METHOD AND CORPUS DEFINITION
The book under review aims to describe the domain of epistemic modality in Old Babylonian. It focuses specifically on particles, other than those associated with conditional constructions. After an introductory chapter that presents the topic of modality and discusses the corpus, the book devotes separate chapters to a range of particles: 1) pigat; 2) midde; 3) wuddi and anna; 4) lu ittum; 5) -man, ibassi, la, and agar for the irrealis; 6) tu sa; 7) kik, ki sa, and kima ki; 8) assurre; and 9) -mi. The book concludes with a synthesis of the main results, followed by indices. In this review, I will discuss each chapter, while offering different or additional data, different views, and new insights.
Wasserman does not adopt any particular method of linguistic analysis. The tools employed are eclectic. Structural analysis is sometimes used, e.g., in the discussion of the syntagmatic compatibility of -ma and -mi (pp. 182-84), but at other times it is not: for example, we do not learn what the implications are when two modal notions co-occur, as when the apprehen-sional particle assurre occurs with or in conditionals.
Nevertheless, the orientation of the book is claimed to be linguistic (p. 1). In linguistics, it is widely accepted that descriptive statements should apply to a specific, defined system. Indeed, the corpus upon which the present study is based is defined in the introduction (pp. 14-15), but only material drawn from the corpus of letters from Mari is actually discussed there. There is no mention of the corpus of letters from central Mesopotamia, the literary texts (Old and Standard Babylonian), or the laws, all of which are used as primary material in this book. More importantly, there is no discussion of whether these different text types constitute a single linguistic corpus or several corpora.
PRESENTATION OF DATA
The examples are presented in transliteration, rather than transcription, as is habitual in Assyriology, which allows the author to remain noncommittal about the interpretation of
Review article of Most Probably: Epistemic Modality in Old Babylonian. By NATHAN WASSERMAN. Languages of the Ancient Near East, vol. 3. Winona Lake, Ind.: EISENBRAUNS, 2012. Pp. xiv + 245. $49.50. Disclosure: As a Ph.D. student (1998-2001), I worked as research assistant on the Modal Particle project, and was involved in collecting and feeding the material into the database. I have since published a number of studies which have to do with various types of modality in Old Babylonian, e.g., Cohen 2005a, 20061% and 2012. specific forms. At times, this also makes it difficult to grasp the author's understanding of examples. Furthermore, the examples are adduced in block form. No typographical means (e.g., boldface, italics, underlining) is used anywhere to direct the reader to the main point of the example. Akkadian words are sometimes broken up, with part of the word carried over to the following line, which is hardly ever done in actual cuneiform texts.
In some examples from literary sources, the citation does not state the tablet, column, and line numbers, preferring instead the notation of the page and line numbers, e.g., "Lambert and Millard 1969:68:370" for Atrabasis I (vii):370.
The examples are hardly ever cross-referenced; perhaps for this reason the examples are not numbered. Since it is necessary to discuss specific examples from the book, I will refer to them by page number and example; for instance, "ex. 133/1" is to be read as "page 133, the first example on the page."
The book is marred by a number of unfortunate technical issues: 1) Breaks in the original cuneiform texts are occasionally not reflected in translations, which may give the impression that a reading is certain when it is merely conjectural, e.g., exx. 19/2, 21/1, 28/1, 29/1, 40-41, 77/2, 78/1, 89/1. 2) In several places, half of the cuneiform sign is carried over to the next line, e.g., exx. 40/3, 50/2, 59/1, 76/4. 3) Ex. 76/4 is identical to ex. 77/4 on the facing page, which is clearly the result of editorial negligence.
Some translations are different from what is expected, but no explanation is provided: Ex. 19/3: ida,s. "is deceiving us"--+ "is deceiving (us)."
Ex. 51/1: "Ibi (PN) . . . (make them) release him- -4 "they will/might have him released" (plural).
Ex. 52/2: ."vuniiti ul taqiup "you will not trust them"--+ "you do not trust them" (unbound object pronouns are highly marked).
Ex. 66/1: a-ta-ap-ra-kum "I kept writing to you" "I have written to you" (the form is not TN but rather the perfect).
Ex. 68/3: a.ga "(only my) field they cause to perish?" and a.git "did the field perish?" are not translated as precatives. Since the functional values of preca-tives is never past and rarely present, I suggest reading the /i-sign as a phonetic complement for a.ga: eqli and eqli Otaliq, which makes better sense syntactically, and even fits the translation.
