Indirect Representation of Questions in Old Babylonian Akkadian

35

Transcript of Indirect Representation of Questions in Old Babylonian Akkadian

eran
Typewritten Text
Babel und Bibel 7

Indirect Representation of Questions in Old Babylonian Akkadian*

Eran Cohen

The Hebrew University [email protected]

To the memory of Gideon Goldenberg

The paper examines whether and to what extent a category “indirect question” is valid for Old Babylonian Akkadian. In order to do so, some fundamental ques-tions are asked about the very phenomenon of “indirect question.” Several strategies are analyzed, and it is concluded that indirect questions do exist in Old Babylonian and occur in two major patterns: either 1. they stand juxtaposed to a nominal anchor (most notably annītam lā annītam), which formally takes the object slot of the question introducing expression, or 2. they take the object slot directly.

Keywords: question, indirect question, free indirect question, reported dis-course, free indirect discourse, embedded direct question, clausal object, syntax, Old Babylonian syntax. 0. Introduction

This paper is about whether and to what extent a category of ‘indirect question’ can be put forward and justified for Old Babylonian Akkadian (henceforth OB). This is not just a matter of assembling and classifying examples from a corpus. It is also necessary to ask rather fundamental questions about the very phenomenon of ‘indirect question.’

In the introduction, I frame the discussion by considering two notions relevant to the domain of indirect questions: (direct) questions (§ 1) and reported discourse (§ 2). The relationship of indirect questions to other types of complement clause will come up along with the discussion of the Akkadian material. The following section deals with the conceptualization of indirect questions (§ 3). I proceed to the situation obtaining in An-

* I would like to express my gratitude to my colleague Eitan Grossman for his

wise remarks and corrections. I would also like to thank the colleagues behind the wonderful ARCHIBAB database (http://www.archibab.fr), which made it pos-sible to both search and approach a great many texts and thus improve my re-search work.

52 Ancient Near Eastern Studies: Articles

cient Semitic (§ 4), and under § 5, the bulk of the paper, indirect ques-

tions in Old Babylonian Akkadian are presented and analyzed. 1. Question: FSP and general nature

A question, when used to perform an interrogative speech act (as op-posed to, e. g., a rhetorical question), shows a mix of characteristics from the domains of both epistemic and deontic modality: like other epistemic expressions (e. g., “maybe”), it expresses non-certainty with regard to some of its components (naturally, the ones at the center of the question). At the same time, it appeals to the addressee for this piece of information, in a manner similar to deontic expressions, such as imperatives.1

There are basically two types of questions: the first type is termed pronominal (or constitutent, or partial, or wh-) questions, asking about a specific component (‘what,’ ‘which’ ‘where,’ etc., each capable of repre-senting an unknown sentence constituent):

[1] mannum īdēšu ‘who knows him?’ (AbB 4, 139:35).

[2] mīnam tugarranni ‘why do you sue me?’ (AbB 6, 41:10).

There is a multifaceted relationship between questions and focus: it oc-curs with constituent questions (Givón 2001 II 237–238, 307–310). This re-lationship is explained by the fact that interrogative pronouns and adverbs are rhematic, that is, they represent the new information yet to come, and these focus constructions mark them as such:

[3] kunuk mannim-ma imma¶¶ar ‘whose seal (is it that) will be ac-cepted? (AbB 11, 90:29) (for OB, see Cohen 2005a:35–36).

[4] mannum ša kīma yâti irammuka ‘who (is it) who loves you like me?’ (AbB 11, 9:3–4).

In both examples, the interrogative pronoun itself is marked as focus, by the particle -ma (ex. [3]) and by cleft consruction (ex. [4]).

The second type is nexus (or yes/no, polar, or total) questions, which ba-sically ask about the polarity of the predicative link (= nexus),2 that which exists between the two parts—the presupposed or given part (theme) and the new part (rheme). In other words, a nexus question in essence requires

1 See various other characteristics in Givón 2001 II 291, dealing with the in-

terrogative speech act. Frajzyngier 1995:476 coins the term “interrogative modal-ity,” but he on the other hand considers it to be part of epistemic modality only.

2 For a full historiographic account of the predicative link and a syntactic proof for its existence, see Goldenberg 1985.

E. Cohen, Indirect Representation of Questions in OB Akkadian 53

some polar response to the possibility of predicating the rheme (new infor-mation) over the theme (presupposed or given information):

[5] amatki anāku3 ‘Am I your servant?’ (AbB 6, 55:4).

[6] ul a¶ūka a[n]āku ‘am I not your brother?’ (AbB 6, 67:3′).

[7] bītum ul bīt[k]a ‘Is the house not [yo]ur house?’ (AbB 5, 84:4′).

Nexus questions occasionally occur with expressions of focus (Givón 2001 II 297–299):

[8] anāku ša unnedukkim šūbulim ‘Am I the one to send the docu-ment?’ (AbB 1, 23:10–11).

Focusing is here marked by a special non-verbal clause pattern (Cohen 2005b:265–266). It does not seem to change the nature of the question, which still seems to lie on the predicative link (namely, the once found between ‘I’ and ‘the one to send …’).

Another aspect of questions is their type of marking, especially whether they are related to the expression of modality in a language (that is, whether interrogative is marked by forms which show any affilia-tion with the modal paradigm in general). The issue, raised by Palmer (1986:30–33, 78–81), is important in the examination of indirect ques-tions, for instance, whether they express such interrogative modality oth-erwise found in direct questions, or not.

To sum up, questions participate in three domains: (1) they are a kind of speech act; (2) they have an important interface with information structure; and (3) they are, at least potentially, part of the modal system of a language. 2. Indirect or reported discourse

Indirect questions are occasionally considered part of the broader do-main of indirect or reported discourse (e. g., Palmer 1986:134). This topic has been intensively covered from many angles.

Jespersen (1924:290–300) mentions two types of indirect speech, de-pendent and represented speech (that is, free indirect discourse). He dis-cusses various types of shift between direct and indirect, involving per-son, tense, and/or mood. He goes on to illustrate the conflicting strategies

3 An underline with a vowel refers to a pertinent lengthening, such as one

that signals a question.

54 Ancient Near Eastern Studies: Articles

in German examples of indirect speech, which are not always subordi-nate (but are nevertheless referred to as “dependent”).

Coulmas 1986 tells us that reported speech is an adaptation of the original utterance to the speech situation of the report as well as to the reporter’s point of view. Tense, mood, pronominal system and reference distinguish direct from reported discourse, as well as word order. In re-ported discourse, more information may be introduced (whose original source, that is, whether it is the original speaker or the reporter, is not always easy to identify). The main issue is that the utterance is “analyzed” by the reporter. This can be done minimally, by providing merely deictic adjustments. In addition to direct and indirect discourse, there is a third kind, free indirect discourse, which is a strategy used in modern narra-tives to express the point of view of a character within the narrative. This kind is often exemplified by free indirect questions, which constitute an important part of free indirect discourse, see below.

Plank 1986 discusses indirect discourse, providing a list of criteria by which it is differentiable from direct discourse: deictic expressions that include personal pronouns, referring expressions, spatial and temporal adverbials, and tense. In addition, subsystems that reflect social and epis-temic distance show differences between the two types of discourse.

Palmer 1986 is a typological investigation of the category of modality. Nevertheless, it contains one of the most succinct—yet precise—descriptions of the essence of indirect speech (Palmer 1986:134–140, 163–167). Palmer, like the others (all published during 1986), insists on deixis as the decisive criterion:

“The most regular, and most important marker of an indirect speech clause is found in the deictic markers it contains. In general there is a switch from the deictics used by the original speaker to those appropriate in the actual speaker, now reporting what was or might have been said” (ibid. 163).

Thus, for Palmer, a shift from ‘hearer’ to 3rd person, or (in Hopi) from proximate to obviate, or (in Navaho) to the “4th” person markers, is enough to claim that the discourse is indirect. Further examples of deixis include tense and spatial and temporal relations that have to do with the speech situation. Palmer also elaborates upon modal expressions in indi-rect speech. The account is unique because it relies on cross-linguistic evidence, and thus carries the most weight.

Goldenberg 1991 discusses Biblical Hebrew as well as other Semitic languages. His general statement consists of: 1. emphasizing the scalar

E. Cohen, Indirect Representation of Questions in OB Akkadian 55

nature of the difference between direct and indirect (with free indirect discourse in between); and 2. breaking off the traditional links between subordination and indirect discourse, on the one hand, and on the other, between syntactic independence and direct discourse. In other words, di-rect speech can be embedded, and indirect speech can be found without the introduction of a matrix verb. 3. Indirect questions

This topic is discussed in descriptions of individual languages, especially English (e. g., Curme 1931:181–183, 211–212 and Jespersen 1961 III 39–49), whether by descriptive linguists or by linguists working in various (generative and functional) theoretical frameworks, and in works dealing with textlinguistics. A typological account does not yet exist.

