NTNU Faculty of Arts
Norwegian University of Department of Linguistics
Science and Technology


Lars Hellan
Telephone no.: +47 73 596526
E-mail: lars.hellan@hf.ntnu.no
Dated: Your reference: Our reference:
16.10.2001 2001/243/423/akf 2001/228/423/LH


The Faculty of Arts


Committee report on the thesis submitted for defence for the degree Dr. Philosophiae
"What's in a Word: Studies in Phonosemantics"
by Margaret Magnus


In the committee's opinion, the dissertation entitled "What's in a Word: Studies in Phonosemantics" by Margaret Magnus meets the six major criteria for a doctoral degree from a research university oriented toward science:

1. An important but neglected problem, defined in a novel way and articulated with a theoretical model of language.
That model is that "most fully expounded in the generative tradition, of language as a natural process which is largely unconscious and whose structures have to be brought to consciousness by empirical methods (p. 185)." Unlike most of today's generativists, however, she rejects the view of language (langue) as somehow autonomous, abstractable from the (parole of the) human user

2. Awareness of the history of the problem and citation of a broad relevant literature.
The historical survey succeeds in relating ideas from different traditions and disciplines, and points the way to further integration.

3. Courageous (testable, i.e., falsifiable) hypotheses and an original research design.
Of the fourteen experiments, less than half have even partial antecedents in the scholarly literature. Experiments 11-l3 feature ingenious field-tests making innovative and economical use of Internet technology.

4. Painstaking marshaling of empirical evidence.
While many of M's predecessors have associated certain sounds or sound-combinations with semantic regularities, M is the first to give equal attention to contexts where the generalizations do not hold, and the first to venture explanations for the exceptions. She shows, for instance, that most monosyllabic English words sharing particular sounds fit into limited sets of functionally defined classes; that the prevalence of some sounds in some semantic clusters (and their absence from others) is too striking to be due to chance. Within each phonetically defined group, however, the number of exceptions hovers around 3%; these are nearly always concrete nouns whose salient referential meanings override the potential connotations of their component sounds,

5. Interesting results (first outlined on p. 4 and summarized in Chapter 5).
The Phonosemantic Hypothesis has been supported by multiple demonstrations of the contribution of sound to word-meaning, and of the inverse relation between referential (denotative) meaning and the connotations contributed by sound. The members of the committee are left willing to believe that "in every language of the world--" there is (there must be) some natural law at work that involves very abstract properties of articulatory features. This pioneering thesis provides insights into these properties. Most of the data are from English, with more limited evidence from other languages. As M is careful to point out, phonemes are language-specific, and the functional classes are, if not language-specific, at least culturally-biased. It is evident from M's work that teasing apart what is universal from what is specific to a particular community is an arduous agenda for generations beyond her own.

M lays to rest the controversies about the arbitrariness or non-arbitrariness of the linguistic symbol. She shows that in all words the relation of sound and meaning is to some extent arbitrary, to varying extents iconic. The arbitrary aspect is strongest in the words (mainly concrete nouns) whose specific reference masks other potential contributors to the word's meaning. M shows convincingly that, while reference is arbitrary, the meaning of a word is not reducible to its reference (or denotation), she demonstrates time and again that words sharing a referential meaning (e.g. jump, leap, hop, -- or look, gaze, glare, stare, --) have different connotations correlated with their sound shapes. The semantic contribution of sounds is to qualify, to suggest what the referent -- in these cases the act of jumping or looking -- is like. Particularly convincing -- and novel -- are M's discussions of the contextual interaction among sounds (/sl/ vs. /gl/ for instance), or the variations due to different positions in the word, e.g., /t-r/ vs./ r-t/, for instance, the brilliant discussion of "true" vs. "right" (pp. 123).

The"phonestheme" phenomenon sporadically observed in clusters such as /gl/ or /fl/ (or in isolated phonemes by one or two earlier scholars), is now shown to extend to the entire inventory of English consonants and even to distinctive features and to their articulatory correlates. Besides new and often astonishing observations and convincing generalizations, M offers bold hypotheses that go a long way toward explaining the phenomena she observes.

Among modern "iconists" M is first to draw a careful distinction between "clustering by semantic association" and "true iconicity" and to provide experimental evidence for their universal character.

