Reduplication and the Old English strong verbs class VII (1).

ABSTRACT

Reduplication, one of regular ways of forming the Indo-European perfect, was inherited into the early Germanic as a marker of the preterite in some strong verbs. Of all older Germanic languages, Gothic as the only one retains this formation systematically; other dialects, due to large-scale modifications in the structure of their preterite, seem to have lost the original reduplication from their grammars. Although no reduplicated forms are properly attested in any of the Northwest Germanic dialects, Old English (not exclusively) displays a number of irregular formations within the seventh class of strong verbs, which have been traditionally considered reflexes of earlier reduplication. The Old English survivals of the originally reduplicated preterits, frequently referred to as r-preterites, are confined to one dialect only (Anglian) and include the following irregular forms: heht, leolc, speoft, beoft, leort, reord, ondreord (the preterite forms of hatan 'command', lacan 'leap', spatan 'spit', beatan 'beat', laetan 'let', raedan 'advise', ondraedan 'dread' respectively).

The present paper attempts to investigate the presumed vestiges of reduplication in the Old English seventh class of strong verbs, concentrating in particular on the provenance of the irregular preterite formations. The analysis hinges on the assumption that the reanalysis of reduplication into a new type of ablaut must have taken place on the way from Proto-Germanic to the early Northwest Germanic stage. Such restructuring in the system of Germanic strong verbs, it seems, obtained in the wake of the interplay of external and internal factors, namely the Germanic accent shift to the initial position and the marked influence of Frankish where the non-reduplicating pattern originated and wherefrom it disseminated to the whole Germanic speaking territory.

1. Introductory remarks

There are many linguistic phenomena, which, despite the amount of attention they have received within various linguistic frameworks, still remain intriguing puzzles, without a fully satisfactory explanation. Such is the Germanic reflex of Indo-European perfect reduplication, and especially its Northwest Germanic development. Investigated in detail, it still lacks unambiguous solution and numerous interpretations of facts offered seem far from being transparent. The present paper is not an attempt at reviewing the many theories (2) that have been advanced; rather, it approaches the data from a slightly different perspective, hoping to shed light on the external conditions of the split which occurred within the Germanic strong verbs class VII. In the paper, I would like to refer to a recent model proposed by Mottausch (1998) which provides a relatively new approach to the problem, concentrating on geographical and chronological circumstances of reduplication in Northwest Germanic.

Reduplication as a means of expressing perfectivity (3) has been employed in various Indo-European languages. Next to ablaut, it served originally as the major pattern of forming perfect in Proto-Indo-European. (4) This original pattern is best preserved in Greek and Sanskrit as in:

(1) Gr. leipo : leloipa (5) 'leave'

Gr. klepto : keklopa 'steal'

Gr. pheugo : pepheuga 'flee'

Skt. vartati : vavarta 'turn'

Skt. budh- : bubodh- 'know'

Latin, on the other hand, retains a slightly modified pattern, displaying, in the majority of forms, regular reduplication without ablaut:

(2) L. cano : cecini 'sing' ( < *kan- : *ke-kan-)

L. pendo : pependi 'hang'

L. curro : cucurri 'run'

A similar situation obtains for Germanic. Available data, furnished primarily by Gothic, allow for the reconstruction of the originally reduplicated preterites, existing along some regular ablaut patterns in Proto-Germanic. The Germanic controversy centres around the conflicting data provided by East Germanic (Gothic) on the one hand and Northwest Germanic on the other.

