hui.
Histoire Épistémologie Langage 13/11 (1991)

LATIN DECLENSIONS AND CONJUGATIONS :


FROM VARRO TO PRISCIAN

Daniel]. TAYLOR

ABSTRACT : Declensions and conjugations are easily the most


recognizable components of Latin grammar, but such was not always the
case. Moreover, their evolution in Roman language science contains a
number of significant discrepancies in enumeration, linguistic analysis, and
terminology. In other words, declensions and conjugations do have a
history, and this paper surveys that history. Thanks to the efforts of the
many grammarians and scholars who collectively comprise the Roman ars
grammatica, which culminates in Priscian's monumental tome, the
embryonic declensions and conjugations first formulated by Varro are
transformed into the declensions and conjugations with which we are so
familiar today.

RESUME : Les déclinaisons et les conjugaisons sont les composantes les


plus aisément reconnaissables de la grammaire latine, mais ce ne fut pas
toujours le cas. En outre, leur évolution, dans la science linguistique
romaine, montre un certain nombre de divergences, concernant
l'énumération qui en est faite, l'analyse linguistique proposée, ou la
terminologie. En d'autres termes, les déclinaisons et conjugaisons ont bien
une histoire, et cet article se propose de la décrire en ses grandes lignes.
Grâce aux efforts des nombreux grammairiens et chercheurs qui constituent
Yars grammatica prise comme un tout, et qui culmine dans l'ouvrage
monumental de Priscien, la forme embryonaire des déclinaisons et
conjugaisons d'abord proposée par Varron se trouvera transformée en celle
avec laquelle nous sommes si familiers aujourd'hui.

Prefatory remarks. Latin declensions and conjugations are


unavoidably familiar to all who study and teach the language, but
their origin and historical development are not. Since declensions
and conjugations are easily the most recognizable components of
modern Latin grammar and since the former have no antecedents in
86 Daniel J. Taylor

Greek language science and the latter only marginal ones (cf. Taylor
1990), we ought to know more about how they came into being and
evolved in classical Roman antiquity.

Given the inflected nature of Latin, it is easy to understand


why declensions and conjugations are the heart and soul of Latin
grammar and the key to describing and teaching the language. Such
was not always the case, however. The earliest extant text to arrange
nouns and verbs into what we nowadays call declensions and
conjugations is Varro's (116-27 B.C.) De Lingua Latina, but his
scheme is not quite the same as ours or even his successors'. In
Priscian's (late 5th-early 6th c. A.D.) Institutiones Grammaticae,
however, the declensions and conjugations appear much the same as
they do in a modern textbook or grammar. In other words,
declensions and conjugations in Latin do have a history, and we
ought to be able to determine that history by studying how the
grammarians subsequent to Varro and prior to Priscian describe
Latin declensions and conjugations.

What follows is therefore a survey of that history. I have


selected and introduced what seem to be the most salient passages
from the extant grammatical literature and have commented on those
features of the texts which are directly relevant to the topic. Since
the conjugations are less complicated, I begin with them and move
on to declensions ; most conclusions and questions are scattered
throughout the commentary and notes, although a few others are
adduced in the concluding section.

1. Conjugations

In De Lingua Latina IX. 109 Varro describes what are, de


facto, the first conjugations in Latin grammar. Varro's conjugations
are embryonic only : that is to say, his presentation is correct in
theory but incomplete~and to that extent, arguably, incorrect~in
practice. The manuscript tradition has corrupted the endings which
he enumerates, but the examples he adduces allow us to reconstruct
the text adequately (cf. Taylor 1977 : 130-31).

I. Varro, LL IX. 109. Thus in the remainder of the paradigm of the


verbs each follows its own pattern. It is important for the purpose of
classifying linguistic similarities to notice whether a given verb form
Latin declensions and conjugations 87

has ay or is or es in its final syllable in the 2nd sg. Therefore the


indicator of analogy is in the second person rather than in the first,
because in the latter form the dissimilarity is covert, as the following
examples attest : meo, neo, ruo ; for different forms emanate from
them as a result of inflection because they are patterned as follows :
meo meas, neo ties, ruo ruis, each of which maintains its own
configuration of linguistic likeness. '

Varro bases his conjugations on the 2 sg. act. prog. ind. près, form,
a strategy that remains constant throughout the tradition (see below),
but he ignores vowel length and thereby conflates what we term the
third and fourth conjugations, with their short and long l\l
respectively. He has, however, successfully distinguished the first,
second, and third conjugations. The task for the grammatical
tradition would therefore appear to be rather straightforward— to
identify the fourth conjugation and ipso facto separate it from the
third by focusing on the quantity of the vowel in the ending— and
that is precisely what happens, though in a somewhat indirect
manner and only in due time. 2
First, however, Varro's successors must recognize the
significance of his discovery, and that does not seem to have

1. ltaque in reliqua forma verborum suam utr <um> que sequitur formam. Utrum
in secunda forma verb[orjum temporale habeat in extrema syllaba as an is
a<u>t <e>s[i], ad discernendas similitudines interest. Quocirca ibi potius
index analogiae quam in prima, quod ibi abstrusa est dissimilitudo , ut apparet in
his, meo, neo, ruo ; ab his enim dissimilia fiunt transitu, quod sic dicuntur meo
meas, neo nes, ruo ruis, quorum unumquodque suam conservât similitudinis
formam.
2. Whether Varro himself contributes to that endeavor is not ours to know. In LL
X.33, after articulating his theory of verbs, he promises to offer more de
formulis verborum [on the paradigms of verbs], but that discussion is not extant.
Yet an off-hand remark by Diomedes suggests that Varro may somewhere have
improved upon his embryonic account of conjugations in LL IX. 109. That text,
as edited by Funaioli (1907 : 277), reads as follows :
haec de quattuor coniugationibus quae pertinent ad verba quae analogiae
parent, quorum exempla passim perscripta sunt et sunt nota, quae si quis
conceperit animo, non facile labetur ; sunt enim evidenter exposita et Varroni
Menippeo [So much for the four conjugations which are a property of verbs that
are subject to analogy, examples of which have been duly recorded and
identified here and there. Whoever has acquired an understanding of these
conjugational issues will not easily make a mistake ; after all, they have been
clearly set out even by Varro, the author of Menippean satire].
In Keil's opinion (GL 1.371), however, those three final and crucial words are
interpolations, and so the text can only be considered suggestive at best, though I
would note that the interpolation, if such it is, is right on the mark.
88 Daniel J. Taylor

transpired automatically. Aufustius (Funaioli 1907 :492) attends to


verbs in -io, but he has no conjugational label for them. In his brief
discussion oi fervere (I.O. I.vi.7-9) Quintilian likewise has nothing
to say about conjugational affiliation even though it would be the
telling point in his argument were he to do so, and his silence may
be instructive. After footnoting, as it were, first Didymus and then
Probus, Priscian (GL 11.445-47 = Mazzarino 1955:107-09)
compares selected Greek and Latin verbs with reference to their
conjugational affiliations, but the one specific reference to Latin
conjugations is clearly Priscian's. Otherwise, however, the hundreds
of grammatical fragments from the early imperial period have little
or nothing to say about the elaboration of Varro's classificatory
schema. The absence of evidence referring to Palaemon's ars is
particularly vexing, for what we learn about it from Juvenal and can
infer about it from Quintilian suggests that its overall organization
becomes the norm for subsequent artes. But the claim that Palaemon
is responsible for the proper classification of (declensions and)
conjugations seems completely unsubstantiated. Nonetheless, the
first and second centuries A.D. are obviously a period of intense
grammatical activity, and it is highly likely that this is the period in
which conjugational affiliations become established, despite the lack
of direct evidence to that effect. 3 At any rate the conjugations
appear in full bloom, so to speak, in Sacerdos' grammar (3rd c.
A.D.), and they obviously did not arise ex nihilo.
Sacerdos (GL VI. 433-35) begins : Coniugationes verborum
sunt très [there are three conjugations of verbs] ; he then enumerates
prima, secunda, tertia correpta, and tertia producta [first, second,

