Inductivism and the emergence of modern descriptive linguistics

Michael Silverstein

The University of Chicago

For nearly 50 years within linguistics, analytic and pedagogic inductivism, have been rejected out of hand, because confused with certain doctrines of behaviorism and operationalism.  Yet the two faces of induction – as a faculty of the rational mind and as a moment in the dialectic of scientific methodology – were central in the institutionally twin enterprises of the applied and the pure in emerging discplinary linguistics from the late 19th century through the post WWII era.  For this span of our discipline’s history, we can compare the theory and materials of “inductive” language pedagogy with the field methods by which the creation of essentially philological materials constituted the basis of linguistic analysis and theorization.  We can see the specific emergence of the institutions of our science before its “deductive” turn.

From the third quarter of the nineteenth century, there was no question among the

linguistically sophisticated but that “inductive” language teaching was the innovative

method, to be preferred to memorization of grammatical rules (from which answers to

lexically specific exercises could be “deduced”). There was as well no question but that

comparative philology – linguistics to the more terminologically aspiring – ranked in the

vanguard of the human sciences because it was, in fact, an “inductive science” just like

botany and geology. And there was no question but that the human mental faculty of

induction, or inductive conceptual reasoning, allowed us both to learn languages and to

make useful and thus pragmatically valid scientific generalizations about linguistic

phenomena. Induction had three effective faces or avatars, a pedagogical, an

epistemological, and an ontological. My aim here is to recuperate some specific

moments in the intertwining of these several faces of induction. I would also propose

that this is, in fact, the fascinating main conceptual plotline of what became disciplinary

linguistics in America (and elsewhere) “B.C.” – before Chomsky – and is emerging again

in various new intellectual trends.

Jan Blommaert, then of Ghent University, now of University of London, posed a

question to me on the very morning, in Winter, 2003, that he arrived in Chicago as a

Visiting Professor: what can we glean about underlying scientific outlooks from the form

of fieldnotes taken by our anthropological linguistic predecessors in their research? I

have no experience on the Africanist side of things, and leave that to Blommaert and to

my Michigan colleague Judith Irvine and others who are exploring the intellectual and

professional histories of scholarly work there. My research has sought to contextualize

the scholarly text artifacts that link professional work on American Native Languages –

as Franz Boas termed them – to the late nineteenth-century emergence of

professionalized disciplinary linguistics itself, an intellectual context which by then had

already been maturing over more than a hundred years. My claim here is that there is

causal continuity in the transfer of inductivisms across both geographical and conceptual

space: from the eastern to western shores of the Atlantic; from Indo-European, Semitic,

Finno-Ugric, etc. to Algonquian, Wakashan, Uto-Aztecan, etc.; from theorizing at first

driven by linguistic comparison and history to the theoretical priority of structural or

descriptive approaches; from found text artifacts in ancient scripts to fieldnote

inscriptions capturing spoken textualities.

The story to be told is, as one might guess, very complex, interweaving large

numbers of scholars and the institutional fields in which they moved for training,

research, teaching, and publication. I approach the problem as one of applied linguistic

anthropology – perhaps even of philology as earlier conceived – opening out such

2

documents as fieldnotes to the “natives’ ” cultural contextualizations that they seem to us

to index. What I thus wish to sketch today is a kind of historical ethnography of

inductivism in linguistics, attempting to discern the cultural system of practitioners in the

field and to connect the dots among the practitioners.

Now cultural systems are, like languages (langages) themselves, sociohistorical

objects observable only in human groups and they are inexorably “drifting,” as Sapir

noted; the sociohistorical fact of scientific or scholarly disciplinarity, organized by

scholarly culture, is no exception. What is significant about the history of inductivism is

the fact that at many different phases of its ascendancy in linguistics its institutional site

shifts as to the location of what we might term peak, influential expression of the cultural

system. Yet in all of these contexts, the very term “inductive” or its semantic relata are

positively valued ones; it is used again and again as a term of triumphant, empirical

modernity, describing a sound, normal-scientific outlook of someone from within the

confraternity of practitioners. It is a deictic term of value-conferral. It is a term that goes

into a complimentary evaluation of someone’s philological work by associating it with

the “other” inductive sciences, like botany, geology, etc. Linguistics – and linguists –

aspired thus to be Baconians.