Jespersen (1924:297–298) mentions both types—dependent questions and represented speech. He mentions marking—the loss of characteristic intonation of the question, but compensation by way of matrix verb (e. g., ‘ask’), an introductory exponent, and a distinctive order of elements. However, he mentions that it has become common to use, in indirect questions, forms peculiar to questions in represented speech (which may be termed “free indirect questions”). Except in quotation of direct speech, these dependent questions are used as substantive clauses with certain verbs. Jespersen also mentions shifting of tense and person from the putative original question. Language specific issues

The perspective in Jespersen (1924) is generally that of Western Euro-pean languages. Of other language groups, Semitic is referred to below (§ 4), and Chadic, a remote branch of Afroasiatic, is discussed by Fra-jzyngier. Discussing direct versus indirect speech, Frajzyngier (1996:173–180) tells us that the complementizer in Chadic languages is not, in itself, indicative of direct or indirect speech. The differences between direct and indirect are not always coded, and when they are, it is by two differ-ent sets of 2nd person pronouns. Accordingly, the differentiation between direct and indirect questions is not very salient, and Frajzyngier (1996: 207–225) is rather interested in the way interrogative modality is en-coded in the complement clause (vis-à-vis main clause interrogative, that is, a direct question). This type of modality is discussed cautiously in Pal-mer (1986:30–33, 78–81).

56 Ancient Near Eastern Studies: Articles

Keevallik 2011 takes a different approach, analyzing various clauses traditionally considered to be interrogative complement clauses in Finnish, as actual questions, along discourse and content lines (such as, for instance, the answer may be found in the text), and the matrix expressions are re-garded as having developed into discourse markers of questions. Generative linguistics

In generative linguistics, indirect questions (“wh-questions” or “embed-ded questions”) are usually treated as complements. The main issues treated are: 1. the matrix expressions that occur with embedded ques-tions; 2. their differentiation from relative clauses; 3. their relationship with negative polarity items and free choice items; as well as 4. various constraints on them (Adger–Quer 2001). Functional linguistics

Givón 2001 I 156–157 refers to indirect questions as “embedded ques-tions,” which are complements of certain verbs (epistemic, of lower cer-tainty), and mentions two types, corresponding to the two types of main questions—nexus (yes/no-question in his terminology) and pronominal (wh-questions), which are further elaborated upon ibid. 2 309–310. Givón does not mention anything about deictic shift.

Frajzyngier (1995) discusses how complementizers can encode mean-ings: this can be exemplified by opposing an object clause with an em-bedded question with the same matrix verb (for instance, a ‘that’ clause with a ‘whether’ clause).

It seems, according to both Jespersen’s German examples (1924:296), that subordination is not a sine qua non feature of indirect questions (this idea becomes even stronger below). Note that most sources, to the exclu-sion of Frajzyngier, hardly deal with anything other than the modern Indo-European type of indirect questions (or, as they are termed in French, “la subordonneé interrogative”).4

It seems that the issue of deixis is hardly ever raised in dealing with “embedded questions.”

4 This (“The interrogative subordinate clause”) is the term in the title of Leo-

narduzzi 2000.

E. Cohen, Indirect Representation of Questions in OB Akkadian 57

Text linguistics

Fludernik 1993 is a comprehensive study of the phenomenon termed free indirect discourse (FID, see § 2 above). Focusing on the syntactic form of indirect and free indirect discourse in the context of (non)subor-dination, Fludernik also addresses questions (ibid. 147–151). In this sec-tion, she describes the gap between what is found in standard language and in oral narrative. In the latter, one finds all types of mixed phenom-ena belonging to both direct and indirect questions, such as inversion and non-subordination. Similar phenomena are shown to exist in French and in German.

Fludernik’s treatment of FID highlights an issue relevant to the topic of the present paper: FID is often illustrated by examples of free indirect questions.

Actually, the phenomenon was already presented and dealt with in Tobler 1894, who considers FID as a mix of both direct and indirect fea-tures. Tobler, like others, illustrates FID with many examples of free in-direct questions (ibid. 6–10). These examples are all at least partly indi-rect, but are not introduced by a matrix expression, which means that they are not subordinate. On the other hand, since they often reflect the inner thoughts of the characters in narrative, they are generally not fol-lowed by any kind of response. The following is an illustration of this phenomenon:

[9] L’horloge lumineuse de la vieille ville égrena dix coups. On appela le commissaire au téléphone.

C’était le maire. — Rien encore ?… Est-ce qu’il s’attendait, lui aussi, à un drame? Mais, au fait, Maigret ne s’y attendait-il pas lui-même?

(Georges Simenon, Le chien jaune, le Livre de Poche 2003:65).

The lighted clock in the Old Town tolled ten. The Superin-tendent was called to the telephone.

It was the mayor. “Still nothing?” It sounded as if he, too, was expecting trouble (lit. was it

that, he too, expected drama?) But, actually, wasn’t Maigret expecting trouble himself ?

(Georges Simenon, The Yellow Dog, Tr. Linda Asher, Penguin 2006:50–51).

[10] He peered around. Which way was his ship? He shielded his eyes against the late afternoon sun until he managed to make out its bent, tubular outline. […]

58 Ancient Near Eastern Studies: Articles

The ship was taken. Three or four immense Caucs stood lounging around it, cigarettes dangling from their slack mouths, white-faced and hairy. Stunned, Sung-wu moved back down the hillside, prickles of despair numbing him. The ship was lost; they had got there ahead of him. What was he supposed to do now? […]

He retraced his steps, mind blank. What could he do? He was helpless; his shivergun had been useless. He was alone, and there was no contact with the Arm. (Philip K. Dick, The Philip K. Dick Reader. Citadel Press, 1997:68).

Such examples are not found in any ancient language, and are consid-ered a relatively recent development, but they do suggest a change in proportions and that the framework for indirect questions is in fact much larger: what about indirect (deictically shifted) questions without matrix expressions, i. e., which are independent? What about direct questions that occur within a matrix expression? Some of these issues are taken be-low, § 5.2.2.

In order to create a working definition for indirect questions, I would like to link the terms “direct” and “indirect” with another domain where they are used: with object complements. A traditional view of the term “direct object” may be viewed as the following:

“Una distinzione basilare che viene tradizionalmente operata è quella tra complemento diretto e complementi indiretti, in riferi-mento all’assenza o presenza di preposizioni” (“a basic distinction traditionally made is the one between direct and indirect comple-ments, in reference to the absence or presence of a preposition”) (Beccaria 1996:155, my emphasis).

This traditional definition is chosen on purpose: nowadays, the “direct-ness” of the object has to do with a principle rather than with form, namely, with the direct effect of the verbal lexeme on this “direct” argu-ment. Admittedly, the traditional idea can only work in certain lan-guages, like Italian or French (where the direct object is unmediated), but not in others, like Spanish (where the animate direct object is pre-ceded by the preposition a) or Latin (where all cases may in principle oc-cur directly, unmediated by a preposition). Such physical “directness” may be borrowed as a scale for the directness of questions: if a question is mediated by a matrix expression, it is indirect; if not, it is direct. This leaves out the issue of deixis shift, which is basically undeveloped in ques-tions in Old Babylonian Akkadian (and in the other ancient Semitic lan-

E. Cohen, Indirect Representation of Questions in OB Akkadian 59

guages). The following definition of indirect questions is especially ap-pealing, since it is an analogue to this view of (in)directness:

“The interrogative is used also in indirect questions, i. e., to ask a question in an indirect way, as in ‘Tell me who did it,’ or to report a question indirectly, as in ‘He asked me who did it.’ ” (Curme 1931:212, my emphasis).

A working definition along Curme’s lines seems practical: asking a ques-tion in an indirect way, without insisting on deictic shift (which, inciden-tally, is often impossible to identify with certainty in many languages as well), allows us to include what could be termed “mediated questions,” that is, questions which are asked, expecting an answer, but asking is done via some question-introducing expression. A question may arise, how an em-bedded direct question (e. g., exx. [33]–[38], § 5.2.2.3) should be regarded: in these cases, it seems that although the direct question seems mediated by a matrix verb, the question can nevertheless be used just as effectively without it. In these cases, the “indirect” is not appropriate. 4. Indirect questions in Semitic

Ancient Semitic does not have a common means of expressing questions indirectly.

Classical Arabic does not differentiate between direct and indirect questions. Indirect questions have the same structure as direct questions, which simply follow the matrix expression: fa-naÏarat hal tarà !aµadan “indem sie ausschaute, ob sie einen sähe” (Reckendorf 1898:513; Fischer 1987:190). This is true for nexus questions and pronominal questions alike: lam !adri mā fa"ala !a-µayyun “ich wusste nicht, was er macht, ob (er) am leben ist …” (Reckendorf 1898:514).

Biblical Hebrew follows the same principles: wayy3šallaµ !ε¬ hay-yōnā me-!ittō lir!ō¬ hă-qallū ham-mayīm “Then he sent forth a dove from him, to see if the waters had subsided” (Gen. 8:8). These interrogative clauses appear to occur in the syntactic slot of the object. There seems to be no syntactic difference between direct and indirect speech:

“It should here be remarked that the distinction between direct and indirect questions cannot have been recognized by the Hebrew mind to the same extent as it is in Latin or English. In Hebrew there is no difference between the two kinds of sentence, either as regards mood (as in Latin) or in tense and position of the words (as in English)” (GKC § 150i, n. 2).

60 Ancient Near Eastern Studies: Articles

Moreover, in both Classical Arabic and Biblical Hebrew there is no spe-cialized exponent for indirect questions: interrogative pronouns as well as exponents of direct nexus questions (Heb. hă-, Cl. Arab. !a-, hal) are used in these questions.

Ancient Ethiopic employs, for indirect nexus questions, an exponent identical to the conditional particle, (la)!3mma(hu~-nu) (Dillmann 1907: 514–515): n3r!ay !3mma y3ma´´3!-nu !elyās “let us see whether Elias comes …” (Matt. 27:49). The “question” marker -nu~-hu is occasionally found in questions and in conditional protases, possibly signaling non-certainty.