3. In this regard the name of Probus, in particular, ought give us some pause. The
immense problems of prosopography, plagiarism, chronology, et al. which
plague our efforts to make sense of the development of the Roman ars
grammatica preclude certainty, but some tidbits of evidence point to a more
prominent role on Probus' part than we have heretofore envisioned. Priscian
refers to him frequently, for example, when dealing with verbs, and Servius (GL
IV. 413), after citing the rules for conjugations, says : quas régulas Probus
artifex tuetur [Probus skillfully attends to such rules]. Moreover, the grammatical
treatises spuriously attributed to him contain some rather primitive features
which may somehow be reminiscent of the earliest attempts to flesh out the
(declensions and) conjugations. Under the circumstances, Probus' specific
contributions, if any, to the development of (declensions and) conjugations are
difficult, perhaps even impossible, to identify, but the remarks of Priscian and
Servius, if nothing else, suggest that we would like to know more about his
version of Roman language science.
Latin declensions and conjugations 89

third short, and third long], 4 identifying each on the basis of its 2nd
sg. verb form, just as Varro had originally done. Later he refers to
tertia producta, quam quidam quartam dicunt [the third long, which
some call the fourth]. The two competing systems of nomenclature
should be understood chronologically : that is to say, the system of
three conjugations with its third short and long precedes the system
of four conjugations. Moreover, the difference between the two
systems is more than merely a matter of terminology (see below),
but the tripartite system dominates late antique grammar. Diomedes,
for example, uses it throughout his treatment of the verb, only twice
referring to the third long as the one quae a quibusdam quarta
dicitur [which is called by some the fourth] and quam quidam
quartam nominant [which some name the fourth] (GL 1.364 & 370
respectively) and only once mentioning four conjugations (see the
text in note 2). Ditto for Donatus, who only once (GL IV. 3 82) says
hanc non nulli quartam coniugationem putant [several consider this
a fourth conjugation]. Nowhere, it seems, are the quidam [some] or
non nulli [several] ever identified by name, but it should be obvious
that the quadripartite system is an elaboration of sorts of the
tripartite one.

Charisius inherits both systems, and so he presents both


systems.

II. GL 1.168-169. The conjugational affiliations which categorize


verbs are four. A verb whose 2nd sg. ends in the letters as, like amo
amas, is a member of the first conjugation. A verb whose 2nd sg.
ends in es, like teneo tenes, is a member of the second conjugation.
A verb whose 2nd sg. ends in is with the vowel short, like ago agis,
is a member of the third conjugation. A verb whose 2nd sg. ends in
is with the vowel long, like munio munis, is a member of the fourth
conjugation. Consequently all verbs belonging to the same

4. The attention which Sacerdos accords vocalic length here and in text VI below
(cf. note 10 also) and which the apocryphal Probus also does in text XII makes
sense historically as well as scientifically, for the study of metrics becomes a
more integral component of Roman grammar during the 2nd c. A. D., as
witnesses the publication of Terentianus Maurus' manual on metrics, published
at the close of that century. I am indebted to Even Hovdhaugen for this
observation.
90 Daniel J. Taylor

conjugation will likewise manifest a similar inflection throughout


their paradigm. . . . 5

III. GL 1.175-176. Comianus, an accomplished grammarian, has


analyzed conjugations, which we referred to as ordines in the
foregoing discussion, as follows. Conjugations, which the Greeks
call suzugiae, are in our language three in number : first, second,
third. ...Verbs of the first conjugation end in as, with the vowel
long, in the 2nd sg. près, ind., e.g., amo amas, canto cantos.
Second conjugation verbs end in es, with the vowel long, like video
vides, moneo mones, sedeo sedes. Verbs of the third conjugation end
in is, with the vowel sometimes short and sometimes long, short as
in lego legis, long as in audio audis, nutrio nutris. 6

Although he offers both systems (and disciples of Barwick 1922 will


note the presence of velut in the former but not the latter text),
Charisius apparently prefers the former or more modern of the two :
in his verbis quae quarti ordinis esse definimus [in these verbs
which we classify as fourth conjugation] (id. 171). And indeed it is
the quadripartite system which ultimately prevails, as we can see in
Priscian, who does not even bother to mention its competitor.

IV. GL 11.442-43. Speakers of Latin therefore have four


conjugations at their disposal all verbs end their 2nd. sg. in as
if they are a member of the first conjugation, like oro oras, sto stas ;
if a member of the second, they end in es, as moneo mones, haereo
haeres ; if third, in is with a short vowel, like cupio cupis, curro

5. Ordines verborum sunt quattuor, qui verba dispertiunt. primi ordinis est verbum
cuius secunda persona as litteris terminatur, velut amo amas, secundi ordinis est
verbum cuius secunda persona es terminatur, velut teneo tenes. tertii ordinis est
verbum cuius secunda persona per is correptam terminatur, ut ago agis, quarti
ordinis est verbum cuius secunda persona productis is litteris terminatur, velut
munio munis, itaque omnia verba quae eiusdem ordinis erunt similem etiam
declinationem habent. . . .
6. De coniugationibus , quas nos ordines praediximus , Cominianus disertissimus
grammaticus ita disseruit. coniugationes , quas Graeci suzugias appellant, sunt
apud nos très, prima secunda tertia. ...primae coniugationis verba indicativo
modo tempore praesenti persona secunda as litteris productis tertninantur , ut
amo amas, canto cantos, secundae coniugationis verba... es litteris productis
terminantur, ut video vides, moneo mones, sedeo sedes. tertiae coniugationis
verba. . . is litteris interdum correpte interdum producte terminantur ; correpte, ut
lego legis, producte, ut audio audis, nutrio nutris.
Latin declensions and conjugations 91

curris ; if fourth, in is with a long vowel, as munio munis, esurio


esuris. 7

So the process which Varro starts with his theoretical breakthrough


ultimately culminates in the conjugational system which we still use.
If we go back and ask how the tradition actually resolved the
problem of conflation which Varro had occasioned by ignoring
vowel length in the -is ending, the obvious answer is that it simply
took vocalic length into account and proceeded accordingly. After
all, the nomenclature, i.e., tertia correpta et producta [third short
and long], would seem to be definitive, and as long as vocalic length
is taken into consideration, the problem is indeed resolved. Such an
answer, however, though it may not be incorrect, may not be
entirely correct either ; that is to say, it is in all likelihood
incomplete, for other factors seem to have been involved. Those
factors are at least four : the future, orthography, the imperative,
and the infinitive.
In citing the conjugations any number of grammarians proceed
directly from, e.g., the present amo amas to the future amabo, and
an observation by [Sergius] may be instructive in this regard. After
noting that the tertia coniugatio...interdum producitur, interdum
corripitur [the third conjugation is sometimes lengthened, sometimes
shortened], he says :

