Re: [tied] Re: Question about rules

If a consistent set of regular historical changes can be proposed to transform A into B and B into C at such time-scales, then B is either ancestral to C or as close to its ancestral lineage as makes no difference. Note that the parent-offspring relation (supported by a battery of correspondences which do not involve any irreversible sound changes) is obvious in the case of Latin and any of the the Romance languages (which is why they are recognised as a genetic unit on _linguistic_, not historical, cultural or acrchaeological grounds), but it would not be possible, even hypothetically, to derive French or Romanian from Ancient Greek, Old English, Sanskrit, etc.

For the above conclusion to be false, we would require a miracle -- the _independent_ development of B and a language ancestral to C along the same path (implementing the same sound changes in the same order and acquiring similar lexical and grammatical innovations), so that after some 2000 years it's hard to tell the difference between them. But then (apart from the fact that such things don't happen) they would be the same language, wouldn't they? And historical linguists, like palaeontologists, prefer the simplest scenario of evolution, one that requires the fewest assumptions and no miracles.

Piotr

----- Original Message -----
From: richardwordingham

To: cybalist@yahoogroups.com

Sent: Tuesday, September 03, 2002 11:20 AM

Subject: [tied] Re: Question about rules

--- In cybalist@......, Paul Alesu <rpales@......> wrote: > > Question about rules: > > Let's assume that a language B develops from language A over 2000 years > following a set of transformation rules AtoB. Independently, a second > language C develops from the same language A over 4000 years following a > set of transformation rules AtoC. > > A ->-> (AtoB) ->->B > A ->-> -> -> -> -> (AtoC) ->->->->->->C > > Due to lack of written and archeological resources we make the honest > mistake to believe that the development followed the sequence > > A ->-> (AtoB) ->->B -> -> (BtoC) ->->C > > where BtoC is the erroneously perceived rule which in reality is the > difference AtoC-AtoB. > > Does any linguistic method exist to help us detect this kind of errors? I believe the biggest indicator of such a problem would be the rules needed to reverse phoneme mergers or other losses of contrast. Changes undoing earlier changes can hint at a problem, but such changes can occur. The prothetic vowel came and went between Latin and Italian, and we can add the example of Guto Rhys's Welsh dialect where at least some of the prothetic vowels have been dropped. They would show up as 'sporadic' changes or weirdly conditioned changes. There would also be many unresolved problems in the evolution. There have been similar cases in the past, and the problems have come to light: 1. Sanskrit as the ancestor of all the other IE languages. Not entirely similar, as language A is unknown. 2. Sanskrit as the ancestor of the Indic languages. 3. Gothic as the ancestor of the Germanic languages. 4. Classical Latin as the ancestor of the Romance languages. I am not sure how much Vulgar Latin has been usefully documented by private inscriptions and the like. I suspect there may have been a similar story with (West Saxon) Old English and Modern English. However Anglian Old English is sparsely documented. Only the first case has time scales similar to your example. However, I think glottochronology can actually give useful results in the case that has inspired the question. The dates it gives may be unreliable, but I believe they give a useable measure of relatedness. I quote the dates given in Renfrew's book on IE origins; the dates appear to come from Rea, 1973. French and Italian split in 1586, as did Spanish and Portuguese. Romanian and Italian split in 1130. I think 1130 AD for 2000 BC (or even much earlier) would be an amazing error. Richard.

Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.