As a concept explicitly invoked in theoretical discourse, though, the inductive

attitude shifts its associability as a metapragmatic, a regimenter of practice, with now

one, now another area of linguists’ practices in their engagement with languages and with

language, sometimes being substituted by another term in a local controversy (thus:

exceptionlessness as an ontological commitment of Neogrammarianism, grounding

epistemological inductivism in philological study with a theoretical program). These

3

shifts or substitutions, always the condition of culture and revealing to the analyst of it,

have perhaps been somewhat invisible to the tendency of historians of linguistics, who

have tended rather to work with a unifying focus in intellectual biography, organizational

history, or emergent functional institutionality (e.g., “schools,” “invisible colleges”

[Merton]). That is why, perhaps, this story has never before been told. But the newer

history of science qua history of ideas, with which I associate any reflexive historical

ethnography of our field, cuts across these foci to study the cultural dynamics immanent

in the way the self-licensing ideas and methods of people “doing science” exist in

complex social fields of interest, power, aspiration, etc.

Let me contextualize this by turning to some fascinating materials to hand,

exhibits of text artifacts from which – inductively – we can, I hope, discern the storyline

of a generalizing cultural theme central to linguistics.

The founding president of my university, William Rainey Harper (1856-1906),

made a scholarly name for himself as scholar of Biblical Hebrew. But perhaps more

importantly he was an entrepreneur who had earlier invented “The Hebrew

Correspondence School” (among many ventures that comprised his American Institute of

Hebrew) while serving as professor at the Baptist Union’s Theological School in Morgan

Park, just south of the then Chicago city limits. [Image 1]. Each fortnight or so the

paying students received a booklet that comprised the lesson [Image 2], which was

essentially a set of word-by-word glossing notes to a verse of the Old Testament [Image

3], together with further “Observations” referenced to a grammar, and followed by a set

of reading-recitation and back-and-forth translation exercises based cumulatively on the

material thus far [Image 4]. Harper – familiar to the Chautauqua circuit of quick tent-

4

sales as well as Baptist Revivalism – also, of course, sold to the Correspondence School

students the Manual on which the lessons were based [Image 5], viz., a book of the Bible

done up as texts in facing-page Hebrew and word-by-word English translation [Image 6],

as well as his own reference grammars, Elements of Hebrew by an Inductive Method

(1881 et seq.) and later Elements of Hebrew Syntax by an Inductive Method (1888 et

seq.), to the paragraphs of which are keyed the explanatory remarks in each of his

lessons.

As you can see, the whole “inductive method” here is text-centered, with the

pupils encountering and at least parroting and trying to use “real Biblical Hebrew” from

day one – appropriately, of course, the day of The Creation! They thus learn any rules or

generalizations about Hebrew in the order in which the language is illustrated by the

particulars of usage in the text. Each such usage encountered is paradigmatically

embedded, in the referenced paragraphs of the Elements, in its declensional or

conjugational frame, as well as its frame of syntagmatic particulars. There is no langue

learned, we might say, outside of its proper context of parole; here, that context is the

text of Genesis. This feature is pervasive. Even Harper’s Hebrew Syntax itself [Image

7], though organized as a systematic discussion of the syntax of various kinds of phrasal

heads, from major to minor – note sections 14 on “pronominal expressions” and 15 on the

syntax of numerals [Image 8] – is essentially a set of actual textual examples keyed back

to the book—chapter—verse loci of the Old Testament, with cautionary remarks and then

a whole set of further textual “references for study,” so as to clinch the generalization by

relying on the inductive capacity of the student.

5

You can see immediately that this is essentially a LEXICOGRAMMATICAL

CONCORDANCE TO THE TEXT. Theoretically, at least, if we collected all of the exhaustive

textual references we should be able painstakingly to reconstruct the text of the Bible.

The text has been segmented into chunks of form according as, collected in one place,

whole sets of such chunks of text consist of exemplifications of a single grammatical fact

about the language, which thus, stated in the grammatical treatise, indexes (points to) its

multiple occurrences in the original text. We might say in the register of philosophy of

scientific language, that the intensional [with-an-ess] statement of the generalization is

extensionalizable as – realizable as – the set of textual exemplifications, locatable

numerically, which are theoretically exhaustive in a finite – even if inspired – corpus

such as the Bible was believed to be in the 1880s (especially by religious folk).

The concept is exactly the same for Hermann Grassmann’s Wörterbuch zum Rig-

Veda, originally published in 1873 [Image 9], though Grassmann is based on

intensionalization by word-root, rather than by construction-type. I recall my humbling

sense of unbounded admiration when, as a student in second-year Sanskrit, I encountered

this incredible work all done by hand in which, note [Image 10], every declensional or

conjugational form of every root in its different senses – including, vacuously,

indeclinables – is classified, grouped, and keyed to the place in the standard numbering of

Rigvedic hymns and the poetic line or śloka in that hymn where the form occurs. From

the data in this work, too, one can reconstruct the entire Rigvedic corpus up to but not

determining the exact order of words in each śloka (for which one would have to use the

guidance of metrics).

