The Blackwell History of the Latin Language - James Clackson.pdf

文档下载格式:pdf 上传:2022-6-21 22:16 阅读:7次 页数:333页 大小:1.44 MB|预览

The Blackwell History of the Latin Language - James Clackson.pdf内容摘要: 9781405162098_1_pre.qxd 8/9/07 11:19 AM Page i




The Blackwell History of the Latin Language
9781405162098_1_pre.qxd 8/9/07 11:19 AM Page ii
9781405162098_1_pre.qxd 8/9/07 11:19 AM Page iii




The Blackwell History
of the Latin Language

James Clackson and Geoffrey Horrocks
9781405162098_1_pre.qxd 8/9/07 3:20 PM Page iv




© 2007 by James Clackson and Geoffrey Horrocks
BLACKWELL PUBLISHING
350 Main Street, Malden, MA 02148-5020, USA
9600 Garsington Road, Oxford OX4 2DQ, UK
550 Swanston Street, Carlton, Victoria 3053, Australia
The right of James Clackson and Geoffrey Horrocks to be identified
as the Authors of this Work has been asserted in accordance with
the UK Copyright, Designs, and Patents Act 1988.
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any
means, electronic, mechanical, photocopying, recording or otherwise,
except as permitted by the UK Copyright, Designs, and Patents Act 1988,
without the prior permission of the publisher.
Designations used by companies to distinguish their products are often claimed
as trademarks. All brand names and product names used in this book are trade
names, service marks, trademarks, or registered trademarks of their respective
owners. The publisher is not associated with any product or vendor mentioned
in this book.
This publication is designed to provide accurate and authoritative information in
regard to the subject matter covered. It is sold on the understanding that the
publisher is not engaged in rendering professional services. If professional advice or
other expert assistance is required, the services of a competent professional should
be sought.
First published 2007 by Blackwell Publishing Ltd
1 2007
Library of Congress Cataloging-in-Publication Data
Clackson, James.
The Blackwell history of the Latin language / James Clackson and Geoffrey
Horrocks.
p. cm.
Includes bibliographical references and index.
ISBN 978-1-4051-6209-8 (hardcover : alk. paper) 1. Latin language–History.
I. Horrocks, Geoffrey C. II. Title.
PA2057.C58 2007
470.9–dc22
2007018802

A catalogue record for this title is available from the British Library.
Set in 10/12pt Galliard
by Graphicraft Limited, Hong Kong
Printed and bound in Singapore
by Utopia Press Pte Ltd
The publisher’s policy is to use permanent paper from mills that operate a
sustainable forestry policy, and which has been manufactured from pulp processed
using acid-free and elementary chlorine-free practices. Furthermore, the publisher
ensures that the text paper and cover board used have met acceptable
environmental accreditation standards.
For further information on
Blackwell Publishing, visit our website:

9781405162098_1_pre.qxd 8/9/07 11:19 AM Page v




Contents




Prefacevi

Chapter ILatin and Indo-European1
Chapter II The Languages of Italy37
Chapter IIIThe Background to Standardization77
Chapter IV ‘Old’ Latin and its Varieties in the
Period c.400–150 BC 90
Chapter VThe Road to Standardization: Roman Latin
of the Third and Second Centuries BC 130
Chapter VI Elite Latin in the Late Republic and
Early Empire 183
Chapter VIISub-Elite Latin in the Empire229
Chapter VIII Latin in Late Antiquity and Beyond 265

Glossary305
Appendix: The International Phonetic Alphabet 316
Bibliography of Reference and Other Works 317
Index 319
9781405162098_1_pre.qxd 8/9/07 11:19 AM Page vi




Preface




The impetus to write this book came after teaching various joint courses
on Latin historical linguistics to undergraduates reading Classics at Cam-
bridge over the past ten years. Although we consistently recommended
L. R. Palmer’s The Latin Language to students as a readable account of
the history of the language, we became increasingly aware of some of the
shortcomings that have become apparent in the 50 years since Palmer’s
book was written. In particular, there have been considerable advances
in linguistic theory and method, as well as important discoveries of texts
in Latin (and in the other languages spoken in pre-Roman Italy), and a
better understanding of the Indo-European background to the language.
Furthermore, Palmer has comparatively little to say about the processes
by which Latin became standardized, nor did he have the advantage of
modern sociolinguistic theory to help explain the interactions between
the spoken language and the Classical standard. Accordingly, we set out
to write a new history of Latin that overcame some of the shortcomings
we saw in Palmer. We decided to model the structure of the work on
Geoff Horrocks’s book Greek: A History of the Language and its Speakers,
including detailed discussion of a number of texts from all periods with
glossing for each word. In this way we hope that the book will be acces-
sible to those who have little or no Latin, but are interested in linguistics
or language history, as well as to classics undergraduates, graduates and
professional Latinists. In order to appeal to this large and diverse con-
stituency we have included expositions of some topics which may be
familiar to some readers, such as the comparative method in historical
9781405162098_1_pre.qxd 8/9/07 11:19 AM Page vii