The terminology employed is eclectic, and it is used in an idiosyncratic way: "text of interlocution" (p. 17), where 'interlocution' is an inelegant alternative term for 'dialogue', the same applies to "spoken utterances between the communicating parties" (p. 44).1 The words "adhesion" (p. 30) and "adhesive" (p. 129) seem to be used to represent the ideas behind the technical terms 'clisis' and Wasserman uses "optative" (p. 43, which is a mood in Greek, expressing, e.g., unreal wishes) when he means 'disjunctive' (the relations between two alternatives, 'or'). The phrase "generic distribution" is used (passim) for the distribution of the particle. One can also find expressions which seem to be oxymoronic, such as "counterfactual certainty" (p. 73), where 'counterfactual', although referring to a state of affairs which is unlike reality, nevertheless presents it as nonfactual. Another pair is "topical comment" (p. 59), which may be fine in everyday language, but not in linguistic terminology: 'topic' is a discourse anchor, containing presupposed or backgrounded information, having two functions: 1) to maintain cohesion between parts of the text and 2) to represent what is being discussed. 'Comment', on the other hand, refers to the new information, above clause level, that is said in reference to the topic. These terms stand for two markedly different, almost opposite notions.
Finally, the oxymoronic term "negative affirmative" (p. 5) refers to the negative forms in the asseverative paradigm in OB. The latter term seems perfectly adequate (see Cohen 2005a: 17-68). When a new term is introduced, it should be explained and justified.
Now I turn to the review of the specific chapters.
piqat is said to be a "weak doubter," and thus to reflect an open possibility, lower than midde on the scale of likelihood (which is termed the "epistemic scale" in Akatsuka 1985). In addition, piqat functions 1) as a disjunctive (like u/Ct 'or', ex. 21/1); 2) occurring in what Wasserman calls "semiconditional constructions";2 and 3) softening the intensity of the utterance (which is termed "downtoner"). Wasserman uses the terms "perspectivization" and "subjectification" to denote the level of the speaker's involvement in what is said. It is important to note that this scale is not necessarily modal. For example, focus phenomena reflect speaker involvement--they are much less common in detached narrative than in dia-logue--but functional sentence perspective, where focus belongs, is not generally considered to be modal by nature.
The difference between piqat and midde is claimed to be related to the available background information (ex. 35/2-3), but this hypothesis is based on a badly broken example. In addition, it is contradicted by an example with midde having no background information (ex. 50/1). Moreover, it is admitted (p. 36): "Nonetheless, it is important to note that in some cases piqat is used where one expects to find midde." In the end, it is difficult to understand what the real weight of this stated difference is.
Contrary to what is suggested in ex. 21/2 ("perhaps this army goes to Karana or to Andarig. I don't know"), piqat seems to introduce an indirect question: piqat ab[tdm ana GNI ul[ri-ma] ana ON. i[//]i ul ide "I do not know whether the army [gods to GNI o[r] to GN2)," very much like gumma 'if' (see for instance, AbB 3,53:14-15 and Cohen 2014: 70-73). This is not the only place where piqat is similar to gumma (see below and ibid. 73-74).
Ex. 34/1 has piqat introducing the apodosis, and the rare iprus form in the apodosis (wilco tells us that this is an inferential conditional: "If they do not come to you. (then) they must have gone to GNI or to GN,."
Two loci are missing for piqat: AbB 3,3:17-19 and 12,40:12-18.
nzidde has been translated in the relatively recent Assyriological literature either as a modal particle of (in)certitude or as a conditional particle. This makes it impossible to state the basic meaning of the particle, as these meanings are perceived by Wasserman as contradictory (p. 46, but see below). The putative source for the particle (man ide), needs to be preceded by an asterisk, as does any reconstructed form (*man does not exist as such in OB). midde is said, on the one hand, to reflect a higher degree of certitude than piyat (p. 47) but on the other, to show different levels of certitude. It seems hard to reconcile these two statements.
Wasserman uses the term "quasiconditional constructions" (p. 52). He neither explains whether there is a difference between "quasiconditional" and "semiconditional," the latter used in the chapter on M4(0 (p. 22). Nor does he explain the terms themselves, namely, whether the conditionality of the constructions is somehow in doubt. In other words, are these real conditional constructions or not? Do they...