The Monophthongization of Latin ae

Author(s): Edgar H. Sturtevant

Source: Transactions and Proceedings of the American Philological Association, Vol. 47 (1916),
pp. 107-116
Published by: The Johns Hopkins University Press
Stable URL: http://www.jstor.org/stable/282830 .
Accessed: 15/05/2014 16:52

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to
Transactions and Proceedings of the American Philological Association.

http://www.jstor.org

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions
Vol. xlvii] of Latin ae
The Monophthongization 107

IX. - The Monoplhthongization

of Latin ae

BY PROFESSOR EDGAR H. STURTEVANT

COLUMBIA UNIVERSITY

MOST English-speakingscholars have for some time past

been accustomed to pronounce Latin ae as theydo ai in Eng-
lish aisle, while scholars of othernationalitieshave employed
the correspondingdiphthongsof theirown languages. There
has been an undercurrentof dissentfromthis practice which
has occasionally come to the surface: Stolz and Schmalz,
Lat. Gram.2p. 27I, King and Cookson, Sound and Inflection,
p. 85, and Sihler, P.A.P.A. xxix, xl-xliv,maintainedthat ae
was pronounced as a monophthong,while Lindsay, Slhort
Historical Latin Grammarl,p. I3, held that the firstmember
of the diphthongwas a sound similarto the a of English man
and the second was like the e of English men. In the second
(I9I5) editionof this book Lindsay modifieshis earlier teach-
ing by assuming (p. I3 f.) that "in the age of Cicero ae"
without quite losing its diphthongal character- "must
have sounded more like a long open e (something like our
interjection eh protracted,or rather doubled)." In these
circumstancesa reexaminationof the evidence may not be
untimely. We shall considerthe various items in chronologi-
cal order, and attempt thus to build up a historyof ae in
Latin.
The diphthong in question was originallywrittenai, but
the spelling ae began to appear about 200 B.C. and became
usual by about 150 B.C. (see Sihler,l.c.,forcitatioils). No one
doubts that the earlier orthographyrepresentedthe pronun-
ciation; and we can scarcely escape the conclusion that the
change in spelling reflecteda change in pronunciation,since
no othermotiveforit has been suggested. Neither can there
be doubt about the general nature of the change of sound.
If the diphthonghad become a monophthongby 200 B.C. (as
Sihler maintains),the new spelling would have been e. The
orthographyae must reflectmerelya more open pronuncia-
tion of the second member of the diphthong; the earlierai

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions
i o8 Edgar H. Sturtevant [I9I6

denoted a diphthong ending in a close i, as in Italian mai,

while the later ae denoted a diphthong ending in a more
open sound approaching a close e. Since this, rather than
a true az, is the diphthong heard in English aisle, zy, etc.,
the currentpronunciationof ae in England and America is
correct; the usual descriptionof it, on the other hand, as
a + i is incorrect.1 This newer pronunciation must have
become established before the beginning of the second cen-
tury B.C.; for the change in orthographycould not begin
until the new pronunciationhad gained standing. Spelling
reformmay lag far behind a change in pronunciation,but
under ordinary circumstances it cannot anticipate such a
change.
In case diphthongal ai was followed by consonantal i, as
in aio and malor (pronouncedai-to, mai-ior), the second ele-
ment of the diphthong remained unchanged,as the orthog-
raphy proves. English furnishesa parallel in such phrases
as mnyuse (pronounced mnai yas), as contrasted with mly
(pronounced mae) in most otherphrases. Latin ais and ami
(from aisne) retained the original diphthongunder the influ-
ence of aio, etc., supportedperhaps by uncontractedals.
The diphthong ai had a very similar historyin Oscan.
The Oscan national alphabet contained a symbol q (trans-
cribed i) to represent the open i-soundwhich resulted from
originale or I, and also frome beforeanothervowel (e.g. likitud
=-llcc7o, iu-k= ea), and a symbol I (transcribedi) to denote
the close i-soundwhich resulted fromoriginalz (e.g. aidilis =
aediles, imaden: imlus). The latter character was also em-
ployed for consonantal i. Hence the orthographyof such
words as aidilis, kvaistur, viai, svai, etc., indicates that the
second member of the diphthong was an i-sound verging
toward an e-sound; Oscan ai must have been similar to
Latin ae. In case, however,the diphthongwas followed by

