University of Virginia Library

Search this document 
Dictionary of the History of Ideas

Studies of Selected Pivotal Ideas

expand sectionV. 
expand sectionV. 
expand sectionV. 
expand sectionV. 
expand sectionV. 
expand sectionV. 
expand sectionV. 
collapse sectionVII. 
expand sectionVII. 
expand sectionIII. 
expand sectionIII. 
expand sectionI. 
expand sectionII. 
expand sectionV. 
expand sectionV. 
expand sectionVI. 
expand sectionII. 
expand sectionV. 
expand sectionV. 
expand sectionVII. 
expand sectionVII. 
expand sectionI. 
expand sectionVI. 
expand sectionVI. 
expand sectionVI. 
expand sectionIII. 
expand sectionIII. 
expand sectionVI. 
expand sectionIII. 
expand sectionIII. 
expand sectionIII. 
expand sectionIII. 
expand sectionIII. 
expand sectionIII. 
expand sectionIII. 
expand sectionIII. 
expand sectionIII. 
expand sectionIII. 
expand sectionIII. 
expand sectionIII. 
expand sectionV. 
expand sectionV. 
expand sectionIII. 
expand sectionI. 
expand sectionVI. 
expand sectionIII. 
expand sectionVI. 
expand sectionI. 
expand sectionIII. 
expand sectionVII. 
expand sectionI. 
expand sectionI. 
expand sectionIV. 
expand sectionVI. 
expand sectionV. 
expand sectionVI. 
expand sectionVI. 
expand sectionIV. 
expand sectionIII. 
expand sectionV. 
expand sectionVI. 
expand sectionIII. 
expand sectionVI. 
expand sectionVI. 
expand sectionVI. 
expand sectionIII. 
expand sectionVI. 
expand sectionVI. 
expand sectionVI. 
expand sectionVI. 
expand sectionII. 
expand sectionII. 
expand sectionII. 
expand sectionVII. 
expand sectionIV. 
expand sectionIV. 
expand sectionV. 
expand sectionVI. 
expand sectionVI. 
expand sectionV. 


For the purposes of this article, and to avoid overlap-
ping with the article on Language, the history of
linguistics will here be viewed as concerned with those
studies and activities in which one's language is con-
sidered as a language rather than as language pure and
simple. This accommodates a good many quite dispar-
ate attitudes; to use modern terms, these might be said
to range from the most extreme structural relativism
under which it is implied that there are no substantive
limits to language diversity to the position taken by
some generative grammarians in whose eyes languages
differ only in surface structure but share the funda-
mental oneness of language.

Thus the idea of distinguishing the study of language
from the study of languages is in itself a possible
touchstone for what is important and for what has
changed in intellectual emphasis. The nineteenth cen-
tury, for instance, would have thought in terms of a
division of labor—frequently a hostile division at that
—between philosophy and linguistics. At other times,
both before and after, this particular borderline has
existed in a far less sharp form.

Multilingualism has been a pervasive fact of life, far
more so than participants in modern technological
civilization and members of large and dominant speech
communities are inclined to think. Besides, far more
than other facets of language (for instance, the internal
properties of one's own principal language, dialect, or
style), language diversification emerges above the
threshold of awareness. Hence mythical and theological
ways of accounting for language diversification are
widespread. They center around such questions as how
languages were created or invented, how “things” got
their names and, indeed, why languages are many.

Among the oldest extant examples of this must be
a Sumerian text (Kramer, 1968) which bespeaks a leg-
endary past in which “the whole universe, the people
in unison(?) To Enlil in one tongue spoke... Then
Enki... changed the speech in their mouths, [brought
(?)] contention into it, into the speech of man that (until
then) had been one.” The idea of a monolingual golden
age crops up again and again. Its biblical version is,
of course, the story of the great dispersal after the
abandonment of the tower of Babel. The familiarity
and authority of this story remained one of the two
controlling influences on linguistic thought in the
Western world.

The other influence was one which allied itself natu-
rally with the first, although its basis lies in a very
different intellectual region. This was Greco-Roman
secular thought. Greek historians frequently use what
they call “resemblance” (however recognized) in
speech as an argument for the descent of one popula-
tion from another. The realities of Greek life supported
this, of course: the ties which linked colonies to their
mother cities and the affinity that existed between the
dialects of the cities so linked were too obvious to
escape notice. It is worth remembering that the stand-
ard (and only half-appropriate) classification of Greeks
into Ionians, Aeolians, and Dorians had its chief appli-
cation among the Greek settlers in Asia Minor and was
at least in part a linguistic one.

When extended beyond the Greek world the argu-
ment became essentially speculative and impres-
sionistic. Considering the deep philosophical interest
of the Greeks in the nature of language, in “etymol-
ogy,” and in logic; and considering further the peripa-
tetic penchant for discovery and taxonomy which sent
along naturalists of all kinds on Alexander's campaign,
it remains remarkable that languages in their diversi-
fication went without systematic attention, one may
say, throughout antiquity. Only glosses, i.e., isolated
foreign words, were collected (to serve us now, in-
cidentally, as scant and uncertain sources of informa-
tion on otherwise lost languages). And this was so
despite constant contact with a variety of populations
and the widespread bilingualism and multilingualism
of the Hellenistic-Roman world; despite the demand
for (and supply of) translations from literary foreign
languages like Etruscan and Punic; and also despite
the never-ceasing literary and rhetorical confrontation
between Greek and Latin with the attendant stereo-
typed value-judgments (about the “poverty” of the
Latin language; Lucretius, De rerum natura, I, 139).
Foreigners appear on the Greek and Roman comedy
stage speaking some amusing gibberish (Persian, Punic,
even, it seems, Kannada of South India). But the only
evidence of philosophical concern seems to be Plato's
casual and irrelevant remark, in Cratylus (409-10), that
Greek words (“names”) may have been borrowed from
barbarian languages (e.g., from Phrygian). If it is really


true that it was the Phoenician background of the Stoic
philosopher Zeno of Kition (ca. 300 B.C.) which
prompted him to introduce certain novel categories
into Greek grammar (Pohlenz, 1926), nothing could
be more characteristic than that the circumstance was
not made explicit. Both Aristotle and the Hellenistic
poet and scholar Callimachus wrote on “Non-Greek
Customs” (Pfeiffer [1968], p. 200), and both appear
to have left language out of the picture. It is also
remarkable that there survive bilingual composition
books (e.g., the so-called Hermeneumata Pseudo-Dosi-
) but no learners' grammars of either imperial
language written specifically in, and for the use of
speakers of, the other. Translations of grammatical
works were made, to be sure. When the Greek-writ-
ten Greek grammar of Dionysius “the Thracian” (first
century B.C.) was translated into certain languages (for
instance, Armenian) the aim was, however, not to
teach Greek but to create a grammar (however awk-
ward) of the target language.

When the argument leading from resemblance to
descent was applied to “barbarian” languages it inter-
sected, in ways which are not clearly understood, with
at least one other principle, namely that of explaining
language differences (or differences between types of
languages, as we would say) through climate. This
begins at least with the Ionian physicians who, to be
sure, may have had no more than the timbre of voice
in mind when they spoke of a division into “clear-
voiced” and “heavy-voiced” people according to the
zones which they inhabit. But as the notion recurs in
the Stoic thinker Posidonius (ca. 135 B.C.-ca. 55 B.C.)
it becomes part of a general and influential system of
ethnography. In the end it enters into that well-known
body of semi-learned linguistic folklore from which
metaphors and ad hoc explanations are available. Not
unfittingly, climate figures in nineteenth-century
scholarship as a potential “cause of change” (see the
amazingly respectful, if critical, discussion in Prokosch
[1939], pp. 55-56).

