Radical sound changes

Some years back while reading a now forgotten book on historical linguistics from some part of the world or other, I came across a citation to the following article as an example of just how radical regular sound changes can be: Ives Goddard, “An outline of the historical phonology of Arapaho and Atsina”, International Journal of American Linguistics 40 (1976), pp. 102–116. I noted this on my list of articles to look up and photocopy one day, and only now have I managed to finally read the paper.

In Goddard’s reconstruction of Arapaho and Atsina out of Proto-Algonquian through Proto-Arapaho-Atsina, some of the shifts are readily understandable for someone working in the Indo-European–Uralic–Altaic field, though one might compare them to the more freakish cases like Armenian or Samoyedic. Other changes, however, are puzzling, like these two cases of nasalization:

  • PA *s becomes *n initially: PA *sipiwi or sipowi ‘river’ > A-A *niki- > Ar nííčíí, nííčíe, Ats nííčééh, PA *sakimewa ‘mosquito’ > A-A *noimen ‘fly’ > Ar nóúbee, Ats nóúbee.
  • Non-post-consonantal *y becomes *n: PA *ayapewa ‘male of large ungulates’ > A-A *eneken ‘buffalo bull’ > Ar enééčee, Ats enééčee (there is a vowel harmony process active as well).

Incidentally, there is a tendency in the sciences for people to assume that all major progress is recent, and that earlier scholarship was very primitive. I seem to suffer from that myself, as while reading this paper I was continually struck by how Goddard’s terminology and sociolinguistic outlook is remarkably fresh for something published in 1976. Where can a cutoff point between old and contemporary be reasonably set?

Swedish encounters with Delaware

I had never known that Swedes had contact with colonial languages until I read The History of Linguistics in the Nordic Countries by Even Hovdhaugen et al. (Helsinki: Societas Scientarum Fennica, 2007) The authors give the following brief account of Swedish missionary efforts among speakers of Delaware (Lenni Lenape).

The first Swedish colonists in America, many of them Finnish immigrants from Värmland, came to Delaware as early as 1538. John [Johannes Jonæ Holmensis] Campanius (1601–1683) stayed and served as a pastor in Delaware from 1643 to 1648. After returning to Sweden, Campanius finished the manuscript of a Delaware catechism in 1656, but by this time a Swedish colony no longer existed in Delaware. Long after his death, Campanius’s grandson, Thomas Campanius Holm, as well as others who planned missionary work among the Delawares, had the Delaware catechism published in 1696, and 600 copies were printed and brought to the Delaware Indians. The article on Campanius in the Swedish Biographical Lexicon (Svenskt Biografiskt lexicon VII:1927 p.262), contains the following informative passage:

The Swedes there, who now knew the language of the Indians as well as their own, read from the catechism to the Indians, and “a number of them observed carefully what was written”. They even asked a Swede, Karl Springer, to teach their children.

This is especially interesting in the light of Holmer’s analysis of Campanius’s Catechism (1946b), in which he convincingly shows that Campanius only had a very rudimentary knowledge of Delaware and furthermore that he had no understanding of the very complicated morphology of this Algonquian language. For instance, he never used plural forms, and he adopted very unidiomatic independent pronouns instead of pronominal suffixes, which he had not observed. Accordingly, he used nux ‘my-father’ (cf. n- ‘my’, ux ‘father’) as the word for ‘father’ and when he wanted to express the sentence ‘honor thy father’, he wrote the equivalent of ‘honor thy my-father’ which may have been unintelligible for the natives. Similarly he used mpa ‘I come’ (cf. m- ‘I’, pa ‘come’) also for ‘thou come(st)’, which is kpa (cf. k- ‘thou’) in Delaware. The syntax was worse. Sometimes Campanius even resorted to desperate devices like introducing the Swedish genitive -s into Delaware.

Holmer is right when he stated that this translation was unintelligible to a native speaker of Delaware. But actually it was not Delaware at all, but a kind of Delaware-based pidgin that had already developed through contact with Europeans (mainly Dutch colonists) before the Swedes arrived. In this respect, Campanius’s catechism may be a very valuable document for linguistics, since it would be the only record of this language.

