A functional profile of Left Dislocation in Biblical Hebrew

The primary aim of this paper is to provide a functional profile of verbal Left Dislocation (=LD) constructions in the Torah and Former Prophets (Genesis–2Kings). As a precursor, however, an abbreviated syntactico-semantic description of LD constructions within the aforementioned corpus is provided. This description consists of seven overlapping constructional schemas (i.e. construction types) that form a radial network around a typologically prototypical schema. Accordingly, each schematic type is more or less prototypical depending on its family resemblance (i.e. shared attributes) to the prototype. In light of this formal profile, and of the fact that grammatical constructions are equal parts form and function, the remainder and central focus of the paper will entail a detailed functional explanation of LD. This functional profile will consist of two complementary parts: 1) the central factors motivating the use of LD, and 2) the prototypical and non-prototypical discourse-pragmatic functions achieved by the construction in discourse. Together with the formal description, this functional explanation reveals a textured profile of the LD category, one that consists of a unified radial network of overlapping constructional types.


Introduction
Despite several studies 1 on the so-called Left-Dislocation (=LD) construction in Biblical Hebrew (=BH), linguists have failed to reach a consensus with respect to the construction's syntactic, semantic or discourse-pragmatic profile.As a result, a fresh investigation was undertaken (cf.Westbury 2014) with the intent of advancing a more unified explanation of verbal LD constructions in BH.Drawing on recent typological and theoretical research on the syntactic, semantic, and discourse-pragmatic attributes of LD constructions across languages, a set of criteria was developed for the identification and classification of LD constructions in BH. 2 These criteria were then applied to the corpus of Gen-2 Kngs, where over 650 instances of LD were identified. 3Further analysis revealed that these tokens reflect an abstract network of approximately seven overlapping constructional schemas (i.e.construction types) 4 where each schema is distinguished based on the clustering together of syntactico-semantic attributes and/or discourse-pragmatic functions. 5Moreover, these attributes and functions expose the degree to which each schema is related to a typologically prototypical LD schema.In other words, each constructional schema is determined to be more or less prototypical depending on its respective degree of family resemblance (i.e.shared syntactico-semantic and discourse pragmatic attributes) to the prototypical schema. 6In short, this analysis reveals a textured profile of the LD category in BH, one that consists of a unified radial network of overlapping constructional schemas.
The focus of the present paper concerns the functional profile of verbal LD constructions in the Torah and Former Prophets.However, as a precursor, I will provide a brief overview of the seven schematic constructions identified in the aforementioned study, giving special attention to their syntactico-semantic attributes. 7The remainder of the paper will concern the functional profile of LD in BH, in which I will discuss both the central factors motivating the use of LD, and the discourse-pragmatic functions achieved by the construction in discourse.

