PA 
 
 267 
 A7 
 1908 
 MAIN 
 
 UC-NRLF 
 
 B M 021 Qta 
 
A, 
 
 r 
 
THE RESTORED PRONUNCIATION 
 OF GREEK AND LATIN: 
 
 WITH TABLES AND PRACTICAL ILLUSTRATIONS 
 
 EDWARD VERNON ARNOLD, Litt.D., 
 
 PROFESSOR OF LATIN AT THE UNIVERSITY COLLEGE OF NORTH WALES: 
 LATE FELLOW OF TRINITY COLLEGE, CAMBRIDGE, 
 
 AND 
 
 ROBERT SEYMOUR CONWAY, Litt.D., 
 
 PROFESSOR OF LATIN IN THE UNIVERSITY OF MANCHESTER; 
 LATE FELLOW OF GONVILLE AND CAIUS COLLEGE, CAMBRIDGE. 
 
 FOURTH AND REVISED EDITION 
 
 {emhodying the schemes app'oved for Latin and Greek by the 
 Classical Association) 
 
 CAMBRIDGE : 
 AT THE UNIVERSITY PRESS 
 
 1908 
 
 Price One Shilling. 
 
 Thohnton 4 SoH, 
 
 Booksellers, 
 
 II The Broad. 
 
 Oxford. 
 
THE EESTOEED PRONUNCIATION 
 OF GREEK AND LATIN: 
 
 WITH TABLES AND PRACTICAL ILLUSTRATIONS 
 
 BY 
 
 EDWARD VERNON ARNOLD, Litt.D., 
 
 PROFESSOR OF LATIN AT THE UNIVERSITY COLLEGE OF NORTH WALES; 
 LATE FELLOW OF TRINITY COLLEGE, CAMBRIDGE, 
 
 AND 
 
 ROBERT SEYMOUR CONWAY, Litt.D., 
 
 PROFESSOR OF LATIN IN THE UNIVERSITY OP MANCHESTER; 
 LATE FELLOW OF GONVILLE AND OAIUS COLLEGE, CAMBRIDGE. 
 
 FOURTH AND REVISED EDITION 
 
 {embodying the schemes approved for Latin and Greek by the 
 Classical Association) 
 
 CAMBRIDGE : 
 AT THE UNIVERSITY PRESS 
 
 :l?Q8 •.:•• 
 
CAMBRIDGE UNIVERSITY PRESS WAREHOUSE, 
 
 C. F. CLAY, Manager. 
 
 HonUon : FETTER LANE, E.G. 
 
 ffilaasoto: 50, WELLINGTON STREET. 
 
 n 
 
 m 
 
 
 mn 
 
 Heipjtg: E. A. BROCKHAUS. 
 
 i^eto Hork: G. P. PUTNAM'S SONS. 
 
 iSombas antJ Calcutta: MACMILLAN AND CO., Ltd. 
 
 First edition — 1895. 
 Second edition — 1896. 
 Third edition— 1^01 . 
 Fourth edition— 1908. 
 
 [All rights rese7yed.'\ 
 
[71 ^^7 
 
 PREFACE TO THE THIRD EDITION. 
 
 IN issuing a third and revised edition of this pamphlet, we think 
 it may be convenient to give a short account of the movement 
 which it is designed to assist. 
 
 As early as 1871 attention was called in authoritative quarters 
 to the unsatisfactory standards of Latin pronunciation in vogue in 
 the United Kingdom, and a definite reform was advocated with 
 the support of such eminent names as those of H. A. J. Munro 
 and Edwin Palmer. This proposal was received with some favour 
 by the English Universities and public schools, with the result 
 that the reformed pronunciation was recognised as a permissible 
 alternative. In practice, however, it was seldom adopted, and 
 appeared only to increase the existing confusion. 
 
 When the University of Wales was founded in 1893, almost 
 simultaneously with the creation of a great number of schools 
 which provided instruction in Greek and Latin, it felt itself called 
 upon to deal with this question. Whatever excuses might be found 
 in England for indecision in dealing with a long-standing tradition, 
 it could not be expected that a new educational system should be 
 burdened with hesitations on so practical a question : least of all 
 in a country in which the reformed pronunciation of most of the 
 symbols concerned was already familiar in the native language. 
 Accordingly with the support of our colleagues we drew up and 
 published this pamphlet in 1895, the circumstances leading us to 
 make use almost exclusively of the English, French, and Welsh 
 languages to illustrate the pronunciation proposed. The scheme 
 was officially adopted by the University, and has since been in 
 regular use in the Principality ; and this experience has shown that, 
 whatever difficulties a change may cause to teachers accustomed 
 to a different system, the system itself causes none to learners who 
 are by it initiated to the study of Latin. 
 
 1—2 
 
 298498 
 
iv / ; FllEifAGE TO TEE THIRD EDITION 
 
 During the last ten years the reform movement has steadily 
 gained strength in England also. New grammars and school books 
 have familiarised Latin teachers in all parts with the proposed 
 change : and their representative associations have with practical 
 unanimity declared in its favour. The restored pronunciation of 
 Latin is now advocated by the Philological Societies of Oxford and 
 Cambridge, and by the Associations of Assistant Masters and of 
 Assistant Mistresses : it has been adopted with practical unanimity 
 at a numerously attended meeting of the Classical Association 
 recently held in Manchester : and the Classical Association of 
 Scotland gives its support on behalf of that part of the kingdom. 
 Within the last few weeks the Headmasters' Conference has by 
 a decisive vote declared in its favour ; and we trust that this 
 re-issue of our pamphlet will help to minimise any inconvenience 
 that may be felt in what remains of the period of transition. 
 
 The reform of Greek pronunciation necessarily lags behind that 
 of Latin, and its intrinsic difficulties are greater. The proposals 
 that we made in 1895 have passed through the test of practice, 
 and have in the main been found feasible. Difficulty, however, 
 is attached to insistence on the ' narrow ' pronunciation of et, the 
 'broad' pronunciation of w, and the pure 'aspirate' values of ^, 
 </), X- In the present edition we have thought it well to allow a 
 certain latitude in the pronunciation of ct and tu, and we have 
 taken a new departure with regard to the 'aspirates.' Some 
 recently discovered evidence (of which an account is given in an 
 Appendix for which, as generally for the statements relating 
 only to Greek, Professor Conway is responsible) has made it 
 probable that the pronunciation of these sounds had become, in 
 part at least, that of fricatives as early as the fourth century B.C. ; 
 and we therefore propose that the pronunciation of that century 
 rather than of the preceding should be accepted as the standard, 
 and reproduced with such approximation to accuracy as circum- 
 stances permit. 
 
 The science of phonetics has made such great advances in recent 
 years that it is necessary for us to make it clear that we are not 
 attempting a scientific treatise on the historical pronunciation of 
 either language. Thus we have left out of account numerous 
 
PREFACE TO THE THIRD EDITION v 
 
 details and distinctions which, though scientifically important, do 
 not greatly concern our immediate purpose. We gladly welcome 
 the efforts made by teachers of modern languages to spread a more 
 precise appreciation of the niceties of correct pronunciation, but w© 
 do not think it practicable or desirable to set up such exacting 
 standards for Latin and Greek, in teaching which our dominant 
 purpose must always be to bring pupils into acquaintance and 
 communion with the great authors in whose pages Greek and Latin 
 for ever live and speak. 
 
 We desire to acknowledge gratefully the assistance we have 
 received from many friends in the preparation of this pamphlet. 
 Prof. J. P. Postgate and Prof. J. Strachan, though in no way 
 responsible for its final form, have generously submitted it to a 
 searching revision, which has removed many obscurities and some 
 errors. And as regards the pronunciation of the modern languages 
 used in illustration, we are especially indebted to Prof. A. H. 
 Fynes-Clinton, Prof. J. Morris Jones, Mr T. Rea, and Dr Frederic 
 Spencer. 
 
 The books which we have most often consulted are the 
 
 following: . ,, .,. 
 
 ^ . /-v /. 1 Authorities 
 
 H. Sweet, Primer of Phonetics, Oxford, 1890, chiefly con- 
 
 and Histori/ of English Sounds, Oxford, 1888. A. J. ^"^««^- 
 Ellis, Early English Pronunciation, London, 1869. 
 W. Rippmann, Elements of Phonetics, London, 1903 (after Victor). 
 And an important paper by Prof. W. W. Skeat, The Testimfiony of 
 English to the Pronunciation of Latin, in Camh. Philol. Sac. Pro- 
 ceedings, 1905, p. 3. 
 
 [More detailed technical information will be found in E. Sievers 
 GrundzUge der Phonetik, Leipzig, 1901 (5^^ verbesserte Auflage) ; W. 
 Viet or, Elemente der Phonetik des Deutschen, Englischen und Franz6- 
 sischen (S*-® Auflage), Leipzig, 1904 ; 0. Jespersen, Lehrhuch der Phonetik 
 (Leipzig and Berlin, 1901).] 
 
 K. Brugmann, Griechische Grammatik (3^ Auflage) in I. 
 Miiller's Handbuch der Altertumswissenschafl, Band Greek. 
 
 2, Munich, 1900. G. Meyer, Griechische Grammatik 
 (3te Auflage), Leipzig, 1887. F. Blass, Greek Pronunciation 
 (translated by Purton), Cambridge, 1890. 
 