With regard to the inference from language resem-
blance to prehistoric descent, posterity has tended to
be equally uncritical. It is not reasonable, for instance,
to expect the judgment of the rhetorician and histori-
ographer Dionysius of Halicarnassus (beginning of the
Christian era) about the Etruscans being “unlike any
other nation in language” to mean what we mean when
we say that Etruscan (or Basque, or Burushaski) is
genealogically isolated. To place such ancient pro-
nouncements exactly is both a more specific and a more
comprehensive task. Its solution depends on our ability
to reconstruct the preconceptions as well as the prac-
tices inherent in entire lines of lost scholarship.

If languages at large were not much touched by
empirical study or even only by speculation, the same
is not quite true of the dialects of Greek including,
in the ancient view, Latin (on which Philoxenus wrote
at the beginning of the first century B.C.; Pfeiffer
[1968], p. 274). The stimulus consisted in the fact that
local dialects, in stylized form, had literary uses (the
Aeolic of Alcaeus and Sappho, the Doric of choral lyric
poetry, etc.). But this was not all: the great Alexandrian
librarian Aristophanes of Byzantium, as others before
and after him, studied the spoken vernacular in detail.
And it was in connection with domestic observation
of this kind that at least occasional glimpses occur in
the direction of a chronological interpretation of dia-
lect differences. Already Plato's Socrates (in the
Cratylus) recognizes change as a normal property of
language, although it is true that he does not emphasize
the point. A more widespread view (and one which
had great persistence in later history) is represented
in Herodotus' famous story (hardly a specimen of
Egyptian linguistic thinking) of King Psammetichus'
experiment in wishing to discover the “oldest” lan-
guage: when two children reared in isolation cried
bekos” and it turned out that the Phrygian word
for bread was bekos the answer was simple. Clearly,
this inference is incompatible with the idea that all
languages change constantly and that all are equally

For the Middle Ages our knowledge is still limited.
In particular, it is difficult to assess the extent to which
the enormous intellectual effort spent on points of
grammatical and semantic theory was matched by
efforts to collect and explain data having to do with
variety. In the West, the theology of dispersal to which
we have alluded, and the corresponding popular as-
sumption that the pre-Babel lingua Adamica was
Hebrew, held dominant sway (see A. Borst's magnifi-
cent work on the subject, Der Turmbau von Babel).
Islamic thinking, it seems, was very greatly concerned
with the primacy of Arabic over other languages. The
two Western European figures standing out from this
general background are, not too surprisingly, two of
the greatest universal thinkers of the high Middle Ages,
namely Roger Bacon and Dante. To Roger Bacon be-
longs the both very medieval (essentially Aristotelian)
and very modern-sounding formulation that grammar
“is the same in all languages in its substance, and that
surface differences between them are merely accidental
variations” (Robins [1967], pp. 76-77). What distin-
guishes Roger Bacon from countless others making
similar remarks is that he, as a writer of a Greek
grammar and as a student of Arabic and Hebrew
looked upon this as a concrete problem, to which there
was an empirical side: there is clear evidence that he
saw the analogy between interlanguage differences of


the grosser kind, on one hand, and the much subtler
relationship between one standard and the various
dialectical forms of a language, on the other, and that
he attached central importance to it. In this he was
akin to Dante. The views set forth in Dante's treatise
De vulgari eloquentia (shortly after 1300) must be seen
against the backdrop provided by the influential
modistae and by the equally popular Doctrinale of
Alexander of Villedieu (1199; a versification of tradi-
tional Latin grammar and syntax) to have their origi-
nality appreciated. Dante faces the contradictions
which arise when observation of living language (in
his case local types of Italian which he classifies with
great acumen) and formulated Latin grammar clash.
His interpretation, to be sure, is unhistorical: to him
authoritarian, “dead” Latin grammar was created, and
always was a “secondary” and artificial language such
as some but not all people possess. But all people, since
the Dispersal, speak their vernacular; each man re-
ceives his “without any rules,” from his nurse. The
vernaculars are more “noble” than (Latin) grammatica.
Their variability and instability is a consequence of the
Dispersal. Before it, all men spoke divine and un-
changing Hebrew (a position from which Dante himself
deviated later in Paradiso XXVI). What is so re-
markable (and what no doubt accounts for the insig-
nificant effect which the work had in its time) is not
only the preference given to the vernaculars “without
rules” but even more the concrete realization that
change is slow but all-pervasive, and that variety is
a function of it; its extent is proportionate to distance
in time and space. (“Here we are investigating some-
thing in which we are not supported by any author-
ity.”) This gives Dante an opportunity to set forth his
grandiose genealogical classification of the languages
of the known world. Here we also find the artifice,
so common later on, of using certain key-words (such
as the word for “yes”) as convenient shibboleths.

Symbolically speaking at least, it was the fall of
Constantinople in 1453, with the ensuing migration of
Byzantine men of letters to western Europe which
opened up a period of feverish activity in the new
medium of print: Lascaris' Greek grammar appeared
in 1476, and thereafter dictionaries and grammars were
produced ceaselessly. Represented are not only literary
languages with their own internal grammatical tradi-
tion like Arabic and Hebrew, but vernaculars of every
kind: Dutch (1475), Breton (1499), Welsh (1511), Polish
(1564), Basque (1587). And shortly after the middle of
the sixteenth century the first “missionary” grammars
of Central and South American Indian languages make
their appearance. Superficially, these works follow a
traditional Latin model. But the assumptions and pro-
cedures involved in fitting newly observed material into
those conventional categories still need to be analyzed
and understood.

Special mention must be made of a particular genre
of scholarly production which begins as early as 1427,
with Schildberger's Pater Noster in Armenian and Tar-
tar: collection after collection, for an ever-widening
circle of languages, of the Lord's Prayer, with appro-
priate commentary, geographic information, attempts
at classification, and so on. It is perhaps sufficient to
mention two of the most famous works in this chain
which bear the same title: Conrad Gesner's Mithridates
of 1555 (with the Lord's Prayer in twenty-two lan-
guages) and J.-C. Adelung's Mithridates in four vol-
umes, of 1806-17), with the Lord's Prayer “in almost
five hundred languages and dialects” (W. von Hum-
boldt took part in this effort). In a sense, the British
and Foreign Bible Society's Gospel in many Tongues
(e.g., 1950), is a present-day sample. These works may
be only compilations, but they occupy a pivotal posi-
tion in the history of linguistic ideas. They are the main
link between observation and speculation (since specu-
lative writers had a way of relying on them in large
part and often exclusively), and at the same time they
reflect, in their commentary, contemporary inter-
pretation with considerable faithfulness.

In the course of the sixteenth century the older
Dantean notions assert themselves in one form or an-
other. Joseph Justus (the younger) Scaliger (1540-1609)
takes up the classification of the European languages
by matrices linguae or “mother languages” (in a tech-
nical sense) of which there are, according to him,
eleven. None of the eleven are related to each other.
With this, the idea of common, monolingual descent,
whether from Hebrew or otherwise, is abandoned as
unprovable at the very least. Each matrix lingua is
identified by shibboleth (see above). The criterion for
classification under a matrix lingua is by vocabulary,
and proceeds by inspection for “common charac-
teristics.” There is as yet (to make this anachronistic
comment) no hint at a more detailed analysis of the
change processes which make the descendants different
from their matrix lingua, and consequently no discus-
sion of how to evaluate common characteristics in
borderline instances.

