One of the bedrock principles of linguistic analysis since the nineteenth century has been the principle of the regularity of cognate borrowing. It forms the basis of the "comparative method" not only in linguistics, but in all of social science. Within the same linguistic family it is expected that a large proportion of linguistic material will be recognizably related due to the derivation of that material from a common linguistic ancestor.

Historical linguists endeavor to show how the patterns of language exhibit regular change as they evolve from mother to daughter languages. The most ambitious theoreticians, such as Morris Swadesh, have claimed to be able to date the divergence of populations from each other "glottochronologically" by correlating presumed migratory movements with rates of linguistic change.

Others have used cognate similarities between languages to speculate on the earliest forms of human language. Stanford linguist, Joseph Greenberg, for example, notes that many languages have a form dik, dig, or tik that refers either to the number "one" or to the index finger, suggesting a bodily origin for words for numbers (the English term being "digit"). Paul Friedrich uses a compilation of common tree names in Indo-European languages to trace a purported geographical origin of the Indo-European people.

However, there is a limited, but powerful countervailing tendency in language behavior--words that absolutely resist borrowing even from their closest linguistic relatives. These words seem to be coined anew by each population group. Because we expect cognate borrowing as a norm, it is surprising when we encounter these fascinating examples. It makes us wonder about the cultural processes that govern the development of communication systems, and the functional differences between segments of vocabulary.

This little discussion started with an inquiry by a group of mathematician friends at Santa Clara University. They noted the curious fact that the word for "butterfly" was different for every European language, including those most closely related--such as Spanish and Portuguese. I later found that the "butterfly problem" is one of those linguistic curiosities that has lurked at the edges of scholarship for some time without much in the way of a full research effort--the linguistic equivalent of the study of yawning by biomedical researchers. The first well-known linguist to note this phenomenon was Emmon Bach, but it has occupied the interest of a surprising number of people (1).

In modern mode, I went to LinguistList.org and posted a query asking for the word for butterfly in different languages. I got fifty replies in two days, and they are still coming in. A large number of people have been collecting their own lists for years. Patricia Black, a Philosopher at Ohio University wrote:

One of the nicest of the compendia I received was a short list from an unpublished paper compiled by linguist Haj Ross, formerly at MIT, and now teaching at North Texas University. Starting with Ross' list, I have added many additional terms contributed by email correspondents.

One astonishing reply came from correspondent Gianfranco Unali, who gives all of the terms found in Sard, the Italian/Spanish/Catalan influenced variety spoken in Sardinia. In addition to farfalla, a clear borrowing, one finds: arrèndza, áspu, bellagasu, bollankàu, kakare, kalare, karabáttula, kárru, kásu, kolare, farina, grattare, lèppore, lìlliri, máma, maría, nòvas,maripòsa,paparéddu, pappagállu, pappare, pompòni, prebelèi, pummeribèlla, pudzòne, sesè, spioni, tutare, and volare.

Ross contrasts these terms with the words for cat in some of the Indo-European languages in this group just to show the stark contrast

The terms for butterfly have several things that generally unite them: they involve a degree of repetitious sound symbolism, (Hebrew parpar; Italian farfale) and they use visual and auditory cultural metaphors to express the concept. Inspecting the list of butterfly terms, it is easy to see how these principles play out in the construction of the terms. In each case, with the many cases of reiterated b's, p's, l's and f's (in widely separated language families) one can almost hear the gentle rustle of butterfly wings and see their repetitive motion.

In those cases where such sound symbolism is not present, more conventional metaphoric processes prevail. To give a sense of this, several examples follow. One internet correspondent, linguist Neal Norrick, gives a definitive explanation for the German term:

Bert Beynan offers this explanation for the Russian term:

Lameen Souag provides the following for the Algerian Arabic bu frtutu

The explanation for this phenomenon defies analysis using the traditional techniques of historical linguistics(2). Looking for more basic principles of cognitive tendencies in human language seems more productive.

The eminent linguist Edward Sapir wrote a prominent paper entitled "A Study in Phonetic Symbolism," in which he suggests that there may be something inherent in human cognition associating particular sounds with particular concepts. His experiments were somewhat imprecise and nearly anecdotal, but no one reading the paper can fail to come away without a feeling that he "had something" in his formulation. In particular Sapir notes that "high" vowels, /i/, /e/ connote small things, and low vowels: /a/, /o/ connote larger things. Sapir also notes that onomatopoeic terms in all languages are iconic in nature, resembling the sound they represent.

The use of metaphor is also a universal process in language, both in the coinage of words as in the construction of discourse. George Lakoff, linguist and Mark Johnson, psychologist, of Berkeley have identified metaphoric construction as the basis for most semantic processes. Linguist Roman Jakobson, and anthropologists Paul Friedrich and James Fernandez point out that aesthetically conceived tropic structures are fundamental building blocks for all cultural materials.

All of these researchers lend excellent support to the idea that the linguistic realization for butterfly might be something welling up from the most basic cognitive creative processes, rather than from a process of inheritance from a mother language or borrowing from a neighbor language.

But why should this process be so universal? And why should butterflies be singled out as unique aesthetic creations of language whereas terms for cats and dogs get borrowed wholesale? Emmon Bach suggests that it has something to do with the creature itself. Whereas cats and dogs are commonplace, however much we may love them, butterflies are uniquely inspirational poetic creatures that by their nature demand special linguistic treatment.

Haj Ross puts this argument beautifully: "the concept/image of butterfly is a uniquely powerful one in the group minds of the world's cultures, with its somewhat unpromising start as a caterpillar followed by its dazzling finish of visual symmetry, coupled with the motional unforgettability of the butterfly's flipzagging path through our consciousnesses. Butterflies are such perfect symbols of transformation that almost no culture is content to accept another's poetry for this mythic creature. Each language finds its own verbal beauty to celebrate the stunning salience of the butterfly's being."

However, one then wonders about the other great poetic images found in so many cultures: flowers, clouds, trees, mountains, birds, and so many other objects of beauty and wonder. Although the terms for individual species of animals or plants, and for specific geographic features may be language specific, the generic terms for the categories are widely borrowed between language. It may be that the butterfly and its unique status in human linguistic cognition will remain as elusive and mysterious as the creature itself.

Notes