 Max Niedermann, Precis de Phonetique Historique du Latin 
 Paris, 1906. Emil Seelmann, Aussprache des ^^^.^ 
 
 Latein, Heilbronn, 1885. W. M. Lindsay, The 
 Latin Language, Oxford, 1894 (or, better, the Short Latin Historical 
 
vi PREFACE 
 
 Grammar by the same author, Oxford, 1896). W. Meyer-Liibke, 
 Grammaire des Langues romanes (traduction frangaise), Paris, 
 1890 — 1906. E. Bourciez, Precis de Phonetique frangaise^ 
 Paris, 1885. H. J. Roby, A Latin Grammar from Plautus to 
 Suetonius, part i, fourth edition, London, 1881. A. J. Ellis, 
 Quantitative Pronunciation of Latin, London, 1874. 
 
 [A convenient summary of the principal points is found in Professor 
 J. P. Postgate's New Latin Primer, revised edition, London, 1890. 
 See especially §§ 251 — 256. A. Bos, Petit traite de prononciation latine, 
 Paris, 1893, and F. Sommer, HaTidhuch der Lateinischen Laut- und 
 Formenlehre, Heidelberg, 1902, should also be mentioned.] 
 
 E. V. A. 
 
 R. S. C. 
 
 Bangor ) _ ,^_„ 
 
 V January, 1907. 
 Manchester ^ 
 
 NOTE TO THE FOURTH EDITION. 
 
 We are glad that the speedy demand for a reprint enables us 
 to make the few changes needed to bring our recommendations for 
 Greek into harmony with the Report of the Classical Association's 
 Committee, the principles of which were approved by the general 
 meeting of the Association at Cambridge last October. In the very 
 few points of practice upon which the discussion disclosed any 
 difference of view, we have been careful to state the different 
 alternatives and the degree of historical truth which they may 
 respectively claim. We have also slightly modified our attitude in 
 regard to the attempt to pronounce the Greek Accents (p. 18). To 
 the list of books consulted should now be added W. G. Hale and 
 C. D. Buck, A Latin Grammar, Boston, 1903. We are also 
 indebted to Prof. Walter Rippmann for comments on several points 
 concerning the pronunciation of modern languages. 
 
 Bangor 1 ^, , ;^^„ 
 
 ,, V March, 1908. 
 
 Manchester j ' 
 
THE PKONUNCIATION OF GREEK 
 AND LATIN. 
 
 It is proposed in the following pages that the method of 
 pronunciation which is almost universal in England objections 
 
 in the case of Greek, and still survives widely in EngUsh %l^- 
 the case of Latin, should be abandoned. This nunciatim. 
 method, like others which prevail locally in various parts 
 of Europe and America, chiefly found acceptance from the im- 
 mediate convenience of giving to the symbols of the Roman 
 alphabet when used to write Latin, and the corresponding 
 symbols in Greek, the sound which is most commonly denoted 
 by the former in the particular locality. But even this con- 
 venience is delusive, since the pronunciation of English varies 
 greatly in dififerent parts of the kingdom, and, as will be seen, 
 where the standard or London pronunciation is most carefully 
 inculcated, there the result is, in fact, furthest from the true 
 Latin sounds. A Scotch or Yorkshire lad will pronounce Latin 
 it much more correctly, and distinguish it far more clearly from 
 Latin a, if left to his own instincts than if trained by a cultured 
 teacher who adopts the unreformed method. Any method of 
 this kind altogether disregards historical accuracy, and ac- 
 
2 TEE PRONUNCIATION OF 
 
 customs the learner to fancy that languages exist as written 
 rather than as spoken : for he is put to no trouble to discover 
 the true sounds of the language he is studying. It is to him in 
 a very real sense a ' dead ' language : he ceases or never begins 
 to realise that by its help men and women lived, felt, and 
 thought : and is directly encouraged in a mistake which defeats 
 che very purpose of his education, the mistake of regarding 
 books as something remote from life rather than as an integral 
 part of the life of mankind and therefore of that for which he 
 himself is preparing. 
 
 But the local ' English ' method of pronouncing Latin and 
 Greek must be condemned also on the following more definite 
 grounds, which involve consequences smaller in themselves but 
 obviously and immediately mischievous : 
 
 (1) It confuses distinct sounds, and hence distinct words : 
 e.g. ceu and seu, caedit, cedit and sedit, caecae, caesae and sese, 
 noscet, nosset and nocet, luceo, lusio and Lucio (to say nothing 
 of so-lutio), Kelrai and 'x^alrav, /caivo), Keivw and KLvat are 
 pronounced alike. 
 
 (2) It obscures quantity : mensis (abl. plur.) is pronounced 
 as mensis (gen. sing.), and very often mensd (nom. sing.) just as 
 mensa (abl. sing.) : mdlum (evil) and malum (apple) are made 
 alike, and so v^nit (present tense) and ve^iit (perfect). The 
 same confusion occurs in the case of Greek, though not to the 
 same extent. 
 
 These two defects largely conceal from the student the 
 musical and rhythmical beauties of the two languages. 
 
 (3) The learner acquires by ear at the very beginning 
 false views as to the relations of languages, and, in particular, 
 fails to recognise the intimate natural tie between Latin 
 and the Romance languages. Thus Latin a instead of being 
 pronounced as French a is made to sound more nearly like 
 French e, following the common value of English a. 
 
GREEK AND LATIN 3 
 
 In this way the historical study of language meets with a 
 needless obstacle even in tracing in a Romance language, such 
 as French, those words which are most immediately derived 
 from Latin. 
 
 (4) A somewhat similar suggestion has especial importance 
 in Wales. The sounds used in the Welsh language are on the 
 whole, and particularly as regards the vowels, of a simpler and 
 more primitive character than those of English: and their 
 expression in the written form is a permanent record of the 
 direct influence of Latin civilisation upon the Welsh people. 
 The * English ' method of pronouncing Latin tends to push out 
 of sight this important historical relationship. 
 
 In any attempt to frame a better system, two conditions 
 should be fulfilled. On the one hand, the scheme Conditions 
 
 proposed should present, if our knowledge can %^ 'ref^'- 
 secure it, at least a reasonable approximation to <^^'"^«cy «^^ 
 
 ^ ^ ease of ac- 
 
 the sounds which actually existed in ancient quirement. 
 times: and on the other, it should avoid placing any really 
 serious difficulty in the way of beginners in Latin or Greek. 
 For it must always be the principle of the study of these 
 languages that the learner shall, as soon as possible, begin 
 to read for himself the works of the great classical authors. 
 
 The progress of philological research has made it possible 
 to meet the first requirement. We can in the 
 
 Accuracy. 
 
 main reproduce with certainty the sounds actually 
 heard at Athens in the fourth century B.C. and at Rome in the 
 first. The margin of doubt that remains, though from the 
 scientific point of view it is considerable, is nevertheless, when 
 seen from the standpoint of the practical teacher, confined 
 within very narrow limits. 
 
 For example : some scholars may feel a doubt whether Latin 
 ^ more nearly resembled French (close) i in Hve (= Eng. ee in 
 queen) or Italian (open) i in civitd ; but that it was immeasur- 
 
 1—5 
 
 y 
 
4 TEE PRONUNCIATION OF 
 
 ably nearer to Eng. ee (though strictly considered this is a 
 diphthongal sound) than to the English diphthong % in line, 
 tide, etc. is clearly demonstrable and universally admitted. 
 
 Accordingly in drawing up the Tables, we have kept in 
 view the distinction between variations of greater and less 
 importance, and have inserted approximate illustrations of 
 some of the sounds from languages where precise equivalents 
 were not easily found: whilst we have tried to guard against 
 any misunderstanding by pointing to more exact equivalents in 
 other columns. We venture to hope that the use of a fairly 
 complete series of English, French and Welsh illustrations side 
 by side may enable even beginners to attain to an approximately 
 correct pronunciation of Greek and Latin, while incidentally 
 rendering some slight service to the teaching of modern 
 languages also. 
 
 In the second place, after careful discussion, and some years' 
 Ease of ac- experience, we feel that the scheme now proposed 
 qmrement. offers no difficulty that can reasonably be called 
 
 serious even to the English-speaking student : while those who 
 are familiar with spoken Welsh (or French) should find it far 
 easier than the local English method. We do not underrate 
 the embarrassment which results from the fact that the pro- 
 nunciation of Latin and Greek here advocated is still unfamiliar 
 to a large proportion of those who are engaged in teaching 
 these languages, and therefore lacks for them the attraction 
 which comes of early associations. But in the case of Latin 
 the weight of authority is now so strongly on the side of the 
 restored pronunciation that those who still adhere to the former 
 custom may fairly be asked to consider whether the cause of 
 Latin in our schools should be hampered by the continuance 
 of this controversy into another generation. Amongst the 
 bodies which are now united in advocating the reform we can 
 point to the Classical Association, the Oxford and Cambridge 
 
GREEK AND LATIN 5 
 
 Philological Societies, the Head Masters' Conference, the 
 Associations of Assistant Masters and Assistant Mistresses, 
 and the Modern Language Association. Further, the Board 
 of Education now requires the adoption of the restored pro- 
 nunciation in all schools in connexion with which it has 
 responsibility for the use of public funds, unless special reasons 
 are given to the contrary ; and it appears that about three- 
 quarters of these schools have now actually adopted it. 
 
