Optimality in the "Grammars" of Ancient Translations

Articles in JHS are being indexed in the ATLA Religion Database, RAMBI, and BiBIL. Their abstracts appear in Religious and Theological Abstracts. The journal is archived by Library and Archives Canada and is accessible for consultation and research at the Electronic Collection site maintained by Library and Archives Canada. ISSN 1203–1542 http://www.jhsonline.org and http://purl.org/jhs Volume 15, Article 7 DOI:10.5508/jhs.2015.v15.a7


INTRODUCTION
Researchers analyzing the techniques employed by the ancient translators of the Hebrew Bible have developed a remarkably sophisticated range of theory, methods, and terminology to capture, analyze, and describe their findings. 1The burgeoning number of studies in the translation technique of the Targumists, the Septuagintal translators, and others testifies to the increasing attention devoted to this subject; in turn, this increased focus has pointed to the importance of the ancient versions in early endeavors to interpret the biblical text.Yet, the proliferation of studies elucidating the various techniques of these early translators has come with a cost: the thickness of the proffered descriptions and the notational systems (if any) used to convey them vary from researcher to researcher, and do not lend themselves easily to systematic collection and cross-corporal comparison. 2 There is no easy way to iden-1 Included in this category are works such as T.A.W. van  2 For the importance of "describing" translation technique, see, e.g., Aejmelaeus, On the Trail, 212.The difference in descriptive endeavors can be seen for example in a comparison of van der Louw, Transformations in the Septuagint, with Boyd-Taylor, Reading between the Lines.Although both offer "thick description" of the translators' techniques, the form of the descriptions can take significantly different shape, making comparison of the results challenging.number of specialists in Septuagint as well, even if they have not traditionally appealed to these principles as "universals." 4If the object of Descriptive Translation Studies is to identify probabilistic, conditional laws, whose conditions are subject to specification and elaboration, 5 the method I elaborate here has the advantage of being able to track and rank the norms employed in various translational works, thereby creating and managing sets of the constraints operative in individual translations.I envision that this systematization of constraints and their intra-corpus "matrix hierarchies" may be helpful in schematizing and conducting cross-corporal comparisons.

Previous Approaches to Translation
Beginning in the 1960's several research paradigms developed, all treating various aspects of translation theory.Biblical scholars are perhaps most familiar with the "dynamic equivalence" paradigm, propounded by Eugene Nida. 6This paradigm stresses the need for translators to invest their literary creations with the greatest degree of equivalence possible, with respect to the text's effects on the reader-whatever, exactly, the most salient intended effects of the original text are taken to be.But this paradigm is hardly the only non of "false friends" (i.e., historically cognate lexemes with divergent semantic values)-can affect translators' decisions.See, e.g., M. Tercedor, "Cognates as Lexical Choices in Translation: Interference in Space-Constrained Environments," Target 22 (2010), 177-93. 4Septuagintalists often explicitly cite or inadvertently hit upon G. Toury's identification of probabilistic "translational laws" (Descriptive Translation Studies-and Beyond [rev.ed.; Benjamins Translation Library, 100; Amsterdam: John Benjamins, 2012], 295-315 [300-1]).Toury identifies two "laws," both of which bear directly on Septuagintal studies: "the law of growing standardization" (ibid., 303-10), which is typically realized in Septuagintal studies as the translator's increasingly stereotyped and formulaic renderings (e.g., Aejmelaeus, On the Trail, 11-29, esp.19: "once the equivalence was established, they did not mind working it to death"); and "the law of interference" (Toury, Descriptive Translation Studies-and Beyond, 310-5), commonly identified as the "Hebraization" of Greek grammar (especially syntax) in the Septuagint.Increasingly, Septuagintalists have adopted the theoretical appliances and terminology of Descriptive Translation Studies (e.g., Boyd-Taylor, Reading between the Lines, esp.55-87). 5See Toury, Descriptive Translation Studies-and Beyond, 295-315 (300-1); see also A. Chesterman, Memes of Translation.The Spread of Ideas in Translation Theory (Benjamins Translation Library, 22; Amsterdam: John Benjamins, 1997).
6 E.g., E. Nida, "Principles of Correspondence," in L. Venuti (ed.),The Translation Studies Reader (3rd ed.; London: Routledge, 2012), 141-55; for fuller discussion, see A. Pym, Exploring Translation Theories (London: Routledge, 2010), 6-42.one being practiced, and translation theorists typically stress a number of problems with the model.Others argue that the translator cannot or should not be bound to crafting a semantically or functionally parallel document, but should instead deliver the translation in a form that performs a function agreed upon by the translator and the "commissioner" (either a person or a group). 7hat both of these theories share in common is their generally prescriptive stance towards translation: practitioners such as Nida saw their goal to be crafting a skilled translation (i.e., a target text) of important documents and literature-e.g., owner's manuals, legal codes, the biblical text, etc.-in ways that were both somehow representative of each respective source text while at the same time being functionally meaningful and socially acceptable within the intended target culture.As is the case in many fields, practitioners of these different paradigms vied for power within academic frameworks, often with the goal of gaining for their methods intellectual legitimacy or, more pragmatically, workplace opportunities. 8lthough my brief account here is an extreme oversimplification of the myriad forces and underlying currents affecting translation theories throughout this period, it is sufficient to provide context for the following statement: Descriptive Translation Studies arose as a reaction against this trend toward prescription.

Descriptive Translation Studies at a Glance
Instead of arguing over the proper way to translate texts in the modern economy or in an evangelistic religious setting, descriptivists prefer to study the principles whereby translations have arisen historically.In choosing description over prescription, these theorists opted out of several established arguments.Foremost among these arguments was the question concerning what, if anything, a translation is.Whereas the equivalence paradigm stressed semantic and functional equivalence as necessary qualifications of a true "translation," other paradigms insisted that such strictures privileged the source text without imputing adequate significance to the target text's communicative and aesthetic roles in the target culture.Traditional equivalence theory, however, disputed such texts' claims to being authentic translations: can a play by Berthold Brecht in which entire scenes are rewritten to convey to an English-speaking audience the characters' emotive utterances in the German originalor, even more drastically, to conceal the latent political theories espoused by the original-really qualify as a "translation"?How does this not simply become a rewriting, an adaptation of Brecht's play? 9Descriptivists dodge this argument by accepting that every-thing that is understood or assumed to be a translation qualifies as a "translation," and thus becomes material eligible for study. 10By casting the net broadly, descriptivists have at their disposal access to the multiplex modes of rendering source texts in target languages that differ from the source language.
Similarly, descriptivists focus on the recipient target culture's appropriation of the target text 11 : they thus dodge the accusation of being unduly committed to the source text (an accusation frequently leveled against equivalence-based theories), while at the same time allowing for the legitimacy of various modes of translation in the target language.In fact, argues Toury, it is the target culture itself that initiates the translation as a way to fill gaps, even if only perceived, in its literary repertoire.These gaps themselves become worthy objects of study, since they too are facts of the target culture.As Toury notes, when multiple alternative translations in a single language are made of a single source text, they "are not likely to occupy exactly the same position and fulfil (sic) the same functions in the culture that hosts them.This in itself is reason enough why no translation should ever be studied outside of the context in which it came into being." 12At the same time, however, descriptivists typically do not shy away from talking about the target text's relationship to its source: The systemic position most relevant to the kind of questions we wish to pursue is of course the one a translation was designed to occupy when it first came into being . . .This would be achieved by weighing the original position of the [target] text against the findings concerning its make-up and formulation, and the way it represents its original, while taking into account what is already bers: Text, System and Refraction in a Theory of Literature," in L. Venuti (ed.),The Translation Studies Reader (3rd ed.; London: Routledge, 2012), 203-19.