1 Oertel, ap. Lane, Latin Grammar2, p. 7, describes the sound of Latin ae cor-
rectly, but fails to identify it wviththe English diphthong. That the English
diphthong really is ae rather than ai is readily seen if one pronounces the dis-
syllabic combination a-e (ah-eli) and then repeats it more and more rapidly until
the two sounds coalesce; a similar experiment with a-i (ah-ee) produces Italian ai.

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions
Vol. xlvii] of Lati;i ae
The Monoplitlhongization I09

consonantal i, its second member is shown by the orthog-

raphy to have been a close i (e.g. Puimpaiianai,Maraiieis,
Mefitaiiais),preciselyas Latin ai retained its close i in similar
circumstances.
Even more importantfor our investigationis the represen-
tationof the Oscan diphthongai by ae in documentswritten
in the Latin alphabet (e.g. suae, aeteis, Bansae). When the
Oscans began to use the Latin alphabet, certainlywell after
200 B.C., Latin ae must have represented a true diphthong;
for if it had represented a monophthong or a diphthong
scarcely distinguishable from a monophthong,the Oscans
would have transcribed their diphthong ai by ai, as they
actually transcribeduii by oi (feihuiis" muris": eizois " eis"),
although the dig-raphoi was foreignto Latin orthography.
That ae was still a diphthong in the time of Lucilius is
shown by his jest (II30 Marx):
Ceciliuspretorne rusticusfiat.
For if the monophthongizationof ae was a mark of rusticity,
ae must have been a diphthongin urban Latin.
Our knowledge of this rustic e for ae is derived largely
from the passage in Varro's de Lingua Latina (vii, 96), in
which is preserved the Lucilian fragmentjust cited:
Apud Matium: ' obsceni interpresfunestiqueominisauctor.'
obsceniumdictumab scena; eam ut Graeciaut Acciusscribitscena.
(in pluribusverbisa antee alii ponunt,alii non, ut quod partim
dicunt(scaeptrum,partim) scep/rumPlautiFaeneratricem,
aliiFenera-
tricem;sicfaenisiciaac f [o]enisicia,
ac rusticipappumm[a]esium,
nonmaesium; a quo Luciliusscribit,' Cecilius(pretor)ne rusticus
fiat'). quare turpeideo obscaenumquod nisi in scaena[m] palam
dici nondebet.
Since these words have often been misunderstood,it may
be well to supply a translation: " In Matius (we read) 'Ob-
sceni interpresfunestique
ominis auctor.' Obscen?um is derived
fromscena; he writes it scena (with a monophthonginstead
of a diphthong) as the Greeks and Accius do. (In a con-
siderable number of words some persons put a beforee, and
others do not; as, for example, some say scaeptrum,others

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions
IIO Edgar H. Sturtevant [I9I6