6

So the concordance, that indispensable product of and tool for the philology of all

the ancient languages, is an intermediate shuffling of the data of textual occurrences

according to some generalizations – the forms of what we now call the grammar – that

permit identification of specific instantiations “in nature,” as it were. It is a meeting point

of both Mr Harper’s “inductive method” of language teaching and of Herr Dr

Grassmann’s textual philology – and that of all our other great predecessors of the time.

The pedagogical “inductivism” of Harper’s Hebrew course is anchored in practice in

precisely the same kind of text artifacts as the epistemological “inductivism” of high-

water Neogrammarianism, the crowning achievement of the previous hundred years of

linguistics. For the ancient languages in particular, then, the procedures by which text,

translation, lexicogrammatical concordance, and teaching lessons are used by students

inculcates them into the inductive world of philology, just as this world produced them as

the product of scholarly induction.

Harper, indeed, produced not only his famous Hebrew curriculum, but,

collaboratively with others, one for Greek and one for Latin as well, both published in

1888. Note the Latin one [Image 11], which is based on Caesar’s Gallic War. Caesar’s

opening sentence is at length completed in Lesson III [Image 12], with specifically keyed

notes and further observations. Interestingly [Image 13], a little lexicon of occurring

words – the “A” set – is supplemented – the “B” set – by denotationally related words,

allowing wider, paradigmatically contrastive translation exercises. Finally, since Latin

grammar is supposed to illuminate one’s language arts, there are questions that point up

the typological contrasts of English and Latin.

7

Now it seems clear that Harper must have been exposed to the paraphernalia at

the heart of the philological program for ancient languages when he did his Ph.D. work at

Yale (1873-1875) principally under William Dwight Whitney, the great Yankee apostle

of the inductive science of language. “It was Whitney,” recollected Harper’s colleague,

the Latinist Charles Chandler (quoted in Goodspeed 1928:39), “who had pointed out to

him that the Semitic languages were a very promising field for exploitation by an

enterprising man, both text-books and methods here and abroad being antiquated,

unscientific, and in America notoriously futile.” Antiquated and unscientific meant non-

inductive, of course. And indeed, what Whitney had done philologically for Vedic and

later Sanskrit – producing definitive editions of texts, creating critical translations,

producing exhaustive concordanced lexicogrammatical databanks, collaborating on

dictionaries, writing a definitive grammar based on these – Harper undertook as the

foundation for his inductive language courses in Hebrew. The genius here is the

reshuffling, the reorganization of all the philological material around the text itself as a

stimulus to learning the language as a recapitulative process of philological induction.

This was exactly what Harper had been doing in Whitney’s introductory and Vedic

Sanskrit classes, of course (at the time Whitney’s exemplary Sanskrit Grammar was still

to be even commissioned of him on a visit to Leipzig in June, 1875). As Harper wrote in

one of his early (1880) circulars from The American Institute of Hebrew’s

correspondence school (U of C Spec. Coll., Amer.Inst.Sacr.Lit., Records, Box 2, fold.1):

The method of instruction employed in the lessons is largely the

inductive – a method which experience has shown to be the simplest and

most attractive, as well as the most thorough and efficient. The facts of

8

the language are first mastered, and then the principles which these facts

teach.

Facts first, principles later. But of course Whitney, too, had been doing this not

only for Sanskrit but for modern languages as well. He was an instructor for some 20-

plus years in French and German in the Sheffield Scientific School at Yale, and later

(1883) a founder of the Modern Language Association. He published teaching grammars

and philologically treated, graded texts in both languages, as well as a teaching grammar

of English keyed, as we might expect, to a collection of examples such as we see in the

lemmata of the OED – attested “usage” as the extended proof of intensional rules or

generalizations. As Whitney says in his preface to his Compendious German grammar

(1869; 61888:iv),

[T]here is a large and increasing class of students whose philological

training has to be won chiefly or altogether in the study of the modern

languages, instead of the classical – and who must win it by methods

somewhat akin with those so long and so successfully followed in

classical study.

He means, of course, text-centered philological induction that he got in the direct line of

Franz Bopp via the mediation of his own two principal masters during his student days in

Germany, Albrecht Weber and especially Rudolph von Roth, with the latter of whom

Whitney had produced the editio princeps of the Atharvaveda in 1855-56.