10 Toury points to the circularity of reasoning engendered by paradigms seeking to qualify what constitutes a "translation" before analysis (Descriptive Translation Studies-and Beyond, 26-31).Yet as Toury notes, the criterion of "assumed translation" can pose some problems, particularly with respect to pseudo-translations (i.e., texts crafted so as to appear as though they were originally composed and published in a language different from the one in which they were actually first written; ibid., 20-1, 47-59).Even so, these texts themselves betray common assumptions in the putative target culture concerning translational norms commonly observed by translators; study of them as filling a particular role in the putative "target" culture is thus warranted.Chesterman provides a thoughtful and cogent survey of the various paradigms (Memes of Translation, 5-17); although he works from a position quite similar to that of Toury, he offers friendly critique of many of Toury's arguments as well.
known about the translation tradition in which it came into being and of which it became part. 13e procedure that Toury employs to determine the precise nature of this relationship actually begins with the acceptance of the postulates underlying a translational act: in order to begin a descriptive analysis of an assumed translation, one must accept that (a) there exists some source text, even if unidentifiable, (b) to which the target text exhibits some "tangible relationships," that (c) have undergone a set of transfer operations. 14The fulfillment of these three postulated criteria constitutes sufficient reason to consider the text under study a "translation," even if no source text can be identified.
Once this set of criteria has been accepted, analysis of the text as a translation proceeds from the point of view of the target side: "Such texts, or aspects thereof, would first be studied on their own terms; namely, in terms of their acceptability on all relevant levels, not only as target language texts, but also as translations into the target culture." 15Although it may not always be the case that a corresponding source text can be identified, it is possible to narrow the focus of the study to smaller-scale mappings between the two texts.In turn, these mappings may be analyzed to determine the degree to which the target member of each pair diverges from its corresponding source member: Once a particular text in a language other than the target language has tentatively been marked as the corresponding source of an assumed translation, the next step is to map the assumed translation onto its assumed counterpart, in an attempt to determine the (uni-directional, irreversible) relations that obtain between the pairs of texts and hold them together….
Owing to many inherent limitations, some of them no doubt cognitive in nature, it will normally be segments of the assumed target text (rather than the text as a complete entity) that would be mapped onto parallel segments of the assumed source text.In the process of mapping, the status of the former as 'translational replacements' would be established, along with what they may be said to have replaced…thus shedding light on translation problems as manifested in the particular act that yielded the target language text under observation…, and on their solutions.Shifts (from a given notion of 'maximal' or 'optimal' rendering) can also be identified and studied, if 13 Ibid., 25; the first emphasis is original, I have added the second. 14Ibid., 28-31.The technical terminology of "transfer operations" is Toury's. 15Ibid., 31 (emphasis original).
deemed justified, interesting and/or feasible in the framework of the research undertaken (sic). 16ury recognizes the ambiguity in the qualification of shifts as being departures from the "optimal rendering"-but how is this "optimal rendering" to be determined?Toury continues: Having been thus established for a series of paired segments, and grouped together on the basis of the comparisons themselves, translation relations could then be referred to the concept of translation that may be said to underlie the text as a whole.This will be done through the mediation of a revised notion of equivalence, conceived of as that translation relationship which would have emerged as constituting the norm for the pair of texts under study. 17 Yet this criterion, too, is ambiguous, or, at least, permissive of a number of alternative formulations.A significant problem arises from the realization that Toury's definition of "norms" is apparently not unequivocal: is a norm to be defined as "a revised notion of equivalence," as in the second block-quote here, 18 or should it be defined as "a given notion of 'maximal' or 'optimal' rendering" (as in the first block-quote), even when that optimal rendering is not one of formal equivalence?To begin with, Toury posits an "initial norm" as the initial choice made by the translator as to the proportion to be aimed at between adopting a stance of adequacy (that is, the perceived faithfulness with which a translation represents the content and function of its source) over against acceptability.This choice is, he argues, logically prior to lower-level translation decisions; 19 in this regard, the initial norm might be compared to the translator's commission (in Vermeer's terminology), whether that commission is selfimposed, negotiated with a "client" of some sort, or imposed almost entirely by an external commissioner.The logical priority of the choice, however, cannot be considered an immutable constant lurking resolutely in the shadows behind any act of translation; 16 Ibid., 32 (emphasis original). 17Ibid., 32 (emphasis and lineation original). 18For shifts as departures from a formally equivalent translation, see J.C. Catford, "Translation Shifts," in L. Venuti (ed.),The Translation Studies Reader (1st ed.; London: Routledge, 2000), 141-7; and K.M. van Leuven-Zwart, "Translation and Original.Similarities and Dissimilarities I-II," Target 1 (1989), 151-81; Target 2 (1990), 69-95.I am particularly concerned here with van Leuven-Zwart's "microstructural shifts," although I do not rule out the utility of her category "macrostructural shifts." 19 Toury, Descriptive Translation Studies-and Beyond, 80.For further discussion of the "norm-governed" nature of translation, see Chesterman, Memes of Translation, 51-85.Chesterman's categories differ from Toury's; although I find both systems helpful, I have deferred here to Toury for the sake of brevity.rather, "the choice between adequacy and acceptability may be (or should I say: is?) repeated time and again during the [translation] act, whereby proximity to either extreme serves as a central feature of lower-level decisions." 20Individual translation decisions (i.e., what Toury calls micro-level decisions) "tend to reflect" the initial norm, even if they are not entirely "made in full accord with one and the same initial norm." 21fter the initial norm is formulated, a variety of operational norms are applied during the process of translation.These operational norms comprise a set of "'instructions' [that] specify what is prescribed and forbidden, as well as what is tolerated and permitted in a certain behavioral dimension . .." 22 The instructions are socially-negotiated, with rewards for the translator's adherence to them-acceptance within the guild, additional opportunities to perform his craft for pay, etc.-and "negative, even punitive" repercussions ("sanctions") "in the case of the violation of a norm, or failure to act in accordance with it." 23Yet despite the viciousness and brutal efficiency with which norms can operate, they need not ever be explicitly articulated to retain their efficacy. 24Insofar as these socially-negotiated norms operate on translators, confining and circumscribing their behavior, it is important to note that their very existence "impl[ies] the need to select from among a series of alternatives, not necessarily a final one, with the additional proviso that the selection be non-random." 25In short, the choices that a translator makes in producing any target text are subject to the operation of norms.Therefore, although I use the term "shift" to indicate a departure from purely formal equivalence, 26 the definition of a translation "norm" is somewhat more complicated.By "norm" I mean here a social, political, literary, or other type of constraint underlying and motivating an established pattern of translation.Norms can, but do not necessarily, motivate any specific degree of [formal] equivalence.