 The reform of Greek pronunciation is naturally not so 
 advanced, but it appears to be immediately urgent that the 
 pronunciation of Greek and Latin should be brought into 
 general harmony. Hence, whilst we have explained as accur- 
 ately as we can the exact sounds of the period of classical 
 Greek, we refrain from advocating generally any changes other 
 than those which the Classical Association is prepared to 
 accept as immediately practicable. 
 
 In conclusion we venture to claim that the restored pro- 
 nunciation has the advantage not only in distinctness and 
 scientific accuracy, but also in simplicity and musical character. 
 Written explanations must always be to some extent tedious 
 and incomplete: but if once an oral tradition is established 
 amongst teachers, it will be maintained with no greater sense 
 of effort than must always be needed by those who, rejecting 
 slipshod and inartistic methods of speech, aim at inspiring the 
 most constant of our occupations with the instincts of force, 
 precision and beauty. 
 
 1—6 
 
TABLE OF GREEK SOUNDS. 
 
 The letters which give the equivalents in English, French 
 and Welsh are printed in heavy type. Examples enclosed in 
 square brackets contain only rough equivalents. 
 
 Greek 
 
 English 
 
 French 
 
 Welsh 
 
 a 
 
 footpsith, Siha 
 
 chaise 
 
 aJ)er, lla.n 
 
 a 
 
 fsither 
 
 B.me 
 
 t&d 
 
 /3 
 
 hut 
 
 hon 
 
 haner 
 
 7 always as 
 
 get, gone 
 
 gateau 
 
 gardd, gynt 
 
 except^ that 
 
 
 
 
 77^ 
 
 anger 
 
 
 Bangor (not as 
 
 - are as 
 
 - 
 
 
 mangor,tyngu) 
 
 7^^. 
 
 ^ahkle 
 
 
 llangc 
 
 h 
 
 [deny 
 
 dette 
 
 du 
 
 e (close ^) 
 
 [getr 
 
 et, ^t4 
 
 [ceginy 
 
 ? 
 
 [adze]' 
 
 
 
 r) (open e) 
 
 nearer to 6ear 
 than to ta.ke* 
 
 pere, il m^ne 
 
 hen' 
 
 ' (spirit us asper) 
 
 1 hoard 
 
 
 hardd 
 
 (spiritus lenis^ 
 
 ) denotes the absence of the i 
 
 spiritus asper in 
 
 initial vowels. 
 
 
 
 6 (see below) 
 
 
 
 
 I 
 
 hit' 
 
 veritd 
 
 dim' 
 
 i 
 
 queen 
 
 rive, eglise 
 
 hin 
 
 K 
 
 cat, come 
 
 eclat, cour 
 
 ci, coed 
 
 \ 
 
 let 
 
 lit 
 
 gweled 
 
 M 
 
 man 
 
 mere 
 
 mam 
 
 V 
 
 name 
 
 nom 
 
 nid 
 
 f 
 
 tax 
 
 fixer 
 
 bocs 
 
 (close 6) 
 
 cannot, consist 
 
 [monologue] 
 
 [colyn, ponty 
 
 TT 
 
 ipit 
 
 "parler 
 
 pen 
 
 p 
 
 [herring] 
 
 [cheri] 
 
 carreg 
 
 r 
 P 
 
 [tree] 
 
 theatre 
 
 rhwng 
 
 <r, 9 always as 
 
 salt, mouse 
 
 savant, russe 
 
 nos 
 
 except that 
 
 
 
 
 a^( 
 
 ' has heen 
 
 
 
 ayi are as 
 
 has gone 
 
 
 
 (Tfl[ 
 
 has made 
 
 - 
 
 
GREEK SOUNDS 
 
 T 
 
 [te/2, tillY 
 
 tete tan 
 
 V 
 
 
 du pain, lutte [North Welsh 
 pump] 
 
 V 
 
 1 ■ 
 
 
 pur [North Welsh 
 [German grun] cun] 
 
 ^\ see below 
 %) 
 
 
 
 ^ 
 
 la-pse 
 
 
 (o (open o) 
 
 nearer to ore 
 than to coke^ 
 
 encore pob* 
 
 The Aspirates 0, <^, ^. 
 
 There is some reason for thinking that in the fourth century 
 B.C. even in Athens the aspirates 6, </>, ;)^ contained more than 
 the pure combination of t'\-h (Eng. antldU), p-\-h (Eng. 
 uphill\ k-\-h (Eng. backhanded); that is, that though the h 
 was still clearly heard, the first part of the sound was no 
 longer a plosive (t, p, k\ but a plosive beginning to change 
 towards a fricative (e.g. in <^ something between the plosive 
 p and the full 'affricate ' pf). Further details will be found in 
 the Appendix. The Committee of the Classical Association 
 recommend for 6, </>, and ;j^ the sounds respectively of Eng. th 
 in thin, Eng. y, and Scotch ch in loch. 
 
 1 A further probable exception, of no great importance, is mentioned on 
 p. 13 below. 
 
 2 For the slight variation in English t, d from the dentals in the Romance 
 languages, see p. 12 below. 
 
 3 Greek e was a 'close' e (see p. 16 below), and may be more nearly heard in 
 the usual pronunciation of college, or in the old-fashioned pronunciation of 
 engine, engineer, entire, than in the approximate examples given in the table, 
 which contain a more 'open' sound. It was the short sound corresponding in 
 quality to the ei of efs, see below. 
 
 ^ For 77 and w the Committee of the Classical Association sanctions, on 
 practical grounds, the sounds of Latin e and o respectively (see p. 10), though 
 commending the open pronunciation in schools where it is already familiar in 
 other languages. Welsh e and o are open or * half-open' sounds, both when 
 short and long : but the degree of openness appears to vary slightly in different 
 districts. See p. 16 below. 
 
 5 But in the fifth century b.c. f had a sound like English zd (e.g. in glazeii). 
 It remained always a double consonant, making the preceding syllable long 
 by position. 
 
 * See note 4 on p. 11 below. 
 
GREEK SOUNDS 
 
 Diphthongs ending in l. 
 
 These should be pronounced simply by combining the 
 sounds of their component vowels. Thus 
 
 aL = a-\- i. Roughly as Eng. I in ride^ more exactly Fr. ail in 
 
 email, Welsh ai in taid. 
 01 — o -\- L. Eng. oi in oil, Welsh oi in troi. 
 vt = v-\- 1. Roughly as Fr. ui in lui. It rarely occurs save 
 
 before vowels and then v has its regular sound 
 
 and the t is simply equivalent to the Eng. 
 
 consonant y. 
 a, y, ft) = a + fc (Welsh ae in caer\ rj-{-i, (o + 1 (Welsh oe in coed) 
 
 respectively. 
 
 The t was probably not pronounced at all in a and « after about 
 200 B.C., just as in modern spoken S. Welsh ae and oe are pronounced (in 
 Glamorganshire) simply as Welsh a and o. 
 
 The sound of ei was originally diphthongal (Welsh ei, half- 
 way between Eng. ^ay and ^ie), but by about 450 B.C. it had 
 become simply equivalent to a long e (' long close e') = Eng. a 
 in day, pronounced without the final y-sound. It was however 
 distinguished from tj down to at least the 2nd century B.C., and 
 from 7) down to at least the end of the 4th century; and if 
 there is any risk of the sounds being confused, it may be 
 necessary to allow et to be pronounced as Eng. eye, though this 
 is certainly incorrect. 
 
 Diphthongs ending in v. 
 
 These should be pronounced by combining the sound of 
 their first vowel with that of Latin u (= Eng. u in full, Welsh 
 w in cwm), not with that of the Attic v, which when it stood 
 alone had undergone a change that it had resisted when pre- 
 ceded by another vowel. 
 
 Thus 
 
 av = a-^ Latin u, as Welsh aw in llawn, nearly as Eng. ou in 
 round, ow in fowl. 
 
GREEK SOUNDS 9 
 
 ev — € + Latin u, as Welsh ew in mewn. The nearest equivalent 
 existing in English is ew in new, but the first 
 element of this is an i- instead of an e-vowel. 
 
 (dv = a) + Latin u, but is of rare occurrence. 
 
 The sound of ov was originally diphthongal (Eng. o in stone, 
 see p. 16), but it became in Attic before 450 B.C. equivalent to 
 a long o (' long close o,' French o in chose), and then, becoming 
 still closer, to a long Latin u (Eng. oo in moon, Welsh w in gwr). 
 
 Accent. 
 For the value of the Greek signs of accent see below, p. 18. 
 
 Note. 
 
 The exact nature of the sound or sounds represented in 
 early Attic writers by (T(t, in later writers by tt, has not yet 
 been precisely determined, though it is practically certain that 
 it denoted some sound closely akin to those which in English 
 we write sh (French ch) and ch (Ital. c before i and e), see 
 Witton, Am. Journ. Phil. xix. p. 420 and Postgate, Classical 
 Review, Dec. 1906. Until a more decisive result is reached, we 
 must be content to give the symbols the sound of a double a 
 and a double r respectively; such inaccuracy as we may be 
 committing in this pronunciation is as great as, but no 
 greater than, the corresponding inaccuracy in spelling allowed 
 by the Athenians themselves. 
 