sceptrurn, some the Faeneratrixof Plautus, others the Fenera-

trix; just so (we hear)faenisicia and fenisicia, and the coun-
trypeople call an old fellowmesizus,not maesius; wherefore
Lucilius writes 'Let's not make the boor Cecilius pretor!12)
Hence what is foul is obscaenzus for the reason that it should
not be mentionedin public except on the scaena."
In v, 97 of the same treatise Varro refers brieflyto the
rusticmonophthonginstead of ae: I;-cus,quod Sabinifi;-cus;
quod illic ftdiis, in Latio rure [/h]cdus, qui in urbe (ut in
multis)a addito (KI)aedus.
In the time of Varro, then,as well as in the time of Lu-
cilius, e was a familiarrusticvariantforthe urban diphthong
ae. As we see fromthe passage last quoted, the Sabine dia-
lect was here in harmonywithrusticLatin; epigraphicalevi-
dence shows that Faliscan, Volscian, and Umbrian also had
simplifiedai to e, and that the same group of dialects had
simplifiedother diphthongsas well in a way foreignto urban
Latin. In this respect several of the old dialects of Latium
agreed with the four Italic idioms just mentionedas against
Roman usage, c.g. Praenestine losna (C.J.L. I, 55) = Iitia
from*louiksniz;Plotina (C./.L. XIV, 3369) = Plauti;za; Ces(K{a)
(C.lL. XIV, 3I93) = Caeszula. It is therefore a dialectic
peculiaritywhich antedates the establishmentof the Roman
dialect as the standardlanguage of Latium.
The passages just citedfromLucilius and Varro prove that
even after the urban dialect had become the norm,country
people continued to use e where they should have used ae.
In fact, a few country(i.e. dialectic) words with e for ae
penetratedthe cityand gained a footholdin standard Latin.
One of the clearest cases is bTvir= Skt. dcvd,Gk. &ctp (from
*8atF7p), whose second vowel is due to the analogical influ-
ence of vir, " husband," and whose initialI ford shows that
the word is of Sabine origin. The traditionin favorof e is
not quite so clear in seJes,pra'sapcs, praesepia, but, as we
shall see, the monophthongis supported by the Romance
languages. Walde (Lat. etym. Wortcrb.s.v.) thinks that
fenum contains an orizinal monoDhthong.but Varro's evi-
"
2 Or perhapsironical," I hope the pretorCeciliusisn'ta countryman!

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions
Vol. xlvii] The Monophzthongization
of Latin ae III

dence in favol of faenisicia is supported by Italian fieno,

whose vowel must represent Latin ae or e. Several other
rustic words with e for ae are evidenced by the Romance
languages; see Meyer-Liibke,Gram. lang. rom.,I, 255.
Varro's preferenceof scaena to scena is supported by the
superior manuscriptauthority(see Sommer, Handb. d. lat.
Laut- und Formenlehre 2, p. 72 and references)and by scaenam
(C.I.L. I, 206, 77), scaena (ib. I, I009, 13), scaenarium(ib.
I, I341), scaenticis (ib. lI, I663), and proscaeniumn(ib. II, I83).
That it contained a real diphthongis shown by scaina (ib. i,
I280). This word and also scaeptrum(Varro, I.c.) for Greek
o-K577r7pov owe theirdiphthongto an "over-correction"; per-
sons who took pains to say praetor instead of pretor"cor-
rected" scena into scaena (so Walde, I.F. xxx, I39; Sommer,
Handb.2 p. 72). Solmsen's suggestion(Unters. zur griech.
Laut- untdVerslehre,279) that, in view of the related forms
cTLta and Skt. chldyd,there may have been a dialectic Greek
*alcatva, cannot be accepted until a similar explanation is
available forscaeptrum.
It is impossibleto suppose (with Lindsay, Latin Language,
p. 42; Claussen, Rom. Forsch.xv, 854; Carnoy,Latin d'Es-
pagne, p. 79, and others) that the ae of scaena and scaeptrum
was a peculiarly exact method of transcribingGreek q.
Even if we could grant that both Latin ae and Greek ij
represented a long open e in the Ciceronian period, there
seems to be no reason why these two words should be
transliteratedmore scrupulously than the hosts of other
Greek loan-words containing q. On the contrary,the Ro-
mans of Cicero's day and later, although careful about the
form they gave to new borrowings(e.g. Ilitliyia), did not
insist upon the correctformof Greek words which had long
been in the language (e.g, comisor,Hercuiles,-tus,ancora,clzo-
ragium). To the latterclass scaena clearlybelongs; it occurs
in Plautine prologues and scaenica is found in Terence, Hec.
I6; scaena was a technical theatrical termwhich must early
have establisheditselfat least as firmly as (Doric) choragium.
There is no proof that scaeptrurnis early,but the very fact
that it was treatedin the same way as scaena makes an early