Once this terminological problem has been settled, a second one emerges: we need to ask, how does the researcher identify the various norms exhibited in the target text?For Toury, the answer to 20 Ibid., 80. 21 Ibid., 80-1. 22Ibid., 63. 23 Ibid., 64 (emphasis original). 24Ibid., 64. 25 Ibid., 64. 26 This is the traditional sense of the term (along with "transformations") as employed by most scholars involved in studying the early biblical versions; e.g., T.A.W. van  this second question relies on a feedback system, in which the "discovery procedures" already performed are confirmed through "justification procedures" moving in the opposite direction (i.e., beginning with the translational segments and moving higher up on the gradient towards analysis of the text as a whole.In this manner, Toury intends to isolate "the considerations that may have been involved in the decisions whose results were first to be identified, along with factors that may have constrained the [translational] act." 27Moreover, this feedback system of justification does not only operate "when the discovery procedures have been exhausted.Rather, in every phase of the study, from the very start, there is room for suggesting tentative explanatory hypotheses, which will then reflect back and affect subsequent questions and discoveries.The normal process of a study is thus helical rather than linear . .." 28 In Toury's method, then, the justification procedures are always working alongside the discovery procedures, the former serving to clarify, reformulate, and sharpen the latter throughout the course of the study.As conceptual entities that leave only traces of their operation, "norms . . .will still need to be recovered from instances of [translational] behaviour, using the observed regularities as a clue . . .[N]orms do not appear as entities at all, but rather as explanatory hypotheses for actual behaviour and its perceptible manifestations." 29n the following section, I describe the basic contours of Optimality Theory, with the explicit goal of elaborating on its usefulness for capturing and depicting many of the insights that Descriptive Translation Studies has made concerning probabilistic tendencies and norms in translation.I also describe how the theoretical application of Optimality Theory has already been anticipated in important ways in the work of Septuagintalist Cameron Boyd-Taylor. arranged, the rules are predictive of derived forms in the language under study, but the rules themselves are unpredictable, and can only be formulated through careful attention to the collected assemblage of linguistic data.The variety of possible rules in this generative model is nearly infinite, and linguistic researchers must formulate rules that fit the evidence in such a way that the proper composition of the rules and their linear ordering operate in tandem to produce the appropriate (i.e., specified) outcome.
During the early 1990's, American linguists Alan Prince, Paul Smolensky, and John J. McCarthy challenged the classical generative model, proposing Optimality Theory as an alternative phonological theory. 31The theory quickly gained popularity in the United States, fueled in no small part by its early tech-savvy propagation and the prominence of its initial proponents. 32Working from a network-based analysis of human cognition, Optimality Theory proposed that the nearly infinite variety of serially-arranged rewrite rules held as normative by generativists could be replaced more economically by a fixed set of requirements constraining the output forms permissible within any given language.These output constraints operate not serially (as in generative phonology), but rather in parallel, with a fixed and limited set of constraints in different arrangements, corresponding to different languages and linguistic systems.This challenge to generative phonology came with at least two theoretical and methodological benefits.First, the proposal limits the number and complexity of possible constraints, allowing-putatively-easier and more fruitful comparisons cross-linguistically than did the generative model, wherein individual, language-specific rewrite rules often required such minute analyses that comparison of rules became difficult at best.The vagueness of Optimality Theory's generally prescriptive constraints (e.g., "Vowguistics; Cambridge, MA: Blackwell, 1994).els must not be nasal") 33 permits not only broad application across many languages, but also simplifies comparison across those languages.Second, the breadth of application (and the ability of the constraints to be ranked by degree of influence) allows the finite set of constraints to be applied across all languages.Linguistic peculiarities are not solved in this model by positing language-specific, language-internal rewrite rules.Instead, linguistic peculiarities are explained by way of language-specific hierarchies of universally valid output constraints. 34n any given language, these hierarchies are comprised of two basic types of constraints: Faithfulness constraints exert pressure on the language to maintain correspondence between the underlying phonological (or morphological, or syntactic, etc.) form of the linguistic unit and its surface representation.Three subsets of faithfulness constraints obtain: MAXIMALITY constraints (abbreviated MAX), preserve the underlying phonemes, discouraging deletion; DEPENDENCE constraints (abbreviated DEP), prevent the insertion of additional phonemes; and IDENTITY constraints (abbreviated IDENT) preserve the underlying forms of the phoneme, effectively maintaining continuity between input and output forms.At the other extreme, markedness constraints exert pressure to differentiate forms from their respective underlying representations-in short, markedness constraints invest the linguistic system with divergences from the underlying forms. 35These two inclinations, faithfulness and markedness, continuity and change, conflict with one another directly-yet both are essential tendencies within languages.In opposition to generative grammar, wherein rewrite rules are presumed to be inviolable, 36 "[t]he basic assumption of Optimality Theory is that each linguistic output form is optimal, in the sense that it incurs the least serious violations of a set of conflicting constraints." 37Constraints are violable, sometimes multiply so, and even optimal output forms will necessarily violate some constraints. 33Kager, Optimality Theory, 9. 34 The description of Optimality Theory in this and the following paragraphs is directly dependent on the introductions provided by Kager, Optimality Theory; and Nathan, Phonology, 144-56.In order to avoid a surplus of citations, I have refrained from citing those volumes except when quoting directly, or when providing examples drawn from Kager.The reader should recognize that all descriptions of Optimality Theory hereexcept those applied specifically to the translation tactics of Tg.Jon.below-derive ultimately from these introductions, unless otherwise cited. 35It is not altogether clear to me that Optimality Theory's theory of markedness corresponds perfectly to that of Markedness Theory (despite Kager's efforts to assert the contrary [Optimality Theory, 2-3]).
36 Phonological rules were presumed to operate immediately and throughout the language by the so-called "Neogrammarian" school; subsequent research has demonstrated that this position requires further nuance, but that discussion lies beyond the scope of this paper.
Most of the actual output forms, whether words, phrases, or of higher rank, will satisfy most constraints in any given analysis, while violating only a few.The most salient set of possible contenders, therefore, will be readily whittled down to a single output form through the application of a few violated constraints in any single analysis.Yet one of the principles of Optimality Theory that has been most subject to criticism in the linguistic community is its insistence that the cognitive faculty tasked with generating candidates for a given realization (symbolized by Optimality theorists as GEN) actually generates an infinite number of candidates for every linguistic unit. 38The putatively infinite capacities of GEN can be safely disregarded in the analysis presented below, in favor of a more modest model in which a constrained number of output candidates (generated both by habitual linguistic replacement and by non-habitual, innovative consideration of possible translation values) are subjected to evaluation (EVAL) by the translator. 39 38 Se e.g., Ibid., 8: "the grammar generates and then evaluates an infinite set of output candidates, from which it selects the optimal candidate . .." (emphasis added); see also the more developed discussion, p. 26-7; but cf. the more thorough description at p. 19: "Gen is a function that, when applied to some input, produces a set of candidates, all of which are logically possible analyses of the input" (emphasis added).Compare also the "possibly infinite" output of GEN in L. Karttunen, "The Proper Treatment of Optimality in Computational Phonology," esp.§ 3.1 (emphasis added; online: http://roa.rutgers.edu/files/258-0498/roa-258-karttunen-2.pdf[accessed March 22, 2013]).
