How Not to Do Comparative Linguistics

The March/April issue of the pop science magazine Archaeology contains an article by UC Press editor Blake Edgar about Cal Poly San Luis Obispo archaeologist Terry Jones and UC Berkeley linguist Kathryn Klar entitled “The Polynesian Connection: Did ancient Hawaiians teach California Indians how to make ocean-going canoes?” The short answer is, “No.”

Edgar cites convincing archaeological counterevidence from Santa Barbara Museum of Natural History archaeologist John Johnson before conceding that “their strongest evidence may rest on a few words.” That’s sad, because the linguistic argument is almost a textbook example of how not to do historical and comparative linguistics. Here’s the essence of it, as presented in the article.

Any language, says Klar, includes words that are native to it, words borrowed from other tongues, and others of unknown origin. When Klar began studying the island variant of Chumash, she found words alien to mainland Chumash dialects. She looked at distant native languages, from Aleut to Uto-Aztecan, whose speakers could have had contact with Chumash. Each time, Klar came up empty--until she tried Hawaiian, a member of the Polynesian language family.

Klar noticed a Hawaiian word that translates roughly as “useful tree,” kumulaa‘au. This bears a striking resemblance to the ancient Chumash word for “sewn-plank canoe,” tomolo‘o, which Klar reconstructed from the terms for “plank canoe” in different branches of the Chumashan language family. The first letters differ, but in Hawaiian “k” words often derive from older words that begin with “t.” Both the Hawaiian and Chumash words contain four corresponding consonants. That’s too many for a coincidence and to a linguist signals the virtual certainty of genetic kinship or borrowing from the native language. Since Hawaiian and Chumash don’t share a traceable ancestry, that leaves borrowing.

So what’s wrong with this linguistic evidence?

• Normally, you’d expect Polynesian linguistic evidence to come from a specialist in Polynesian, or at least Austronesian. Not Celtic.


• Polynesian ocean-going canoes are either double-hulled (like catamarans) or have outriggers. In fact, outriggers are standard all over Oceania. The Chumash canoes were single-hulled, without outriggers.


• Hawaiians call their canoes wa‘a, from Proto-Polynesian *waka, which has cognates all over Polynesia--and Oceania more generally. They don’t call their canoes ‘useful trees’.


• The Hawaiian word mistranslated as ‘useful tree’, kumulaa‘au (where aa indicates a long a), translates better as something like ‘tree (with trunk)’, from kumu ‘bottom, base, foundation, trunk’, plus laa‘au ‘tree, plant, wood, timber, forest, stick, pole’.


• The semantics are a stretch. If the Chumash islanders were going to borrow a form for ‘sewn-plank canoe’, why didn’t they borrow Hawaiian wa‘a ‘canoe’ or kaula ‘rope, cord, string’, or even humuhumu ‘to sew’, instead of the word for ‘tree (with trunk)’. ‘Treetrunk’ would be a better match for ‘dugout canoe’.


• The sound correspondences are haphazard. The “ancient Chumash” form tomolo’o is paired with modern Hawaiian kumulaa‘au. But the “ancient” Eastern Polynesian form for the latter would have been something like *tumu ‘trunk’ plus *la‘akau ‘wood’. It wasn’t just Polynesian *t that shifted to k in Hawaiian (and that fairly recently); Polynesian *k had earlier shifted to ‘ (glottal stop). So, if we’re going to restore Hawaiian kumu to Polynesian *tumu to make it match Chumash tomo- then we also need to restore Hawaiian laa‘au to Polynesian *la‘akau, where *la‘a would have to match Chumash -lo and *kau would match Chumash -‘o. Let's not even talk about the vowels, which famously count for little in historical and comparative linguistics.

It wouldn’t be at all surprising if ancient Polynesians had visited the coast of North America. They apparently reached the coast of South America, from which they brought back the sweet potato. But the evidence they had anything to do with Chumash ocean-going canoes is far from convincing.

SOURCE: Hawaiian Dictionary, revised and enlarged edition, by Mary Kawena Pukui and Samuel H. Elbert (U. Hawai‘i Press, 1986).