Groucho Marxism

In the context of Indo-European linguistics, the term ‘gutturalwechsel’ – literally ‘velar interchange’ – refers to alternations between palatovelars and plain velars across different satem languages, with the same word or root appearing with a palatovelar in some languages and a plain velar in others, or even a palatovelar and a plain velar within the same language. It can also take on an extended meaning which encompasses alternations between plain velars and labiovelars in the centum languages, with the same word or root appearing with a plain velar in some languages and a labiovelar in others. Such alternations occur frequently and suggest that Proto-Indo-European (PIE) only contained two types of velar consonant.

The reason is that these alternations suggest that the split between palatovelars and plain velars occurred separately in the satem languages, after the breakup of PIE, with palatalization occurring in one language but not in another, or in one phonological context but not in another. It is difficult to explain such alternations using a three-velar reconstruction. In a previous blog posts I have given several examples of these alternations: before *e ~ *H₂ in *gheH₂ns ~ *ghH₂ens- ‘goose’; before *e ~ *H₂ ~ *u  in *keH₂u ~ *kH₂u- ~ *kuH₂-  ‘owl’; before *e ~ *l in *kelH₂- ~ *klH₂- ‘straw’; and before  *e ~ *o in *gher- ~ *ghor- ‘enclose’, *kei- ~ *koi- ‘settle’, and *bherghe- ~ *bhergho- ‘protect’. In this blog post I will provide a few more.

One position where such alternations are common is in so-called ‘thorn clusters’: a sequence of a dental stop followed by a velar stop. (These were once thought to contain a phoneme represented by a letter referred to as ‘thorn’, although this idea has now been abandoned.) For example, *dhghom ‘earth’ yields Sanskrit kṣam alongside Avestan zā̊, Albanian dhe, and Greek khthṓn. The explanation is that the cluster *TK underwent metathesis in the satem languages and became *KT, which yielded *kṣ in Indic, and the velar was then blocked from palatalizing by the following *ṣ. Voiceless *tk yields *š rather than the expected **s in Iranian. For example, *ktitis ‘home’ yields Avestan šiti alongside Sanskrit kṣití and Greek ktísis.

Two examples derive from the root *tetk- ‘create’: *tetketi ‘create’ yields Sanskrit tákṣati, Avestan tašaiti, and Latin texō; and *tetkon ‘carpenter’ yields Sanskrit tákṣan, Avestan tašan, and Greek téktōn. The sequence *tk apparently yields ǰ rather than the expected **s in Armenian; for example, *H₂rtkos ‘bear’ yields Armenian arǰ alongside Sanskrit ṛ́kṣa, Avs. arša, Grk. árktos, and Latin ursus. Otherwise, the reflexes of *tk are identical to the reflexes of the sequence *ks; for example, *kokseH₂ ‘joint’ yields Sanskrit kakṣā́, Avestan kaša, and Latin coxa. This sequence yields *š in Baltic and *s in Slavic; for example, *deksinos ‘right (side)’ yields Sanskrit dákṣina, Avestan dašina, Old Church Slavonic desnŭ, and Lithuanian dẽšinas.

Another position where such alterations are common is before *r. For example, *krH₂uos ‘horn’ yields Old Prussian sirwis alongside Old Prussian curwis, Old Polish karw, Albanian ka, and Welsh carw. The explanation is that palatalization was blocked before *r in the satem languages but later spread analogically from the variant *kerH₂uos, whence Greek keraós and Latin cervus. Similarly, *suekruH₂ ‘mother-in-law’ yields Sanskrit śvaśrū́ alongsode Old Church Slavonic svekry, Latin socrus, Old High German swigar, and Welsh chwegr. The explanation is that palatalization was blocked before *r in the satem languages but later spread analogically from the variant *suekur, whence Sanskrit śváśura, Lthuanian šẽšuras, and Greek hekurós.

In a 2014 paper, the Australian linguist Robert Woodhouse showed that palatalization was not blocked after *s in Iranian when followed by a front vowel, unlike in other satem languages. This explains the gutturalwechsel in words such as *skidros ‘split’, which yields Avestan sidara alongside Sanskrit chidrá, Latvian šk’idrs, and Greek skidarós. In a 1978 paper, the Dutch linguist Frederik Kortlandt argued that palatalization was blocked before *m and *n in Balto-Slavic. This explains the gutturalwechsel in words such as *H₂ekmon ‘stone’, which yields Sanskrit aśman, Avestan asman, and Lithuanian ãšmens, alongside Old Church Slavonic kamy, Lthuanian akmuõ, and Greek ákmōn. To explain this alternation we can posit that palatalization was blocked before *m but later spread from variants of the root *H₂ek- ‘sharp’.

In his 1978 paper, Kortlandt also argued that palatalization was blocked before *uo in Balto-Slavic. This explains the gutturalwechsel in *kuoitos ‘bright’, which yields Old Church Slavonic cvětŭ alongside Sanskrit śvētá, Avestan spaeta, and Old Church Slavonic světŭ, To explain this alternation we can posit that palatalization was blocked before *uo but later spread analogically from the variant *kueitos, whence Gothic hweits; or from the variant *kuitros, whence Sanskrit śvitrás and Lithuanian švitras. Another example is given by *pekus ~ *pekuos ‘livestock’, which yields Lithuanian pẽkus alongside Sanskrit páśu ~ paśváḥ, Avestan pasu, Latin pecū, and Gothic faihu. Here we must posit that the plain velar spread analogically from the genitive *pekuos in Baltic.