178 Semantics

15. Can you contextualize ‘He had a pain in her stomach?' so that
utterancec would be explicable and the proposition
position iti expresses
conlmr.'i,lcmry (cf. p. 161)? Must e and her necessarily refe'r’m the o
6
person? Language-change
16. Explain what is meant by the truth-conditions of a sentence.
T . Ithas been said of *You are the cream in my coffee’ that it is **
el

which is neceslsavily false” (Kempson, 197;: 71). Do yo\i:t ;spea:’?c?:d

your ?nswer in relation to: (a) a particular interpretation of ‘neces-
sarily’; (b) the meaning of ‘you'; (c) the distinction between sentences
and utterances; (d) the author’s view of the interdependence of lmth"‘
6.1 Historical linguistics
conditionality and the literal interpretation of sentences. ‘What is now called historical linguistics was developed, in its main
. Discuss the validity of the notion of characteristic i lines at least, in the course of the nineteenth century (cf. 2.1).
to the analysis of indirect speech-acts. ik Scholars had long been aware that languages change with time.
19. )What distinction, if any,. would uld you draw between reference and They also knew that many of the modern languages of Europe were
descended, in some sense, from more ancient languages. For
20, “deixis is all-pervasive in the grammar and vocabul f example, it was known that English had developed out of Anglo-
languages™ (p. 170). Discuss. A
Saxon, and that what we now refer to as the Romance languages —
French, Spanish, Italian, etc. — all had their origin in Latin, How-
ever, until the principles of historical linguistics were established it
was not generally realized that language-change is universal, con-
tinuous and, to a very considerable degree, regular.
Each of these three aspects of language-change will be discussed
in some detail later. Here it may be noted that the universality and
continuity of the process of language-change — the fact that all living
languages are subject to it and that the process itself is going on all
the time — was obscured, for most people, by the conservatism of
the standard literary languages of Europe and by the prescriptive
attitudes of traditional grammar (cf. 2.4). The status of Latin is
particularly important in this respect. It had been used for centuries
in Western Europe as the language of scholarship, administration
and international diplomacy. Since the Renaissance, it had gradu-
ally yielded ground, in these functions, to the emergent Romance
languages, as well as to others that were not derived from Latin:
English, German, Dutch, Swedish, Danish, etc. By the nineteenth
century Latin was close to being a dead language, butitstill enjoyed
aprestige that set it apart from most other languages. Anditdoesso
to this day for traditionally minded grammarians. The importance
of the special position of Latin in the present context is that until
well after the Renaissance scholars could think of it as having
 180 Language-change 6.1 Historical linguistics 181
existed as a living language, more or less unchanged, for some habits of literacy that most of us still find it difficult, without special
2
years and as having been preserved from corruption, through training, to think in these terms.
ol
this period, by the usage of the educated and by the rules We often find it difficult, for example, to appreciate the full
anc
precepts of the grammarians. As we have seen, similar significance of the fact that, although languages may become extinct
attitug
were adopted in respect of the modern literary languages of at a particular point in time, so that, speaking metaphorically, we
Euro
when they came into being - or, more precisely, when they can talk of them as dying, there is no sense in which it is reasonable,
came to
be recognized as languages that could be used for literary making use of the same organic or biological metaphor, to talk of
purposes
— in the post-Renaissance period. languages as being born.! The point is worth making because, as we
Literary languages were more highly regarded than non-lit shall see, the terminology of historical linguistics is consistently
erary
languages and dialects; and any differences that were metaphorical. We group languages into families by virtue of their
noted by
grammarians between the literary and the colloquial, or betwee common descent from an earlier parent-language; and we say that
n
the standard language and non-standard dialects, tended languages that can be traced back to a common ancestral language -
to be
condemned and attributed to slovenliness or a lack of education, as the Romance languages can all be traced back to Latin - are
Few, if any, realized the significance of the fact that the transmi genetically related. When these terms were introduced into linguis-
s-
sion of the literary languages of Europe from generat tics in the nineteenth century, they were frequently given a more
ion to genera-
tion is highly untypical of the way in which people literal interpretation - under the influence of German romanticism,
acquire, as
children, their native language. Nor was sufficient attentio on the one hand, and of Darwinian evolutionism, on the other —
n paid to
the fact that in the case of many modern languages, notably than is generally the case nowadays. It must be appreciated that
English
and French, the spelling-system, which is still based on the pronun- there is no point at which, say, Anglo-Saxon suddenly was trans-
ciation of centuries ago, conceals most of the phonetic formed into, or gave birth to, English; and no point at which Latin
and phono-
logical changes that have taken place in them. If we are literate gave birth to the Romance languages, whilst continuing to exist
in
English or French we have relatively little difficulty in reading itself as a scholarly language for several centuries. And yet this is
Shakespeare or Ronsard; we would find their works more how the origin of languages is commonly conceived by non-
or less
incomprehensible if we heard them spoken as they were spoken linguists.
by
their authors. It was only after a great deal of detailed work The truth of the matter is that the transformation of one language
had
been done during the nineteenth century, in what we may into another is not sudden, but gradual. It is largely a matter of
now
think of as the classical period of historical linguistics, from convention and arbitrary decision that leads us to divide the history
the
1820 to the 1870, that scholars came to a better understanding of English into three periods — Old English (or Anglo-Saxon),
of the relation between written and spoken languages, Middle English and Modern English - and to consider these alter-
on the
one hand, and between standard and non-standard languages, natively as three different languages or as three stages of the same
on
the other. language. There are both linguistic and non-linguistic reasons for
On the basis of this detailed research and by applying the making the divisions where we do. What is now Standard English is,
so-
called comparative method (which will be explained in 6.3), in the essential features of its phonology and grammar and in much
it was
demonstrated beyond doubt, not only that all the great literary of its vocabulary, a descendant of the dialect of London, which
languages of Europe had originated as spoken dialects, but
also that
their origin and development could only be explained in terms ! Pidgins and creoles might perhaps be said to be born of the union of two parent-
of
principles which determine the acquisition and use of the associa languages, each of which continues to exist during the life-time of their offspring
ted (cf. 9.3). But this more or less acceptably metaphorical interpretation of ‘paren-
spoken language. Such is the force of traditional attitudes
and of the tage’ and 'birth' is not what is at issue here,
182 Language-change 6.1 Historical linguistics 183
being close to the point where three of the four major Anglo tion and business, not only in the United States, Canada, Australia
dialects came into contact - Mercian, West Saxon, and Kentish and New Zealand, where it was now the first language of most of the
contains features from all three, It also contains a few more isolate politically and economically dominant settlers and their descen-
features which derive from the fourth major dialect, Northumbri: dants, but also in India and other Asian and African countries
~notably the forms they, their, them and most of the words written within the British Empire. English in the post-Renaissance period
with initial sk- (‘skill’, ‘sky’, ‘skin’, ‘skirt’, ete.) — which was strony has become a world language in much the same way that Latin had
influenced, from the ninth century, by the language of the Vikings. become a world language (in the so-called old world of Europe,
For about a century and a half after the Norman Conquest in North Africa and parts of Asia) almost 2000 years earlier; and for
1066, the language of the ruling classes was French, as far much the same reasons. But both Latin and English were in origin
literature and administration were concerned; and when Eng nothing other than the local dialects of small tribes, Italic in the one
came to be used again as a literary language at the beginning of th case and Germanic in the other, and did not differ in any linguisti-
thirteenth century, it was very noticeably different from Anglo cally relevant detail from the related Italic and Germanic dialects
Saxon of the earlier period. Apart from other developments thal of neighbouring tribes.
had taken place, what we now refer to as Middle English had com The brief and highly oversimplified account of the evolution and
under the influence of Norman French and been deeply affected b expansion of English that has just been given will serve to illustrate
it in vocabulary and grammar. Chaucer, for example, wrote in the the general point that, although there may be good reason to divide
London dialect of Middle English, which, by virtue of the political both the external and the internal history of a language into more or
and economic importance of the capital, was now beginning i less distinct periods, the process of language-change itself is con-
emerge as a standard national language. By the end of the Hundred tinuous. What produces the illusion of discontinuity between, for
Years’ War in the fifteenth century, England had become very example, Anglo-Saxon and Middle English or, to a lesser degree,
conscious of its own independent national identity and had been between Middle English and Modern English, is the coincidence of
transformed from a feudal state to one with an educated, wealthy several factors, including, on the one hand, gaps in the historical
and increasingly powerful bourgeoisie. This was an important record between identifiable periods and, on the other, the relative
factor in the formation and increasing standardization of literary stability of literary languages over quite long stretches of time. We
Middle English. have very little in the way of non-literary written records for the
The period of Middle English is separated from that of Modern various dialects of Anglo-Saxon and Middle English. But we can be
English by the Renaissance, which reached England in the late sure of two things: first, that from the earliest times the dialects of
fifteenth century, One of the most important consequences in the spoken English were less homogeneous and less neatly separable
sphere of education and culture was the revival of Latin as a literary one from another than traditional accounts of the history of English
language. But this was a relatively short-lived phenomenon. based on the evidence of literary texts represent them as being; and,
Although Latin continued to enjoy great cultural prestige until well second, that, if we had a full historical record of any one spoken
into the nineteenth century, the greatest literary works of the dialect, whether it was the dialect of London or of some small
Elizabethan and post-Elizabethan period, including the plays of village in a remoter part of the country, we should be unable to
Shakespeare and Milton’s Paradise Lost, were written in English. identify any definite time at which the dialect in question suddenly
Meanwhile, Britain was beginning to play an increasingly important changed from being that of one period to being that of another.
role in world affairs. English-speaking colonies were established in Languages change more rapidly in certain periods than they do in
North America in the seventeenth century. And by the nineteenth others. Even literary languages change in the course of time; and
century English was the language of administration, higher educa~ spoken languages acquired in childhood and employed throughout
184 Language-change 6.2 Language-families 185