TABLE OF LATIN SOUNDS. 
 
 The letters which give the equivalents in English, French 
 and Welsh are printed in heavy type. Examples enclosed in 
 square brackets contain only rough equivalents. 
 
 Latin 
 
 English 
 
 French 
 
 Welsh 
 
 a 
 
 footpath, a,ha 
 
 chs^se 
 
 Siber, lla.n 
 
 a 
 
 fsither 
 
 kme 
 
 tsid 
 
 b 
 
 hut 
 
 hon 
 
 haner 
 
 bs as ps, urhB as urps 
 
 
 
 c always as 
 
 cat, "kitten 
 
 eclat, cour 
 
 coed, ci 
 
 d 
 
 [deny 
 
 dette 
 
 du 
 
 e (open ^) 
 
 get 
 
 berger 
 
 pen 
 
 e (close e) 
 
 [tay, bsLneY 
 
 passee^ 
 
 [henf 
 
 f 
 
 fox 
 
 fifre 
 
 Son 
 
 g always as 
 
 get, gone 
 
 gateau 
 
 gardd, gynt 
 
 h 
 
 hoard 
 
 
 hardd 
 
 i 
 
 hit 
 
 4 
 
 dim 
 
 I 
 
 queen 
 
 rive, eglise 
 
 hin 
 
 i consonant 
 
 yoke 
 
 bataillon 
 
 iaith 
 
 1 
 
 let 
 
 lit 
 
 gweled 
 
 m 
 
 man 
 
 mere 
 
 mam. 
 
 n 
 
 name 
 
 nom 
 
 nid 
 
 n before c," 
 
 &, q 
 
 
 
 
 song^ 
 
 
 llangc 
 
 o (open o) 
 
 dot 
 
 reconnu 
 
 pont 
 
 o (close o) 
 
 [low, loney 
 
 chose, chaxide 
 
 [pob]' 
 
 P 
 
 pit 
 
 parler 
 
 pen 
 
 qu 
 
 quiz 
 
 quoi 
 
 
LATIN SOUNDS 
 
 11 
 
 r [herring] 
 
 [cheri] 
 
 carreg 
 
 s always as hiss, pace, manse 
 
 savant, russe 
 
 nos 
 
 t [tiny 
 
 tete 
 
 tan 
 
 u pyill, wood 
 
 nouvelle 
 
 cwm 
 
 fl pool, wooed 
 
 resoudre 
 
 cvrn, gvrr 
 
 u consonant® vrine 
 
 Nord-Ouest 
 
 (g)vreled 
 
 X tax 
 
 fixer 
 
 bocs 
 
 The following sounds, used in Latin in words borrowed from 
 the Greek, are to be pronounced as in Greek : 
 
 Latin 
 
 Greek 
 
 y 
 
 V 
 
 y 
 
 V 
 
 z 
 
 ?\ 
 
 ch 
 
 % 
 
 ph 
 
 <!>' 
 
 th 
 
 ^1 
 
 rh 
 
 p 
 
 English 
 
 see p. 7. 
 
 [^ree] 
 
 French 
 du pain 
 pur 
 [German grun] 
 
 Welsh 
 [North Welsh pump] 
 [North Welsh cun] 
 
 theatre 
 
 rhwng 
 
 ^ Latin and French t, d are more strictly dental than the corresponding 
 English sounds. 
 
 2 Latin e may be obtained from English a in ba.y, ba.ne by omitting the i 
 sound which follows immediately on the English vowel : and so o from English 
 in low, lone by omitting the u sound. English air, oar give the corresponding 
 open vowels. French ee in pass6e is a close vowel, and was formerly long, 
 though in modern pronunciation it is short. See further p. 16 below. 
 
 3 Welsh e and o are open or * half-open ' sounds, both when short and long : 
 but the degree of openness appears to vary slightly in different districts. 
 
 4 Latin i differs slightly from the corresponding Greek vowel : it is an open 
 sound, as are the English and Welsh equivalents. To this sound there is no 
 precise equivalent in French. On the other hand, the Greek vowel is close, and 
 corresponds exactly to French i, e.g. in veiAte. 
 
 5 Accordingly Latin ng corresponds to the double sound {'ng' + g) in English 
 finger, and not to the single sound heard in singer: to Welsh ng in Bangor, 
 not in tyngu, 
 
 8 A distinctive symbol v is still often found in Latin printed texts, althongh 
 j for the i consonant has been generally abandoned. These distinctive symbols 
 for the consonants came gradually into use towards the end of the Middle Ages, 
 when the sounds had undergone considerable change (see p. 14), and had become 
 comparatively remote from those of the corresponding vowels. 
 
12 LATIN SOUNDS 
 
 Diphthongs. * 
 
 These should be pronounced simply by combining the 
 sounds of their component vowels. Thus 
 Latin English French German Welsh 
 
 ae^ side email, hat^jllon t?dd 
 
 oe 6oi^ hUMser troi 
 
 ui [ruiri] [oui] [mwyn] 
 
 (The vowel-sounds in ruin, etc., if run closely together, will 
 correspond to Latin ui in Auic, cui.) 
 
 au cow /iau5 ^/awn 
 
 eu [new] mevrn 
 
 (English new would be in Latin letters niu: but if we 
 substitute the sound of e for that of i, we obtain the Latin 
 diphthong.) 
 
 1 See p. 17. 
 
 EXPLANATIONS OF SOME POINTS IN PHONETICS. 
 
 .» 
 
 These remarks are intended as comments upon the tables of 
 Greek and Latin pronunciation, amplifying the very 
 brief directions there given. They aim only at dealing 
 with certain practical difficulties, and not at presenting even the 
 elements of the Science of Phonetics. Technical terms in common 
 use are therefore not, as a rule, explained. Fuller information may 
 readily be found in the recognised text-books of Phonetics (see the 
 list of authorities in the Preface). 
 
 The plosive sounds or stops are found written 
 
 onsonan s. q^^^q^^ uniformly in modern European languages : 
 
 those common to Greek and Latin with them are 
 
 7rp, T t^ KG (breathed sounds) and /S b, S d, y g (voiced 
 
 b. voiced. , v 
 
 sounds). 
 
 But t and d are more strictly dental in the modern Eomance languages 
 than in English or German, where they are formed rather above than on 
 the teeth ; and in this particular the Romance and Welsh sounds cor- 
 respond more closely to Greek r 8, Latin t d. 
 
 Besides these six sounds Greek possessed aspirates, and Latin 
 
 used them in words borrowed from Greek : ch ph. th, 
 
 c. aspirate. t mi • . n i . \ ^ 
 
 X ch. These also were originally plosive sounds, only 
 
 differing from the corresponding breathed sounds tt p, r t, k c 
 
EXPLANATIONS 13 
 
 respectively in adding a slight emission of breath, in sound like the 
 
 English A, before the next vowel or consonant in the word. They 
 
 were gradually replaced by the continuous (fricative or 'spirant') 
 
 sounds which appear in modern Greek, as well as in other 
 
 European languages, such as ph =/ in English, French and Welsh, 
 
 th in English and Welsh, ch in Welsh and German. On the date 
 
 at which this change began in Attic Greek, see the Appendix. 
 
 These fricatives are not the only modern developments of the 
 
 plosives of classical times : and in English and other 
 
 ModcTTi 
 modern European languages (though not in Welsh) values {in 
 
 c, g, t have come, in certain positions, to represent ^^^"* cases) 
 
 corresponding continuous sounds, for which other 
 
 symbols also stand. Thus English city^ gin, nation are pronounced 
 
 just as if sity, jin, nashun were written. These secondary values 
 
 of. modern c, g, t are in no case to be given to the corresponding 
 
 Greek or Latin symbols. 
 
 Amongst continuous sounds the nasals m and n corresponding 
 
 to the voiced h and d are found in all European 
 
 languages : but observe that Greek v and Latin n consonants^ 
 
 may perhaps have more closely resembled the sound 
 
 heard in modern French and other Romance languages, and have 
 
 been more strictly dental than English, Welsh, or 
 
 m^ . ,. . i. 1 (a) Nasals. 
 
 German n. The nasal corresponding to g is found 
 
 in English and Welsh, and written ng. In Latin and Greek the 
 
 sound is heard before the sounds k c (q), y g, x ^^^ (^^ Greek probably 
 
 also before //, and i/, though certainly not before m and n in Latin). 
 
 It is denoted by y in Greek, by n in Latin. Hence Greek yy (for 
 
 example) = Latin ng = English or Welsh ng + g, or ng in the examples 
 
 given in the tables. 
 
 Final m was pronounced lightly in Latin. In verse, when the 
 
 next word began with a vowel, we find the vowel before -in elided : 
 
 yet when the next word began with a consonant, the syllable ending 
 
 in -m is counted long by position. The best explanation seems to 
 
 be that the -m was so far weakened, that when a vowel followed, the 
 
 -m was only heard as a nasal afiFection of the vowel before it, such as 
 
 is given to o and a in French hon, plan : and therefore the vowel 
 
 was subject, like others, to ehsion. The exact sound before following 
 
 consonants is doubtful (see Hale and Buck, Lat. Gr. § 34). For class 
 
 purposes it seems sufficient to pronounce m as in English, but rather 
 
 more lightly when it occurs at the end of a word. 
 
