Review Of: S. Schumacher & J. Matzinger, Die Verben Des Altalbanischen. Published In: Kratylos 60, 1-17.

Transcript

Kratylos Kritisches Berichts- und Rezensionsorgan für indogermanische und allgemeine Sprachwissenschaft Jahrgang 60, 2015 Dr. Ludwig Reichert Verlag  Wiesbaden KRATYLOS Kritisches Berichts- und Rezensionsorgan für indogermanische und allgemeine Sprachwissenschaft Im Auftrage der Indogermanischen Gesellschaft herausgegeben von OLAV HACKSTEIN Jahrgang 60, 2015 Redaktioneller Beirat: Tamaz Gamkrelidze, Tbilisi José Luis García Ramón, Köln Heinrich Hettrich, Würzburg Wolfgang Hock, Berlin Michael Job, Göttingen Joshua T. Katz, Princeton Martin Kümmel, Jena Gerhard Meiser, Halle/Saale Torsten Meissner, Cambridge H. Craig Melchert, Los Angeles Hans-Jürgen Sasse, Köln Rüdiger Schmitt, Laboe Klaus Strunk, München Die Autoren des KRATYLOS sind angehalten, die Formen sachlicher Diskussion zu wahren. Erwiderungen werden grundsätzlich nicht aufgenommen. Unverlangt eingesandten Werken kann eine Besprechung nicht zugesagt werden. Redaktionelle Korrespondenz und Besprechungsexemplare erbeten an: Prof. Dr. Olav Hackstein, Ludwig-Maximilians-Universität München, Lehrstuhl für Historische und Indogermanische Sprachwissenschaft, Geschwister-Scholl-Platz 1, D-80539 München E-Mail: [email protected], http://www.indogermanistik.uni-muenchen.de ISSN 0023-4567 © 2015 Dr. Ludwig Reichert Verlag, Wiesbaden http://www.reichert-verlag.de Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urhebergesetzes ist ohne die Zustimmung des Verlages unzulässig und strafbar. Dies gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Speicherung und Verarbeitung in elektronischen Systemen. Printed in Germany I. REZENSIONSAUFSÄTZE Schumacher, Stefan und Matzinger, Joachim: Die Verben des Altalbanischen. Belegwörterbuch, Vorgeschichte und Etymologie. Unter Mitarbeit von AnnaAnna-Maria AdaktyAdaktylos. los Albanische Forschungen, Band 33. Harrassowitz Verlag Wiesbaden, 2014. 1055 Seiten. Gebunden, 198,00 EUR. ISBN: 978-3-447-06448-4. This impressive volume (henceforth: VerbAlt) is a long-awaited milestone in Old Albanian linguistics, and it fully lives up to expectations. The book is the outcome of a research project, funded by the Austrian Science Fund, of which the authors were the main applicants. Content-wise, the main parts of the book are the lexicon of attestations, p. 287–963, which was jointly compiled by both authors, and the seven linguistic chapters written by Stefan Schumacher (henceforth: S.). They entail the analysis of the verbal system, a historical grammar of the verb, a historical phonology of Albanian (p. 19–286), and a selection of etymologies (p. 965–1010). The book is concluded by a bibliography and an index on the variants of the Albanian verb stems discussed in the lexicon (p. 1023–1055); the latter index was compiled by Anna-Maria Adaktylos. The Old Albanian sources on which this study is based are the Old Gheg works of Buzuku (1555), Budi (1618–1621), and Bogdani (1685), and the Old Tosk writings by Matranga (1592) and Variboba (1762). The authors give different explanations for the absence of other important pre-1800 texts from their corpus (p. 24). The Latin-Albanian dictionary by Bardhi (1635), the text of Kuvendi i Arbënit (‘The Council of Albania’, 1703) and Kazazi’s Dottrina cristiana (1743) were excluded because they are available in recent editions, all of which contain full concordances. The dictionaries by Da Lecce (Old Gheg, 1716) and Kavalliotis (Old Tosk, 1770), and the Codice Cheutino (Old Tosk, 1736–1739), were not available in facsimile, so that the correctness of the existing editions could not be checked. Finally, the Italian-Albanian dictionary of Grottaferrata (Old Gheg, 1710) and Da Lecce’s Albanian grammar of 1716 were left out of consideration for lack of financial and personal resources. Particulary the absence of Bardhi’s material is a pity, as it would have taken little extra time to incorporate it (there are not many 2 DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen inflected verb forms in the dictionary), the text is relatively early, and it would have added some interesting lexical material to the study. Chapter B (p. 25–126) deals with the structure of the OAlb. verbal system. One of the main novelties in this chapter is a new division of the verb stems which differs from the systems hitherto in use. I think that the new system is more practical and is destined quickly to become accepted. The authors distinguish four verb classes, which they name by one or two typical representatives: 1. The class di/djeg, in which the present stem usually has the same form as the unsuffixed basis or verbal stem. It can end in a vowel, as in di ‘to know’, or in consonant, such as djeg ‘to burn’. 2. The class kujton/ecën, in which the basis takes an nsuffix which can be preceded by a stressed vowel (as in kujton ‘to think’) or a schwa (ecën ‘to walk’). 3. The class vret/përket, in which the basis takes a stressed Vt-suffix to form the present. The basis can end in a vowel (/vra-/ → vret ‘to kill’) or a consonant (/prek-/ → perket ‘to touch’). 4. Irregular verbs, viz. with suppletive present and aorist stems (e.g., është – kle ‘to be’), and some other verbs with an irregular aorist stem. The discussion of the verbs /ënderi/ ‘to honour’ and /rreni/ ‘to lie’ at the end of section 1.2.4 is unfortunate, as these are not suppletive verbs. They show a stressed í-suffix in the present in Buzuku (in both verbs) and Budi (in /rreni/), whereas other authors treat them as djeg- (e.g., Budi and Matranga /ndēr-/, Bogd. /rren-/) or ecën-presents (Bardhi and Bogd. /ndērën/). The aorist stems are /ndēr-/ and /rrēn-/. The verb ‘to honour’ stands beside the nouns /ndēr/ (Buz. Budi) ‘honour, praise’ and /ndere, -ja/ (Buz. Budi Bogd.) ‘honesty’; the variant present /ndēron/ in Budi, Bogdani and Variboba is a denominal derivative. S. (p. 31) claims that /ënderi/ and /rreni/ are the last two representatives of a present class with stressed í-suffix. This may be true, though VerbAlt itself classifies Budi’s 3sg. pres.sub.act. /thëthijë/ ‘sucks’ as an instantiation of a stem /thëthi/, another í-present (p. 880). But it can also be observed that the occasional ín-presents (a subtype of class 2 in VerbAlt) may be a remake of the í-type. Compare Buzuku’s 3pl.ind. /ënderinë/ ‘they honour’ (di-class) beside 3pl.ind. /ënderinjënë/ (ín-class) (Fiedler 2004: 726). Hence, /ënderi/ and /rreni/ rather belong to the di-class. Of ín-presents in Old Albanian I find /djersīn/ ‘to sweat’, /andërrin/ ‘to dream’, /gjëllin/ ‘to live’, /lëpīn, lypīn/ ‘to lick’, /mbërrīn/ ‘to arrive’, /përtërīn/ ‘to renew’, /cqyrëtin/ ‘to investigate’ (in Budi), /regëzīn/ (Buz.) ‘to tear’, /trokullin/ (Buz.) ‘to knock’, /thëthin/ ‘to suck’, and possibly Bogd. 1s.impf.act. /krehinjë/ ‘I combed’. Most of these stems have a denominal origin (to djersë ‘sweat’, ëndërr ‘dream’, gjallë ‘alive’, ri ‘new’, rragje ‘quarrel’, thithë ‘nipple’, ndēr ‘honour’), but /lëpin/ (to lap ‘to slobber’), /mbërrin/ and /cqyrëtin/ (to qyr ‘look’) are deverbal formations. DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen 3 Section 1.4 (p. 31–47) provides a long and excellent discussion of the use of the middle voice in Old Albanian. Sections 2 and 3 (p. 50–109) provide a synchronic classification of the verbal categories of Old Albanian, with overviews of the endings and with – sometimes detailed – discussions on the meanings of the various categories. Some of the footnotes deal with etymological questions. Section 4 is a brief but highly interesting addition on the verbal complex. The latter is defined as the syntactic cluster of verb plus adjacent particles, such as (in traditional terms) the object pronoun, reflexive pronoun, progressive particle, negation, and the subjunctive (/të/), possibilitative (/mundë/ vel sim.) and jussive (/lê/) particles. S. (fn. 106) argues that all these particles can be regarded as affixes, since they have a fixed placed in the order of constituents. Hence, he refers to the object pronouns as “object particles” and to the reflexive as the “non-active particle” /u/. The different combinations of verb plus particle(s) are then discussed for each verbal category separately. It seems to me that this chapter will provide a fertile basis for future research into Albanian syntax and for comparisons with other Balkan languages. A taste of Balkan linguistics is given by S. himself in the section on the verbal complex in the imperative (p. 120), where he compares Albanian with Modern Greek. Parts C, ‘On the prehistory of individual verbal categories’ (p. 127–197), E, ‘Historical Phonology’ (p. 205–276), and H, ‘Selected Etymologies’ (p. 965–1010) will draw most attention from historical linguists. Part C focuses on the middle, imperfect, aorist and optative. The inflection of the present indicative is not discussed separately, though some of S.’s views can be pieced together from the historical phonology. First, however, I must dwell on an issue of general nature. On p. 49, and again on p. 127, S. explains that his reconstructions are guided by what he calls the “Hereditätsprinzip”, a term borrowed from Meiser (2003: 5), who introduced it in his work on the Latin verb. Meiser formulates this principle in the following way (in my translation – MdV): “In the case of unbroken historical continuity, we can expect the regular continuation of linguistic (e.g., morphological) elements which were inherited from a previous stage. This principle is not affected by general and trivial changes which exceed a specific case (such as thematization, the replacement of endings, regular sound change). Wherever we observe deviations from this rule, they must, as a matter of principle, be explained – as we would explain the ‘exceptions’ to a sound change.” At first sight, this seems a reasonable proposal. For instance, if PIE had a middle voice and an aorist tense, and Albanian has these categories as well, it is reasonable to entertain the idea that the Albanian middle and aorist in some way continue the PIE middle and aorist. S. hastens to add that his hereditary principle is neither a “Denkverbot” nor a “Denkgebot” but merely a “Denkerlaubnis”. He also wishes to combat a certain tradition 4 DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen in Albanology which regards the morphological continuation of PIE categories in Albanian as little likely. Still, I find the “hereditary principle” problematic for both theoretical and concrete reasons – as much as I find the name uninformative. At the principal level, it is a basic rule of our branch of science that, when we are faced with the choice between two or more possible explanations for a linguistic change, the safe bet is always the chronologically more recent development. It is recent origin which must be falsified by proving that it cannot be true; only then are we allowed to assume a solution from an earlier chronological stage. This is the best way to avoid the introduction of ghost developments, of pretending specific knowledge about a more distant past when the correct explanation lies in a plausible development of a more accessible period. In other words, the chronological principle of recent-overancient is an extension of Occam’s Razor. Of course, Schumacher is well aware of this caveat and hedges himself in against criticism. Still, I disagree with his claim (p. 49) that “es spricht aber nichts dagegen, es [viz., the hereditary principle] unverändert auf das Albanische anzuwenden. Wenn sich also im Albanischen verbale Kategorien finden, die gleiche oder ähnliche Funktionen haben wie grundsprachliche Kategorien, dann ist es als wahrscheinlich zu betrachten, dass die albanische Kategorie die grundsprachliche Kategorie fortsetzt”. Here and in several other instances, I would prefer to see the word “wahrscheinlich” replaced by a weaker claim, such as “möglicherweise”. For instance, S. claims that the hereditary principle allows us to assume that the Albanian middle voice directly reflects the PIE middle (p. 127–131), that the Albanian subjunctive directly reflects the PIE subjunctive (p. 55), and he bluntly posits that “the Albanian imperfect indicative continues the PIE injunctive of the present stem” (p. 53). My main concern is the absence, in VerbAlt, of any discussion of possible alternatives. Even accepting Meiser’s principle as a null hypothesis for our investigations, the main task would be to falsify the hypothesis, that is, to investigate possible counterarguments. And this step is rarely undertaken by VerbAlt. For instance, the subjunctive endings 2s.act. /-sh/ and 3s.act. /-ë/ are derived from PIE thematic subjunctive forms, although the postulated PIE preforms *-ēsi and *-ēti (see below for details) are by no means the