life in a variety of situations

- living languages in the full sense of the structs, for whose existence there is no direct evidence, but which
term - change far more obviously than literary languages do. are postulated as having existed and as being of such-and-such a
Furthermore no living language is completely uniform (cf. 1.6). structure, in order to account for the genetic relatedness of two or
And this fact, as we shall see later, is crucial for the explanation of more attested languages. For example, Proto-Germanic is postu-
language-change. lated as the ancestor of the Germanic languages (English, German,
In what follows, I will begin by giving an account of historical Dutch, Danish, Icelandic, Norwegian, Swedish, etc.); and
linguistics of the kind that might have been given (except in respect Proto-Slavonic as the ancestor of the Slavonic languages (Russian,
of details that have come to light more recently or relate specifically. Polish, Czech, Slovak, Serbo-Croatian, Bulgarian, etc.). In each
to the present day) by one of the so-called Neogrammarians or their case, we have documentary evidence relating to the earlier history
successors, The Neogrammarians (Junggrammatiker, in German). of the family. For Germanic, apart from a number of earlier frag-
were a group of scholars, based at the University of Leipzig in the mentary inscriptions, we have the fourth-century a.p. translation of
late nineteenth century, who were largely responsible for formulat- the bible in Gothic (as spoken by the Visigoths who were settled at
ing the principles and methods of historical linguistics that have the time on the lower Danube); quite extensive literary texts in the
since governed most work in the subject. These principles and various dialects of Anglo-Saxon (or Old English) covering the
methods were controversial when they were first proclaimed; and period from the sixth to the eleventh century A.D.; the texts of the
much of the criticism that was directed against them we now know. 0ld Icelandic (or Old Norse) sagas of the twelfth century A.p.; Old
to have been justified. However, they have reigned supreme for High German texts dating from the second half of the eighth cen-
almost a century; and they underlie much of the linguist’s everyday tury A.D.; and so on. For Slavonic, the earliest evidence that we
thinking about language-change, as well as being the basis for all the have to go on is that of the ninth-century A.D. texts written in Old
standard treatments of language-families in encyclopaedias and Church Slavonic. In neither case do we have anything as close to the
other works of reference. In later sections of this chapter one or two. postulated ancestral proto-language as the language of the Latin
of the Neogrammarian principles will be called into question and texts which have come down to us must have been to that presum-
reformulated in the light of recent work. ably more popular dialect of Latin (often referred to as Vulgar
Latin) which we are calling Proto-Romance.
6.2 Language-families On the basis of all the available evidence, and by applying the
To say that two or more languages belong to the same family — that principles which were claborated in the course of the nineteenth
they are genetically related - is to say that they are divergent century and formulated in their essentials by the Neogrammarians,
variants, descendants, of a common ancestral language, or proto- scholars can reconstruct with a fair degree of confidence much of
language. the sound-system and some of the grammatical structure of both
In most cases we have no direct knowledge of the proto-language Proto-Germanic and Proto-Slavonic. They can also reconstruct
from which the members of a particular family, or subfamily, are intermediate stages in the development of the attested members of
descended. The Romance languages are in this respect highly un- a particular language-family from their assumed common ancestor.
typical: although the dialect of Latin from which they are descended For example, Figure 3 gives a schematic representation of the
must have differed in many details of grammar and vocabulary from development of the officially recognized Germanic languages
that of even the more colloquial texts that have come down to us, we spoken today and of Gothic, which went into decline in the early
have a much better idea of the structure of what we might call Middle Ages and finally died out (yielding to one or other of the
Proto-Romance than we have of most other proto-languages. Slavonic dialects) some centuries later. It will be noted that English
Generally speaking, proto-languages are hypothetical con- which, as we saw in the preceding section, was already dialectically
 186 Language-change 6.2 Language-families 187
*Proto-Gme evidence can be dated to about the middle of the second millennium
B.C. — we can partially reconstruct the phonology and some of the
*Proto-WGme *Proto-NGme *Proto-EGme grammatical characteristics and vocabulary of Proto-Indo-
European, the hypothetical ancestor of Proto-Germanic, Proto-
*Proto-AF *Proto-NethGmn *Proto-WScand *Proto-EScand Gothic Slavonic, Proto-Celtic, Proto-Italic, Proto-Indo-Iranian, etc., and
English Frisian Icelandic Norwegian ultimately of all the Indo-European languages, ancient and
Dutch German Danish Swedish modern.
We can even locate Proto-Indo-European, more or less con-
Figure 3. The Germanic languages. Reconstructed proto-languages
are fidently, in space and time — to the plains of South Russia in the
indicated by means of a preceding asterisk; the names of extinct
languages are italicized. Gme = Germanic; WGme = West Germanic fourth millennium B.c. ~ and, by combining the linguistic and
;
NGme = North Germanic; EGmc = East Germanic; AF = archaeological evidence, say something about the culture of those
Anglo-Frisian; Neth-Gmn = Netherlandic German; WScand = West who spoke it. For example, most of the earliest-attested Indo-
Scandinavian; EScand = East Scandinavian, (Much of the detail is European languages have words that can be traced back to
omitted: e.g. High German is not distinguished from Low German, and
hypothetical source-words meaning “horse”, ‘“dog”, ‘“‘cow”,
older attested ancestors of modern languages ~ Anglo-Saxon, Old High
“sheep”, etc. That the reconstructed vocabulary of Proto-
German, etc. - are not included.)
Indo-European contains these words, as well as words relating to
spinning, weaving, ploughing and other agricultural and pastoral
differentiated at the time of our earliest surviving records, is shown
occupations, indicates fairly clearly that the speakers of Proto-
as being more closely related to Frisian than it is to Dutch or Indo-European led a relatively settled existence, Words denoting
German, and as being more closely related to both of them than itis
flora and fauna, climatic conditions, etc. , enable us to identify, within
to the Scandinavian languages. Frisian was once far more widely limits, their geographical habitat, whilst common vocabulary relat-
spoken than it is today. Although it is not a national language in the
ing to social and religious institutions makes it possible to draw
sense that all the other modern Germanic languages shown in
inferences about the more abstract features of their culture. It is
Figure 3 are, it has official status in the province of Friesland in the
quite clear, for example, that their society was patriarchal and that
northern part of the Netherlands, where it is being heavily in-
they worshipped a sky-god and other deified natural phenomena.
fluenced, however, in vocabulary at least, by Standard Dutch. Not
As to the archaeological evidence, it has recently been suggested
only English, but all the modern languages shown in Figure 3 exist
that this points to the Proto-Indo-European speakers being the
inseveral dialects; and very often the transition between one dialect
bearers of the so-called Kurgan culture, a Bronze Age culture
and another is not sharp, but gradual. As we shall see later, the
which spread westwards from South Russia in the earlier half of the
conventional family-tree-diagram of language-relatedness tends
to fourth millennium s.c. and eastward into Iran somewhat later. This
oversimplify the facts, if not to distort them completely, by failing to
hypothesis, though perhaps the most plausible one so far produced,
give recognition to the phenomena of convergence and diffusion
is not universally accepted; and many scholars are sceptical about
and by representing language-relatedness as being the result of
the possibility of saying anything very definite, on present evidence,
necessary and continuous divergence.
about the habitat and culture of the speakers of Proto-Indo-
Going further back and taking a wider range of evidence into
European.
account, including that of the Hittite inscriptions of Asia Minor
The reason why I have mentioned this question at all is that the
(deciphered in 1915), the Mycenaean Greek tablets (deciphered in
Indo-European family of languages occupies a rather special place
1952) and, for the earliest Sanskrit, the Vedic hymns ~ all of which
in historical linguistics. This is in part due to the fact that many of
 188 Language-change 6.2 Language-families 189
the Indo-European languages, as we have seen, have written re- Celtick, though blended with a very different idiom, had the same
cords going back hundreds, if not thousands, of years. Although origin with the Sanscrit; and the old Persian might be added to the
many of the relationships that hold within the Indo-Europe: same family.”
family could undoubtedly be established on the basis of modern
There is much in this famous quotation that is worth noting.
spoken languages, the details of these relationships could not be
However, the point that must be given particular emphasis is that
worked out—and Proto-Indo-European could not be reconstructed
what seemed so obvious an explanation to Jones, at the end of the
to the degree that it has been reconstructed - without the evidence
eighteenth century, of the remarkable similarity that he observed
of the older texts. between the classical languages of Europe and Sanskrit — the
But to say that it would be possible to group many, if not all, the hypothesis of family-relationship — might not have seemed so ob-
modern Indo-European languages into a single family even if we vious in another age, or indeed to someone of different educational
had no records of the earlier stages of these languages is to presup-- background and a less liberal outlook. Evolutionary ideas were in
pose that the idea of grouping languages into families should the air and from the mid-eighteenth century had been applied to
already have occurred to us, and furthermore that we should have language by such scholars as Condillac (1715-80), Rousseau (1712~
at our disposal a reliable method for comparing languages and 78) and Herder (1744-1803), not to mention James Burnett (1714~
demonstrating genetic relatedness. This brings us to the second 99) with whom Sir William Jones was in correspondence at the time.
reason why the Indo-European family of languages has pride of By the end of the eighteenth century, as a consequence of the
place in historical linguistics: it was the reconstruction of Proto-Indo- post-Renaissance expansion of Europe, a lot more was known
European, and of the intermediate proto-languages for the sub- about the diversity of the languages of the world. It was no longer
families of Indo-European (notably for the Germanic subfamily), possible to maintain with the same degree of plausibility as previous
which provided the motivation and ultimately the methodology upon generations of classically trained scholars had done that all lan-
which historical linguistics, as we know it, now depends. It is arguable guages must be similar in structure. The resemblances between
that, not only historical linguistics, but linguistics itself as an indepen- Greek and Latin had been taken for granted for centuries. But in
dent and scientific discipline, originated in what might be described, the context of what was known about language-diversity the fact
somewhat romantically, as the nineteenth-century quest for that Sanskrit was strikingly similar to Greek and Latin cried out for
Proto-Indo-European. explanation; and the explanation that seemed so natural to Sir
It is customary to date the beginning of Indo-European scholar-
William Jones, and to his contemporaries once he proposed it, was
ship with the statement made in 1786 by Sir William Jones (1746~ the one that general movements in European thought at that time
94) about Sanskrit, the ancient sacred and literary language of made available to him.
India, and its relatedness to Greek, Latin and other languages: Something should also be said about the importance of the new
““The Sanscrit language, whatever be its antiquity, is of a wonder- spirit of romanticism, which was particularly strong in Germany,
ful structure; more perfect than the Greek, more copious than the and its connection with nationalism. Herder had maintained that
Latin, and more exquisitely refined than either, yet bearing to both there was an intimate connection between language and national
of them a stronger affinity both in the roots of verbs and in the forms.
character. This idea took deep root in Germany and contributed to
of grammar, than could possibly have been produced by accident; the development of a climate of opinion in which the study of earlier
so strong indeed, that no philologer could examine them all three, stages of the German language was seen as being an integral part of
without believing them to have sprung from a common source
the assertion and authentication of the national identity of the
which, perhaps, no longer exists: there is a similar reason, though German-speaking peoples.
not quite so forcible for supposing that both the Gothick and the In this connection it is important to emphasize the difference
190 Language-change 6.2 Language-families 191

