Proto-Indo-European “thorn”-clusters

(1) Introduction Since the beginning of Indo-European linguistics, the group of words in which the Skt. cluster kṣ corresponds to Gk. KT have received much attention. According to Brugmann (1897: 790), these clusters must reflect a combination of a PIE velar plus “þ-Laute” (hence the name “thorn”-clusters), which was the standard view for many decades. For instance, in Pokorny 1959, the word for ‘bear’ (Skt. ¶koa-, Gr. ἄρκτος) is reconstructed as *¶§þo(875); the word for ‘earth’ (Skt. koám-, Gr. χθών) as *ǵðem(414); etc. When in 1932 Kretschmer equated the words for ‘earth’ in the newly found languages Hittite (tēkan) and Tocharian B (tkaṃ) with the thus far common reconstruction *ǵðem-, he was able to convincingly show that the initial cluster must originally not have contained a “thorn”, but rather consisted of a dental and a velar stop, *dǵ-. According to Kretschmer, the original order of these stops was retained in Hittite and Tocharian, but in Greek and Indo-Iranian the cluster was metathesized to *ǵd-, with a subsequent development of *-dto -sin Indic “weil ihm zwei Verschlußlaute im Wortbeginn ungewohnt waren” (1932: 67). In the other languages, *dǵwas simplified to *ǵ-, yielding Lat. hum-, Lit. žem-, OCS zem-, etc. Burrow (1959) argued, however, that assuming a metathesis in Indic is unnecessary. In analogy to Skt. koumánt‘having cattle’ ~ Av. fšūmaṇt‘id.’ < *p§u-mént-, where an initial cluster *p§yielded Skt. ko-, showing a development of palatovelar *§ into the retroflex sibilant ṣ, Burrow argued that we may assume a similar change for the “thorn”-clusters: *H¶t§o> *¶tśa> *¶»oa> ¶koa‘bear’ and *dǵém> *dj́ám> *dźám> ḍẓám> »oám> koám‘earth’. In his famous 1977 article ‘A thorny problem’, Schindler therefore concluded that the assumption of a separate phoneme *þ or *ð “is superfluous for an early stage of IndoEuropean” (1977: 34). According to him, all words with “thorn”-clusters reflect a cluster *TK (the one word where he reconstructs *KT, namely ‘yesterday’, will be treated in detail below). Moreover, he assumed that already in the PIE mother language this cluster was reduced in some environments, for instance before a syllabic nasal: *TK±C > *K±C. Recently, Lipp, in his book Die indogermanischen und einzelsprachlichen Palatale im Indoiranischen (2009), devotes a 350 pages long chapter to “Das Problem des Ansatzes von idg. þ (Thorn)”, in which he provides a very detailed account of all problems, proposed solutions and material regarding the “thorn”-clusters. Although this chapter is extremely elaborate, I still have the feeling that not all details regarding this topic have