Prefacevii

linguistics, but we hope that all readers will find something new. We have
included a glossary of some linguistic terms for Latinists who are new to
the subject; these readers may also find an introductory volume to his-
torical linguistics helpful, such as Herbert Schendl Historical Linguistics
(Oxford, 2001) or Mark Hale Historical Linguistics: Theory and Method
(Oxford, 2007).
There are other changes of emphasis from Palmer’s work. On the whole
our focus has been on Latin as a language, and its relationship to the
history of Rome and Roman imperialism. We have in consequence con-
centrated more on linguistic than stylistic issues, and we have been less
concerned to describe the particular idioms and vocabulary choices of
Roman literary figures, except in so far as they have proved important for
the subsequent development of Latin. Accordingly, the reader should not
expect to find here very much in the way of appreciations of the music
of Vergilian verse or descriptions of the metrical patterns found at the end
of periodic sentences (clausulae) in Classical prose, though the latter are
very briefly considered in chapter VI. We have not presented a system-
atic overview of the phonology, morphology, and syntax of the language,
concentrating instead on specifics where these are relevant to the history
of linguistic innovations and replacements. The necessary basic informa-
tion can be easily found in standard grammars and handbooks. Finally,
we devote proportionately much less space to the development of the Latin
vocabulary than Palmer, who was writing before the publication of the
Oxford Latin Dictionary and when the ongoing Thesaurus Linguae
Latinae was still in its infancy. These works now allow the reader to trace
word histories in a much more systematic way than was possible when
The Latin Language came out, and we have preferred to restrict the space
given to lexical discussions in order to allow a corresponding increase in
the exposition of syntactic changes.
We hope that readers will benefit from being able to appreciate the
history of Latin in its entirety, from the pre-historic origins to the end
of its existence as a language with native speakers. But we are aware that
people have many different needs from a book such as this, and some
may prefer to read a chapter at a time. In view of this, we have decided
to present specific bibliographies for each chapter. We have also
appended a further bibliography of standard and useful reference works
at the end of the volume.
Of the eight chapters of this book, we each took individual respons-
ibility for four: James Clackson wrote the outer chapters, I, II, VII and
VIII and Geoff Horrocks wrote the inner core: chapters III, IV, V and
VI. All translations of Latin texts cited are our own. We each read, com-
mented on and discussed the other’s work in draft. We have also
benefited from the input of successive groups of students taking the ‘E3’
9781405162098_1_pre.qxd 8/9/07 11:19 AM Page viii




viiiPreface

course at the Classics Faculty who were unwitting guinea pigs as readers
of many of the Latin texts that appear in this book. We are particularly
grateful to Jim Adams, for allowing us to use material from his forth-
coming book Regional Diversity in Latin, and to Michael Crawford, for
giving us access to his new reading of CIL I2 5. An anonymous and well-
informed reader for Blackwell’s saved us from innumerable errors and made
many welcome and constructive suggestions for improvement. The usual
disclaimers apply, of course. And finally we wish to thank Anna Oxbury
for copyediting our manuscript so expertly and professionally, and for devis-
ing a range of excellent solutions to rather complex problems of layout
and presentation.
We dedicate our book to Gill Horrocks and Véronique Mottier.
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 1




Chapter I
Latin and Indo-European




1.1 Introduction
Latin is an Indo-European language. This means that Latin is genetically
related to most of the modern (and the ancient) languages of Europe, as
well as many languages of India, Iran and Central Asia. The genetic rela-
tionship accounts for the large numbers of similarities, both in vocabu-
lary and in grammar, between Latin and Greek, Sanskrit, Gothic, Old Irish
and many other ancient languages which no alternative hypothesis (such
as chance similarity, linguistic borrowing or convergence) can explain. Over
the last 200 years linguists have undertaken a systematic comparison of
the similarities between the Indo-European (henceforth IE) languages
to build up a picture of what the non-attested parent (‘Proto-Indo-
European’, henceforth PIE) must have looked like. The reconstruction
of PIE is, in places, highly abstract and highly complex, and for many
individual features there is still considerable debate amongst experts in
the field as to which reconstruction is the most plausible. Even so, it is
possible to arrive at a picture of the parent language which is widely
accepted, and use that to set the background to the development of Latin.
The reader will have to take much of what is said about PIE on trust in
this chapter, since this is not a book about PIE, but about Latin.
Why should the historian of Latin be interested in PIE? Apart from
the intrinsic interest of knowing the relationships between Latin and other
languages, we can suggest a number of reasons. Firstly, in order to under-
stand the development of Latin, it is necessary to see what it started out
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 2