V. GL IV. 506. There is this difference between the long and the
short : the short always terminates its future tense with am in the 1st
sg., e.g., dicam and scribam ; but the long employs both am, as in
audiam and nutriam, and bo, as in the well known quotation from
Terence's Hecyra [line 495] « matris servibo commodis » [I shall
devote myself to my mother's interests].-8

And it is a fact that eo and its compounds with their futures in -bo
receive special attention from the grammatici Latini. So the

7. sunt igitur coniugationes quattuor apiid Latinos omnia verba...si primae


sint coniugationis , in 'as ' efferunt secundam personam, ut 'oro oras ', 'sto stas ' ;
sin secundae, in 'es ', ut 'moneo mones ', haereo haeres ' ; sin tertiae, in 'is '
correptam, ut 'cupio cupis ', 'curro curris ' ; sin quartae, in 'is ' productam, ut
'munio munis ', 'esurio esuris '.
8. inter productam et correptam hoc interest, quod correpta futur urn tempus semper
in am mittit, ut dicam scribam ; producta vero et in am mittit, ut audiam nutriam,
et in bo, ut 'matris servibo commodis ' : Terentius in Hecyra.
92 Daniel J. Taylor

formation of the future may have played a role in distinguishing the


third and fourth conjugations, to use the modern terminology.
Sacerdos testifies eloquently to the twin role of orthography
and the imperative in disambiguating the two conjugational
affiliations at issue.

VI. GL VI.434. Some think that the conjugational affiliation of


every verb can be ascertained on the basis of its ending in the 2nd
sg. près. ind. But they are wrong. This principle of theirs will be
valid in the case of the first and second conjugations, for the first
ends in as, e.g., amas, and the second in es, with the vowel long, as
doces. But the third short and the third long, which latter some call
the fourth, can not thereby be confirmed as separate conjugations :
for each ends in is, e.g., scribis and munis. Therefore, so that the
most definitive principle may be adduced in every conjugation, the
2nd sg. of the present imperative must be taken into consideration.
That imperative ends in a in the first conjugation, e.g, ama, and in
long e in the second, e.g., doce ; but in the third short conjugation it
ends with a short e, as in scribe, whereas in the third long, the
ending is an i, as in muni. 9

The 2nd sg. verb forms in the two conjugations are spelled the
same, regardless of phonology, but the imperatives differ in both
respects. 10
Even though we nowadays determine conjugational affiliation
on the basis of the infinitive, the role of the infinitive in the
historical process of sorting out the conjugations is less certain. It
would appear that only a few statements attest to its potential

9. Quidam putant de secunda persona modi indicativi temporis praesentis omne


verbum cuius sit coniugationis posse cognosci. sed errant, na/n poleril .haec
eorum ratio in prima et secunda coniugatione constare : nam prima as, ut amas,
secunda es producta syllaba terminatur, ut doces. terlia vero correpta et tertia
producta, quam quidam quartam dicunt, non poterunt separatae firmari : nam
utraque is syllaba terminatur , ut scribis munis, ergo ut in omni coniugatione
firmissima ratio tradatur, consideranda est modi imperativi numeri singularis
secunda persona, nam primae coniugationis modi imperativi temporis praesentis
numeri singularis secunda persona a terminatur, ut ama, secundae e producta,
ut doce, tertiae vero correptae e correpta, ut scribe, tertiae productae i, ut muni.
10. Sacerdos seems oblivious to the fact that his analysis of imperatives, while
resolving the conflation of the third and fourth conjugations, produces exactly
the same sort of ambiguity with regard to the second and third conjugations, for
doce and scribe differ only in the length of the final vowel. His overall logic
may be faulty, but his argument does direct our attention to the role of
orthography and the imperative in the matter at hand.
Latin declensions and conjugations 93

significance. Charisius notes merely in passing (GL 1.172) : cuius


autem ordinis sit verbum facile intellegi potent ex infinitivorwn
observatione [the conjugational affiliation of verbs can be readily
ascertained by observing their infinitives], and Donatus {GL IV.
359 ; cf. Servius, id. 413) observes : haec in imperativo et in
infinitivo statim discerni possunt, utrum i littera correpta sit an
producta [these verbs can be immediately identified by determining
whether the letter i in their imperative and infinitive is short or
long]. Neither elaborates. It therefore seems that several factors, not
just vocalic length in the 2nd sg. verb form, played a role in
resolving the problem Varro caused. The tradition opted for the
locution tertia correpta et producta [third short and long] because it
was the simplest under the circumstances, not because it was
necessarily more definitive than any of the other criteria adduced
here or there.
In fine, the Latin conjugations have a lengthy history, which
is relatively clear in outline, albeit somewhat unclear in all its
details and especially so in its earlier stages, and which is
cumulative, though not strictly uniform, in nature. The irony of it
all is that this history would not have occurred had Varro only taken
vocalic length into account.

2. Declensions

In LL X.62, after articulating what is arguably one of the most


sustained discussions of linguistic, specifically, morphological,
theory in ancient language science, Varro describes what are the
first declensions in extant grammatical literature, Greek or Roman.
As was the case with his conjugations, Varro's declensions are
embryonic only, he plays fast and loose with vocalic length, and he
does not elaborate. Nonetheless the declensions are all there in
theory.

VII. LL X.62. But if someone does in fact wish to set out from the
singular, he would be well advised to start from the sixth case [sc.
the ablative], which is unique to Latin, for the several different
morphemes of this one case will make it easier for him to classify the
combinations and permutations of the remaining cases. The endings
at issue are : A as in terra, E as in lance, I as in levi, O as in caelo,
94 Daniel J. Taylor

and U as in versu. The method whereby we establish declensional


paradigms can therefore proceed in two directions. "

Varro has therefore set the agenda re : declensions for his


successors in the ars grammatica. That agenda includes three major
tasks, two of which are designed to « correct » Varro and a third
which the tradition adopts unilaterally : 1) consonant-stems and i-
stems, which, with their ablatives in short Id and long I'll
respectively, Varro identifies separately, must be combined into the
third declension, to anticipate the nomenclature ; 2) the fifth
declension, which, with its ablative in long Id, Varro fails to
identify because he ignores vowel length, must be created, or
discovered if you will ; and 3) the ending in the genitive singular,
rather than that in the ablative singular, becomes the basis for
distinguishing the declensions, a strategy which further entails a
switch in the enumerated order of the declensions, i.e., from
Varro's a, e, i, o, u order to ae, i, is, us, ei. That the tradition
accomplishes these tasks successfully can be clearly seen in Priscian.