between language and race. Terms like ‘Germanic’ and “Indo- family. On the other hand, the several quite striking structural
European’ apply primarily to language-families. They do not apply correspondences among the languages of the world, which at
to anything that a physical anthropologist would regard as geneti- first sight would seem to support the hypothesis of monogenesis,
cally distinct races. There is no such thing, and never has been, asa are no less readily accounted for in terms of diffusion and
Germanic or Indo-European race. In so far as the use of these terms convergence.
in historical linguistics implies the existence of a language- The transmission of language from one generation to the next is
community speaking Proto-Germanic or Proto-Indo-European at partly a matter of biology and partly a matter of culture. We may be
some time and in some place in the past, it is reasonable to assume genetically programmed, as human beings, to acquire language; but
that the members of these language-communities may have thought we are not genetically programmed to acquire a particular
of themselves as belonging to the same cultural and ethnic groups. language. It follows that, given the right social and cultural condi-
The possession of a common language is, and presumably always tions, not only individuals, but whole communities, can acquire a
has been, an important mark of cultural identity and ethnicity. But language or dialect which differs from that of their parents. The
there is no connection, other than partial and coincidental, between great nineteenth-century founders of historical linguistics, to whom
race, genetically defined, and either culture or ethnicity. we are indebted for the notion of language-families with which we
This point is worth emphasizing for two reasons. The first is that still operate, did not give to this fact the theoretical importance that
terms like ‘Germanic’ and ‘Indo-European’ - or alternatively ‘Nor- we now know it should have. All too often the spread of languages
dic’ and ‘Aryan’ - have often been given a racial, and indeed racist, over a large area was assumed to imply great movements of people.
interpretation. It is up to the linguist and the anthropologist to This is, to say the least, an unnecessary assumption. We shall see
correct the misconception upon which this particular kind of racism later that cultural diffusion and convergence are no less important
is based. There is no warrant at all for belief in the racial distinctive- for the explanation of language-change than are migration of
ness of the speakers of Indo-European languages; still less for the peoples and divergence. The traditional family-tree model of
use that was made of the assumption of racial superiority by Nazi language-relatedness does not allow for anything other than the
propagandists in the 1930s. The same holds true in respect of terms continuous divergence of languages from a common ancestor,
like “Celtic’, ‘Slavonic’, or ‘English’; or indeed of any terms that The Indo-European family of languages is but one of very many
apply primarily to language-families and languages. different language-families so far identified. Some scholars have
The second reason for emphasizing the fact that there is no argued for a classification of all the languages of the world into
intrinsic connection between race and language - and it reinforces about thirty major families, of which a number of the more gener-
the first — is that it gives us a better understanding of how language- ally accepted families would be subfamilies. But much of this more
families are formed and therefore of the nature of language. We do comprehensive genetic classification and subclassification of
not know whether there was once a single proto-language from languages is still controversial. For example, something like a
which all human languages are descended, in the way that the thousand different languages are spoken in Africa. All of these
Germanic languages are descended from Proto-Germanic, and (except for English, French, Spanish, Afrikaans, etc., which were
Proto-Germanic in turn from Proto-Indo-European. We cannot brought to Africa in the period of European colonization) have
even relate the Indo-European languages with certainty to any of recently been grouped into four large families, one of which,
the other major language-families so far established. It is quite Hamito-Semitic (or Afro-Asiatic), comprising all the indigenous
possible that all languages go back in the far distant past — perhaps languages spoken north of the Sahara, contains the traditionally
half a million years ago—to a single ancestral language and are thus, recognized Semitic family, whose best known members are Arabic,
in the technical sense of the term, members of the same language- Hebrew and Ambharic. Similarly, the Bantu languages (including
 192 Language-change 6.3 The comparative method 193
Swahili, Xhosa, Zulu, etc.) are now generally, though not univer:
Table 4. Some systematic correspondences of form in Latin and
sally, regarded as a subfamily of the Niger~Congo family. Much th
three Romance languages
same situation obtains in respect of the languages spoken elsewhe
in the world. Progress is gradually being made in the grouping of a
larger number of subfamilies into a smaller number of what might Latin (L) French (Fr) Italian (It) Spanish (Sp)
be called super-families (or language-phyla). Since the evidence for
(1) “thing” causa chose cosa cosa
the larger groupings is frequently very scanty, the resulting genetic “head” caput chef capo cabo
classification which depends upon it is correspondingly tentative, “horse” caballus cheval cavallo caballo
and must be treated as such. Not all the language-families identified “'sing” cantare chanter cantare cantar
and named by linguists are equally well established. “dog" canis chien cane
“‘goat™ capra chévre capra cabra
6.3 The comparative method
(2) “plant™ planta plante pianta llanta
The standard way of demonstrating the genetic relatedness of “key” clavis clef chinvg Ilavu_:
languages is by means of the so-called comparative method. This “rain" pluvia pluie pioggia lluvia
was developed and refined in what was referred to earlier as the
classical period of historical linguistics: between the 1820s and the (3) “eight” octo huit otto ocho
1870s (cf. 6.1). It rests upon the fact that many of the most obviously “night” nox/noctis nuit notte noche
“fact” factum fait fatto hecho
related words across languages can be put into systematic corres- “milk"” lacte lait latte leche
pondence in terms of their phonological and morphological struc-
ture. By the 1870s scholars had achieved such a high degree of (4) “daughter” filia fille figla hija
success in the application of the comparative method to the more “beautiful formosus hermoso
obvious instances of genetic relatedness that they felt confident of
its reliability in the case of languages whose relatedness was far
and vertically, by adding many more sets of related words. Limited
from obvious.
though it is, the information given in Table 4 will serve to illustrate
I will exemplify the principle of systematic correspondence, in what is meant by the principle of systematic correspondence.
the first instance, from the Romance languages. This has the adyan- The first point to note is that the words brought together in each
tage, not only that the fact of their relatedness is beyond dispute,
row of Table 4 are related, not only in form - to which the principle
but also that we have direct evidence of the proto-language from
of systematic correspondence applies — but also in meaning. Now,
which they are derived, Latin, However, as we shall see, there are
words can change their meaning in the course of time. For example,
instances where the Proto-Romance forms which, given evidence
the ordinary Classical Latin for “horse” was not ‘caballus’, which
from the Romance languages themselves, can be reconstructed by
had the more specific meaning of *“‘pack-horse™ and was also used
means of the comparative method differ from the attested Latin forms,
pejoratively to mean “nag” or “hack"”, but ‘equus’. Howeyer.
Table 4 brings together for comparison several sets of obviously
‘caballus’ and ‘equus’ are obviously related in terms of meaning;
related words (in their normal orthographic citation-forms) from
and it is plausible to suppose that ‘caballus’ lost both its specific
Latin and three of the Romance languages: French, Italian and
meaning and its pejorative overtones in late Latin (i.e. in Pmlf»
Spanish. The table could be extended both horizontally, by includ-
Romance) and became the general and stylistically neutral word in
ing the appropriate words from other Romance languages and
place of ‘equus’. Conversely, the descendants of the Latin word
dialects (Rumanian, Portuguese, Catalan, Sardinian, Ladino, etc.),
‘caput’ shown in the table have all acquired a set of narrower or
194 Language-change 6.3 The comparative method 195
metaphorical senses: e.g. Fr. ‘chef’ means “head” in the sense of pronunciations for fait, However, in so far as French orthographic
““chief"” or “boss", as does It. ‘capo’; and Sp. ‘cabo’ means “‘cape” practice is based on the pronunciation that was current some cen-
(i.e. “headland™), “corporal”, “end", etc. But once again there is turies ago (as is also the English spelling-system), we can take the
an intuitively obvious connection between the meaningof L. ‘caput’ French written forms, for present purposes, at their face value. Nor
and the meanings of its descendants. need we be concerned with the fact that it is not usually the Latin
None of the words listed in Table 4 causes any problems as far as citation-form of a lexeme that is the source of the diachronically
the fact of their being semantically related is concerned, even related forms of French, Spanish and Italian: it is almost always the
though there may be room for disagreement, in particular in- accusative form of nouns and adjectives, rather than their nomina-
stances, about the nature of their relationship. Very often, how- tive form, which provides the source of the Romance stem-forms -
ever, itis not clear — especially in the case of languages for which we canem, caballum, etc. (the final [m] being lost in late Latin, or
have far less evidence than we do in the case of the Romance Proto-Romance).
languages — whether two words are semantically related or not. Itis If the words in the different sections of Table 4 are compared, it
for this reason that the comparative method gives priority to re- will be observed that there are regular correspondences holding
latedness of form. It should also be noted that words may not only among related forms (i.e. among the forms of related lexemes).
change their meaning in time, but also, for various reasons, fall into These correspondences are indicated in bold type and may be
disuse and be replaced. This explains the gaps in Table 4. Moaern represented, in terms of sounds rather than letters, but, as ex-
Spanish has replaced the word derived from L. ‘canis’ with ‘perro’; plained above, taking the orthography at face value, as:
and neither Italian nor French preserves in its vocabulary any
descendants of L. ‘formosus’. (1) L. [K] = Fr. [f] = It. (k] = Sp. [K]
To turn then to the correspondences of form exemplified in the (2) L. [pl], [kI] = Fr. [pl], [kI] = It. [pi], [ki] = Sp. [4]
table. The words are all given in their written citation-form. It must (3) L. [kt] = Fr. [it] = It. [tt] = Sp. [tf]
be remembered, therefore, that we are concerned, in principle, not (4) L. [f] = Fr. [f] = It. [f] = Sp. [h]
with letters, but with sounds. As far as Latin, Spanish and Italian Both Fr. <it> and Sp. <h> are here given a phonetic value which we
are concerned, there is relatively little discrepancy between spelling know to be appropriate for earlier periods: this is what I mean by
and pronunciation. We have to keep in mind the fact that in Modern taking the orthography at face value. We could just as well operate
Spanish there is no phoneme in correspondence with the letter <h>; with phonetic transcriptions (or with phonological representations)
that in both Spanish and Italian the letter <¢> is pronounced dif- of the modern spoken forms. The systematic correspondences of
ferently in different positions; that <ch> is pronounced (k] in Italian, form could still be formulated. But they would be less immediately
but [tf] in Spanish; and so on. But these are minor discrepancies; obvious. It will be noted that, in addition to (1)-(4), other corre-
and we can operate, without doing too much violence to the facts, spondences can be extracted from the words given in Table 4:
on the assumption that there is a one-to-one correspondence be-
tween single letters (or, in certain instances, groups of letters: Sp. (5) L. [b] = Fr. [v] = It. [v] = Sp. [b]
<>, <ch>; It. <ch>, <ggi>) and phonemes. The situation is very (6) L. [a] = Fr. [¢] = It. [a] = Sp. [a]
different for French. For example, there is no way of knowing on and so on. How do we explain these systematic correspondences?
the basis of the orthographic conventions of French that clef is The answer given by the nineteenth-century inventors of the
pronounced [kle], but chef is pronounced [fef]; or that huit is comparative method was that the sound-changes which take place
generally pronounced with a final [t], whereas nuit and lait are not in a language in the course of its history are regular. The principle of
(except in certain fixed expressions), and that there are alternative the regularity of sound-change was not emphasized, however, until
 196 Language-change 6.3 The comparative method 197