Prototypical LD
Typological evidence suggests that the prototypical LD schema consists of a constituent, typically a NP, that both precedes and is detached from the core clause with which it is 3 As an initial point of departure, this investigation assumes, in line with a Construction Grammar approach to linguistic analysis, that a grammatical construction consists of a conventional form-function pairing (cf.Hoffman and Trousdale 2013). 4A constructional schema is a generalized template that emerges from entrenched patterns of instantiated usage (Evans and Green, 2006: 754).They are achieved by removing points of difference between a variety of instantiations leaving only their common attributes.Although constructional schemas entail syntactic, semantic, and discourse-pragmatic information, from a structural standpoint, they can be thought of as generalized templates consisting of a sequence of ordered slots that may be filled by a variety of words and phrases of differing grammatical categories (Taylor 1995: 198).By contrast, an instantiation of a constructional schema is a lexically specified instance of a more abstract schema.Both constructional schemas and instantiations of schemas coexist in the grammar, the schema being an abstraction that emerges from instantiated usage patterns (Evans and Green  2006: 755). 5In full disclosure, the seven schematic construction types identified and discussed below represent a heuristic parsing of the data.In other words, where one draws the boundary lines between different construction types is somewhat arbitrary.The point is to show that the LD category in BH consists of a variety of differing construction types that are related, in varying degrees, to a typologically prototypical schema.In other words, the LD category consists of a cluster of LD schemas that form a radial continuum, where schemas that align with the prototype are located in the center of the category, while schemas that differ from the prototype occur at varying degrees of remoteness from the prototype (cf.Boas 2013). 6This is in line with Lakoff's (1987: 483-487) proposal that "regards grammar in general as a radial category, in which the schematic and more regular constructions constitute prototypes while more specific and idiosyncratic constructions are located at the periphery and inherit properties from more central instances of constructions" (Boas 2013: 245). 7In reality, form and meaning (or function) are inextricably connected such that one cannot be expressed without the other.Thus, the practice of describing a construction's syntactico-semantic profile apart from its functional profile is done for practical reasons bearing in mind that both are intricately linked and interdependent to one another.associated. 8Moreover, the referent of the detached constituent is resumed within the core clause by a co-referential element, typically in the form of either a regular or clitic pronoun, or an epithet.Approximately 20% of the tokens in the dataset met this description.The syntacticosemantic attributes that comprise the prototypical schema are as follows: 1.A referential constituent both precedes and is dislocated from a core clause with which it is associated.2.An alternative position for the dislocated constituent exits within the associated core clause.3. The alternative position is filled by an anaphoric co-referential resumptive element in the form of either a regular or clitic pronominal, or an epithet.In other words, the referents of the dislocate and resumptive element stand in a total identity relation to one another 4.The dislocated constituent is accompanied by a separate intonation contour Moreover, an example of the prototype is found in Gen. 2:17: (1) Gen. 2.179 ["From the tree of the knowledge of good and evil]i, you shall not eat from iti, for on the day you eat from it, you shall surely die"
Despite the lack of an overt co-referential resumptive, the dislocate in the Non-resumptive LD schema is often semantically or pragmatically related to another intra-clausal element, or to the clause as a whole.These semantic/pragmatic relations are characterized as: 1) a metonymic identity relation 11 (as in [2]) a partial identity relation 12 (as in [3]) a pragmatic relevance relation 13 (as in [4]) (cf.Tizón-Couto 2012: 38-39, Westbury 2014: 104-105  11 A metonymic relation is one in which the referent of the dislocated constituent stands in either a hypernemic relation (e.g.whole/part) or hyponymic relation (e.g.part/whole) with another referent of a constituent within the core clause.Often the clause internal element with which the dislocated constituent is metonymically linked takes the form of a NP modified by a possessive determiner.Unlike the prototypical schema, the referent of the determiner is co-indexed with the dislocate, rather than the referent of the NP. 12 A partial identity relation entails a partial overlap in referents denoted by the dislocated constituent and another clause internal element, respectively. 13A relevance relation is one in which a direct semantic link, either metonymic, partial, or otherwise, does not hold between the dislocated constituent and the core clause.The referent of the dislocated constituent in these constructions rather stands in a relevance relation to the ensuing proposition.In other words, the referent of the dislocated constituent functions as a framing device (cf.Fillmore 1982; Croft and Cruse 2004: 7-32) that constrains the interpretation of the following proposition to a certain semantic domain. 14Further examples of Non-resumptive LD include: Gen. 9.6; 17.15; 21.12; 34.8; 44.9; Exod.4.9; 9.6; 9.21; 26.12;   30.33; 30.38;Lev.11.33; 13.45; 15.18; 20.10; 20.12; 20.13; 25.33; 26.36; Num.3.46-47; 21.8; 22.11; 23.3; 34.6;  35.8;Deut.21.3; 28.54; 28.56; Josh.21.40; Judg.11.31; 19.30 1 Sam.2.10; 3.11; 11.7; 11.11; 20.4; 25.27; 2 Sam.2.23; 21.5-6; 24.3; 24.17; 1 Kng.8.41; 11.26; 17.20; 2 Kng.11.7; 16.14.Moreover, like with the prototype, it is possible for LDs of this type to occur embedded in a larger clause.Cf. 1 Sam.3.11, and 2 Kings 21.12, note that each subordinate dislocate is headed by the quantifier ‫ָל‬ ‫.כּ‬ 15 Note that in (4)-( 5) the dislocated status of the initial constituent is established both by the lack of an available valency slot, as well the position of the dislocate preceding the clause initial conjunctive ‫.ו‬ Also included in this schema are constructions that do not possess a semantic or pragmatic relation to another overt clause-internal element.In these constructions, the dislocated status of the initial constituent is confirmed solely on the basis of its position before a grammatical clause boundary marker.In (5), for example, the clause boundary position (i.e. the so-called CPposition) is filled by the interrogative particle ‫ה‬ ‫. מֶ‬ Likewise, in (6)

Pronominal LD
The pronominal LD schema is distinguished by a dislocated pronoun, which may be either strong or weak (i.e.clitic).The dislocated status of the initial pronoun is confirmed by either a clause-internal co-referential pronoun or epithet, and/or a clause boundary marker (e.g. a conjunction ‫,ו‬ etc.) at the juncture between the dislocate and the matrix clause.These constructions make up roughly 2% of the dataset.Furthermore, Pronominal LDs, as we will see below, are always used to accomplish a highly non-prototypical function.As a result, this schema is positioned further still from the prototype.Example ( 9 While all constructional schemas consist of generalized features by definition, cognitive linguists and construction grammarians have long observed that constructions tend to occur on a continuum between those that are more schematic (e.g.abstract) and those that are more substantive (cf.Croft and Cruse, 2004:225-247, Goldberg, 2013:15-31).More schematic constructions consist of abstracted generalized templates where sequential open slots may be filled by a variety of different words or phrases (cf.Taylor, 1995:198).By contrast, constructions that are more substantive entail fixed components that occur in the same slot in every instance.The Prototypical, Non-resumptive, and Pronominal schemas described above are all fairly schematic.20However, the final three schemas that emerged from the data are more substantive, with each construction type possessing one or more fixed elements.Moreover, as we will see at the end of this paper, these "substantive schemas" have also conventionalized various aspects of their functional profile to the degree they have become virtually formulaic in their interpretation.