VIII. GL 11.331. The ablative of the third declension ends in short e


in some words, in / in others, and in yet some others in both e and
i.12

IX. GL III. 446. But Latin nouns, if they have an / before the es in
the nominative, are members of the fifth declension, as witness the
following examples : nom. fades, gen. faciei ; nom. series, gen.
seriei ; nom, dies, gen. diei. 13

X. GL III. 443. All the nouns which the Latin language utilizes are
inflected in five declensions, which are enumerated in accordance
with the order of the vowels forming their genitives. The first
declension is therefore the one whose genitive ends in the diphthong
ae, as nom. poeta, gen. poetae ; the second is that in which the

1 1 . Sin ab singulari quis potius proficisci volet, initium facer e oportebit ab sexto
casu, qui est proprius Latinos : nam eius casu[sh]is litterarum discriminibus
facilius reliquorum varietate <m> discernere poterit, quod ei habent exitus aut
in A ut hac terra, aut in E ut hac lance, aut in I ut hac levi, aut in O ut hoc
caelo, aut in U ut hoc versu. Igitur ad demonstrandas declinationes biceps via
haec.
12. Ablativus tertiae declinationis in aliis per e correptam, in aliis per i, in aliis et
per e et per i profertur.
13. Latina vero, si ante es i habuerint, quintae declinationis, ut haec fades huius
faciei, haec series huius seriei, haec dies huius diei.
Latin declensions and conjugations 95

aforementioned case ends in /, as nom, doctus, gen. docti. The third


ends in is with a short vowel, as nom. pater, gen. patris ; the fourth
ends in us with the vowel long, as nom. senatus, gen. senatus ; the
fifth ends in ei, which is pronounced as two syllables, as nom.
meridies, gen. meridiei. 14

The differences between the Varronian and the Priscianic analyses


encapsulate, as it were, the history of the declensions in the Roman
ars, but not completely, for two features peculiar to Varro's
approach continue to figure prominently in the later tradition and to
impact considerably upon it, namely, the role of the ablative
singular ending (and the alphabetical order attendant thereupon) and
the locutions by which he refers to the ablative case.
As Diomedes (GL 1.302) informs us, ablativum Graeci non
habent. hunc tamen Varro sextum, interdum Latinum appellat, quia
Latinae linguae proprius est... [the Greeks do not have an ablative
case. Varro on the other hand calls it the sixth, sometimes the Latin,
case, because it is unique to the Latin language]. It would therefore
appear that Varro is responsible for using the adjectives sextus
[sixth], Latinus [Latin], and proprius [unique] to describe the
ablative. We can find similar statements in Donatus {GL IV. 377),
his commentator Pompeius (GL V.181), Consentius (/J.351), even
in Priscian (GL 11.187 : igitur ablativus proprius est Romanorum....
[the ablative is therefore uniquely characteristic of the Romans]),
and undoubtedly elsewhere. Even without the footnote, so to speak,
to Varro in Diomedes (and likewise in Consentius) the verbal echoes
ring loud and clear, especially that of proprius. The persistence of
the terminology throughout the tradition not only testifies to the
implicit continuing presence of Varro in that tradition but also
provides continuing verbal reminiscence of two competing systems
of analysis (see below). The ablative may ultimately surrender its
premier analytical role to the genitive but never its unique
terminological identity.

14. Omnia nomina, quibus Latina utitur eloquenlia, quinque declinationibus


flectuntur, quae ordinem acceperunt ab ordine vocalium formantium genetivos.
prima igitur declinatio est, cuius genetivus in ae diphlhongon desinit, ut hie
poeta huius poetae ; secunda, cuius in i productam supra dictus finitur casus,
ut hie doctus huius docti. tertia in is brevem, ut hie pater huius patris ; quarta
in us productam, ut hie senatus huius senatus ; quinta in ei divisas syllabas, ut
hie meridies huius meridiei.
96 Daniel J. Taylor

The tradition also ascribes the discovery of the ablative's role


in determining nominal inflection, at least in part, to Varro, as Pliny
attests :

XI. (Funaioli 1907 :275 = Mazzarino 1955 : 27 8-79) According to


Pliny, « Varro's rule specifies that unusual words like toreumata,
enthymemata , noemata, schemata, poemata, and all words similar to
these must end in bus in the dative and ablative plural because their
ablative singular ends with the letter e. » l5

As time passes, so too does the tradition's memory, and the


Varronis régula [rule of Varro] becomes simply the régula ablativi
[rule of the ablative] ; but we know better. 16 The Varronis régula is
the obvious precursor not only of the régula ablativi and the
widespread attention paid to the ablative on the part of almost all
grammarians but also of the various and assorted catholica (se. de
casu ablativo singulari) [general principles regarding the ablative
singular case] which abound in our texts. Diomedes, for example
(GL I.303ff), after enumerating in only one paragraph his seven
declensions based on the genitive, follows with seven paragraphs on
the ablative which are in effect a consummate elaboration of LL
X.62 and in which our modern declensions are clearly set forth but
in alphabetical order based on the vowel in the ablative singular. In
time the role of the ablative in determining declensional affiliation
was formally usurped by the genitive, but the tradition-Diomedes is
conspicuous but by no means unique—never forgot the significance
of the ablative, though it did constrain and redefine its role (see
below). Sorting out the chronology of the changing role of the
ablative may be difficult, but the persistence of that role readily
testifies both to the acknowledged brilliance of Varro's original
formulation and to his unacknowledged but continuing influence on
the Roman ars grammatica.
As was the case with verbs, the grammarians of the early
empire seem not to have realized the practical importance of what

15. 'glossemata ut toreumata enthymemata noemata schemata poemata et his


similia omnia Varronis régula ' inquit Plinitis 'dativo et ablativo plurali in bus
derigit, quia singularis ablativus e litterafiniatur. '
16. Actually, one fragment from Caesar (Funaioli 1907 : 149 #6 [11] = 484, #3)
suggests that the descriptive role of the ablative may have been known or at
least suspected prior to Varro's institutionalization of it in the De Lingua
Latina.
Latin declensions and conjugations 97

Varro had accomplished in theory. Any number of fragments in


Mazzarino (1955) and Funaioli (1907) attest to the grammarians'
inability to classify nouns into declensions, and declensional
classification is likewise conspicuous by its absence in the
grammatical chapters of Quintilian's Institutio Oratoria. On at least
a half- dozen occasions Quintilian has ample opportunity to cite
declensional affiliation as either a or the salient point at issue, but
not once does he do so. Argumenta ex silentio are notoriously
insufficient as evidence, but Quintilian's silence probably speaks
volumes. Pliny too, despite his knowledge of the Varronis régula
(the scope of which already seems to have been reduced) and of the
fact that one case ending presupposes another in the inflection of a
given noun or set of nouns, does not take his knowledge that one
step further. It therefore seems that the elaboration of Varro's
declensional scheme occurs later in the tradition, but it is not clear
when.
The Instituta Artium once ascribed to Probus, in many
respects quite fascinating, is anything but modern in respect to its
analysis of declension. The lengthy discussion of nominal inflection
(GL IV. 75- 130), admittedly entitled de ablativo casu [on the ablative
case], manifests no understanding whatsoever of a genitive-based
analysis and never once mentions declensional affiliation. Instead it
discusses paradigms solely on the basis of the ablative, and it does
so, as we might expect, in alphabetical order. It does, however,
distinguish the long and short Id endings :