Accordingly, the nearly-infinite majority of this infinite number of candidates is immediately excluded through their egregious violation of numerous faithfulness constraints, and only a few salient contenders are subjected to the more detailed analysis of the evaluation faculty (EVAL).This principle has been alternately challenged and defended by computational linguists (e.g., J. Eisner, "Efficient Generation in Primitive Optimality Theory," in Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics [1997] posited by most versions of Optimality Theory.In any event, the application of EVAL seems to me to present more problems than it solves, at least as it pertains to the generation of translation variants. 39For an example, see Chesterman's application of Karl Popper's philosophy of scientific inquiry.Chesterman advocates viewing translating as The optimality of actual output forms is taken for granted in Optimality Theory; the researcher's task is to arrange the systemthe hierarchy of constraints-such that it accurately predicts these putatively optimal forms.In service to this end, Optimality Theory's notational system involves tableaux (a technical term employed by Optimality theorists) and a variety of symbols that serve as a shorthand for the various evaluations reached over the course of the evaluation process. 40These charts offer a synchronic diagram of the full process undertaken by EVAL: the linguistic input and the most salient output candidates appear in the left-most column, and the remaining columns are headed by the constraints against which the candidates are to be tested.(The constraints are arranged in the order that the researcher has determined to be most conducive to selecting the optimal [i.e., actual] output candidate, with each abbreviated in small capitals.)An asterisk (*) marks each violation of the pertinent constraint incurred by the output candidate in the left-most cell of the row, and when the point of disqualification has been reached, an exclamation point (!) marks the "fatal" violation (i.e., the violation beyond which the candidate is definitively disqualified, marked with darkened cells; the cells of the optimal candidate are also darkened once they are no longer diagnostic for the analysis).The optimal candidate is marked with an arrow (→), or, in several notational systems, a pointing hand (☞).Candidates are eliminated from contention as soon as they have incurred a violation of a highly-ranked constraint if there are other candidates that have not incurred the same number of violations of the same constraint.When two or more candidates incur violations a recurring series of successive stages of proposing a "tentative theory" and then "error elimination" (Memes of Translation, esp.16  -8).Pym seems to account for (near-)simultaneous generation of multiple tentative texts, with the subsequent elimination of "non-optimal T[arget] T[ext]s" (Translation and Text Transfer, 175-6), comparable to the contemporaneous generation-evaluation process of Optimality Theory, but Gile and Sager both seem to opt for a process comprising serially recursive generation-evaluation pairs. 40My lexical choices in this description are particularly indebted to Kager (Optimality Theory, 13-4), and overlap to a large extent with his terminology, although I have attempted throughout to refrain from using his precise locution. of the same constraint, their respective violations of the next-highest constraint are evaluated.Similarly, when two or more candidates incur violations of the same constraint, the candidate with more violations is disqualified.
I provide the following tableaux using the input *malk, which Semitists will recognize as the (reconstructed Proto-Hebrew) form underlying Tiberian Hebrew ‫ְך‬ ‫לֶ‬ ‫מֶ‬ (mɛlɛχ) "king." 41Although several pertinent constraints could be proposed, I have evaluated the candidates with respect to three constraints: a markedness constraint against word-final consonant clusters (i.e., a variation of the rule that produces anaptyxis between R 2 and R 3 in traditional analyses of Hebrew, abbreviated here as *CC#) 42 ; a markedness constraint favoring the segholation of the base vowel (FAVOR-Ɛ); and input-output constancy of the linear order and identity of segments (a combination of two standard constraints of Optimality Theory, abbreviated here as IO-IDENT43 ).Tableau 1 presents the analysis when the constraints are organized in the order given above, with the constraint *CC# dominating (i.e., being ranked higher than) FAVOR-Ɛ and both of these constraints dominating IO-IDENT.The relationship of domination (siglum: ≫) is presented in (1):44  Ultimately, the only way to determine which of these three proposed constraint hierarchies is correct would be to work painstakingly through the corpus of Biblical Hebrew, testing each of them against the data, and adducing particular examples that would target certain pairs or orders of constraints.In the analysis given in section 4 (below), I provide an example of this process.What is clear, however, from Tableaux 1-3, is that the constraint IO-IDENT must not be ranked as the highest of the three, since if it were accorded domination over both of the other constraints, then the correct (actual) output candidate would incur a violation before candidate (a): A significant point of convergence between historical, generative, and Optimality Theory approaches to phonology should be recalled at this point, which suggests that my use of *malk here is heuristic and unrepresentative of how an Optimality Theory analysis would appropriately deal with an ancient language.Historical linguistics holds a model of the serial application of developmental rules through time, as in (2), with successive generations slowly adopting the successive stages of the reconstructed development: (2) *malk > *malɛk > *malɛχ > (Tib.)mɛlɛχ From the standpoint of the study of the historical development of Hebrew, this gradual development can be proven through recourse to texts transcribed in Greek (i.e., Origen's Secunda), and vocalized in pre-Tiberian (i.e., Palestinian and Babylonian) pointing systems.Most historical linguists do not examine the cognitive operations working to effect this development, and would probably accept a model in which speakers have as their underlying representation simply the form which they have come to learn as grammatical (whether it be /malk/, /malɛk/, etc.). 45Although the continual and unending application of the processes actually occasioned each development, leading to the (phonemic) forms preserved in the text, it is my understanding that neither generativists nor optimality theorists would claim that we have access to the surface realizations or output forms that speakers may have (inadvertently and habitually) applied to the underlying phonological form. 46 45 Somewhat differently, both generative and optimality phonologists admit a historical development of pronunciation, but argue that speakers applied, respectively, rewrite rules or optimality constraints to the underlying phonemic representation of their lexicon.Nathan, for example, differentiates between the historical phonological developments that changed the phonemic realization of the underlying form, as in (2), and the phonetic processes that speakers would have applied to those inputs (Nathan, Phonology, 103-13).In his discussion of opacity, Kager presents a similar theory, in which opacity can be traced to the operation of slightly reconfigured constraint hierarchies at different levels (Optimality Theory, 381-85), but he does not connect this view explicitly to historical linguistics, as does Nathan.Idsardi ("Clarifying Opacity," 346) addresses this historical perspective briefly, but dismisses it because "this [solution] gives Optimality Theory two distinct computational devices with which to do phonology . . .In addition, the levels must be empirically motivated . .."Although I am generally sympathetic to Idsardi's objections to Optimality Theory's less than adequate handling of opacity, it is not clear to me that he has effectively dismantled an "intermediate levels" approach to opacity with this argument.As Nathan points out (Phonology, 153) concerning recent applications of "usage-based theory," "[J.]Bybee . . .has argued that words (and probably larger units too) are stored exactly as heard (and as produced)."This view of language storage conflicts with both Optimality Theory and serial-based phonological theories, and, in my view, might be cause to subordinate both to a unifying theory in which serial derivations at the historical level can be explained, in the words of Idsardi, by "Optimality Theory as the theory of the internal structure of a process" ("Clarifying Opacity," 349). 46The primary difference between generative, derivational theories and Optimality Theory is in how these two models conceptualize the production of surface representations (or output forms) from the underlying representations (or input forms).For derivational theorists, the application of rewrite rules is habitual, ingrained, and hardly problematic for the immensely powerful computing resources of the human brain.Optimality theorists concur that the human brain is immensely powerful, but hold that all pertinent output constraints are run on the underlying forms at the same time, in parallel.This theoretical model is belied by the appearance of the Tableaux and the work-flow involved in processing them: we work linearly from left to right, recognizing that fatal violations of constraints in one column render irrelevant violations in "subsequent" columns.But this appearance is unrepresentative of the way our minds actually process phonological data, argue Optimality theorists: because the output form

Application of Optimality Theory to Descriptive Translation Studies
For my present purposes, it is sufficient to express my own lingering dissatisfaction with the application of Optimality Theory to phonology, while at the same time suggesting that Optimality Theory's attempts to systematize and universalize tendencies of speakers (or translators) may prove to be a useful heuristic and notational tool for a descriptive approach to translation studies.In particular, Optimality Theory's affirmation of the cross-linguistic universality of output constraints, combined with the language-specific peculiarity of constraint hierarchies, may be a major breakthrough in the cross-corporal analysis of translation.Two salient points arose in my summary of Toury's methodological reflections in section 2, both of which justify the application of Optimality Theory in service to Descriptive Translation Studies.First, the centrality of operational norms-as codified in the "prevailing concept of translation"-in the translational methodology that Toury effectively (and, probably, rightly) imputes to translators anticipates the two major categories of constraints operating in Optimality Theory.Toury reminds us that we are forced to recognize and account for the fact that "actual translation decisions will normally be found to involve some combination of, or compromise between, the pressures of the two extremes [adequacy and acceptability], the choice between which constitutes the initial norm." 47The tendency toward adequacy evokes the so-called faithfulness-constraints, in which the impulse is to preserve the initial phonological identity of the lexeme under investigation (IDENT), without adding (DEP) or deleting segments (MAX).In addition to recognizing a translator's attempts to achieve adequacy, Toury seeks to identify the underlying "considerations" or "constraints" underlying the departures (shifts) from each source text segment; Toury connects these shifts-sometimes obligatory, sometimes not-with the impulse toward acceptability in the target language.Some such shifts are inevitable, 48 but "encountering 100% regularity [of shifts] undergoes the evaluative process (EVAL) in a single step, there are no intermediate representations to serve as inputs for an ever-lengthening chain of rewrite rules.This theoretical stance is one of the tenets of Optimality Theory most frequently cited and argued against by members of opposed theoretical camps; see especially the discussion of Optimality Theory approaches to opacity in Kager, Optimality Theory, 372-424; and Idsardi, "Clarifying Opacity," 337-50.There are, of course, many other issues of contention, and the debate has proven to be somewhat intractable. 47Toury, Descriptive Translation Studies-and Beyond, 81.

