Franz Boas's Introduction to the Handbook of American Indian Languages

by Suzanne Kemmer

Background on 19th century linguistic typology

In 1836, Wilhelm von Humboldt laid out an influential view of the "essential types" of language, which is a division into structural types based on what we now call morphological typology, i.e., the typical morphological structure of the word in a language: for example, do words in a language usually consist of a single morpheme or multiple morphemes?

Until the twentieth century, this was effectively the only type of typology of language, that is, if we define language typology as a division of languages into structural types (with or without a semantic aspect to the types). Typology, then, can be viewed as language classification based on aspects of linguistic structure. The other major type of classification in the 19th century was of course genetic classification, based on hypotheses about historical relationships of cultural transmission through the generations and not on cultural borrowing. Both modes of classification, genetic and typological, were seen as important, but logically distinct as they are based on different criteria. (Sometimes the two classifications can show empirical overlap, however. Genetically related languages can be typologically similar if they have not diverged structurally in any significant way through areal contact with typologically different neighboring languages.)

Morphological typology is based on word structure, i.e. the typical degree of complexity of words in a language compared to other languages. Why was morphological typology seen as so important? Why not, for example, a division based on what kind of passive structure a language had, or what kind of syllable structure?

There were different reasons for the focus on word structure, but one was, I think, that it is an instantly recognizable difference once you look at the languages through a type of writing system with spaces between words. You can actually see the difference in word sizes on paper if the languages are written like western languages. Some languages have really long words, and the length comes from morphological complexity; while others have really short words, which are morphologically simple.

A second reason is that morphological typology seems to divide languages into types on a global or "entire-language" basis. Every language has words and morphemes and there has to be some relation between morphemes and words. Morphological typology is interesting because it is not a necessary fact that many or most words in a language should correlate with each other roughly in terms of degree of complexity. It is an observable empirical fact that a language as a whole had a particular and characteristic kind of structure permeating all its utterances. Such a fact seems to beg for an explanation. (We still don't really have one today, other than a suggestion that something about the way humans process languages limits the possibilities for complexity of word structure to a certain range, much as our auditory range limits the pitches of the sounds found in human languages).

Add to this noticeability a folk model in which the structure of language maps directly to the structure of thought (also a model held by grammarians of the 17th-18th centuries), and that magnifies differences among the languages even more.

Another reason for the intensive focus on morphological types was the prevalent belief that the differences in word complexity among languages are indicative of complexity of non-linguistic aspects of their speaker populations, such as cultural complexity and/or degree of complexity of thought. Some of these beliefs are described in the next section.

Von Humboldt had some other very specific philosophical reasons why it was important, which I won't go into.

Although boundaries between the morphological types are hard to draw no matter what criteria you use (since linguistic structures fall on continua rather than in sharply separate categories), there are still prototypical examples that show up the differences clearly. Chinese approximates one morpheme per word, so it is isolating; Turkish is agglutinating because it has words consisting of a root plus at least one affix and often many separate affixes; Greenlandic Eskimo is polysynthetic because it has a very large number of morphemes put together into a word, so many, in fact, that utterances that express fully specified events can easily consist of just one word, with one or more roots modified by a number of affixes. The occurrence of complete sentences expressed in one long word is facilitated by the fact that Greenlandic is an incorporating language, that is, the verb can incorporate lexical objects. The result is the occurrence in many instances of more than one root in a single word. English has multi-root words as what we call compounds--but these do not generally make up a whole utterance; they usually serve as just one part of speech in a sentence, such as a noun or a verb.

Earlier beliefs in correlation between morphological type and mental/cultural type

The view about morphological typology that was developed by the mid-19th century was: the more complex the morphology (i.e. the structure of the word), the higher on an imagined scale of culture and civilization are its speakers.

In the West, Latin and Greek were considered the greatest civilizations there had ever been; and since their languages had lots of morphology, everyone assumed that this was something that went along with high civilization. The fact that modern European languages had changed in the direction of loss of morphology and therefore more analytic/isolating form was seen as correlating with a general cultural degeneration since classical times, as seen, at least, by humanities specialists. Those more attuned to technological achievement, in contrast, saw 19th century culture and civilization as progressing nicely, although they did not usually make claims about progress in language. A late exception was Otto Jespersen who saw the sloughing off of morphology to be a sign of cultural vigor and "masculinity" that went along with the industrial revolution.

However, attempts at correlation of word form and culture never really provided a coherent story. Chinese was morphologically simple, but Chinese culture and civilization were recognized as advanced, because of agriculture, social organization, and level of technology. (Europe began to gain a technological advantage over Chinese society only in the 17th century and an educational advantage - mass literacy and mass education - as late as the late 19th-early 20th century). Eskimo languages were highly agglutinating, but their speakers were seen as low on the civilization hierarchy because they had no agriculture and no metal technology. Similarly for American Indian languages and African languages, some of which had plenty of complex morphology, but whose speakers in some cases had stone age technology (or bronze age in the case of Africa).

The "cycle" of morphological type change

Von Humboldt suggested that languages go through a "cycle of agglutination": isolating languages become agglutinating languages, and these become synthetic or polysynthetic, and then the structure breaks down and these become isolating again. This theory captured some correct observations about grammaticalization processes; lexical items become more grammatical, and part of that process involves losing phonetic substance and morphological independence. Hence, former word boundaries become morpheme boundaries. But these processes happen regardless of whether a language changes type; the changes seem to be pegged to the development of specific units rather than a whole language (but there are still open questions about the relation between grammaticalization and change of morphological type).