Syriac (Nöldeke 1904:303–305) does not differentiate between direct and indirect pronominal questions, not even when the pronoun d- is found (which is quite frequent). However, nexus questions are marked differently: no overt mark in the case of direct (to the point that it is in-distinguishable in writing from a declarative clause, Nöldeke 1904:267), and en (otherwise the main exponent of conditional protases) in the case of direct question: neµzē en ā¬yā m"addrā lā− “we shall see if it comes and helps thee.” 5. Indirect questions in Akkadian

This section is devoted to the synchronic description of the indirect ques-tion. After a short literature review, the OB material relating in various ways to indirect questions is analyzed and classified. This material is mainly epistolary, taken from both central as well as peripheral Mesopo-tamia. In addition, when relevant, examples are drawn from the OB lit-erature as well. 5.1. Review of sources

Up to this point the literature on Akkadian has turned up very few ex-amples of indirect questions—GAG § 180a provides one instance in OB (our ex. [58]), which is not quite an indirect question (see discussion be-low). GAG § 180c has another (our ex. [34]), which is really an embedded direct question.

Deutscher (2007:137–148, chapter 9) is the only source devoted so far to indirect questions. It focuses on the historical aspect of indirect ques-tions in all periods of Babylonian. For this reason Deutscher concentrates on strategies that reappear in post-OB periods, and posits three strate-gies for OB:

E. Cohen, Indirect Representation of Questions in OB Akkadian 61

1. Concealed questions constitute the largest class: bal¢ūssu ul īde-ma ‘I did not know that/whether he lived …’ but lit. ‘I did not know his liv-ing’ (our ex. [66], Deutscher 2007:141, ex. 334); PN šāl-ma ašar i¶liqu liqbīkum-ma AbB 1, 122:25–27 (our ex. [60], Deutscher 2007:143, ex. 354);

2. Direct question: u eqlam mannum ana PN iddin bīr-ma ‘also, verify who gave the field to PN?…’ (our ex. [37], Deutscher 2007:140, ex. 332);

3. Indirect questions constitute the smallest, almost negligible class: Deut-scher provides only two examples with a postposed kīma clause (our ex. [61], Deutscher 2007:144, ex. 357).

Other than these, Deutcher provides a good example of the particle êma (ibid. 144, n. 56), which he deems to be a locative particle, like ašar.

Two examples of indirect questions with šumma are found in Cohen 2006:564 and in Cohen 2012:28, n. 45, which are taken up below (exx. [45] and [47]). This type is considered as central in the description below. 5.2. Strategies

The strategies used are diverse: questions introduced by the direct quote marker (§ 5.2.2.1); nexus questions preceding question-introducing ex-pressions (§ 5.2.2.2); pronominal questions preceding question-introduc-ing expressions (§ 5.2.2.3); indirect questions with šumma (§ 5.2.3.1), with modal particles (§ 5.2.3.2), with êma (§ 5.2.3.3), with kīma (§ 5.2.3.4), or alternatives and reduced interrogative clauses (§ 5.2.4). 5.2.1. Direct questions

Some general issues related to direct questions have already been raised above, in § 1. In this section, we will focus on points that are relevant to OB.

Direct questions are of two basic kinds in OB, and are marked in two ways: in case of a pronominal, or constituent question, interrogative pro-nouns or adverbs are used:

[11] awātam annītam ina qāti mannim tāmur ‘where (lit. in whose hand) have you seen such a thing?‘ (AbB 6, 208:9–10).

In some cases, for reasons enumerated in § 1, the interrogative particle is marked as the focus of the clause, either by a cleft construction (ex. [12]) or other means (ex. [13]):

[12] [išt]u* PN ana šīmtišu illiku mannum ša bītka ukallu ‘[Sinc]e PN died, who (is it) that takes care of your house?’ (ARM 1, 61:27–28).

62 Ancient Near Eastern Studies: Articles

[13] ina ayyitim-ma libbaka imra ´ ‘in what way (lit. in whichfs) have you been hurt?’ (AbB 14, 64:17–18).5

In fact, ex. [13] is doubly marked—once by the interrogative adjective and once by the lengthening used for nexus questions in OB: nexus questions, or polar questions, are often marked by a lengthening a vowel in one of the entities of the clause (marked by an underline):

[14] [ ¢]ēma[m g]am[r]am ana ´ēriya š[up]r[am] ¶arrānum ayyītum ana alāk ´ābim damqat ¶arrānum elîtu[m d]amqat ¶arrānum qablītum damqat ¶arrānum šaplītum damqat ištu ma¶rīka ana alāk ´ābim damqat ‘Se[nd me] a compl[ete re]port: which route is good for the army march? Is the high route [g]ood? Is the middle route good? Is the low route good? Is right in front of you good for the army march?…’ (ARM 1, 85+A.1195; MARI 5:163–167, l. 31–35; LAPO 17:19–23).

This is a model example in marking each and every nexus question graphically. Ex. [15] is similar:

[15] ālum GN1 GN2 GN3 u māt GN4 ša mannim ul kûm-ma ‘To whom belong the city of GN1, GN2, GN3, and the land of GN4? (Are they) not yours?’ (FM 7, p. 157–161, No. 47:54–56).

Questions in OB are compatible, to an extent, with other modality types. For instance, both nexus and pronominal questions are often used as deliberative questions, asking the addressee for advice:

[16] [umma] anāku-ma lušpur ana [ ´]ēr šarrim umma šunu-ma [š]upur ‘I (said): “should I write to the king?” They (said): “write!” ’ (Du-rand 1992:117–118, l. 41–43).

Note that the order is somewhat irregular (since the complements follow) in both exx. [16] and [17]:

[17] anna PN1 ša šum[eram ¶ī¢u] aṭarradakkum PN1 u […] PN2 šumera[m ¶ī¢] u [an]a têrtim šaki[n] līzib-ma têrta[šu] u ana ´ērika li[llik] ‘Certainly, (there is) PN1 who [reads Su]merian, shall I send you PN1? … PN2 [reads] Sumerian, but he is on an ad-ministrative post, should he leave his post and [go] to you?’ (Ziegler–Charpin 2007:69–71, l. 10–15).

This type is important for the rest of the discussion, because this type of question is very common in the OB letter genre. However, the precative form liprus in this deliberative function in nexus questions is hardly ever

5 ina ayyitim-ma could be interpreted as an indefinite pronoun: ‘have you been

hurt by anyone?’

E. Cohen, Indirect Representation of Questions in OB Akkadian 63

found in central OB (see Cohen 2005a:108–112),6 and even in peripheral OB it is not very common. Various alternative constructions are used in-stead. For one, the present, iparras is used instead of liprus (as in the first clause of ex. [17] above):

[18] ´ubātam mimma eleq[q]e ul atta-ma teleqqe ‘Shall I take some gar-ment? (Is it) not you (who is to) take it?’ (ARM 1, 46:15–17; LAPO 18:171).

Another construction is the modal ša parāsim in questions. This construc-tion, although not a finite form, should be considered as modal anywhere it occurs, whether as the predicate or as a complement:7

[19] umma atta-ma ammīni unnedukkaka lā illikam anāku ša unnedukkim šūbulim ‘Thus you (said): “Why hasn’t your let-ter arrived?” Am I the one to send the letter?’ (AbB 1, 23: 7–11).

The question in ex. [19] is a nexus question with a focused element, anāku.8 In pronominal or constituent questions, since the constituent at the center is the rheme of the question, another focus is hardly ever marked:

[20] aššum PN awīlum warad bīt [D]N ištu ina ālini wašbu ina sartim matīma šumš[u] ul ¶asis awīlum pānānum ul ¶asis inanna mannum i¶susakkuššu ‘As for PN, the man is a slave of the DN temple. Ever since he resides in the city, his name has never been mentioned in connection with any fraud. The man was not mentioned previously. So who mentioned him to you now?’ (AbB 14, 144:24–30).

Note that in ex. [20] too, there is an unmarked contrast between “previ-ously” and “now.” This contrast probably goes unmarked due to the ten-sion created by the natural function of the interrogative word as rheme. Focus marked in addition to the interrogative is quite rare in OB.

However, in nexus questions, any entity could be the rheme or the fo-cus. The following example has the same structure, but the infinitive con-struction is here the rheme:

6 One example of deliberative precative is: ana sūqim lu´i ‘should I go out to

the street?’ (AbB 14, 164:8). 7 The form is described, in all its syntactic positions, in Cohen 2005a:180–200. 8 See Huehnergard 1986:225–226, n. 34. Also, Cohen 2005b:265–266, dis-

cussing exx. 56–57 (our exx. [21] and [19] respectively).

64 Ancient Near Eastern Studies: Articles

[21] ana ar¶im šuāti ša kalîša ū lā ša kalîša ša rašê šī adi šinīšu ašpu-rakku-ma ‘As to this cow, is it to be kept or is it not to be kept? Is it to be acquired? I wrote to you twice …’ (AbB 9, 174:9–12).

This example is analyzed as an embedded direct question (namely, the question is the content which had been written and sent twice), but it is nevertheless a deliberative question, which may also occur independently of the lexeme √ špr.

These (direct) question types are important as a basis when setting out to describe indirect expressions of question. Unlike deliberative ques-tions, informational questions about the past are not very important in the current context:

[22] anāku ina bītika mīnam elqe ‘What did I take from your house?’ (ARM 10, 3:15′).

Such questions are at the heart of reported questions, which are quite rare in OB. 5.2.2. Embedded direct questions?