OPTIMALITY IN THE GRAMMARS OF ANCIENT TRANSLATIONS 19
should arouse immediate suspicion as being too good to be true." 49hese departures correspond strongly to Optimality Theory's second macro-category of constraints: the so-called markedness-constraints (which typically fall into a wider variety of sub-categories than do the IDENT constraints).
A second significant point connects these two paradigms.In attempting to identify regularities in the features of continuity and in systematic changes in a given translator's decisions, Descriptive Translation Studies effectively tries to articulate the degree of systematicity employed in crafting each translation, identifying the areas of abnormality-weaknesses in those putatively-monolithic representations-and attempting to explain them.Optimality Theory too aims to provide a comprehensive, systematic statement of a language's grammar, which it conceives of as the hierarchy of output constraints, while at the same time explaining those points of anomaly.It is therefore not out of line to describe translational systems as "grammars" (as I have done in the title of this study).Optimality Theory, I will argue, provides researchers with a "grammatical" architecture and notational system capable of capturing and accounting for any departures from Toury's "initial norm" that the researcher encounters.

Anticipation of Optimality Theory in LXX-Studies
As noted above, I point here to an important recent study of Septuagintal translation style in which many of the principles associated with Optimality Theory that I apply here explicitly are adumbrated.Cameron Boyd-Taylor's 2011 monograph, Reading between the Lines, adopts a sophisticated theoretical stance in the service of understanding the Septuagint as a product of the culture that produced it-in Toury's terms, as a fact of the target culture. 50lthough a thorough discussion of Boyd-Taylor's study remains beyond the scope of this study, it is worthwhile pointing to several theoretical principles he employs, because they intersect with the proposals laid out here.
First, just as Toury's application of Descriptive Translation Studies employs the idea of socially-imposed norms of translation as constraints, so too does Boyd-Taylor stress the importance of translational norms in constraining how the translator has gone about moving the text from one linguistic system into a different linguistic system. 51Boyd-Taylor's project throughout the book is to recover and identify "the relevant conventions [i.e., norms] that were operative in that culture when the text was produced." 52He reminds us further that these norms "do not operate in isolation," 49 Toury, Descriptive Translation Studies-and Beyond, 81. 50Boyd-Taylor, Reading between the Lines, 55-87. 51Ibid., 39, 57-68. 52Ibid., 60. but instead "are to be viewed in terms of their inter-relations." 53In this assertion, Boyd-Taylor's theoretical stance accords well with that of Optimality Theory, noted above.It will be recalled that the network model assumed by Optimality Theory presumes a set of constraints operating in parallel, rather than serially; the same simultaneousness of application pervades Toury's norms.
Second, Boyd-Taylor adopts Toury's concept of a hierarchy of norms.Some of these norms are nearly mandatory for the translator to obey in order to avoid censure or sanction, others less so.Toury (and Boyd-Taylor following him) schematizes three groups of norms-as summarized by Boyd-Taylor, these categories are "basic or primary norms," "secondary norms or tendencies," and "tolerated (permitted) choices." 54A similarly schematic dissection of the constraints governing translation are found in Boyd-Taylor's relaying of Christiane Nord's "constitutive conventions" (i.e., socially-imposed norms) and "regulative conventions" (norms more broadly applicable in translation). 55In both nomenclatures, those norms, conventions, and choices lower in the hierarchy are more violable than are those ranked higher in the hierarchy (as in Optimality Theory).I have adopted neither Toury's nor Nord's schema in the present study, although my optimality-based analysis similarly attempts to identify those norms that are ranked higher than others.I have not attempted to schematize the constraints into particular categories for the simple reason that Optimality Theory conceives of the constraint hierarchy as lying on an undifferentiated continuum.Nonetheless, my approach remains consistent with those cited by Boyd-Taylor: Toury's and Nord's respective schematizations provide heuristic models allowing researchers to refer to variously-ranked constraints (Toury) emerging from different sources (Nord) with a simple short-hand.
This brief summary of Boyd-Taylor's work has pointed to a few points of tangency between his theoretical framework and the basic principles of Optimality Theory.A more thorough review would enumerate many more such connections; for example, one might compare Boyd-Taylor's notions of "quantitative fidelity" with Optimality Theory's MAX and DEP constraints. 56A more involved study would also note a few differences.Moreover, I leave 53 Ibid., 61. 54Ibid., 60, and passim. 55Ibid., 71, citing C. Nord, Translating as a Purposeful Activity: Functionalist Approaches Explained (Manchester: St. Jerome Publishing, 1997), 58.
56 I have not attempted to formulate these constraints below, because the text under scrutiny does not exhibit substantial shifts in quantity that would lend themselves to such investigation.My sense, however, is that what Boyd-Taylor has called "quantitative fidelity" (Reading between the Lines, passim) actually involves two closely aligned constraints, one of the MAX type (e.g., "represent in the target text everything that is present in the source text"), and one of the DEP type (e.g., "add nothing to the target text that is not present in the source text").aside discussion of the larger thrust of Boyd-Taylor's study, namely, that the Septuagint was composed according to a principle of interlinearity. 57The position I have adopted here, in which I view optimality as "a description of the translator's sense of an appropriate analogue in the target language" for what can be found in the source text 58 does not commit me to the principle of interlinearity.
In section 4, I present a sketch of what an Optimality Theory approach to Descriptive Translation Studies might look like.