This would not be worth remarking on but for the
fact that a few decades earlier an important step in
that direction had already been taken, quite incon-
spicuously, by Claudio Tolomei (1492-1555; quoted by
R. A. Hall, Jr. [1964], p. 301, with appropriate com-
ment) who for the first time (so far as our knowledge
of such priority goes) not only notes the difference
between (Latin) grammatica and (Italian) vernacular
but interprets it historically; he does so in a method-
ologically impeccable fashion through a confrontation


of doublets, within Italian, such as arise from a Latin
word as it has come down to “the middle of the town
square of Tuscany” (e.g., pieve from Lat. plebem) and
then again as the same word is “set up... by someone
who wished to enrich the language, preferring to use
[it] in the form in which he found [it] written in Latin”
(e.g., plebe).

The seventeenth century saw a good deal of theoret-
ical debate on grammar. It has been remarked that
the thinkers of Port-Royal wrote as though Mithridates
did not exist: no reference is made to languages at large
(Mounin [1967], p. 129), perhaps because, for their
purposes, none needed to be made. But in any event
this is not true of one of the greatest intellects of that
century, G. W. Leibniz. His function in the history of
linguistics is twofold: he continues the tradition of
language classification into families (though he enter-
tains the notion of a possible common descent for all);
an important aspect of this classification is its chrono-
logical-historical interpretation (he dwells on the im-
portance of such studies as a tool for the writing of
history). And secondly, in this field as in others, one
of his claims to fame is that of having been an organizer
of research. He tried to interest Peter the Great in
making a highly sophisticated language survey of his
dominions in Europe and Asia. The task was taken up
in earnest in the reign of Catherine II on an even wider
basis: word lists were obtained from governors, diplo-
mats, and scholars from all over the world and pub-
lished without delay in 1786 by the German traveler
P. S. Pallas under the title Comparative Vocabularies.
The work made a deep impression. We may be partic-
ularly grateful that it prompted a review by the
Koenigsberg political economist, Chr. J. Kraus (Arens
[1955], pp. 118-27). This review is the voice of a
well-informed man who, incidentally, had had some
practical experience in data-gathering. He sympathizes
with the curiously antispeculative, positivistic frame
of mind which characterizes the whole Russian effort
in the first place: he pleads for more careful elicitation
(“if there are mistakes in the material, no conceivable
reasoning is available to correct them”) and for a more
accurately planned, cartographic presentation of the
results. He also uses the term “comparison” almost like
a technical term (more so than others, e.g., Leibniz,
before him). Comparison, he says, has two aspects. In
part it is philosophical and teaches us how men think
and how differently they think: relativity in matters
of both sound and meaning is stressed as never before.
On the other hand, comparison is also a tool of history.
Just because structural differences can be so great, an
agreement or similarity between two languages proves
historical connection. In this respect we can appreciate
how the general outlook had changed, partly through
the enormous increase in empirical breadth, since the
days of Scaliger. What is still missing in this view of
languages is their role in history as objects (or agents)
of change rather than as mere witnesses towards the
establishment of historical truth. In this, Kraus was a
spokesman for his age.

The age, to be sure, was more varied than that. It
was also rife with controversy over the origin of lan-
guage, on the nontheological basis which the Enlight-
enment had provided. Thus, Rousseau and Lord
Monboddo, to select these two figures somewhat arbi-
trarily, were preceded—though apparently not influ-
enced—by G. B. Vico (1668-1744) and followed by
J. G. von Herder (1744-1803). The importance of men
like these lies in the fact that they were historians (or
philosophers of history) by temperament and thus had,
in the light of later events, something decisive to offer.
But the gap was great. Their influence was therefore
not always palpable and certainly not steady. It took
time and a personality and career like Jacob Grimm's
to weld Herder's feeling for universal and national
history to the substantive tradition. On the surface the
observer has the impression that the French Revolution
and the age of Napoleon, with the academic reforms
that followed, simply put an absolute end to all that
“prescientific” theorizing—an impression which was
allowed to harden into the familiar myth of the sudden
emergence of serious linguistics around the year 1800.

Pallas' collection, mentioned above, as well as the
one made largely from missions material by Lorenzo
Hervas in 1800-05, had the incomparable merit of
overwhelming the scholarly community with data
which were fresh and not encumbered with literary
history and artificial tradition. In the best style of the
doctrine of the Noble Savage this new knowledge
helped to gain a new perspective on more familiar
materials. For the same reason, the knowledge which
then existed of the more marginal European and Near
Eastern areas was especially instructive. Already in
Leibniz' time, the Semitist Job Ludolf (1624-1704) had
formulated a rather clear view of how to recognize a
“family” of languages (namely, not merely by words—
an allusion to the traditional fascination with vocabu-
lary lists—but by grammaticae ratio). Presumably he
was aided in this by his familiarity with an especially
close-knit and superficially diversified (as well as, of
course, literarily accessible) stock. Even more impres-
sive was the technical excellence with which the Finno-
Urgic family of languages had been explored, by Scan-
dinavians, like Ph. Strahlenberg in 1730, then in 1770
by the Hungarian J. Sajnovics, in Danish service, in his
Demonstratio, of the identity of the Hungarian with
the Lapp language, and finally by his far better known
compatriot S. Gyarmathi (1751-1830). Gyarmathi's


treatise on the relation between Hungarian and Finnish
appeared, under the aegis of the historian A. L. von
Schlözer (1735-1809) of Göttingen, and previously of
Saint Petersburg, the influential champion of the Slavs
and their East and North European neighbors.

Novelty of a different, and, as a matter of fact, very
ancient sort, was provided by the celebrated spread
of the knowledge of Sanskrit in the West. Sir William
Jones (1746-94) and other Englishmen studied Sanskrit
in India, that is, in natural conformity with the impos-
ing national tradition of the Hindu grammarians. Sir
William's enthusiasm was born in the general cultural
climate in which that distinguished man had spent his
earlier life, and was fed not only by the intellectual
excellence of Indian grammar but also by the attitude
of religious reverence with which it approaches its
subject, the “accomplished” (saṁskṛta) language. “The
Sanscrit language,” said Jones in 1786 “is of a wonder-
ful structure; more perfect than the Greek, more
copious than the Latin.” Comparisons of this sort were
indeed very much part of the eighteenth-century out-
look, except insofar as the new knowledge of indige-
nous languages had exerted a dampening effect. Such
comparisons remained in vogue much later (and are,
of course, still with us in folklore); thus W. von
Humboldt, writes, casually, in 1822: “Sanskrit is among
the oldest and first to possess a true edifice of gram-
matical forms..., but it is Greek which has undoubt-
edly attained the highest perfection of structure.”