only conceivable reconstructions which would lead to the required PAlb. endings. S. tries to strengthen his argument by adducing the semantics of the Albanian subjunctive, which would be remarkably similar to those of the PIE subjunctive. Yet it can be questioned how relevant such a similarity is: one could say the same of the Latin subjunctive, which also does not – at least in my view – continue the PIE subjunctive (see Schrijver 2006 for a critique of the views presented by Meiser 2003 on the Latin verb). DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen 5 In addition, the historical linguist of Albanian is in a much tighter spot than her or his colleague dealing with Latin. Latin is separated from PIE by 3,000 to 3,500 years, and by using the Sabellic languages we can reconstruct a Proto-Italic stage which is between 500 and 1,000 years closer to the parent language. Taking into account a possible Proto-Italo-Celtic stage, the Celtic verb may further help to understand the Latin categories. For Albanian, the situation is much less advantageous. Buzuku post-dates the first extensive Latin texts by 1750 years. Even if we take Proto-Albanian, the ancestor of Gheg and Tosk dialects, as a point of departure, and date it to around 500 AD, there is still a gap of 4,000 years with PIE. No light is shed on this gap by an intermediate stage shared with another IndoEuropean branch. And our knowledge of the PAlb. verb is much less detailed than that of the Latin verb of Plautus’ time. Hence, we must be even more careful in applying a hereditary principle to Albanian than to Latin. Now to some specific issues in parts C, E and H. For the present middle (OAlb. ind. sg. /-em/, /-ē/, /-etë/, pl. /-emi/, /-ī/, /-enë/), S. deduces (p. 129) from 2s.ind. /-ē/, 2s.sb. /-ēsh/, and 2p.ind. /-ī/, that the suffix vowel was always stressed, since posttonic vowels in OAlb. are never long. This is not the only way to look at it. It is true that OAlb. long vowels are usually found in stressed position (but note derivatives such as /bēkón/ ‘to bless’, /ndērón/ ‘to honour’), but word-final long vowels can contrast phonologically with short ones in monosyllables, as in /klē/ ‘you were’ vs. /kle/ ‘he was’, /u thā/ ‘has dried’ (intransitive) vs. /tha/ ‘said’. The long variants are usually explained from a recent vowel contraction. I see no way to exclude the possibility that non-active 2s. -/ē/ or 2p. /-ī/ also reflect a contracted sequence of vowels, even if other polysyllabic words ending in these long vowels are absent. This is not to say that S.’s theory is impossible, but it is important to realize that there is an alternative, since S. himself draws some far-reaching conclusions. He generalizes that the suffix vowel would have been stressed in all middle forms, and that this would reflect a forward stress shift from the initial syllable to the second one at some prehistoric stage, e.g., from *dégamai ‘I burn myself’ to *degámai, whence *degámẹ > *degámi > digjem. He (p. 129) even postulates a sound law which would have shifted the stress in trisyllables in *-ai from the first to the second syllable, as exemplified by the well-known alternation between Alb. gjarpër ‘snake’, pl. gjërpinj ‘snakes’ < *serponoi. Thus, this accent shift would confirm that a majority of the middle forms ended in PAlb. *-ai from PIE *-oi or (1s.) *-h2ei. But if the accent rule – which is typologically peculiar – loses its foundation, the reconstruction of the middle endings also becomes much more uncertain. In the chapter on the non-active imperfect (p. 140–49), S. builds on his previous conclusions. In the non-active imperfect, long vowels appear in all endings 6 DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen in Budi (1s. /-ēshë/, 2s. /-īshe/, 3s. /-ē(j)/, 1p. /-īshmë/, 3p. /-īshnë/), and 3s. /-ē/ also appears in Bogdani. S. concludes that the OAlb. suffix vowel was stressed in this paradigm, which led to vowel lengthening. His reconstructions on p. 142 show that he explicitly regards this lengthening as post-PAlb., and he assumes that the suffixal stress would have been adopted from the non-active present (p. 148). The surprising restriction to Budi could be due to analogical reformations in the other authors. Thus, S.’s theory on the accent could be right. But if the endings of the non-active imperfect indeed go back to the imperfect of ‘to be’ which was suffixed to the present stem (as S. is not the first to assume), then another possibility presents itself. The imperfect of ‘to be’ is reconstructed on p. 142 as sg. *ješǩ, *iše, *je, pl. *išm, *ištǩ, *išnnǩ, but there are indications that the 3s. may have been *jē, viz. in the 3s.admir.ipf. ending /-kē/, on which see below. If correct, *jē may directly explain the 3sg. non-active imperfect /-ē(j)/, and the length may have spread to the other person-number forms. In the reconstruction of the middle endings (p. 130–1), I note the discrepancy between the different outcomes of PAlb. *-ai which are assumed, viz. zero (in 1s. -em < *-amai) and /ë/ (in 3s. /-etë/ < *-etai and 3p. /-enë/ < *-antai). The same vacillation is explicitly noted by S. in section E (p. 227) for the nom.pl. of masculine nouns, but remains without explanation. Since -në occurs as a 3p. ending in a variety of categories, an analogical origin of final /-ë/ in at least this ending would be unsurprising. Thus, the claim “die Form [viz. /-enë/] setzt unmittelbar die uridg. Primärendung fort” is too strong, even aside from the fact that the reconstruction of the ending as *-ontoi is based on the implicit assumption that Albanian introduced the same innovations in the middle paradigm as Sanskrit and Greek (of course, it could have shared such innovations with its Balkan neighbour, ProtoGreek). The subjunctive gets a meager deal compared to the other categories (p. 55-58, 214, 223). The 2s.act. ending /-sh/ is explained from PIE *-ēsi or *-eh1esi plus an unknown particle starting in a consonant (see below for criticism of the phonology), the 2s.non-active ending /-ēsh/ is derived from analogy of 2s.non-active ind. /-ē/ with 2s.active sb. /-sh/, and 3s.act. /-ë/ is regarded as the outcome of PIE *-ēti or *-eh1eti. The variant endings /-të/, /-etë/ of the 3s.sb. forms /jētë/ ‘would be’, /kētë/ ‘would have’, and /thoetë/ ‘would say’, are left undiscussed, though they could be used to support the claim made by Schumacher 2007: 232ff., viz. that