Anaphoric ‫ֵן‬ ‫כּ‬
The first substantive construction is characterized by a dislocated PP or NP (rare) and is coindexed to an anaphoric adverbial ‫ן‬ ֵ ‫כּ‬ in a fronted position within the associated matrix clause.Moreover, in all but two instances, the dislocate is headed by either ‫ר‬ ‫שֶׁ‬ ֲ ‫א‬ ַ ‫כּ‬ or ‫.כְּ‬This schema comprises approximately 7% of the dataset, and is represented by example (10) below.
(10) 2 Sam.9.11 21 ‫נִ‬ ‫ֶ�׃‬ ‫ל‬ ֽ ‫מֶּ‬ ‫הַ‬ "Then Ziba said to the king, ['According to all that my lord the king commands his servant]i, soi your servant will do.'So Mephibosheth ate at David's table as one of the king's sons."

Conditional LD
A second substantive LD schema consists of those constructions that contribute to the formulation of a conditional clause.These constructions are characterized by a dislocated NP or indefinite pronoun that precedes the protasis of a conditional clause.The dislocated status of the initial constituent is confirmed by ‫י‬ ‫כִּ‬ (or more rarely ‫ם‬ ‫אִ‬ or ‫)ו‬ marking the beginning of the protasis clause.The dislocate may or may not be semantically linked to a correlative overt element in the protasis or apodosis of the conditional clause.Furthermore, in virtually every instance, tokens of this type occur in legal contexts as casuistic precepts.Conditional LDs make up approximately 5% of the dataset.Example (11) represents the Conditional LD schema.
(11) Num.5.6 22 ‫֥פ‬ ‫נֶּ‬ ‫הַ‬ ‫וא‬ ֽ ‫הִ‬ ‫׃‬ "Speak to the Israelites and say, ["A man or a woman]i, if they commit any of the sins that people commit by breaking faith with Yahweh, that personi is guilty."
Furthermore, the cognitive motivations identified will provide the basis for explaining the communicative goals and pragmatic effects achieved by the use of LD in particular contexts.As with the syntactico-semantic profile sketched above, the functional profile of LD also reflects prototypically effects.Discourse-pragmatic functions are construed as either more or less prototypical depending on the degree to which they overlap with a prototypical function.
According to Lambrecht, the PSRR dictates that a theoretical distinction be drawn between the grammatical strategies employed in coding a referent as a topic expression in a sentence (ibid.: 186-187).This distinction results in two types of topic expressions: Reference-oriented, and Role-oriented.Reference-oriented topic expressions merely serve to designate a topic referent, prototypically by means of a nominal phrase, while on the other hand, role-oriented topic expressions designate the topic referent anaphorically or deictically by way of a pronominal expression (ibid.: 186-187).Furthermore, while the semantic role of reference-oriented topic expressions is not discernable from their form or position in a sentence, role-oriented expressions serve as "grammatical links between the topic referent and the proposition by indicating the semantic role of the referent as an argument, i.e. as a participant in the action, event or state expressed by the proposition" (ibid.: 187).With respect to the LD construction in particular, the dislocated constituent prototypically serves to activate the referent about which the following proposition conveys some new information.Lambrecht, therefore, refers to the dislocated constituent as a referent-oriented topic-expression.The resumptive element, in turn, serves as a roleoriented topic expression that satisfies an argument position in the clause proper (ibid.).While the cognitive constraints that constitute the PSRR are universal, their grammatical manifestation in individual languages is subject to typological variation. 25Cf.Westbury (2016)  (1993), represented in Figure 1.The Givenness Hierarchy is a linear entailment scheme that maps particular referring expressions to corresponding cognitive statuses, such that each status entails every other status to the right on the scale.For example, a referent that is Uniquely Identifiable is also Referential, and Type Identifiable.Accordingly, this model was used as an empirical tool for plotting the accessibility status of dislocated referents in the dataset. 26This results of this analysis reveal that approximately 92% of the dislocated expressions register no higher than Uniquely Identifiable on the Givenness Hierarchy.In other words, the dislocated 26 It is worth noting, however, that although each form-status mapping is represented as a discrete category, the reality is likely much more complex.In particular, Ariel, has argued that the cognitive basis of referential forms should not be thought of as a disjunction of discrete statuses, as per the Givenness Hierarchy, but rather as overlapping spheres of accessibility that exist along a continuum (Ariel 2001: 63).Although this is most certainly more accurate, the Givenness Hierarchy nevertheless functions as a useful tool for estimating the accessibility of referents in discourse.I have therefore adopted the cognitive categories reflected in the Givenness Hierarchy for heuristic purposes, although in light of Ariel's arguments, they are re-conceptualized as non-discrete spheres along a continuum. 27The distribution of the dislocates in my dataset on the givenness hierarchy are as follows: focus (0%), activated (4%), familiar (4%), uniquely identifiable (51%), referential (26%), type identifiable (15%).

Type Identifiable:
The addressee is able to access a representation of the type of object described by the expression.

