William Diver

The Columbia School of Linguistics started in the mid-1960’s at Columbia University under the leadership of Professor William Diver.  He and his students formulated a comprehensive framework for linguistic analysis, early on called “Form-Content”, whose tenets contrast sharply with those of generative grammar, which emerged around the same time.  Our philosophy has evolved considerably over the past 30 years, and continues its development in many parts of the world today.

Rather than formalizing traditional sentence grammar as a set of autonomous syntactic rules, Diver sought to understand uses of linguistic mechanisms in actual discourse.  Why do verbs appear to “govern” cases in languages like Latin and Greek?  Why did Greek authors use sometimes the dual, sometimes the plural when talking about two of a thing?  What explains seemingly whimsical occurrences of subjunctive, passive or middle morphology in texts?  Why do “reflexive” pronouns have non-reflexive uses?  Why do some verbs (“deponents”) have a passive form but an active sense?  Why are particular sounds or combinations of sounds favored in individual languages or across languages, or disfavored, sometimes to the point of non-occurrence?

Diver demonstrated the inappropriateness of traditional categories of grammar to analysis of language.  The Sentence and its parts - subject, predicate, direct/indirect object, clause, phrase, etc. - as well as the parts of speech stem from the classical interest in logic, and represent an analysis of the structure and content of thought.  Traditional grammar, Diver realized, began as an attempt to correlate linguistic structure with this logical structure.  For example, in the traditional account of the Greek and Latin nominative, accusative, and dative cases, there was an attempt to explain the occurrence of these morphologies in terms of categories of the structure of thought: the nominative is the case of the subject, the accusative is the case of the direct object, and the dative is the case of the indirect object.

When, as frequently happens, direct objects turn up in the dative, and predicates in the nominative, this attempt to correlate the two structures has failed empirically.  However, rather than abandon the enterprise, traditional grammars set up what Diver regarded as mere escape clauses: "government of the dative", "predicate nominative".  Through such maneuvers, the theoretically unmotivated part of traditional grammars came to dwarf the theoretically motivated part, leading ultimately to a picture of language as a collection of arbitrary devices, a type of human behavior not comparable to other, more readily understandable types of behavior.

Thus, Diver regarded the categories of syntax simply as artifacts of an unsuccessful attempt to explain linguistic phenomena in terms of the logic-derived parts of the sentence and parts of speech, not as a revelation of some unique human cognitive process.  He took issue with generative grammar on the grounds that, rather than recognizing these categories as a consequence of analytical failure, generative grammar bought heavily into the traditional scheme and went on to build up a school of analysis which took it for granted, thereby developing a view of language as having an important component of arbitrary relations of the "government" type.  This was an unjustified conclusion, Diver said, because the assumptions about linguistic categories lying at its very base were faulty.

In Diver's view, the task of grammatical analysis is not to seek manifestations of universal categories in languages, but to discover the unique categories articulated by each language.  Here, his position was similar to the anti-nomenclaturist view propounded by Saussure.  Diver wanted to explain the outward face of language, what we actually observe, ultimately, the shape of the sound waves of speech.  Diverian grammatical analyses focus on occurrences of forms in texts and discourse, the distribution of forms being regarded as the best overt clues to underlying categories of language.  Morphs and morphemes are examined as potential bearers of linguistic meaning, so that grammatical hypotheses very often take the form of signals and meanings.

Diver articulated an innovative view of grammatical meaning, which has come to be called an 'instrumental' view of meaning, in contrast to the traditional compositional view.  In the compositional view, everything in a linguistically communicated message is attributed to some element of linguistic input, and a direct mapping between input and output is required.  Diver recognized that communicative output can often be traced not to the form with which compositional analysis associates it, but rather to some other element of linguistic or extralinguistic context.  A compositional analysis may build into the meaning of a form all sorts of communicative effects for which that form is actually not responsible at all.

The instrumental view, in contrast, recognizes that not everything communicated with language is encoded linguistically; that people use their inferential powers to jump to conclusions on the basis of a relatively small amount of actually encoded linguistic information.  Diver thus saw the effects of human intelligence as pervasive in the functioning of language, and speakers' use of linguistic meanings as comparable to other kinds of human tool use.  This led to the distinguishing of two different kinds or levels of function:  the meaning of a form - that sparse element which the form encodes and consistently contributes to the communicative process, and the message - the totality of communicative effects which may at one time or another be associated with the occurrence of a form, but which is actually the resultant of human inference operating with many different kinds of input, both linguistic and non-linguistic.

Grammatical analysis thus becomes a search for that which languages actually encode, these sparse, hint-like meanings.  In this enterprise, then, meaning is not something studied in the abstract, without reference to a particular language; meaning is rather a device of explanation, invoked to account for facts of morphemic distribution in individual languages.  In Diver's own words:

The general picture of human language is that of a particular kind of instrument of communication, an imprecise code by means of which precise messages can be transmitted through the exercise of human ingenuity.  The code and the ingenuity must be kept clearly separate; most of the difficulties encountered in the various schools of linguistic analysis result, simply, from the attempt to build the ingenuity into the structure of language itself.

Diver liked to give his own twist to the well-known analogy of Sapir, saying:  "Language is only powerful enough to run a light bulb; but we use it to run an elevator."

By pursuing this view of language as being driven by meaning and by ordinary human behavioral and perceptual characteristics, Diver and his students were able to develop explanations not only for those parts of language which have traditionally been regarded as basically semantic--verb tenses, demonstratives, aspect, etc.-- but also for those that have always been seen as lying within the central core of syntax, such as government, concord, and ordering phenomena.  He took particular issue with the attempts of descriptivism and generativism to see language as having an autonomous structure that can be described algorithmically.  Analyses of a great variety of languages have been carried out in the framework Diver innovated.