XII. ld.%1 . E. An ablative case which in the singular number ends in


e comprises a binary declensional pattern. For there is one which is
understood to result from a long e, but there is another which is
perceived to result from a short e ; and that's why we formulate an
account of each pattern. 17

In other words this text is effectively a logical elaboration of Varro's


original analysis in LL X.62, and it points to that intermediate stage
of the tradition wherein the declensions, based on the ablative, are
all there and have simply not yet been rearranged, reordered, and

17. De e. e litter a ablativus casus numeri singularis terminatus duplicem


dedinationis continet fonnam. una enim est, quae ab e producto venire
intellegitur , alia vero, quae ab e correpto venire reperitur ; et ideo utriusque
formae rationem subicimus.
98 Daniel J. Taylor

numbered on the basis of the genitive. All in due time. Only one
paragraph of Sacerdos' account of nominal inflection in Book I of
his Ars is extant, but it clearly presupposes the ablative singular as
the point of departure for inflecting the oblique plural cases.
Moreover, he (GL VI. 427) distinguishes the long and short Id
endings : quae e finiuntur, ea aut corripientur aut producentur
[those declined forms ending in e will either be shortened or
lengthened]. Book II, however, commences with an extensive
discussion (zY/. 47 1-83) in which the five declensions are enumerated
here and there—correctly, I might add-on the basis of the genitive
singular, but when he lists the genitive singular endings (z<i.483), he
includes only ae, i, s, and u, thereby conflating the second and fifth
declensions as well as, apparently, the third and part of the fourth.
(N.B. The grammarians regularly cite the gen. sg. of 4th declension
neuters as simply u.) Sacerdos would therefore seem to be an
advocate of a genitive- based system (which may perhaps require
some refinement) but one who is not unaware of the role of the
ablative, especially insofar as it 'rules' the genitive and dative-
ablative plurals. Such a position, which may be viewed as a sort of
compromise, ultimately becomes the norm for the ars in the
subsequent centuries.
Charisius, Diomedes, and Donatus provide strikingly different
accounts of declensional affiliation. Charisius' would appear, in
retrospect at least, to be the more standard or more modern account,
but Diomedes' is either representative of the later intermediate stage
or else just plain idiosyncratic (or perhaps both), while Donatus' is
so out of sorts that his commentators not only remedy but even seek
to explain its deficiency of scope. And Consentius seems to be
rethinking many of the issues involved from a fairly enlightened
historiographical perspective. What is so obvious and so remarkable
is simply that each account differs so markedly. The texts can
almost be allowed to speak for themselves.
Charisius puts the issue of declensional enumeration succinctly
enough {GL 1.18) : Or < dines omnium nominum, qui quidem>
ratione et observatione <inveniuntur> , numéro sunt quattuor vel,
ut quibus < dam placet >, quinque [the declensional affiliations of
all nouns, which are in fact identifiable by reason and observation,
are four in number or, in the opinion of some, five]. The genitive
singular is at the basis of his classification, and the declensions are
numbered accordingly ; e.g., id.2\ : Secundae declinationis
Latin declensions and conjugations 99

nominativi sum hi quorum genetivus i facit [nominatives of the


second declension are the ones whose genitive ends in i]. Cf. also
JY/.537 : omnis autem declinatio a genetivo casu cognoscitur
[moreover, every declension is recognizable by its ending in the
genitive case]. After working his way through four declensions, he
concludes :

XIII. Id.3l. There is another declensional set, which some


grammarians claim is affiliated with the second declension since it
forms its genitive with the letter i, but which others have regarded as
a member of the third since it forms its accusative in em and its
dative and ablative plural in bus ; for that reason some others do not
assign it to one of the given declensions on the basis of a particular
case-ending but instead have felt compelled to establish it as a fifth
declension. 18

We are, quite literally, present at the birth, as it were, of the fifth


declension in this text. Given a genitive-based analysis, four
declensions are uniquely and unambiguously defined, but one is
not ; it is therefore constituted as a separate and distinct entity. The
scientific method is working at its best, but we should note that the
role of the genitive in determining declensional affiliation is both
chronologically and logically prior to the establishment of the fifth
declension, at least as Charisius describes the issues, whereas such
was not the case with the long-e declension identified solely on the
basis of the abl. sg. ending in the Instituta Artium. Charisius' text
takes us back to an earlier stage of what we may think of as the
modern period in the history of declensions. The role of the ablative
is not completely forgotten, however, for Charisius devotes an
entire section {id. 147-50) to it ; it has, however, been restricted to
determining the plural inflection rather than the declensional
affiliation : (/d.148) ablativus casus numeri singularis observata
novissima syllaba litterave facile monstrabit quo modo declinare
pluralem numerum debeamus [the ablative singular, once its final
syllable or letter has been observed, will easily reveal how we ought
to decline the plural number]. Charisius has elegantly (re)stated the
Varronis régula to which Pliny had earlier alluded (see text XI

18. Est et alias ordo declinationis, quern alii ad secundum ordinem per liner e
dicunt, quoniam genetivwn in i litteram facit, alii tertii putaverunt, quoniam
accusativum in em, item dativiun et ablativum pluralem in bus facit ; quern ideo
nulli parti tribuentes quintae declinationis dicendum esse putavere.
100 Daniel J. Taylor

above), though his ordering is not strictly alphabetical (see below on


Donatus).
Diomedes treats his declensions rather summarily and only
under the rubric of the cases at that, and his system is based on the
genitive singular.

XIV. GL 1.303. The patterns of nominal declensions, or so it


seems to some grammarians, are seven. The first is that which ends
its genitive singular in ae and includes the masculine as well as the
feminine gender, e.g., masc. Aeneas Aeneae, fem. Latona Latonae.
The second forms its genitive in i in every gender, e.g., masc. puer
pueri, fem, laurus lauri, neut. caelum caeli. The third is that which
employs a double / in the genitive but encompasses only masculine
and neuter genders, e.g., masc. Vergilius Vergilii, neut. ingenium
ingenii. The fourth ends in is in each gender, e.g., masc. orator
oratoris, fem. oratio orationis, neut. sidus sideris. The fifth has a us
for both the masculine and the feminine gender, e.g., masc. portus,
fem. porticus. The sixth ends in ei, which likewise suffices for both
masculine and feminine, e.g., masc. dies diei, fem. acies aciei. The
seventh ends in u and is restricted to the neuter gender, e.g., genu
and cornu ; neuter nouns of this sort have only that one ending in the
singular number. "
Diomedes'
seven declensions are either an anticipation or an
elaboration of the not quite yet canonical five. Diomedes then
immediately launches into his de casu ablativo singulari catholica
which was mentioned earlier and, like Charisius, restricts the role of
the ablative : (id. 303) Ablativus cas us numeri singularis observata
novissima syllaba litterave facile demonstrabit quo modo pluralem
numerum declinare debeamus [the ablative singular, once its final
syllable or letter has been observed, will easily reveal how we ought
to decline the plural number]. There follow six paragraphs, one for
each of the plural declensions as determined by the ending in the