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions
I I2 Edgar H. Sturtevant [I9I6

date probable. That the spelling ae in these words was not

regarded as an approximationto the Greek formis made per-
fectlyclear byVarro's citationof the spellingwithe, not only
fromAccius, who undertookto followGreek usage accurately
(see Varro, L.L. x, 70), but fromthe Greeks themselves.3
The rustice fromai was no doubt at firstan open e, as in
Umbrian (see Buck, Grammarof Oscaniand Umibrian, p. 44),
but the " over-correction" of scte;aand sceptrumto scaena and
scaepzrutmindicates that it had become as close as ordinarye
by the firstcenturyB.C. For the open e of Greek a-Kc yVi had
undoubtedlybeen assimilated to the Latin close e duringthe
centuryand more in which it had served as a technical term
of the motleycrew who made up the theatrical professionat
Rome. There is no doubt at all that the rustice of sepesand
fe"numdid ultimatelybecome identicalwithordinarye in most
of the Roman world; for the two yield the same result in
most of the Romance languages. OI1ly in Italian do these
words show formswhich indicate Latin open e. The follow-
ing table presentsthe facts:
Open e Rustice forae Close e
Latin me1 sepes fenum verum
Italian miele siepe fieno vero
French miel soif foin voire(vere)
Spanish miel seto(s-eptum) heno vero
Rumanian miere fin plin(plenum).
A similarlyinconsistentdevelopment has been observed in
three or four other words (see Meyer-Lubke,Ic.), and we
3 Other Greek loan-words also occasionally show ae for -q in Imperial times
(see Carnoy, op. cit. p. 8i; Hammer, Die lokaZl Verbreitungfriihesterromnaniscizen
laiztwandlungent im alten Italien, p. 9-I4; and the indexes to the C.I.L. under the
caption Grazmitmatica quaedczami); but these cases are much fewer than the cases
of ae for Greek E, and a majority of them occur in final syllables, in which posi-
tion ae is often substituted for original Latin e. There is no reason to doubt,
therefore,that these words are for the most part instances of the later monoph-
thongization of Latin ae, which is presently to be discussed. The vulgar Latin
genitives such as Aquilliaes (C.I.L. 1, I025), Laudicaes (ib. I, 12I2) may repre-
sent a contamination of the Latin ending -ae and the Greek -- s. To regard such
forms as meticulously accurate transcriptions of Greek -'vs (Sihler, i.c.) is quite
absurd, in view of the carelessness of the inscriptions in which they occur and of
the fact that many of the words concerned are Latin.

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions
Vol.xlvii] The Monophthongizationof Latin ae 113

must assume that they too spread over the Roman world
in a formwhich originallybelonged to the countrydistricts
of central and northernItaly. The inconsistencybetween
Italian and the other Romance languages indicates thatthe
open e-sound was permanentlyretained in the region where
it originated,whereas open e became close e in the city of
Rome and in the provinces.
In support of his theoryof virtuallymonophthongalpro-
nunciation in the time of Cicero, Lindsay (Short Historzcal
Latin Grammar2,p. I4) adduces a new argument: " When the
prepositionprae was shortened before a followingvowel it
came to be writtenpre in prehendo,a clear proof that ae was
the long (more correctly,the diphthongal) form of e (the
short open e-sound)." But prehendohas a short initial syl-
lable as early as Plautus: e.g. Epid. i (the crucial word is
preservedin the Ambrosian Palimpsest):
Heu's adul6scens. quisproperantem
m6repreh6ndit
pai1ho?
It is possible, of course,to read reprendithere, as we must
read prenditin Bacch. 696; but the contractionseen in this
and similar forms presupposes a monophthongal e in the
prefix. If prehendoproves anythingabout the pronunciation
of ae, such proof holds for Plautine or pre-Plautinerather
than for Ciceronian Latin. Now, since we have seen that
the spelling ae, which began to be used during Plautus' life-
time,,clearly indicated a diphthongal pronunciationat the
time of its adoption,we must-lookfor a differentexplanation
of prehendo, and two satisfactorysuggestions have in fact
been made. Sommer (Handb.2 p. I 12) iS inclined to the
opinion that the a of prae was assimilated to the vowel of
the followingsyllable (*prai-hendo> *praiendo > *preiendo
> pre(h)endo). No difficulty is caused by praeda and prae-
mizum,because they show contractionof *praiida and *prai.
imi'um. Such wordsas praeest and praeeo are re-compositions,
although of somewhat earlier date than praeemino, etc.
Schwyzer, Bern. Phil. Woch. XXIII, 439, suggests that the
formprehendooriginatedin the compounds comprehlendo and
reprehendo, where ai stood in an originallyunaccented sylla-