2Latin and Indo-European

as. Thus the development of the Latin verbal system, or Latin word order,
has its roots already in PIE structures. Secondly, a knowledge of the
background to Latin can help assess the question of its relationship to
neighbouring languages – as we shall see in the next chapter. Thirdly,
a knowledge of PIE may actually help us to understand some features
of Latin vocabulary or grammar. To take a single example, one of the
earliest Latin inscriptions known is a sixth century graffito scratched around
a pottery vessel, known as the ‘Duenos vase’ (CIL I2 4, see 1.4.5 below).
The final 15 letters of the inscription read nemedmalostatod, and this was
long recognized as ne med malos (s)tatod an earlier form of ne me malus
stato not me-ACC bad-NOM set.3sg.IMPER ‘let no bad man set me’,
although the use of the verb sto ‘I stand, set’ in this context was unex-
plained. Comparative Indo-European linguistics, however, offered a solu-
tion to this problem (first proposed by Rix 1985). Other IE languages,
such as Irish and Hittite, share a verbal root which can be reconstructed
as *ta- (*teh2-) and which means ‘steal’. If we assume that this verbal root
also survived into an early stage of Latin, then we can interpret the sequence
as ne me malus *tato not me-ACC bad-NOM steal.3sg.IMPER ‘let no
bad man steal [me]’, a commonplace formula on inscriptions on move-
able objects in the ancient world. This attestation remains the only
appearance of this verbal root in the whole Latin corpus, and its mean-
ing is only recoverable through IE comparison.


1.2 The IE Language Family
The IE language family comprises over 80 different languages and varieties.
All of the living languages, and most of the varieties which are no longer
spoken can be assigned to one of the subgroups of the family. Some ancient
languages have left such scanty remains that their position in the family,
and in some cases, even their membership of the family, remains in doubt.
As we shall see in the next chapter, some of these scantily attested lan-
guages are relevant to the early history of Latin, and we shall discuss them
in more detail there. Here we shall confine ourselves to giving an
overview of the different branches of the IE language family.

1 Anatolian. The Anatolian branch is the earliest attested branch of
Indo-European. The best attested language in the Anatolian family
is Hittite, which is written in the cuneiform script, adopted from
Semitic languages of the ancient Near East and for which the
earliest texts date from the sixteenth century BC. A number of other
languages are now also recognized to belong to the Anatolian fam-
ily. Two others are recorded from the period before 1000 BC, both
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 3




Latin and Indo-European 3

in cuneiform (Palaic and Luwian – Luwian is also attested in a hiero-
glyphic script which is not used for any other language), and from
a later period other languages are recorded in alphabetic scripts, includ-
ing Lydian, Lycian and Carian.
2 Greek. The Greek branch of Indo-European is the second earliest
attested, with texts written in the Linear B syllabary surviving from
the fourteenth century BC and later. Greek is extensively attested in
alphabetic script from the eighth century BC onwards.
3 Indo-Iranian. The two large language families termed Indic and
Iranian share a number of common innovations which guarantee that
they both derive from the same branch of Indo-European. The first
evidence for Indo-Iranian is also in the second millennium BC, and
consists in the inclusion of some terms and phrases relating to rid-
ing and horsemanship in cuneiform sources. The major early textual
remains of the Indic branch are the hymns of the Rg-Veda (written
in an archaic form of Sanskrit, often termed Vedic), and of the Iranian
branch the Gathas, the hymns attributed to Zarathrustra in the Avesta
(their language is known as Avestan or Gathic Avestan). Both of these
texts were orally transmitted for centuries before being written
down, but internal evidence suggests that they are both of great anti-
quity, and scholars generally assign a date to around 1000 BC for
the composition of the Gathas and a couple of centuries earlier for the
oldest Vedic hymns. Indo-Iranian, Greek and Anatolian are the three
most important branches for the reconstruction of PIE.
4 Latin and the languages of Italy. As we shall see in the next chapter,
it is a moot point how closely the IE languages of Italy are related
to each other. Several subgroups are recognizable: Latino-Faliscan,
comprising Latin and the neighbouring language Faliscan which are
attested from the seventh–sixth century BC, although the early
inscriptions are short and difficult to interpret in both languages; the
Sabellian group, known principally through Oscan and Umbrian and
attested first in the South Picene inscriptions which date from the
sixth century BC; Venetic, attested in short inscriptions from the sixth
century BC is also IE. The Messapic language, attested from inscrip-
tions from the same date in the area at the extreme south-west of
Italy shows greater divergence from the other languages of Italy.
5 Celtic. The only surviving languages of the Celtic branch are Irish,
Scots Gaelic, Welsh and Breton, but the family once extended over
a much wider section of Western Europe. The earliest attestations
of Celtic are inscriptions from France, Italy and Spain in the centuries
immediately before and after the beginning of the Christian era.
Extensive textual evidence for Celtic is much later, with the first Old
Irish glosses recorded in the seventh century AD. The interrelationship
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 4