these three verbs, and ‘to say’ in particular, took their endings from the non-active paradigm (which has 3s.ind. -etë). The development of the imperfect gets an extensive discussion (p. 132–149). The different endings per person-number form are given for Budi, Bogdani, and Variboba. Bogdani’s ending /-ke/, which occurs in /flitke/ ‘spoke’ (beside unsuf- DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen 7 fixed /flit/), /pritke/ ‘waited’ (beside /prit/), /itke/ ‘remained’, /rridhke/ ‘streamed’ and /flĩke/ ‘slept’, is omitted in the table on p. 133 for reasons that are only explained in footnote 8. S. rightly regards the use of /ke/ as phonotactically conditioned, because in 4 out of the 5 tokens, the ending is added to a stem in a dental obstruent. In other words: whereas other stems show the suffixation of /-te/ to the originally endingless 3s.impf.act., these four stems probably added /-ke/ for dissimilatory reasons. But this cannot explain /flĩke/. In fact, it is unclear to me why S. declares the origin of /-ke/ to be unknown, since he does acknowledge the connection with the 3s.impf. in /-ke/ of the admirative, found in Buzuku, Budi and Bogdani. This issue requires some explanation. The admirative was grammaticalized in the OAlb. period from univerbation of inverted perfects, e.g., ka klenë ‘has been’ vs. klenë-ka ‘is!’. The admirative suffix usually has the same forms as the present and imperfect of kam ‘to have’, but there are a few exceptions in the imperfect, such as 3s.impf. /kish/ or /kishte/ ‘had’ vs. adm.impf. /-ke/. Such variation suggests that the admirative here retains the older forms of ‘to have’, as S. indicates on p. 147: “[es kann] als wahrscheinlich betrachtet werden, dass die Imperfektformen von ‘haben’ im Admirativ eine ältere Flexion reflektieren als das Normalparadigma.” For the 3s. impf., Klingenschmitt (1981: 121) and Schumacher (2007: 237) already concluded that the vocalism of adm. /-ke/ presupposes the erstwhile existence of a 3s.impf. *je ‘was’. This solution also yields a satisfactory explanation for /flitke/ ‘spoke’ etc.: apparently, in Bogdani’s dialect, the ending /ke/ was not yet completely relegated to the admirative, but lingered on in dental-final stems where it was useful for phonotactic reasons. A few lines must be devoted to the vowel length in 3s.adm.impf. /-ke/. There are graphic indications for /-kē/ in Buzuku and Bogdani, ignored by VerbAlt. The reason probably is that Fiedler (2004: 400–1) decided that Buzuku’s ending must be interpreted as /-kej/, but S. himself (p. 145, fn. 18) disagrees with Fiedler’s interpretation of 3s.impf.n-a. <-ee, -eh> as /-ej/, and posits /-e/ instead. Since the 3s.adm.impf. is only attested with <-ee> and <-e> (and zero), the interpretation as /-ej/ is even less likely here. The attestations in Buzuku are: 7x -<-chee> in clenechee (3x) ‘had been’, ξanechee ‘had given’, ξanechee ‘had said’, paschee ‘had had’, vdecunechee ‘had died’, 4x <-che> in muiteche (2x) ‘had been able’, chleneche ‘had been’, dasche ‘had loved’, and 2x reduced <-ch>, viz. in clenech and clẽech ‘had been’ in a single phrase (Fiedler 2004: 392–5). If reduction was the synchronic tendency, then the long vowel must be the most original variant. In Budi, the form is five times attested in Speculum Confessionis, always with short <-ke> (qenëke ‘had been’, vraake ‘had killed’, ndodhke ‘had happened’, gjetke ‘had found’, paske ‘had done’). In Bogdani, most tokens have <-ke> (banke ‘did’, pritke ‘waited’, flitke ‘talked’, ξξanèke ‘had said’, kjanèke ‘had been’) but two 8 DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen forms in <-kee> suggest a long vowel: bajtèkee ‘supported’ (1.115; vs. 4x bajteke, 3x bajtekè, 1x bajtèkè ‘portava’) and mujtekee ‘prevailed’ (2.156; vs. once munitekè and mundkè). Thus, it is quite possible that the earliest OAlb. 3s.impf. of ‘have’ was *kē. The origin of the 3s.impf. ending /-te/ (p. 136–7) is regarded by S. to be a local adverb of the meaning ‘thence’ or similar. This would first have been used in the present tense of /ve/ ‘to go’, for which we find OAlb. 123s. /vete/ ‘goes’, 3p. /venëte/, and thence spread to the 3s.impf. of other verbs. Yet S. admits that the motivation for this spread is unclear. I therefore wonder whether /-te/ may rather have spread independently, to the 3s.impf. on the one hand, and to /ve/ ‘to go’, on the other. Rather than directional ‘thence’, the meaning may have been locational ‘then’ (in the impf.) or momentary ‘now’ (in the present), meanings which we can derive from the preposition te ‘to, at’ or the adverb tej ‘over there, beyond’. The temporal meaning of tej is usually ‘later’ in Modern Albanian, but it could easily have had the mirror meaning ‘then, earlier’ in other contexts. S. makes a convincing case for interpreting the geminate in the impf. plural endings Buz. <-gniim> and Budi <-gnijme>, <-gnijte>, <-gnijne> as a long /ī/. When he concludes that these endings represent “Late Proto-Albanian” plural endings of the form *-īm, *-ītǩ and *īnnǩ, however, with a long vowel *ī of “completely unclear” prehistory (p. 140), I again call for caution. The element /nj/ in /-njīm/ etc. is clearly due to metanalysis from present stems in a nasal. As S. himself indicates on p. 135, the absence of /nj/ in Arvanitika and other Tosk dialects points to a post-Proto-Albanian origin of the grammaticalization of nj-endings in the imperfect. Furthermore, the most basic ending of the 3pl.impf. is ī-less /-në/ in all OAlb. authors. Hence, deriving the /ī/ in the plural impf. forms from the PIE optative suffix *-ih1- (as per Klingenschmitt 1994: 228–30) is problematic. In view of the duty to look for the most recent source possible, a connection with the 3s.impf. ending OAlb. postvocalic /-j/ (as in Budi /duoj/ ‘wanted’, Bogdani /drōj/ ‘feared’) is more attractive. The plural endings of the impf. would in that case have been formed on the basis of a 3sg. in *-i, to which the productive preterite endings *-m, *-tǩ, *-nnǩ were added. The vowel lengthening may be analogical to the plural of the vocalic aorists – which also have a long suffix vowel, as opposed to the singular endings. Next follow separate treatments of the v-, t- and suffixless aorist of Old Albanian. S. explains the v-aorist from PIE s-aorists to roots ending in *-eH > PAlb. *-ā-, of which OAlb. /shton/ ‘to add’ (< *steh2-) and /mblon/ ‘to fill’ (< *pleh1-) may be