Referential:
The addressee is able to access a representation of the type of object described by the expression.
Uniquely Identifiable: The addressee can identify the speaker's intended referent on the basis of the nominal alone.

Familiar:
The addressee can uniquely identify the intended referent on the basis of an existing representation in memory.

Activated:
The referent is represented in current short-term memory.
In Focus: The referent is at the current center of attention.

Highly Accessible
Minimally Accessible expressions are generally highly informative, rigid, and unattenuated, signaling a relatively low degree of cognitive accessibility. 28wever, these results are potentially obscured by the intentional use of low accessibility markers to encode highly accessible referents, a discourse-pragmatic device referred to as "referential over-encoding."For example, although pronouns are typically used for active referents, or those at the current center of attention, such a referent may instead be encoded by a full lexical expression in order to produce a particular pragmatic implicature.Indeed, this is precisely why the Givenness Hierarchy is formulated in terms of an entailment scheme in which each form-status category entails every other category to its right.For instance, a referent with the cognitive status Activated or In Focus may be encoded by a referring expression typically used for Uniquely Identifiable referents since the former entails the later in the hierarchy.It is therefore necessary to account for false-positives resulting from referential over-encoding when attempting to assess the cognitive accessibility of a referent based solely on the type of referring expression used.
To account for false positives resulting from over-encoding, a second measure examining the source of a given referent's accessibility is applied to the dataset.In particular, discourse referents are made accessible by one of three general sources: First, they may be textually accessible by virtue of having been explicitly mentioned in the prior discourse.Second, they may be situationally accessible by virtue of their presence within the shared text-external context of the interlocutors.Or, third, they may be inferentially accessible by virtue of their implicit relation to either another entity or conceptual frame in the discourse, or their location within the long-term memory of the addressee (e.g.Sun, Moon, etc.).By accounting for the source of accessibility, a more accurate assessment of the referent's cognitive status is obtained, in addition to isolating false positive form-status pairings due to the potential for over-encoding.
This analysis reveals that approximately 25% of the dislocates analyzed are found to be textually accessible, meaning that the referent had been explicitly introduced or referred to at some previous point in the discourse.Moreover, virtually all of the referents are encoded by expressions registering as Uniquely Identifiable or lower on the Givenness Hierarchy, ostensibly indicating a low degree of accessibility.Yet, despite this use of low accessibility markers, only 20% of these referents possessed antecedents occurring at distances substantially greater than 15 clauses, while 80%, have antecedents at a relatively close proximity, within 15 clauses.This suggests that the majority of textually accessible referents represent instances of over-encoding, a point that will prove significant in explaining the discourse function of LD in BH below.
In addition to textually accessible dislocates, approximately 2% were found to be situationally accessible by virtue of their presence in the text-external context.In virtually every instance, the referent was one of the speech participants, and was therefore presumed to be highly accessible, or even active.As a result, situationally accessible referents are predominately encoded as pronouns, an accessibility marker reserved for active referents. 29 28 For further discussion of this analysis, along with examples and percentages for each category, see Westbury   (2014: 286-293). 29By contrast, the dislocate in 1 Sam.25.27 is a situationally accessible inanimate object and arguably entertains a low degree of accessibility.For further discussion of this case, see Westbury (2014: 304).
Finally, approximately 73% of the dislocates were found to be accessible by virtue of an inference relation prompted by one or more contextual factors.An overwhelming majority of these (83%) were determined to be inferable based on their membership in a so-called Partially Ordered Set Relation, while some (6%) were inferential via a conceptual frame/script (6%), or via an anchoring relation (4%).Lastly, a remaining (7%) were inferentially accessible as a result of their inherent saliency as part the general knowledge of the universe, that is knowledge pertaining to the natural, social, and cultural universe shared by the writer and addressees.For instance, some referents in BH narrative (e.g.Moses, Abraham, Sun, Moon, etc.) are assumed to be accessible due to their salient position within the long-term memory store of the reader. 30hile it is difficult to specify the precise degree of accessibility of inferred referents, it is reasonable to assume that they entertain lower degrees of accessibility than entities explicitly mentioned in the preceding discourse (textual) or referents that are accessible from the textexternal context (situational).This assumption is supported by the tendency for inferred entities to possess looser cohesive links to their triggers, therefore ostensibly demanding a greater degree of processing cost to activate them.For instance, inferred referents require the addressee to accomplish two separate cognitive tasks simultaneously: that of identifying or creating a cognitive representation, and that of accessing that representation. 31e results of measure described above are summarized in Figure 2, below: 1. Textual: The referent has been explicitly mentioned in the prior discourse.25% (Accessibility Marker ≤ Uniquely Identifiable) 80% Over-encoded (antecedent < 15 clauses) 20% Low Accessibility (antecedent > 15 clauses) 32 2. Situational: The referent exists within the text-external context shared by the interlocutors.2% (Accessibility Marker ≤ Uniquely Identifiable) 30 It is important to keep in mind that ancient BH narratives recount the lives of famous and infamous men and women of Israel's past, in addition to famous locations and artifacts of Israel's heritage.These figures, places and artifacts are often assumed by the writers to be identifiable due to the universe of discourse shared by the writer and reader, or that there existed only one notable referent which could be denoted by these NPs, e.g.Abraham, Moses, David etc.The reader is assumed to know these unique referents and therefore they need not be introduced every time they appear in a new story.Rather, they are considered inactive referents stored in the long-term memory of the reader (Heimerdinger 1999: 133-134). 31It is conceivable, as Ariel (2001: 33) points out, that some inferred referents may entertain higher degrees of accessibility than some textually accessible referents, depending on the context.This is perhaps most likely with referents that are a permanent fixture of the natural, social, or cultural universe.Nevertheless, inferentially identifiable/accessible entities, on average, entertain a relatively low degree of accessibility as compared to other recently mentioned referents, or those immediately accessible from the text-external context.Furthermore, their low degree of identifiability/accessibility is substantiated by the types of expressions (i.e.accessibility markers) used to encode these referents.In virtually every instance, the inferentially accessible referents are encoded by accessibility markers that rank no higher than Uniquely Identifiable on the Givenness Hierarchy. 32Moreover, these referents were often separated from their antecedent by one or more discourse boundaries (e.g.chapter breaks, paragraph breaks, or even book boundaries).