TESTING THE SYSTEM
It is necessary now to synthesize and apply the theoretical and methodological frameworks that I have summarized above.The goal of this section is to demonstrate that Optimality Theory and its admittedly enigmatic notational system can be applied fruitfully to a study informed by and participating in the project of Descriptive Translation Studies.By fruitful application, I mean that Optimality Theory provides the schematic and notational means whereby we can (a) determine, represent, and quantify significant constraints comprised by the target literary system's norms; (b) rank those constraints in their order of dominance; (c) apply them to a translated biblical text subdivided into appropriately-sized segments of translation replacement; and (d) compare the segments' respective constraint hierarchies in order to determine whether more specific subcategories of constraints need to be established.
Optimality Theory provides us with an investigatory system whereby the sub-classification of constraints permits a principled method of exploring and reconstructing the precise hierarchical ordering of constraints by reference to subsets of micro-level translational decisions in a given text.Insofar as a single hierarchical constraint arrangement can be shown to constitute the dominant constraint-hierarchy operative in a given text, this arrangement constitutes a quantification of Toury's initial norm; I call this the matrix hierarchy, and we may think of it as the "grammar" of the translation under review.The primary task of an Optimality-based Descriptive Translation Study consists of the reconstruction of this "grammar."It is important to note here that this investigation is primarily concerned with establishing the optimality of the translation in the translator's processes of cognition, insofar as those processes are discoverable from the information provided by the text at hand. 59Previous biblical studies have already gestured toward this principle in identifying literal transla- 57 Ibid., 88-111. 58I have drawn this locution from personal communication from Ronald Troxel. 59A similar problem is alluded to by Boyd-Taylor (Reading between the Lines, 24), who distinguishes between the mental processes of the translator(s) and those of the later recipients of the text.I am unconcerned here with developing a profile of "optimal" translation in the modern day.tion as an "easy technique," 60 or lexically-based interlinearity as an explanatory "concept of equivalency" driving the formation of the Septuagintal text. 61et, when we encounter instances where our established matrix hierarchy does not predict the correct (i.e., actual) replacement candidate, we are alerted to a possible hierarchy anomaly.This prompts us to analyze possible causes for the anomaly.In some cases, we may infer "grammatical" reasons for the anomaly having to do with the formal or dynamic equivalence that the translator attempted to encode with the replacement text (i.e., grammatical constructions required by the target language, genre conventions inserted to satisfy the target polysystem, and so on).In other cases, we may find that the hierarchy anomaly has occurred because of the translator's concern to supply "non-grammatical" (contextual or pragmatic) information such as theological explication and so on.In reconstructing the causes of these departures, we gain additional insight into the interpretive strategies of the translators. 62As will be seen, in the hierarchy anomaly examined below, the solution is a purely mundane one, relating to the linguistic constraints of the target language.
The text used in the following experiment is Tg.Jon. of 2 Sam 11:1 (see text-box). 63I have selected this short sample text because of the relative simplicity of the study that it ensures.Most of the segments have been translated with such a high level of formality that they permit concise formulation of the basic faithfulness constraints.Yet, there are enough dynamic ("expansionistic") tendencies in the target text that we may analyze at least a few shifts from the source text as well.
This short sample text has the advantage of having as its target language another member of the Northwest Semitic language group, closely related to Biblical Hebrew.This filial relationship mitigates the number of obligatory shifts to be considered, because morphological and syntactic categories remain relatively consistent between the two languages; lexical cognates also mitigate the number of shifts to be accounted for. 64  6 Boyd-Taylor, Reading between the Lines, 99 (italics original). 62Compare the similar concerns of van der Louw, "Linguistic or Ideological Shifts?" 107-25. 63See the more traditional analysis of Tg.Jon. 2 Sam 11:1 provided by E. van Staalduine-Sulman, The Targum of Samuel (SAIS, 1; Leiden: Brill, 2002), 545-46.
64 Contrast the frequent need within studies of the Septuagint to evaluate the degree to which the Greek is representative of "translationese"; N. Fernández Marcos, The Septuagint in Context: Introduction to the Greek Version said so far concerning Optimality Theory's faithfulness constraints and adequacy to the source text, we may postulate faithfulness constraints for syntactic identity (siglum: IO-IDENT:SYN); morphological identity (siglum: IO-IDENT:MORPH) 65 ; phonological identity (siglum: IO-IDENT:PHON); and semantic identity (siglum: IO-IDENT:SEM). 66As we will see in the following sections, these faithfulness constraints interact with various markedness constraints to enact the "grammar" of Targum Jonathan's translation technique. 65As one anonymous reviewer helpfully pointed out, it is unwise to describe the Hebrew and Aramaic forms as "morphologically identical," since formally the Hebrew participle is not the same as the Aramaic participle; for a similar theoretical position, see W. Croft, Radical Construction Grammar: Syntactic Theory in Typological Perspective (Oxford: Oxford University Press, 2001), esp.45-61.Nonetheless, I use this term to describe cognate forms that are historically related (i.e., derived from the same Proto-Northwest Semitic form) and whose similar (but not necessarily identical) phonological realization would have bolstered a cognitively-routinized association between the two, especially given the similarity of semantic values.Therefore, I retain the principle of "morphological faithfulness" in the context of translation even while conceding its theoretical limitations.Accordingly, while I speak of "morphological correspondence" below (see, e.g., the term "morphosyntactic correspondence" employed by Boyd-Taylor, Reading between the Lines, passim), that phrase should be recognized as Optimality Theory shorthand for "morphological category commensurateness informed by phonological similarity and semantic identity." 66I use IDENTITY constraints here as a catch-all category encompassing MAX and DEP constraints as well; a more thorough analysis could conceivably consider these constraints separately, especially with regard to syntax.The relative ranking of two of the constraints posited above-IO-IDENT:MORPH and IO-IDENT:SEM-can be tentatively established as a "working norm" by examining the first word of 2 Sam 11:1.The Hebrew narrative wayyiqtōl (*way-yaqtul) form ‫‬ ‫י‬ ‫הִ‬ ‫יְ‬ ‫וַ‬ begins the verse.It is unnecessary to presuppose that the cognitive framework of the translator generates an infinite number of possible replacements, as Optimality Theory often supposes.I find it much more likely that several possible replacements are generated; 67 the number of these possible replacements is already inherently limited by the two basic general input-output identity constraints posited above, and we may simply begin with the reduced set of the two most salient forms selected by the two IO-IDENT constraints, the semantically identical ‫‬ ‫ה‬ ‫וָּׁ‬ ‫הְ‬ ‫,וַ‬ and the morphologically commensurate ‫י‬ ‫יהֵ‬ ‫.וִ‬ 68 In this [replaced] + [replacement] pairing, we make the empirical observation that the translator has selected the semantically identical 69 Aramaic *w-qatal form ‫‬ ‫ה‬ ‫וָּׁ‬ ‫הְ‬ ‫וַ‬ over its most salient alternative, the morphologically correspondent *w-yaqtul form 67 Moreover, the process of generation may entail the use of habitualized or routinized lists and flowcharts; see, e.g., J. Levý, "Translation as a Decision Process," in L. Venuti (ed.),The Translation Studies Reader (1st ed.; London: Routledge, 2012), 148-59; and Toury, Descriptive Translation Studies-and Beyond, 296-7.For an attempt to classify types of replacements, see Chesterman, Memes of Translation, 89-116. 68Strictly speaking, ‫י‬ ‫יהֵ‬ ‫וִ‬ is the Aramaic reflex of *w-yaqtulu, not of *wyaqtul.However, I am unable to find any examples in Tg.Onq. and Tg.Jon. of the latter form of ‫הוה√‬ (using the basic search functions of Accordance 8.0.4), and Biblical Aramaic uses a different morph for the volative ‫א(‬ ‫וֵ‬ ‫הֱ‬ ‫;לֶ‬ e.g., Dan 2:20).Tg.Jon. 2 Sam 24:17 renders Heb.‫י‬ ‫הִ‬ ְ ‫ת‬ with Aram.‫י‬ ‫הוֵ‬ ִ ‫,ת‬ suggesting that the long form, as given here, may have been the closest approximant available, morphologically speaking.Additionally, Tg.Job 1:21 does render the Hebrew apocopated form with ‫יהא‬ (cf.also Tg.Qoh.10:19), which may allow for an apocopated Aramaic cognate, but without additional pointing we cannot identify the precise form used. 69 These morphs are "semantically identical" insofar as both denote verbal past tense in combination with a conjunction.In interlinear morphemic notation, both may be represented as: COP-be.PST.A morphological distinction obtained, of course, between the Hebrew wayyiqtōl form (< *wayyaqtul), which typically indicates narrative past-tense, and what will be seen to be its habitual Aramaic replacement, the wǝ-qǝtal (< *w-qatal) form.I simplify here, since in BH the wayyiqtōl form is, synchronically speaking, a single morpheme denoting past tense (e.g., J. Cook, Time and the Biblical Hebrew Verb.The Expression of Tense, Aspect, and Modality in Biblical Hebrew [LSAWS, 7; Winona Lake, Ind.: Eisenbrauns, 2012], 256-65).
