Published online by Cambridge University Press: 22 April 2026
In discussions of Proto-Indo-European phonology few attempts have been made to state the distribution of phonemes in the proto-language. Handbooks give lists of the phonemes, with examples of their occurrences; this is the presentation that we find in both Brugmann's and Hirt's descriptions of PIE phonology. From such lists, especially from those given by Brugmann, students can draw conclusions about the distribution of PIE phonemes; the conclusions of course are valid only to the extent that the compiler was aware of the need for a comprehensive list. But accurate descriptions of the distribution of PIE phonemes are lacking.
1 Brugmann gives the PIE ‘Lautbestand’, Gdr.2 1.92–3; examples occupy the next 700 pages, and these are followed by a discussion of clusters. Hirt catalogs the PIE ‘Laute’ in the first two volumes of his Indogermanische Grammatik.
This paper was presented in essence at the 1950 summer meeting of the Linguistic Society and profited from comments made there, especially from those of Émile Benveniste, Franklin Edgerton, Hans Kurath, and Werner Winter.
2 See for example Hirt, IG 1.215.
3 Hethiter und Hethitisch 81 (Stuttgart, 1947). Most Indo-Europeanists seem to assume that the lack of initial /r/ in Greek, Armenian, and Hittite is an innovation. See for example É. Benveniste, Problèmes de morphologie indo-européenne, Actes du quatrième congrès international de linguistes 62–3 (Copenhague, 1938). Part of the discussion of this paper, as reported in the Actes loc.cit., was as follows:
'M. Bartoli souligne une idée de M. Benveniste qui est importante et juste: qu'il n'y a pas eu
“racines” ario-européennes du type ereg-. En effet, le type ereg- est une innovation vis-à-vis le type reg- ...
‘M. Pokorny: Auch das Baskische kennt kein r- im Anlaut. Da Griechisch, Armenisch, Hettitisch ein armenoides ethnisches Substrat aufweisen, besteht die Möglichkeit, dass der Vokal vor r- eine dem Substrat zuzuschreibende Neuerung sei.‘
Benveniste's views at that time are not explicitly stated in the brief summary of his paper, nor did he comment on the remarks of Bàrtoli and Pokorny in his rejoinder.
4 Origine indoeuropéenne du redoublement attique, Eos 30.206–10 (1927), esp. 209 fn. 2. This view has been repeated by other scholars, without reference to Kurylowicz. Benveniste, for example, deals briefly with the subject in Répartition des consonnes et phonologie du mot, TCLP 8.27–35 (1939); I cite his entire statement so far as it is pertinent to the subject (33): 'L'accord du grec, de l'armenien et du hittite dans l'exclusion de r- initial prolonge en réalité un trait du plus ancien phonétisme indo-européen. Alors que i. e. l- initial est bien attesté ..., on ne trouve pas de racine ni de mot commençant sûrement par r-: la “prothèse vocalique” du grec et de l'armenien ainsi que diverses considérations de morphologie comparée conduisent à penser que les mots qui, en latin par exemple, ont r-initial, commençaient par un élément sonantique (šva) en indo-européen.1 Seules des combinaisons telles que *Әr- ou *sr- étaient normales. C'est donc à l'indo-européen ancien qu'il faut reporter le caractère que l'on croyait distinctif du grec ou du hittite ou limité aux langues de l'aire asianique.' Footnote 1 to this passage reads: 'Nous avons indiqué ce principe dans un exposé sur la morphologie indo-européenne au IVe Congrès des linguistes (Copenhague, 1936).' Benveniste's words were somewhat hasty. Not all words beginning in Latin with /r/ had an initial laryngeal in PIE: for some of them we must assume initial /w/.
5 Walde-Pokorny, Vgl. Wb. d. idg. Spr. 2.341–75 (Berlin and Leipzig, 1927–32), list 61 ‘roots’ with initial /r/; those mentioned here are cited in WP's transcription. Among the ‘roots’ listed with initial sr- (2.702–6), the /s/ is nowhere definitely ‘movable’; these ‘roots’ have therefore been excluded from the present study.
The 28 entries for which the evidence is too scanty to be seriously considered are the following, in the order of their listing by WP: rab-; rē-; rē(i)- ‘give’; rei-, roi-; reiq-; reig-; rēig-; reid-; reis-; reu- ‘seek’; reubh-; reu-men; reus-; rebh-; reg- ‘choose’; reg- ‘see’; reĝh-; regh-; red-; rēd-; rēp-, rӘp-; rebh- ‘move’; rebh- ‘arch over’; rem- ‘strong’; remb-; renq-; rent-; rendh-.
6 See Kluge-Götze, Etymol. Wb. d. deutschen Spr.11 488–9 sub Rübe (Berlin and Leipzig, 1934).
7 Aside from these three ‘roots’, such variation is found also in req-, rēq- ‘arrange’ beside red- ‘arrange’ (listed in fn. 5). One may also compare req- ‘pole’ and reig- ‘pole’, both of which have reflexes only in Baltic and Germanic, with rēt-, rōt-, rӘt- ‘pole’, with reflexes cited from Germanic, Slavic, and Latin. I use the term ‘PIE root’ to refer to a meaningful element composed of two consonants, which may be accompanied by a vowel in various ablaut grades, e.g. /tm-/ (also /tern-, torn-/), /dhr-/, /ky-/. The second consonant of a PIE root shows no variation. Of all the forms listed by WP, there is a possibility of such variation only in keg-, kek- (1.382), ĝep(h), ĝebh- (1.570–1), and
-,
- (1.675–8, i.e. /
-,
-). For the first two of these one may assume sandhi variants, for the third two different PIE roots similar in form and meaning.
8 Entries with original initial /w/: reṷos; renghus; rug-;
.
9 Entries with original initial /s/: reg- ‘dye’; rēp-.
10 These are: rās-; rē(i)-, rei-; rei-; reidh-; reibh-; reir(ēi)-; reu-, rēu-,
; reu-; reug-; reugh-m(e)n-; reudh-; reu-men-; reĝ-;
; rēd-, rōd-, rӘd-; rep-; rem- ‘rest’; rezg-; roino-. Only one of WP's ‘roots’ is thus left unaccounted-for: reth-.
11 P. Persson, Beiträge zur indogermanischen Wortforschung 637–8, 773–4, 839–42 (Uppsala, 1912); see also WP 1.142–3.
12 E. H. Sturtevant, IH Laryngeals 40–1 (Baltimore, 1942), connects Hitt. barrai ‘break, pulverize, ruin’ with Gk. άρόω, Lat. arō ‘plow’, and posits the initial laryngeal /x/. On p. 62 he again connects these words, but admits the possibility of having here initial /γ/. Reflexes of WP's root er- agree in meaning much more closely with cognates of Hitt.
; I do not exclude the possibility that the words meaning ‘plow’ developed from this root, but suggest that it was a secondary development.