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions
114 Edgar H. Sturtevant [I9I6

ble and was weakened to J. That e beforea vowel became e

is shown by dieus from *djus and ol?um from *oijum. At
any rate, we may be sure thatpre- in prehendois not a mere
shorteningof the familiarpraeof Ciceronian Latin; it is far
too ancient forthat.
When ae became a monophthongelsewhere than in the
old Praenestine-Sabine-Umbrian district,it promptlyshowed
affinitieswith open e rather than with close e. Pompeian
graffitiof the firstcenturyA.D. show an extensiveconfusion
of e and ae, e.g. etati, maeae, l/aberae(C.I.L. iv, 1684; for
otherinstancessee Hammer,Op.cit. p. 11-14, and the Index to
C.I.L. Iv, Suppl.). A few plebeian inscriptionsof the city of
Rome show that a similarconfusionwas beginningtherealso
in the firstcenturyA.D., e.g. Cliarie(dat., C.I.L. VI, 5i8o),
saenatius(ib. VI, 2066; for other citations see Hammer, i.c.).
In the second centurythe confusionbecame much moreex-
tensive in Rome and appeared in the provinces(see Carnoy,
op.cit. 7I-74, and the indexes to C.I.L.). Since open e and
close e were customarilywrittenin the same way, the pho-
netic confusionbetweene and ae led to an occasional graphic
confusion between e and ae, e.g. aegisse (C.I.L. IV, 24I3 f.),
caeteri(ib. VI, I585 b).4
That mistakes in orthographyof this latter sort did not
reflecta confusionbetween close j and ae is proved by the
fact that the Romance languages keep the two sounds dis-
tinct. Latin open e, however, everywhereyields the same
resultas ae. The followingformsare typical:
Open e ae Close e
Latin me1 caelum verum
Italian miele cielo vero
French miel ciel voire(vere)
Spanish miel cielo vero
Rumanian miere cier plin(plenum).
4 As the final of a polysyllable ae for e is relatively common, e.g. sanctae, opti-
mae (adverbs, C.I.L. ii, 6278, 4405). We may infer that final e tended to become
an open e; and this inference is supported by the fact that, while the grammarians
are chieflyconcerned to distinguishae frome (quaeri/ur queritur, vae . ve, prae-
iniurn .-pretium), Servius, adAen. I, 344, thinks it necessary to distinguishbetween
mise;-ae and misere.

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions
Vol. xlvii] of Latin ae
The Monopht/zongization II5