4 Latin and Indo-European

of the Celtic languages is still debatable, but the following subgroups
are recognized:
(a) Goidelic: the branch which comprises Irish, Scots Gaelic and Manx.
(b) Brythonic: the branch which comprises Welsh, Cornish and
Breton.
(c) Celtiberian, known from inscriptions in Spain.
(d) Gaulish, known from inscriptions mainly in France.
(e) Lepontic, known from inscriptions in northern Italy.
Owing to an imperfect knowledge of branches (c), (d) and (e), it is
difficult to be sure whether a ‘continental Celtic’ sub-group, com-
prising all the Celtic languages from outside the British Isles with
the exception of Breton, actually reflects any linguistic reality.
6 Germanic. The Germanic language group is first known from
sources in the first millennium BC; the first extensive text is the Gothic
Bible translation made in the fourth century. Old English is attested
from the eighth century, and Old High German, Old Saxon and Old
Norse from the following century.
7 Armenian. The Armenian branch comprises just one language,
known in its classical form from the Bible translation and theological
and historical works written in the fifth century AD.
8 Slavic. The first texts to record a Slavic language are the Bible version
and translations of Greek texts made by Cyril and Methodius in the
late ninth century.
9 Baltic. The Baltic subgroup comprises Lithuanian, Latvian and
the now extinct Old Prussian. The first texts were written in the
fourteenth–sixteenth centuries.
10 Albanian. Albanian has only a relatively shallow time depth, being
first attested in texts written by missionaries and others from the late
Middle Ages to the early modern period.
11 Other poorly attested languages. There are also a number of languages
which are only known from short inscriptional texts or glosses
recorded by Classical authors which are reckoned to be IE, but whose
relationship to other languages remains in doubt. These include
Phrygian, Thracian, Illyrian, Sicel and Lusitanian.


1.3 Reconstructed PIE
The reconstruction of PIE entails the assumption that a single language
was spoken at some point in time from which all the different IE varieties
have evolved. However, the reconstructed picture can never reach the stage
of giving an adequate description of PIE. This is due to the nature of
reconstruction through the comparative method (CM). The CM operates
9781405162098_4_001.qxd 8/10/07 11:15 AM Page 5




Latin and Indo-European5

through identification of sets of correspondences in languages which are
known to be related, and forming of hypotheses to explain the corres-
pondence. We may, for example construct a correspondence set of word
forms with identical meaning and similarity of form, as follows:

Latin Greek SanskritEnglish
pater patbr pitar-father
pes poûspad-foot
plenusplbrbsperná-full
pro pró prá for
pellispélas Old English fell ‘hide’
pecupauúOld English feoh ‘livestock’
piscisfish

In all these words (and several others) we see a correspondence between
initial p- in Latin, Greek and Sanskrit, and initial f- in English. The words
are not limited to one particular lexical field and they represent core items
in the lexicon. We can reconstruct a single PIE phoneme as the forebear
of these sounds in the daughter language, traditionally denoted *p. The
same process is used to reconstruct the whole phonemic system for PIE.
When we come to morphology, however, we find reconstruction is not
so straightforward. To take a notorious example, we can compare the gen-
itive singular of the o-stem declension (the Latin 2nd declension; in this
table we have added further IE languages to those given above):

Latin(Homeric) Greek SanskritOld English HittiteLithuanian
-i -oio-asya -es -as-o

These forms are not reconcilable to a single prototype, and in order to make
sense of the differences one must hypothesize motivations for replace-
ment of an earlier form in one language branch or another. The Hittite
ending -as is identical with the nominative singular ending -as (both can
be derived from *-os), and this is usually seen as an especially archaic form,
and one which would be liable to be replaced in order to disambiguate
the two categories. The Greek and Sanskrit forms can both be derived
from an extended form *-osyo, which is found in other branches of IE,
suggesting that the replacement of *-os already took place within the par-
ent language. It is clear that in order to explain these different genitive
singular endings, we must reconstruct a proto-language with diachronic,
or dialectal, variation. It can thus be difficult to reconcile the reconstructed
morphology to the reconstructed phonology – are we to assume that the
different chronological or dialectal variants of the proto-language shared
the same phonemic system? This seems unlikely from what we know of
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 6




6 Latin and Indo-European

attested languages, but there is no way to restrict the reconstruction of
phonemes to one particular morphological reconstruction.
The CM gives the impression that reconstructed PIE is a single point
from which the daughter languages all derive separately. In actual fact, it
is unlikely that the reconstructed data that linguists operate with were
concentrated together in an actual speech-community at one time and
place. It is more likely that the reconstructed items are diffusely arrayed
in time and space and across the speaker population. The comparative
linguist must therefore draw up a framework in order to fit the recon-
structed data into plausible temporal and spatial slots. For example,
in order to explain the reconstruction of both a genitive singular marker
*-os and *-osyo, one model would propose that the language ancestral to
Hittite and the rest of the Anatolian branch split off from PIE at an ear-
lier date than other languages. The only check on whether this model is
correct is its own explanatory power and internal consistency, and it may
be possible to construct two, or more, separate models which both give
adequate explanations of the reconstructed data. In dealing with hypo-
theses about the Indo-European language family, it will be necessary to
bear these methodological points in mind.