14 EXPLANATIONS 
 
 There is some variety in modern languages' in the sounds denoted 
 
 by r and I: but in most they are voiced sounds clearly 
 
 pronounced, and r is trilled. Yet English r is always 
 
 a weak sound, and often entirely inaudible : and even English rr 
 
 represents only imperfectly the full sound of Greek p, Latin r. 
 
 A breathed sound corresponding to the voiced r is found in 
 Greek p, Latin rh (in words borrowed from Greek or Celtic), Welsh 
 rh and French r in certain positions \ as in theatre. This sound can 
 often be recognised in the pronunciation of the English tree. 
 
 In most modern languages the symbols s z are found to represent 
 
 a breathed and a voiced sibilant respectively. But 
 
 whilst Welsh on the one hand possesses the s sound 
 
 only, English on the other frequently employs the voiced or z 
 
 sound, even where s is written : so that lies comes to rhyme with 
 
 size., and cheese with freeze : whilst the breathed sound (when final) 
 
 is often represented by c or s, as in pace, manse. Greek a (except 
 
 before /S, y and /x) and Latin s always, represent the breathed 
 
 sounds, and care should be taken to give them this value even 
 
 when final ; for example, the endings of ovtods and reges should not 
 
 be pronounced as in English toes and gaze, but as in dose and chase. 
 
 In Latin a consonant as well as a vowel value was given to each 
 
 of the symbols I, V : but that the consonantal sound 
 
 could not have differed widely from the vowel in 
 
 either case is shown by the easy passage from one to the other; 
 
 as, for example, in siluae (silvan) and siluae (siluae) ; neue (neve) 
 
 and neu; and by the well-known play on words between caue ne 
 
 eas {cave ne eas) and cauneas 'figs from Caunus.' 
 
 The sounds given to j in English and French respectively, and 
 to V in most European languages (Welsh f), are historical develop- 
 ments of Latin consonantal i (j), and u (v), as well as of other 
 sounds : but they are much later in date than the classical period, 
 and should not be introduced in reading classical Latin. 
 
 The number of distinct vowel sounds used in modern European 
 languages is very considerable, whilst the number of 
 symbols available is relatively much smaller than in 
 the case of consonantal sounds. Accordingly in all modern European 
 languages, but very especially in English, the written representation 
 of vowel sounds is entirely inadequate. Not only is one symbol 
 used to denote several distinct sounds, but it is quite common to 
 
 1 See Victor, p. 132, § 93 and Anmerkung 1. 
 
EXPLANATIONS 15 
 
 find a double or diphthongal sound represented by a single symbol, 
 and a single sound by a double symbol. 
 
 If however we put English (and Dutch) on one side, we find 
 that in modern languages generally an approximately uniform 
 character is given to the sounds represented by a, e, i, o. These 
 sounds are single vowel sounds, and we have every reason to 
 attribute them to Latin a, e, i, o, and to Greek a and i respectively. 
 Latin u corresponds to Greek ov = French ou (for Greek € and o, 
 see below). Greek v, Latin y, seems to correspond most closely to 
 French u : and sounds somewhat resembling this are found in 
 German and Welsh. Most of these sounds occur in English also, 
 but the symbols by which they are denoted are not uniformly 
 employed, and often have a value peculiar to this one language. 
 Accordingly it is better to start from French and Welsh than from 
 English in studying the character of the Greek and Latin vowels. 
 
 The distinction between short and long vowels was more plainly 
 
 marked in the classical languages than in modern 
 
 ° . Quantity. 
 
 English : and this difference must be kept entirely 
 
 distinct from that between short and long syllables ; thus in co-Trepos, 
 Vesper, the first vowel in each case is short, the first syllable 
 long. In a reformed pronunciation the distinction between short 
 and long vowels should be carefully observed : and the student 
 should endeavour to master the quantity of the vowels in new 
 words in Greek and Latin, as he learns the words themselves. But 
 where in Latin a vowel is followed by two consonants, its quantity 
 is only known to us in a certain number of cases ^ : e.g. vowels are 
 long before nf, ns, as in Injuns : in other positions we consider that 
 the only practical course is to follow the English method of making 
 the vowel short. 
 
 Further distinctions can be traced with the help of English 
 illustrations : but to use these it is necessary first to 
 understand the nature of the sounds represented in ^^j^^^ "^ " 
 English by the vowel symbols and those used in con- 
 nexion with them. For shortness' sake we may denote the values 
 given to a, e, % o, u in Latin, as stated in the table above, as the 
 
 1 See Lindsay, Latin Language, p. 133 ff. ; and observe that the difference 
 of quantity in Latin produced differences of quality in the derived vowels in 
 Eomance ; thus Lat. sthlla gives Fr. etoile, Ital. stella with close e ; while Lat. 
 imum 'beautiful' gives Fr. hel, Ital. 6eWo with open e (Meyer-Liibke, Gram. 
 Lang. Rom. i. pp. 91 and 153). 
 
16 EXPLANATIONS 
 
 normal values of these symbols. English ' short a, e, t, o have 
 then, in the examples given as equivalent to Latin a, e, ^, o, 
 the normal values. 
 
 It is, however, not easy to find clear examples of normal short a 
 in English : the English short u (as in 6u^, Uncertain), does not 
 greatly differ from it, but a more exact equivalent is found in the 
 first element in each of the English diphthongs heard in mine 
 (main), how (Aau). The English spellings er in better, e in the 
 man, a in Sigain, alike represent an indistinct vowel sound, which 
 is chiefly found in unaccented syllables, and is denoted in phonetics 
 by e reversed (a). It closely resembles French e in me, te, and 
 Welsh y in yn, yma. 
 
 English long a and o in most cases represent a long vowel sound 
 followed by a slight i and u sound respectively : this sound is 
 written as a consonant in 5ay, low, and, though unwritten, is none 
 the less heard in ba.ne (bein), stone (s^oun). Hence English long a 
 may be represented by ei, long o by ou. Although in each case two 
 distinct sounds exist, it is not easy without practice for an Englishman 
 to pronounce the first separately. 
 
 English long i in most cases would, on the same principle, be 
 denoted by the combination ai (a -hi), and long u by yu : from the 
 latter normal u can easily be deduced, as it is common in English 
 under the form oo. 
 
 English ai, ea, when followed by r are equivalent to normal e 
 lengthened : e.g. in air and bea.r : and so oa to normal 6 lengthened, 
 in oar. In these words the final r is but slightly heard, and nearly 
 equivalent to the indistinct vowel sound, so that air (English) or 
 (b)ear (English) is little more than ed (normal), and oar (English) 
 than 09 (normal). 
 
 English ee = ea in Zean^ normal i followed by 9. 
 
 The short vowels described^ are uniform in quality, and are 
 
 known as ' open ' or ' wide ' vowels (see below). But 
 
 Open {wide) i\^q lonff vowels are of two kinds. Some differ from 
 and close (nar- 
 tow) vowels. the short vowels just mentioned only in quantity or 
 
 duration, and these are open long vowels. Such 
 
 are e in English air (eo), beSir (ed), French phre, il mhne, Greek ^v 
 
 and Welsh hen (in many districts) : o in English oar {od), bore (od), 
 
 Greek wkv9, Welsh pob in many districts, Italian popolo. 
 
 But more often the long vowels differ also in quality : e being 
 
 ^ Observe that Greek e and o are close and were therefore omitted above. 
 
EXPLANATIONS 17 
 
 somewhat nearer to i than e is, and so o to u. [This may readily 
 be felt by pronouncing in succession (1) a, e lengthened, e, I ; (2) a, 6 
 lengthened, o, w.] Thus we obtain long close (or narrow) vowels, 
 so called because the channel through which the stream of voice 
 passes is specially narrowed in their formation ; e.g. e in English 
 6a?/, {hei), ba,ne (bein), German wer : this sound occurs in Greek ets 
 (at 450 B.C.), Latin sedi. Similarly 5 in English bone (boun), French 
 chose, Latin nonus, is formed with ' narrowing of the voice.' The 
 same tendency can be observed (but less clearly) in i in French rive, 
 and uin English rVide, etc. 
 
 So far as e and o are concerned, the distinction is of some 
 practical importance in Greek and Latin. Greek distinguishes in 
 writing both the short and long close vowels c and et, o and (in 
 early Attic ^) ov from the open long vowels {-q and w). Latin e and 
 o were close; but in the fourth century a.d. ae had come to be 
 pronounced as the open long vowel corresponding to e, so that equus 
 and aequus differed only in length of the first vowel. The change 
 seems from Yarro, Ling. Lat. vii. 26, to have begun even in classical 
 times : but this passage does not give us enough data to determine 
 exactly how far it had then gone : and we have therefore recom- 
 mended in the table that ae should be pronounced as a diphthong, 
 since it is of great importance that the syllable ae, so common in 
 inflexional endings, should be at once recognisable in the oral work 
 of a class. 
 
 In vulgar Latin the classical distinctions of quantity were much 
 obscured by the effect of the stress-accent, with the result that (close) e 
 and (open) i became identical in pronunciation, and similarly ^ (close) o 
 and (open) u ran together. This development is of importance in tracing 
 the connexion between Latin and the Komance languages. Thus the 
 following examples show the normal representation of the Latin vowels 
 e e i I, occurring in accented syllables, in modern French. 
 
