Glottochronology

Glottochronology is a method that tries to calculate when two languages separated in the past. It is analogous to a kind of linguistic Carbon-14 test, but it usually cannot give absolute dates. The later term lexicostatistics is often used synonymously with glottochronology, though occasionally it simply refers to any kind of statistical comparison of lexical items from two languages.

It is obvious that languages change over time. Consider the opening lines of this poem by Chaucer written circa 1360. (The spelling has been standardized, and a modern rendition given to the right):

• Almightly and all mercy-able queen[Almighty and merciful queen],
• to whom that all this world fleeth for socour [to whom everyone in the world runs to for help],
• to have release of sin, of sorrow, and teene [for release from sin, sorrow, and hurt],
• Glorious virgin of flowers flower [Glorious virgin and flower of all flowers],
• To thee I flee confounded in error [To you I come, confused and in error].

Even though this poem was written almost seven centuries ago, it is no doubt transparent to any native English speaker today, and only a few archaic words are found. However, if we go back another few hundred years to Old English, things begin to look more and more like a foreign language as more and more of our familiar words drop out. Proponents of glottochronology argue that we might be able to measure the time depth of languages if we examine this rate of disappearance.

The glottochronology project rests on a number of assumptions. The first is that some words are more stable than others. Thus, an examination of basic core vocabulary will be more reliable than words that are more likely to change. There have been several proposed lists, based on several criteria, but a 100- or 200-word list is most commonly used. It includes terms referring, for example, to body parts, numbers, pronouns, and universal geographic features.

The second major assumption is the rate of change—that is, the loss of basic cognates (forms that have a common historical origin)—is the same for all languages at all times. Early proponents like Robert Lees and Morris Swadesh, comparing languages for which historical time depths were known, like the divergence of the Romance languages from Latin, conclude that anywhere from 80% to86% would be retained over a thousand years. Using languages for which the time of separation was accurately known, they proposed the following logarithmic formula:

where time depth (in years) is represented by t, cthe percentage of cognates shared between the two languages in question, and r the assumed retention rate (percentage).

As an example, we might compare five modern words from German and English, all core vocabulary items:

• animal (Tier)
• four (vier)
• head (Kopf)
• I (ich)
• sun (sonne)

We find 60% agreement, as two pairs (animal/Tier and head/Kopf) do not appear to be related. If we assume that languages change at the rate of 85% [Page 1091]every 1,000 years, after looking up the appropriate logarithms, the above formula gives:

In other words, English and German diverged 1,561 years ago, around the year 400 CE.

...