No CrossRef data available.
Published online by Cambridge University Press: 22 April 2026
Ever since the seventies of the last century scholars have given a great deal of attention to various kinds of evidence that has seemed to indicate a reduced grade alongside of the zero grade in the light bases. The early discussions concerned chiefly Osthoff's theory of the ‘nebentonige Tiefstufe’, but one should note also Chr. Bartolomae's attempt to establish the presence of ə in the e-series. While no one today would support either theory in its original form, some of the material collected by Osthoff and Bartholomae still has significance.
1 Hermann Osthoff, MU 4 (1881), preface and pp. 1–406; Karl Brugmann, Grundr. 1.252–5 (1886); Paul Kretschmer, KZ 31.326–45 (1892); Jakob Wackernagel, Ai. Gramm. 1.97–9 (1896); and references contained in these works.
2 BB 17.108–24 (1891).
3 The most important contributions are Herman Hirt, Der indogermanische Ablaut, especially 11–9 (1900); Osthoff, MU 6, especially 208–19 (1910); Hermann Güntert, Indogermanische Ablautsprobleme, Untersuchungen über Schwa Secundum, einem zweiten indogermanischen Murmelvokal (1916); Hirt, Der indogermanische Vokalismus, especially 76–102 (1921); Alois Walde, Streitberg Festschrift 152–99 (1924). Of secondary importance are Manu Leumann, Lateinische Grammatik, especially 69 f. (1928), and Antoine Meillet, Introduction, Ed. 7, especially 102 (1934). Among arguments against assuming such reduced vowels may be mentioned Brugmann, Grundr. 12.393 f. Anmerkung (1897), IF 28.369–71 (1911); Sigmund Feist, DLZ 40.355–60 (1919); Franklin Edgerton, Lang. 10.235–65 (1934), 19.83–124 (1943); and Walter Petersen, Lang. 14.39–59 (1938).
4 Since Walter Petersen's article answers Hirt only, it seems to me to miss the mark.
5 KVG 141 f. (1904): ‘e bei Geräuschlauten wurde irgendwie reduziert, ohne dass die Silbe verloren ging .... Es fiel aber mit dem intakten e wieder zusammen.‘
6 Lang. 14.41 (1938).
7 Brugmann and others set up such formulae as *kwtwr- in cases like this, but they can scarcely intend to say that such consonant groups actually existed in Proto-IE or in any of the historical languages. At any rate the impossibility of precisely this group is now clear from Edgerton's paper in Lang. 19.96–8 (1943); before a consonant we should have in Proto-IE *kwtuwr- and before a vowel *kwtur-.
8 So Brugmann, KVG 520; Sommer, Hdb.2 48.
9 See below, §18.
10 See above, fn. 3.
11 It seems particularly important to do this, since W. Petersen (Lang. 14.43 f.) has claimed that the evidence alleged for our reduced vowel is different in kind from that on which the reconstruction of other IE vowels is based.
12 Comp. Gramm. of the Hitt. Language (1932), The Indo-Hittite Laryngeals (1942), and the articles there cited.
13 Lang. 19.120–1, §§90–3.
14 Cf. Austin, Lang. 18.22–5 (1942) and Sturtevant, The IH Laryngeals 29 f.
15 HG 98 f.; IH Laryngeals 31 f. For Gk. πίτνημι, etc., see fn. 24.
16 Brugmann and others would write IE kwétwrres. My reasons for preferring to write ьra, ьna, etc. instead of rra, nna, etc. are stated in §§64–72.
17 Ber. d sächs. Ges. d Wiss. 53.89–100 (1901).
18 Lang. 19.98, §31b (1943).
19 So Osthoff, MU 6.212; and Meillet, Le slave commun 49. Holger Pedersen, KZ 38.420, identifies these West Slavic forms with OCS cetyre on the assumption that after the palatalization of unaccented initial kwe to če the consonant assimilated the vowel to ь. He has to admit, however, that the syllable če often remains under undetermined conditions. If we trace the Proto-Sl. ь to IE ь, on the other hand, OCS četyri causes no difficulty; OCS četvrьtь, Russ. četvértyj ‘fourth’ proves that a source for analogical e in this word was present in Proto-Sl.