2. Gothic and Northwest Germanic facts

Preserved intact in Gothic, reduplication in the seventh class of strong verbs is in fact one of the features which renders this language distinct from other Germanic dialects. Regularly reduplicated forms appear here to indicate systematically the preterite of about 27 Gothic strong verbs. It is formed by prefixing a vowel (/[dpsilon]/) to the verbal root, possibly together with a consonant -- if the root begins with a single consonant, or a consonantal cluster -- if the root begins with such, unless one of the consonants is a sonorant. Some of the attested Gothic examples include (Bech 1969: 14):

(3) haitan : haihait 'command'

fahan : faifah 'catch'

fraisan : faifrais 'tempt'

laikan : lailaik 'jump'

letan : lailot 'let' (6)

maitan : maimait 'cut'

The vowel of reduplicating syllable (spelled ai) which is a reflex of IE */e/ (Gr. [epsilon], Skt. a) occupies a mid-open position of /[epsilon]/ (7) and is expected to occur regularly before /h, h, r/ when bearing primary or secondary stress in Gothic. (8) The fact that Gothic as the only Germanic dialect preserves reduplicated preterites systematically can be traced back to a number of Gothic-specific developments, such as the early elimination of the effects of Verner's law (which thus did not have a chance to obscure reduplication), extension of reduplicated vowel ai /[epsilon]/ to all forms (ignoring the contextual restrictions such as /h, h, r/ environment) or finally, extension of the stem of preterite singular to preterite plural (hence pl. lailotun for expected **lailtun < PG *leltun) (Jasanoff 2002: 1).

Reduplication was evidently abandoned in Northwest Germanic dialects where the forms corresponding to Gothic reduplicated preterites show no regular traces of it. Instead, class VII of strong verbs displays a unique pattern of ablaut, with the generalised *a :*e alternation. The new alternation gives rise to /[e.sup.2]/ and /eo/ which serve as the preterite forms of the originally reduplicated verbs. The former corresponds to Germanic /e/, /ai/ and /a/ in the present, the latter to /au/ and /o/ (D'Alquen 1997: 69) as in:

(4) OE slaepan : slep 'sleep'

OHG bluozan : blioz 'sacrifice'

OHG heizzan : hiaz 'command'

ON hlaupa : hljop 'run'

ON heita : het 'command'

ON blasa : bles 'blow'

There seems to be little doubt that the Northwest Germanic forms are closely related to the reduplicating verbs in Gothic and that they formed their preterite by reduplication in Germanic. (9) Such assumption is corroborated by a number of isolated relic forms which can be easily and unambiguously traced back to earlier reduplication, e.g.,

(5) ON roa : rera 'row'

ON snua : snera 'turn'

OHG stozan : steroz 'push'

OHG scrotan : screrot 'cut'

The main problem with this class of verbs is that of accounting for the split in tense formation between reduplication and ablaut alternation. Providing a plausible and satisfactory account of the developments which occurred on the way from Proto-Germanic into the early Northwest Germanic stage proved to be problematic, if not impossible. The change in the original system of reduplication appears to be closely related to the pattern of Germanic stress placement and in particular to stress shift. As suggested by Voyles (1980), the split between Gothic reduplicating verbs and Northwest Germanic ablauting verbs of class VII is to be looked for in their different stress patterns. Available data and phonology of Gothic (10) indicate that the root syllable was evidently stressed, while the prefix remained unstressed and as such was still felt to be a separate entity, not incorporated into the verb. The opposite proved to be the case in Northwest Germanic where the stress shifted to the reduplicating syllable which this time came to be interpreted as an inseparable part of the verb, rather than a prefix (Voyles 1980: 119). Placing the stress on the initial syllable then was the motivating force which initiated such reinterpretation of the reduplicating unit, to the effect that the syllable started to serve as the onset of the verbal root.

In terms of absolute chronology, the accent shift is traditionally dated to c. 500 AD and is assumed to have spread gradually over the whole Germanic-speaking territory (both continental and insular), leaving Gothic and East Nordic as relic areas with the stress still preserved on the verbal root.