direct survivals. Intervocalic *s became *h (before back vowels), and the sequence *-āha- was then generalized as aorist marker. Of the plural endings /-uom/, /-uotë/, /-uonë/, all of which are regarded as regular outcomes of the PAlb. DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen 9 preforms by S. (p. 152), I would argue that only /uom/ (and maybe */-uon/, before it was remade into /-uonë/) regularly reflects diphthongization of *ō before a resonant. Analogical spread of /uo/ to the 2p. /-uotë/ is trivial. For the t-aorist, S. assumes an origin in the verbal adjectives in *-to-, specifically in intransitive verbs. Again I find the claim that we can say “mit hoher Sicherheit” how this class developed rather daring. The suffixless aorists are partly derived from PIE root aorists, partly from reduplicated perfects (if one accepts the proposal that a PIE sequence *T1ēT2- may reflect earlier *T1e-T1T2-),1 and partly by internal productivity within Albanian. Suffixless aorists with o-vocalism, which reflects earlier *ē or *ā, occur with ca. 30 roots of the structure *CeT-, *CeL- or *CeRT-, e.g.,2 djeg, pjek, vjell, del, hje(l)k, djerg, the present of which usually has /je/ from PIE *e (exceptions: /dal, del/, /marr, merr/). S. derives the long vowel from the PIE reduplicated perfect (see above), and dismisses an s-aorist origin. He does not consider the possibility that the PIE root aorist could have provided the lengthened grade that is needed. Yet PIE lengthening in monosyllabic verb forms, such as in the 2s. and 3s. inj. of the root aorist, has been around some time as an explanatory device (see Kortlandt 2009: 51–60). There is nothing “nicht systemangemessen” (p. 172) about exploring this possibility for Albanian, especially if one regards – as S. does – its verbal system as “comparatively conservative” (p. 170). Note that S.’s rule of consonant simplification will work only for the roots of the structure *TeT (ca. 11 of the oaorists), whereas in roots of the types *CeL, *CReT, *CeRT, the weak reduplicated forms *CeCL, *CeCRT, *CeCRT would not yield a two-stop cluster. Vowel lengthening in monosyllabic forms of the root aorist, however, would have affected all of the roots with o-aorist. Hence, this solution has more explanatory force. The aorist endings must be discussed together. Since the OAlb. endings of the imperfect and the aorist are very similar, and since the PIE endings of both categories were identical (*-m, *-s, *-t, *-me, *-te, *-nt), it seems worthwile to compare them explicitly: 1 2 This explanation was proposed by Schumacher 2005: 601–603, but did not meet with general acceptance. S. now (p. 163) proposes a modification that was suggested to him by Martin Kümmel, viz., that either *T1 or *T2 had to be a PIE voiced stop (*b, *d, *g, *ǵ, *gw) for the consonant loss and vowel lengthening to take place. A slight inconvenience is the absence of lists of OAlb. verbs that have a v-, t- or suffixless aorist, and mutatis mutandis the same goes for other verbal categories with morphological variation. The reader can collect the examples from the Lexicon, but that takes some time to accomplish. 10 DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen impf. v-aor. t-aor. zero-aor. PIE athem. 1s. /-ë/ /-a/ /-a/ /-a/ *-m 2s. /-e/ /-e/ /-e/ /-e/ *-s 3s. -Ø /-j/ act. /-i/ and /-u/ non-act. -Ø act. /-i/ non-act. /-ë/ -Ø *-t 1p. /-m/ /-m/ /-m/ /-m/ *-me 2p. /-të/ /-të/ /-të/ /-të/ *-te 3p. /-në/ /-në/ /-në/ /-në/ *-nt It is clear that the core of the plural endings continues corresponding PIE forms (with /-në/ from PAlb. *-nn- < PIE *-nt-), even if final /-ë/ must be of analogical origin. In the singular, the endings of impf. and aor. are very similar, but they cannot be linked directly to the PIE preforms. In 1s., impf. -ë vs. aor. -a recalls the difference between indefinite /-ë/ and definite /-a/ in feminine a-stems, which is commonly explained by cliticization of a fem. pronoun in the definite form, e.g., -ë < *-a vs. -a < *-aha < *-a-sa. S. tries a similar trick for aor. -a, assuming that PIE *-s-m became *-han which gave *-ǩ, to which the pronoun *u(ð) ‘I’ was suffixed. Contraction of *-ǩ-u would subsequently have given *-ā. In S.’s view, this contraction happened in the v-aorist, hence the ending 1sg. *-Ǥ.ǩu would have given *-Ǥā and then, with v as a hiatus filler, /-ova/. But this scenario is phonetically unlikely: How would *-ǩu contract into *-ā? Would *-Ǥǩ not have contracted at an earlier stage, when *h was lost? A smoother alternative is to assume that /-a/ in the v-aorist was taken from t- and/or zero-aorists. In general – as S. admits –, most of the vocalic presents with a v-aorist are secondary, so that it would be unsurprising if this category adopted the endings of a more original aorist type. On the basis of a few non-active variant forms in /-të/, S. posits for the t-aorist an original nonactive ending /-ë/ as opposed to active /-i/. He then explains the ending /-të/ from the PIE nom.sg. in *-tos of the verbal adjective in *-to-, the meaning of which would nicely explain the non-active semantics. Note, by the way, that an original ending *-s-os (that is, from the secondary suffixation of non-active *-os to the saorist) might explain the 3sg. endings /-i, -u/ of the v-aorist (puno-i, le-u), compare the masc. definite affix nom. -i / -u < *-os (p. 213). Since there is no good alternative explanation for the t of this aorist in general, S. posits that the active t-aorist DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen 11 was derived from the non-active source form by the addition of 1s. *-a, 2s. *-e, 3s. *-i, etc. This certainly seems an attractive solution for the t-aorist, though one would like to see it tested by an investigation of the semantics of all OAlb. t-aorists. If the endings of the t-aorist are also secondary, they might be more original in the zero-aorist. Future research might therefore concentrate on that category. The optative is derived from the aorist stem by means of the endings /-sha/, /-sh/, /-të/, /-shim/, /-shi/, /-shinë/. These are directly suffixed to the stem of zeroand t-aorists (where the ending appears as /-të-sha/ etc.), while in the v-aorists, the v was devoiced, yielding 1s. /-fsha/. Since /sh/ can result from PIE *s before front vowels, S. understandably explains the optative suffix from an earlier aorist optative in *-s-ī-. While this may be attractive