It must, however, not be overlooked that Jones's
observation, conventional as it was, carried a special
force (likewise conventional in nature). Perfection and
copiousness were much-discussed properties with
which to place languages in the scheme of things. The
argumentation has theological roots, although these
were not necessarily visible under the eighteenth-
century guise of early evolutionism. A typical writer
of a hundred years earlier, Daniel Morhof, had put it
as follows: “It is most credible that the first language
was not one of the languages now known (which owe
most to art), but rather some language different from
them.” He was, then, one of those (like the Dante of
the Paradiso) who did not believe that man first spoke
Hebrew—because Hebrew is too “perfect” and not (as
we might now say) “primitive” enough. Clearly this
is why Jones goes on to say that the “first language”
from which the three perfect languages are descended
must be “a common source which perhaps no longer
exists.” Furthermore, when he states that the (later
so-called) Indo-European languages have a “stronger
affinity” than could be produced by accident, he takes
sides in another popular controversy, namely that of
possible polygenesis. Not only was it possible to doubt
that the language of Adam had been Hebrew, but
many, Lord Monboddo and Adam Smith among them,
were quite satisfied that more than one primeval lan-
guage could have existed. Like so many of the earlier
authors these men accepted the idea of an irreducible
number of separate families, each one with a primitive
ancestor. It was, therefore, important to Jones to de-
clare that in his opinion those languages had “a
stronger affinity” than could have been produced by
“accident” (when a more obvious source would have
been monogenesis!). Later in the nineteenth century,
such a statement would have carried the implication
that there exists a statable method for excluding acci-
dent, or that the proof of common descent lies in the
performance, in detail, of a consistent reconstruction.
It is excusable that such thoughts were also read into
Jones a hundred years later; but unless one takes a
fanciful view of preformation and premonition in the
history of scholarship this remains a distortion. This is
not to say that Jones was not interested in thinking
about what the “common source” may have been like.
Indeed, his inclinations went in that direction; he also
believed that “Pythagoras and Plato derived their
sublime theories from the same fountain with the sages
of India” (Edgerton [1946], p. 236)—an idea which,
by itself, should warn us against facile linear inter-
pretations of simple precursorship. Jones' enormous
scholarly distinction lay elsewhere; in matters of lan-
guage he shared the view of the best of his older as
well as younger contemporaries who, in turn, differed
less radically from one another than it appeared in

Nor should the position of F. Schlegel (1772-1829)
and his dithyrambic Sprache und Weisheit der Indier
(Language and Wisdom of the Indians), published in
1808, be oversimplified. Interestingly, the title contains
echos of “Pythagoras and Plato,” but this is less impor-
tant than the new turn given, apparently, to the con-
cept of comparison which, as we have seen, had slowly
come into prominence, at least since Leibniz. F.
Schlegel calls for a “comparative grammar” which will
furnish “completely novel insights into the genealogy
of languages,” on the express analogy of the then much
admired field of comparative anatomy. It must be
remembered that G. Cuvier's Leçons d'anatomie
had just appeared (1801-05) and that
Schlegel was one of those who found themselves in
Paris during the first years of the century, in contact
with the rising school of French Oriental scholars, but
also with similarly stranded Anglo-Indians like Alex-
ander Hamilton (1762-1824), the British orientalist. It
is most intriguing, also, to realize that it is precisely
in the names of those two disciplines, comparative
anatomy and comparative (grammar or) linguistics that
the word “comparative” carries, to this day, a technical


meaning which is no longer self-descriptive and indeed
quite often misunderstood. “Comparison” is here not
comparison for comparison's sake (i.e., what in linguis-
tics is usually called typology or typological compari-
son) but for the sake of retrieving a past, linguistic or
evolutionary as the case may be.

It is not for nothing, however, that Schlegel speaks
of comparative “grammar”; the term was not (as it later
became) simply a synonym for comparative linguistics.
His thinking about grammar is akin to that of Ludolf
and Gyarmathi. The choice is, in a sense, a negative
one. It stems from a dissatisfaction with arbitrary,
lexical etymology. Words travel easily, or come to
“resemble” one another by accident. Grammatical
structure goes deeper. Besides, Schlegel believed in
qualitative criteria both for classification and for the
determination of antiquity; to his mind languages re-
tain traits that are characteristically archaic and ac-
quire traits that are characteristically innovative. At
the same time, there are two classes of languages: the
inflected Indo-European ones which are born and have
developed “organically” (Mounin [1967], pp. 160-62),
and all others. For the above-named typological rea-
sons he is convinced (unlike Jones; and, of course, in
our view, “wrongly”) that Sanskrit is “the oldest” of
the Indo-European languages and that the others are
descended from Sanskrit, but (still, it appears, in defer-
ence to the idea that perfection is the result of growth)
Sanskrit is said to go back, in turn, to an older form
of speech. Language contact has produced “mixed”
idioms; non-inflected languages have thereby improved
their nature—the term is applied even though Schlegel
takes great romanticist care to disavow value-judg-
ments put forward too explicitly.

Schlegel can be a textbook case of the pitfalls of
“presentism” (Stocking [1965], p. 211)—so much so
that it is useful to stop and ponder some of the detail
by way of looking ahead. “Comparative” refers to a
typological comparison (in the ordinary sense of the
term) which enables the investigator to reconstruct
because he possesses qualitative knowledge to tell the
old from the new. At a later period, the term will
become entirely technical and will refer to a non-
qualitative, essentially formal, binary matching proce-
dure through which some features which, taken sever-
ally, might be retentions as well as innovations, are
shown to be either one or the other. Those features,
as we shall discover, are precisely not “grammatical”
ones in the ordinary (and still familiar) sense of the
word, but they are, more often than not, phonological
and can be discovered within the lexical material itself.
Hence the twofold quaintness of the term “compara-
tive grammar” as it is retained later on, or rather, as
it will have been transferred to activities of which
Schlegel knew and could know nothing. The typologi-
cal comparison, on the other hand, which for Schlegel
is fraught with chronological and classificatory mean-
ing, will lose its intimate connection with chronology
as well as with the so-called comparative method of
the later nineteenth century. It will be found that
genealogical classification and typological taxonomy
can intersect, and besides it will be clear not only that
the “inflected” sort of some of the better known and
older Indo-European languages is not typologically
unique, and also that the noninflected types one might
almost say, to the naked eye are as different from one
another as they are from what seemed the acme of
human perfection to Schlegel. In other words, if Jones
differed less than is thought from some of his contem-
poraries, so did Schlegel, although the forces that
impinged on the young German, writing fourteen years
after Jones' death, were naturally not the same.

One of the best-studied figures of that great period
is the Dane Rasmus Rask (1787-1832). The somewhat
grotesque and even tragic circumstances of his life are
not merely anecdotal. Apparently Rask was the man
he was because he went through an intellectual crisis
of an interesting sort; in Mounin's words: “In order
to understand [his career] in all its complexity one must
realize that while he started out as a romantic in the
German fashion and immersed himself in Scandinavian
antiquities for the same reasons, no doubt, as did his
contemporaries, he changes his vision rapidly and
ceases to be interested in text philology and in history
in order to turn to the problem of describing the system
of languages—a notion to which he is led by his
eighteenth-century education and by his personal bent”
(Mounin [1967], pp. 166-67). Rask was an admirer of
the botanist and classifier, Linnaeus. Thus, he became
a typologist avant la lettre, quite unlike some of his
deeply historical and romanticist contemporaries in the
heart of continental Europe. Hence perhaps some of
his more surprising “aberrations” such as his initial
refusal to recognize Celtic as an Indo-European sub-
family: Celtic, though demonstrably a descendant of
the common Indo-European ancestor, is typologically
one of its most deviant descendants. And, most intrigu-
ing of all: Rask's famous “anticipation” of “Grimm's
Law,” which from the point of view of cumulative
knowledge was an epoch-making accomplishment in-
deed, and one that figures rightfully in the subsequent,
linear histories of the profession, was probably intended
as a taxonomic observation. As Rask compares Greek
and Icelandic he finds (1) that both languages when
pronounced “correctly” have the same “letters,” i.e.,
sounds, (2) that they obey the same “euphonic” rules
whereby sounds alternate within paradigms in each
language, and (3) that certain transitions from sound


to sound and from language to language, such as be-
tween Greek p and Icelandic f in patér: faðir “father,”
appear very frequently. His biographer says: “It is
characteristic of Rask that he adduces all these 'iden-
tities' as equally valid proof of relationship” (Dide-
richsen [1960], p. 236). A later generation would have
regarded the first two as either merely universal or
typological, while according full status to the third
as an argument in favor of the specific historical con-
nection called common descent or “relatedness.” To
Rask, the sound correspondences (in the recognition
of which he was guided by J. G. Wachter's Glossarium
of 1737) stood as regular and recurrent
characters rather than as the result of events in time.
Nevertheless, it was the discovery of the recurrence
“set forth for the first time completely, and without
heterogeneous admixtures” (Diderichsen [1960], p.
236), that made all the difference.