The originally unstressed reduplicating syllable created appropriate conditions for the operation of Verner's law in Germanic. The process operated regularly on the root initial consonants of Proto-Germanic, bringing about the following alternations: h : [gamma], f: [beta], s : z, as in: PG *haitan : *he[gamma]ait, *hropan : *he[gamma]rop, *fraisan : *fe[beta]rais. In Gothic and possibly in East Nordic the Vernerian alternations became soon obliterated and the effects of the process were eliminated (through analogical levelling of [gamma] [beta] z with their voiceless equivalents in the present). Available evidence seems to prove that by the early Northwest Germanic times Verner's law ceased operating in the above-mentioned context, leaving only exceptional relic forms. The well-known Gothic relic of the process is saizlep (< *sezlep) 'slept' (cf. Olcel. sera < *sezo) attested next to the regular saiso, the latter with no traces of Verner's law. Precisely such vestiges indicate that the reduplicating prefix could not have bee n stressed either in Gothic or in the Pre-Germanic period. Northwest Germanic dialects, on the other hand, preserved alternations by Verner's law, which in turn blurred the original reduplication, thus making its identification not so straightforward. Such state of affairs contributed directly to the collapse of the original system of reduplication functioning as a preterite marker (Mottausch 1998: 53). Chronologically, once Vemer's law left its clearly recognisable traces in the system, the shift of the accent from the root syllable onto the initial reduplicating syllable occurred.

On the basis of the attested data, Mottausch (1998) groups the Northwest Germanic verbs of class VII into four major types, using chronological and geographical criteria as the basis for the classification. (11) The present paper, rather than discussing all the four types in detail, concentrates on those relevant to Old English data exclusively. The other types will be mentioned only marginally, as long as they can add to the discussion of the Old English data. Each type corresponds to a given Germanic dialect. Accordingly, type I is represented by East Nordic/Scandinavian and reflects an archaic state of affairs, comparable to Gothic, with the original Proto-Germanic accent-placement (on the root syllable). This type can be disregarded in the discussion of the Old English conditions since none of the attested verbs evince its existence in Old English. So is the case with type III (allegedly the infixing type) which is represented primarily by Old High German and Old Nordic (West Scandinavian) data, the so ca lled r-preterites. Investigation of Type II and IV is of primary importance here since these two types are well represented by the Old English material. As has been mentioned, the emergence, development and spread of all the four types can be chronologically ordered and traced back to particular geographical regions.

3. The case of Old English

Direct and unambiguous vestiges of earlier reduplication in Old English are confined to a small group of strong verbs belonging to class VII, attested in Anglian and in poetic texts (in majority still Anglian). Their preterite forms are strikingly different from the regular preterite of Old English class VII. The attested forms are the following (Brunner 1962: 201-202):

(6) hatan : heht 'command, order'

lacan : leolc 'play, jump'

spatan : speoft 'spit'

beatan : beoft 'beat'

raedan : reord 'advise'

ondraedan : ondreord 'dread'

laetan: leort 'let'

They have been traditionally divided into: (a) Anglian syncopated r-less type (i.e. heht, leolc, speoft, beoft), and (b) Anglian syncopated r-type (with reord, ondreord, leort) (cf. footnote 10). The earlier reduplicated forms of these verbs are reconstructed as follows (D'Alquen 1997: 77):

