(Originally appeared in Lognet 97/2)

The article below is an extended abstract of a talk given by Prof. Kevin Miller at a February 1996 Carnegie Mellon Psychology Department colloquium. Prof. Miller’s permission to reprint it here was obtained for us by James Salsman, the logli who first learned of it and who decided that the Loglan community should know of these interesting results.

A “Human Factors” Approach to Language and Cognition

By Kevin Miller
Department of Psychology and the Beckman Institute
University of Illinois at Urbana-Champaign

Research on the effects of language on cognition has generally been organized around a framework, the Sapir-Whorf hypothesis, that reflects archaic views of both language and cognition. This talk will present a newer framework for thinking about how symbolic systems may affect cognitive development, which will be illustrated in terms of research on the cognitive development of Chinese and American children.

Symbol systems (such as number words, calendars, orthographies, etc.) can affect cognition in three main ways. First, the organization of such a system may impede or facilitate its acquisition. Second, the structure of a symbol system may have consequences for its use by skilled users’ in performing cognitive tasks that require use of the symbols. Finally, the organization of symbols may affect the conceptual understanding users have of the domain the symbols represent. Only the final type of symbolic structure effect falls within the scope of the Sapir-Whorf hypothesis, yet I will argue that effects of the first two types are both more pervasive and more clear-cut.

The Symbolic Structure Effects framework will be used to describe U.S./Chinese cross-cultural research on symbolic development in children’s mastery of numbers, calendar systems, and orthographies.

A clear example of the effects of symbol structure on initial acquisition comes from research on children’s early counting and later learning of Arabic numerals. Both Chinese and English number words [the author uses the, to logli, misleading word names here; I have taken the liberty of replacing his word name with the word word throughout] up to ten consist of a list of words that must be learned by rote (i.e., there’s no way to derive six from five or liu from wu). [There are, of course, just such ways of deriving Loglan digit-words from their 5 prefixed (n- t- f- s- v-) and 3 postfixed (-i -e -o) elements.] After ten, the systems diverge. Chinese has a clear base-ten structure of words consistently used, whereas English number words in the second (“teens”) decade follow a more complex structure used only for numbers in this range. After twenty, both systems of words converge on a roughly base-ten word-formation scheme. These differences in structure are reflected in children’s difficulty in learning to count in Chinese and English, predicting when differences emerge and the kinds of mistakes children make. They also predict the kinds of errors children make when first learning to write Arabic numerals (which are a strict base-ten system).

Even after children acquire symbol systems, the structure of those systems can continue to affect cognitive performance. As an example of this, calendar naming systems in Chinese and English will be compared. Chinese uses a basically numerical system for naming days of the weeks and months of the year [as does Loglan]. These differences are reflected in dramatic differences in strategy and speed by both children and adults in tasks where participants were required to perform basic calendar calculations (e.g., Which month comes seven months after February?). The Chinese system [and the Loglan one] lends itself to arithmetical calculation; American subjects tend not to use such algorithms.

Finally, the structure of symbol systems may serve to make certain concepts more or less accessible. Comparisons between Chinese and English in the ability to identify and manipulate linguistic units [have been made]. Comparisons between the languages and orthographies support the following predictions: Chinese children should find it easier than Americans to identify and manipulate syllables, but harder to identify both smaller (phoneme-level) and larger (word-level) units. Even Chinese college students show considerable disagreement among themselves in identifying words. These results indicate that naive conceptions of linguistic units such as phonemes, syllables, and words are at least partially artifacts of [the] symbolic systems that [we] may choose to represent language in terms of these units.

Symbolic systems are one of the most basic resources that support cognition, as well as one that is relatively amenable to change. Understanding the contribution that symbolic structure makes to cognitive development is a matter of both practical and theoretical importance, but it requires a framework broad enough to serve as a foundation for spanning the gamut of symbol system effects.


All the structural causes attributed by Prof. Miller to the “symbolic system”, and ably argued by him to have cognitive effects, are, on my own 40-year-old and distinctly wider interpretation of the Sapir-Whorf hypothesis—I guess that that qualifies it as “antique”, soi crano!—definitely “Whorfian” ones. Any structural feature of language is—on the interpretation of S-W that has guided the design and engineering of Loglan over these last 40 years—grist for the Whorfian mill. Thus, Loglan’s number system, with its systematic use of CV-form digit words, /e/-final words for the 5 odd digits, /o/-final ones for the even, non-zero digits, its use of a distinctive vowel, /i/, to mark the recycling of the system after each 10 integers, and its grouping of the 10 digit words into 5 pairs of serially adjacent words, each pair characterized by a distinguishing consonant (/n t f s v/) ...all this bespeaks an obvious effort on the part of the designer to build both learnability and manipulability into the “symbolic structure” of the Loglan number system. To Miller, however, these are all “non-Whorfian” ways in which language can affect the cognitive system! Clearly the design effort that went into the Loglan number system presumes, as do hundreds of other features of the language, an hypothetical “symbolic structure effect” that is at least as powerful as any of those mentioned by Miller in this talk. I wonder if this modern cognitive psychologist could put aside his evident contempt—Is his use of the word archaic contemptuous? I think it is—for Whorf’s hypothesis long enough to take a look at a total “symbol system”, namely Loglan, the design of which has been guided from the beginning and in all particulars by a much broader scientific interpretation of the S-W hypothesis (where by scientific I mean a non-metaphysical one) than he seems willing to accord to it. Or at least mine is a broader view of Whorf’s thinking than is evidently useful for Miller’s own scholarly purposes...which include, I suppose, a desire to detach himself from the past. Someone should send him a copy of L1 and this issue of Lognet. (It will not be me. I have neither taste nor talent for polemics.) But it is useful for logli, at least, to know that many of these modern “anti-Whorfian” arguments, such as Pinker’s (see LN 95/1: 22ff), create for their polemical purposes a metaphysical Whorf, an anti-scientist, who is indeed easy to shoot down. That Whorf, whom I believe to be largely imaginary, has, wherever he resides, never played a role in my own thinking. From the very beginning, forty-one years ago, both the facilitative and learnability effects Miller talks about as if they were the products of fresh insights—indeed, are being presented by him as modern improvements on Whorf’s “antique” thinking—have been major parts of my own conception of what both Sapir and Whorf were saying about the possible effects of language on cognition since I began to read the works of these scholars nearly fifty years ago. It was that interpretation, not this modern metaphysical one, that has guided the design and subsequent engineering of Loglan, as well as my earliest contemplation of its uses, both as an experimental instrument and as a tool for broadening human understanding. Indeed, Chapter 7 of the 4th Edition of L1 gives a non-polemical account—a catalog—of the “symbol system effects” that Miller now tells us are “non-Whorfian”. —JCB


Copyright © 1997 by The Loglan Institute. All rights reserved.