*trisu

While discussing the Anatolian multiplicatives of the numeral ‘two’, Eichner 1992a:61f. suggests that the marker -su originates from an endingless locative singular of a thematic *-u̯o- derivation, i.e. *du̯isu̯é ‘twice’. However, a more straight-forward explanation (EDHIL:873, LIPP II:803, 818 n. 1) views the multiplicative of HLuw. (“3”)tara/i-su-u and Lycian B trisu ‘three times’ as originating from the locative plural of the numeral ‘three’ with the regular case ending -su, which would have been lexicalised by hypostasis, i.e. “in twos, in threes, in fours” etc., as an adverb. This is considered here to be the superior hypothesis. Therefore, it can be analysed as the locative plural of the numeral *téri-/tri-´ + *-su.

[AH.B.] [D.S.] [E.R.]

For the lemma head see PAnat. *téri-/tri-´.

If we interpret the form PLuw. *trisu ‘thrice’ as a locative, there are two very close semantic comparanda. Firstly, the prepositional phrase OIr. fo thrí *‘(lit.) under three’ is used in multiplicative sense ‘three times’, cf. Buck 1949a:944, 947, Thurneysen 1946a:250.According to eDIL s.v. fo, fa, fá (i), this construction is attested early.Secondly, Lith. trisù ‘in a set of three’ is lexicalized from the locative and makes a full formal match with Lyc. B trisu, Luw. t(a)risu ‘thrice’. According to Forssman 2003a:147, the semantics of Lith. trisù started as locative *‘in three (vel sim.)’, which developed hence further into an iterative adverb *‘three times’ and finally became a quantitative adverb ‘by three, in a set of three’. Thus, it is perfectly possible on semantic grounds that a locatival form PIE *trisú ‘in three (vel sim.)’ is the preform of Lyc. B trisu, Luw. t(a)risu ‘thrice’.Cf. also Hoffmann 1965a:252f. on the connection of locatives with multiplicatives.
As a consequence of the formal difficulties of this solution (cf. below), Dunkel offers another but semantically similar interpretation (LIPP II:804, 818) and takes it as a univerbation *trís=u ‘(lit.) thrice=there’ consisting of the multiplicative PIE *tris ‘thrice’ (cf., e.g., Ved. trís ‘id.’) and the deictic particle *u ‘there’ (LIPP II:817–822). The evidence for the particle, however, is mainly “aptotological” and thus the conclusions drawn thereof have to be handled carefully.
Thirdly, Eichner 1992a:61f., 74, 93 proposed the preform *tris-u̯é, an endingless loc. of a u̯o-stem derived from *tris ‘thrice’ (cf., e.g., OHG driror ‘thrice’, YAv. θrižuuat̰ ‘id.’). This solution is quite possible (cf. EWAhd II:786.34–787.2), but the suffix of the adduced evidence, the (Late) Young Avestan example is more complex than just *-u̯o-, viz. YAv. -uuat̰ < *-u̯n̥t.

All three proposals (loc. pl. *tri-su, *tris + loc. ptcl. *u, loc. sg. *tris-u̯é) are based on locative semantics, which makes them equally possible in semantic respect.

PLuw. *trisu ‘thrice’ formally looks like the locative (plural) of the numeral ‘three’, cf. the respective forms sub Transmission above. However, as Dunkel (LIPP II:803 n. 1) points out, the loc. pl. ending *-su has been unattested in Anatolian so far. It still seems natural to assume that some predecessor of this ending (cf. LIPP II:717: *so, *su ‘in the midst of; together with’) already existed in the (pre-)proto-language.
Nevertheless, in view of the persuasive agreement of Lith. trisù ‘in a set of three, by three’ and PLuw. *trisu ‘thrice’ in both semantics (cf. above) and phonology, it is plausible to a certain degree that PLuw. *trisu ‘thrice’ goes back to PIE *tri-sú, (a preform of) the locative of the numeral ‘three’.
Phonologically and semantically it is not unreasonable either to follow LIPP II:804, 818 in segmenting PLuw. *trisu into PIE *tris ‘thrice’ and PIE *u ‘there’ (LIPP II:817–822), cf. above. However, this solution, as mentioned above, leads deep into the field of “aptotology” and may thus be not commonly accepted.
The third solution presented above, viz. the (endingless) locative *tris-u̯é of a u̯o-stem derived from *tris ‘thrice’ (cf. Eichner 1992a:61f., 74, 93) is acceptable in many aspects, but there seems to be no evidence for an e-grade of thematic vowels apart from the vocative.

Any of the three proposals has its own disadvantages, be it morphological (loc. ending *-su), lexical (particle *u ‘there’) or morphonological (thematic vowel *-e-). In sum, one may be prone to the first solution, viz. the locative *tri-su.

[A.O.]

Ved. loc. triṣú ‘3’
PIIr. *tris ‘thrice’, cf. Ved. trís, Av. θriš ‘id.’
EWAia I:675–678, Grassmann 1996a:555f., Bartholomae 1904:811, Emmerick 1992a:166f. (OInd.), Emmerick 1992b:293–295 (Iran.)
YAv. θrižuuat̰ ‘thrice’ (hapax, V.9.31)
Bartholomae 1904:812, Martínez & de Vaan 2014a:69, Emmerick 1992b:329

Gk. dat. τρισί, Myc. dat./loc. ti-ri-si ‘3’
Gk. τρίς ‘three times’ (Il.+)
GEW II:921f., DELG:1131, EDG:1502, Waanders 1992a:371, LSJ:1822

Arm. erir ‘third’
Olsen 1999a:482, Winter 1992b:349

OLat. terr (Pl.Bacch.1127), Lat. ter ‘three times, thrice’
WH II:668f., EM:701, EDLIL:628f., OLD²:2121, Coleman 1992a:416

PGerm. *þrizu̯ār ‘thrice’, cf. ON þrys-var, OHG driror (Murb. H.) etc. ‘id.’
EDPG:—, de Vries 1977a:624, EWAhd II:784.42–787.9, Ross & Berns 1992a:649f.

Lith. dial. adv. (< loc.) trisù ‘in a set of three, in three’
OCS loc. trĭxŭ ‘3’
SEJL:689f., Kurschat 1968–1973 IV:2526

Alb. trish, tresh adv. ‘threefold’
Demiraj 1997a:387f., Meyer 1891a II:435