(7) *heyait- > heht

*lelaik- > leolc

*spespait- > speoft

*be[beta]aut- > beoft

*rered- > reord/ondreord

*lelot- > leort

The development of these forms must be attributed to specifically Old English conditions. The majority of explanations offered so far (e.g., Flasdieck 1936, Campbell 1959, Bech 1969, D'Alquen 1997) hold Anglian syncope responsible for the appearance of such forms. Accordingly, syncope which occurred as a result of strong initial stress reduced the second syllable (root vowel) and thus rendered the preterite stem monosyllabic, as in *heyait- > *heyt > heht, *lelaik > leolc. Such monosyllabic forms comply with the syllable shape of the strong verb classes one through six, all of which display monosyllabic structure. An alternative approach, which allows even more straightforward explanation, adduces the original ablaut pattern to account for the Old English developments. Specifically, the origin of the Anglian preterites is to be looked for in the behaviour of two verbs: *lelot/*lelt-, later leort, (12) and *rerod/*rerd-, later reord. The monosyllabic preterite stems *lelt- and *rerd- originally proper to the p lural, alternated with forms *lelot and *reroo, characteristic of the singular (Jasanoff 2002: 1). The root vowel of the singular reflects the original o-grade while the non-singular form, the reduced or zero grade. They constitute, together with later ondreord, the so called, primary verbs in the sense that they are the immediate continuations of the earlier reduplicated preterites and that they initiate the development of a new pattern ('type') within the seventh class of strong verbs, forming the basis for analogical spread. On this model further forms, constituting secondary formations/spread, (13) were built; these include: *lelak/*lelk-, *heyat/*heht-, *beBaut/*beft-, *spepat/*speft-. Soon, following the pattern of other monosyllabic preterites, the short non-singular form was generalised, i.e. *lelt *rerd, *lelk, *heht, *beft *speft. In all of these forms the initial syllable was stressed. The new consonantal clusters which emerged ran against some phonotactic principles of Old English. This necessitat ed consonantal simplifications such as: gt> ht, bt > ft, spt >ft (Bech 1969: 23). Such was the case with Anglian speoft, originating in *spespait, where an unpronounceable form *spespt would be expected; yet through a reduction of the cluster spt > ft, speoft obtained.

As regards the vocalism of these forms, the short diphthong co, present in all of the attested forms, except for heht, can be explained as an outcome of breaking before r + the following consonant in the r-type verbs. The other type, the r-less forms have been frequently attributed to velar umlaut operating before a liquid or labial + back, rounded vowel (14) (D'Alquen 1997: 87). It is most likely, that the diphthong soon came to be interpreted as a characteristic feature of the new preterite, also in verbs where it normally would not be expected.

One additional form which can be considered a relic of reduplicated preterite and which deserves an individual treatment is Anglian blefla, the past form of blawan. The original reduplicated form is reconstructed as *beBlaw. Once the initial stress weakened the old root vowel, the old reduplicated consonant (/b/) came to be associated with the initial cluster of the present blawan, hence *beBlaw> *bleBla > *blevla spelled blefla (15) in Anglian. The shift of the clusters then from the root initial to absolute initial position occurred as a part of the reanalysis of the reduplicated form.

The emergence and spread of this type, identified as type II in the adopted classification, and characteristic of Old English exclusively, is dated to the period of the Anglo-Saxon invasion, to the 5th (c. 450), possibly the 6th century. The later date could suggest that this type developed more or less simultaneously with the East Nordic type I.

This system of strong preterites, relatively unstable, was soon to be replaced by the newly emerged, continental type IV. The new type was to serve as a successful way of regularising strong verbs of class VII, not just in Old English but also in the other dialects of the Northwest Germanic subbranch. In Old English the type is represented by the non-reduplicating verbs of class VII which form two groups depending on the vocalism of the preterite: traditionally referred to as: -e-preterites and -eo-preterites. The vocalism of these verbs poses numerous difficulties; they systematically display the new pattern of ablaut and their preterites can be analysed as: /e/ + root vowel of the present where /e/ would serve as a kind of infix before the original root vowel and would soon be generalised as an infix preterite marker, e.g., aikan : e-aik, aukan : e-auk. The new pattern is modelled on the original vowel-initial verbs as the source of the e-infix. The basis for analogical spread was formed by six such verbs: aikan, aran, alpan, aukan, ausan, aupan (referred to as the aukan-group) (D'Alquen 1997: 80). Assumedly they generated a wholesale restructuring of the reduplicated preterite. Analogical levelling produced forms like h-e-ait (< *haitan), bl-e-ot (< *blotan), skr-e-aud (< *skraudan), f-e-all (< *fallan), etc. As a result of monophthongisation of unaccented diphthongs /ai/ and /au/ into /e/ and /o/, the sequences *e-ae (ae < e) and *e-o obtained, and soon after the weakening of the second (unaccented) syllable and subsequent contraction, /[e.sup.2]/ emerged:

(8) *e-ae (< *e- e) > /[e.sup.2]/

*e- o > /eo/

                     infixation             merger into one syllable

*haitan :            *h-e-ait >             *heet > h[e.sup.-2]t
laetan (< *letan) :  *le-et > h[e.sup.-2]t
*hropan:             *hr- e-o-p > hreop

This new type spread first into the archaic reduplicating preterites: het, let, red for heht, leort, reord, and later disseminated to the other verbs of this class (Mottausch 1998: 71).