I will return to the epistemology of philological induction later, but wish now to

point out that there was, in fact, a second strand of inductivism leading to Harper’s

lessons. It is the whole tradition of pedagogical inductivism that is, like the scientific,

9

ultimately sprung from the great Baconian instauration. As a pedagogical theory, it was

vigorously and influentially advocated by Comenius (Jan Komenský [1592-1670]) and

taken up as a practical program with extraordinary effect during thelate 18th – early 19th

century by Johann Heinrich Pestalozzi (1746-1827), the Swiss educator. Pestalozzi ran

several of what we would today call demonstration, or “laboratory” schools for young

children in which subjects like natural philosophy, mathematics, and language were

induced as knowledge in the young through a kind of step-by-step, prioritized exposure

crafted developmentally. Pestalozzi’s ideas influenced generations of reformers as well

as designers of curriculum, in which the adjective Pestalozzian became synonymous with

inductive. Among his followers were, in the United States, the great educational

reformer Horace Mann, who created the idea of the normal school – that is, teacher-

training college – to create a corps of Pestalozzians, and in Europe such figures as

Friedrich Wilhelm August Fröbel (1782-1852), who was central to Prussian elementary

educational reform. (We will pick up this story below, when Franz Boas’s mother Sophie

Meyer Boas [1828- ] starts a Fröbel Kindergarten in Minden, Westphalia, in 1860.)

After Pestalozzi, “inductive” courses for elementary phases of subjects were all

the rage. Warren Colburn’s (1793-1833) 1833 An introduction to algebra: upon the

inductive method of instruction, characterizes “the best mode” of instruction this way

(1833:3):

[T]o give examples so simple as to require little or no explanation, and let

the learner reason for himself, taking care to make them more difficult as

he proceeds. This method, besides giving the learner confidence, by

making him rely on his own powers, is much more interesting to him,

10

because he seems to himself to be constantly making new discoveries.

Indeed, an apt scholar will frequently make original explanations much

more simple than would have been given by the author.

The pedagogy of discovery allied a subject taught with the most prestigious of post-

Romantic ideas of pedagogy for the epistemological growth of the student; indeed, the

“inductive sciences” – what we today call the empirical natural sciences – were at the

center of 19th century philosophizing about and consideration of the history of science.

Note the preface to this end-of-century Introduction to qualitative chemical analysis by

the inductive method by one Delos Fall, a professor of chemistry at Albion College in

Michigan (1891:4):

On entering the laboratory, whether physical, biological, or

chemical, the student should thoroughly appreciate the changed relations

and different surroundings in which he finds himself. He has studied

literature, history, mathematics, language, or philosophy in the privacy

and quiet of his own room; he has studied from books, but, on taking up

the study of the natural sciences his books have been taken from him and

in their place he has been given specimens and objects of natural history.

He is led to look at these objects, observe as many points as possible,

make notes, arrange and classify the facts and truths thus obtained, draw

his own conclusions, and discover underlying principles and laws.

What is this method like when transferred to languages? It is precisely the encounter

with usage itself, as for example texts that stimulate the inductive discovery of the

“principles and laws” underlying.

11

It is remarkable that in Mr. Harper’s personal library, each item numbered, we

find item 563, Introduction to the teaching of living languages without grammar or

dictionary (1874), by one Lambert Sauveur (1826-1907), one of the Pestalozzians a

generation removed from the master who widely publicized the oral-aural methods of

inductive training, to which the inductive method of Harper for biblical Hebrew is

perhaps the closest correspondent, formed in the Whitneyan laboratory of Sanskrit

philologizing.

And, in the epistemological tradition of Bopp, a laboratory it was, as were also

the scholarly enterprises of all of Whitney’s European counterparts. The story to be told

here, thus far not clearly enough articulated, is the gradual emergence and “normalizing”

of the mode of inductive study of the Indo-European languages individually and as

members of a language family sparked by, and institutionally focusing upon, the fact of

autonomous phonological change, a.k.a. Lautgesetz. Lautgesetze had both an

epistemological and an ontological manifestation, not carefully enough distinguished

either in the instance or in the later historiography of the period. More importantly, the

most significant aspect of this story is that as comparative-historical linguistics morphed

into descriptive-structural linguistics, the breakthrough integrally depended on the

continued focus on the plane of phonology and on the continuity of the methods for its

study across such profound metatheoretical shifts. Given the retrospective sainthood of

Saussure, this is all the more ironic, in that the Ascended Master had nothing of interest

to say directly about phonology, a level of structure in language that is outside his

concern in theorizing the actual linguistic sign at the lexicogrammatical plane. Yet it was

only because of the continuity of inductive methods in the diachronic and then, later, the

12

synchronic study of sound systems that Saussurean structuralism could take hold in the

Euro-American world of linguistics.