19. Formae declinationum nominum, ut quibusdam videtur, sunt septem. prima est
quae genetivum singularem tnittit in ae genere dumtaxat tarn masculino quatn
feminino, ut Aeneas Aeneae, Latona Latonae. secunda genetivum facit in i in
omni genere, ut puer pueri, laurus lauri, caelum caeli. tertia est quae genetivo i
geminata genere tantum masculino et neutro terminât ur ; masculino ut Vergilius
Vergilii, neutro, ut ingenium ingenii. quarta in omni genere in is, ut orator
oratoris, oratio orationis, sidus sideris. quinta in us masculino dumtaxat genere
et feminino ; masculino, ut portus, feminino, ut porticus. sexta in ei similiter
dumtaxat genere masculino et feminino, ut dies diei, acies aciei. septima in u
genere tantum neutro, ut genu cornu ; et huius modi nomina neutra numéro
singulari tantum monoptota sunt.
Latin declensions and conjugations 101

ablative singular, arranged alphabetically : a, long e, short e, i, o,


and u. 20 Thus it seems that by this time in the tradition, if not
earlier, the genitive has once and for all displaced the ablative as the
determiner of declensional affiliation and has relegated it to a
secondary though by no means unimportant role in Latin grammar,
namely, as a predictor of the plural.
Donatus offers us no information whatsoever, for surprisingly
he does not discuss declensions. The nouns he declines in the ars
minor (GL IV. 355-56) exemplify only the first three of our five
declensions, and both declensional affiliation and enumeration are
conspicuous by their absence. The paragraph which follows does
define the restricted role of the ablative, however.

XV. (ibid.) What ending is employed in the genitive plural of those


nouns which in the ablative singular end in a or o ? The gen. pi.
ends in rum, the dat. and abl. in is. What ending is used in the
genitive plural of those nouns which in the ablative singular end in e
or i or ul If the e is short, the gen. pi. ending is um ; if it's long,
rum ; if /', then ium ; if u, the gen. pi. ends in uum with a double u.
What is the ending for their dative and ablative plural? They all end
in bus. 21

Donatus has abandoned the strict alphabetical order of Varro's


original formulation and its elaborations in the Instituta Artium and

20. Diomedes concludes his account of the ablative in a decidedly curious manner :
(id : 308) meminerimus autetn quaedam nomina vel auctoritate veterum vel
euphonia modo secundo modo quarto ordine declinari, ut domus ficus laurus
quercus et conplura arborum nomina [let us recall, however, that some nouns,
either by virtue of the ancients' authority or because of euphony, are sometimes
declined in accordance with the second declension and sometimes the fifth, like
domus 'house', ficus 'fig-tree' laurus 'laurel', quercus 'oak', and any number
of other names of trees]. Either the sentence is an interpolation or Diomedes
hasn't been honest with us, for the quarto ordine [fourth declension] here does
not correspond to the fourth of his seven declensions enumerated earlier but
does correspond with our fourth. If the sentence is his, then he either made a
mistake (quarto [fourth] for quinto [fifth], which could be more readily
explained as a scribal error), or else he was aware of a schema which
encompassed only five declensions and had earlier deliberately elaborated it to
a system of seven for reasons which are beyond our ken.
21. Quaecumque nomina ablativo casu singulari a vel o fuerint terminata genetivum
pluralem in quid mittunt ? In rum, dativum et ablativum in is. Quaecumque
nomina ablativo casu singulari e vel i vel u fuerint terminata genetivum
pluralem in quid mittunt ? Si e correpta fuerit, in um ; si producta in rum ; si i
fuerit, in ium ; si u, in uum geminata u littera. Dativum et ablativum in quid
mittunt ? In bus omnia.
102 Daniel J. Taylor

Diomedes and has opted instead for a quasi-alphabetical order such


as we can also find in Charisius and which can be explained in
either of two ways. Either he is aware of what we now consider the
five canonical declensions and has altered the inherited Varronis
régula accordingly (by listing o-stems second), or he has combined
a-stems and o-stems in his first question because they share the same
genitive, dative, and ablative plural endings and therefore elicit the
same response from his students. 22 In either case long-e-stems, our
fifth declension, as has been their wont all along in these explicit
and implicit catholica de ablativo casu, are treated side-by-side with
consonant stems with their short e ending in the ablative singular.
Donatus elaborates somewhat in the ars maior (GL IV. 378-79), but
there he retains a strictly alphabetical order : a, short e, long e, i, o,
and u. Nowhere, however, does he allude to a genitive-based system
of five declensions, and that bothers Servius.
Servius takes great pains to explain what he obviously
considers a monumental omission on Donatus' part.

XVI. GL IV. 408-09. All nouns are subsumed within five rules,
which are not in Donatus because his ars is only an epitome, but
they must be mastered anyway. These rules are deduced from the
ending in the genitive singular, for that case is characterized by five
different endings : the diphthong ae as in Musa Musae ; i as in
doctus docti ; is as in pater patris ; us as in nom. fluctus, gen.
fluctus ; and ei as in masc. or fern. nom. dies, gen. diei. Therefore
when we come across some noun whose declension is a matter of
doubt, we must first of all inquire as to its genitive ; if the genitive
ends in ae, for example, the noun will be declined just like Musa.
And so on and so forth. B

Not only does Servius explain Donatus' otherwise inexplicable


omission (and we may choose to doubt the accuracy of that
explanation), but he also provides the clearest and most succinct

22. If the former explanation happens to be correct, it provides the only evidence
I've found so far to suggest that Donatus knows a system of five declensions.
23. Omnia nomina...quinque regulis continentur, quae regulae apud Donatum
quidem non sunt propter conpendium, tamen tenendae sunt. colliguntur autem
istae regulae de genetivo singulari : nam is casus quinque finibus terminatur,
aut ae diphthongo, ut Musa Musae, aut i, ut doctus docti, aut is, ut pater
patris, aut us, ut hie fluctus huius fluctus, aut ei, ut hie vel haec dies huius diei.
ergo cum invenerimus aliquod nomen, de cuius declinatione dubitatur,
quaerendus nobis erit praecipue genetivus ; qui inventus si aefuerit tenninatus,
ad similitudinem Musae declinabitur. ita et in reliquis fiet.
Latin declensions and conjugations 103

statement to date of the genitive-based scheme of declensional


affiliation. So either Servius is more knowledgeable than Donatus in
this respect, or the principle that hindsight is better than foresight is
at work in his text.
Much more instructive is Servius' immediately preceding
paragraph (zd.408) on the ablative.