 Latin 6 = French ie : pedem, pied ; hen, hiet: 
 
 el . , jtres, trois ; habere, avoir 
 
 I) ' [viam, vole ; /idem, foi 
 
 i i : filum, fit ; venire, venir. 
 
 Similar distinctions are found in the history of Latin o 6 u u, see 
 Meyer-Liibke, Grammaire des Langues romanes, §§ 118 ff'. and 183 ff. 
 
 1 It is probable that even as early as the classical period ov had become still 
 closer, and = Lat. u. 
 
 2 Except in one Eomance language, see Meyer-Liibke II. cc. 
 
X 
 
 18 ACCENTUATION 
 
 English ee, ea are usually followed by an indistinct vowel sound 
 before r : the r is also sounded except when final, so 
 
 ea {i) before r. ^^^^ peeress, dearest, are pronounced almost as pwres, 
 didrest. This should be avoided in Latin : e.g. audirem 
 
 should be pronounced (save for accent), as audi rem, not as 
 
 audiQrem : for the double sound would have been represented in 
 
 Latin by ie as le in fterem. 
 
 The short i sound is favoured in English at the end of words, 
 as in lady, cheery : but this sound should not be 
 
 V iiy. ** introduced for Latin final e, as in the English pro- 
 
 nunciation of triste, posse. Except for the different 
 
 position of the break at the end of the word parce tibi and parcel 
 
 ibi are pronounced alike. 
 
 Accentuation. 
 
 There is no doubt that in the Classical period of Greek 
 the accented syllables were marked by a higher 
 Greek Accent, pitch or note than the unaccented, and not by 
 more stress, not, that is, with a stronger current 
 of breath and more muscular effort. Therefore, unless the 
 student is capable of giving a musical value to the Greek 
 signs of accent, it is doubtful whether he should attempt to 
 represent them in pronunciation ; for in many cases we should 
 make our pronunciation more, not less remote from that of the 
 Greeks themselves if we gave to their accented syllables the 
 same stress as we do to the accented syllables in English; 
 for example, in paroxytone dactyls (/ce%p7y/^ei^09) when the 
 penult is stressed, the quantity of the long antepenult is apt 
 to be shortened and its metrical value destroyed \ But where 
 there is no conflict between accent and quantity {a^aOo^), 
 something may be said for stressing moderately the accented 
 syllable, and so distinguishing e.g. KaXaj<; and koKco^, Al6<; 
 and Btof;, ravrd and ravra. 
 
 1 This had actually happened in spoken Greek at least as early as the 
 2nd century a.d., as is proved by the frequency of the confusion between w and o 
 on inscriptions {^AvTiyobvip for older ^Avriyovip, ^Xopov for ^\d}pov, and such 
 prosody as Ner/co/x^S^s). See, e.g., Meisterhans-Schwyzer, pp. 24 and 68. 
 
ACCENTUATION 19 
 
 In Latin the accent had undoubtedly a musical character, 
 but it also implied stress, though not such forcible 
 Latin Accent, stress as in English, nor such as involves any slurring 
 or indistinct articulation of the unaccented syllables. 
 The student should therefore be careful not to shorten the vowel 
 of the syllable which precedes the accented syllable in such 
 words as audiehdmus, amdbdtur. The most important laws are 
 (i) that words of two syllables are accented on the first, (ii) that 
 words of three or more syllables are accented (a) on the last 
 syllable but one if that is long, (6) on the third syllable from 
 the end, if the last but one is short. These laws are correctly 
 observed in the ' English ' pronunciation : e.g. dmat, vituperdre, 
 regerem, compedibus. Welsh-speaking students, accustomed in 
 almost all words to accent the last syllable but one, need to 
 be careful in the two instances given last, and to avoid such 
 mispronunciations as regerem, compedibus. 
 
 It is necessary to guard the English student by pointing out 
 that the Greek and Latin vowels possess the qualities 
 which have just been described in whatever position Quality in 
 
 unaccented 
 
 of the word they may occur ; for instance the three syllables. 
 syllables of e^ere and the three of regere should 
 be pronounced with exactly the same respective vowel-sounds, 
 € and e. But in English almost all vowels in unaccented 
 syllables are pronounced (except in special cases, where the 
 nature of the following sound affects the vowel) simply as the 
 indeterminate, colourless vowel d which was described above ; 
 as in villa., better, the wind, author : in careless speech even i 
 in authority, etc. is given the same sound. The student should 
 distinguish the vowels in the unaccented syllables of tlvu, rive, 
 mensa,, imber, turtur, as well as those in accented syllables like 
 vir, ver, fur. 
 
APPENDIX. 
 
 NOTE ON THE GREEK ASPIRATES. 
 
 What was the pronunciation of 0, <f), and x ^^ Athens in the 
 fourth century B.C.? It cannot be said that the evidence is full 
 enough to provide us with a complete and precise answer to this 
 question. How much is certain and how much we must for the 
 present be content to regard as matters of probability merely will 
 best appear from a summary of the chief considerations which bear 
 upon it. The general limits of our knowledge are stated very clearly 
 by Brugmann {Grundriss d. Vergleichenden Sprachwiss. ed. 2, Vol. i., 
 p. 655) from whom the following sentences may be reproduced : 
 
 " The Tenues Aspiratae of pro-ethnic Greek ph, th, kh, no doubt 
 remained unaltered in most dialects down to a period within the 
 historical epoch. In course of time they were ultimately converted 
 into Breathed Spirants (/, ]> (Eng. th in thin), ch (in German or 
 Welsh), probably through the intermediate stage of Aflfricatae 
 (pf, t]}, kch). But we have not now the means of tracing precisely 
 in its various stages at different times and in different localities the 
 gradual progress of this conversion. The different dialects did not 
 
 all advance in this matter at the same rate The genuine 
 
 Aspiratae were maintained best and longest in Attic. This is 
 especially clear from the appearance of such forms as cxw [instead of 
 €^(0), see below] with the first syllable aspirated through assimilation 
 to the second ; from the representation of <f>, 6, Xj ^J P) ^> ^ re- 
 spectively in Latin of the early period [e.g. purpura = -n-opcfivpa, 
 tus = 6vo^, calx = xaXi^] and by ph^ th, ch in the writing of the 
 
 educated class at Rome from about 100 B.C. onwards In Egypt 
 
 even in the second century a.d. they were still aspirates, except only 
 6 before t [as is shown by their transcription into Egyptian 
 characters], see Hess, Indogerman. Forschungen 6. 124." To this 
 Brugmann adds a few cases in which the fricative pronunciation is 
 
APPENDIX 21 
 
 proved for certain dialects (in the third and second centuries B.C.; 
 see his Griech. Grammatikj ed. 3 (1900) p. 105 where some further 
 details are given). 
 
 Can we hope to add anything to Brugmann's discreet silence as 
 to the earliest date at which the Affricate stage was developed in 
 Attic Greek 1 
 
 It is acknowledged by all scholars that the sounds in Attic of <f), 
 6, X ii^ ^^® sixth century B.C., when these symbols themselves had not 
 yet been completely established in use in place of the older fashion^ 
 of writing ttA (FlB) etc., and when in metre they never made a 
 preceding vowel long by position, were pure aspirates. Now^ let vis 
 ask first : 
 
 A. What grounds are there for supposing that these sounds 
 remained unchanged in Attic till the fourth century B.C. 1 
 
 1. The most important evidence is that to which Brugmann 
 briefly refers in the passage just quoted, namely the very frequent 
 mis-spelling of words containing aspirates in the fifth and 
 fourth centuries B.C. in Attic inscriptions, the record of which is 
 collected in Meisterhans-Schwyzer, Grammatik d. Attischen Inschrr. 
 ed. 3, 1900, p. 102. But to interpret the evidence correctly it is 
 necessary to sift it a little more closely than has yet been done. 
 
 (a) Such spellings as exoi for €xw show quite clearly that the 
 Aspirates contained an h sound which careless speakers added to 
 other syllables in the same word. 
 
 (b) Such spellings as ;j(t^ajv for ;)(it(joV and cvopxovvTL for evopKovvri 
 show that the combination of this h sound with a preceding plosive 
 was written (by the authors of the particular insc.) by means of the 
 aspirate symbol, r + h being written in x*-^^^- 
 
 (c) Such spellings as kiOwv (and sometimes kitwV) show that when 
 an h was subtracted from an aspirate the result was written (by the 
 authors of the particular insc.) as a plosive, k instead of the x which 
 had lost its h to the following syllable. 
 
 Now if we had only such cases as (c) we could maintain, I think, with 
 absolute confidence, that they established a quite enormous probability in 
 favour of the view that at this date K + h was the only value of x. and so with 
 6 and <f>. 
 
 1 On this see Kirchhoff's Studien zur Geschichte d. griech. Alphabete (ed. 4, 
 1887, e.g. p. 172) or Koberts' Greek Epigraphy, i. almost passim, i.e. wherever 
 the history of <f> and x is discussed in the various localities ; for <p<T, xo" io the 
 old Attic alphabet see Meisterhans-Schwyzer, p. 3. 
 