If there was a man who was responsible for the
feeling of continuity with which the practitioners of
the discipline have looked back over the nineteenth
century (but not beyond it!) it was Jacob Grimm
(1785-1863). His and his brother's labors for the cause
of Teutonic antiquities—legal, folklore, and mytholog-
ical as well as linguistic—characterize him as the
paragon of the new romantic attitude. His Deutsche
(German, i.e., Germanic, Grammar) is a
“historical grammar,” perhaps the first of the many
works explicitly so named. Grimm's theoretical utter-
ances have been called vague, and so they seem at first
blush. But the impression disappears once an effort is
made to read them in the light of the accompanying
positive work. For once there is harmony between the
two. For all his metaphorical style, Grimm was averse
to speculation. He was taken with the concreteness and
individuality of language phenomena, and hence with
history. He combatted “general logic” and pre-
scriptiveness; he insisted on the primary importance
of dialects. The lexical aspect of language, with its
irrational abundance of life, fascinated him increasingly
as he matured. It was towards the end of his life that
he began to plan his famous German dictionary. Con-
sequently, he turned to etymology, conceived in the
modern sense as histoire des mots rather than as the
recovery of “real” meaning, as the literal (the “etymo-
logical”!) implication of the term has it. This emphasis,
combined with a characteristic desire to embrace all
the phenomena and see the typical while at the same
time keeping an eye out for aberrancy explains a new
and growing interest in matters of sound. The first
edition (1819) of the German Grammar does not deal
with phonology. The second (1822) contains 595 pages
of it where Rask's observation about the consonant
shift is not only incorporated but given the appropriate
chronological interpretation. Nor was this an isolated
instance; Grimm paid the same kind of attention to
the characteristic vowel alternations of Germanic
(those of the types sing: sang: sung, and foot: feet,
mouse: mice
) which still bear their Grimmian names,
ablaut and umlaut.

Wilhelm von Humboldt (1767-1835), most of whose
linguistic writings come from the years after his retire-
ment from politics in 1820, is the author of two lengthy
works both (1827-29 and 1830-35) concerned with the
“differences in language structure.” These highly liter-
ary treatises are difficult to evaluate because Humboldt
can so rarely be observed dealing systematically with
the language data to the collection of which he devoted
such vast time and energy. We notice at once, however,
how thoroughly Grimm had impinged on Humboldt's
thinking. There are references, in the spare manner
of the period. More important, unacknowledged pieces
like the important passage at the end of the chapter,
“Language and Nations” (in the 1827-29 work) show
how quickly innovation had spread. “In order to estab-
lish anything firm with regard to relationship between
languages it is necessary to... separate the similarities
which may be found among them.... I have used
historical connection as my main criterion.... As the
only proof of historical connection I have recognized
sound...” (Werke [1963], III, 357). This is not to say
that Humboldt does not contain other types of theoriz-
ing, both traditional and original, as well. In a sense,
he was as much of a typologist as he was a historian;
“the development of languages interested him only
insofar as it allows us... to go back to their origin,”
that is, to the point in time at which their typological
complexion was set (Mounin [1967], p. 189). What is
most remarkable of all, however, is the degree to which
Humboldt can be considered, legitimately or decep-
tively, to be the precursor of later linguists. It is a
tribute to his suggestive style, though it is also more
than that, that both the Whorfian relativists and the
deep-structure universalists of the twentieth century
have been able to claim Humboldt as their intellectual
sponsor: the former by quoting him to the effect that
a language will condition the speaker's understanding
of the world, and the latter by an appeal to the many
passages in which all language is admired for its re-
cursive energeia (“process”) rather than stationary
ergon (“accomplished act”) properties.

As Grimm's German Grammar had been epoch-
making by aiming at the mastery of massive rather than
intuitively (and conveniently) selected data, so did the
writings of Franz Bopp (1791-1867), especially his
1839 Vergleichende Grammatik (Comparative Gram-
), aim at “comparing and summarizing all that is
related” in the major Indo-European languages. A.


Meillet thought that Bopp “went out to explain Indo-
European [e.g., in terms of reducing affixation to
compounding with old, formerly independent roots, in
a way that was quite reminiscent of Platonic etymo-
] and ended up by finding comparative linguistics
[in the modern sense] just as Columbus had gone out
to travel to India and had discovered America” (Meillet
[1937], p. 458). To recognize obvious similarites is not
enough; the task, says Bopp, is rather to reduce “the
more or less extensive differences between them to the
laws to which they are due” (quoted after Arens [1955],
p. 199), that is, to discover those laws. And rather like
Rask, but very much unlike the majority of his con-
temporaries, Bopp turns his back on the idea that the
study of language is but a tool. “In this book, the
languages with which it deals are treated for their own
sake, that is as an object, rather than as a means, of
knowledge” (Arens [1955], p. 198).

Yet Bopp was centrally concerned with declension
and conjugation, that is, with what was still compara-
tive grammar in the narrower sense, and not, as Grimm
was, with the vocabulary as well. It fell to one of the
more peculiar figures of the period, the long-lived,
individualistic, and difficult A. F. Pott (1802-87), to
develop this side of things, and with it the heart, as
it were, of the coming “comparative method.” He
began in 1833, simultaneously with Bopp's work, to
publish his Etymologische Forschungen which contains
many of the etymologies that are still considered cen-
tral. Nor are statements of principle missing, some
quite running counter to ordinary cant; “the letter [i.e.,
the sound] is a surer guide in the labyrinth of etymol-
ogy than is meaning which is so often subject to sudden
leaps.” He is conscious of his intellectual kinship with
Grimm as he lauds him for “his historical exposition
of the sound changes in the Germanic languages
[which] has more value than many a philosophical
grammar full of one-sided and futile abstractions,” and
for “restoring 'letters' to their rightful place.”

A. Schleicher (1821-68) was younger than Pott by
a decisive nineteen years. Both were trained as classical
scholars but only Schleicher studied with F. Ritschl,
a contemporary of Pott. Ritschl's influence was enor-
mous; he presided over a school in which the art of
textual criticism by formal method was cultivated and,
as the younger generation came to feel, ultimately
mechanized and well run into the ground. Schleicher's
relations with Ritschl were close, and there is no doubt
that among his first intellectual experiences must have
been the exposure to such doctrines as the principle
of the shared error. If two or more manuscripts exhibit
some gaps or copying mistakes in common and to the
exclusion of other manuscripts (and if certain other
conditions are fulfilled), they are thereby proved to be
direct or indirect copies from one (sub-)archetype. The
established device to represent such dependencies was,
of course, the genealogical tree, or stemma. Schleicher
also had an early interest in the biology of the day
with its pre-Darwinistic evolutionism. His linguistic
work includes a number of intriguing, rather abstract
typological studies. He was also the first Indo-
Europeanist to abandon the traditional concentration
on the great literary languages as they appear in the
earliest texts. During a sojourn in the Baltic region he
collected texts of spoken Lithuanian and organized
them into a grammar and chrestomathy. By this he
added decisively to the knowledge of Indo-European
sources, and incidentally raised problems of a theoret-
ical sort for the discussion of which there had been
no real opportunity since the days of Pallas. In 1861
he published, in its first edition, his most famous work,
known in English as the Compendium of Comparative
Grammar of the Indo-European Sanskrit, Greek, and
Latin Languages
(2 vols., 1874-77).