The confusion of ae with open e-firstappears, as we have

seen, in Rome and in Pompeii (whose original language was
Oscan), both of them in regions which did not share the
earlier monophthongization of ai. It now appears that while
the earlier monophthongizationled to a confusionbetween ae
and close e, the later monophthongization led to a confusion
between ae and open A. It is thereforemost unlikelythat
the two processes had any connectionwith each other.
It remains to fix the date when the monophthongalpro-
nunciation of ac made its way into standard Latin.5 Teren-
tius Scaurus, a leading grammarian of Hadrian's reign, is
unusuallyexplicit in his remarkson ae (vii, i6 Keil): A igi-
tur litterapraepositaest u et e litteris.... et apud antiquos i
litterapro ea scribebatur,ut testantur,pEra7rwXaokot in quibus
est eius modi syllabarum diductio, ut pictai vestis et aulai
medio pro pictae et au/ae. sed magis in illis e novissima
sonat. If his ear told Terentius that the second memberof
the diphthong is e ratherthan i, we may be assured that he
heard a diphthong. In the early part of the second century,
then,ae was still a diphthongin standard Latin.
Terentianus Maurus, who probably wrote not far from
200 A.D., includes ae in his list of diphthongs(vI, 338 Keil);
but since he includes also ei (which he illustrateswitheitur,
oveis,and omneis)he is evidentlygiving merelythe traditional
teaching of the schools. That ae was really a monophthong
in his day may be inferredfrom Terentianus' account of ui
in the dative singular of qzi. He devotes more than a hun-
dredlines(VI, 345 ff.Keil, 11.67 I-777) to an involveddiscussion
of the questionswhethercui is a dissyllableor a monosyllable,
whetherit contains a diphthong or not, and whetherits u or
its i is to be regarded as a consonant. A careful examination
of the argumentshows prettyclearlythat cui reallycontained
a diphthong,and the question arises: " Why did Terentianus
hesitate to call it a diphthong?" We can scarcely findbut
5 I attach little importance to the evidence of loan words such as Welsh praidd
and Gothic kaisar, partly because of the notorious laxity of foreign pronunciation
and partly because we know little of the phonetic character of Celtic and Ger-
manic in the firstcentury.

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions
II6 EEdgarH. Sturtevant [19I6

one answer: In Terentianus' time ae and oe, as well as ei,

were really monophthongs,but theywere traditionallycalled
diphthongs. Terentianus saw that the old definitionof a
diphthongfittedthe ni of czii, but the sound was so different
in characterfromae, etc., that he hesitated to call it by the
same name. It is, then,probable that ae was a monophthong
in standard Latin as early as the end of the second cen-
turyA.D. This conclusion is supported by the fact,already
noted,that the confusionbetween ae and e became common
duringthe second century.
Various passages in grammarians of the fourthcentury
show clearly thatae was a monophthongat that time. We
need cite only the followingfromMarius Victorinus(vi, 66 f.
Keil): ConsimilirationequaeriturOrpheutsin metro,ut
vincatnec ThraciusOrpheus,
carminibus
Non nme
utrum trisyllabuman disyllabum sit, an idem nomen duplici
enuntiationepromatur,aut sine a littera,ut Pe/euus,Pent/zeus,
aut cum a, ut ita declinetur Orphzaezzsut Aristaezus. visum
est tamen hoc posse discerni,ut illa sine a litteraGraeca sit
enuntiatio,haec Latina quae per diphthongoneffertur.
The historyof Latin ae may be sketched as follows: The
orthographicalchange of ai to ae in the firsthalf of the sec-
ond centuryB.C. reflecteda change of the second memberof
the diphthong from a close i (as in Italian mai) to a more
open sound approaching an e (as in English aisle). In many
parts of Latium ai became e in prehistorictimes, and this
rustic e made its way into urban Latin in a few country
words such as sepes and fenzim. The attemptof dwellersin
the city-particularly, no doubt, those who had come from
the country- to avoid rustic e led to an "over-correction"
in the case of scaena and scaeptrnim. This rustice became in
the citya close e, like originalLatin J.
The monophthongization of genuine Latin ae, on the other
hand, led to a confusionbetween ae and open e. It began in
southern Italy and Rome in the firstcenturyA.D., and made
its way into the standard speech probablyin the latter part
of the second century,certainlybeforethe fourthcentury.

This content downloaded from 193.104.110.129 on Thu, 15 May 2014 16:52:13 PM

All use subject to JSTOR Terms and Conditions