the mid-1870s, when the Neogrammarians proclaimed it in its Now Grimm’s Law, thus formulated, covers a large number of
strongest and most uncompromising form: “The sound changes observed correspondences. For example, it accounts for the fact
which we can observe in documented linguistic history proceed that English (E.) has [f] where Latin (L.), Greek (Gk), Sanskrit
according to fixed laws which suffer no disturbance save in accor- (Skt), etc., have [p]: cf. E. father: L. pater, Gk pater, Skt pitar-; E.
dance with other laws.” At first sight, the thesis that sound-laws (as foor: L. pesipedis, Gk pousipodus, Skt pdt/padas. 1t also accounts
they had now come to be called) operated without exception was for the initial and medial consonants of Gothic tathun: L. decem,
patently false. There were many instances of obviously related Gk deka, Skt dasa - E. ten preserves no trace of the medial conso-
words which did not exhibit the expected correspondences. Let us nant, but cf. Modern German ze/n, as well as Old High German
take a famous example — one which turned out to be no more than zehan and Old Saxon fehan (the initial [t:] of German represented
an apparent exception when the problem that it gave rise to was by the letter <z> in the orthography results from the so-called High
brilliantly resolved by the Danish scholar, Karl Verner, in 1875. German Sound-Shift, which probably took place about the sixth
In 1822, Jacob Grimm (one of the two brothers who are more century A.p.). The [J]-sound of Skt dasa, represented here by <>,
widely known for their interest in Germanic folklore) pointed out results from a palatalization of PIE [*k], which, at a very early
that there is a systematic correspondence holding between the prehistoric time, affected many of the eastern subfamilies of the
consonants of the Germanic languages, on the one hand, and of the Indo-European languages, including all the Indo-Iranian, Baltic
other Indo-European languages, on the other. He was not the first and Slavonic languages, as well as Armenian and Albanian: there
to notice this correspondence: the credit for the original observa- are certain complications attaching to the reconstruction of what 1
tion must go to the Danish scholar, Rasmus Rask. But Grimm’s have, for simplicity, taken to be velar stops, [*g®, *g, *k]. in PIE;
work, being written in German, was more readily accessible to but they do not bear upon the general formulation or validity of
international scholarship; and the sound-law that was postulated to Grimm s Law. Due allowance being made for subsequent develop-
account for the observed correspondences is generally known as ments in particular languages, or intermediate proto-languages,
Grimm’s Law. Reformulated in the terms of modern articulatory Grimm's Law, as summarized in the previous paragraph, is sup-
phonetics (and simplified in certain respects), Grimm's Law says ported by an impressively large number of instances of systematic
that: correspondences.
(a) Proto-Indo-European (PIE) voiced aspirates [*bb, *d, *g¥] But there were many apparent exceptions. Of some of these,
became voiced stops [*b, *d, *g] - or possibly voiced frica- Grimm himself had commented: “The sound-shift succeeds in the
tives [*B, *8, *y] - in Proto-Germanic (PGme); majority of cases, but never works itself out completely in every
(b) PIE voiced stops [*b, *d, *g] became voiceless stops [*p, *t, individual case; some words remain in the form they had in the
*k] in PGmc; older period; the current of innovation has passed them by." For
(c) PIE voiceless stops [*p, *t, *k] became voiceless fricatives example, the [p] of E. spit, spew corresponds to a [p] in other
[*f, *6, *h] in PGmec. languages, in apparent violation of Grimm's Law: L. spuo, etc.
Similarly, for Gme [t] = L. [t], Gk [t], Skt [t]): cf. E. stand: L.
The asterisks indicate, according to the convention that has long stolstare, etc. Here indeed, as Grimm said, the current of innova-
been established in historical linguistics, that the soundsin question
tion has left the Germanic consonants unchanged. It will be
are reconstructed, rather than being directly attested. We shall take
noticed, however, that in each case the voiceless stops, [*p. *t. *k],
up the notion of reconstruction presently. Both Proto-Indo- occur as the second segments of two-segment consonantal clusters.
European and Proto-Germanic are, of course, hypothetical con-
All that we need to do, therefore, is to modify the formulation of
structs (cf. 6.2).
Grimm’s Law given above, so that it is assumed not to have applied
 6.3 The comparative method 199
198 Language-change