20 For the semantics, cf. the use of Eng. learn in the same sense as teach.
21 KZ 43.185 and fn. 1 (1910).
22 Glotta 4.349 (1913).
23 See fn. 33.
24 See §17. The t instead of α in this word, σκίδναμαι, and some other verbs in -νā- is ascribed by Güntert, Schwa Secundum 31, to the well-known alternation of the present suffix neu with nā (IE pbtneu- yielded Gk. *πιτνῦ- and this induced *πιτνᾱ- instead of *πaτνᾱ); at any rate we actually meet π∊τάννυμι and σκ∊δάννυμι (with analogical c from the aorist stem). Quite possibly, however, some modification of my formula in §16 will seem preferable to some scholars. The important point is that the ι of these verbs obviously belongs to the same pattern.
25 One other explanation of Latin a suggested in Petersen's article (p. 49) is even more absurd than this one. He says that crassus owes its vowel to the rhyme word lassus on account of the obvious association of words for ‘fat’ and for ‘lazy’; but lassus never means ‘lazy’ and crassus means ‘fat’ in only a minority of its occurrences.
26 Esquisse d'une grammaire comparée de l'arménien classique 21 (1903); 2d ed. 42 (1936).
27 KZ 39.416 (1904).
28 Lang. 11.175–84, especially 176–81 (1935).
29 PBB 16.236 f. (1892).
30 With OE styde compare Gk. στ∊υται, Skt. sthaviras, etc.; with OE dyde compare Lat. duim.
31 Gr. Gramm. 699 fn. 3.
32 Cf. Brugmann, Grundr. 22.3.292–5.
33 Leumann, Lat. Gramm. 155 and references; cf. Sommer, Hdb.2 235, who considers the change Indo-European.
34 Gr. Gramm. 700 f.
35 VGKS 1.39,160, 2.256.
36 Güntert, Schwa Secundum 29 and references.
37 Osthoff, MU 6.213 and ref.
38 CGG 102. Prokosch assumes (ibid. 54 f.) that in this word and in dis- ‘asunder’ IE d remained in Goth. This irregularity seriously impairs the evidential value of Goth, du; perhaps its apparent correspondence in meaning and vocalism with Arcad. βθρδα is accidental.
39 Sturtevant, The IH Laryngeals 67 f., 71.
40 Sturtevant, The IH Laryngeals 55 f.
41 Sturtevant, The IH Laryngeals 55 f.
42 So Weidner and Friedrich, AOF 6.223; Sommer, ZA NF 12.46 f., doubts the transcription.
43 IF 7.54 (1897).
44 H. Güntert, Schwa Secundum 53.
45 Osthoff, MU 6.211 f.; H. Güntert, Schwa Secundum 54.
46 Osthoff, MU 6.216.
47 Sievers, Ags. Gramm.3 13 §31A, suggested that the short y of this word might be due to the long y of the contrasting term lȳtel, but that seems difficult.
48 Feist, Vgl. Wörterbuch d. got. Spr.3 s.v. *griþs, would avoid IE ь by tracing Goth. i and Lith i to IE i; but this involves a complete separation of Lat. gradior.
49 MU 6.1–69, especially 34–69.
50 Cf. βραχύς βραχίων ‘short’ beside Lat. brevis brevior. The troublesome gravis ‘heavy’ instead of *garvis, like Gk. βαρύς, or *gervis, with corresponding full-grade vowel, is probably due to contamination with the nearly synonymous grandis ‘large’ (cf. Skt. comp. gárīyas-). Otherwise, IH Laryngeals 32.
51 Osthoff, MU 6.210.
52 Güntert, Schwa Secundum 87; Walde-Pokorny 2.374.
53 Pedersen, VGKS 1.77, 2.616; Walde-Pokorny 2.540.
54 Güntert, Schwa Secundum 44, 56, 61.
55 Of course Skt. an- is ambiguous; it is only the accent of the word that suggests a weak first syllable.
56 Walde's doctrine (Über älteste sprachliche Beziehungen zwischen Kelten und Italikern 54 f.) that IE n yielded Osc.-Umbr. an- in initial position is improbable on account of the well established agreement between these languages and Lat. in showing en and em for IE n and m in medial and final position. No better than this is Buck's suggestion, Gramm. of Osc. and Umbr. 65, that Osc.-Umbr. anter contains a prior element cognate with Gk. ἀνά.
57 Gr.-Lat. etymologisches Wörterbuch 1.30 (1877).
58 Cf. Brugmann, Grundr. 22.1.329 f. Prellwitz' connection of ἂντρον with Skt. ániti ‘breathes’ etc. scarcely deserves mention.