22 APPENDIX 
 
 But the examples like (a) where the h is added to another syllable, but 
 leaves the x unaffected in its original place, do not exclude (though of course 
 they do not directly support) the supposition that in the pronunciation of the x 
 (or or 6) some slight fricative sound had already crept in after the 
 plosive and before the h. And examples like (6) do not necessarily exclude it 
 either, because while it is clear that a fricative pronunciation is not likely (at all 
 events not nearly so likely as the /i-sound) to be transferred from one syllable to 
 another, yet the fact that d was used in xt^wj' (being the only available symbol) 
 to denote merely an h added to r does not amount to an absolute proof that it 
 did not commonly denote r + a faint th + h. 
 
 Now from the fourth century and later Meisterhans-Schwyzer gives 15 ex- 
 amples of (c) to 11 of (6) : in the sixth and fifth centuries there are 11 of {a), 10 
 of [h) to only 2 of (c). Altogether the proportion in his examples is 
 
 11a : 216 : 17c. 
 
 In other words the examples of (a) and (6) together are nearly twice as 
 numerous as those of (c), in spite of the fact that no examples of (a) can 
 occur after 403 B.C. since the symbol for the h went out of use with the rest 
 of the Old Attic alphabet in that year, when the Ionic alphabet was adopted 
 by law^. These numbers are significant 2. 
 
 Subject, however, to these comments it is clear that from these Attic 
 spellings taken alone we should judge the pure aspirate pronunciation dis- 
 tinctly more probable for the fourth century in Athens. It is worth noting, 
 however, under (c), that while there are 7 examples of x losing its aspirate and 
 appearing as k, and 4 of ^ appearing as r, the only words in which appears as 
 IT are Uwacpopos and Il€pa-€<p6p7), there being three examples of each. 
 
 2. The development of ovOeU (fern. ovSe/Aia) from ov8' ct? and 
 the occasional spelling ovO* ot for ov8' ot (see Meist.-Schwyzer, 
 pp. 104, 258). These are roughly parallel to the examples under 
 (b) just discussed. 
 
 3. The writing in early Latin of words derived from Greek 
 words containing aspirates, like calx 'cup' from x<^^'^- This re- 
 presentation is not exact, but it clearly suggests that there was a 
 definite plosive element in the Greek aspirates. Observe that the 
 examples of this are very numerous and that a large number of 
 them are of the class that came over uiua uoce, by actual spoken 
 intercourse, e.g. purpura, apua, tus, calx. 
 
 ^ It is well to state that I have regarded xf-^^^^ i^oXxVj J^o.Xx'^Sojv and 
 <^ep<r€^6vT] as the normal forms, and have only counted variants from them. 
 
 - For statistics it seemed best to start from Schwyzer's examples, though his 
 collection is not complete. He omits all examples from Vases, referring merely 
 to Kretschmer's Gr. Vaseninschriften, pp. 149 ff., 156 ff. Those that Meisterhans 
 has included from this source (in his second edition, 1888) raise the examples of 
 (b) to 33 instead of 21 ; for instance, the signature of the vase-painter Xaxpv\lo}v, 
 which I have only counted as one of the 33 examples, though it occurs on a 
 number of vases {KaxpvXicav is less frequent). 
 
APPENDIX 23 
 
 4. The adoption of ph, th, ch by educated Romans, beginning 
 about 133 B.C., and regular from 100 B.C. onwards, in transliterating 
 Greek words, or spelling words derived from Greek in which the 
 aspirate was customarily pronounced, as in Corinthua = Kopivdos. 
 Cicero's very interesting remark^ (^Orator § 160) shows that an h 
 sound was clearly heard in pulcher and triumphus, both words 
 borrowed from Greek; but it does not necessarily follow that 
 there was no fricative element as well. He clearly reflects the 
 instruction of his teacher. Priscian prescribes a pure aspirate 
 pronunciation {p + h, etc.) six centuries later, when it is absolutely 
 certain that this represented only grammatical tradition, not the 
 pronunciation of his day. The statements of Diomedes (fourth 
 cent. A.D.) show conclusively that by that date (f>, 0, x were mere 
 fricatives. See Lindsay's Latin Lang. pp. 58 and 100. 
 
 5. Finally we have the evidence of the transcription of Greek 
 aspirates into the alphabets of non-European languages, e.g. on some 
 Graeco-Indian coins of the second century B.C. (Bendall, Proceedings 
 of the Gamb. Philol. Soc. Nov. 12, 1903, in the Camb. Univ. Eej)orter 
 of Nov. 24), and in Egyptian papyri of the second century a.d. (cited 
 above by Brugmann). Now in both cases the Greek <^, 0, x '^^^ 
 represented by the symbols of the pure aspirate sounds of the 
 languages into w^hich the words are borrowed. 
 
 This gives us reason for thinking at least that the h sound clearly 
 survived in Greek, and probably also some part at least of the plosive 
 element. But it cannot be pressed so far as to prove that there was 
 no other element in the Greek sound ; any more than the fact that 
 the spirant f of Latin and the other languages of ancient Italy was 
 always and only represented in the Greek alphabet by ^ (e.g. 
 ^\<xovLo<i = Flauius) proves that ^ had exactly the sound of /; 
 
 1 Quin ego ipse, cum scirem ita maiores locutos esse ut nunquam nisi in 
 uocali aspiratione uterentur, loquebar sic, ut pulcros, Cetegos, triumpos, Carta- 
 ginem dicerem. Aliquando, idque sero, conuicio aurium cum extorta mihi 
 ueritas esset [i.e. 'when the pronunciation I heard all round me broke down 
 my obedience to the grammarians' rules which I had been taught '], usum 
 loquendi populo concessi, scientiam mihi reseruaui. Orciuios tamen et Matones, 
 Otones, Caepiones, sepulcra, coronas, lacrimas dicimus quia per aurium iudicium 
 semper licet [i.e. *in these latter words the addition of h is a mere vulgarism ']. 
 The origin of this vulgarism has recently been shown to be Etruscan, see my 
 notice of Schulze's Lat. Eigennamen in the Classical Review, Nov. 1906. On 
 triumphus see Ital. Dial. p. 230; on pulcher (from TroXi/xpoi/s, through the 
 probably Oscan TroWaxpo-) see Ital. Dial. p. 48. Carthago is of course Semitic, 
 Cethegus probably Etruscan or Etrusco-Greek. 
 
24 APPENDIX 
 
 we know (e.g. from Quintil. 1. 4. 14 and 12. .10. 29) that there was 
 an audible difference. All that these transliterations prove is that 
 the degree of resemblance between the transliterated sound and the 
 transliterating symbol was closer than between the same sound and 
 any other symbol in use in the given locality. 
 
 But why, the reader will naturally ask, should we seem 
 anxious to reduce this evidence to its narrowest compass 1 Why 
 not give it the benefit of a liberal instead of a grudging inter- 
 pretation ? The answer is : in order to reduce to as narrow limits 
 as possible the apparent contradiction between the conclusions which 
 would, from it alone, seem reasonable, and the evidence of an 
 opposite nature, to which we must now briefly turn. Some of 
 this was put forward by Dr Elizabeth Dawes in her dissertation 
 on the Greek Aspirates (London, 1895), though her argument is to 
 some extent weakened by lack of critical method. 
 
 B. What grounds are there for attributing a partly fricative 
 pronunciation to ^, ^, x ^^ Attic in the fourth century B.C. 1 
 
 1. The cases, beginning^ with Aristophanes (<f>iX6(T6<f>ov Eccl. 
 571) and continuing down to the Christian era, in which ^, Xi ^ 
 appear as double consonants, making a preceding vowel long 
 by position, as in the Latin bracchium from Ppa^iiav, Low Latin 
 struppus from arTp6<fio<;. For some other examples see Lindsay, Lat. 
 Lang. p. 58. It is reasonable to assume with Gustav Meyer (Gr. 
 Gi^amm. Ed. 3, p. 287) that this innovation indicates the beginning 
 of the affricate stage. It is clear that a single case of such a prosody, 
 which defies the traditional poetical usage in Attic, is weightier 
 evidence than a great multitude of cases in which the traditional 
 usage is obeyed; but there are parallel examples of the metrical 
 lengthening before (f>, e.g. Kaibel, Epigrammata Graeca ex lapidibus 
 collecta, 60. 1 ; Notiz. d. Scavi di Antich. 1888, p. 282 {philosopha 
 in a Latin metrical epitaph of the last century B.C.); and others 
 given by Schulze, Quaestiones Epicae, pp. 256, 232 footnote. Pro- 
 bably all of these are later than the fourth century B.C., but in 
 Attica itself we have the names IltT^evs, HiT^tos^, probably derived 
 
 1 Cases from earlier poets must hardly be quoted as evidence of Attic use ; 
 and even Aeschylus (Ghoeph. 1049) may be copying the Epic or AeoUc examples. 
 Homeric lines beginning with e.g. Ze<pvp€r], iwiTovos, or ending with e.g. 6<f)iv 
 probably represent variations of metre, not of pronunciation: see Schulze, 
 Quaestiones Epicae, p. 430. 
 
 ■^ This third century example would be very strong evidence but for the fact 
 that some names, originally pet-names, were formed by an intentional doubling 
 of the consonant, e.g. Sa7r^c6, '^evvdo. 
 