to the Germanic reflexes (i.e. descendants) of PIE [*p, *t, *k] in this (iii) *fadér-: *broar-
(iv) PGmce *fdder-: *bro6ar-
kind of phonetic (or phonological) environment. In effect, we are
saying ~ to introduce some modern terminology — that what
Grimm's Law is traditionally held to account for the transition from
Grimm’s Law accounts for was a phonetically conditioned sound-
(i) to (ii), and Verner's Law for the transition from (ii) to (iii). Both
change. Provided that it is so formulated, the preservation of a
laws, it will be noted, are held to have operated prior to the period
voiceless stop in words like E. spit/spew, stand, eight, etc., can be
which we identify as that of Proto-Germanic, which is characterized
seen as a regular development.
by having the word-stress on the initial syllable of all words. The
More interesting is another class of apparent exceptions. If we.
sound-changes that are accounted for by Grimm’s Law and Verner's
take the words for “father” and “brother” in various Germanic
Law taken together might be explained somewhat differently
languages other than English, we see that they differ in respect of
nowadays. This is of no consequence in the present connection. The
the medial consonant: Go. fadar : bro6ar, G. Vater : Bruder, etc.
point is that a whole class of apparent exceptions to Grimm's Law
And Old English shows the same difference: OE faeder : bropor.
were shown by Verner to fall within the scope of a supplementary
The fact that German, like its ancestor Old High German (fater :
generalization: another sound-law.
bruodar), has a voiceless stop in the word for “father’ and a voiced
stop in the word for “brother”, can be explained, once again, as
Several other so-called sound-laws were formulated about the
same time as Verner’s Law. Taken together, they gave scholars a
being the consequence of the High German Sound-Shift. Let us
much better idea of the relative chronology of developments within
grant that, on the basis of all the evidence, we can reconstruct as the
the different branches of the Indo-European family of languages.
Proto-Germanic sources of the words in question, *fader- and
More important, they made the famous Neogrammarian principle
*bro6ar-, what is represented by means of <d> being either a stop (d]
of the absolute regularity of sound-change far more plausible than it
or a fricative [8], but in either case voiced and thus different from
would have seemed to an earlier generation of historical linguists.
the voiceless fricative [0] of the word for “brother”. Since the
This principle was highly controversial when it was put forward in
corresponding words of the non-Germanic Indo-European
the 1870s. However, it soon came to be accepted, by most of those
languages show no such difference (L. pater: frater, Skt pitar-:
in what we may think of as the mainstream of scholarship, as the
bhratar-, etc.) and, according to Grimm’s Law, PIE [*t] should
very basis, not only of the comparative method, but of the whole
yield PGme. [*0], the word for “father” appears to be irregular as
discipline of historical linguistics. We shall have occasion to look
far as its medial, but not its initial, consonant is concerned.
more critically later at the principle of the regularity of sound-
It was this problem that was solved by Verner. He demonstrated
change, and at the use that the Neogrammarians made of the term
that, on the assumption that the PIE words for “father” and
‘law’ in connection with it. But nothing that is said in qualification
“brother” had differed with respect to the place of the word-stress,
of it should be taken as detracting from its methodological
as they do in Sanskrit (pitdr- : bhrdtar-), the apparent exceptional
significance. It forced those who subscribed to it to draw a
status of PGmc *fader could be satisfactorily explained in terms of
distinction between conditioned and unconditioned sound-change
what is now called Verner’s Law: intervocalic voiceless fricatives,
and to formulate as precisely as possible the conditions under
e.g. [0], become voiced unless they are immediately preceded by
which a particular conditioned sound-change took place. And it
the word-stress. What is assumed is a sequence of stages that can be
laid upon them the onus of providing an explanation for the forms
represented as follows:
which had not developed in accordance with the sound-laws
whose conditions they appeared to fulfil. Two explanatory factors
(i) PIE *patér-: *bbrdter-
to which the Neogrammarians and their followers appealed in
(i) *fabér-: *bréfar-
200 Language-change 6.4 Analogy and borrowing 201

this connection, as we shall see, were analogy and borrowin) form for all the attested forms. It follows that the reconstructed
(cf. 6.4). ) language-system is likely to be, not only morphologically more
We are concerned in this section with the technique of historic regular, but also dialectically more uniform than any actual
reconstruction by means of the comparative method. It is fit language-system. Furthermore, we have no way of knowing
that the reader should now be cautioned against misinterpreting t whether all the sounds that occur in a starred form did in fact
so-called starred forms (i.e. the hypothetical forms to which co-oceur at the same time and in the same dialect of the proto-
asterisk is prefixed: e.g. PIE *patér- or PGmce *fader-) which re language.
from the application of the technique of reconstruction. They mus| For these and other reasons, reconstructed proto-languages must
not be identified with the actual forms of Proto-Indo-European of be taken as hypothetical constructs, whose relation to actual spoken
of any other proto-language. There are several reasons why this i languages of the past is rather indirect. We cannot go further into
S0. the technicalities of this question — or into the several criteria that
First of all, the comparative method tends to exaggerate the must be weighed in the process of reconstruction. It suffices, for our
degree of regularity in a reconstructed language-system. This point. purposes, to have drawn attention to the fact that all historical
may be illustrated with reference to the differences between certain reconstruction tends to idealize and oversimplify the facts. As far as
attested Latin forms and what we take to be the Proto-Roman particular starred forms are concerned, some parts of the recon-
forms from which the corresponding forms in French, Italian, § struction may be more soundly based than others; and no part can
ish, etc., are derived. The attested Latin word for “*head” had caput be sounder than the evidence that supports it. The evidence is
as its citation-form and capit- as its stem-form. None of the highly variable.
Romance languages preserves any evidence of the stem-final [t]. In conclusion, it should be mentioned that, although we have
They indicate, instead, that the Proto-Romance form was been concerned throughout this section with lexical reconstruction,
*capu(m): see Table 4 above. Now it is quite likely that the irregular it is possible, in favourable cases, to reconstruct features of the
noun ‘caput’ was regularized in late Latin. But we have no direct grammatical structure of proto-languages. It was of course the
evidence that it was. It may have been regularized independently, morphological correspondences between related languages that so
but at a relatively early date, in different branches of the Romance impressed the first generation of historical linguists, since such
family. The point is that irregularities tend to disappear in the grammatical features as inflectional endings, they felt, were un-
course of time and, generally speaking, the comparative method is likely to spread from one language to another by borrowing
unable to reconstruct them. (cf. 6.4).
Secondly, the comparative method operates on the assumption
6.4 Analogy and borrowing
that each member of a family of related languages is in a direct line
of descent from the proto-language and has been unaffected, The concept of analogy goes back to antiquity. The term itself is
throughout this time, by contact with other related languages and from the Greek word ‘analogia’, which meant “regularity” and
dialects. This is, to say the least, an unrealistic assumption. All more especially, in the usage of mathematicians and grammarians,
languages are, to a greater or less degree, dialectically differenti- “proportional regularity”. For example, the proportional regular-
ated. There is no reason to believe that Proto-Indo-European, ity that holds between 6 and 3, on the one hand, and 4 and 2, on the
Proto-Germanic, Proto-Slavonic and the other proto-languages other, is an analogy, in the intended sense of this term: it is a
that we hypothesize as the source of families and subfamilies of relation of four quantities (6, 3, 4, 2) such that the first divided by
attested language were dialectically undifferentiated. Wherever the second is equal to the third divided by the fourth (6:3 = 4:2).
possible, the comparative method will reconstruct a single proto- Analogical reasoning was used widely by Plato and Aristotle, and
202 Language-change 6.4 Analogy and borrowing 203