So what gradually emerges in the context of German Sprachforschung and

becomes reflexively obvious to its practitioners around 1875-1880, is the mutual

necessity of the ontological assumption of uniformitarian regularity of autonomous

“sound change” and the epistemology of the method of residues, one of the chief modes

of inductive reasoning set out by all the philosophers of science of the time. Whether

operating in the realm of a single language or of multiple languages, one gathers textual

examples, and one segments and sifts them under a generalization that, if adequate to the

classification achieved, covers them as phenomena. The generalization for historians of

language is ultimately a statement of how a proto-form changes under sound laws

specific to each dialect into the multiply attested forms gathered and systematized as

correspondences. Note that by following this logic inexorably in the inductive method of

residues, exceptions at one level of generalization become special cases the exceptionless

regularity of which at their own level one sees when the exceptions, too, are segmented

and sifted as their own group. (Think of Grassmann’s Law of deaspirating the first of

two aspirate stops in syllabic sequence in Indo-European roots that explains certain

otherwise glaring exceptions to Grimm’s Law regularities of how Indo-European stop

types are realized in Germanic.) This is what came to acute reflexive consciousness in

Brugmann and Osthoff in the later 1870s.

In the second part of the book Language (1933), Leonard Bloomfield – himself

what we might term a neo-Neogrammarian – triumphantly, even misleadingly,

recapitulated the epistemological logic of induction involving the Grimm, Verner, and

13

Grassmann “laws” as an actual intellectual history of revelation of the true natural world

of phonological change. But of course, like all results gotten by inductive “method,”

what is involved in Bloomfield’s suggestive pseudo-history is really other. Bloomfield’s

account assimilates the real Grimm, Grassmann, and Verner, historical figures in the

development of linguistic scholarship, to the anachronistic terms of the scientific logic of

induction itself. Why? In his aspirations in this textbook to present linguistics as a real

science, Bloomfield loses the subtle distinction between methodology and method: the

methodological crierion for evidentiary sufficiency in inductive science demands of

hypotheses that they be reconstructible in principle as though reached by step-by-step

generalizations-over-data; induction as method does not require a report of how one

actually did go about – or should go about – discovering regularities. (Compare the

Chomskyite denunciations of neo-Bloomfieldian “discovery procedures,” which they

completely misunderstood, knowing nothing of the influence on Bloomfield of

Bridgmanian “operationalism” nor, indeed, of the formal continuity from inductive

historical linguistics to structuralism.) As Bloomfield had earlier formulated the lesson in

his “Postulates” of 1926, “phonemes change,” i.e., in the double articulation of language,

forms at the phonological plane change over time independent of what

grammaticosemantic forms they are exponents of (or “occur in,” in an inaccurate idiom).

Actually, it is allophones, classes of contextualized sounds-in-syntagmatic-sequences –

what Saussure termed phonèmes, actually, in the “Appendice” to part I of the Cours! –

that constitute the locus of “sound change,” though the implications for actual

phonological structure were finally worked out in the early 20th century.

14

As Anna Morpurgo Davies nicely points out (1994:135-36) in her account of

scholars and schools of 19th century linguistics, all the epistemological elements of

inductive reasoning about autonomous – and autonomously evolving – phonological form

are discernible in Bopp’s magnum opus, the Vergleichende Grammatik of 1833.

Thereafter, a discipline of linguistics emerges that is centered on the issues that

eventually became Neogrammarianism, even if sometimes contesting them. It is the

working through from this Boppian intellectual site to all its necessities and

compatibilities, to a richer and more algorithmic “normal scientific” formalism, to a

deeper theorization of the ontological model of the Stammbaum implied in it. This last

was especially so after Darwin’s Origins of Species (1859) brought inductive

classification as taxonomy together as the epistemically available evidence for

uniformitarian theories of the history of life, something the uniformitarianism of Lyell in

geology could not until then do. (Note Schleicher’s enthusiastic extremism in this

respect.)

Hence, the two important corollary interests of Neogrammarians and their critics

that emerge from the tremendous controversies about the ontological sense in which

Lautgesetze, “sound laws” posited to get from state A to state N of a form, could come to

be Ausnahmslos ‘exceptionless’. Studying living, especially “exotic” languages may

reveal the phonetic and other tendencies in pronunciation that we might understand as the

leading microdiachronic edge of inductively inferred changes that were posited to have

occurred in the undocumented, prehistoric past of the ancient languages. But how do you

– inductcively! – find the units of “sound” of exotic, nonstandard forms of Indo-European

languages, let alone those of languages of non-European peoples in small language

15

communities that have no standardized, orthographically fixed register? In the ancient

languages there are the apparent Buchstaben, the letters of the alphabetic or syllabary

graphic modes of inscription. Transcriptional work on several Asian languages like

Sanskrit had as well “reduced” them to alphabetic letters. But notice how this is

completely problematized for contemporary, “living” linguistic phenomena in their

spoken manifestation: can we find help in the methods of philology as it had been

constituted to be an inductive science?