Because there exists in Aramaic a root cognate to Hebrew ‫,ׁשחת√‬ 73 this variation might recommend the separation of SEMANTIC FAITHFULNESS (IO-IDENT:SEM) from LEXICAL FAITHFULNESS (IO-IDENT:LEX), with the former dominating the latter. 74

The Participle and Infinitive
Other replacement verbal forms in the sample text demonstrate adherence to the same set of constraints.For example, the translator rendered the Hebrew participle ‫ב‬ ֵ ‫יֹוׁש‬ with the semantically-identical Aramaic cognate ‫יב‬ ֵ ‫ת‬ ‫.יָּׁ‬In this first case, all semantic elements of the replacement demonstrate faithfulness to those of the replaced word; there would presumably have been little resistance to this replacement.Similarly, we see in the segment contained in (4) that the translator has replaced the Hebrew infinitive construct with a semantically-identical morphological form (the Aramaic Gstem infinitive 75 ) bearing the same syntactic and semantic relations to the surrounding replacements.term: Heb.
‫א‬ ‫לָּׁ‬ ‫בָּׁ‬ ַ ‫ח‬ ‫לְ‬ ‫ין‬ ִ ‫ת‬ ְ ‫ׁש‬ ‫תעַ‬ ‫מִ‬ ‫א‬ ָּׁ ‫ׁשּור‬ "plotting to destroy the wall."In short, the morphologies have been preserved while the constituent roots have been shifted around; this example serves to demonstrate the complexity encountered in studying translational norms. 73A lemma search of the Comprehensive Aramaic Lexicon (online: http://cal1.cn.huc.edu/[accessed Feb. 8, 2013]) produces a meaningful cognate found in several Aramaic dialects.Three separate senses are given for the D-stem of the root: "to stain" (Syr.), "to ruin" (OA, Sam.), "to delete" (Syr.).Further study would be necessary to explore why this root was not used to render the Hebrew in Tg.Jon. 74The familial relationship between Aramaic and Hebrew renders possible and necessitates a constraint IO-IDENT:LEX.The isolation of this constraint permits circumspection towards etymological "false friends" (i.e., cognate roots with variant semantic fields or discursive distributions, such as Aram.‫ענה√‬ "to begin to speak," which influenced the later [LBH] semantic value of Heb.‫ענה√‬ "to answer").The same constraint, obviously, would also not be accorded high position in an Optimality Theory analysis of the "grammar" of LXX translation technique, since nearly every lexical unit undergoes IO change. 75 "at the time of the departing of kings" → "at the time when kings went out (to war)" 79 This correspondence occurs despite the fact that Hebrew and Aramaic infinitives in the G-stem are formed with different morphological augmentations.In this second case, the IO-IDENT:MORPH (i.e., MORPHOLOGICAL CORRESPONDENCE 80 ) con- 76 The two lexemes used in Hebrew and Aramaic meaning "time, season" are typically realized in different genders.Hebrew uses the feminine singular ʿēt, whereas Aramaic uses the masculine singular ʿiddān.I have collapsed this distinction in the interlinear morphemic notation by omitting reference to the gender of the noun entirely.It is important to note that this transformation is an obligatory one, following on normal lexical usage within each linguistic system. 77The division of hammǝlākîm into three (morphemes) is, of course, a simplification of the actual structure of the word, in which the nominal base *mal.khas been supplemented with the masculine plural morpheme *a.îm, after which a process of phonetic reduction took place.For our purposes, it is sufficient to mark the base noun and the MPL suffix as separate. 78I disregard here the obligatory structural shift in which the Aramaic counterpart malk-ay-yā(ʾ) exhibits the form: king-MPL-DEF.ART.This structural shift will be assumed in the discussion below.I also disregard the consonantal text ("messengers"), preferring to read the Masoretic raphe as intending "kings."See, e.g., P.K.

Summary and Conclusions
In this paper, I have provided brief overviews of the principles and methodologies of Descriptive Translation Studies and of Optimality Theory.In the detailed analysis of Tg.Jon. 2 Sam 11:1, I have tried to show: (i) how these two theoretical approaches can be used in tandem to track and rank the norms employed in various translational works, thereby creating and managing sets of the constraints operative in individual ancient translations of the Hebrew Bible; (ii) that the theory and formalisms of Optimality Theory may be used to capture and organize the translation descriptions provided through Descriptive Translation Studies, even if statistical methods as a whole can or should be rejected88 ; and (iii) that the notational system employed by practitioners of Optimality Theory allows us to formalize observations in a straightforward manner, capturing anomalies in translation and subordinating translation norms to universally-valid principles, even if those principles receive different degrees of authority in the variant approaches to translation employed in antiquity.
throughout the world's languages, my own application may need to revise this view slightly.I am forced on the one hand to posit a group of faithfulness constraints mandating (or merely recommending) the transmission of grammatical, semantic, and pragmatic features of the source text into the target text (compare Toury's concept of "adequacy").Further study may allow the qualification of these constraints as common across translation practices as a whole, in a way similar to Christiane Nord's "regulative conventions."On the other hand, it would seem that the constraints I have identified as markedness constraints must be language-and culture-specific (compare Toury's concept of "acceptability").This tentative differentiation of translation-universal faithfulness constraints and language-specific markedness constraints requires further elaboration in future studies.

Directions for Further Investigation
A few complications exist here that have not yet been addressed, but which would be pertinent for further investigations into the possibility of Optimality Theory analysis of translation techniques.

