From: Piotr Gasiorowski
Message: 19123
Date: 2003-02-24
----- Original Message -----
From: "Miguel Carrasquer" <mcv@...>
To: <cybalist@yahoogroups.com>
Sent: Monday, February 24, 2003 12:57 AM
Subject: Re: [tied] PIE *ts ?
> I still prefer to avoid *þ and *ð as well as *ts and *dz from PIE reconstructions: *h1ed-ti and *dhg^ho:m are sufficient.
... as long as you are interested in the abstract morphophonemic input and the attested outputs only. Some surface-true reconstruction is necessary, however, if one is interested in understanding what happened in between. Since we are talking about phonological innovations occurring between PIE and the individual branches, such apparently trifling details may have some bearing on things like the genetic sub-classification of IE.
Piotr