Unsolved problems in 19th century morphological typology

Quite apart from the problems in correlating structural type with cultural factors, scholars also had some trouble classifying languages as to morphological type from a purely structural standpoint. It turned out they were conflating two different (but partially empirically overlapping) criteria: the average number of morphemes per word (the lower the number, the more isolating the language), and the degree to which the morphemes were clearly divisible. If a language generally had multiple morphemes in its words, it could have either morphemes with clear boundaries, like Turkish, Bantu languages, etc., or "fused together" morphemes such as those in the Indo-European morphological systems (e.g. Latin dative plural endings, in which the endings are not clearly divisible into a dative part and a plural part.) It took until the 20th century for linguists to figure out how to separate the two criteria (first Sapir, then Greenberg, who refined and quantified Sapir's ideas.)

Background on assumptions on race, culture, and language

Ordinary people often did not (and some still do not) distinguish between physical type (physical characteristics determined by genes), culture, and language. It was thought that people of a given race or ethnic group "most naturally" speak a particular language or languages of a particular family; or have a "natural affinity" to a particular culture, which is seen to persist regardless of environment. For example, new immigrants to the United States in the late 19th and early 20th century were seen as groups defined by ethnicity and culture (and language) indiscriminately. It was thought that southern and eastern European immigrants would never be able to assimilate to American culture because they were a) ethnically distinct and b) their culture and language would remain pegged to their ethnic group rather than changing in response to the surrounding American culture.

Similar fears are found today with respect to other more recent groups of immigrants. People who consider themselves "non-immigrants" (even though their own ancestors were immigrants, some from not that long ago) tend to severely underestimate the speed with which cultures and languages change: the Italians and Jews who were so worrisome to early 20th century Americans because they looked, sounded, and acted so foreign have lost their ancestral languages and keep only a much watered-down remainder of their "old country" cultural customs and beliefs.

Probably because people reason by prototypes (or stereotypes, which is the more traditional term for prototypes based on categorization of human groups), they fail to notice cases where language, race, and culture do not line up and in which they fail to divide people into fairly neat groups with 3 sets of correlating characteristics. Some people think for example that because two populations live near each other, e.g. in West Africa, they must speak the same language or languages from related families, they must have the same culture, and must be of the same physical type.

Boas's arguments for distinguishing the three concepts logically and empirically

It is notable that even people who were supposed to be scientists observing people objectively (a new idea in the 19th century) arrived at classifications of people derived from a mixture of racial, cultural, and linguistic criteria. Boas pointed out that this is why there were so many different classifications proposed.

Boas made the point that any given classification you come up with for humans that mixes the criteria is arbitrary, rather than a "real world" set of pre-existing categories.

Different categorizations may be appropriate for different social purposes, Boas argued, but the only appropriate "scientific" classification was one which focused solely on physical type (race) without conflating the confounding factors of language and culture; and the aim of such a classification was simply to reconstruct the distant past history of humanity: the current gross division into 3 main races and several subgroups reflects the earlier unity and later breakup of successive groups of people through history.

In the earlier section of this chapter Boas goes through many examples of cases in which groups share one of the three features but not the others. The upshot is that knowing a group's racial type, you cannot predict its culture or its language/language family; knowing its cultural properties or affinities, you can't predict the racial type of its speakers or the language family of the language they speak; and knowing the language family in which the group's language is classified doesn't let you predict anything about either their race OR their culture.

For example, Hungarians are culturally central European, but their language is Finno-Ugric, related to Finnish and Lapp, and not to the languages around them. Many African Americans are (or were then, anyway) racially still relatively close to the populations of Africa they were pulled from, but culturally, Boas stated, they are more like southern American whites than like Africans, and their language is of the Germanic family (since they are English speakers).

A side note: the northward migration of blacks since Boas's time has in the meantime effected an increased divergence between African-American culture and language compared to those of southern whites--in fact to whites in general. (See for example Labov's studies of African American vs. non African-American varieties of English, and their increasing divergence.)

The culture (and language) of southern American Blacks was already probably more different from southern white culture and language than Boas realized -- or wanted to emphasize. The usual view at that time, entrenched as standard belief until a generation or so ago, was that black people were culturally and linguistically inferior to whites (characteristics seen as going along with their supposed racial inferiority). So it was probably shocking to many people that Boas placed southern Blacks together with southern whites, culturally and linguistically. He maintained (and modern genetic scientists agree with him in this respect) that the only difference between the two populations was some accidental and ancient differences arising some extremely slight genetic variations in the two populations that control superficial physical traits such as skin color, shape of a few facial features, hair color and type, and a few differences in the population distribution of bone and body types. Any such differences are from a scientific standpoint trivial, and outweighed by a huge array of cultural, including linguistic commonalities due to living in contact with one another on this side of the Atlantic. Differences in behavior of the groups, to the extent that they really are observable and not just part of the cultural view of the opposite group, are due to cultural factors rather than any genetic factors.

This view accords with that of modern cultural anthropologists, who have come to understand racial categories as largely socially constructed rather than physical categories.

Contributing to the blurring of populations, we can add the fact that there have now been 300-400 years of significant genetic admixture in these populations in the Americas, which has blurred the physical distinctness of the populations in the western hemisphere by introducing some genes originally associated with one population into the other.

[ jump to top ]