Campanius’s grandson published an account of the Swedish colony in Delaware (T. Campanius Holm 1702), which was influential and widely read in Sweden. The book contained extensive material, word lists and short phrases as well as a dialogue, mainly based on his grandfather’s work. He also tried to prove the hypothesis going back to Governor William Penn that the Delaware Indians are one of the ten lost tribes of Israel and accordingly that their language is related to Hebrew. To prove this Holm provided about twenty etymologies of very questionable quality (T. Campanius Holm 1702:115-120).

The purely musical language

Back in April an article about the weirdness of Pirahã appeared in The New Yorker, written by John Colapinto. It has come to my attention now that popular news sites like Slashdot and Digg have featured stories on the supposed lack of numbers in the language.

While in many respects the article is informative, especially about the inter-personal dramas of the field, it evidently just isn’t possible for a reporter to ever get all the details right and clear in writing about linguistics.

Unrelated to any other extant tongue, and based on just eight consonants and three vowels, Pirahã has one of the simplest sound systems known. Yet it possesses such a complex array of tones, stresses, and syllable lengths that its speakers can dispense with their vowels and consonants altogether and sing, hum, or whistle conversations.

Is the author suggesting that Pirahã is some kind of Solresol-like freak where one can just communicate through pure tones? It sure seems like that’s what he wants to say. He even repeats the point later:

The key to learning the language is the tribe’s singing, Keren said: the way that the group can drop consonants and vowels altogether and communicate purely by variations in pitch, stress, and rhythm—what linguists call ‘prosody.’

Yes, prosody is an important part of language, but it works around vowels and consonants, not without them. What could the author possibly mean? The farthest any flight of fancy would take me is that each part of speech has its own tone in Pirahã, so the mother being described in the latter portion would be forming valid constructions for her child in that the tones for all parts of speech for a coherent sentence would be present and in the right order, but without any actual lexicon involved. But I know nothing about Pirahã, so that’s just a wild shot.

Last speaker of the Alaskan language Eyak dies

From an article at BBC News comes this sad news:

A woman believed to be the last native speaker of the Eyak language in the north-western US state of Alaska has died at the age of 89.

Marie Smith Jones was a champion of indigenous rights and conservation. She died at her home in Anchorage.

She helped the University of Alaska compile an Eyak dictionary, so that future generations would have the chance to resurrect it.

It’s depressing to think that if I pursue a career in the Uralic languages, at current rates I’ll be reading stories like this about a good many of the languages of that family before retirement.


I’ve been doing more with Indo-European within my university studies now, enough that it is starting to seem like “work”, so in my free time I’ve been reading more about other language families. I’ve stumbled upon an fascinating connection between Athabaskan, an American Indian language family, Chinese, and Vietnamese, concerning the development of tones.

In Lyle Campbell’s American Indian Languages: The Historical Linguistics of Native America (Oxford University Press, 1997), p. 113, we find:

While many Athabaskan languages have tonal contrasts, Proto-Athabaskan lacked tone—a trait that can be shown to have developed from (Pre-)Proto-Athabaskan differences among *V and *Vʔ (and *V:)

In Chinese too, a word-final glottal stop is responsible for tonogenesis. In Jerry Norman’s survey Chinese (Cambridge University Press, 1988), he notes that in Vietnamese one tonal category derives from words which in the surely toneless proto-language ended in a final stop. Vietnamese words with a rising tone have cognates in Khmu and Riang which still end in glottal stops, e.g. Vietnamese ‘leaf’ cognate with Khmu hlaʔ and Riang laʔ.

Furthermore, the shǎng (rising) tone of Middle Chinese seems to derive from Old Chinese words ending in a glottal stop. Old Chinese *pang gives the píng (high) tone and, according to the “s-hypothesis”, *pang-s ends up with the (departing) tone, while a form ending in a stop, e.g. *pak, leads to the (entering) tone. Finally, *pang-ʔ is responsible for the fourth, rising tone. As Norman writes, Glottality still survives as a feature of the rising tone in several modern dialects.

If ending glottal stop is so productive in creating tones, I would be interested in the tonal ramifications are of Proto-Indo-European reconstructions ending in -h1, widely believed to have been a glottal stop. However, while reconstructed forms with the endings -h2 and -h3 abound, I cannot think of anything ending in the first laryngeal.

Incidentally, I would like to thank the anonymous reader who recently bought Curta’s The Making of the Slavs through the referral link here. As that book is somewhat expensive, the referral fee will provide three full meals today to this impoverished student.