Zompist Bboard Again

Otto Kretschmer wrote: ↑Thu Mar 25, 2021 1:55 pm

alice wrote: ↑Thu Mar 25, 2021 1:52 pm Is there any realistic way to actually *quantify* linguistic change and reduce it to a number?

You can calculate average rate of linguistic change and vocabulary replacement. It was calculated that relative to Old Norse, Norwegian is 800 years old while Icelandic is only 200 years old.

Sorry for the potentially quarrelsome question, but what units are average rate of linguistic change and vocabulary replacement measured in? It's my understanding that the only way to quantify something is to measure it in some units. And how were those two figures calculated?

Zju wrote: ↑Thu Mar 25, 2021 3:12 pm Sorry for the potentially quarrelsome question, but what units are average rate of linguistic change and vocabulary replacement measured in? It's my understanding that the only way to quantify something is to measure it in some units. And how were those two figures calculated?

It's based on an analogy to radioactive decay - the theoretical basis is that vocabulary mutates essentially at random, so if you measure the rate of change over time by looking at the number of conserved lexemes in the Swadesh list you can calculate the divergence time between two related languages. Initial attempts at glottochronology assumed a constant "lexical half-life" across all languages, which led to the absurdity that Icelandic is somehow less old than Norwegian by comparing both to Old Norse.

To pursue the analogy, how did these calculations take into account the different half-lives of different nuclides (i.e. meanings)?

And for that matter, how do you quantify change in phonology, grammar, and syntax?

you don't. like all attempts to quantify non-quantifiable data it's a fool's errand

If you drop the chronology side of it, measuring the comparative rate of loss of vocabulary seems like an interesting thing to do though. I wonder whether Italian has dropped less vocabulary than Spanish, for example (probably yes). Although one problem is that sometimes words survive but only as fancy uncommon synonyms (piélago in Spanish for the sea), or are consciously reborrowed from an old language, and it's not even easy to tell... And there's the question of whether derived words should count (does French soleil < sōliculus count as a survival of Latin sōl?).

I have not read any work of glottochronology, but I wonder how well the many issues of methodology are addressed...

Glottochronology is pseudoscience, about on a par with astrology. It doesn't work. Not even relative chronologies can be achieved that way, because the thing can be fouled by some languages replacing vocabulary faster than others. I have seen IE family trees constructed that way where the topology of the tree was obviously wrong, e.g. by placing Polish within East Slavic or Albanian deep within Romance. No matter how the thing was dated.

WeepingElf wrote: ↑Fri Mar 26, 2021 9:51 am Glottochronology is pseudoscience, about on a par with astrology. It doesn't work. Not even relative chronologies can be achieved that way, because the thing can be fouled by some languages replacing vocabulary faster than others. I have seen IE family trees constructed that way where the topology of the tree was obviously wrong, e.g. by placing Polish within East Slavic or Albanian deep within Romance. No matter how the thing was dated.

Using distance metrics to build phylogenetic trees seems prone to this sort of error. Albanian as part of Romance looks like Romance loans not being identified - did you look into whether that happened?

Polish as East Slavic need not be so wrong. One should expect problems when working on dialect continua. The Anglo-Frisian group is far from visible when looking at modern languages, though numeric hybridisation detection techniques can pick up that something non-branching went on there.

Kuchigakatai wrote: ↑Fri Mar 26, 2021 8:28 am If you drop the chronology side of it, measuring the comparative rate of loss of vocabulary seems like an interesting thing to do though. I wonder whether Italian has dropped less vocabulary than Spanish, for example (probably yes). Although one problem is that sometimes words survive but only as fancy uncommon synonyms (piélago in Spanish for the sea), or are consciously reborrowed from an old language, and it's not even easy to tell... And there's the question of whether derived words should count (does French soleil < sōliculus count as a survival of Latin sōl?).

Glottostatistics works by looking at the words attached to meanings, not the words themselves. Derived words have to count as retentions - one can get massive levels of derivation. Also, where would one draw the line with shifts of declension?

Richard W wrote: ↑Fri Mar 26, 2021 6:54 pm

Kuchigakatai wrote: ↑Fri Mar 26, 2021 8:28 am If you drop the chronology side of it, measuring the comparative rate of loss of vocabulary seems like an interesting thing to do though. I wonder whether Italian has dropped less vocabulary than Spanish, for example (probably yes). Although one problem is that sometimes words survive but only as fancy uncommon synonyms (piélago in Spanish for the sea), or are consciously reborrowed from an old language, and it's not even easy to tell... And there's the question of whether derived words should count (does French soleil < sōliculus count as a survival of Latin sōl?).

Glottostatistics works by looking at the words attached to meanings, not the words themselves. Derived words have to count as retentions - one can get massive levels of derivation. Also, where would one draw the line with shifts of declension?

Although that sounds reasonable, I was thinking that derivation could also indicate something about how much a language has changed. Namely, French has soleil where Spanish has sol, Portuguese sol, Italian sole, Romanian soare, Sardinian sole. That may be saying something about French... The descendant of sōlem would be seul, a homophone of *seul < sōlus 'alone' and *seul, seuil < solum 'floor' (seuil seems to come from solium 'seat', maybe motivated by homophony too). Spanish distinguishes the three as sol, solo and suelo...

Was high literacy and urbanization rate in Greece a factor of why Greek has changed relatively little since the time of Biblical Koine?