by their followers, not only in mathematics, but also in the develoj synchronically productive rule. Evidence for the synchronic pro-
ment of other branches of science and philosophy, including gram- ductivity of the rule in question comes partly from language-
mar. Unless this fact is appreciated, it is impossible to understand acquisition by children and partly from the adult speaker’s ability to
one of the basic principles of traditional grammar: that of (‘ construct the past-tense form of new verbs that he first meets in
paradigm. Given, let us say, the paradigm jump, jumps, 1ump their present-tense (or present-participle) form. As far as language-
and jumped (i.e. the forms of the model English verb ‘jump’; the acquisition is concerned, the fact that the child has mastered the
term ‘paradigm’ comes from the Greek word meaning “model”or rule for the formation of regular past-tense forms by suffixation is
“example”), we can construct such proportional equations as the proved by his production, not only of a large number of correct
following: jump : jumps = help : x; jump : jumped = help : y; etc. forms (e.g. jumped, walked, loved), but also of the occasional
And we can solve these equations, assigning to the unknown quan- incorrect form such as rided or goed. In fact, paradoxical though it
tities (x, y, etc.) their appropriate values (helps, helped, etc.). may appear at first sight, the production of such incorrect forms by
This, then, is what is meant by ‘analogy’ in traditional grammar, analogy with some typical member of the regular class of weak
and more particularly in the controversy between the analogists and verbs (jump : jumped = ride:x; therefore x = rided) constitutes a
the anomalists, which arose in the second century B.c. and was to more convincing proof that the child is applying a rule than does his
endure, in one form or another, until modern times, exerting a production of any number of correct past-tense forms, all of which
profound influence upon the development of linguistic theory. might, in principle, have been remembered and imitated as un-
Roughly speaking, we can say that the analogists took the view that analysed wholes (cf. 8.4).
the relation between form and meaning was governed by the prin- There are a few instances of weak verbs having been made into
ciple of proportional regularity, and that the anomalists adopted strong verbs by the force of analogy in the history of English. For
the contrary view. We have no need to go into the details of this at example, in some American dialects the past-tense form of ‘dive’ is
times confused, and confusing, controversy, It is important to real- dove, rather than dived; and, contrary to what one might suppose, it
ize, however, that it is all part of the background against which the is dove that is the result of innovation. In the vast majority of cases,
Neogrammarians established their own notion of analogy and its however, analogy has operated in the reverse direction, increasing
role in the historical development of languages. the weak verbs at the expense of the strong: e.g. Middle English
Let us take an example. English, like German, draws a distinc- holp (cf. gof) was replaced with Modern English helped. It should
tion between what are conveniently referred to as weak and strong be noted that dived — dove is just as much a result of analogical
verbs. The former, which constitute the majority of all the verbs in pressure as holp — helped is. The language exhibits two patterns of
the language, form their past tense by adding a suffix to the present- formation, either of which might serve as the paradigm for ana-
tense stem (cf. E. jump-s, jump-ed; G. lieb-t, lieb-te); the latter logical extension.
exhibit a difference, of one kind or another, in the vowels of 1t is worth observing at this point that the fact that many obviously
corresponding present-tense and past-tense stems, and usually lack related verbs of English and German show the same phenomenon
the past-tense suffix characteristic of the weak verbs (cf. E. ride-s, of vowel-alternation is a particularly striking piece of evidence
rode; sing-s, sang; G. reit-et, ritt; sing-t, sang). The strong verbs fall in favour of the hypothesis that these two languages are, in fact,
into several subclasses according to the nature of the vowel alterna- genetically related: cf. E. begin-s, began, begun : G. beginn-t,
tion which distinguishes their present-tense and past-tense forms; begann, begonn-en; E. bring-s, brought : G. bring-t, brach-te, ge-
and they are commonly regarded as irregular. They are certainly brach-t; E. find-s, found:G. find-et, fand, ge-fund-en; E. give,
less regular than the weak verbs, which have been on the increase gave, giv-en:G. gib-t, gab, ge-geb-en. (I have added the past-
for many centuries and which conform to what has long been the participle form, for both German and English, when it differs from
 6.4 Analogy and borrowing 20§
204 Language-change

. the past-tense form, as it almost always does in German.) Anal were”", German, on the other hand, has remodelled the singular
has operated independently in both English and German, for stem by analogy with that of the plural: ich war: wir waren. In this
eral hundred years, to reduce the incidence of vowel-alternatios instance the historically regular final [s] of the singular has been
that, for example, whereas ‘help’ is weak in Modern English, replaced with the historically irregular [r]. Interestingly enough, in
related verb ‘hilfen’ in German is strong (hilf-t, half, ge-holf- n) very early Latin intervocalic [s] also became [r]. Hence the contrast
Sound-changes that have taken place independently in the sevel in Classical Latin between the nominative singular form honos
Germanic languages have also had their effect, increasing the nus “honour” and the other forms of the same noun: honorem, honoris,
ber of distinct sets of vowel-alternations and making the corres etc. (from *honosem, *honosis, etc.). And in later Latin honos was
dence between the forms of individual verbs less systematic, i replaced by honor, so that honor- was generalized as the stem
detail, than it was in earlier periods. But there are still dozens of throughout the whole set of inflectional forms. It is also worth
verbs which exhibit a similar vowel-alternation. The same is true adding that analogy is responsible for the fact that the verb ‘to be’ is
Dutch, which, as we saw earlier (cf. 6.2), is more closely related the only verb in Modern Standard English in which there is a
German than itis to English: ¢f. begin-t, begon, begonn-en; breng difference between a singular stem and a plural stem in the past
brach-t, ge-brach-t (“bring"); vind-t, vond, ge-vond-en (“find"); tense. In Middle English, many of the strong verbs showed a similar
ete. Even the North Germanic languages have strong verbs wbos:' difference. Analogy has, once again, generalized either one stem or
past-tense and past-participle forms can be related to the present- the other (or, in some instances, the past-participle form); and this
tense forms in terms of more or less regular vowel-alternations: cf, accounts for the very considerable fluctuation that there is across
Swedish skriver, skrev, skrivit (“write”); kryper, krop, krupit English dialects and even in the spontaneous usage of individual
(“creep”). Indeed, this pattern of vowel-alternation goes back speakers.
ultimately to the Proto-Indo-European period: cf. Greek peith-6, The second point to be made about analogy is that it is a much
pe-poith-a, e-pith-on (“persuade”); leip-6, le-loip-a, e-lip-on more potent factor in language-change than the Neogrammarians
(“leave™); etc. As was mentioned at the end of the preceding held it to be. In fact, the Neogrammarians were inclined to invoke
section, it was this kind of correspondence ~ what Sir William Jones the influence of analogy only when it enabled them to explain away
referred to as “‘a stronger affinity both in the roots of verbs and in apparent exceptions to one of their postulated sound-laws. Further-
the.forms of grammar than could possibly have been produced by more, some of them drew a distinction between sound-change as a
accident” (cf. 6.2) - that so much impressed the founding fathers of physiologically explicable process and analogy as something that
comparative philology. But we are here concerned with analogy; resulted from the sporadic and unpredictable intervention of the
and in this connection there are two points to be made. human mind. For those who took this view, the sound-laws were
The first is the one that was given particular emphasis by the seen as being comparable with the so-called laws of nature. Itis now
Neogrammarians: that analogy often inhibits (or subsequently re- more clearly realized, first of all, that no such sharp distinction can
verses the effect of) otherwise regular sound-changes. For ex- be drawn, as far as language is concerned, between the physical and
ample, after the operation of Verner’s Law (cf. 6.3), but prior to the psychological; and secondly that analogy ~ provided that the
our earlier texts, intervocalic [s] became [r] in Germanic. It is this traditional term is interpreted according to the spirit, rather than
sound-change which accounts for the letter <> - still pronounced as the letter, of the tradition — operates on both the phonological and
[r]in some dialects - in the plural of the past-tense of the verb ‘to be’ the grammatical levels of language-structure. What was traditionally
in English, in contrast with what the orthography shows was once an described in terms of proportional regularity can be subsumed
[s] in the singular: were : was. Dutch shows the same contrast (but under the more general principle of regularization on the basis of
without a difference in the vowels): ik was “I was” : wij waren “‘we existing patterns of correspondence between form and meaning.
206 Language-change 6.5 The causes of language-change 207