1.4Latin and IE
A presentation of reconstructed PIE is beyond the scope of this work. In
this section, we shall present some of the salient features of PIE for the
history of Latin, in order to give an idea of what Latin has inherited from
PIE and where it has diverged.


1.4.1Phonology
The reconstruction of PIE is most secure in the domain of phonology.
This is because the phoneme system contains a small, finite and ordered
set of elements. Phonological change is, on the whole, regular, well studied
and well documented. This means that it is usually possible to compare
two cognate sounds, such as Latin p and English f in the example given
above, and identify the sound which is most likely to be ancestral to them.
In the case of p and f, for example, we know of many secure examples
of the change p > f in the world’s languages, but far fewer of f > p, so
we can reconstruct the ancestral sound as *p (written with an asterisk since
it is a hypothetical, unattested form). However, we must be aware of the
limits of our reconstruction; we may be able to reconstruct the phon-
emic system without complete certainty about the phonetic realization of
those phonemes. We have no way of knowing, for example, whether a
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 7




Latin and Indo-European7

reconstructed *d was a true dental or an alveolar or some other linguo-
dental consonant, although we can be sure that it was opposed to two
other consonants with the same place of articulation, *t and *d h, and other
consonants with the same manner of articulation, but a different place of
articulation, such as *g. Nor do we know for certain that the reconstructed
phonemes *d and *g were distinctively voiced, and some models of PIE
claim that they had a different manner of articulation. In reconstructed
PIE, it is the oppositions between the phonemes that are important, rather
than the distinctive features per se that articulate these oppositions. The
standardly reconstructed phoneme system of PIE is as follows:

Consonants
Stops:
Labial Dental Palatal VelarLabio-velar
*p *t *k′ *k *kw
(*b) *d *g′ *g *gw
*bh*dh*g′h*gh*gwh
Fricatives: *s
‘Laryngeals’: *h1, *h2, *h3
Nasals: *m, *n
Continuants: *r, *l, *y, *w
Vowels
*e, *o, *a (*b, *d, *a)
*q, *r, *s, *p, *i, *u (*c, *e)

Some explanatory points should be made about the above tables:

1 ‘Labio-velars’ is the term given to a series of consonants which
have reflexes in Eastern IE languages (Indo-Aryan, Iranian, Slavic, Baltic,
Armenian) as velars or palatalized velars, but which in the earliest stages
of Western IE languages (Greek, Germanic, Celtic, Latin) appear as velars
with concomitant lip-rounding, or sometimes as labials. Typical cognate
sets are the following:

*kwo-/kwi- ‘who?’: Sanskrit ká-, Greek tís, Germanic (English) who, Latin quis
*g wem- ‘come’: Sanskrit gam-, Greek baínd, Germanic (English) come, Latin
uenio
*g wow- ‘cow’: Sanskrit gav-, Greek boûs, Germanic (English) cow, Latin bos

(Note that in some of the Greek and English cognates, the labio-
velars have been further obscured by specific sound-changes: tís shows a
characteristic Greek development to a dental before a front vowel, and
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 8




8Latin and Indo-European

in English cow and come the labial element has been lost before a back
vowel.)
It can be seen from the above examples that Latin qu- derives from
PIE *kw, but *g w develops differently. In most words it is continued by
Latin u [w] but there are also cases, as bos, where it appears as a labial
stop. The words which show this development (and also forms which have
p in place of PIE *k w ) are normally explained as borrowings from other
IE varieties spoken in Italy which regularly develop labial stops from ori-
ginal labio-velars. These will be discussed more fully in the next chapter.
Alongside labio-velars, there are also ‘velar’ and ‘palatal’ series, which have
different outcomes in some IE languages, but merge as velar consonants
in prehistoric Latin. Schrijver (1991: 425–36) has suggested that the two
series had different effects on a following *e: *ke- giving Latin ca- (as carpo
‘I pluck’ < *skerp-), and *k′e- giving Latin ce- (as cedo ‘give!’ < *k′e-).
Unfortunately, there are only six etymologies to support Schrijver’s claim,
and a few counterexamples, so Schrijver’s theory remains unproven at
present (see Meiser 1998: 82f.).
2 ‘Voiced aspirates’ is the traditional term for a series of consonants
which are reconstructed from the comparison of voiceless aspirates p h, t h,
kh in Greek, voiced aspirates b h, d h, g h in Sanskrit, and voiced consonants
in Germanic (English b, d, g), Iranian, Armenian, Baltic and Slavonic. Note
the following examples of cognate sets for PIE *b h and *d h:

*b her- ‘carry’: Greek p hérd, Sanskrit b hárami, Germanic (English) bear,
Armenian berem
*neb h- ‘cloud’: Greek nép hos, Sanskrit náb has-, Germanic (German) Nebel
‘fog’, Old Church Slavonic nebo ‘heaven’
*d huh2mo- ‘vapour, smoke’: Greek t hemós, Sanskrit d hemá-, Old Church
Slavonic dymj
*rudhro- ‘red’: Greek erut hrós, Sanskrit rud hirá-, Germanic (English) red,
Slavonic (Russian) rudyj ‘red-haired’.

It can be seen from the table that these consonants are not opposed to
a voiceless aspirate series (as the voiced aspirates of Sanskrit are), and it
may be better to envisage them as originally ‘breathy-voiced’ in PIE,
although we shall retain the traditional terminology of ‘voiced aspirate’.
The reconstruction of voiced aspirates without voiceless aspirates has been
held to violate a linguistic universal, and has led to attempts to refashion
the PIE consonant stem entirely. One such attempt, independently
proposed by the American Paul Hopper and the Georgian Thomas
Gamkrelidze (see Szemerényi 1996: 152), involves re-casting the recon-
structed voiced stops as ejectives (or ‘glottalics’), and then interprets the
opposition between the other two series as only reliant on the feature
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 9




Latin and Indo-European9

[voice], with aspiration not a distinctive feature. The question of the recon-
struction of the PIE stops is still under debate, but the ‘glottalic’ model
does not seem to have any extra explanatory power when it comes to the
derivation of the Latin consonant system from PIE, since PIE *p, *t, *k,
etc. are continued as voiceless stops in Latin and *b, *d, *g, etc. as voiced
stops, whereas the voiced aspirate series develop either to Latin fricatives
in word-initial position or to voiced stops word-medially. Thus the Latin
cognates to *b her-, *neb h-, *d huh2mo- and *rudhro- are fero, nebula, fumus
and ruber. We shall return to investigate these Latin developments more
fully in the following chapter, but for our present purposes we need only
state that the Latin reflexes are most economically derived from original
‘voiced aspirates’: the word-initial development to fricatives can be
accounted for by the original feature [aspiration] (cross-linguistically
the move from aspirates to fricatives is widely attested), whereas in word-
internal position the feature [voice] is preserved.
3 ‘Laryngeals’ is the traditional term used to refer to three consonants
which are hypothesized to have existed from their effect on neighbouring
vowels, and whose presence can be detected by systematic vowel alterna-
tions in different morphological environments. Laryngeals have no direct
reflexes as consonants in any IE language outside the Anatolian branch,
where they are sometimes continued by velar or pharyngal fricatives
(and even there *h1 may leave no trace). Despite their widespread loss,
laryngeals appear to have had different outcomes in different language
branches, and they must be reconstructed for early, prehistoric stages
of Latin in order to explain certain developments. The treatment of
laryngeals in Latin is generally similar to that found in neighbouring IE
languages, although aspects of their behaviour is complex, and there are
still areas of disagreement (Schrijver (1991) gives a detailed treatment of
laryngeal developments in Latin, in a book of over 500 pages). The basic
effects of laryngeals on neighbouring vowels is as follows, we have omit-
ted the details of the development in Anatolian languages, citing Hittite
or Luwian forms only where the laryngeal has a consonantal outcome (note
that H = any of *h1 *h2 *h3):

(a) following vowels laryngeals are lost with lengthening of a preced-
ing short vowel; the three laryngeals have differing effects on the
vowel *e:

*iH > Latin c, Greek c, Sanskrit c
*gwih3wo- ‘alive’ > Latin uZuus, Sanskrit jZvá-
*uH > Latin e, Greek e, Sanskrit e
*dhuh2mo- ‘vapour, smoke’ > Latin f\mos Greek t h\mós, Sanskrit
d h\má-, Hittite tuhhai-
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 10