As regards the verbs with the structure: /ea/ + nasal or liquid + C (e.g., feol:feallan, heold:healdan, beonn:bannan), the long diphthong /eo/ present here can be traced back to the earlier *e-a vocalism which could be analysed in the following way:

(9a) *e - a > /ea/ > /eo/ (where /eo/ obtains as a result of breaking before /r, 1/ + C (possibly before /w/and /h/) for the sequence with liquid, and as a result of rounding caused by /n/ for the sequence with a nasal) subsequently:

/eo/ > /eo/ (where /eo/ occurs as a result of analogical replacement and is not environmentally restricted)

e.g., haldan 'hold' and spannan 'stretch':

(9b)

                                           analogical
 infixation       merger       breaking    replacement

[down arrow]   [down arrow]  [down arrow]  [down arrow]
 *h-e-ald >      *heald >      *heold >       heold
*sp-e-an(n) >  *spean(n) >   *speon(n) >      speon

4. Chronology of the developments. Final remarks.

The reanalysis in the system of originally reduplicating verbs in Northwest Germanic is to be dated long after the departure of Goths from the region of the Baltic Sea (around the 3rd c. AD). Available data suggest that initially, strategies of reinterpretation differed regionally. Accordingly, Old English employed type II, Old Norse (its West branch) and Old High German adopted type III. Type II can be dated roughly to the late 5th century, while type III to the 6th century, both then immediately or shortly after the accent shift. Soon both of them were to yield to the emerging type IV pattern, readily accepted as the common strategy of preterite formation in all dialects. This particular type, it seems, was originally a local strategy, employed in Frankish (esp. Salish and Ripaurish), which emerged there around 500 AD. It was precisely due to the powerful, both political and social influence of Franks, and thus Frankish, that this type, around the 7th century, came to be interpreted as the prestigious form and spread to the Northwest Germanic dialects. Mottausch considers this to be a case of Sprach-Darwinismus, where out of a number of competing types, one, strongest emerges as the prominent, and soon begins to dominate. Frankish, being a language of highly influential socio-political status with no relic forms representing the earlier stages of development (e.g., type II or III) preserved, seems to fulfil all the criteria necessary to serve as the Kern gebiet and to nurture the later successful type. Assuming that this was the case, Old English, just as Old High German and Old Norse, must be considered to have been a region of secondary adaptation of type IV, which spread there systematically under the powerful influence of Frankish.

The dialectal distribution of the four major types according to the Raum-Zeitmodell by Mottausch (1998) can be presented as follows:

(10) Type I -- East Nordic

Type II -- Old English

Type III -- Old High German, West Nordic

Type IV -- Old English and all the other Northwest Germanic dialects

To conclude, undeniably tense formation through ablaut was the productive pattern in Germanic in contrast to reduplication which was used as a preterite marker only in a small number of the attested forms. The split between Gothic reduplicated preterites, representing earlier state of affairs, and Northwest Germanic ablauting forms, relatively 'recent', indicates that the reanalysis of reduplication into a new kind of ablaut must have taken place. The interplay of two crucial factors may be held responsible for such radical restructuring in the system of Germanic strong verbs. One of them, internal in nature, was the accent shift to the initial position, which apparently occasioned morphological reanalysis and indirectly led to the elimination of reduplication. The other, of external nature, has to do with the contribution of Frankish, where the non-reduplicating pattern originated and where from it disseminated to the whole Germanic-speaking territory.

(1.) An earlier version of this paper was presented at the 37th International Congress on Medieval Studies, Western Michigan University, Kalamazoo, 2002. The research was generously supported by a Kosciuszko Foundation grant for which I would like to express my deepest gratitude.