Indeed, it would not be unreasonable to identify both the Saussi say that, in the course of its history, it has borrowed enormously,
rean notion of structure and the generativist's notion of . not only in its vocabulary, but also in grammar and phonology, from
governed creativity with an appropriately modernized version other languages and dialects.
the traditional concept of analogy. But this is a bigger, and mi But does it make sense to talk as if a sharp distinction can always
controversial, question (cf. 7.4). be drawn between native and non-native forms? It has long been
Another phenomenon to which the Neogrammarians appealed evident that the conventional family-tree-diagrams of language-
order to explain some of the apparent exceptions to the sound-l; development and language-relatedness can be seriously misleading
was borrowing. For example, in addition to the word ‘chef’, which if they are taken to be realistic models of historical processes. More
we identified above as the French descendant of the Latin ‘caput’, recent work in dialectology and sociolinguistics has made clear the
whose Proto-Romance citation-form may in fact have been importance of synchronic dialectal and stylistic variation within a
*capu(m) (see Table 4), we also find, in Modern French, the word language-community as a causal factor in language-change. In
‘cap’ (cf. ‘de pied en cap’, literally “from foot to head™). The form conditions of synchronic variation — and more especially of
cap quite clearly violates all three of the sound-laws (apart from the bilingualism and diglossia (cf. 9.4) ~ the traditional concept of
loss of the final vowel) required to derive chef from *capu. The borrowing is perhaps inapplicable.
explanation is that it was borrowed into French (at a fairly early However that may be, it is certainly the case that the Neo-
date) from Provengal, to which the sound-laws in question did not grammarians drew too sharp a distinction between what could be
apply. Similarly, there are many words in English that begin with handled in terms of sound-laws and what was to be explained by
sk- in their written form (cf. sky, skill, skirt, etc.) which are appar- means of analogy and borrowing. Nevertheless, most general
ent exceptions to the sound-law that changed Old English [sk] to accounts of the historical development of languages still follow the
Modern English [I] (cf. shirt, ship, shed, etc.). Such words were Neogrammarians in this respect.
borrowed from one or other of the Scandinavian dialects, which
6.5 The causes of language-change
were brought to England at the time of the Viking invasions and had
a considerable influence on the speech of the Danelaw region. (To Why do languages change in the course of time? There is no
this day, much of the vocabulary of the local dialects of Northern generally accepted answer to this question. Several theories of
England and Southern Scotland is of identifiably Scandinavian language-change have been put forward. But none of them covers
origin; but we are here concerned with what may be regarded as all the facts. The most that can be done here is to mention, and to
borrowings into Standard English.) Pairs of co-existent cognate comment briefly upon, some of the main factors that scholars have
native and borrowed words, like English ‘skirt’ and ‘shirt’ or French referred to in the explanation of language-change.
‘cap’ and ‘chef’, are often called doublets. Lexical doublets, it may It is customary, in discussions of this question, to operate with
be noted, are very rarely even descriptively synonymous (cf. ‘skirt” ; two separate distinctions: (a) between sound-change, on the one
‘shirt’, ‘skipper’ : ‘shipper’, etc.). hand, and grammatical and lexical changes, on the other; (b) be-
The same point can be made about borrowing as was made about tween internal and external factors. But neither of these two
analogy: that it is a far more important factor in language-change distinctions should be pressed too hard. As we have seen, the Neo-
than the Neogrammarians (and many of their successors) have grammarians’ view that sound-change is radically different from
taken it to be. In particular, like analogy, it should not be seen as other kinds of language-change is, at best, no more than a half-
merely providing an explanation for apparent exceptions to the truth. Even such more or less physiologically explicable processes
sound-laws. If English is considered to be a purely West Germanic as assimilation (which results in successive sounds being made
language — and it is conventionally so regarded (cf. 6.2) - we have to identical, or more similar, to one another in terms of place or
208 Language-change 6.5 The causes of language-change 209

manner of articulation: cf. Italian, otto, notte, etc., in Table 4 of described, macroscopically and in retrospect, as a regular sound-
section 6.3), or haplology (the loss of one of two phonetically simils change.
syllables in sequence: e.g. Old English Engla-land “‘country of the It is not being suggested, of course, that all sound-change can be
Angles” > England) require the support of other more general explained in this way. We must still allow for the possibility of
factors, if they are to produce permanent changes in the sound- gradual and imperceptible phonetic drift over time throughout all
system of a language. As for the distinction between external and the words in which a particular sound occurs. The point being made
internal factors, which depends upon the abstraction of the is simply that a variety of causal factors may interact to produce the
language-system, as such, from the cultural and social matrix in same kind of end-result: something that is usually regarded as
which it operates, this too breaks down, in the last resort: the regular sound-change and, in the Neogrammarian tradition at least,
communicative function of language, which interrelates form and contrasted with such allegedly sporadic phenomena as analogy and
meaning within a language-system, also relates that language- borrowing.
system to the culture and society whose needs it serves. Scholars who emphasize the distinction between internal and
Two of the most general factors of language-change were men- external factors — and more especially those subscribing to the
tioned in the preceding section: analogy and borrowing. It may now tenets of structuralism and functionalism (cf. 7.2, 7.3) - tend to
be emphasized that much of what the Neogrammarians accounted attribute as much language-change as they can to what are classified
for in terms of sound-laws can be brought within the scope of the as internal factors: especially to the continual readjustments that
joint action of these other two factors. The sound-laws themselves are made by a language-system as it moves from one state of
have no explanatory value: they are no more than summaries of equilibrium (or near-equilibrium) to another. One of the most
what happened in a particular area (more precisely, in a particular influential proponents of this point of view has been the French
language-community) between two points in time. Considered in scholar, André Martinet, who has tried to account for language-
retrospect, and macroscopically, the change that has taken place change, and more particularly sound-change, with reference to his
may appear to be regular enough (in the sense in which the principle conception of languages as self-regulating semiotic systems,
of regularity was understood by the Neogrammarians and their governed by the complementary principles of least effort and
followers). However, the investigation of sound-changes that are communicative clarity. The former principle (under which one can
taking place at the present time has shown that they can originate in subsume such physiologically explicable phenomena as assimilation
one or more borrowed words and can then spread by analogy into and haplology, referred to above, and also the tendency to shorten
others over a period of time. forms of high predictability) will have the effect of reducing the
One of the symptoms of this process of language-change is what is number of phonological distinctions and of maximizing the work
commonly called hypercorrection. An example of this is the analo- that each of them does. It will be held in check, however, by the
gical extension of the Southern English vowel of butter into words necessity of maintaining a sufficient number of distinctions for the
like butcher by speakers from the north of England who have purpose of keeping apart utterances that might otherwise be con-
acquired (i.e. borrowed) the RP pronunciation of the former class fused in the acoustic conditions under which spoken languages are
of words. Phonetic hypercorrection of this kind does not differ, as normally used. This notion has a good deal of intuitive appeal and a
far as its causation is concerned, from the hypercorrection which certain number of sound-changes have been explained in terms of
has resulted in middle-class, and often educated, speakers of it. So far, however, it has not been convincingly shown to have all
Standard Southern English saying berween you and I. Tt will be the explanatory power its proponents claim for it.
readily appreciated that the former, though not the latter, kind The main contribution that the structuralists and functionalists
of hypercorrection could eventually lead to what might well be have made to historical linguistics comes from their insistence that
 6.5 The causes of language-change 211
210 Language-change