To understand Schleicher one must note that he had
two preoccupations, one with phonology (his Lithuan-
ian grammar had a large section on this subject), and
one with the scheme of the tree to represent rela-
tionships within a language family. The two are closely
related in an operational sense for the following reason.
It was a consequence of the work of Grimm and Pott
that the nonrandom nature of sound correspondences
between “sister” languages was more and more seen
as the result of original sound differences having been
either retained or blurred in one but not in the other
of two descendant languages. In the simplest case this
means that when, say, a t in language I is in some words
answered by a t, but in other words (in essentially
similar kinds of syllables) by a d in language II, the
difference must be ascribed to the common source,
with the proviso that a sound change has eliminated
it in language I. Language I has innovated (by merging)
where language II has retained a feature. True, not
all innovations are phonetic; in fact those occurring
in other areas such as grammar and meaning are more
varied and in many ways more significant. But only
in the realm of sound is there a procedure of matching
which by itself determines which is retention and
which innovation.

Schleicher, in effect, represented innovating lan-
guages, that is, descendant languages sharing at least
one common innovation, on a separate branch of his
genealogical tree. In conformity with the prevailing
climate and with his own avowed interests he put an
evolutionist interpretation on the matter; first a pre-
Darwinian one, and then, when his attention was called
to Darwin's work, one in which he hailed Darwin as
a kindred spirit. Naturally both his followers and his


detractors followed suit. But it may be argued that the
real basis lies deeper and that Schleicher had, in fact,
acted on the logical analogy between phonemic in-
novation (in languages) and scribal error (in manuscript
copying). There is a lesson, then, in the manner in
which objective and biographical factors are inter-
woven in his story.

Like the oversimplified manuscript histories pro-
duced by Ritschl's students, the oversimplified language
families of Schleicher's trees do not allow subfamilies
to intersect; if A shares an error (a sound law) with
B to the exclusion of C, then it cannot significantly
share another error (another sound law) with C to the
exclusion of B. If it appears to do so nevertheless, this
must be a matter of independent duplication by acci-
dent or by a factor of intrinsic probability for that
particular error (or sound change) to occur more than
once—or else there was collation between sister man-
uscripts (contact between sister languages). To guard
against these possibilities it is therefore necessary to
amass all the correspondences and even subject them
to some qualitative judgment. Thus everything must
be accounted for: all the errors (sound changes) must
be explained. The goal, in any event, is clear. It is the
reconstruction of all the features of what is in one case
the archetype manuscript and in the other the proto-
language. In Schleicher's ideal view, the very act of
reconstituting the complete history of a phonology
from the descendants upward in time is identical with
the act of determining the distribution of the languages
among subfamilies. Hence his concern with the
“asterisked” proto-forms which he introduced into
scholarly use, and hence, quite naturally, even his
much-derided attempt at setting down on paper a
complete proto-Indo-European fable.

Thus Schlegel's comparative grammar, even while
retaining its name, was replaced by a powerful if
technically limited triangulation procedure. The limi-
tation was to the phonemic shape of dictionary items,
a concern that had not been particularly strong with
many earlier linguists. The power rested on the fact
that this limited reconstruction can be used as a sure
means, or so it was claimed, to resolve questions of
ancestry and descent, complete with subfamilial rela-
tionships, in a unique fashion. What was more, such
a claim necessarily went beyond the narrow confines
of linguistics; it was bound to interest historians and
archaeologists as well. There developed, in fact, a
borderline field of research devoted to the extralinguis-
tic exploitation of schemes of language relationship,
and the old romanticist attempts at reconstructing the
religion and mythology of the ancestral Indo-Europeans
were taken up again under the new auspices and this
time in subordination (not parallelism) to language,
especially by A. Kuhn (1812-81). As other aspects of
culture were added, however, and as the need to rec-
oncile “linguistic palaeontology” with regular archae-
ological work increased, it became painfully clear that
the Comparative Method as outlined above did not
primarily aim at the reconstruction of meanings. Still,
the Words-and-Things technique, as one particular side
of that special effort came to be called, was extended
to fields other than Indo-European and led to valid

But the principal results were intradisciplinary. Now
that reconstituted proto-forms were something con-
crete, the “uniformitarian” nature of knowable lan-
guage history was beyond the slightest doubt: proto-
languages were different from their descendants, but
no more so, necessarily, than the descendants were
from each other. They were “just languages,” and any
idea that fundamental alterations in the history of the
species had occurred during the shallow interval
accessible even to the newly refined method was
plainly wrong. It was in this vein that a good many
of the once familiar topics were discouraged, and that
the question of the origin of language, in particular,
was banned by statute from the proceedings of the
Société de Linguistique of Paris. The profession turned
in on itself, and a period of Victorian sobriety, marked
by indefatigable collection of data and by the testing
of the now orthodox procedures, set in.

One must, of course, distinguish between the actual
working principles and their formulation. The fact that
Schleicher himself did not and could not see his own
position clearly does not detract from his contribution,
nor from the fact that the changes that had come to
a head in the sixties were indeed great. The great
“neo-grammarian” debate came a decade later, and it
was little more than a somewhat murky expression of
what had already occurred. The total accountability
principle which had won the day in matters of phonol-
ogy did indeed make it necessary for the researchers
to distinguish replacements (from older to later stage)
which may be stated without reference to specific word
lists (e.g., all kn- is replaced, in spoken southern and
standard English by n- at a certain time in history;
hence (k)not (k)night (k)now (k)nee) from others requir-
ing such listing (inwit yields to conscience; rooves, in
some styles of speech, to roofs, while knives retains
its -v-). In a somewhat misleading way, the former may
be labeled “regular.” Alternatively, they were also
labeled with the technical term “sound change.” Taken
in this way, the regularity of sound change, or as A.
Leskien (1840-1916) put it in 1876, the exceptionless-
ness of sound laws, is only a tautology. The fact that
it became and remained a battle cry for the next sixty
years of theoretical debate was unfortunate because


it clouded the real issue which was important enough
and which had to do with the typical historical and
demographic settings in which sound changes are likely
to take place. But the problem could not be clearly
stated until the advent of synchronic phonemic analy-
sis. Thus it is worth remarking in passing that Leskien
had more reason than his classicist and Sanskritist
colleagues to worry about the nature of sound replace-
ment; he stands in the Schleicher tradition of East
European studies, where alphabetic representation is
not always a philological datum but part of the
scholar's work—and a theretofore unformulated part
at that.

However that may be, some of the important new
discoveries were being made by men who did not
recognize Leskien's dogma but who nevertheless ad-
hered to the notion of reconstruction underlying that
dogma. Karl Verner (1846-96) discovered the hidden
“regularity” of “exceptions” to Grimm's Law (in the
very same year, 1876) and it became one of the most
potent arguments in favor of exceptionlessness. Yet
Verner remained pointedly aloof himself. Others were
more belligerent in their attacks and pointed to situa-
tions in which the tree model had important factual
weaknesses. Subsequent contact between descendants
may be extensive; there may be no clear cleavage. This
does not invalidate the comparative method, but it may
pose serious problems to its application. After Johannes
Schmidt (1843-1901) had pointed this out, Schmidt's
“wave theory” was long regarded as a kind of antidote
to Schleicher's tree. More interesting than Schmidt's
opposition was that of Hugo Schuchardt (1842-1927),
who was not an Indo-Europeanist but more specifically
a Romance scholar. The injection of fresh and different
material into the theoretical debate had great impor-
tance. It was primarily in the Romance and the
Germanic fields in which the graded nature of dialect
areas was investigated, and men like Jost Winteler
(1846-1929) and Jules Gilliéron (1854-1926), the initi-
ator of the French Linguistic Atlas, contributed de-
cisively to the theory of change in ways which, on
mature reflection, turn out to be far less destructive
to the real core underlying the “neo-grammarian”
position than is sometimes thought (Bloomfield [1932],
p. 231).