for the suffix itself, the endings are not so easily explained. S. assumes that the inherited forms led (with some analogical remakes) to PAlb. sg. *-shǩ, *-sh, *-sh, pl. *-shim, *-shitǩ, *shinnǩ. Subsequently, the 1s. replaced final shwa by /-a/, the 3s. replaced *-sh by /-të/, and the 2p. replaced *shitǩ by /-shi/. The latter form would have been adopted from the 2pl.imperative, which seems acceptable. The 1s.opt. /-sha/ would be analogical to the ending /-a/ of the zero-aorist, though S. himself admits that this is not a formally perfect analogy (p. 179). Nor would the motivation for such an analogy be clear. The 3s. /-të/ is derived by S. from a four-part analogy between the sb. and the opt. of ‘to be’: sb. 2s. *tǩ jēš, 3s. *tǩ jētǩ :: opt. 2s. *klofš, 3s. *klofš >> *kloftǩ. Yet this is hardly credible. The f of the opt. stem klof- probably has the same recent origin as that of the productive o-presents, viz. from devoicing of hiatusfilling *v. Note that ‘to be’ is the only verb of which the aorist stem (/kle-/) and the optative stem (/klof-/) have different vocalism. As long as this difference is not explained, it seems wiser not to rely on /kloftë/ for the explanation of 3s. /-të/. If one accepts S.’s claim that the ending /-të/ cannot have arise from */-shtë/ (that is, that 3s. /kloftë/ cannot reflect */klofshtë/), a different solution must be sought (see below on this topic). S. (p. 181–182) also points to optative forms such as Budi /shkuosh/ ‘you would go’, Matranga /u shëjtëruatë/ ‘you would be sanctified’, Variboba /gëzuash/ ‘you would rejoice’, which seem to be based on the aor.pl. vocalism /uo, ua/ of the on-presents. S. therefore posits a simple analogy between the 3p.aor. active and the 3p.opt. to explain the formation of /shkuosh/ for */shkofsh/, etc. This explanation is somewhat awkward if these forms are synchronically archaisms, as S. himself proposes. Maybe it is better to take the 3s.aor. non-active with Ø-ending as a starting point, e.g., Buz. /u firmuo/ ‘changed’, /u harruo/ ‘was forgotten’, also in view of the role ascribed by S. to the 3s. non-active in the formation of the t-aorist (see above). 12 DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen Finally, another problem that VerbAlt mentions (p. 75) but does not dwell on is the derivation of the optative from the present stem (or, as VerbAlt rightly formulates, from the verbal “basis”, which is also found in the verbal adjective in all these verbs) in the case of zero aorists with o-vocalism. Some of the attested forms are Buz. 3p.opt. non-act. /u ëmbëljedhshinë/, Bogdani 3s.opt. non-act. /u mbëledhtë/ to /mbledh/ ‘to collect’, aor. /mblodh/; Buz. 3s.opt. /bdjerrtë/, Budi /vdjerrtë/ to aor. /bdor-/, /vdor-/ ‘to lose’; Buz. 3s.opt. /sjelltë/, 3p.opt. /sjellshinë/ to /sjell/ ‘to bring’, aor. /soll-/ (Fiedler 2004: 210–211). Other verbs in which the optative is reported to have the vocalism of the present are marr ‘to take’ (opt. /marr-/ in Buz., Budi, Bogdani vs. aor. /mor-/), del ‘to come out’ (opt. /dal-/ in Budi), heq ‘to pull’ (Buz. opt. /hjek-/ vs. aor. /hoq-/), ndjek ‘to follow’ (Buz. opt. /ëndjek-/ vs. aor. /ndoq-/), and djeg ‘to burn’ (Buz. 3s.opt. /djektë/ vs. 3s.aor. /dogj/). This correlation between opt. and pres. is still the rule in Modern Albanian. If one does not accept that these optatives are old, the only way to explain them is by analogy with the verbal adjective. After all, in all other verbs with vowel ablaut, the optative and the verbal adjective share the same stem (abstracting away from diphthongization of *ō to /uo/ etc.), and hypothetical opt. forms such as *muortë, *duoltë could have been replaced by /marrtë/, /daltë/ on the model of the ptc. /marrë/, /dalë/. In the light of all this – that is, in view of the retention of formal connections between the optative and either the aorist or the verbal adjective –, it is unlikely that the opt. basis /klo-/ of ‘to be’ would be an analogical remake on the model of the optative of the on-presents, as suggested by Matzinger 2006: 201, unless it already had o-vocalism. If the optative had been *klefsha (aor. /kle-/), there would have been no impetus to reform it, least of all for this most frequent of all optatives. I have no final solution for the optative, but one approach would be the following. The 2s.pres.sb. /marrsh/ and 2s.opt. /marrsh/ are identical in form. Since the second person is frequent in direct wishes and curses, the optative could be explained from the addition of the 2s. (present) subjunctive ending /-sh/ to the aorist/participle stem. Cross-linguistically, that is a frequent method of building a potentialis or irrealis. The plural endings /-shim/, /-shi/, /-shinë/ and 1s. /-sha/ then arose by addition of the preterital endings of the present (that is, the endings of the imperfect) to /sh/. Only the 3sg. /-të/ remains to be explained. It could have a nominal origin (as in the Lithuanian 3s.opt. in -tų from *-tum), for instance, from the t-participle which is still around in remnant forms in OAlb. Altogether, S. judges the verbal system of OAlb. to be comparatively conservative if one takes into account the late date of attestation. This impression logically results from S.’s confessed initial assumption that the verbal categories are likely to continue the exact same PIE categories. Unlike S. (p. 170), and as argued in the above lines, I see no irrevocable proof that the Albanian present subjunctive con- DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen 13 tinues the PIE pres.sb., that the non-active must necessarily continue the PIE middle, that the v-aorist continues a PIE s-aorist, or that we can be sure about the PIE predecessors of most of the OAlb. verbal endings. While many of S.’s proposals are good possibilities indeed, they must be reevaluated against a much wider scope of other theoretical solutions. I would avoid creating the impression of certainty in these matters. Three interesting pages are devoted to the etymology of verbal prefixes (p. 188–190), which are rarely discussed in other historical surveys of Albanian. There remains ample room for further study. For instance, in modern Standard Albanian, the preverb ‘un-, off-, apart’ has the forms ç- /tsh-/ before word-initial resonants, sh- /sh-/ before voiceless stops, and zh- /zh-/ before voiced stops and v: ç-mend ‘to make insane’, sh-qetëson ‘to worry’, zh-duk ‘to disappear’. Both