The University of Texas at Austin; College of Liberal Arts
Hans C. Boas, Director :: PCL 5.556, 1 University Station S5490 :: Austin, TX 78712 :: 512-471-4566
LRC Links: Home | About | Books Online | EIEOL | IE Doc. Center | IE Lexicon | IE Maps | IE Texts | Pub. Indices | SiteMap

Proto-Indo-European Phonology

Winfred P. Lehmann

7. The Old High German r-preterites

7.1. The forms attested in our documents

In the last three chapters I have cited forms which give evidence that laryngeals or their reflexes survived into Gmc. in the neighborhood of resonants. I shall now examine Gmc. forms, some of them from roots whose forms have already been discussed, that have preserved an unexplained r in some rare forms.

Preterite forms with -r- are found in OHG for stōzan ‘push’, scrōtan ‘cut’, bluozan ‘offer’, būan ‘dwell’, a preterite participle for spīwan ‘spit’, and a preterite participle as well as preterites for scrīan ‘cry’. The forms are: anasteroz 3d sg. ind., anasterozun 3d pl. ind. (and the later glosses steraz, stiriz, farsterc), kiscrerot 3d sg. ind., pleruzzun 3d pl. ind., capleruzzi 3d sg. subj., biruun 3d pl. ind., biruuuis 2d sg. subj., pespiren pret. ptc., scrirun 3d pl. ind., scriri 2d sg. subj., and erscrirena pret. ptc.1 The preterite forms made from the cognates of these verbs, when not weak, follow in the other Gmc. dialects the usual patterns of strong verbs belonging to classes VII, e.g. OS stōtan, steot, steotun, gistōtan ‘push’, and I, e.g. OS stīgan, stēg, stigun, gistigan ‘climb’. In OHG we find in addition to the r-preterites, regular preterite forms, some of them weak, for these verbs. From the limited number of occurrences of r-preterites and their restriction in these verbs to one Gmc. dialect one can infer that they either were aberrant developments from regular patterns, or survivals from IE form categories that otherwise were lost in OHG. Although few r-preterites are attested, they are not limited to any single monastery, author, or scribe.

7.2. Proposed explanations

Two types of explanation have been suggested for the origin of the r: 1. that it is a phonetic development in OHG; 2. that it has survived from a tense formation that elsewhere in OHG has been lost.

The oldest explanation falls into the first classification. Lachmann suggested in his notes to the Nibelungenlied that the r was a hiatus-breaker; his explanation has been adopted by Müllenhoff, Weinhold, Zarncke, and Flasdieck.2 Scholars who adopt it cite in support of it forms with epenthetic r; Zarncke, for example, cites from the Ludwigslied wolar, abur and also some German dialect forms where r is found internally in words: rueri ‘peaceful’ and strārete ‘strewed’. Loewe objected to this explanation and his arguments are still pertinent: why is r found only in certain forms, and why was it introduced in a diphthong where there is no hiatus?3 Moreover, the forms cited in support of this theory are random examples that do not belong to one category, whereas the OHG preterite forms are part of one morphological system. One may readily find examples of intrusive sounds, such as the r of Eastern American English, but such sounds fall into specific phonetic patterns rather than morphological ones. If an intrusive r was used as a hiatus-breaker in OHG, as it is in Eastern American English, we should not expect to find it established only in strong verb preterites, but also in other forms with a similar succession of vowels. Although this explanation may be the best suggested—Flasdieck, op.cit. 275, arrives at it by eliminating other suggestions—it is very weak.

In other theories, various forms, later lost, have been cited as the origin of the r-preterite. Osthoff, Schmidt, Loewe, and Janko consider the r-preterites remnants of reduplicated perfect forms; -spiren is derived through various intermediary forms from *spe-spu-me.4 Among the weaknesses of this theory is the lack of reduplicated preterite forms in OHG; there is no evidence that any OHG verb forms developed from reduplicated forms like *spespume. Moreover the dissimilations are unusual; Loewe posits a dissimilation of *stestaute to *stesaute and of *spespume to *spesume. In the Gmc. languages st and sp are treated as units, both in sound shifts and in alliterative verse; consequently dissimilation of sp to s is as little likely as dissimilation of b to p.

Knoblauch suggested origin for scrirun in an s-aorist; his suggestion has been adopted by Schmidt, Streitberg, and Brugmann.5 His theory is very weak for we have no other evidence that an s-aorist was ever found in the Gmc. languages. Proponents of this explanation must further explain the r of the pret. ptc.; they may assume spread by analogy in the form erscrirena, but we have no evidence that r-preterites were made from spīwan from which the r may have spread to pespiren.