10Latin and Indo-European

*oH > Latin d, Greek d, Sanskrit a
*d hoh1- ‘put, place’ in Latin sacer-d[s ‘priest’
*eh1 > Latin b, Greek b, Sanskrit a
*dheh1- ‘put, place’ > Latin fYci, Greek tít hYmi, Sanskrit dádhXmi
*eh2 > Latin a, Greek a (in Doric and other dialects), b (in Attic and
Ionic), Sanskrit a
*peh2- ‘pasture’ > Latin pXsco, Hittite pahs-
*eh3 > Latin d, Greek d, Sanskrit a
*deh3- ‘give’ > Latin d[s, Greek díd[mi, Sanskrit dádXmi
(b) before vowels laryngeals are generally lost, but, again, the three laryn-
geals have differing effects on the vowel *e:
*Hi > Latin i, Greek i (but *h2i may go to ai), Sanskrit i
*h2im- ‘copy’ > Latin imitor ‘I copy’, Hittite himna- ‘substitute’
*Hu > Latin u, Greek u (or possibly eu, au, ou), Sanskrit u
*h1us-to- ‘burnt’ > Latin ustus, Sanskrit utWá-
*Ho > Latin o, Greek o, Sanskrit a
*h2owi- ‘sheep’ > Latin ouis, Greek ówis, Sanskrit ávi-, Luwian hawis
*h1e > Latin e, Greek e, Sanskrit a
*h1esti ‘is’ > Latin est, Greek estí, Sanskrit asti
*h2e > Latin a, Greek a, Sanskrit a
*h2ent- ‘front’ > Latin ante, Greek antí, Hittite hant-
*h3e > Latin o, Greek o, Sanskrit a
*h3ekw- ‘eye’ > Latin oculus, Greek ómma (< *óp-m-), Sanskrit ákti
(c) when laryngeals stand between other consonants, they develop to
vowels:
*Ch1C > Latin a, Greek e, Sanskrit i, other IE language branches a
or lost
*dhh1(k)tó- ‘put’ > Latin factus, Greek t hetós, Sanskrit hitá-
*Ch2C > Latin a, Greek a, Sanskrit i, other IE language branches
a or lost
*sth2tó- ‘standing, stood’ > Latin status, Greek statós, Sanskrit sthitá-
*Ch3C > Latin a, Greek o, Sanskrit i, other IE language branches a
or lost
*dh3tó- ‘given’ > Latin datus, Greek dotós

4 The reconstruction of the PIE vowel *a and the long vowels. As
the above tables relating to laryngeals show, the reconstruction of these
vowels is closely related to the reconstruction of laryngeals. If Latin a
can go back to PIE *h2e or a laryngeal between two consonants, then is
there any need to reconstruct a separate PIE phoneme *a? If Latin long
b can be derived from a sequence *eh1, can we then dispense with the
reconstructed vowel *b in the PIE phoneme inventory? We have followed
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 11




Latin and Indo-European 11

a model of PIE which holds that both *a and the long vowels should be
reconstructed, although the reasons for this are dependent upon phenomena
in the Indo-Iranian and Anatolian languages. For the history of Latin,
however, the difference between original *h2e and *a, or between original
*b and *eh1, is irrelevant, since the laryngeal consonants were lost at such
an early stage in prehistory that they make no difference to the language.
5 The short vowels *q, *r, *s, *p, *i, *u have a special status in PIE,
since they act as allophones of the consonants *m, *n, *r, *l, *y, *w respec-
tively, depending on their position in the word. For example, consonantal
*r occurs adjacent to a vowel sound: as *ph2terq ‘father-ACC’ (Greek
patéra, Sanskrit pitáram), and the vocalic *s between two consonants:
*ph2t|su ‘father-LOC.PL’ (Greek patrási, Sanskrit pit|tu). As the example
of *ph2tssu shows, Sanskrit has retained the vocalic allophone of *r but
in Greek it regularly developed to ra or ar. Latin u and i still retain some
vestiges of this alternation between consonant and vowel: for example,
in the paradigm of the verb uoluo, uoluit ‘he rolls’ (with u = [w]), but
uolutus ‘rolled’, However, in Latin vocalic u and consonantal u are now
separate phonemes (note the minimal pair uoluit [wolwit] ‘he rolls’ and
uoluit [woluit] ‘he wanted’). In Latin the PIE short vowels *q, *r, *s,
*p have developed to combinations of vowel and consonant, *em, *en,
*or and *ol respectively, as shown by the following etymologies:

*k′qtom ‘hundred’ > Latin centum, Greek hekatón, English hundred
*trto- ‘stretched’ > Latin in-tentus, Greek tatós, Sanskrit tatá-
*k ′sd- ‘heart’ > Latin cord-, Greek kardía
*mpd- ‘soft, weak’ > Latin mollis, Greek bladús, Sanskrit msdú-

1.4.2 Latin morphosyntactic developments from PIE
The term ‘ablaut’ (also known as ‘vowel gradation’) describes a system-
atic alternation of vowels within a morphological paradigm. For PIE the
following types of ablaut can be reconstructed.