“Embedding” is often used as synonymous with “subordination”: “An embedded clause is a clause functioning as a constituent of another clause” (Cristofaro 2003:16, discussing the term as representative of the current usage). However, here we need some more syntactic finesse. Since subordination in OB has a clear form (it is always the equivalent of a genitive construction, where the genitive, or rather attribute slot of the construction is a clause, which is marked as such by subjunctive forms whenever possible), “embedding” will be used here as in Van Valin (1984): subordinate clauses are characterized as 1. dependent upon an-other clause (whereas no such dependence occurs in [symmetrical] coor-dination), and 2. as having a function as a part of another clause, that is, subordination involves embedding (ibid. 542). For Van Valin, “embed-ding” is merely a parameter of subordination, alongside dependency. This way it is possible to define subordination as characterized by the pa-rameters [+embedded +dependent], whereas chaining (Van Valin’s “co-subordination,” with the example of -ma chains in Akkadian) is character-ized as [–embedded +dependent], etc. In the same vein, it could be ar-gued that a direct quote, when taking the syntactic slot of an object com-

E. Cohen, Indirect Representation of Questions in OB Akkadian 65

plement, is embedded ([+embedded]), but not dependent ([–depend-ent])—since it can stand outside this matrix.9

Accordingly, we can start this survey by examining a range of phe-nomena that are relevant, in various ways, to indirect questions. 5.2.2.1. Question introduced by the direct quote marker In but a few cases, a question-introducing expression is followed by the direct quote marker:10

[23] PN [ašt]āl-ma umma anāku-ma ali ´ābum ‘[I ask]ed PN saying: where is the army?’ (ARM 26, 26:11–13).

[24] kīam išâlšu ummāmi ayyikêm t[allik] ‘He might ask him thus, saying: where [did] you [go]?’ (ARM 27, 163:5′).

Note that the direct quote marker and the direct speech in its scope al-ways follow the verb, often interconnected with it by -ma (as in ex. [23]). In this case it is nothing but the original direct question with a special syntactic adaptor which allows it to fit in any syntactic environment. The question itself is not embedded, nor is it part of the clause of the ques-tion-introducing expression, in this case šâlum ‘ask.’ The relationship is rather one of chaining, as the explicit marking with -ma in ex. [23] indi-cates (aštāl-ma umma x-ma). The order is not a good parameter, since there are examples (§ 5.2.3.4 below), which have post-verbal kīma clauses.

[25] inanna muškēnum … ul išāl ummāmi i[št]u erbum šêtini ikkalu nīnu annikīam wašābam nile!!i ‘Now, the dependant … does not ask as follows: “since the locust devours our crops, can we dwell here?” ’ (ARM 27, 27:26–29).

This strategy is quite rare. 5.2.2.2. Nexus questions preceding question-introducing expressions In this strategy, which is not very common, a direct question precedes the verb and its object, and the latter is a nominal capable of reflecting a content:

9 “The fourth combination … defines a potential relation in which a clause

functions as part of another clause but is fully independent of it. … Another pos-sible candidate is direct discourse complements, since they are part of but in all respects independent of the main clause” (Van Valin 1984:547, my emphasis).

10 For a discussion of the syntax of the direct quote marker, see Cohen 2012: 95–96.

66 Ancient Near Eastern Studies: Articles

[26] iddinakkim annītam lā annītam šuprīm ‘… did he give (them) to you? Write me either this or that’ (AbB 7, 13:23–25).

[27] inanna awīlû šunu ittiqū annītam lā annītam bēlī lišpu[ram] ‘Now, can these men pass? Let my lord write [me] either this or that’ (ARM 6, 22:19–21, LAPO 16:621 ‘doivent-ils continuer leur chemain?’).

The pronominal expression annītam lā annītam ‘either this or that’ is cen-tral to the notion of indirect expression of a question (for instance, vis-à-vis conditionals, where it does not occur): it specifically asks the addressee to decide about which mode of action he prefers. This expression serves as a phoric anchor inside the matrix clause for the question. It may be found with other expressions as well:

[28] u i´´ū ša innaksū ma´´arū qišātim ikkisū ina qātim a¶ītim innaksū warkatam purus-ma ‘ “And the trees that were cut, did the forest keepers cut them? Were they cut by another party?” Investigate the circumstances …’ (AbB 4, 20:21–26, quoted in Deutcher 2000:139, ex. 327).

In ex. [28] the question-introducing expression is warkatam parāsum ‘in-vestigate the matter.’ warkatum ‘matter, case’ is the phoric anchor. In the following example the anchor is ¢ēmum:

[29] [kaspam ša ri]bbāt būlim [ša ter]dītim wak[lū amu]rrim inaddinū waqlū ¶a¢¢im la[pp]utū u rēdû inaddinū ¢ēmam bēlī lišpuram ‘Should the ge[ner]als pay [the silver of the ad]ditional [ar-r]ears in cattle? As to the captains, ser[gea]nts and soldiers, should they pay? Let my lord send me a decision’ (AbB 1, 1: 5–10).

All these cases (exx. [26]–[29]), have in common what looks like a direct nexus question which precedes a potentially question-introducing ex-pression. However, the question does not take the actual place of the ob-ject. That place is formally taken by a nominal in the expression, either the pronominal syntagm annītam lā annītam (‘this or that’), or content-holding substantives such as warkatum (‘matter’) and ¢ēmum (‘decision’). The relationship between the question itself and these nominals (which we have termed “anchors”) is notional: first of all, all these expressions stand for the answer, rather than the question.11 It may be perceived as

11 Compare the following example, where the demonstrative represents the

question itself: ‘You can even ask just that: “What are the next steps?” ’ (http://blog. arrowstaffing.com/2011/09/is-it-ok-to-ask-if-you-got-the-job/) (April 2013). Here the demonstrative is directly related syntactically to the direct quote. The following is

E. Cohen, Indirect Representation of Questions in OB Akkadian 67

some kind of an apposition, or rather parallelism corresponding to the one that exists between the interrogative element and the answer. This strategy could, at best, be regarded as an early precursor of indirect ques-tions. The fact that they are quite rare can partially be explained by the availability of šumma clauses (see below, § 5.2.3.1). 5.2.2.3. Pronominal questions preceding question-introducing expressions We have more examples for this category, pronominal questions, than for the previous category; the question-introducing expression may in-clude, in addition to the group of expressions presented in the previous section, a less limited set of expressions as well.

[30] kussûm [šī] ayyānum iššakka[n] annītam lā annītam bēlī lišp[ur]am-ma ašar kussûm iššakkanu [lišp]ur ‘[This] throne, where should it be set? Let my lord write me this or that, [let him wri]te where the throne is to be set (lit. the place the throne is set)’ (FM 8: 55–57, No. 11:41–45).

Note that, if the reconstruction is good, we have some kind of an equiva-lent: the first part has a direct pronominal question followed by the an-chor expression annītam lā annītam, which stands for the answer. So does the complement in the second part, which is headed by ašar and again, represents the answer. ašrum stands for the yet unknown place, but unlike ayyānum “where? …,” or even êma “where …” it does not even hint at a question (see below, § 5.2.3.2). The ašar syntagm is part of what Deutscher (2007:140–144) calls “concealed questions,” the largest group he discusses.

[31] še!am kī ma´i tušaddin [k]ī ma´i u¶¶ur ¢emam gamram me¶er ¢uppiya šūbilam-ma ‘Send me as an answer to my tablet, a com-plete report, (viz.,) how much barley did you collect? [Ho]w much is left? …’ (AbB 14, 87:22–25).

Here too, the direct question has some link with the anchor ¢ēmam ‘re-port,’ which stands for the answer and functions as formal object. An-other example is the possible prototype of these constructions:

similar except that it is indirect: ‘I received a call from Camilla who wanted to know just that: did I, in fact, phone the police?’ (http://denversyntax.com/issue24/fiction/ keating/contest.html) (April 2013). The last is in French: ‘C’est à toi qu’il faut de-mander cette question: “Pourquoi ils sont si nombreux?” ’ (http://fr.answers.yahoo. com/question/index?qid= 20080827123056AALmSN3) (May 2013).

68 Ancient Near Eastern Studies: Articles

[32] aššum PN1 PN2 ša eqlam ibquru šu kī qāssu ina mīnim ilikšu [mā]ru mannim šū eqlam šâti kī ibquršu [x] PN3 šū [x] PN4 šū magana ¢ēmšu šupram ‘Concerning PN1, (is it) PN2 who claimed the field from him?12 How is his ‘‘hand’’? In what does his service consist? Whose [so]n is he? How did he claim that field from him? Is he [x] of PN3? Is he [x] of PN4? Now then, send me a report about him’ (AbB 9, 199:5–16, cited in Deutscher 2007:140, ex. 331).

Because of the large number of questions (seven in all) and consequently their physical distance from the anchor ( ¢ēmum), this link seems weaker.

In the foregoing examples, the anchor is the formal direct object. The rest of the examples in this section are of a different structure: there is no anchor to fill the slot of the direct complement, and the question imme-diately precedes the verbal form and take this slot:

[33] ana māti[m] ayyitim irrubū [n]aplisā u [ ¢]ēmam terrā[nim] ‘[F]ind out: which land do they (intend to) penetrate? And return [me] a report’ (FM 2:68–69, No. 34:14–15).

[34] ina PN1 u PN2 ayyum abūšu u eqlētušunu ayyum eqel bīt abīšu … birram ‘Now then … verify: between PN1 and PN2, who is his father? And as to their fields—which is his family field?’ (AbB 4, 118:27–31, cited in Deutscher 2007:140, ex. 330).