Feist suggested that the r-preterites started from forms like OIcel. *rōan : rerō; on the pattern of this was formed blōzan : *bleroz. Michels and Karstien have adopted his view.6 But again the objection may be raised that the forms from which r-preterites are said to develop are not found in OHG.

None of the theories proposed, or combinations of theories,7 can be considered satisfactory. Karstien characterized the problem as ‘ein ganz unleidliches gebiet,’ but favored Feist's theory. Hirt made no choice among the various theories.8 It is obvious that no adequate explanation has been suggested.

7.3. The two groups of r-preterites

The r-preterites fall into two groups according to form: those with -eru-, or u-umlaut of -e- to -i-, e.g. pleruzzun, biruun; those with -ir-, e.g. scrirun. The differences in preterite formation agree with the difference in verb classes. Preterites with -eru- belong to seventh class verbs, those with -ir- to first class verbs. It is further noteworthy that the r-forms are found for seventh class verbs in the singular and plural, indicative and subjunctive of the preterite, but not in the preterite participle. For first class verbs they are found only in the plural indicative and in the subjunctive of the preterite, and in the preterite participle, that is, in those forms with zero grade of the root vocalism.

7.4. The origin of the preterite of 7th class verbs in NWGmc

The origin of the preterite formation in NWGmc. of seventh class verbs is disputed; the ablaut grades found in these preterites have been the subject of much discussion. For although these preterites follow a definite pattern, this pattern is not found in other IE dialects. Its origin must be explained from evidence found in Gmc.

Two theories have been proposed: 1. that the seventh class preterites are from reduplicated forms like those of Gothic; 2. that they and the seventh class presents developed from forms with ablaut grades different from those of the first five classes.9 According to the second theory the preterite forms of the seventh class verbs are based on forms with an ablaut grade contrasting with that of the present, which have also acquired a contrast in meaning. In formation then, the seventh class differs from the first five classes only in its development from an ablaut pattern that was less widespread than the /e : o : e/ ablaut of the first five classes. This theory accounts both for forms and for the development of meaning. According to the elaboration of the theory which Prokosch presents, the contrast in meaning between present and preterite developed in ablauting forms contrasting between punctual and durative force.10 The strength of the first theory lies in the presence in Gothic of reduplicated preterites for seventh class verbs; its weakness lies in the impossibility of suggesting a development in accordance with sound laws from such forms to the preterite forms that occur in NWGmc. Consequently I follow the theory in which the preterite is derived from contrasting ablaut forms.

Prokosch, CGG 175ff., suggests that the verbs belonging to class VII are made primarily from diphthongal heavy bases. For the present tense he posits the reduced grade in all Gmc. dialects, for the preterite in NWGmc. the normal long e-grade; e.g. the present tense of OHG stōzan is from Gmc. staut- which developed from PIE stəud-; the preterite tense from Gmc. stēut-, which developed from PIE stēud-. Under the terms of the laryngeal theory the alternance would be stated as between /steXwd-/ and /steXwd-/.

7.5. Evidence for laryngeals in cognates of r-preterites

In chapters 4 to 6 we have noted that laryngeals, or reflexes of laryngeals, survived in Gmc. dialects long enough to produce modifications of sounds. According to Prokosch's theory, at least the four seventh class verbs with r-preterites developed from forms with laryngeals; examination of the r-preterite forms may disclose the treatment of these.

OHG stōzan, Goth. stáutan, OIcel. stauta ‘push’ are cognate with Skt. tundate, tudáti, Lat. tundō ‘push’. Although WP assume an IE root steu-, this is not adequate on the basis of Prokosch's analysis of seventh class verbs. WP recognize as related forms OIcel. stúta ‘wear down’, stútr ‘worn-down horn’; the ū in these forms would be best explained from the zero grade of /-eXw-/. Consequently Persson's suggestion, Beitr. 714 fn.1, that steu- is derived from steA- ‘stand’ is very plausible. WP object to Persson's suggestion on the dubious grounds that the meaning ‘push, strike’ is not to be derived from that of ‘stand, be stiff, stand up stiffly’, but are unable to separate the developments of their two roots. Since forms from the ‘two roots’ overlap, I consider WP's objection weak and derive OHG stōzan from the PIE root /steA-/ plus /w/ and /d/ determinatives.

OHG scrōtan ‘cut’, OE scrēadian ‘cut off’, OIcel. hrjóþa ‘peel off’ are from PIE /(s)ker-/ ‘cut’. Assumption of a laryngeal base is supported by Lat. scrūtor ‘examine’ and OE scrūd ‘dress’ which point to zero grade of /-eXw-/. That the root /(s)ker-/ was extended by means of laryngeals is clear from Skt. kr̥ṇāti ‘harms’, which is derived from PIE /kr-n-eX-/. I assume that the Gmc. verbs developed from /(s)kr-eX-/ plus /w/ and /t/ determinatives.