6. The promise of further fruitful research. M provides enough evidence from other languages and from hypothetical words to make her Phonosemantic Hypothesis a strong candidate for a language universal and outlines a research program for testing her hypotheses across languages. Some interesting suggestions are made in passing, for instance, for testing if the positional effects in VSO or SOV languages are as in the SOV languages which provide most of her evidence (p. 114). The distinction M makes between the classificatory (metonymy-hyponymy) referential meaning (which has had considerable attention from linguists) and the still tentative functional classes that provide a home for clustering sound shapes is an important one. M's inventory of classes (which she carefully warns us is based on what is codified in English) requires further exploration, if only to sort out the universal from the culture-specific. M's evidence is drawn mostly from English, with some support from two other languages she speaks well (Norwegian and Russian), and a range of data from other languages culled from dictionaries or from earlier studies. Even as she muses about universals, she brings up interesting points about possibly predictable effects of sound changes on word-frequencies, and about the differences among the phonemes of different languages. (For instance, if American English /r/ is labial (p. 102) and bears a connotation of rupture in certain contexts, one wonders if that connotation is stronger in the more intensely articulated alveolar /r/ of Spanish or Russian? Does the uvular /r/ of French or German differ significantly?).

The earlier neglect of Phonosemantics is due in part to the inaccessibility of the data and to other inherent difficulties which have led the author to devise an unusual expository format. We believe it is worth-while to say a few words about that here.

A linguist writing about the (abstract) meanings contributed by the components of words is forced to record observations by means of whole words, whose meaning will necessarily specify more than is needed or wanted. To transcend this difficulty, M resorts to (quite entertaining) metaphors and analogies, both in the names assigned to semantic/functional classes and in discussion of specific examples. Rather than interrupt the exposition with floods of data, and perhaps to accustom the reader to her terms, she has chosen a kind of spiral structure for the body of the thesis, with some redundancy: first in a brief introduction, and a few well-chosen illustrations (pp. 4-5), then a more detailed illustration (/gl/ pp. 40-45) and an overview of the experiments (45-46), then in accounts of each experiment with selected illustrations, and a final chapter summarizing the results and a most interesting section on future research. The full data quantified in the experiments are contained in separately bound appendices, the bulky Appendix I classifies all the English monosyllabic words (over 3000), first according to how words containing each consonantal phoneme (not only word-initial) lend themselves to semantic groupings, and then according to how all those monosyllabic words distribute among semantic classes. A third volume contains Appendices II-XIV. The first nine of these, like Appendix I, record tests for semantic clustering and for the productivity of semantic associations; the last four test for true iconicity.

Beyond its importance as an example of "scientific linguistics," this thesis exemplifies the role of linguistics as a discipline that is potentially at the center of the academy, a meeting ground between the natural sciences, the social sciences, and the humanities. Most saliently, M's thesis explores the nature of human language by using the experimental "scientific method." It also provides grist for the philologist in its consideration of allophonic variation and language change; and it should engage the anthropologist and sociolinguist with its proposal for semantic classes and its implied invitation to refine the inventory and sort out the universal from the culture-specific. The thesis is also a most remarkable contribution to humanistic scholarship, first of all in its use of the Cratylus dialogue as a unifying theme, and then in her discussions of types of meaning. For M, every word has a triad of meanings whose saliency varies; her terms -- iconic, classificational, referential -- have (iconic) antecedent in Peirce's hierarchy: firstness, secondness, and thirdness. Even as she draws a careful distinction between true iconicity and clustering, she points out that clustering is not only indexical (classificational) but also iconic, since similarity is also involved.

It is no accident that the most comprehensive survey of iconicity in language cited in this work (Genette 1976) is by a literary theorist; for M's thesis promises important applications to literary criticism. M's discussion of contrasts due to sounds (bull, buffalo vs. bison (pp. 70-71); or right vs. true (p. 123) compares well with discussions found in the best poetic interpretations (including Jakobson's), but where such literary statements tend to be intuitive and apodictic, M's research provides theoretical and empirical underpinnings.

On the basis of these considerations, the committee wholeheartedly recommends that Margaret Magnus' thesis be accepted for a defense for the degree Dr. Philos.

Yours sincerely,

Catherine M. Chvany
Professor Emerita
Department of Foreign Languages
Gregory N. Carlson
Department of Linguistics
University of Rochester, New York
Lars Hellan
Department of Linguistics