[35] adi inanna eqlum šipram [k]ī ma´i īpuš13 ina ¢uppika pānam šuršiam-ma [š]upram ‘Up until now, as to the field, [ho]w much work has he accomplished? [W]rite to me clearly in your let-ter’ (AbB 10, 182:7–10, cited in Deutscher 2007:139, ex. 329).

Note that the same question-introducing expression may occur without an anchor (the previous example) and with an anchor (the following ex-ample):

[36] kiriam šuāti ana PN mannum iddin ¢ēm kirêm šuāti gamram pānam šuršiam-ma šupram ‘Who gave this orchard to PN? Write to me clearly your full report of this orchard’ (AbB 4, 13:12–17).

The following example contains a pair of syntactically analogous struc-tures: two types of complements for the lexeme burrum ‘verify, establish, convict’: one of which is a noun syntagm and the other is a direct question:

12 This clause is interpreted as a question with a cleft structure (#x ša+V#,

compare with ex. [4]), in order to account for the lengthening in ibqurušu. 13 See ibid., note a to the translation.

E. Cohen, Indirect Representation of Questions in OB Akkadian 69

[37] awātam šuātu bīrši14 u eqlam mannum ana PN iddin bīr-ma ‘Ver-ify this matter and also verify: who gave the field to PN? …’ (AbB 11, 189:23–26, cited in Deutscher 2007:140, ex. 332).

The two verbal clauses (bīr) are connected by u, which also marks the be-ginning of the second clause. Where in the first clause we find the object complement awātam šuātu ‘this matter,’ we find, within the second clause, between the beginning of the clause and the verbal form at the end a di-

rect question. Analysis of these twin structures reveals that the question clearly takes the slot of the object, in Van Valin’s terms, it is embedded, but not dependent: in other words, it is not marked as subordinate, but it does have a function as part of the matrix clause (the issue of subordina-tion is further discussed below, § 5.2.3.3). Yet another example of a simi-lar structure exhibits a clefted question:

[38] mannu ša še!am ana ukulle bītika ni¶abbatam šupram-ma ‘Write to me: who (is it from) whom we can borrow barley for provi-sions to your household? …’ (AbB 14, 164:8–10).

The claim here, then, is that the direct question is in paradigm with an object such as awātum (ex. [37]), and accordingly has the same function: it functions as the object complement. 5.2.2.4. Another kind of question Only in the Mari letters do we find a clause-final kī, which seems to mark nexus questions:

[39] inanna [awīlû] ¶[an]û ina māt GN innassa¶ū kī annītam lā annītam bēlī lišpuram ‘Now, are the Ú[an]a to be evacuated from the land of GN? How about it? Write me this or that’ (ARM 26, 358:13′–15′).

[40] [k]appān imittam u šumēlam [d]amda m idūkā ina abunnatim idūkū kī annītam lā annītam ¢ēmam gamram šupram ‘Did the [w]ings on the right and on the left bring [v]ictory? Was it brought by the cen-ter? How about it? Write me this or that, a full report’ (ARM 4, 85+A.3090; FM 3:145–152, No. 14:13–17; LAPO 17:308–309).

As far as I can tell, this particular function of kī is not commented upon in the literature; yet it is different from the usual function as an inter-rogative manner adverbial, which is clearly not the case here.

14 An even better analogy exists with the following example, where no re-

sumption occurs with the finite verbal form: attunu ālum u šībūtum awâtim ša eqlim šuāti ma¶ar ilim birra-ma ‘You, city and elders, verify the matters of this field in front of the god …’ (AbB 4, 40:33–35).

70 Ancient Near Eastern Studies: Articles

5.2.3. Indirect questions?

There are three candidates for the indirect expression of a question in OB: the particle šumma (for nexus questions) and the conjunctions êma and kīma (for pronominal questions). The following sections examine the evidence and analyze it, so as to reach a conclusion whether there is a category of indirect questions in OB. 5.2.3.1. šumma clauses The most promising exponent is šumma, which is known primarily from the expression of various conditional protases, but also from the rare ex-pression of oath as well as its disjunctive function (“or”). In all these func-tions, šumma generally introduces clauses which do not reflect facts (not even in oaths, see Cohen 2005a:56–59), which makes it a natural candi-date to introduce another state of affairs whose certainty is low. In this section the examples for this strategy are discussed and the rationale is explained.

Biclausal constructions with šumma are first and foremost conditional. In some cases, however, the construction is not conditional, but rather what seems like an indirect expression of a question, as is found in G3"3z (la!3mma) and in Syriac (en). In these cases, šumma clauses occur with the previously discussed anchor expression annītam lā annītam:

[41] inanna šumma … ama¶¶ar-ma awīlam šâtu uwaššar annītam lā annītam bēlī lišpuram ‘Now, let my lord write me this or that, whether I should accept … and release this man’ (ARM 14, 17:45–49; LAPO 17:640–642).

[42] u šumma allakam annītam lā a[nnītam] bēlī lišpura[m adi me¶er ¢uppiya] ša alākiya ū lā alāk[iya] iturram ´ābam a[kla] ‘Also, let my lord write me this or [that], whether I should come. [Until an answer to my letter] to come or not to come returns, I [detain] the people’ (ARM 14, 71:11′–14′; LAPO 17:661–662).

Exx. [41] and [42] are rather convincing cases of indirectly asked ques-tions. First, although they look like the most common conditional pat-tern, they are different semantically: for instance, it is clearly not the case that the addressee should write this or that only in case the writer comes; rather, his instructions should be sent unconditionally (unlike any apo-dosis of an ordinary condition). In fact, the anchor expression annītam lā annītam is usually not found in conditional apodoses. Another important point is the sequence of events: in both of these examples, writing is to take place before the actions found inside the šumma clause. This clearly

E. Cohen, Indirect Representation of Questions in OB Akkadian 71

contradicts the spirit of ordinary conditional structures, where generally the protasis temporally precedes the apodosis.

These, as well as other cases of šumma clauses, all exhibit the same ba-sic structure: they look like conditional patterns but have the annītam lā annītam anchor in the apodosis. Some are translated as direct questions in the different editions, for instance, for example [42] we find in LAPO 17:661 “Dois-je tout de même partir?” This translation choice is not ex-plained anywhere. šumma basically does not, as far as I know, introduce direct questions.15

Another group is similar, but the šumma clause does not have any an-chor:

[43] šumma bā!erū ša ušēbirū[š]unū[t]i nišpīšunūti-ma ‘We asked them whether (it was) fishermen who helped them cross …’ (AbB 3, 53:14–15, mentioned as such in SLB 4:175–176).

This example is ideal: it contains a report of a question, which is very rare, plus a deictic shift from the putative original question (*‘were these fishermen who helped you cross?’). In addition, the pattern is clearly not a conditional, for the simple reason that iprus forms do not form part of the apodosis (see Cohen 2012:61–64 for the possibilities). However, the ex-ample is unique in that it reports a question indirectly. Otherwise, re-porting a question is basically done directly:

[44] [ina q]āti PN1 [kaspam] āmur-ma [umm]a anāku-ma [m]annum iddinakki umma šī-ma PN2 ‘I saw [the silver] in PN1’s possession and I (said): “[W]ho gave (it to) you?” She (said): “PN2” ’ (AbB 1, 133:15–19).

There are a few more such examples:

[45] šumma ina pilkātim [š]a GN [eq]lum šū [an]a PN ša¢er [b]irrā-ma ‘[V]erify whether in the sectors of GN this [fie]ld is registered [to] PN …’ (AbB 4, 99:8–12).

15 Nevertheless, there is a strong tendency to render these šumma clauses by a

direct question: 1. ARM 2, 87:32–35; FM 2:225, No. 121b:32–35; LAPO 16:298–299, No. 163

(the form is it-ta-´í, and nevertheless is regarded as deliberative); 2. ARM 14, 12: 11′–13′; LAPO 17:633; 3. ARM 14, 48:51–53; 4. ARM 27, 158:4′–6′; 5. ex. [48]; 6. ex. [69] and 7. ex. [70].

Rendered by an indirect question: 1. ARM 26, 294:16–18 (negative!); 2. FM 2:285–286, No. 130:24–27 (reconstructed); 3. ex. [33] (even though the question is directly posed!).

72 Ancient Near Eastern Studies: Articles

[46] šumma š[al]im ilam šāl-ma alik ‘Consult the god whether it [sa]fe and leave’ (AbB 12, 38:21–22).

[47] šumma adi u¢¢ētika akammisu wašbāku šupram-ma ‘Write to me whether I (should) stay until I collect your wheat …’ (AbB 14, 141:25–27).

In ex. [47], there are temporal differences between the šumma clause and the question introducing expression: the former (staying) occurs later than the latter (writing), which is irregular for an ordinary conditional.

[48] šumma nēšum šū adi alāk <bēliya> ina rugbim-ma uššab bēlī lišpuram ū šumma ana ´ēr bēliya ušarrâššu bēlī lišpuram ‘Let my lord write me whether this lion is to dwell in the loft (of all places) until <my lord’s> return, alternatively, let my lord write me whether I should direct it to my lord’ (ARM 2, 106:8–13; LAPO 16: 345–347 has “Ce lion doit-il rester dans la réserve elle-même jusqu’à l’arrivée de <mon Seigneur>? … Ou bien dois-je le faire conduire chez mon Seigneur?”).