OHG bluozan ‘offer’, Goth. blōtan ‘honor’, OIcel. blóta, OE blōtan ‘offer’ are related by WP to Lat. flāmen ‘priest’, and are therefore derived from /bhleAd-/. This derivation fails to account for the usual preterite forms, OHG plioz and OE blēot. To account for them we must assume, as did Prokosch CGG 182, that the preterite was made on the analogy of second class verbs. I assume that the preterite forms developed from /bhleA(w)d-/ which contrasted with /bhleAd-/ of the present.

OHG būan, OIcel. búa, OE būan ‘dwell’ are from PIE /bhewX-/. Cognates have been listed above, 4.41a. The forms of this verb are not parallel to other seventh class verbs in the Gmc. dialects or in the PIE etyma. The PIE root was clearly /bhew-/, usually enlarged by a laryngeal to /bhewX-/; the pattern of other seventh class verbs is /eX/ plus resonant. The present tense vowel ū may be from the zero grade of /ewX/ as well as /eXw/. Except for the OHG r-preterites and OIcel. bió the preterite forms are weak, e.g. OE būde. Prokosch, CGG 182, finds that OIcel. bió is the regular e-grade to be expected in the preterite of seventh class verbs, and concludes that this verb, although having short e plus w in the normal grade, was shifted to the seventh class by analogy. We must find the origin for OHG biru- in such an analogical form. Although the IE cognates of biruun and biruuuis give us no evidence for assuming an etymon /bheXw-/, the presence of seventh class preterite forms indicate that such forms were made from this root.

All r-preterite forms made from seventh class verbs are found where we should expect an earlier sequence /-eXw-/; in three of the verbs this sequence was followed by an obstruent of the stem, in būan by the endings.

r-preterite forms of the first class verbs on the other hand are found where we expect zero grade of the root vocalism, that is, i. Both first class verbs with r-preterite forms developed from roots with laryngeals.

Cognates of spīwan give clear evidence for laryngeal, see above 5.4b. In OHG, spīwan ‘spit’ is a regular first class strong verb with the principal parts spīwan, spē, spiuun, gispiuuan.11 Since the PIE root differs from that of other first class verbs, these forms must have been made in Gmc. by analogy with other first class verbs. I follow Collitz' suggestion, Mod. Phil. 15.103ff., that spīwan < *spiXwan became a first class strong verb when PIE ei became Gmc. ī; the other forms were then made on the pattern of first class verbs. I assume that pespiren on the other hand developed from a verbal adjective in -no- which was older than these analogical forms, /spiX(w)-e-no-/.

OHG scrīan ‘cry’ is a regular first class strong verb, with, apart from the r-preterite, one aberrant form, the preterite singular screi. The vocalism of screi can only be explained by assuming lengthened i; as has been indicated above, this developed from a form with laryngeal. The long i of Lat. crīmen supports derivation from a laryngeal base. Since scrīan is a first class verb, not a seventh class verb, I assume PIE /(s)kreyX-/. Moreover there are only zero grade forms giving evidence of a laryngeal; normal grade forms with a sequence /eXy/ are less likely to preserve evidence of laryngeal than are forms with a sequence /eyX/. I therefore derive scrirun from /skryX-/.

7.6. The conditions under which r developed

The sequence assumed here for seventh class verbs is PIE /eXw/ [eXu], for first class verbs /yX/ [iX]. I suggest that in these sequences the laryngeals were preserved, and that their reflexes fell into the OHG r-phoneme.

To support such an assumption we must find evidence for the preservation of the laryngeal, and for its representation as r in OHG.

We have already noted that Gmc. is remarkably conservative in the treatment of laryngeals when these were contiguous to resonants. The upper dialects of OHG give further evidence of such conservatism.

In Alemannic we find the forms āmar rather than jāmar ‘suffering’ and enēr rather than jenēr ‘that one’.12 On the other hand the initial j is preserved in jesan ‘ferment’, jetan ‘uproot’, and jehan ‘say’. While the cognates of jetan and jehan have not been determined, it is clear that (j)enēr is cognate with Skt. yás, Gk. ὅς, (j)āmar with Gk. ἥμερος ‘tame’, but jesan with Gk. ζέω ‘seethe’. We may also adduce OHG joh ‘yoke’, Gk. ζυγόν as further evidence for suggesting a correlation between Alemannic and Greek in the treatment of ‘initial y.’ Where j-was dropped in Alemannic we find a rough breathing in Greek; where it has remained, we find ζ. Sapir13 has demonstrated that rough breathing in such Greek words corresponds to PIE [y] preceded by voiceless laryngeal. It seems plausible to assume for Alemannic too that the initial sound of (j)āmar and (j)enēr had become voiceless after a voiceless laryngeal and was then lost.