Notice, then, what was at issue at the end of the 19th century. As an inductive

intellectual project, Neogrammarianism – and with it, 19th century linguistics, note – was

sent, respectively, to the field where spoken languages lived and to the natural science

laboratories of the psychophysicist and the physiologist. In short, the task of shoring up

the foundations of linguistic theory, notwithstanding its historical focus, fell to the field

worker and to the phonetician. Notice that in the history of linguistics, the first, the

empirical fieldwork track, led to descriptive and structural linguistics as we know it. As

for the second, refined micro-measurement, a pointillist intellectual cul-de-sac for

phonetics as for physics, was eventually reincorporated in linguistics when it was seen

through the perspective of the first. This breakthrough reimagination of what, in fact, are

the “sounds” of language played a central role in the birth of a new kind of inductivist

linguistics in which phonetics - in relation to phonology - can take its place.

Inductivist linguistics, in short, awaited Franz Boas and awaited the transition to

his intellectual progeny dealing with living languages who worked out the ontology and

epistemology of the phoneme, the foundational scientific entity of modern linguistics as a

science, as Bloomfield noted in1927. And not only the development via the substantive

16

problem of what “sounds” are, and how they work in language. When we examine the

work of these key figures, we can observe that across the transition of theoretical focus

from diachronic comparative philology to synchronic descriptive-structural linguistics

precisely the same methods carried over to create corresponding theoretical entities.

Scientific practices about etymological sounds become those concerning phonemic

segments; phonetic laws thus become (morpho)phonological rules; earlier vs. later forms

becomes “inner/organic” vs. “outer/inorganic”; etc. (Notice, please, the Haeckelesque

recapitulationism in all this in which synchronic derivation must, of necessity, resemble

historical provenance.)

The extraordinary irony in all this is that Saussure, who was a real

morphophonologist avant la lettre in his pathbreaking Mémoire of 1878, was not an

explicit theorist of phonology in the posthumous Cours, in which the hero of the story is

the grammaticosemantic significant/signifié correlation mediated by distributional

structure (valeur, or “valence,” as I prefer to translate it with proper Mendeleevan

allusion). Yet modern 20th century linguistics started with the application of the

inductivist techniques of analyzing valeur in precisely the realm of sound – ‘figurae’ of

phonic form, in Hjelmslev’s view, and then building up a view of grammar in its image.

But back to Boas. As historical studies by my colleague George Stocking and

others have shown, Boas was not trained in any real sense in anthropology; his university

background was in mathematics, psychophysics, and geography. His biographer Douglas

Cole (1999:51-57) makes clear the reasons why he attended an inferior university, Kiel,

for his 1881 doctorate. Here, however, he came under the influence of a philosophical

neo-Kantian and Helmholtz and Steinthal student, Benno Erdmann (1851-1921), just

17

seven years older than Boas, whose work, among other things, focused on

epistemological aspects of logic and inductive science. This affiliation is central, it

seems to me, in Boas's turn from physics to psychophysics to his key epistemological

position on what we might term, with him – echoing Alexander von Humboldt – the

“cosmographic” nature of culture, phenomena of mind in sociohistorical context.

Retrospectively, we see that this cultural psychology was initially articulated in a paper

on “Alternating Sounds” (1889), aimed directly at the work of, and linguistic views of

Major John Wesley Powell, head of the Smithsonian Bureau of American Ethnology.

The critical problem he takes up is how to insure the accurate, inductive scientific

discernment of the sound system – and hence, the proper means of transcribing and

spelling – of languages such as the exotic-sounding ones of Arctic and Northwest Coast

North America, which Boas had already begun to work on. In an evolutionary view of

the universal history of the human mind, Powell and others took their own reflexive

consciousness of Indo-European sounds, as written alphabetically, as the endpoint of

development of clear categories. They reported hearing the sound categories of other

languages as "fluctuating" or "alternating" across their own, or as "compounded" of two

or more of them, seeing this as consistent with the putative lack-of-abstraction and lack-

of-fixity of so-called primitive thought. As everyone from Stocking (1968; 1996) to

Rulon Wells (1974) to Michael Mackert (1993, 1994) and so forth have recognized the

solution Boas proposes. He proposes that what we now term a phonological system of

any perceiver’s – and would-be transcriber’s – language is a biasing filter on the

perception of proper intensional category boundaries that classify the sound stream one

hears. Such a system of categories is in essence a structure of (Herbartian)

18

apperceptional norms relative to which a speaker has been trained to hear not only his or

her language, but other languages as well.

