*sh₃óg^u̯-/sh₃g^u̯-´

Besides PAnat. *sHeg^u̯‑, the meaning ‘to see’ is also found in PGerm. *seχ^u̯‑, *saχ^u̯- ‘id.’ and Alb 1. sg. shoh ‘id.’ as well as in the root underlying OIr. rosc ‘eye’.
The agentive root noun Core IE *h₃ó/ék^u̯- ‘eye’ may be paraphrased as ‘the seeing one’ > ‘eye’. Thus, PAnat. *sHeg^u̯- ‘to see’, Core IE 1.*sek^u̯- ‘to see’, and *h₃ek^u̯- ‘eye’ go back to the same root PIE *(s)h₃ek^u̯- ‘to see’.

Further connections to other roots remain tentative, but an original meaning ‘to attach’ provides a persuasive basis for the putative cognates:

1. resultative: ‘to have one’s attention/eyes attached to’ > ‘to watch, see’, hence PIE *(s)h₃ek^u̯- ‘to see’ > PAnat. *sHeg^u̯‑, Core IE 1.*sek^u̯- (LIV²:525f.), Core IE *h₃ek^u̯- (NIL:370–383, LIV²:297f.)Similarly LIV²:525f., esp. 526 n. 7, but with a different primary meaning (“sich anschließen”). Furthermore, cf. Skt. netra- (n.) and nayana- (n.) ‘leading, conducting; eye (as the instrument of leading or guiding)’ (Monier Williams 1974a:511, Böhtlingk 1855a:4.313).
2. reflexive: ‘to attach oneself to’ > ‘to accompany, follow’, hence Core IE 2.*sek^u̯- [mid.] (LIV²:525f.)For parallels, cf. ModIr. leanaim ‘to follow’ < OIr. lenim ‘to hang to, adhere’ etc. (Buck 1949a:699) and ModE to stick to someone(’s heels), to join (oneself to) someone. Grestenberger 2016a:129–131 investigates the deponent status of 2.*sek^u̯- ‘to accompany, follow‘. Usually, the meaning ‘to accompany, follow’ is assumed to be the original one, cf. LIV²:525.
3. iterative: ‘to attach words to each other’ > ‘to recount, narrate’, hence Core IE 3.*sek^u̯- (LIV²:526f.)Cf. LIV²:526 n. 1. For the meaning ‘to narrate; to announce’, cf. Risch 1985a. The semantics of the root PIE *d^heh₁- to put, place; to say’ (LIV²:136–138, esp. 137 n. 1) is comparable. Cf. also Buck 1949a:1257 and in general 1253–1256 (s.v. speak, talk) and 1256–1258 (s.v. say).

In view of the attestations, it is recommendable to assign the meaning ‘to attach; to see’ to the oldest layer (Anat., Core IE). Since it is remarkable that parts of the Anatolian material morphologically correspond to formations of the semantic field ‘to follow’ (cf. PAnat. *sh₃óg^u̯-o-, Lyd. šawtaar-), the roots may be condensed to polysemous PIE *(s)h₃ek^u̯- ‘to see; to follow’, on the one hand, and PIE *sek^u̯- ‘to recount’, on the other.

Throughout all associated lemmata (PAnat. *sh₃óg^u̯-o-, HLuw. /tawasha-/, and Lyd. šawtaar-), common PIE inheritance is assumed.

The phonologically closest comparandum of the root PAnat. *sHeg^u̯- ‘to see’For the assumption of a biphonemic onset of the root PAnat. *sHeg^u̯‑, cf. above sub PAnat. Reconstruction.is the verbal root underlying PGerm. *seχ^w‑, *saχ^w- ‘to see’, Old Irish rosc ‘eye’, and Alb. 1sg shoh ‘to see’ and the nominal root Core IE *h₃ek^u̯- ‘eye’. Thus, a root PIE *(s)h₃ek^u̯- ‘to see’ with s‑mobile can be reconstructed.Cf. already Hoenigswald 1952a, esp. p. 183, and also Bjorvand & Lindeman 2000a:764, 785 (s.v. sjel). Oettinger 1979a:395f. sets up the root *sh₃ek^u̯- for Anatolian alone, but regards the anlauting *s- as secondary (ibd. 396), discarding the proposal of mobile s. ALEW I:27 is indecisive regarding *(s)h₃ek^u̯‑. On semantics and s‑mobile, cf. Shields 1996a:253. Cf. also PIE *(s)h₂ei̯k- in Eichner 1982a:18.

Furthermore, there are two further roots, viz. Core IE 1.*sek^u̯- ‘to accompany’ (LIV²:525f.) and 3.*sek^u̯- ‘to tell’ (LIV²:526f.). Since they can semantically be well connected with PIE *(s)h₃ek^u̯- ‘to see’ (cf. above), they may be combined with PIE *(s)h₃ek^u̯- ‘to attach’.
To explain cases without laryngeal reflex like Core IE *sék^u̯‑e- ‘to follow’ (not †sok^u̯‑e‑), two scenarios are possible:

1. PIE *sh₃ek^u̯- lost the laryngeal before the coloring effect of *h₃, viz. PIE *sh₃ek^u̯- > Core IE *sek^u̯- ‘to attach’
2. PIE *sh₃ek^u̯- lost the laryngeal after the coloring effect of *h₃, viz. PIE *sh₃ek^u̯- > *sh₃ok^u̯- > *sok^u̯‑, and consequently the e/o/Ø‑ablautable root vowel was reintroduced, viz. → Core IE *sek^u̯‑/*sok^u̯‑/*sk^u̯- ‘to attach’, cf., e.g., Bjorvand & Lindeman 2000a:764f.

The s‑less variant *h₃ek^u̯- ‘eye’ developed no verbal formations reconstructable.The verbal stems listed in LIV²:297f. can hardly be traced back to the proto-language. Thus, it is preferable to assign nominal status only to Core IE *h₃ek^u̯- ‘eye’, cf., e.g., NIL:373 n. 1 referring to GEW II:408.
In the way presented above, the two to three homophonous roots Core IE 1.*sek^u̯- ‘to see’, 2.*sek^u̯- ‘to accompany’, and 3.*sek^u̯- ‘to tell’ as well as the nominal root *h₃ek^u̯- may be united as PIE *(s)h₃ek^u̯- ‘to attach’.

[A.O.]