preverbs come in different OAlb. variants, which S. divides in two groups: /ç-/, /xh-/, /sh-/, /zh-/ would reflect a borrowing of Latin dis- (though /xh-/ is unattested in OAlb. and is only substandard in the modern language), whereas /c-/, /x-/, /s-/, /z-/ would stem from a later Romance form *des-. This claim would need more detailed argumentation to be credible. Dialectal developments can complicate the picture, and this can also be observed in the OAlb. material collected in VerbAlt. A comparative study of the fate of inherited shC- and compound ç-C- in the different OAlb. authors would probably yield some interesting observations. Not all of the etymological proposals in this section carry conviction. The claim that *an- in /an-grë/ ‘s/he ate’ and /an-shtë/ ‘s/he is’ derives from PIE *on, an o-grade form of *h 1(e)n ‘in’, is based on the view that Slavic *vъ ‘in’ derives from *on (Schumacher 2007: 229, fn. 17). However, Derksen 2008 s.v. *vъ(n), prefers to derive it from the same Balto-Slavic preform *in < PIE *h1n as Lithuanian į ‘in’. This obviates the need to reconstruct PIE *on for Albanian /an-/, since the meanings of ‘ate’ and ‘is’ would also be compatible with a derivation from *ana ‘on, up’ (Gr. ana-). Initial f- in /ftoh-/ ‘to cool’ and /p(ë)-/ in /pëvet/, /pyet/ ‘ask’ could come from *apu, as S. proposes, but *upo seems equally likely. The range of possible preforms of the root /vet-/ in /pëvet/, by the way, is not so small as to allow for the claim that the verb must “undoubtedly” (p. 190 and 992) be derived from the same PIE root form *woit- as OCS věštati ‘to say’. In the same vein, I am surprised by S.’s trust in the etymology of marr ‘to take’ as *me ‘with’ plus *arnu- ‘to take’ (p. 190, 983). The explanation is ingenious and theoretically possible, but no more than that. Chapter E (p. 205–276) deals with the historical phonology, describing the main developments of PIE phonemes to Old Albanian, as well as some specific OAlb. changes. From the bibliography on p. 205 the main omission of recent date is Vermeer 2008. Because of the small number of reliable examples for nearly all 14 DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen sound changes, writing a historical phonology of Albanian has always been a hazardous enterprise. S.’s survey ranks among the longest and most reliable treatments of this subject. It contains a number of interesting new proposals, in particular on the relative chronology of various changes, the conditions for vowel lengthening in the Late PAlb. system (after the collapse of inherited vowel quantities), the vocalization of PIE resonants, the prehistory of the noun /ujë/ ‘water’ (though without mentioning of Matzinger 2007), the paradigmatic alternation explaining i madh ‘big’, the etymology of hurbë ‘gulp’, the paradigm of dhjes, dhjet ‘to defecate’, and many other details. In chapter E.11, final devoicing is traced back to Late PAlb., and the vacillating reflexes in OAlb. are convincingly ascribed to different kinds of paradigmatic leveling. To my satisfaction, the unlikely etymology of /dēt/ ‘sea’ as PIE *dheub-eto- ‘depth’, which has been passed on by generations of Albanologists, is firmly dismissed by S. (p. 233), as is the claim that Albanian /h-/ could be the direct reflex of a PIE laryngeal consonant (p. 267). Perhaps the least convincing novelty in E is the proposal (p. 214) that wordfinal *-i of PIE was apocopated early in the prehistory of Albanian but only in trisyllables ending in *-ti, *-si or *-nti. This would not have been a sound law “im engeren Sinne”, but an undefined “Vorgang” (‘process’) affecting unstressed, clause-final verb forms. This rule is intended to explain the zero ending of the 2s. and 3s. pres.ind. from *-ih < *-eh < *-esi resp. *-it < *-et < *-eti. But this rule then forces S. to assume that the 2s.sb. ending -sh derives from *-ēsi plus an unknown particle, since its *-i must have been retained when it was lost in the indicative endings. To my mind, this proposal is based on the unwarranted assumption that the pres.ind. endings have to be based on a combination of thematic and athematic endings of the Sanskrit type -ā(mi), -asi, -ati; one needs only to look at Greek. 3s. pres.ind. -ei to see that things are not so simple. The hypothesis again presupposes more clarity about the development of the PIE endings in Albanian than we actually possess. More on the history of Albanian is contained in chapter H, ‘Selected Etymologies’, which discusses 116 Old Albanian verb stems. Interesting new solutions are proposed, among others, for /duketë/ (to *deuk-), është and its variants, heq, and others. A detailed discussion of all proposals in this chapter would require a separate, lengthy review. The only suggestion which I will make here is that vret, aor. vra- ‘to kill’ might contain the same preverb v- < *wi- as vdek ‘to die’ (as hinted at by S. on p. 190), and that the aorist stem /vra-/ might consist of *vë- plus the aorist /ra-/ of ‘to fall’. In other words, */vë-ra-/ would have meant ‘to fell’ > ‘to kill’, and the present stem */vratje-/ would represent */vë-ra-/ plus /-tje-/, compare /pres/ ‘I cut’, /pret/ ‘he cuts’, aor. /preu/. DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen 15 Part F, ‘Hinweise zum Gebrauch des Wörterbuchs’ (p. 277–286), is given in German and in an Albanian translation by Genc Lafe. In the lexicon (Part G), the OAlb. verbs are cited in the 3sg. present indicative, as is done in the dictionaries of Buchholz/Fiedler/Uhlisch 1981 and Newmark 1999. Each entry is subdivided according to the different OAlb. authors that use the verb and all finite and non-finite forms of the stem are then given. Each verb is accompanied by a German translation or by an English gloss taken from Newmark 1999 (if the verb is absent from the German dictionary). In addition, verbs from Matranga are also cited with the Italian translation. The same is true for part of the verbs occurring in Bogdani, which are sometimes cited with their Latin counterpart, too. For verbs occurring in Matranga and Variboba the Italian or Greek source word from which they were borrowed is often provided. The actual verb forms are presented in two ways, viz. in their original spelling as attested in the documents and in a phonological interpretation. Exact references to the page numbers or paragraphs in the OAlb. texts enable the reader to retrieve the evidence in its original context. Further, grammatical information, such as on the transitivity of verbs, and on different syntactic constructions into which they enter, is also provided for many forms. As indicated in the introduction, the Lexicon gives all the verb forms occurring in the texts of Budi, Bogdani, Matranga and Variboba. Of Buzuku, the attested stems are only cited in the 3s.pres.ind. together with their meaning, and a reference to the page in Fiedler 2004 who cites all attested tokens. It is inevitable that a Lexicon with thousands of forms contains some small errors, but (after intense perusal of at least the present and imperfect forms) I encountered only a very small number of them. On p. 381, s.v. /dobit/ (Budi), replace the first “Prs. Ind. nicht-akt.” by “Pres. Ind. akt.”