Inferential:
The referent is implicitly related to either another entity(s) or conceptual frame explicitly evoked in the discourse, or its location within the long-term memory of the addressee.73% (Accessibility Marker ≤ Uniquely Identifiable) 83% Inferential via partially ordered set relation 6% Inferential via conceptual frame/script 4% Inferential via anchoring relation 7% Inferential via long-term memory store Figure 2: Source of Accessibility In short, these two measures-the type of referring expression used and the source of referent's accessibility-suggest that the dislocates in the dataset generally entertain low degrees of accessibility.This evidence supports the aforementioned hypothesis, that the use of LD in BH is iconically motivated by the need to talk about a referent which the writer assumes is identifiable, but only minimally accessible to the addressee.Moreover, this constraint provides the key point of departure for discussing the overall communicative goals and pragmatic effects achieved by the construction in BH discourse.

The Core Function of LD: (Re)Activation of Minimally Accessible Referents
In light of the constraint on accessibility just outlined, I proposed that, at its most basic level, LD in BH is a discursive strategy used to (re)activate referents or propositions which are assumed to entertain a low degree of accessibility in the mind of the addressee.This function is motivated by the iconic form of the LD construction which, as stated above, entails a uniquely partitioned structure which facilitates the processing of two cognitively costly tasks: that of activating a minimally accessible referent and interpreting that referent's role within the proposition.
While its core discursive function is that of (re)activation, LD constructions that accomplish this function also fall into one of three sentence-level pragmatic categories.These are: Topic Announcing LD, Focus Announcing LD, and Framing LD.To which category a particular token belongs depends on the resumptive or otherwise semantically linked element's pragmatic relation to the proposition.I will discuss each of these categories in turn.

Topic Announcing
A topic Announcing LD construction is one in which the newly (re)activated referent fulfills a topic relation within the associated proposition.Research shows that sentences with active topic referents are easier to process (cf.Lambrecht 1994: 165, Givón 1983: 17; inter alia).As a result, writers employ LD constructions in order to (re)activate an inactive referent with a low degrees of accessibility, thereby ratifying the referent as an acceptable topic for the associated proposition.Moreover, the (re)activation + topic announcing function is widely held to be the typologically prototypical function for LD, and therefore serves as the exemplar function for this analysis. 33rthermore, Topic Announcing LDs may involve either primary or secondary topic relations.Primary topics usually coincide with the syntactic subject, while secondary topics tend to coincide with the direct object.The examples in ( 13) and ( 14) illustrate each of these types.

Focus Announcing
It addition to topics, the dislocated constituent also frequently announces the focus of the following proposition.That is to say that the resumptive element satisfies a focal relation to the proposition, rather than a topical one.Although not uncommon in BH, Focus Announcing LDs represent a less prototypical discourse function than that of topic-announcing LDs. 37evertheless, Focus Announcing constructions are also motivated by the PSRR.Rather than ratifying a topic referent, however, Focus Announcing LD mitigates the processing cost associated with activating a referent and interpreting that referent's role within the propositional assertion.
Furthermore, Lambrecht and others have shown that, across languages, focal constituents often carry a prosodic accent (1994: 218ff).Along these lines, Khan has proposed an additional motivating constraint for the use of Focus Announcing LDs.When the focal constituent is long it prohibits a speaker/writer from evenly distributing the stressed accent over the constituent (Khan 1988: 94).Thus, an LD construction is employed so that the shorter resumptive expression can carry the accent by proxy.Consider the example from Deut.20.20 in (15)  below: 38 (15) Deut.20.20 39 Only the trees that you know are not used for food]i, themi you may destroy and cut so that you may build siege works against the city that is waging war against you, until it falls." The dislocated referent "TREES NOT USED FOR FOOD" supplies the missing argument in the presupposed open proposition inferred from v.19: "you may cut down X".However, the low degree of accessibility entertained by the dislocate triggers the use of a long and complex NP, thus prohibiting the even distribution of a focal accent.By using a Focus Announcing LD, the author is able to mitigate two cognitively costly tasks.The intended referent is activated by the dislocated constituent, after which the fronted resumptive pronoun ‫תוֹ‬ ‫אֹ‬ marks it as the focus of the open proposition.39 F40