APPENDIX ' :;. [ : :•*.;';. :*.; :\: l^ 
 
 from the deme 111^17 (though not used in a geographical sense) at the 
 end of the third century B.C. (C. I. A. 11. 977 uv. 9, 250—150 B.C.) ; 
 and in imperial times it is clear that IltT^cvs is used in the geo- 
 graphical sense (C. I. A. iii. 908 ; iii. 1230 b; iii. 226) ; the regular 
 spelling in Attic inscriptions being Ilt^evs (Meisterhans-Schwyzer, 
 p. 80, to which add Gustav Meyer Gr. Gramm. Ed. 3, pp. 288 ff.). 
 
 ' Such borrowings into Latin as hracchium are no doubt later 
 in date than those like apua, Corintus etc., and point to a 
 change in Greek pronunciation. Yet hracchium is older than 
 Plautas\ 
 
 2. The evidence that in Aeolic (from an early date, as the forms 
 BaK;(09, jSpoKxo's, and the later fX€Tr)WaK)(OTa show), Laconian, Cretan 
 and Boeotian, in or before the fourth century cf>, and ^ had become 
 affricatae or fricatives. (See Brugraann's Gr. Gram. Ed. 3, p. 105 f., 
 and Gustav Meyer Gr. Gram. Ed. 3, p. 287.) Now it seems at 
 least likely that such a change, once established in Boeotia, would 
 affect to some extent the speech of the commoner class of their Attic 
 neighbours. The pronunciation of Laconia, which Aristophanes 
 ridicules (vat to> o-tw for vat rw ^cw), is hardly likely to have affected 
 the speech of Attica. 
 
 3. The remarkable use of <^ in the Eteocretan fragments 
 (written in Ionic alphabet and dating from about 400 B.C.) to 
 denote what can only be some kind of spirant arising from p before 
 t {ado(i>te probably meaning ' optione,' ' libenter,' like Osc. ufteis = 
 Latin opt(ati)), as well as before s {-o<j>sano). See the Annual of 
 the Brit. School at Athens viii. p. 146, with the footnotes. 
 
 It is difficult to separate this from the early Greek use of (per for \^ (in an 
 Attic inscription as late as of 439 b.c, see Meisterhans p. 3), or, more generally, 
 to believe that f and \f/ had precisely the sound of ko- and ira- ; for if so, why were 
 these additional symbols needed ? And this brings us very near the unsolved 
 question of the sound denoted by x, &, <t> in the combinations x^» 0^. which is 
 however too complex to be discussed here. See Brugmann and Meyer U. cc. and 
 the authorities they cite. 
 
 4. Why did the Romans change their transliteration of the 
 Greek Aspirates from the simple jO, <, c to pA, th, ch 1 Was not the 
 additional sound which they heard so clearly in the second and first 
 centuries B.C. something more than merely an A, though this was 
 the only means they had of writing it ? 
 
 1 Its use in Miles Glor. i. 1. 26 shows that it was a completely naturalised 
 word. In Cato it is even applied to trees. 
 
^'fjdl :;/: V; V '■<• i^^: /;, APPENDIX 
 
 5. Finally, in the first century a.d., while all the Greek school- 
 masters at Rome were busy (as they were for one or two more 
 centuries) inculcating the pure Attic pronunciation of the aspirates, 
 we have on the walls of Pompeii (overwhelmed in 78 a.d.) the indis- 
 putable evidence of the graffiti of certain persons content to spell as 
 they spoke that to some people in Italy at any rate <^ sounded like 
 / {Dafne 'litteris sat uetustis' C. I. L. iv. 680; Fyllis ibid. 1265 a; 
 Trofime 2039 ; Filetus 2402). 
 
 None of this evidence demonstrates that a fricative pronunciation 
 had established itself even on the most vulgar lips in Athens in the 
 fourth century B.C. But it does appear a possibility that under 
 certain conditions the pronunciation was that of affricates. 
 
 My own present interpretation of the data, if it be worth setting 
 down, might be represented by some such conjectural table as this, 
 where/ denotes a labiodental spirant,/' a bilabial spirant ("like the 
 sound made in blowing out a candle ") : 
 
 At 450 B.C. at Athens <f>=p + h, and so in educated speech down 
 to at least 150 B.C., and in the teaching of Atticist grammarians at 
 Rome for another two or three centuries ; 
 
 but at 400 B.C. in colloquial speech medially at least <^=pf' + h 
 at 150 B.C. in vulgar speech in Italy <f>=/' + h 
 
 at 50 A.D. „ „ „ <^=/' 
 
 at350A.D. cfi=f. 
 
 R. S. C. 
 
 CAMBRIDGE: PRINTED BY JOHN CLAY, M.A. AT THE UNIVERSITY PRESS. 
 
PRESS NOTICES OF THE FIRST EDITION. 
 
 " The foundation of the University of Wales, which deserves more attention 
 than it has received outside the Principality, has already led to the first serious 
 effort to reform the current pronunciation of Greek and Latin. The circum- 
 stances were exceptionally favourable. Owing to the fact that Welsh sounds 
 are only represented conventionally in the Koman alphabet, Welsh speaking 
 students have no prejudice in favour of the English practice ; while the 
 classical professors happen to belong to the advanced school of modern 
 philology.... We must congratulate the authors and their colleagues upon the 
 boldness of their enterprise and express a hope that Wales will show the way 
 to England in adopting the reformed system thoroughly, not as an alternative, 
 or in parts." Academy, Nov. 23, 1895. 
 
 "Diese kleine Schrift geht eigentlich nur die Englander an, kann aber 
 diesen allerdings warm empfohlen werden ; und das konnte noch eindringlicher 
 geschehen, wenn wir nicht selbst in derselben Richtung noch sehr viel vor 
 unserer eigenen Thiire zu kehren hatten. Unter den vielen grausamen Aus- 
 sprachen, mit denen die Werke der Griechen und der Rdmer in den Schulen 
 Europas und Amerikas vorgetragen werden, ist die in England iibliche wohl 
 der grausamsten eine und der Abanderung am meisten bediirftig. Die beiden 
 Verfasser der kleinen Schrift, als linguistisch tiichtig gebildete Latinisten 
 bekannt, haben Vorschlage fiir die Aussprache der beiden klassischen Sprachen 
 gemacht, die auf den wissenschaftlichen Ermittelungen iiber die zum Teil etwas 
 verwickelten Fragen hexvihen" —Berliner Philologische Wochenschrift, 1896, 
 No. 13. 
 
 " Die Frage der lateinischen und griechischen Orthoepie steht auch in 
 England seit langerer Zeit auf der Tagesordnung, ganz besonders brennend 
 scheint sie in dem zwiespraohigen Wales geworden zu sein. Zwei Professoren 
 der dort neu gegriindeten Universitat haben mit Unterstiitzung ihrer KoUegeu 
 und im Einverstandnis mit den regierenden Korpern der Universitat, gestiitzt 
 auf die besten engliseheu und deutschen Werke iiber Phonetik und Aussprache, 
 diesen kurzen Abriss verfasst, der eine sichere Grundlage fiir den Unterricht 
 an Universitat und. Mittelschule bilden soil. Massgebend ist natiirlich fiir das 
 Griechische das 5. Jahrhundert, fiir das Lateinische das 1. Jahrhundert vor 
 
 Christ Alles kurz und zuverlassig und durchaus geeignet den Zweck zu 
 
 fordern welchem es dienen soil." — Deutsclie Litter aturzeitung, May 23, 189(). 
 
THIS BOOK IS DUE ON THE LAST DATE 
 STAMPED BELOW 
 
 AN INITIAL FINE OF 25 CENTS 
 
 WILL BE ASSESSED FOR FAILURE TO RETURN 
 THIS BOOK ON THE DATE DUE. THE PENALTY 
 WILL INCREASE TO 50 CENTS ON THE FOURTH 
 DAY AND TO $1.00 ON THE SEVENTH DAY 
 OVERDUE. 
 
 HE 
 
 MAY 15 1933 
 
 JAN 31 1^45 
 
 \#V^ 
 
 RECEJVF 
 
 DEC 2r67 42^1 
 
 
 delopment. By E. 
 
 J6. 12.S. net, 
 
 ' of Sanscrit scholars, to 
 jcialist papers on Yedic 
 |ie work of a long series 
 p who care for whatever 
 
 ious literature In its 
 
 ;he book is worthy of the 
 IS a wrangler and senior 
 
 ndustry and labour we 
 es supposed, the peculiar 
 a splendid specimen of 
 Ihe patient investigation 
 ftnd logical use to which 
 jstem of metre for the 
 the text and a basis for 
 jj Herald. 
 
 bimar and Glossary. 
 
 times. 30s. 
 
 nd the Minor Dialects, 
 
 ;ers, and the Local and 
 
 lects, Appendix, Indices 
 
 ;a in usum academi- 
 
 studuit E. S. Conway. 
 
 LD 21-50m-l,'33 . -- . . 
 
 in Comparative 
 
 Philology. By the late Herbert Dukinfield Darbishire, M.A., Fellow 
 of St John's College, Cambridge. Edited by R. S. Conway, Litt.D., with 
 a Biographical Notice by J. E. Sandys, Litt.D., Fellow and Tutor of 
 St John's College and Public Orator in the University of Cambridge. 
 Demy 8vo, 279 pp. 7s. Gd. 
 
U. C. BERKELEY LIBRARIES 
 
 CDM7DMEDfi7 
 
 ^,4 8'4:^^ 
 
 L 
 
 UNIVERSITY OF CAIvIFORNIA IvIBRARY