And as goes phonology, so also categories coded in grammaticosemantic forms of

the higher, meaning-bearing planes of language. You can see the epistemological – and

ultimately ontological – revolution that this implies for linguistics as an inductive

science. Suddenly, we know that mere phonetic form, a physical fact, is not the proper

framework for the determination of what in language count as the “sounds,” just as the

most accurate scientific descriptive appreciation of phenomena of experience is not the

framework for determining grammatical categories and their meanings. A stunning

analogical – inductive – transfer, this, from Boas’s cumulated research experience and

training in his fields of study – not, however, from philological training as such, other

than perhaps having heard some lectures of Steinthal.

Yet Boasian practice is precisely the philological inductivism that we have

already noted in both pedagogical and research realms. We have only to look at

examples of the Boasian published corpus to realize that his anthropological linguistics is

a philology of the unlettered, an inductive science of the times. In the Handbook of

American Indian Languages [Image 14], first appeared in 1911, each grammatical

sketch, done up in numbered sections, is followed by a text – here [Image 15] is the

sample Kathlamet one from Boas’s own Chinook sketch. Observe the apparatus: an

interlinear translation of a word-by-word closeness, fluent translation reserved for

separate printing; superscript-indexed notes to the text, giving the analysis of the

elements of each word as it occurs in the text, together with explanations keyed to the

general statements that apply from the sections of the grammar. Haven’t we seen this

19

before in the teaching materials of Harper in relation to the philological inductivism of

Whitney, Grassmann, and others? But where did Boas learn this? Surely he did not

invent, or reinvent, philological inductivism himself.

He might well have read various works in the realm of philology; for example, he

makes reference to Richard Lepsius in some of his explanations of his use of phonetic

signs, and Steinthal is referred to in later writings. But it is the specific philological

inductivism common to all serious Indo-European and nearby linguistic research that is

so startling in this autodidact. Here are some considerations to pinpoint the historical

connections.

First, that Boas was committed, almost religiously, to inductive science is

something that appears in his youthful letters to family and friends, and is part of the

multi-generational commentaries of all of his students and their interlocutors. There is no

question but that the patient gathering of facts of whatever sort – in language, of course,

spoken word forms and texts – must, for Boas, be the starting point for all research.

Hence his publication of hundreds of pages of recipes in Kwakiutl, that critics point to as

a sign of a disordered ethnographic sensibility. Boas truly believed as well that the

anthropologist (“ethnologist”) is the culture historian on behalf of peoples without such a

scholarly tradition in the Western sense. The foundation of culture history is comprised

of great bodies of texts and other art forms, precisely as the 19th century philologists of

the great, writing-based cultures thought. When we look at the earliest notebooks thus far

available to me, for Lower Chinook and Kathlamet in 1890, the layout is already

designed for this philological linguistic intervention: facing-pages of left-hand side text-

with-interlinear and right-hand side keyed notes with contrastive vocabulary of semantic

20

sets, paradigmatic and derivational material, etc. that pinpoint each form at that point

needing elaboration for the researcher. It is the page of the printed Kathlamet text I

showed before, instead of upper and lower halves, on facing sides. Observe that his best

and most linguistically useful volumes of texts have precisely the form of text-with-

interlinear and fluent or running English translation nearby. Here are two versions.

[Image 16] Title page of Kathlamet Texts (BBAE 26)

[Image 17] Printed page 39 of KT

[Image 18] Printed page 220 of KT

[Image 19] Title page of Chinook Texts (BBAE 20)

[Image 20] Printed pages 92-93 of CT

[Image 21] Printed pages 244-45 of CT

A second consideration is this. Boas may have gotten his specifically linguistic

methodological inductivism from inductive study of Latin and Greek languages in his

youth at the Minden Gymnasium, a school then highly influenced by Pestalozzian

pedagogical technique through its late former Director, Siegmund Imanuel (1792-1847).

This remains to be explored in collections in situ and perhaps some Boas juvenilia as yet

unexamined.

A third and more intriguing idea is that in his earliest fieldwork, with the language

Inuktitut of The Central Eskimo (Boas 1886) in Baffinland in 1883-84, we know that his

publications evolve under the influence of his collaborator, Hinrich Rink (1819-1893).