each postulated change in a language-system must be evaluated: theory and methodology of historical linguistics by generativists can
terms of its implications for the system as a whole. For example be seen as a refinement and development of the structuralists’
they have made it clear that the several parts of Grimm’s Law (or conception of language-change. Preference is given in both cases to
the Great Vowel Shift, which took place in the transition fr what are classified as internal factors. The stgucturalists’ notion of
Middle English to Early Modern English) must be considere self-regulation has been replaced with that ofsthe restructuring of
together. And they have raised interesting questions about the rules of the language-system and a tendency towards simplifica-
kinds of chain-reactions that seem to take place at certain periods tion, It is difficult to see any fundamental differences between these
the historical development of languages. To take Grimm’s Law two notions.
our example: did the PIE voiced aspirates, [*b¥, *d®, *gt], in I However, one difference between the Chomskyan competence/
their aspiration, cause the PIE voiced unaspirated stops, [*b, *d, performance distinction and the Saussurean distinction of langue/
a
*g], to lose their voice and this in turn cause the PIE voiceless stops parole is that the former lends itself more readily than the latter to
[*p, *t, *k], to become fricatives? Or was it rather that the PIE psychological interpretation. Generativists, as we shall see, have
voiceless stops initiated the process, pulling the others behind been much concerned, for various reasons, with the problem of
fact
them, as it were, into the place that they were leaving vacant? These language-acquisition by children. They have emphasized the
languag e, is not
q'nestions may not be answerable. But they do at least give recogni= that the child, as he begins to acquire his native
taught the rules of the underlying system, but must infer these from
tion to the fact that the several changes summarized in Grimm’s which
Law may be causally connected. the patterns of correspondence between form and meaning
‘What is now referred to as internal reconstruction (in contrast he detects in the utterances that he hears around him. What is
y to
with reconstruction by means of the comparative method) may also traditionally regarded as false analogy (e.g. the child’s tendenc
vist as part of the
be set to the credit of structuralism. This is based on the conviction say goed rather than went) is seen by the generati
that synchronically observable partial regularities and asymmetries more general process of the acquisition of rules.
can be explained with reference to what were fully regular, produc- Generativists are not the first to have sought an explanation of
tive processes at an earlier period. For example, even if we had no language-change in the transmission of language from one genera-
comparative evidence to go on and no records of earlier stages in tion to another. But they have looked more carefully than others at
t}?e development of English, we might infer that the partial regular- the process of language-acquisition in terms of the nature of the
ities evident in the English strong verbs (cf. drive: drove : driven, rules that are required at identifiable stages in this process. Further-
ride : rode : ridden; sing :sang : sung, ring : rang:rung; etc.) were more, they have begun to investigate in detail syntactic, as well as
the relics, as it were, of an earlier more fully regular system of phonological and morphological, change: syntactic change was
verb-inflection. Internal reconstruction is now a recognized part of hardly dealt with at all, other than occasionally and unsystemati-
that
the methodology of historical linguistics and it has proved its worth cally, until recently. Most important of all, however, is the fact
on several occasions. generativism has provided historical linguistics with a more precise
conception of formal and substantive universals, in relation to
A§ we shall see later, generativism develops out of, and in part
continues, a particular version of structuralism. It is characteristic which postulated changes in prehistoric and unattested stages of a
of generativism that it should see language-change in terms of the language can be evaluated as more or less probable.
a_ddition, loss or reordering of the rules that determine a speaker’s On the debit side, both structuralism and generativism have been
l|'ng'uislic competence. In so far as the competence/performance inclined to pay insufficient attention to the importance of synchro-
distinction can be identified with the langue/parole distinction of nic variation as a factor in language-change. Apart from anything
Saussurean structuralism (cf. 7.2), the contribution made to the else, this has given rise to such pseudo-questions as the following: Is
212 Language-change Questions and exercises 213

sound-change sudden or gradual? Does language-change origi with another. Indeed, it is now coming to be accepted that bilingual-
in competence or performance? As far as the first of these ism and diglossia — and even pidginization and creolization — may
questions is concerned, it is now over a hundred years since Johar have played a much more extensive role in the formation of the
nes Schmidt challenged the family-tree concept of langu: language-families of the world than was once thought to be the case
relatedness favoured by the Neogrammarians, and pointed out (cf. 9.3, 9.4).
innovations of all kinds, and more particularly sound-changes, ca We began this section with the question: Why do languages
spread out from a centre of influence, like waves on a lake, losing change in the course of time? We may conclude it by repeating what
force as they reach points further and further from the centre. In was said in an earlier chapter (cf. 2.5): the ubiquity and con-
decades that followed, it was demonstrated, especially by scholars tinuity of language-change is far less puzzling once it is realized that
working on the history of the Romance languages, where there was. no natural language is ever stable or uniform and that much of
abundant evidence, both synchronic and diachronic, that what has what is describable, macroscopically, as language-change is the
come to be called the wave theory of language-change provided a. product of socially conditioned synchronic variation. This is not to
more satisfactory account of the facts, in many cases at least, than say that all language-change is to be accounted for in this way,
did the more orthodox family-tree theory, with its inbuilt assump- but only that social factors are undoubtedly far more important
tions of sudden and thereafter continuous divergence between re- than they were once thought to be.
lated dialects. It was also shown, by dialectologists, that, far from
applying simultaneously to all words in which they were applicable, FURTHER READING
sound-changes might originate in just one or two words and then Most general textbooks and introductions to linguistics have chapters on
spread to other words and, along the lines of communication, to language-change. In particular, Bloomfield (1935), chapters 18-27, is still
other areas. If this is generally the case, it is clear that the question well worth reading for an essentially Neogrammarian view, with many now
classic examples from English and other languages.
whether sound-change is gradual or sudden loses much of its point.
Since it also turns out to be the case that individuals may fluctuate in Recent introductions to historical linguistics as such include Aitchison
(1981); Bynon (1977): Lehmann (1973). Bynon (1977: 281-2) gives addi-
their usage, between an older and a newer form, so too does the
tional references, by topic, for everything dealt with in this chapter; Aitchi-
question whether language-changes originate in competence or son (1981) emphasizes the role of social factors in language-change.
performance. On the history of English (at various levels of detail and technicality):
More recently, sociolinguists have shown that what is true of the Barber (1972); Baugh (1965); Francis (1967); Lass (1969); Potter (1950);
geographical diffusion of phonological, grammatical or lexical Strang (1970); Traugott (1972). For other languages and language-families,
variants also holds true of their diffusion through the socially dis- the Encyclopaedia Britannica, 15th edn (1974) is the most convenicnt work
tinguishable classes of a given community. In general, it has become of reference.
clear that social factors (of the kind that we shall consider in QUESTIONS AND EXERCISES
Chapter g) are far more important in language-change than was
. What grounds are there for recognizing three different periods in the
previously realized. After all, it is not only geographical, or even history of English: Old English (Anglo-Saxon), Middle English and
political, boundaries that set limits to the degree of intercommu- Modern English?
nication among people living in the same area. Social dialects may 2. On the basis of information to be found in encyclopaedias or other
be as distinct from one another as geographically based dialects. On works of reference, list the principal extant members of the Germanic,
the other hand, given the right social conditions (breakdown in a Romance and Slavonic families of languages.
traditionally stratified society, the imitation of upper-class forms or . What is a proto-language?
expressions, etc.), one social dialect may be modified by contact Explain the purpose of reconstruction in historical linguistics.
P
 Questions and exercises 215
214 Language-change

5. English, like French, Russian, Hindi, etc., is said to be an 17. Compare and contrast the family-tree theory and the wave theory
European language. What does this mean and is it true? How al (Wellentheorie) of language-development.
Finnish, Hungarian, Basque, Turkish, Tamil? 18, Evaluate the contribution that structuralism and generativism have
‘Why do you think Sir William Jones attached so much importance made to the theory and methodology of historical linguistics.
what he called “the roots of verbs” and the “forms of grammar™ 19. What contribution can the study of (a) language-acquisition and (b)
(cf. p. 188)? pidgins and creoles make to historical linguistics? (This question is best
. “One of the strongest reasons for adopting the assumption of regullf tackled after reading Chapters 8 and 9.)
=

phonetic change is the fact that the constitution of the residues . . .

throws a good deal of light upon the origin of new forms" (Bloomfield,
1935: 405). Discuss.
. Give an account of Grimm’s Law (using examples other than those in
the text) and show how Verner’s Law relates to it.
. Explain what is meant by the analogical regularization of synchronically
©

irregular forms.
10. What inferences can be drawn about the history of a language from the
existence of lexical doublets? Make a list of ten such pairs of lexemes in
Modern English other than those used as examples in the text. What
distinction, if any, would you draw between lexical doublets and co-
existent grammatically equivalent forms of the same lexeme (got:
gotten, learnt : learned, dived : dove)? How would you classify brothers
: brethren, mediums : media, struck : stricken in relation to such a
distinction?
. Comment upon the following more or less fossilized forms and con-
structions: Rest in peace!, Bless you!, If it please your Majesty . . . (cf.
also If your Majesty pleases . . .), methinks (vs. I think). What light do
they throw on earlier stages of English?
12, Such fixed expressions in Modern English as ‘meat and drink’
and ‘neither flesh nor fowl (nor good red herring)' preserve older
meanings of some or all the constituents. Can you think of similar
examples?
. Show how borrowing can account for apparent exceptions to the regu-
lar operation of a sound-law.
14. “Language change, then, offers important evidence about the nature
of human language — namely that it is rule-governed” (Akmajian,
Demers & Harnish, 1979: 226). Discuss this statement in relation to the
generativists' notion of restructuring.
15. “What is perhaps the most important contribution towards an under-
standing of the actual mechanism of language change has come from
the detailed sociolinguistic investigation of living speech communities
(Bynon, 1977: 198). Discuss.
16. Explain and exemplify the notion of internal reconstruction.