XVII. The ablative singular determines the genitive, dative, and


ablative plural endings in accordance with grammatical rules. But
that statement referring to « an ablative ending in either long or short
e » must be elaborated as follows : the e is long whenever the
genitive singular ends in ei, but when the genitive singular ends in
is, the e in the ablative is short, as is exemplified by gen. patris and
abl. patre. This, moreover, is the reason why the Latin grammarians
have decided to formulate their rules on the basis of the ablative
singular, since it is a Latin case and doesn't exist in Greek ; and by
all means that case which is uniquely characteristic of speakers of
Latin ought to provide the rules for Latin nouns. 24

Servius' paragraph elegantly epitomizes the history of declensions in


ancient Latin grammar. The last sentence, in which the role of the
ablative is not restricted, refers to the earliest or Varronian stage ;
the second sentence refers to the intermediate stage, wherein the
quantity of the vowel e in the abl. sg. is taken into consideration and
in which the genitive has begun to play a more prominent role ; and
the first sentence is a statement of the status quo contemporary with
Servius, in which the role of the ablative, the longstanding Varronis
régula, has been constrained or restricted and which presupposes
both the genitive-based account of declensional affiliation and the by
now canonical five declensions. [Sergius] supports Servius on all
counts. 25

24. Ablativus singularis regit genetivum dativum et ablativum pluralem secundum


régulas in arte positas. sed Mud quod dicit 'ablativo e producta vel correpta
terminata', ita intellegendum est, ut tune producta sit, quotiens genetivus
singularis ei terminatus sit ; cum vero is fuerit tenninatus idem genetivus
singularis, < correpta >, ut huius patris, ab hoc patre. ideo autem Latini
voluerunt ab ablativo singulari régulas sumere, quoniam hic casus Latinus est,
nee eum habent Graeci ; et ille ulique casus debuit régulas Latinis nominibus
dare, qui proprius Latinorum est.
25. I do not think, however, that Servius and [Sergius] are historically correct in
explaining why Varro based his theoretical and embryonic declensions on the
ablative, though they may be correct insofar as the anonymous Latini to whom
they refer are concerned. Varro's schema is predicated exclusively on an
104 Daniel J. Taylor

XVIII. GL IV. 495-96. There are six cases : nominative, genitive,


dative, accusative, vocative, ablative. Greeks base their rules on the
nominative, but Romans have opted for the ablative so as to derive
rules for the other cases from it, inasmuch as we know that the so-
called rule of the ablative does not pertain to Greek nouns.
Moreover, the ablative singular ends with the five vocalic letters a,
e, i, o, u : a as in Musa ; e, which is sometimes short and sometimes
long, short as in pariete and long as in die ; i as in puppi ; o as in
docto ; u as in versu. There are five letters but six rules on account
of the e, which is shortened as well as lengthened.

The foregoing are the rules which Donatus has touched upon
in the first section of his artes. What follows are those which master
Servius has prescribed by consulting other sources. All Latin nouns
employ endings for the genitive singular in accordance with five
rules : the diphthong ae as in Musae ; i as in docti ; is as in patris ;
us as in versus ; ei as in diei. Therefore whenever we want to decline
any noun, we first ask what its genitive is ; once that has been
determined, then we check to see which of the above mentioned
nouns it is similar to and decline the noun in question in exactly the
same way. M

So at this point the tradition has pretty much exhausted itself, and
Priscian can take the declensions for granted.
Consentius' style and independence are noteworthy, and his
observations on nominal inflection may be read almost as a
summary of the history of declensions as reconstructed herein. After

insightful analysis of Latin morphophonemics. The terminology he originates


(see above), however, directs his successors' attention to Greek inflection and
leads quite naturally to the historical explanation offered by Servius and
[Sergius] .
26. Casus suni sex, nominativus genitiviis dativus accusativus vocativus ablativus.
Graeci régulas a nominativo habent ; Latini ablativum elegerunt, ut de ipso
régulas ducerent ad alios casus, ut sciamus quia ista régula ablativi casus ad
nomina Graeca non pertinet. ... ablativus autem singularis quinque litteris
vocalibus terminatur , a e i o u : a, ut ab hac Musa ; e partim correpta partira
producta, correpta, ut ab hoc pariete, producta, ut ab hoc die ; i, ut ab hac
puppi ; o, ut ab hoc docto ; u, ut ab hoc versu. quinque sunt litterae, sed sex
regulae propter e, quae nunc prod licit ur, nunc corripitur. ...
Haec sunt quae Donatus in prima parte artium tractavit. haec magister
Servius extrinsecus dictavit. ornnia nomina Latina genitivo singulari quinque
regulis terminantur, aut ae diphthongo, ut huius Musae, aut i, ut huius docti,
aut is, ut huius patris, aut us, ut huius versus, aut ei, ut huius diei.
quotienscumque igitur aliquod nomen declinare volumus, primum genitivum
eius requiramus ; quo invento quaerimus similitudinem nominum praedictorum
et pari ratione declinamus.
Latin declensions and conjugations 105

differentiating between analogia [analogy] and régula [rule] (GL


V.353), he says : nunc igitur de regulis, quae Graeci canones
appellant, dicemus, ac prim de his quas ablativus casus numeri
singularis ostendit, quoniam hae breviore quadam via ad
emendationem loquendi sufficere posse creduntur [we shall therefore
now discuss grammatical rules, which the Greeks call canons,
beginning with those which the ablative singular manifests, since by
virtue of their particular and briefer formulation they are deemed to
be able to suffice for the removal of errors from speaking],
Consentius predicates the priority of the ablative's regulae on
linguistic economy, just as Varro had originally done (cf. text VII
above). He then adduces those regulae in a rather novel 27 order : a,
o, long e, short e, u, and finally i. He soon (id.351) turns to another
topic :

XIX. But nonetheless, lest anyone get the idea that the rules for the
ablative singular case itself are determined by magic rather than
reason, I shall explain the method whereby one can proceed properly
to that case. Therefore a few things must be said about the rules
pertaining to inflection in the singular number. Insofar as that topic
is concerned, it seems to me difficult and troublesome to use the
nominative singular as a point of departure. For that case manifests
in and of itself no adequately defined systematic principle which can
function compellingly as a sufficient condition for inflecting the
other cases. a

Consentius almost seems to be recapitulating Varro 's arguments in


LL X.51ff, which lead to Varro's ablative-based schema of
embryonic declensions in paragr. 62. 29 In any case he concludes
quickly (id. 35$) : ergo ob has difficultates ab eo [sc. nominativo]
ordinare régulas supersedimus, quia neque ratione certae sunt et
numéro inexplicabiles [therefore on account of these difficulties we

27. The order would appear to be predicated on the endings in the genitive plural ;
I am indebted to Vivien Law for this suggestion.
28. verum tamen ne in ipso ablativo casu singulari divinatio potius quam ratio
puietur, qua via veniri ad eum oporteat explicabo. igitur de regulis singularis
numeri pauca dicenda sunt. in quo mihi difficile et arduum videtur a nominativo
incipere singulari. is enim nullam ex se fere certain rationem emittit, cui servir e
necesse sit condicionem cas mon ceterorum.
29. I would also suggest, on the basis of this and other regulae-tcxts, that the
régula ablativi (earlier, the Varronis régula) is the first of the regulae and that
its descriptive success is what engenders the grammarians' search for and
exploration of regulae for the other cases.
106 Daniel J. Taylor