Here too, like the examples above, the temporal sequence is opposite to the one expected in ordinary conditions. Note, that here as well, bēlī lišpuram is an unconditioned request.

These šumma clauses can occasionally be compared with other types of complements:

[49] kīma PN abi sukkukim eqlam šuāti lā i´batu eqlum šū ´ibitni-ma ubirrūnim-ma ‘They established that PN, the deaf man’s fa-ther, did not seize the field (and that) this field is our prop-erty …’ (AbB 4, 40:12–14).

Note that here, with the preterite of the lexeme burrum (ubirrū), the kīma clause reflects a factual statement, rather than an expression of non-certainty, as we find in the examples deemed as indirect questions.

The cases presented in exx. [43] and [45]–[48] are the heart of our in-vestigation: these are, in all probability, legitimate examples of embedded, indirect questions. It is true that by the standards of OB they are not sub-ordinate, but they are dependent in other ways: šumma clauses, except perhaps when expressing oaths and perhaps disjunction (‘or’), are never independent, that is, they do not occur on their own. In addition, these šumma clauses (unlike conditional šumma protases), are reducible to syntac-tically simpler syntagms (see § 5.2.4 below). It is curious that this group has not developed any further (at least based on Deutscher’s account of this phenomenon in Middle Babylonian, Deutscher 2007:145–146). Deutscher, in fact, does not mention these cases at all, perhaps because they are not found in later periods. It seems safe to claim that these šumma clauses,

E. Cohen, Indirect Representation of Questions in OB Akkadian 73

which occur without an anchor, take the slot of the object complement, and are thus the object complement of the question introducing verbs.

Except for ex. [43], which reports a question, the other examples ([41]–[42], [45]–[48]) are all introduced by precative and imperative forms. This is important because these forms generally refer to the cur-rent situation, the here and now of the utterance, which means that the question is real in that it is part of an interrogative speech act, expecting an answer. It is just posed indirectly. 5.2.3.2. Modal particles A conjectural, marginal case is possibly the case of pīqat ‘maybe.’ The former is a modal particle, whose path crosses šumma in several points. First, there is a conditional pattern with pīqat (for which see Cohen 2012: 93–97); both particles express non-certainty; both are capable of express-ing disjunction.16 It is possible that it can also participate in the expres-sion of an indirect question:

[50] lu¢¢ul pīqa[t] eleppum imaqqutam-ma še!am anniam ušarkab [ele]ppum ul imqutam ‘ “Let me see, perha[ps] a ship comes my way and I can load this barley.” (But) a [sh]ip did not come my way’ (AbB 6, 125:19–22).

This example is unique, and the order is opposite to what is expected if the indirect part were part of the matrix clause. The following example seems to conform with the opposite order of elements:

[51] pīqat ´āb[u]m dšūe ana GN1 ūl[ū-ma] ana GN2 i[ll]i ul īde ‘I do not know whether the army [goe]s to GN1 o[r] to GN2’

17 (FM 6:252–255, No. 25:22–25).

The same may apply when the expression mannum lū īde ‘who knows’ oc-curs instead of ul īde:

[52] u pīqat ina kī[nā]tim-ma idabbub mannum lū īde ‘Who knows whether he speaks frankly’ (ARM 26, 489:43–45).

These cases then may be analyzed 1. as some kind of semantic equiva-lence, where the idea of not knowing is expressed once with ul īde ‘I do

16 See, for instance [pīq]at ana GN1 pīqat ana GN2 pānūšu šaknū ‘he is headed

[eith]er to GN1 or to GN2’ (ARM 2, 135 17–19 = ARM 26, 354 17–19). 17 The translation in the edition interprets this occurrence as a regular occur-

rence of the modal particle: “Peut-être cette armée montera-t-elle à Karana ou bien à Andarig? Je ne le sais pas,” which reminds one of the way šumma clauses are translated (see n. 15 above).

74 Ancient Near Eastern Studies: Articles

not know’ and once with pīqat ‘maybe’; 2. as an expression of non-certainty which takes the slot of the object, like šumma clauses. However, it is impossible, with the few examples at our disposal, to decide how close this structure is to the one described in the preceding section. 5.2.3.3. êma Deutscher (2007:140) wonders why direct pronominal questions (as the ones discussed § 5.2.2.3 above, e. g., ex. [37]), have not developed into indirect questions. He suggests that the difficulty was the distinction be-tween subjunctive and indicative: namely, without the subjunctive these questions cannot be regarded as indirect. This view does not hold: (in)di-rectness does not have to do with formal subordinate marking, as is plain from both Bibilical Hebrew kī and Syriac d-, both are subordinating par-ticles that can precede direct discourse (see Goldenberg 1991). Secondly, for interrogatives to be able to occur with the subjunctive, one more step was needed: the interrogative had to have a construct state form, mark-ing it as the nucleus of the construction. Only in these syntactic condi-tions does Akkadian allow the subjunctive (which is just a formal subor-dination marker). Two interrogative particles, to the best of my knowl-edge, have construct state forms: êma ‘where(ever)’ and kīma ‘how’ (for which see the following section). The former derives from *ayy- ‘where’ (AHw. 23b) and is found in various interrogative pronouns (ayy-ānum ‘where,’ ayy-iš~êš ‘where(to),’ ayy-išam~êšam ‘whereto,’ AHw. 24b, 25b) and in indefinite pronouns which are derived from them (basically, with the addition of -ma to the interrogative pronoun: ayy-išam-ma ‘wherever to, irgendwohin’). êma seems to continue this line of expressing indefi-niteness. However, unlike all other interrogatives~indefinites, it devel-oped a construct state form: *!ayy > ê (with the habitual ending -ma). êma is primarily an adverbial conjunction which heads an indefinite

locative clause:

[53] êma bēlī išapparanni allak-ma ‘wherever my lord sends me, I go …’ (AbB 1, 128:7–8).

[54] êm tamma rušu ´abassu ‘wherever you find him, seize him’ (AbB 2, 143:7′–8′).

Note that, in these cases, the clause specifically takes the slot of an adver-

bial, rather than object complement. Deutscher (2007:144, n. 56) mentions one case of êma (ex. 56 below)

but does not analyze it as an indirect question. He rather equates it with ašar, regarding it as a “locative particle.” However, êma seems to be able

E. Cohen, Indirect Representation of Questions in OB Akkadian 75

to provide us with another type of indirect question, of pronominal ques-tions:

[55] ī¢ū-ma ūmu êmi allaku ul īde ‘the day grew dark and I did not know where I was going’ (Gilg. Schøyen 2:39, êmi is a variant of êma) (LIT).

[56] êma ¢uppū šunu šaknū ul nīde ‘We do not know where these tablets are placed’ (VAB 5, 281:31–32, mentioned in Deut-scher 2007:144, n. 56) (LEG).

These cases are almost self-explanatory. The following case is less so:

[57] [š]umma i[n]a bītim ib[aš]ši kupram liddinū šumma êma ibaššû līmurū-ma maturram lip¶[û] ‘[I]f the[r]e is (some) i[n] the house, let them provide asphalt. Alternatively (= šumma), let them find out where it exists and seal a small boat’ (AbB 3, 40:14–16).

Note that the second occurrence of šumma in ex. [57] is not conditional, or it would be difficult to explain why there is no (main) clause within its scope. To verify these cases, two sets of comparisons need to take place. One, against the adverbial êma clauses, and the other, against other com-plement clauses of ul īde.

When comparing the adverbial êma clauses (exx. [53]–[54] above) with the ones claimed to represent indirect questions (exx. [55]–[57]), it is plain that the latter clauses are different syntactically: whereas the former are indefinite locative adverbial clauses, the latter take the slot of the ob-

ject complement, namely, what is not known, or what is to be found out. This is important: embedded questions basically take the slot of core ar-guments, and accordingly constitute substantival clauses. The indefinite component is very much alive in the first, adverbial group, but weaker in the second.

The second comparison is against other complement clauses, with the same matrix expressions:

[58] [pā]nūšu [aš]ar šaknū [u]l īde ‘I don’[t] know [wh]ere (lit. the place) he is [he]aded’ (ARM 5, 33:11–13).

[59] u ašar pānūšu šaknu ul iššeme ‘and it is not known (lit. it was not heard) where (lit. the place) he is headed’ (ARM 26, 430:31).

[60] PN šāl-ma ašar i¶liqū liqbīkum-ma ‘Frage den PN und er soll dir den Ort wo er/sie/es verschwunden ist sagen …’ (AbB 1, 122: 25–28).

In these three cases, the content of the matrix verb (know, hear and say) is built around the construct form of the substantive ašrum ‘place.’ This is perfectly reflected in Kraus’s translation of ex. [60], explicitly using Ort ‘place.’ êma, on the other hand, relates to the interrogative *ayy, still found

76 Ancient Near Eastern Studies: Articles

in interrogatives and indefinite pronouns. For this reason, whereas ašar clauses are considered “concealed questions,” in substantival êma-clauses the question is not concealed, just converted to an object clause. 5.2.3.4. kīma Another group of examples, which possibly expresses an indirect question is the occurrence of kīma clauses with the verbal lexeme šâlum ‘ask.’ This type is mentioned as basically the only kind of indirect questions in OB (Deutscher 2007:144). Two examples are cited, one is re-adduced here:

[61] [m]ārīka u GN kalašu [š]āl kīma ana ¢âtim [u] ´ibit ¢âtim akī ⅓ mana kaspum watrum gamru ‘[A]sk your [s]ons and the entire GN how for bribe [and] bribe-taking some ⅓ mana silver extra cost is paid’ (AbB 6, 73:18–24; Deutscher 2007:144, ex. (357); also AbB 13, 134:5–6, cited in Deutscher 2007:144 as ex. (356), which is devoid of context).

