The Grammar of West Middle Iranian (WMIr) by Durkin-Meisterernst (DM) is designed to meet a long-standing and much needed desideratum in Iranian studies. Over 602 pages, DM gives a detailed survey of WMIr corpora (25 pp.), secondary sources (3 pp.), writing systems and palaeography (55 pp.), sound systems and sound changes (64 pp.), word formation (49 pp.), and a description of word classes, their categories and historical derivation (64 pp.), of formation and usage of grammatical categories (143 pp.), and of syntax sensu stricto (53 pp.). The appendix contains chapters about verbs of the mixed Parthian and Persian text Draxt asūrīg, and the attestations of numerals. The index of quoted material facilitates use of this book, and the bibliography is in itself a helpful guide. Because of its clear structure, it should also be easy to handle for non-native speakers of German.
Based on a large body of material throughout, DM's descriptive approach yields a large amount of reliable data, neatly arranged. The transliteration of examples is given in the appendix, together with their sigla, although the sigla would be better placed in the text as helpful information on localizing linguistic features. This leads to the general problem of how to describe a language which is attested over several centuries. Such a corpus is by no means homogeneous in either time or space. DM offers a broad classification by specifying the examples as Manichaean (MpT/PaT), Zoroastrian (MpB/PaB), and inscriptional corpora (MpI/PaI). However, this information is not always used for the evaluation of grammatical data. A general summary could have combined the observations and offered an outline of language development or a guideline to localize WMIr text in time and space by means of grammatical features. Chronological differences within WMIr are not specifically addressed, and some variations are considered New Persian (NP) influence (e.g. interchangeability of kē, ka and kū, p. 441 n. 265, or the use of rāy as object marker, §753). However, NP forms have a predecessor. So when there is a fitting variety, the differences between MpB and MpT may be dialectal or chronological rather than correct vs. corrupted MP. When MpB shows in most texts a clear difference of the 1sg. -m and 1pl. -ym in verbal endings (which matches the NP state), it is unlikely to assume that this is due to an orthographical convention as represented in MpI (where both are -m). Besides, 1sg. endings other than -ym are also attested for MpT (e.g., phypʾrwm in M842 V.2). Nevertheless, DM's focus on Manichaean texts is well justified due to the much clearer data.
Although Parthian and Persian are close in many respects, DM conscientiously keeps them apart. His fine-grained analyses of the semantic differences make this grammar a most useful handbook for future editors. It is a key task of grammars to differentiate lexical meaning from grammaticalized function. Very occasionally one might disagree with DM's analysis in this respect. For instance, the supposed inchoative notion of the verb kām- “to wish” (§825) results from its lexical meaning (a striking example for a grammaticalized inchoative would be “the sun wished to sink” in the sense of “the sun began to sink”). A diachronic representation would help to put such statements into perspective, yet DM is rather hesitant in this respect. On the one hand, the chapter on the phonological system shows a solid etymological background (however, inner-WMIr sound changes are not systematically presented). The presentation of inherited differences and common features and of new inventions is very helpful. On the other hand, the section on word formation offers little historical information (e.g. etymology of affixes). For instance, xwarxšēd is unlikely to be a MIr compound because of the preserved -x- of Old Iranian xšāita-.
DM does not accept Agnes Korn's analysis that the letters ‹c› and ‹z̈› represent historical *č, and ‹j› *ǰ or *ž (“Parthian ž”, BSOAS 73/3, 2010, 415–36). However, his counterexample ‹ʾbʾj› (fragment S32 B3) is clearly ‹ʾbʾz̈›, the verb twc- may show pseudo-historical spelling (Antoine Ghilain, Essai sur la Langue Parthe – son système verbal (...), 1939, 43 quoted by DM), and wjydg/wcydg “Chosen, Elect” could be MP with Parthian orthography (Korn, p. 422) or later spelling. Hence, it would be good to maintain the otherwise consistent orthographic differentiation of etymological *č vs. *ǰ and *ž in the transcription.
DM states that there is no marking (“Kennzeichnung”, p. 384) of grammatical aspect in WMIr. The question remains whether the so-called simple past and present tense had aspectual functions. The simple past is attested for expressions of coincidence (like “herewith I give” in the MP law book). The sudden change of present tense and simple past in some texts can well be explained by imperfective and perfective aspect (background vs. foreground action, ongoing vs. momentary action), while a temporal or stylistic interpretation seems less motivated.
A peculiarity deserving close scrutiny is the development of relative pronouns (RP) to subordinators without actant reference (relative particles). DM mentions (p. 416) that enclitic pronouns (EP) can express possessiveness of the RP (e.g. kē-š “whose”), but he does not conclude that the RP can no longer be pronominal. This development explains the spread of these pronouns to animate (čē instead of kē) and inanimate (kē instead of čē) antecedents. Free relative clauses (called indefinite by DM) usually occupy a position in the matrix clause (ex. 904, abar hō in the following close is co-referential, but not a correlative in the strict sense).
Interrogatives usually appear in front of the finite verb (as in ex. 177 and see Dādestān ī Dēnīg or Mēnōg ī Xrad for further examples), not clause-initially as stated by DM, whose examples 173–4 simply happen to contain only interrogative and verb in a clause.
Regarding adpositional phrases (§605 ff.), I would like to add that the functional prepositions MP az, ō, and pad always require a complement, while lexical ones (like andar) can be used adverbially. Consequently, enclitic pronouns attach to the first, but not to the second group. With the latter they may continue the function of a dativus commodi. When functional prepositions are combined with a 3sg. EP, they become adverbials (“therefrom”, “thereby”, etc.). Imitating the word order of lexical prepositions, EPs with dative function can appear as well so that it seems that the complement of the preposition appears twice (u-š ... az-iš, even u-m ... az-iš). The dative function may also account for problematic cases with seemingly oblique subjects of intransitive verbs. In the case of ex. 163 u-šān hamāg may be analysed as “all of them” rather than “they all”. These questions, and DM's very elaborate study of adpositional phrases, are worth connecting with the issue of valency, which DM addresses only briefly. It seems that DM uses valency for what is also called “case frame” because he considers the logical subject of an ergative construction a third actant (“indirect object”). However, according to semantic or syntactic valency this constituent remains the same (agent or subject). A different explanation may be favoured for modal verbs, which are not subjectless, but take the infinitive as their subject.
Regardless of these few critical comments, this marvellous study is a model of accuracy. Its detail (over 1,000 examples) forms a solid and inspiring base for future investigations. It is a treasure of information diligently compiled and nicely arranged by a true expert in MIr philology. This book is also highly important for neighbouring fields.