We can only tentatively suggest reasons for the representation of the reflex of laryngeals as r in OHG. For as yet only general phonetic descriptions have been suggested for the laryngeals. Evidence in one Hittite text points to a phonetic similarity with r; in this text waḫnu- ‘turn’ is confused with warnu- ‘burn’. Götze and Pedersen have concluded that the Hittite reflex of laryngeal in waḫnu- was pronounced like the Hittite r and they suggest that , was a weakly articulated spirant; see above 3.7.

We may note that OHG scorra ‘crag’, scorrēn ‘jut out’ has an inexplicable double r. WP derive these forms from the root from which scrōtan developed but are unable to account for the lengthened r. Since it is clear from Skt. kr̥ṇāti that the base is /skerX-/, scorra seems to provide further evidence for the development of the reflex of laryngeal to OHG r.

7.7. The history of the r-preterites after OHG

It is difficult to say how wide-spread was the r-preterite formation in Gmc. The r-preterites did not fit into the regular patterns of the Gmc. verb and were eventually displaced by the regular forms. In scrīan, spīwan, scrōtan, and stōzan the strong preterite forms predominated; in būan and bluozan the weak. Very few r-preterites survived into OHG; some of these were obscure to the later scribes and were corrected to the regular forms. By MHG times traces of the formation were almost obliterated.


1 J. Schatz, Althochdeutsche Grammatik, (Göttingen, 1927) lists the forms and gives references for them on pages 282 and 294; he also lists the regular preterite forms of these verbs. For the preterite forms of spīwan and scrīan that have survived into MHG, see H. Paul and E. Gierach, Mittelhochdeutsche Grammatik14 117 (Halle, 1944).

2 K. Lachmann, Zu den Nibelungen und zur Klage, 66 (Berlin, 1836); K. Müllenhoff, Angebliche aoriste oder perfecta auf r im altnordischen und hochdeutschen, ZfdA 12.397-9 (1865); K. Weinhold, Alemannische Grammatik 167 (Berlin, 1863) (on page 326 Weinhold cites Bopp's explanation for schriren. Bopp in his Vergleichende Grammatik3, I.36 (Berlin, 1867), assumed a general phonetic interchange between iw and ir); F. Zarncke, Zu den reduplizierenden Verben im Germanischen, Anglia 60.241-365 (1936), especially 279ff. and the references cited.

3 R. Loewe, Das starke präteritum des Germanischen, KZ 40.266-351 (1907). See also J. Schmidt, Die germanische flexion des verbum substantivum und das hiatusfüllende r im hochdeutschen, KZ 25.592-600 (1881).

4 H. Osthoff, Zur reduplicationslehre, PBB 8.540-67 (1882); J. Schmidt, Zur Geschichte des Indogermanischen Vocalismus II.429 (Weimar, 1875); Loewe, KZ 40.343-51; J. Janko, Über germanische ē2 und die sog. reduplizierenden Praeterita, IF 20.229-316 (1906-7).

5 K. von Knoblauch, Die germanischen perfecta auf r, KZ 1.573-6 (1852); J. Schmidt, KZ 25.598-600; Streitberg, Urgermanische Grammatik 281 (Heidelberg, 1896). K. Brugmann, KVG 541, lists under ‘unsichere Reste des s-Aor.’ wissun and scrirun; 596-8 he connects scrirun with the r-formations found in Indic and Italo-Celtic; I have found no scholar who has adopted this suggestion.

6 S. Feist, Die sogenannten reduplicierenden Verba im Germanischen, PBB 32.447-516 (1907); V. Michels, Zur deutschen Akzentgeschichte, Germania 87-8 (Halle, 1925); C. Karstien, Die reduplizierten Perfekta des Nord- und Westgermanischen 153f. (Giessen, 1921).

7 Schmidt, KZ 25.598f and Streitberg, UG 281, explain the first class forms as aorists and the seventh class forms as perfects.

8 Karstien, Redup. Perf. 153; Hirt, HU 2.143.

9 Flasdieck, Anglia 60, reviews in detail the theories and reexamines the data.

10 Prokosch discusses the seventh class verbs CGG 144-6, 148-9, 176-82.—It has been suggested that a similar development from aspect to tense was involved in the development of the weak preterite, Language 19.25 (1943).

11 Kögel discusses the various forms, PBB 9.535ff. There is also a MHG preterite 3d sg. spei, but this seems to be an analogical development on the pattern of schrei ‘cried’.

12 E. Sievers, Grammatische Miscellen. 5. Das pronomen jener, PBB 18.407-9 (1894); see also W. Braune, Althochdeutsche Grammatik, 5th ed. by K. Helm, 96,241 (Halle, 1936)

13 Language 14.248-78; see 10.3b.