Rink was a colonial official and expert on Greenland and amateur linguist of Greenlandic

and dialectology, who put Boas’s early published texts into canonical philological shape

with interlinear, linguistic and dialectological explanatory notes, etc., Boas supplying

21

cultural glosses on the significance of the texts [Image 22, Image 23]. These pages

come from Journal of American Folklore published in 1889. By 1894, note, Boas

himself, without Rink, was republishing texts [Image 24] he had previously included in

his Bureau of American Ethnology report of 1886 with essentially identically formatted

translation, annotation, etc. This is not to say that Rink had a magic method; he was

merely doing what was being done by all European ethnological philologists, even

missionaries like the great Samuel Kleinschmidt, whose Greenlandic grammar Boas, too,

used.

My point, then, is that in matters linguistic, Boas the empirical, inductive scientist

of language moved into the orbit of philologists of texts, where the generalizations of a

concordance are essentially the next step of generalization over these texts. The

grammaticosemantic aspect of concordancing is the grammar; the lexicosemantic and

lexicopragmatic aspect of concordancing is the text-apparatus and dictionary or lexicon.

Documenting a language inductively starts from the texts, and anchors all rearrangements

of the data, including generalizations over it, to them. Methodologically, how language

is observed, analyzed, and presented in its synchronic aspect is precisely continuous with

its treatment in its diachronic aspects. Theoretical frameworks of explanation change;

method continues. This is obvious in the field notebooks of Boas and of all his real and

fancied students. It is the method of Sapir, of Bloomfield – interestingly, who got it as

much, he claimed after his Tagalog fieldwork but before his Algonquian, from his teacher

Jacob Wagernagel (1853-1938) as from Boas. And it is the method of every one of their

students whose notebooks I have examined.

22

Observe these images of some pages of Boas’s Chinook and Kathlamet

notebooks, courtesy the Library of the American Philosophical Society. The first [Image

25] shows a whole set of numbered micro-elicitations that give paradigmatic and

semantically related lexical material. The next [Image 26] gives a good view of a verso

and recto of text dictation in Kathlamet in the 1894 field season; note how few words are

glossed in the notebook version, though the earlier image of the published text we saw

[Image 17] is fully glossed interlinearly as well as in running English. (Boas had

inductively mastered much of the language by this time.)

Precisely the same method is revealed in Sapir’s notebooks, with halting

consistency at first in his maiden-flight Wishram notebooks of the summer of 1905

[Image 27], and with a masterly oderliness in his1914 Nootka notebooks, here, a

dictation on 3-4 January of that year, written out by Sapir with few corrections in black

ink, not pencil [Image 28] on the recto side of the fold, with superscript-indexed

clarifying notes on the verso side. The physical medium through which language was

investigated, as well as learned, is continuous across all theoretical revolutions, across all

schools of inductive science.

By mid-nineteenth-century William Dwight Whitney delighted in the fact that

linguistics – comparative philology – had become an inductive science. Boas was

brought to it from physical and psychophysical science in precisely this spirit as well.

Bloomfield held rigorously to this view, in fact setting the methodological inductivism of

how linguists work into an epistemological framework of Bridgmanian operationalism

and a psychological commitment to behaviorism. (Yet, of course, Bloomfield continued

23

to declare as irrelevant to linguistics as a science all psychological and philosophical

commitments of this type!)

Bloomfieldian inductivism, we might note, came full circle in the period of the

second World War, when the inductive method of teaching living languages for the

Allied war effort was the applied recompense of the pure science of linguistics. The

personnel of 165 Broadway, all trained in inductivist linguistics in the anthropological

manner, even if working on languages with long graphic traditions and literatures, were

the orchestrators of Pestalozzian courses of presentation, imitation, memorization, and –

one hoped – inductive generalization, guided by the prior linguistic analysis of the trained

Bloomfieldian analyst of distributional form.

It was this long and continuous history that Chomsky and others misread and

attacked, causing as much acrimoniousness as had been caused by the Neogrammarian

assertion of inductivist principle and as had been by Bloomfieldians’ earlier erroneous

dismissal of philology. (Those who live by the sword…) Yet, inductivism was indeed

what made a science of linguistics, and in many quarters it is what will keep it that way –

even inductivism over formal models of phenomena in languages (when linguists looks

for “the best generalization” over data). The computer has brought a new power to the

methods of text-based philology, able to gather, i.e., concordance, vast amounts of text,

whether in digital audio files or digital graphics. Contemporary variationist

sociolinguistics as much as so-called corpus linguistics depend on such examination of

language material as data for possible generatlizations – even where those generalizations

take the form of abstract rules, of conditions on tree structures, or similar formal

24

proposals. We might say that linguistics or linguists have never abandoned inductivism

in the way we handle language material.

The great historical puzzle, then, is why so many linguists have apparently

thought that in the last 40 or so years we have. And with what institutional consequences

for the place of linguistics among the sciences?

25