>Nik Taylor wrote: > >>Have any of you clicked on the link nearby, the "Caveman words", at >>http://www.discovery.com/news/briefs/20000420/misc_babtable.html ? It >>gives the following "protowords": > >[snip list] > >This looks like the work of Ruhlen and Bengtson. I've looked at their work on >this subject, and they just don't have a clue. The data they use for their >"reconstructions" are so semantically and phonetically loose, that you could >fit almost anything in. > >For example, not included in this list is TIK "one; finger" >Allowed correspondances: >T - s, t, d, d’, ts, z, t’, tj, tl, tl’, lh (lateral fricative), nt, dl, th >(as >in English), c, ts’, ch, sh >I - i, e, o, @ (schwa), u, E, a, ei, ai, ay, y (front rounded), barred-i, ia, >ea, ï, ö >K - k, x, g, h, kk, ?, hw, q, kh, k’, N, kw, xw, qw, ch, j, X > >As you can imagine, some words that are created from these sets hardly like >TIK. To make matters worse, they allow completely free metathesis and >deletion, and insertion of all kinds of sounds (mostly nasals and glottal >stops). Worst part - they are willing to take any string with these sounds, >whether or not they are separate morphemes. Try Imihita _meuxtsekoa_ >"finger". Claiming that there is a connection between this and Japanese _te_ >"hand" really stretches their credibility. To be precise: The fault with these reconstructions lies not in the phonetic discrepancy between alleged cognates, but in the lack of systematic sound correspondences. I would be willing to buy a sound correspondence of, say, /k/ and /j/ if it could be demonstrated that this correspondence shows up consistently in cognate after cognate. From a linguistic standpoint, the real problem with Proto- World reconstructions is not their implausibility, but their ad hocness. Matt.