refrain from predicating rules on the nominative singular, because


its forms are not systematically definitive and are too numerous and
complex].
Consentius' next paragraph is rather striking ; it begins as
follows : Ordinemus ergo a genitivo singulari, qui cum et ipsa
natura quodam modo per se valeat (non enim rationi alicui
praecedentis casus débet, cur hoc aut Mo modo ecferatur), cetews
ex se gignit atque explicat [let us therefore base declensional
affiliations on the genitive singular, which, since even its very
nature is in some manner independently valid (for it does not require
any principled rationale determined by some preceding case to
explain why its desinence is expressed in one way or another),
generates and develops the other cases from itself]. In his
parenthesis Consentius has explained why the Latini came to prefer
declensional affiliations based on the genitive (a question so far
avoided), notwithstanding the descriptive economy of the ablative,
which he had just acknowledged only a few pages earlier. The very
irregularity of the genitive (historical as well as descriptive, as we
now know) and its unpredictability, especially from a student's
perspective, make the genitive an ideal candidate for the role of
determining declensional classification. Consentius continues : atque
hoc ausus sim dicere, quod omnis ilia varietas, quant plerique a
nominativo conati sum explicare, tribus quattuorve a genetivo certis
regulis includitur [and I have ventured to make this assertion
because all that declensional diversity, which most grammarians
have attempted to resolve on the basis of the nominative, is
subsumed within three or four well defined rules predicated on the
genitive]. The tribus quattuorve [three or four] ought give us pause.
Consentius lists only three declensions, again in a rather novel
order, based on their ending in the genitive singular : is, i, ae. But
as he says, sed quaedam extra has in paucis admodum nominibus
aliter eveniunt, quae et ipsa breviter attingemus [but in the case of
at least a few nouns their inflections fall outside the scope of these
rules, resulting in different patterns, and these too we shall briefly
address], and he is as good as his word ; the u-stems and long-e-
stems are subsumed within this latter class.
Consentius would appear to be recapturing or transmitting that
stage of the tradition when the genitive is only just beginning to play
the preeminent role in determining declensional affiliation and when
the ramifications of that innovation are just being explored, as
Latin declensions and conjugations 107

witness the facts that he nowhere numbers or enumerates the


declensions and that he shows no awareness of the canonical five
declensions which ultimately evolve from a genitive-based account
of declensional affiliation. It therefore comes as no surprise that he
then (id. 359ff) proceeds to analyze the genitive in more detail, but
not without a parting shot, as it were, at the ablative. Haec sum
declinationis discrimina in numéro singulari, quae nos via quadam
ad ablativum ducunt, qui declinationem rursum numeri pluralis
informat [these then are the salient characteristics of declension in
the singular number, which direct us methodologically to the
ablative, which delineates in turn the declension of the plural
number.] Consentius has, in effect, explained how the ablative's
role, the Varronis régula if you will, has been assimilated into a
genitive- based system of declensional classification. The nominative
is a given ; the genitive, which by virtue of its descriptive
uniqueness is now the basis of the declensions (regardless of how
many or how ordered), leads through the singular paradigm to the
ablative ; and the ablative generates, to use his earlier metaphor, the
plurals in their entirety. Consentius' observations obviously hearken
back to that intermediate stage in the tradition for which we have
little or no direct evidence but which must have occurred.
In fine, declensions in Latin also have a lengthy history. They
begin as theoretical constructs in Varro's De Lingua Latina, and the
ending in the ablative singular is their point of departure. Varro's
embryonic system is slowly but surely revised, at least to the extent
that our fifth declension is discovered, as it were, and the role of the
ablative remains central. A new schema based on the ending in the
genitive singular comes to the fore, and the ablative's role becomes
diminished but still remains significant. In due time the canonical
five declensions emerge. That history is not as cumulative or as
unilinear as a summary makes it appear, however, for our extant
texts contain any number of discrepancies, omissions, and
competing claims. The ancient ars grammatica is therefore a
dynamic rather than a static endeavor, but its auctoritas [authority]
is such that most contemporary Latin teachers and scholars are not
even aware of any system other than the one which they have
ultimately inherited from the last of the grammatici Latini.

Concluding remarks. Let me reiterate that this is a survey and


that I am therefore not wedded to any of the conclusions or
108 Daniel J. Taylor

interpretations adduced herein. Indeed, I have scrupulously and


consciously avoided references to secondary literature, manuscript
traditions, typology, and matters of Quellenforschung and have not
perused all the primary sources. Nor have I even mentioned various
and sundry attendant issues : alphabetized lists of nominal analogiae
[analogies] would seem to have proliferated at an early date ;
Varro's use of ordo in LL X seems to anticipate its widespread use
in later grammatical literature ; the inclusion of suzugia
[conjugation] under the rubric of verbal accidence in Greek
grammar would appear to explain why the Latin conjugations,
which differ in kind from those we found in Dionysius Thrax's
Techne, occur where they do in the Roman ars, whereas the absence
of declensions in Greek grammar apparently explains why the Latin
declensions occur in no fixed order in our texts and why dedinatio
[declension] is not listed under nominal accidence in the artes. And
so on and so forth. Even so, however, this survey does highlight a
number of issues, some of which may come as a surprise, and raises
a number of others.
Though the specific results of my survey are somewhat
tentative, they do relate in a general way to the results of other
studies. For example, what I purport to see in the history of
declensions and conjugations from Varro to Priscian supports what
Hovdhaugen (1987) has to say about the treatment of voice by the
grammatid Latini. The Roman ars grammatical one of the most
significant cultural artifacts transmitted throughout the centuries, is
simply not the monolithic structure such as some of us, at least,
have been led to believe it was. Likewise, the obvious disparity in
the analysis of declensions which characterizes the texts of
Charisius, Diomedes, and Donatus supports Barwick's (1922) broad
groupings of the earlier traditions. And of course it goes without
saying that the accounts of declensions and conjugations in our
extant texts confirm what we already know about chronology :
neither the floruit of a grammarian nor the date of his publication
provides an unequivocal criterion for dating the intellectual content
of everything he transmits. At any rate, when ideas change, as they
so obviously do in the case of declensions and conjugations in
ancient Latin grammar, it is incumbent upon us to take notice.
Latin declensions and conjugations 109

REFERENCES

Barwick, Karl. 1922. Remmius Palaemon und die rômische ars


grammatica. (= Philologus Supplementband 15 :2.) Leipzig :
Dieterich. (Repr., Hildesheim : Olms, 1967.)
Funaioli, Hyginus, ed. 1907. Grammaticae Romanae Fragmenta. Leipzig :
Teubner. (Repr., Stuttgart : Teubner, 1969.)
GL = Grammatici Latini. 7 vols. + supplement. Ed. by Heinrich Keil.
Leipzig : Teubner, 1857-1880. (Repr., Hildesheim : Olms, 1961 and
1981.)
Hovdhaugen, Even. 1987. « Genera Verborum Quot Sunt? ». The History
of Linguistics in the Classical Period ed. by Daniel J. Taylor,
133-147. (= Studies in the History of the Language Sciences, 46.)
Amsterdam : Benjamins. [= Historiographia Linguistica 13 (1986)
307-321.]
Mazzarino, Antonius. 1955. Grammaticae Romanae Fragmenta Aetatis
Caesareae. Torino : Loescher.
Taylor, Daniel J. 1977. « Two Notes on Varro ». American Journal of
Philology 98, 130-132.
Taylor, Daniel J. 1990. « Dionysius Thrax vs. Marcus Varro ».
Historiographia Linguistica 17, 15-27.

Reçu Octobre 1991 Department of Classics


Lawrence University
Appleton, Wisconsin, USA

Вам также может понравиться