Delimitation of Segments
One such complication is the accurate identification of [replaced] + [replacement] segment couplings.The method that I have established here seems best to apply to translations that are "integral," that is, translations that "[contain] no additions or deletions transcending the sentence level." 89Although the corresponding segments of the investigation conducted here were easily identified, the issue is not always so straightforward. 90From the cognitive standpoint of the translator, translation problems are never simply solved and left behind, but rather they establish a residual cognitive network of previous equivalence-based solutions, often with the effect of creating entrenched lexical and morphological correspondences, even if the [source text segment] + [target text segment] coupling does not provide for semantic adequacy in some circumstances. 91Two theoretical conclusions follow.First, in theory, the boundaries between segments are significantly more fluid than is implicit in the presentation given above; and, as Toury recognizes, correspondence units are not limited to "identical" scope and rank. 92Second, we must take into account cognitive linguistics' reaction to the hard-and-fast distinction between semantics and pragmatics held by more formalist, generative approaches to lin-89 Van Leuven-Zwart, "Translation and Original (I)," 154. 90Toury (Descriptive Translation Studies-and Beyond, 115-29) provides a well-reasoned description of the precise means by which corresponding segments of source text and target text are determined. 91Ibid., 116-7. 92Ibid., 117.
guistics.Ronald Langacker disputes the concrete division between semantics and pragmatics, arguing instead that the two facets of meaning-making in linguistic utterances are ends of a continuous, gradable continuum, with no established boundaries between the two. 93If this account is accepted, then we must also accept the possibility that different linear (syntactic) orderings project variant pragmatic structures (including some phonetically realized with no additions of morphological or phonological data 94 ), and that these structures may necessarily be rendered in the target language in vastly different ways.That is, their solutions may violate IO-IDENT:SYN and IO-IDENT:MORPH while at the same time satisfying IO-IDENT:PRAG.If so, how are we to analyze IO-IDENT:SEM?And are we certain that we have correctly evaluated the boundaries of the segment?It may, at times, be exceedingly difficult to make individual arguments for segmental correspondences.An example of this difficulty is posed by a case mentioned by Lefevere, in which the translator has added dialogic exchanges between the two interlocutors in order to capture more fully-more dynamically-the snippet of conversation: Yvette: Dann Können [sic] wir ja suchen gehn, ich geh gern herum und such mir was aus, ich geh gern mit dir herum, Poldi, das ist ein Vergnügen, nicht?Und wenns zwei wochen dauert?(Then we can go look, I love walking about and looking for things, I love walking about with you, Poldi, it's so nice, isn't it?Even if it takes two weeks?)becomes Yvette: Yes, we can certainly look around for something.I love going around looking, I love going around with you, Poldy… The Colonel: Really?Do you?Yvette: Oh, it's lovely.I could take two weeks of it!The Colonel: Really?Could you? 95 The expansion of the dialogue here suggests that the translator felt that equivalence-of whatever sort he was attempting to convey-could only be achieved by blurring the lines of the segments, amplifying and clarifying the component segments of material of the original by re-voicing the questions as spoken by the Colonel.This passage exemplifies P. Zabalbeascoa's claim that "it may be the case that not all [translation] solutions are purely segmental in their nature." 96

"Non-Grammatical" Hierarchy Anomalies
Another major complication arises when we discover hierarchy anomalies that cannot be explained as simply as the one encountered in section 4.2.2.As Toury points out, we should not expect a single hierarchical ordering of constraints to obtain ubiquitously throughout any given text. 97There will inevitably occur some instances where the operative constraint hierarchy of a specific segment departs from the text's matrix hierarchy.I agree with Toury in his assertion that "norms themselves are far from monolithic: not only are some of them more binding than others at any given moment, but their validity and potency may not be fixed for a very long time." 98I would hypothesize that the larger the text, the more such hierarchy anomalies are likely to occur.
However, Optimality Theory provides us with an interpretive framework to understand these temporary rearrangements of constraints.Because rearrangements of constraints occur between small subsets of the total constraint inventory, we might posit that the remainder of the inventory-the constraints that remain satisfied-do not experience reordering, thus maintaining the vast bulk of the matrix hierarchy intact.Therefore, there is no need to consider the boundaries between the various matrix hierarchies we encounter in any given text to be hard and fast.Thus, although the precise orderings of constraints may shift minimally between segments, we should be attentive to Toury's larger critique of positing rigid differentiation between variant matrix hierarchies: "the borderlines between adjacent types of constraints are diffuse: there is no fixed point of passage from one to the other." 99inally, it would be worthwhile to consider briefly the bearing this concept may have on putative cases of a translator's theological intentions.In recent years, scholars of Septuagint (and the other versions) have grown increasingly wary of attributing transformations to the deliberate (or, sometimes even subconscious) theological decisions made by translators.For example, Anneli Aejmelaeus provides guidance for claiming theological intent of the much broader textual sample, and would likely encounter much more difficult problems to solve.In all probability, more questions have been raised than have been answered, and I have only gestured in the direction of where additional analysis might lead.Nonetheless, I hope that this study is illustrative of the type of rigorous, principled analysis rendered possible by an Optimality Theory approach to Descriptive Translation Studies.
indicates, the first output candidate (a) is maximally faithful to the linearity of the phonological segments of the input /malk/.However, because it incurs a violation of the highestranked constraint, *CC#, it is eliminated from candidacy already at the point of the second column; its violation of the constraint FAVOR-Ɛ is irrelevant in this analysis.The second output candidate (b) incurs a violation of the second constraint, whereas the third candidate (c) does not; candidate (b) is fatally eliminated by this violation, and its multiple violations of the constraint IO-IDENT, which it shares with (c), do not matter.But this is not the only possible solution to the problem, since the two highest-ranked constraints in Tableau 1, *CC# and FAVOR-Ɛ, could have been reversed with little effect on the outcome, since the optimal candidate (c) violates neither.Tableaux 2 and 3 present alternative analyses.In Tableau 2, both candidates (a) and (b) incur a violation of the highest-ranked constraint and are thus definitively removed from contention before candidate (c) has incurred any violations.Ultimately, the two candidates are eliminated in the same order as in Tableau 1.In Tableau 3, similarly, candidates (a) and (b) are eliminated at the same time, since both incur a violation of the highest constraint.Candidate (c)'s violation of the next-highest constraint is irrelevant in this analysis.
In this case, an apparent hierarchy anomaly points to a grammatical phenomenon in the target language.If we run a search in a computerized database of the Targum's text for infinitives (absolute and construct) augmented by a personal possessive suffix and followed by ‫ד‬ + NOUN-DEF.ART, we find overwhelming evidence that this syntagm was non-productive in the Aramaic dialect employed by the translator of Samuel.Of 299 exemplars of infinitives from Tg. Jon. 2 Samuel, 86 none of them appears in this syntagm.87ThisThedivision between infinitives construct and absolute is an arbitrary one in Aramaic, since both have the same underlying form and the distinction is made only on the basis of the form's morphosyntactic environment.87Some infinitives construct are suffixed with a personal pronoun, when the person is the subject or, less frequently, the object of the verb (2 (γ) the prepositional phrase that serves as the indirect object of ‫רּו‬ ‫צֻ‬ ‫יָּׁ‬ ‫וַ‬ ~ ‫רּו‬ ‫צָּׁ‬ ‫:וְ‬ ), the close correspondence between the replaced Hebrew text and the Aramaic replacement is clear.Here, too, the syntax of each of these phrases (and of the clauses of which they are constituent) is standard, in both Hebrew and Aramaic.As in (α), all the lexical choices the Aramaic translator used in his replacements are cognates of their corresponding Hebrew segments, with the sole exception of ‫,חבל√‬ which was treated above in 4.1.2.In each case, the constraints IO-IDENT:SYN, IO-IDENT:SEM, IO-IDENT:MORPH, IO-IDENT:PHON, and IO-IDENT:LEX all exert control, but do not conflict with one another.