59 See Sturtevant, The IH Laryngeals (1942), and the earlier literature referred to on pages 15–22 of that book. The theory was originated in 1878 by Ferdinand de Saussure, Mémoire sur le système primitif des voyelles dans les langues indo-européennes.
60 For a refutation of Kurylowicz' assumption of *dhAьtós and the like, see Lang. 17.181 f. (1941).
61 So Saussure, Mémoire 135 (1878); Cuny, Rev. de Phonétique 2.101 f. (1912); Pedersen, Hittitisch u. die anderen indoeuropäischen Sprachen 179 f. (1938). Cf. Edgerton, Lang. 19.107–8, §63.
62 Rev. de Phonétique 2.101–32, esp. 101–4 (1912).
63 Edgerton, Lang. 19.94–103, esp. §§27, 28, supplies the proof for the statement in this sweeping form.
64 Evidence for reduced-grade IE dhəhy- has recently been extracted from Goth. daddjan; see H. L. Smith Jr., Lang. 17.93–8 (1941); Sturtevant, The IH Laryngeals 81.
65 Lang. 14.107 f. (1938), The Indo-Hittite Laryngeals 35 f., 72, 73.
66 IH Laryngeals 71, 73.
67 CP 36.356–64 (1941), IH Laryngeals 78–80.
68 IH Laryngeals 79–81.
69 IH Laryngeals 76–86.
70 Op.cit. 87.
71 Lang. 16.273–84 (1940), IH Laryngeals 87–9.
72 Lang. 17.184–6 (1941).
73 I assume that Edgerton's formulae for two semivowels after a consonant (Lang. 19.97–8, §31) are valid for IH, with the modification suggested in §§63–72; for Edgerton's [iya] and [uwa] I substitute [ьbya] and [ьwa]. Upon the change of ьA to ə in pre-IE, stəwьrós and dhəyьnéu- (in Edgerton's notation stəwrrós and dhəynéu) automatically became IE stəwrós and dhəynéu- by Edgerton's §26.
74 See Wackernagel, Ai. Gramm. 1.81–3 and references. Cf. H. Pedersen, Hitt. u. die anderen IE Sprachen 185; Sturtevant, IH Laryngeals 75–8.
75 The contrast between Skt. h and Gk. γ does not interest us at the moment; one way of explaining it has been suggested in IH Laryngeals 86.
76 IF 12.182 f. (1901).
77 IH Laryngeals 75.
78 Cf. Wackernagel, Ai. Gramm. 1.82.
79 IH Laryngeals 75 and cross references.
80 Edgerton assures me that it is easy to find analogical explanations for these Skt. plural forms and for all other supposed zero grade forms in Skt. from the roots dā- and dhā-. The chief reason why I find such explanations unsatisfactory is the existence of confirmatory evidence for zero grade in related languages.
81 Walter Petersen, Lang. 14.41, says: 'Of course the adherents of the schwa-secundum theory maintain that e is here IE ь, although otherwise they assume that ь becomes almost everything else than e in the various IE languages.' Edgerton, Lang. 19.120, §89 says: ' . . . this magic word is used to exorcise any vowel in any IE language which the theorizer finds troublesome; even accented vowels in Gk. are often thus “explained”. No consistent picture is presented, and there is great discrepancy among the several forms of the belief.'
82 Lang. 10.235–64 (1934); 19.84–124 (1943).
83 Initial and final αρ, αλ.
84 Lang. 10.237–44 (1934), 19.87–8, 119–21, §10 and fn. 14, §§88–92 (1943).
85 AJP 50.360–9 (1929), TAPA 60.33 f. (1929), Lang. 7.170 (1931), HG 114–6 (1933), IH Laryngeals 26 f. (1942).
86 On the assumed simplification of ww, cf. below fn. 88.
87 Lang. 19.87–8, §10.
88 If this occurred in Indic we should speak, no doubt, of shortening of a consonant after another consonant. If it occurred earlier, it was probably a part of a larger process by which all long consonants were shortened. It is well known that the process of reconstructing Proto-IE has not led to the positing of long consonants. Instead we have had to assume the secondary development of s between two dental mutes that were brought together by morphological processes, and there is at least one instance of the shortening of a long consonant between syllablcs, namely IE esi ‘you are’ (Skt. asi, Gk. el).
89 Bloomfield JAOS 16.xxxiv f. (1896); Wackernagel, Ai. Gramm. 1.278–80 (1896).
90 IH Laryngeals 32.