(2.) Some of the previous studies on the Germanic reflexes of reduplication include: Brugmann/Wood (1895) - ablaut theory, Lehmann (1952) - laryngeal explanation, van Coetsem (1956) - reverse analogy, Bech (1969) - infixation, Voyles (1980), Fulk (1987) - e-infix Vennemann (1994) - phonological explanation.

(3.) Expressing perfectivity is only one of the many functions reduplication can have; understood in a broader, non-Indo-European context it can well indicate plural or collective (Malay, Aztec, Hausa, Dakotan, Japanese, Mandarin Chinese), it can serve the purpose of intensification, increase (Turkish, Celtic, Mandarin Chinese), diminution or attenuation (Cantonese), etc. All these functions of reduplication are discussed in some detail in Moravcsik's (1978) survey of reduplicating systems. A detailed treatment of perfect reduplication is offered by Niepokuj (1992).

(4.) Reduplication can be used here not only as a preterite marker but also in present formations, e.g., in a small group of Greek mi-verbs: didom:-mi 'I give', tithe:-mi 'I place', piple:-mi 'I fill'; in thematic verbs, e.g., L. sisto 'stand' (<sti-[sth.sub.2]-o) L. sido (si-sd-o-);

in the aorist: Gr. epepHne 'he killed', Skt. avocam 'I spoke' (<*e-we-uk (w)-o-m) (Szemerenyi 1996: 280-1).

(5.) A special kind of ablaut is present in Greek, whereby the stressed c-grade of the root appears in the singular and the unstressed nil or reduced grade in the plural: le-loip-/le-lip- < sg *le-[loik.sup.w]\ p1*le.[lik.sup.w-].

(6.) In a very few archaic Gothic verbs, including this one, reduplication is found next to qualitative ablaut, even though these two markers of the original preterite seem to have appeared essentially in complementary distribution (also saian: saiso 'sowed').

(7.) The origin of the vowel and especially its occurrence in certain contexts, namely before consonants other than /h h r/ continues to pose difficulties. Analogical levelling has been adduced to explain the appearance of /[epsilon]/ in forms such as: saiso, laailaik, taitok, in which /i/ rather than /[epsilon]/ would have been phonologically regular (cf. below) (Bennett 1967: 662).

(8.) Gothic /i u/> /e o/ when stressed and when before /h h r/.

(9.) An alternative account was posited, among others, by Lehmann (1952) who explains the Northwest Germanic perfect forms by recourse to laryngeal theory, assuming that the aberrant preterite forms in NWGmc have no genetic relationship to the reduplicated formations in Gothic. The origin of these forms is to be sought instead in a type of secondary conjugation, productive in the early IE dialects. Lehmann's analysis hinges on the assumption that the OE and ON preterites are found in verbs whose roots ended in laryngeals or had laryngeal extensions, hence the following pattern of development is suggested: e.g., inf. laetan (Gr. ledein, Lith. leiden, Lat. letum), with original preterite *leXyd> leXyd-> leyt- > l[e.sup.2]t (pres. *leX-). Lehmann assumes that "in some forms the laryngeal may have survived and fallen together with the r-phoneme, giving rise to *lert, whence leort" (Lehmann 1969: 284). Attractive as the interpretation was, it has not gained widespread recognition and was soon put aside in favour o f the theories relating the NWGmc forms to the Gothic reduplicating verbs.

(10.) E.g., Gothic rule of vowel deletion, lowering of the short high vowels /i u/to /e o/, or relic forms of Verner's law (cf. below) (Voyles 1980: 94-95).

(11.) An alternative classification, remaining largely within the traditional framework, is offered by Bech (1969). According to his grouping, which includes Gothic evidence as well, the Germanic reduplicated verbs can be subdivided into six major types (reduced later to four): (a) the Gothic reduplicating type, (b) the Old Norse verbapura type in -r-, (c) the Old High German verba impura type in -r-, (d) the Anglian syncopated r-less type, (e) the Anglian syncopated r-type, (f) the Northwest Germanic e-type (Bech 1969: 3) (cf. Kortland 1991: 97).