Framing LD
A common trait of both Topic and Focus announcing LDs is that the resumptive is co-referential with the dislocate in every instance.However, a third function exists in which this is not the case.Framing LDs accomplish the basic level function of (re)activation, but lack a coreferential resumptive within the core clause.It is sometimes the case that the dislocate is pragmatically related to the following proposition by way of a particular relevance relation.As such, the dislocate functions to restrict the interpretation of the following proposition to a certain semantic domain.The example in ( 16), from 1 Sam.2.13, is a representative of this type: ["Now, the custom of the priests], [anyone offering a sacrifice], the priests servant would come while the meat was boiling with a three-pronged fork in his hand." The LD construction in this example consists of two dislocated NPs, each with a topic framing function.The first dislocate activates a referent that provides a general semantic frame (i.e.priestly customs), that constrains the interpretation of what follows.In turn, the second dislocate, likewise, activates a type identifiable referent that provides a narrower semantic frame within the more general one that precedes (i.e. the event of offering a sacrifice).Both dislocates, then, exhibit a framing function: the first is more general, while the second is more narrow.Together, they function to constrain the interpretation of the following proposition(s) to the relevant domains.
In other tokens, Framing LDs not only constrain the interpretation of what follows, they also establish a pragmatic anchoring relation that facilitates the easy activation of a semantically linked element within the clause.Unlike the previous example, the coherence relation between the dislocate and the linked element within the clause is one of partial identity or metonymy.Moreover, the linked element may function in either a topic or focal relation to the pragmatically structured proposition.For example, consider the instance of LD in Gen. 9.6, represented in example ( 17 The dislocate and the clause-internal topic expression stand in a metonymic (i.e. a whole-part) relation as a result of the framing function accomplished by the LD construction.In other words, the referent HIS BLOOD is an acceptable topic expression because it is semantically anchored to the newly activated dislocate "WHOEVER SHEDS THE BLOOD OF MAN". 43

Activation of a New Attribute/Profile
Earlier I observed that some of the dislocates in the dataset exhibit over-encoding.That is, a low accessibility marker is used to encode a referent that is highly accessible.Many of these expressions involve a referential phrase modified by a restrictive relative clause which functions to predicate some new attribute to the referent.The use of LD in such cases derives not from the need to (re)active the referent, but rather, by extension, from the need to activate a new conceptual profile, according to which the referent is recast.A good example of this is found in Gen. 3.12, represented in example ( 18).The referent THE WOMAN has already been established as a salient participate in the discourse and was recently mentioned in v.8.This, along with the fact that the modifying relative clause whom you gave to be with me is semantically non-restrictive, indicates that this referent is highly accessible, despite the highly informative expression used to refer to her.Therefore, a Focus Announcing LD (cf.§4.1.2) is used in this context, not to reactivate the referent, but rather to activate a new conceptual profile.In short, by recasting the woman against this new conceptual profile, the man is able to implicitly implicate Yahweh as the one culpable for his own actions.

Less-Prototypical Functions
The functional profile describe thus far has consisted of constructions that are directly motived by the PSRR.That is to say, that they all involve the (re)activation of a referent, or attribute thereof, that is assumed to entertain a low degree of accessibility.Alternatively, however, some constructions involve the dislocation of referents that entertain a high degree of accessibility.In these cases, expectations associated with the PSRR are exploited in order to disrupt cognitive processing.In other words, where, under prototypical discourse conditions, the use of LD serves to facilitate cognitive-processing, the opposite effect is created in contexts where the processing 42 Further examples include: Gen. 17.15; Exod. 9.6;26.12;Lev. 11.33;26.36 (secondary topic);1 Sam. 3.11. 43 I interpret the first of the two fronted constituents (i.e.‫ם‬ ‫דָ‬ ‫בּאָ‬ / "by a man") as a marked focal constituent.costs are marginal. 44Moreover, disruption in processing serves as an ostensive stimulus triggering additional pragmatic implicatures, such as: 1) marking a development shift in the discourse, 2) marking a contrastive/comparative inference, 45 or 3) foregrounding the associated proposition.Furthermore, these non-prototypical functions represent a more advanced stage on the grammaticalization cline as LD begins to gradually overlap, both formally and functionally, with other construction categories (e.g.fronting).