The particle kīma has many functions, but only two likely etymological sources: the interrogative particle kī and the adverbial kīam. The particle kīma attested in these examples is plausibly the construct state form of the interrogative particle kī with an appended -ma, the mirror image of êma, which is also the construct state of an interrogative element with an ap-pended -ma.18 The particle kīma in all other conjunctional functions (indi-cating object, comparative and temporal clauses) is rather the construct state of kīam ‘thus.’ Two more examples are found in Mari:

[62] bēlī rē!âm u mārī PN lišāl kīma immerī šina[t]i ilqē-ma bēlī dlāe ukallimu ‘Let my lord ask the shepherd and PN1’s sons how he (= PN2) took these sheep but did not show my lord’ (‘que mon seigneur interroge le berger et les fils de PN1, comme quoi il (= PN2) a pris ces moutons sans les donner à mon seigneur’) (MARI 8, pp. 347–351:38–40).

[63] u anumma bēlī PN warassu lišāl kīma eqlam šêtu [šā]pi¢um ālik pāniya ´abit ‘Now, let my lord ask PN his servant how the dis-trict [gov]ernor, my predecessor, holds this field’ (‘Voilà qu’il faut que mon Seigneur interroge son serviteur Aqba-Ahum, comme quoi le gouverneur qui me précédait détenait ce champ’) (ARM 14, 81:40–42; LAPO 17:537–541).

These examples are good, but they are few and are apparently specific to the lexeme šâlum ‘ask’ in the precative (see another strategy, with a direct quote, § 5.2.2.1 above, e. g., ex. [24]). Note, additionally, that in the ex-

18 For a suggestion for the original function of this -ma see Cohen 2000.

E. Cohen, Indirect Representation of Questions in OB Akkadian 77

amples from Mari, the subjunctive occurs only some of the time. They can be confronted with ex. [46], re-adduced here:

[64] šumma š[al]im ilam šāl-ma alik ‘Consult the god whether it [sa]fe and leave’ (AbB 12, 38:21–22).

The most salient difference between kīma and šumma examples of indirect question, except for the irregular order in exx. [61]–[63], is clearly the difference between a pronominal question (‘how’) and a nexus question (‘whether’).

A structurally similar example is found in Tell ar-Rīmaµ as a comple-ment of ul īde:

[65] kīma ´e¶ertum šī ina ekallim bašêt ul īde ‘I do not know whether that servant girl is in the palace’ (OBTR 142:7–9).

The context clearly indicates not-knowing, very similar to the context in the following example:

[66] PN mārī ištu 8 šanātim i¶liqanni-ma bal¢ūssu ul īdē-ma kīma mītim kispam aktassipšum ‘My son PN disappeared on me 8 years ago, and (since) I did not know whether he was still alive (lit. his being alive), I have made funerary offerings as if he were dead’ (AbB 13, 21:5–9).

However, with this matrix verb, kīma generally expresses factual, indica-

tive content:

[67] [k]īma wašbu ul īde ‘I did not know [th]at he was (there)’ (FM 9:138–140, No. 28:8–9).

[68] kīma nazqāta u i(na) GN atta adi inanna ul īde ‘I did not know till now that you were angry and (that) you are in GN’ (AbB 10, 2:8–11).

That is, an objective kīma clause with ul īde generally means ‘not know that’ rather than ‘whether.’ The fact that the different conjunctions be-have differently with different matrix verbs means that the language sys-tem is still in the process of sorting out a strategy for the indirect expres-sion of a question, or else kīma would have had one function after a cer-tain kind of predicates to express non-certainty. 5.2.4. Alternatives and reduced clauses

This section is devoted to related alternative constructions, which are oc-casionally a reduced expression of indirect questions. The following ex-ample features an alternative, packaged as twin indirect questions:

eran
Typewritten Text

78 Ancient Near Eastern Studies: Articles

[69] [u] šumma i´´abbat-ma [ana] nepārim ušerrebšu [bēlī] lišpuram ū šumma i´´abbat-ma ana ´ēr bēliya ušarrâššu annītam lā annītam bēlī lišpuram ‘(when a Benjaminite is caught), let my [lord] write to me whether, having been caught, I should put him [in] prison, and let my lord write to me this or that, whether, having been caught, I should direct him to my lord’ (ARM 2, 102:22–27; LAPO 17:423–424).

It is clear that these two structures do not contradict in their second part (bēlī lišpuram), which is to say that they are not conditional (where one expects, when having different protases, to have different apodoses ac-cordingly). The main point of this double construction is in asking for advice about two contradicting issues—incarcerating a Benjaminite or sending him to the addressee. This example exhibits two options, each in the form of an indirect question. In the following example, the structure is simplified:

[70] šumma ittiqū ūlū-ma ikkallû [ann]ī[t]am lā annītam ar¶iš bēlī lišpuram ‘Let my lord write to me quickly [this] or that, whether they should move onward or rather be detained’ (ARM 6, 19:10–11; LAPO 16:623–624).19

In this case, the two options are found within the same indirect question. A reduced expression of an embedded indirect question may be found

with other expressions of an alternative between two courses of action, when the latter is expressed by the infinitive construction epēšam ū lā epēšam:

[71] inanna bēlī etēqšunu ū lā e[tēqšunu] annītam lā annītam bēlī l[išpuram] ‘Now, let my lord [write to me] this or that, whether they should move onward or not (lit. their moving onward or not [their moving onward]’ (ARM 14, 97:12–14).

This is the only example of expressing an alternative with an infinitive in the accusative, namely, the alternative is a formal object which relates directly to the verbal lexeme. The rest of the examples of this kind con-sist of an infinitive construction expressing an alternative, but which functions as an adnominal expansion. The noun is the anchor found with those (indirect) questions discussed above.

19 In LAPO 16:623 we find “Doivent-ils continuer leur route ou être arrêtés?

Mon Seigneur doit m’écrire rapidement ce qu’il en est,” that is, as a directly posed question, analogous to ex. [27] above (see n. 15, for more cases).

eran
Typewritten Text

E. Cohen, Indirect Representation of Questions in OB Akkadian 79

ša epēšim ū lā epēšim

The construction ša parāsim is a modal infinitive construction anywhere it occurs (Cohen 2005a:180–200). As such, it naturally expresses various shades of obligation in both independent and subordinate environments. In this specific case, in order to express doubt, which is a component of the deliberative question, disjunction is used, viz., “to do or not to do”:

[72] pu[r]u[ssâm ša] ´āb bēliya ša wašābišu ū lā w[ašābišu] itti bēliya ul [elq]iam ‘I did not [rece]ive from my lord the de[cision con-cerning] the army of my lord, whether it (is) to stay put or not t[o stay put]’ (ARM 26, 390:5″–7″).

In this example, like the following ones, there is an adnominal expan-sion, where the nominal is analogous to the anchor discussed above: purussûm ‘the decision’ is expanded by ša wašābišu ū lā wašābišu ‘his staying put or not.’ We know about the doubt because on the same letter we find, a few lines below, another alternative, this time expressed as two disjunc-tive direct questions:

[73] … ´āb bēliya ina libbi ālim u´´êm-ma ittallakam ūlūma ina li[bbi] ālim-ma ikkalla bēlī puru[ssâm ša] ´ābišu annītam lā annītam li[špuram] ‘… should the army of my lord exit the city and go away or should it be retained right in[sid]e the city? Let my lord [write me] his deci[sion concerning] his army, either this or that’ (ARM 26, 390:14″–17″).

This example consists of directly posed alternative questions with an an-chor (purussâm), and it is thus analogous to the strategy of direct question plus an anchor (§ 5.2.2.2). Note that the contrast between ‘exit and go away’ and ‘be retained’ is analogous to the one expressed in ex. [72], ‘stay put or not.’

The reduction in these cases is always to a deliberative question: [74] u šumma allakam annītam lā a[nnītam] bēlī lišpura[m adi me¶er

¢uppiya] ša alākiya ū lā alāk[iya] iturram ´ābam a[kla] ‘Also, let my lord write me this or [that], whether I should come. [Until an answer to my letter] to come or not to come returns, I [de-tained] the people’ (ARM 14, 71:11′–14′; LAPO 17:661–662) (= ex. [42]).

Ex. [74] is another key for understanding the function of this syntagm, since it contains in the first part an explicit indirect question with ap-proximately the same contents as are found in the alternative-delibera-tive infinitive construction—whether the writer should come or not. There are two more examples of this kind:

eran
Typewritten Text
eran
Typewritten Text
eran
Typewritten Text
eran
Typewritten Text
eran
Typewritten Text
eran
Typewritten Text

80 Ancient Near Eastern Studies: Articles

[75] [b]ēlī ¢ēm awīlê [š]unūt[i] ša waššurišunu ū lā waššurišunu lišpuram ‘Let my lord write me an order concerning [t]hese people (namely) whether (they are) to be released or not to be re-leased’ (ARM 27, 10:10–13).

[76] ¢ēm še!im šâti ša turri ū lā turrim ammī[nim l]ā tašpurīm ‘Wh[y] did you [n]ot send me instruction(s) regarding this barley, whether (it is) to be returned or not to be returned?’ (AbB 4, 156:7–9).

aššum epēšim ū lā epēšim

A similar example is found with aššum:

[77] ¢ēm ´ābim ša ikaššadam aššum ana GN etēqim ū lā etēqim annītam lā annītam bēlī lišpuram ‘Let my lord write me this or that in re-gards to the army that approaches, about moving onward or not moving onward to GN’ (ARM 14, 112:19–22).