(12.) The later form leort owes its existence probably to the process of dissimilation: lelt>lert>leort (Campbell 1959: 320); it may well be analogical formation (cf. Flasdieck 1936: 259-60).

(13.) In fact, Mottausch distinguishes three groups of originally reduplicated preterites for the pre-Old English state of affairs; next to the two mentioned groups (primare Verben and sekundare Ausbreitung), he postulates 'tertiare Ausbreitung' with OEsweop, weop and sweog (Mottausch 1998: 57). Unfortunately, the originally reduplicated preterites of these verbs, as a result of a number of subsequent phonological developments, were very early obliterated.

(14.) Cf. Flasdieck (1936: 255-6).

(15.) The form has been long considered a scribal error (e.g., Brunner 1965); it was very early replaced by gebleou by analogy to verbs whose infinitives have the form of C(C)awan (D'Alquen 1997:85).

REFERENCES

Bech, Gunnar

1969 Das germanische reduplizierte Prateritum. (Det Kongelige Danske Videnskabernes Selskab, Historisk-filosofiske Meddelelser 44, 1.) Copenhagen: Munskgaard.

Bennett, William H.

1967 "Effects of pre-Gothic juncture", Language 43: 661-5.

Brunner, Karl

1962 Die englische Sprache II. Tubingen: Max Niemeyer Verlag.

Campbell, Alistair

1959 Old English grammar. Oxford: Clarendon Press.

Coetsem, Frans, van

1956 Das System der starken Verba und die Periodisierung im alteren Germanischen. (Mededelingen der Koninklijke Nederlandse Akademie van Wetenschappen, Niuwe Reeks, Deel 19,1.) Amsterdam: North Holland Publishing Company.

D'Alquen, Richard

1997 "Non-reduplication in Northwest Germanic". NOWELE 31-32: 69-91.

Davis, Garry W. -- Gregory K. Iverson (eds.)

1992 Explanations in historical linguistics. Amsterdam-Philadelphia: John Benjamins Publishing Company.

Flasdieck, Hermann M.

1936 "Die reduplizierenden Verben des Germanischen", Anglia 60: 241-366.

Fulk, Robert D.

1987 "Reduplicating verbs and their development in Northwest Germanic", Beitrage zur Geschichte der deutschen Sprache und Literatur 109: 159-179.

Greenberg, Joseph H. (ed.)

1978 Universals of Human Language, vol. 3. Word structure. Stanford: Stanford University Press.

Jasanoff, Jay H.

2002 "Reduplication in Germanic" (abstract).

www.fas.harvard.edu/~humit/Master/Jasanoff.pdf

Kortland, Frederik

1991 "The Germanic seventh class of strong verbs", NOWELE 18: 97-100.

Lass, Roger (ed.)

1969 Approaches to English historical linguistics. An anthology. New York: Holt, Rinehart and Winston, Inc.

Lehmann, Winfred P.

1952 Proto-Indo-European phonology. Austin: University of Texas Press.

Lehmann, Winfred P.

1969 "Old English and Old Norse secondary preterites in -r-", in: Roger Lass (ed.), 276-286.

Moravcsik, Edith A.

1978 "Reduplicative constructions", in: 3 H. Greenberg (ed.), 297-334.

Mottausch, Karl-Heinz

1998 "Die reduplizierenden Verben im Nord- und Westgermanischen Sprachen. Versuch eines Raum-Zeitmodelle", NOWELE 33: 43-91.

Niepokuj, Mary

1992 "The development of perfect reduplication in Indo-European", in: G.W. Davis - G. K. Iverson (eds.), 193-206.

Szemerenyi, Oswald J. L.

1996 Introduction to Indo-European linguistics. Oxford: Oxford University Press.

Vennemann, Theo

1994 "Zur Entwiklung der reduplizierenden Verben im Germanischen", Beitrage zur Geschichte der deutschen Sprache und Literatur 116: 167-221.

Voyles, Joseph B.

1980 "Reduplicating verbs in Northwest Germanic", Lingua 52: 89-123.