Marking a Thematic Shift in the Discourse
The unexpected dislocation of highly accessible referents is sometimes used to signal a thematic shift in the discourse.This is the case, for example, in Lev.13.45, represented  Verses 42-44 introduce a person unclean from a grievous skin disease.Because of the referent's active status, there is no need to (re)activate it in v.45.However, the writer chose to use both dislocation and over-encoding, referring to the active participant with a Definite NP modified by a restrictive relative clause: the person who is inflected with a skin disease.This effectively disrupts cognitive processing, thereby signaling a discontinuity in the flow of discourse.Moreover, this discontinuity serves as ostensive stimuli, triggering a specific implicature in this context, namely that a higher level development shift has occurred.In particular, there is a shift from the initial introduction of the diseased person and declaration of their uncleanness, in vv.42-44, to instructions on how they are to behave, in vv.45-46.

Marking a Comparative/Contrastive Inference
The over-use of LD may also serve to enhance a contrastive or comparative relation between propositions that would have otherwise been left implicit.The relations of contrast and comparison are typically not encoded by any one lexico-grammatical feature, but are rather pragmatically inferred on the basis of contextual assumptions drawn from the content of two or more juxtaposed propositions (cf.Blakemore 2002: 89-148, Vallduvi and Vilkuna 1998).Nevertheless, these implicit relations are often made more explicit by the use of connectives, and word order devices.For instance, a speaker or writer may dislocate a highly accessible/active referent in a particular context in order to pragmatically enhance the otherwise implicit contrast between two entities or propositions. 47The over-use of LD in Deut.18.14b,  represented in (20) The contrastive relation between the nations-characterized as those who listen to those who practice witchcraft or divination (v.14a)-and the Israelites, who are prohibited from such activities is contextually inferable from the content and juxtaposition of the two propositions in v.14.Indeed, if one rereads v.14b without the dislocated pronoun, the contrastive relation is still inferable, albeit not as conspicuous.Dislocating the second person plural pronoun you, an expression reserved for active referents, creates a discontinuity in the flow of discourse.This, in turn, has the effect of enriching the implicit contrastive relation between the activities of the nations, and those prohibited by the Israelites.

Foregrounding Salient Information
The third non-prototypical function pertains to the over-use of LD as a "foregrounding mechanism" (cf.Heimerdinger 1999: 223-225) used to draw attention to information that the writer considers especially salient in the immediate discourse context.This function derives from the same exploitative processes involved in the over-use of LD to mark thematic shifts or to inflate contrastive/comparative implicatures.In these contexts, however, the unexpected disruption in the flow of discourse functions to pragmatically highlight the propositional information following the dislocated constituent. 49This explains the use of LD in Exod.4.9, represented in (21): 47 It is worth noting that this function (i.e.marking a comparative/contrastive inference) represents a prototypical function of fronting constructions across languages (Erteschik-Shir 2007).Therefore, the fact that LD is also used to accomplish this function, albeit non-prototypically, showcases the advancement of LD along the grammaticalization cline, as it gradually functionally overlaps with fronting.See also Andrason (2016a). 48Further examples include: (contrast) Gen. 4.22,21.13,22.24,47.21,Lev. 14.6;(comparison) Gen. 6.21;17.17;19.38;21.13;22.24;24.27;4.22;Lev. 25.44;Deut. 12.22;28.56;Judg. 11.24;1 Sam. 13.2. 49 For additional examples of this function in Neo-Aramaic, see Khan (2016).
Consistent with its structure (see §2.7), the functional profile of Temporal LDs is highly nonprototypical.In particular, Temporal LDs generally function to mark a discontinuity between mainline events in a narrative by activating a "temporal stage topic"-a temporal expression that specifies or updates the reference time 56 of a scene or episode (cf.Van der Merwe et al. 1999;  Floor 2004). 57The updating of the reference time may occur at the onset of a scene or an episode, a development within an episode, or the concluding scene of an episode.For example, in 2 Kng.4.11, represented in (24) It is also somewhat common in BH for a temporal expression to occur in a fronted position within the clause, without being detached from it.Fronted temporal expressions are distinguished from dislocated expressions by the lack of an intervening ‫ו‬ between the temporal expression and the matrix clause.Indeed, this typically constitutes the only syntactic difference between the two constructions.
However, the precise functional difference between fronted and dislocated temporal constructions is still in need of further research.Nevertheless, I hypothesized that the LD construction functions to signal a more marked discontinuity in the discourse than its fronted counterpart.Additionally, the Temporal LDs may result from the choice to use a wayyiqtol/weqatal verb form as opposed to qatal/yiqtol in particular contexts.Accordingly, the LD construction would then be a byproduct of the semantic and pragmatic motivations associated with the use a particular verb in context, rather than being motivated by the discourse-pragmatic function of the construction as a whole.However, further research is needed in this regard.
Lastly, the dislocation of temporal expressions also functions as a pragmatic foregrounding device.This generally happens when the content at the end of a proposition is immediately repeated in the form of a dislocated temporal expression at the beginning of the following proposition.The result is a cohesive device referred to in the literature as "Tail-Head Linkage" (cf.Hengeveld and Mackenzie 2008: 3, de Vries 2005, Levinsohn 2000: 200ff).The over-use of LD in these contexts slows down the processing of the discourse, while reinforcing continuity 56 "Reference Time refers to that point in time that provides one with a vantage point from which an event is viewed" (Van der Merwe et al. 1999: 94).This is distinguished from "Event Time" (i.e. the time of the event itself), and "Speech Time" (i.e. when reference to an event is made) (ibid.). 57In his study on the formal and functional profile of ‫י‬ ‫הִ‬ ‫ַיְ‬ ‫ו‬ in 1 Samuel, Van der Merwe et al. (1999: 112) writes, "...breaking the wayyiqtol sequence in biblical Hebrew is normally associated with some type of discontinuity between events on the time-line." 58Further examples include: Gen. 6.21; 22.21; 27.1; 38.1; 48.1; Josh.1.1; 5.1; Judg.1.1; 5.1; 5.13; 9.1; 11.1; 23.1;   24.29; Judg.1.1; 6.25; 2 Sam.1.1; 10.1; 15.1; 2 Sam.11.1; 1 Kng.9.1; 18.1; 2 Kng.2.1; 6.24.For tokens that signify a development within an episode, see: Gen. 26.8; 26.32; 30.25; 31.10;38.24; 38.27; Exod.16.13; Judg.3.27; 8.33; 16.4; 1 Sam.16.6; 18.6; 20.35; 2 Sam.2.21; 8.1; 11.14; 15.7; 17.21; 21.18; 31.8; 1 Kng.13.23; 18.36.For those that signify the concluding scene of an episode, see.Gen. 39.7; 40.20;Deut.31.24; 2 Kng.19.35.with what precedes.This has the effect of throwing the subsequent proposition into sharp relief. 59 ‫ה׃‬ ֽ ‫פֶ‬ ‫סָּ‬ ‫ן-תִּ‬ ‫פֶּ‬ "But he lingered.So the men seized him and his wife and his two daughters by the hand, because of Yahweh's compassion for him; and they brought him out and set him outside the city.17 [And as they brought them out], one said: Escape for your life!Do not look back or stop anywhere in the valley.Escape to the hills or you will be swept away." Despite the angel's warning in v.15, Lot resists leaving Sodom, and therefore, he and his family are seized by the visitors and brought outside the city (v.16).The initial dislocate beginning v.17 is preceded by ‫י‬ ‫הִ‬ ‫ַיְ‬ ‫ו‬ and repeats the final proposition of v.16 by way of a temporal expression headed by ‫.כְּ‬This updates the reference time of the subsequent event as one of immediate temporal proximity with what precedes, while maintaining continuity between the mainline events of the scene.Moreover, the use of tail-head linkage slows down the processing of the discourse which has the effect of foregrounding, or highlighting the following propositional event.In this case, it is the insistent urging by the angel that Lot and his family flee the city at once.