. On p. 430, s.v. /falet/ (Variboba), replace “akt.” by “nicht-akt.”. On p. 520, s.v. /këput/ (Budi), replace the first “Prs. Ind. nicht-akt.” by “Prs. Ind. akt.”. On the following line, replace “3.Sg. /këput/” by “2.Sg. /këput/”. On p. 566 s.v. /lëvizën/ (Budi), replace “Prs. Ind. akt.” by “Prs. Konj. akt.”. On p. 657, s.v. /ndivënen/ (Bogdani), replace “3.Pl. /ndivënjenjinë/ by “3.Pl. /ndivënenjinë/”. On p. 779, s.v. /rref/ (Budi), replace “3.Pl. /rrafnjëmë/” by “1.Pl. /rrafnjëmë/”. On p. 802, s.v. /soditën/ (Bogdani), replace “3.Pl. /sodisëne/ sodissenè” by “3.Pl. /sodisënë/ sodissènè.” Corrigenda elsewhere in the text: on p. 181, last two lines, read “3.Pl.” four times for “3.Sg.”; on p. 186, line 4, replace “Imperativs” with “Infinitivs”. On p. 526, the interpretation of Bogdani’s 1s. impf. act. krehigne /krehinjë/ as belonging to a stem /krehën/ is possible, but /krëhín/ cannot been excluded (see above on ín-stem presents). On p. 662, s.v. /ndjetë/ (Bogdani), I would interpret the attested 3s. form ndȣtè /ndytë/ as aorist rather than imperfect because of the /y/-vocalism, which is found in Bogdani’s other aorist forms and in the ptc. /ndȳm/ 16 DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen in Budi and Bogdani (see VerbAlt p. 986 for the background of this vocalism). True, the ending /-të/ is not otherwise found in the 3s.aor. other than with ‘to be’, ‘to have’ and ‘to say’, which may explain why S. regards /ndytë/ as an imperfect. But Buzuku’s pres. /ëndotë/ ‘it seems’ has the same ending. This form is plausibly explained by S. as a secondarily created o-aorist to an inherited deponent, the present *ǩndjétǩ; the model was provided by the presents in /je/ plus aorists in /o/ discussed above. Budi has the 3s.aor. /ndotë/. Hence, Bogdani’s hapax /ndytë/ may tentatively be interpreted as a remake of earlier */ndotë/ on the model of the forms in /ndy-/. On p. 683, the interpretation of the present stem of ngjit in Bogdani as /ngjitën/ rests on the 2p. imperative active /ngjitëni/, which could also belong to a stem /ngjit/. The 2p. ending /-ni/ sometimes alternatives between /-ni/ and /-ëni/ after obstruents, as in Bogd. /shifni/ next to /shifëni/ ‘you see’ (p. 813). This vacillation is obviously the result of shwa-insertion in the cluster, which has the effect of obliterating the difference between the 2p. ending of djeg-presents and ecënpresents. Hence, it seems best not to rely on the 2p. ending for determining the OAlb. conjugational class. The points raised in this review confirm how interesting the field of Albanian historical linguistics is, and how much work, philological and linguistic, can still be done. I have tried to show that some of S.’s choices of reconstruction can be contested on methodological grounds, but also, that many of his solutions are convincing. I have great admiration for the meticulousness and the acumen of the book presented by Schumacher and Matzinger, which clearly is the work of reference for the years to come. It is, to all intents and purposes, a great leap forward. References: Buchholz, Oda, Wilfried Fiedler & Gerda Uhlisch. 1981. Wörterbuch AlbanischDeutsch. Leipzig: Enzyklopädie. Derksen, Rick. 2008. Etymological Dictionary of the Slavic Inherited Lexicon. Leiden: Brill. Fiedler, Wilfried. 2004. Das albanische Verbalsystem in der Sprache des Gjon Buzuku (1555). Prishtina: Akademia e Shkencave dhe e Arteve e Kosovës. Klingenschmitt, Gert. 1981. Albanisch und Urindogermanisch. MSS 40, 93–131. Klingenschmitt, Gert. 1994. Das Albanische als Glied der indogermanischen Sprachfamilie (Tischvorlage). In: J.E. Rasmussen & B. Nielsen (eds.), In honorem Holger Pedersen. Wiesbaden: Reichert, 221–233. Kortlandt, Frederik. 2009. Long vowels in Balto-Slavic. In: Baltica & BaltoSlavica. Amsterdam/New York: Rodopi, 51–60. DE VAAN, MICHIEL: Schumacher/Matzinger, Die Verben des Altalbanischen 17 Matzinger, Joachim. 2006. Der altalbanische Text [E] Mbsuame e Krështerë (Dottrina cristiana) des Lekë Matrënga. Eine Einführung in die albanische Sprachwissenschaft. Dettelbach: Röll. Matzinger, Joachim. 2007. Altalbanisch < ieh>/ujë/ und die Kategorie der Massennomina bei Buzuku. In: B. Demiraj (ed.), Nach 450 Jahren. Buzukus »Missale« und seine Rezeption in unserer Zeit. 2. Deutsch-Albanische kulturwissenschaftliche Tagung in München vom 14. bis 15. Oktober 2005. Wiesbaden: Harrassowitz, 169–190. Meiser, Gerhard. 2003. Veni Vidi Vici. Die Vorgeschichte des lateinischen Perfektsystems. München: Beck. Newmark, Leonard. 1999. Albanian-English Dictionary. Oxford UP. Schrijver, Peter. 2006. Review of Meiser 2006. Kratylos 51, 46–64. Schumacher, Stefan. 2005. “Langvokalische Perfekta” in indogermanischen Einzelsprachen und ihr grundsprachlicher Hintergrund. In: G. Meiser & O. Hackstein (eds.), Sprachkontakt und Sprachwandel. Akten der XI. Fachtagung der Indogermanischen Gesellschaft, 17.–23. September 2000, Halle an der Saale. Wiesbaden: Reichert, 591–626. Schumacher, Stefan. 2007. Kontinuanten urindogermanischer Wurzelaoriste im Albanischen. Teil 1: Wurzelaorist mit frühuralbanischem Stamm auf Vokal oder auf *ś. International Journal of Diachronic Linguistics and Linguistic Reconstruction 4, 207–280. Vermeer, Willem. 2008. The prehistory of the Albanian vowel system: A preliminary exploration. In: A. Lubotsky, J. Schaeken, J. Wiedenhof (eds.), Evidence and Counter-Evidence. Essays in honour of Frederik Kortlandt. Amsterdam/New Yok: Rodopi, 591–608. Michiel de Vaan Université de Lausanne Faculté des Lettres Section des sciences du langage et de l'information Unil, Anthropole-3134 CH-1015 Lausanne Switzerland [email protected]