These infinitive constructions which express an alternative are the re-duced parallel of those šumma clauses which express indirect questions since they occupy the same syntactic slot.

6. Conclusions

1. Structure~pattern

No single pattern expresses indirect questions in OB: several strategies may introduce questions indirectly: šumma is the most common exponent but it does not head a formally subordinate clause, since there is no sub-junctive. êma and kīma do head formally subordinate clauses, but show different orders: whereas the former is always within the confines of the clause, occurring preverbally, the latter occur post-verbally and in that it is similar to the direct speech marker (where the clauses are actually chained, rather than subordinate). These exponents are easily explained: šumma is a connective that by and large expresses low certainty; êma and kīma are similar: they are both interrogative particles in the construct state, i. e., marked as conjunctions. 2. Syntactic status

The syntactic status of the different strategies (to the exclusion of the di-rect speech marker) is considered the same, where all clauses which take the slot of the object complement are deemed embedded, regardless of whether they are formally subordinate or not. The question, this way, is

E. Cohen, Indirect Representation of Questions in OB Akkadian 81

the complement of the matrix expression (first and second columns on the left of the table below). 3. Negation

It is notable that the examples for indirect representation of a question (including embedded direct questions) are basically affirmative, with only two exceptions (one mentioned under n. 15 and another is found in ex. [62]. This is very different from directly posed questions which are of-ten negative.

4. The function of the indirect question in OB is hardly ever to report a question. Basically, it is used as a strategy to actually ask a question but in an indirect way, namely, mediated by another expression.

5. The matrix expressions used (summarized on the rightmost column in the table below) are various verbal forms with a natural tendency for such complements: asking, not knowing, finding out, order (this or that), etc. More curious are the categories that characterize these expressions: besides a couple of expressions which refer to the past, e. g., exx. [43] and [55]), the present~future is the prevalent framework. More impor-tant: the most common form of matrix expressions is the precative or the imperative. These forms create a domain which has its own syntax (for details see Cohen forthcoming).

6. Despite the fact that OB does not have a single way of expressing indi-rect questions, we have seen that the category of indirect questions in OB is richer than what has been assumed. It is true that this expression is not yet crystallized, and not for every type of direct question is there a correspond-ing indirect one (namely, for pronominal questions, we have no strategy for who or what, but we do for where and for how). Moreover, these ques-tions are not really characterized by a shift of all deictic elements of the pu-tative original questions. Nevertheless, compared to Biblical Hebrew and Classical Arabic, it can be stated that the category no doubt exists.

Embedded question Nominal anchor Matrix clause

nexus ALA,* warkatam ‘matter’ šapārum ‘write,’ parāsum ‘decide’

ALA, ¢ēmam ‘information’ šapārum ‘write,’ šūbulum ‘send’ Direct

pronominal —

naplusum ‘observe,’ burrum ‘find out,’ šapārum ‘write’

eran
Highlight

82 Ancient Near Eastern Studies: Articles

ALA šapārum ‘write’ šumma ‘whether’ —

šepûm ‘ask,’ burrum ‘find out,’ šapārum ‘write,’ šâlum ‘ask’

êma ‘where’ — ul edûm ‘not know,’ amārum ‘find’ Indi-

rect kīma ‘how’ — šâlum ‘ask’

reduced ALA, purussâm ‘decision,’ ¢ēmam ‘information’

*annītam lā annītam

References

Adger–Quer 2001

Banfield 1973

Beccaria 1996

Cohen 2000

Cohen 2005a

Cohen 2005b

Cohen 2006

Cohen 2012

Cohen forthcoming

Coulmas 1986

Cristofaro 2003 Curme 1931

Deutscher 2007

Dillmann 1907

Durand 1992

Adger, D.; Quer, J. The Syntax and Semantics of Unse-lected Embedded Questions. Language 77:107–133. Banfield, A. Narrative Style and the Grammar of Direct and Indirect Speech. Foundations of Language 10:1–39. Beccaria, G. L. Dizionario di linguistica e di filologia, met-rica, retorica. Torino. Cohen, E. Akkadian -ma in Diachronic Perspective. ZA 90:207–226. Cohen, E. The Modal System of Old Babylonian (HSS 56). Winona Lake. Cohen, E. Addenda to Non-Verbal Clauses in Old Baby-lonian. JSS 50/2:247–279. Cohen, E. Syntactic Marginalia in Old Babylonian. JAOS 126:551–565. Cohen, E. Conditional Structures in Mesopotamian Old Baby-lonian. Winona Lake. Cohen, E. The Domain: A Formal Syntactic Unit above Sentence Level. Isaksson, B.; Persson, M. (eds.). Strategies of Clause Linking in Semitic Languages. Proceedings of the In-ternational Symposium on Clause Linking in Semitic, Kivik, Sweden, 5–7 August 2012. Wiesbaden. Coulmas, F. Reported Speech: Some General Issues. Coul-mas, F. (ed.). Direct and Indirect Speech. Berlin. Pp. 1–28. Cristofaro, S. Subordination. Oxford. Curme, G. O. Syntax: A Grammar of the English Language. Vol. III. New York–London. Deutscher, G. Syntactic Change in Akkadian: The Evolution of Sentential Complementation. Oxford. Dillmann, A. Ethiopic Grammar, second edition enlarged and improved by Carl Bezold, translated with additions by J. A. Crichton. London. Durand, J.-M. Unité et diversité au Proche-Orient à l’époque amorrite. Charpin, D.; Joannès, F. (eds.). La

eran
Highlight

E. Cohen, Indirect Representation of Questions in OB Akkadian 83

Ewald 1881 Fischer 1987 Fludernik 1993 Frajzyngier 1995 Frajzyngier 1996 Givón 2001 Goldenberg 1985 Goldenberg 1991 Goldenberg 1998 Huehnergard 1986 Izre’el–Cohen 2004 Jespersen 1924 Jespersen 1961 Keevallik 2011 Leonarduzzi 2000 Nöldeke 1904 Palmer 1986 Plank 1986 Reckendorf 1898

circulation des biens, des personnes et des idées dans le Proche-Orient ancien (CRRAI 38). Paris. Pp. 97–128. Ewald, H. Syntax of the Hebrew language of the Old Testa-ment (reprint 1891). Edinburgh. Fischer, W. Grammatik des klassischen Arabisch2. Wiesba-den. Fludernik, M. The Fictions of Language and the Languages of Fiction: The Linguistic Representation of Speech and Con-sciousness. London–New York. Frajzyngier, Z. A Functional Theory of Complementiz-ers. Bybee, J. L.; Fleischman, S. (eds.). Modality in Gram-mar and Discourse. Amsterdam–Philadelphia. Pp. 473–502. Frajzyngier, Z. Grammaticalization of the Complex Sentence: A Case Study in Chadic. Amsterdam. Givón, T. Syntax: An Introduction. 2 vols. Amsterdam. Goldenberg, G. On Verbal Structure and the Hebrew Verb. Bar-Asher, M. (ed.). Language Studies. 1. Jerusalem [In Hebrew; English translation in Goldenberg 1998: 148–196]. Goldenberg, G. On Direct Speech and the Hebrew Bi-ble. Jongeling, K.; Murre-van den Berg, H. L. (eds.). Studies in Hebrew and Aramaic Syntax. Leiden. Pp. 79–96. Goldenberg, G. Studies in Semitic Linguistics: Selected Writ-ings. Jerusalem. Huehnergard, J. On Verbless Clauses in Akkadian. ZA 76:218–249. Izre’el, Sh.; Cohen, E. Literary Old Babylonian (Lincom Europa Languages of the World: Materials 81). Munich. Jespersen, O. The Philosophy of Grammar. London. Jespersen, O. A Modern English Grammar on Historical Principles. 7 vols. London. Keevallik, L. “Interrogative Complements” and Question Design in Estonian. Laury, R.; Suzuki, R. (eds.). Subordi-nation in Conversation: A Cross-Linguistic Perspective. Amster-dam. Pp. 37–68. Leonarduzzi, L. La subordonnée interrogative en anglais con-temporain. PhD. diss., Université de Provence-Aix-Mar-seille I. Nöldeke, T. Compendious Syriac Grammar. Tr. by T. A. Crich-ton. London. Palmer, F. Mood and Modality. Oxford. Plank, F. Über den Personenwechsel und den anderer deiktischer Kategorien in der wiedergegebenen Rede. Zeitschrift für germanistische Linguistik 14:284–308. Reckendorf, H. Die syntaktischen Verhältnisse des Arabi-schen. Leiden.

84 Ancient Near Eastern Studies: Articles

Tobler 1894 Van Valin 1984 Ziegler–Charpin 2007

Tobler, A. Pieç’a, guere n’a, peut-être, est-ce que, c’est que, c’est … que und ähnliches ohne temporale Be-stimmtheit. Tobler, A. Vermischte Beiträge zur Französi-schen Grammatik. II2. Leipzig. Pp. 1–14. Van Valin, R. D. A Typology of Syntactic Relationship in Clause Linkage. Berkeley Linguistics Society 10:542–558. Ziegler, N.; Charpin, D. Amurritisch Lernen. WZKM 97 (Festschrift für Hermann Hunger). Pp. 55–77.