Conclusion
By way of conclusion, the functional profile presented here provides a typologically informed explanation of the LD construction in BH.This profile accounts for both the cognitive and contextual motivations for the use of LD in BH, in addition to the prototypical and less prototypical communicative goals and pragmatic effects achieved by the construction in discourse.
In short, LD in BH, as in other languages, is iconically motivated by the need to talk about or assert a referent that entertains a low degree of cognitive accessibility.In light of this constraint, it is evident that the construction principally functions as a discursive strategy for (re)activating inaccessible referents, or attributes thereof.These referents may function as the pragmatic topic or focus of the associated proposition, or they may serve as a semantic frame according to which the proposition is understood.Finally, LD may be used in non-prototypical contexts where the dislocate is already highly accessible or even active.In these cases, expectations associated with the prototypical use of LD are exploited in order to disrupt cognitive processing, thereby 59 Van der Merwe, et al. (forthcoming) avers that, in cases where the temporal proximity of two events on the time-line is involved, the ‫י‬ ‫הִ‬ ‫ַיְ‬ ‫ו‬ preceding the dislocate functions to minimize the inherent discontinuity triggered by the dislocated temporal expression, thus allowing speakers to "describe the immediate temporal proximity of two events, without breaking the mainstream of events in the scene."Furthermore, these authors note that "these constructions are often used at the climax of a scene in order to signal what triggered a climactic event." 60Further examples include: Gen. 24.30; 39.13; 39.15; 39.19; Num.11.25; Josh.6.20; Judg.11.35; 1 Sam.4.18;   1 Kng.14.6; 15.21; 16.18; 19.13; 21.16.triggering additional pragmatic implicatures.Lastly, these non-prototypical functions represent the outer fuzzy edges of the LD category and thus a more advanced stage on the grammaticalization cline, where LD begins to overlap with other constructional categories.

"Whoever is found with it from your servants], he
must die, and we also will be my lord's servants."

But every slave that is bought for money]i, [after you have circumcised him]j, thenj, he
. may eat of it." Blessed be Yahweh, the God of my master Abraham, who has not forsaken his steadfast love and his faithfulness toward my master.
[As for me]i, Yahweh has led mei in the way to the house of my master's relatives." in focus → activated → familiar → uniquely → referential → type

Whoever sheds the blood of man], by
a man shall his blood be shed, for God made man in his own image."