The precariousness of a methodological discussion
which was not always relevant to the real scholarly
issues was most keenly felt by some of the most brilliant
pragmatic workers. The clearest cases in point were
perhaps the Italian Graziadio I. Ascoli (1829-1907)
and, a generation later, the Swiss scholar Ferdinand
de Saussure (1857-1913). Both had made decisive
technical contributions to what was then the center
of the field (in 1870 and 1879, respectively), both were
dissatisfied with the theory, as it was then stated, of
what they were doing, and both, being physically
somewhat removed from the center of things in the
German universities, found original and effective ways
of speaking out. Ascoli's strength was that he controlled
Semitic and Romance materials in addition to being
in the mainstream of the Indo-European work. He
wrote with a detachment and self-knowledge that are
almost without parallel at the time. The concept which
has made him famous is that of the ethnic substratum
as a “cause” of linguistic change. By asking, in effect,
“How must we picture the genesis of different lan-
guages on the soil of the Roman empire?” (Arens
[1955], p. 333), he took an important step toward filling
the theoretical gap which had been left since Hum-
boldt by the increasing neglect of the synchronic study
of speech communities and of language contact.

Hermann Paul (1846-1921), the great and consci-
entious theorist of the neo-grammarians, still held in
his Principles of the History of Language that only a
historical study of language had scholarly and scientific
value when a new interest in nonhistorical questions
began to stir. In part this interest seemed innocuous
because it was clearly limited to an area which, it
seemed, could safely be relinquished to specialists. This
was phonetics. There was some overlapping, especially
in the persons of E. Sievers (1850-1932) and Henry
Sweet (1845-1912); but especially in England, with
Sweet's successors, phonetics became indeed a disci-
pline by itself. The reason why this was possible lay
in the fact that the gap between synchrony and
diachrony (history) remained to be bridged: it was not
clear just how the physical description of speech sounds
related to the entities which had somehow come to
be represented in the (syllabic or) alphabetic records
in which the various well-studied literary languages
were recorded. Alphabetic representation was obvi-
ously a simplification of endless physical fact, yet it
was just as obviously an appropriate and congenial
simplification. The old confusion between letter and
sound was, for some reason, not all that destructive.

The problem had come to the special attention of
those few leading linguists who had combined histori-
cal-comparative studies with “field” activities. This was
true of Schleicher, Leskien, and Winteler, as was
pointed out before; it also held for O. Böhtlingk (1815-
1904) with his fascinating combination of achievements
in Sanskrit lexicography, Hindu grammar, and, the
Russian ethnographic-linguistic tradition, with his ex-
pert description of Yakut, a Turkic language of Siberia.
Students of European dialects were less troubled by
it, since many of them were more interested in con-
founding the neo-grammarian enemy by judicious use
of phonetic detail than in clarifying the issue. Some


extra-Indo-European fields, however, suffered consid-
erable damage; Finno-Ugric studies, for example, were
seriously hampered by the deliberate and perverse
introduction of phonetic raw material into compara-
tive problems.

Characteristically, the problem was first seen in all
its depth on East European soil, by J. Baudoin de
Courtenay (1845-1929) and his collaborator N.
Kruszewski, both Poles. By defining the role of opposi-
tions or contrasts between classes of sounds they laid
the groundwork of phonemic theory and suggested,
among many other things, that alphabetic records of
the familiar sort do not, in the ideal case, falsify lan-
guage just because the phonetic detail is simplified,
since the phonetic detail can be classified into func-
tionally unimportant and functionally important
(“phonemic”) fact. It turned out that those who had
worked with literary languages had in reality always
had an implicit analysis performed for them by the
orthographic tradition. It was on the basis of this un-
derstanding that de Saussure proceeded in his lectures
(published by his students as the famous Course in
General Linguistics
in 1916) to explain the relation
between synchrony and history. Out of this pair he
made, partly for pedagogical reasons, a rather rigid
dichotomy. In this sense de Saussure is hailed as the
originator of structuralism in linguistics. The particular
phonemic facets of the movement were hammered out
with great vigor, especially by N. Trubetskoy (1890-
1938), with strong attention being paid to nonliterary
languages (Trubetskoy had a deep knowledge of the
languages of the Caucasus) and, in general, to syn-
chronic description as much as to the analysis of

The impact of this work was very great, not only
on the philosophical view of language, but on typology
which it helped to come to new life. On the whole,
the study of nonliterary and “primitive” languages had
languished since Humboldt. Outside of the occasional
activities mentioned earlier the only major figure to
approach such material from a linguistic point of view
was F. N. Finck (1867-1910). His amazing Haupttypen
des Sprachbaues
(1910) stood, to all intents and pur-
poses, alone, and it is quite unjust to go on blaming
philosophers from W. Wundt to Ernst Cassirer for
building on poor empirical foundations when nothing
better was being offered. The change came, partly with
Trubetskoy's Prague School, and partly when F. Boas
(1858-1942), then professor of anthropology at
Columbia University and also Honorary Philologist at
the Bureau of American Ethnology, began his prodi-
gious work on American Indian languages, both in the
Handbook of American Indian Languages (2 vols.,
1911-12) and elsewhere. Building on very respectable
domestic traditions he became the founder of a highly
successful school of American descriptive linguistics
dedicated to careful phonetic recording and to an
objective and strictly relativistic conception of gram-
mar and semantics in which as little as possible was
taken for granted, and any analogies with language
categories familiar from traditional (i.e., European)
language structure were suspect. E. Sapir (1884-1939)
and L. Bloomfield (1887-1949) had been closer than
Boas to academic linguistics of the conventional kind;
both brilliant field workers, they spent considerable
effort on problems of genealogical as well as typologi-
cal classification. All three were considerable thinkers
on the nature of language as a faculty and as an institu-
tion, and all three had their outlook profoundly deter-
mined by their widened typological knowledge. Some
of the implications of all this were formulated in the
highly stimulating writings of B. L. Whorf (1897-1941).
In a manner reminiscent of Humboldt's ideas on the
relation between language structure and world view,
Whorf propounded the primacy of the semantic struc-
ture of the language, with its formal properties, over
much that passes in ordinary, and not so ordinary,
philosophy as content such as is amenable to expression
in any language but in itself preexistent. Considering
the observations made along the line, and the value
of the ensuing debate (Hoijer, 1954), Whorf's innocence
of the prehistory of this old philosophical problem did
little damage.