1 Shift of word-accent within a paradigm with a concomitant loss of
the unaccented vowel (quantitative ablaut). For example, the recon-
structed paradigm for the noun ‘god’:
*dyéw ‘god’ vocative Greek Zeû, Sanskrit dyàus
*diw-és ‘god’ genitive Greek Di(w)-ós, Sanskrit div-ás
(note that *y/*i are allophones of a single phoneme, as are *w/*u).
Or the reconstructed present tense paradigm of the verb ‘go’:
*éy-mi ‘I go’Greek eî-mi, Sanskrit émi
In Sanskrit, e derives from *ei
*i-mé ‘we go’Greek í-me-n, Sanskrit i-má-si
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 12




12 Latin and Indo-European

2 Change in vowel quality within the same syllable (qualitative ablaut).
For example the vowel of the suffix in the word meaning ‘family’ or
‘stock’ (the suffix is usually represented as *-e/os-).
*g´énh1-os ‘family, stock’ nominative Greek gén-os, Latin gen-us
*g´énh1-es-os ‘family, stock’ genitive Greek gén-e-os, Latin gen-er-is

The processes of quantitative and qualitative ablaut mean that every mor-
pheme in PIE has (at least) three alternative morphs, one with the vowel
*e, one with the vowel *o and one with no vowel. There are also recon-
structed ablaut forms with a long vowel, *b or *d (for example, the nom-
inative singular of the word for ‘god’ is sometimes reconstructed *dybws).
However, these forms are more restricted in distribution, and the motiva-
tion for them is disputed.
In many roots, the effects of ‘laryngeals’ or other sound changes have
disguised the original pattern and obscured the relationship between differ-
ent ablaut forms. Latin datus ‘given’ < *dh3-to- and dos ‘gift’ < *deh3-t-,
for example, are respectively reconstructed with the root in an ablaut form
without a vowel, and one with the vowel *e, although the vowels in Latin
are a and long o. Compare also the attested Latin form with the recon-
structed ablaut variants in the following:

factus ‘made’ < *dhh1k-to-
feci ‘I made’ < *dheh1k-
imitor ‘copy’ < *h2im-
aemulus ‘rival’ < *h2eim-

The reconstruction of laryngeals thus enables many different vowel alter-
nations to be reconciled to either an *e/*o or an *e/zero alternation.
In order to illustrate the operation of ablaut and its fate in Latin we
can take the suffix which is used to form comparatives of adjectives. In
PIE this could take the form *-yos- (cf. Latin melius ‘better’ which con-
tinues an earlier *mel-yos), *-is- (continued in Latin mag-is ‘more’), and
*-yes- (probably continued in Latin mulier ‘woman’ < *mp-yes-, perhaps
originally part of the paradigm of melior, but with a later shift in mean-
ing (from ‘the better woman’ to ‘the best woman in the house’, hence
‘wife, woman’, see Klingenschmitt 1992: 130). There was also a form with
lengthened vowel -ids which was restricted to the nominative singular. The
Latin paradigm of the comparative, outside of the neuter nominative/
accusative singular, alternates between -ior in the masculine and feminine
nominative singular (as melior) and -idr-, with lengthened vowel, in the
rest of the paradigm. This pattern is a completely new development, and
shows the spread of the lengthened form of the suffix throughout the
9781405162098_4_001.qxd 8/9/07 11:22 AM Page 13




Latin and Indo-European13

paradigm, with subsequent phonological changes. It is worth sketching
out the hypothetical development of the paradigm in Latin prehistory,
since this shows not only the analogical developments which led to the
loss of ablaut as a regular process, but also the interchange between sound
change and analogy in the creation of Latin paradigms.

Stage I*-ydsnominative singular masculine
*-is nominative/accusative singular neuter
*-yos-qaccusative singular masculine
etc.

The first change to affect this paradigm in the prehistory of is the spread
of the ablaut form *-yos from the masculine forms to the neuter; Latin
melius thus represents a very early replacement of *melis. The old form
of the neuter in *-is survives only in magis ‘more’, which early became
isolated from its paradigm owing to its widespread use as an adverb.

Stage II *-ydsnominative singular masculine
*-yosnominative/accusative singular neuter
*-yos-qaccusative singular masculine
etc.

The next change to affect the paradigm was the spread of the long d from
the nominative to the rest of the paradigm, with the exclusion of the neuter
singular, which did not enter into any of the subsequent paradigmatic
changes (hence Classical Latin -ius).

Stage III *-ydsnominative singular masculine
*-yds-eq accusative singular masculine

Stage III must have been reached at the beginning of the historical period,
sin

第1页 / 共333页

第2页 / 共333页

第3页 / 共333页

已预览3页，请下载后阅读全文

文档下载

©版权说明： 本文档由用户提供并上传，收益归属内容提供方，若内容存在侵权，请进行举报或认领！