Thus, for Latin and Greek, discarding notions of sentence structure and syntactic government, and taking occurrences of case morphology themselves as the data to be explained, Diver found that these cases, for one thing, have to do with communicating the degree of control exercised by participants over events.  Similarly, he analyzed subjunctive morphologies in these languages as indicating particular levels of the probability of occurrence of the event denoted by their attached lexical item; other forms turned out to have to do with attracting greater or lesser degrees of attention to an associated item. He posited that word order in English can function as the signal of a meaning, like the morphological signals of Latin and Greek.  Moreover, he and his students discovered that the meanings attached to these signals often organize themselves into closed systems in which the meanings exhaustively divide up a semantic substance.  So, for instance, the Latin cases denote relative degrees of control over an event, in the order (from highest to lowest) nominative, ablative, dative, accusative.

In phonology, Diver was concerned with explaining the shape of the sound wave of speech below the level of the signal, that is, the nonrandom distribution of distinctive units of sound within a language's lexical and grammatical morphemes.  He accounted for these skewings in part by appealing to facts of articulatory and acoustic phonetics, some of which had gone neglected in previous phonological research, which, indeed has minimized the role of phonetics to begin with.  But the theoretical significance of Diverian phonology is more profound, for complete explanation of this non-randomness has required an appeal to the same principles of communication and human behavior which underlie grammar, two external orientations that had not previously played so explicit a role in phonological theory.  Diver, then, proposed a non-autonomous phonology, just as he proposed a non-autonomous, non-modular grammar.

The communicative factor requires speakers to maintain distinctions among sounds; yet speakers show a tendency, here as in other aspects of human behavior, to economize effort.  Diverian phonology, as it studies both the frequencies of phonological units and the ways in which they combine, gives evidence of the dynamic interplay of these competing pressures.

The following is one of Diver's examples.  It is well known that in many languages, such as German and Russian, final stops are voiceless.  In English, although the skewing is not absolute, voiceless stops in word-final position heavily outweigh voiced stops.  English is thus merely a less extreme example of what is found in German and Russian; the difference of a few percentage points is not important since one explanation covers both situations.  Diver proposed that it is the task of coordinating two active articulators (the tongue or lips which create the stop, and the vocal folds which provide voicing) that accounts for the lower frequency of voiced stops as compared to voiceless, where only one of these articulators has to be controlled.  However, the beginning of the word, where the hearer does not yet know the identity of the word, bears a greater communicative burden than the end of the word, which the speaker will likely be able to figure out for himself once it is reached (cf. people's tendency to chime in at the ends of words).  This example shows the interplay of factors: the communicative factor motivates the greater distinctiveness afforded by having both voiced and unvoiced stops; but the human factor--ordinary laziness--carries the day when one can get away with less distinctiveness.

In following this route, Diver developed an epistemology intended to bring the practice of linguistics into line with that of other attempts to understand natural phenomena in the scientific era.  For Diver, explanation was not a matter of simply demonstrating that a particular item is a member of a more general class; he wanted to get at the "Why" of things.  This meant seeking motivations for observations one does not understand in terms of areas of knowledge one does understand, not embarking on a speculative program.  It meant adhering to highly demanding standards of validation and fit between hypothesis and data.  Diverian analyses are heavily textually oriented; large quantities of data from actual texts and extensive use of counts are their hallmarks.  Diver was very skeptical of a-prioristic schemes, such as universal grammar.  He insisted that theory be always guided by analysis, not the other way around, no matter how unfamiliar the resulting theory might appear.

This approach is of course quite the opposite of contemporary mainstream linguistic thought, and thus did not get much press.  Diver was a man far ahead of his time; in a world obsessed with modularity and syntax, it is a rare voice which asserts that language is an instance of ordinary human behavior, and that linguistic structure can and must be understood without reference to syntax. Nonetheless, a great many in-depth analyses of a wide range of languages by Diver and his students have borne him out, and the scholarly mechanism he established has quietly pursued its work, with little public fanfare.

From 1975, Diver edited the Columbia University Working Papers in Linguistics, in which many of his own writings appeared.  He gave invited lecture series in numerous countries of Europe and Asia, in addition to speaking at conferences in the USA and Canada.  The Columbia School has held biennial international conferences since 1989 at Columbia, the University of Virginia and Rutgers, and a Summer Institute of the Columbia School was held at the City College of New York in 1996. Even after his retirement to Emeritus status in 1989, Diver remained an active participant in an ongoing Linguistics Seminar at Columbia and in the Conferences, giving generously of his time, and continuing to attract new adherents through his writings and lectures.

William Diver perished on August 31, 1995 while sailing solo in Nantucket Harbor.   His relations with other people were characterized by unlimited generosity, tolerance, and gentlemanliness.  The role model he exemplified made as profound and lasting an impression as did his ideas.  Few scholars have evoked such heartfelt grief at their passing as has this extraordinary man.

Linguists who received the Ph.D. at Columbia under Diver's advisorship include Erica Garcia, Robert Kirsner, Flora Klein, David Zubin, Wallis Reid, Abdul Azim, John Penhallurick, Robert Leonard, Ellen Contini-Morava, Anita de la Garza, Alan Huffman, Bonny Gildin, Radmila Gorup, Barbara Goldberg, and Joseph Davis.  A dissertation written under Diver's sponsorship received the Edward Sapir Award in Linguistics from the New York Academy of Sciences in 1985.  Books presenting Columbia School analyses or discussing Columbia School ideas have been written by Garcia, Kirsner, Reid, Contini-Morava, Gorup, Huffman, Zubin, and Yishai Tobin.