The need to catch up, as it were, with “classical”
comparative linguistics, and to give it the theoretical
foundation it had not had, had other aspects, too. On
one of them we have touched already. It was one thing
to classify change processes, even successfully so far
as methodological goals were concerned—this had
been accomplished in the 1860's or shortly thereafter;
it was another thing to interpret the classification,
again in synchronic terms, by preserving historical
concreteness. Ascoli had pointed the way, and others
followed. Among them were a number of distinguished
Frenchmen who knew Durkheim's sociology, like
Antoine Meillet (1866-1936), whose famous essay on
How Words Change Their Meaning (1958, pp. 230-71)
shows him at his best on generalized topic, although
his strength lay on the whole in his finesse on individual
points and his ability to synthesize vast fields of factual
knowledge—his Introduction to Indo-European Com-
parative Linguistics,
while dependent on the gigantic
neo-grammarian achievement of K. Brugmann (1849-
1919) and B. Delbrück (1842-1922), is nevertheless the
one enduring interpretive work on this classical subject.
In time this line of endeavor was to develop into the
field of sociolinguistics in which the structure of speech
communities is studied in such a way as to enclose


within itself as a special case as it were, the study of
linguistic change. This is especially easy to illustrate
with instances of borrowing; that is, with certain con-
sequences of contact between language communities—
the phenomenon to which U. Weinreich (1926-67), in
particular, directed his attention. But there is no doubt
that the point of view is far broader. Here, too, very
old concepts are coming to the fore again: language
mixture, for one, as reinterpreted in the light of such
seemingly marginal phenomena as the pidgin and
creolized languages which captivated the original and
unorthodox mind of O. Jespersen (1860-1943).

It would be strange if the discipline of linguistics
with its constant thrust toward greater and greater
formalization had remained untouched by develop-
ments in mathematics and in logic. Much of this (espe-
cially what belongs with the much-discussed relation
between linguistic and philosophical semantics) has,
almost by definition, little to do with language diver-
Statistical methods have, of course, been tenta-
tively applied to problems of linguistic change—most
spectacularly perhaps, in the effort of M. Swadesh
(1909-67) to calculate the degree of time depth behind
divergent members of a family or of an area through
glottochronology, that is, by assuming constancy for
the rate with which “basic” vocabulary is replaced.
The demand for content analysis or translation by
machine has had very little direct effect on the theory
of natural languages, although it has led indirectly to
the asking of novel questions on the subject.

Transformational syntax as developed in the 1950's
and 1960's is, however, a development of such general
importance that it is inevitable that it should bear on
the theory of language diversity. The claims vary
greatly, but in the view of those who prefer a “genera-
tive” formulation it is evident that the relativity of
natural language structure has receded into the back-
ground or rather gone to the level of “surface struc-
ture”; all languages share much or all of their deep
structure which is often regarded as genetically given.
In an older, slightly ambiguous, terminology the deep
structure is a “universal,” but not quite in the sense
in which the tag has been applied to empirically wide-
spread surface features or, especially, their concatena-
tion. This relegates both typological and historical
diversification to a low station in the hierarchy of
concepts and, for that matter, of grammatical state-
ment. But just because of that, manners of changing
genealogical relationship, and the grading of typologi-
cal areas created by diffusion (“contact”) processes
must be reformulated. What intrigues the historian of
methodology is the fact that in addition to sound
change, the other varieties of linguistic change may
find a more solid berth in this framework than they
have had in the past. It is an almost traditional com-
plaint in the literature that neither historical syntax
nor semantic change had had any really systematic
treatment and that even a classical work of J. Wacker-
nagel (1853-1938), Vorlesungen über Syntax, or of
C. D. Buck (1866-1955), Dictionary of Selected Syno-
mainly serves to show the lack of proper concepts.

Linguistics is a self-conscious field. Throughout its
existence it has not only developed working theories
about its subject, language (or, in our special cases,
languages); it has also theorized about itself. More so,
perhaps, than in some other areas of knowledge, the
results have tended to be unhappy. Self-description,
of which there is a good deal in the form of forewords,
popularizations, and polemics must of course be taken
with a grain of salt always; much the same is true of
historiography. The tendency to look upon the people
of the past with the simple question, what did they
know, or fail to know, that we, now, consider true,
is all too frequent. The alternative attitude, that is, one
in which past error becomes as important as antici-
pated truth, is of no immediate apologetic value, and
yet it alone guarantees an understanding of how truth
is found. Some of the pitfalls are only verbal, but others
are more subtle: few things, for example, are as in-
structive as the manner in which the biblical and
classical notion of ancestry and descent, amenable as
it is to the metaphor of the tree with its successive
bifurcations, is filled, almost unbeknownst to the writer,
with a fresh content; or how the concept of “compari-
son” (which is of course not limited to concern with
languages) changes so decisively.


H. Arens, Sprachwissenschaft (Freiburg and Munich,
1955). A. Borst, Der Turmbau von Babel (Stuttgart, 1957-63).
B. Delbrück, Einleitung in das Studium der indogermani-
schen Sprachen,
6th ed. (Leipzig, 1919). P. Diderichsen,
Rasmus Rask og den grammatiske tradition, Hist. Filos.
Medd. Dan. Vid. Selsk. 38, no. 2 (Copenhagen, 1960). F.
Edgerton, “Sir William Jones: 1746-1794,” Journal of the
American Oriental Society,
66 (1946), 230-39; reprinted in
Portraits of Linguists, ed. T. A. Sebeok (Bloomington and
London, 1966), I, 1-18. R. A. Hall, Jr., Introductory Linguis-
(Philadelphia and New York, 1964). Harry Hoijer, ed.,
Language in Culture (Chicago, 1954). Also published as
Memoir No. 79 of the American Anthropological Associa-
tion. S. N. Kramer, “The Babel of Tongues: A Sumerian
Version,” Journal of the American Oriental Society, 88
(1968), 108-11. A. Meillet, La méthode comparative en
linguistique historique
(Paris, 1924); trans. G. B. Ford as
The Comparative Method in Historical Linguistics (Paris,
1967); idem, Introduction à l'étude comparative des langues
8th ed. (Paris, 1937); idem, Linguistique
històrique et linguistique générale,
latest ed. (Paris, 1958).


G. Mounin, Histoire de la linguistique des origines au XXe
(Paris, 1967). H. Pedersen, Sprogvidenskaben i det
nittende aarhundrede
(Copenhagen, 1924); trans. J. W.
Spargo, as Linguistic Science in the Nineteenth Century
(Cambridge, Mass., 1931); reissued as The Discovery of
(Bloomington, 1962). R. Pfeiffer, History of Clas-
sical Scholarship
(Oxford, 1968). M. Pohlenz, “Zeno von
Kition und die Tempora,” Neue Jahrbücher für Wissenschaft
und Jugendbildung,
2 (1926), 259-60. E. Prokosch, A Com-
parative Germanic Grammar
(Philadelphia, 1939). R. H.
Robins, A Short History of Linguistics (Bloomington and
London, 1967, 1968). T. A. Sebeok, ed., Portraits of Linguists
(Bloomington and London, 1966). H. Steinthal, Geschichte
der Sprachwissenschaft bei den Griechen und Römern,
ed. (Berlin, 1890, 1891). G. W. Stocking, Jr., “On the Limits
of 'Presentism' and 'Historicism' in the Historiography of
the Behavioral Sciences,” Journal of the History of the
Behavioral Sciences,
1 (1965), 211-18. V. Thomsen, Sprog-
videnskabens historie
(Copenhagen, 1902); trans. H. Pollak,
as Geschichte der Sprachwissenschaft (Halle, 1927). W. von
Humboldt, Werke in fünf Bänden, ed. A. Flitner and K.
Giel (Darmstadt, 1960, 1961, 1963, 1964, —).

Several of the works cited make good introductory read-
ing, especially those by Meillet (Introduction, Appendix I),
Mounin, Pedersen, Robins, Sebeok, Steinthal, and Thomsen.


[See also Evolutionism; Language; Linguistic Theories;
Myth; Primitivism; Romanticism; Structure; Uniformitar-