Is Biblical Hebrew a non-configurational language? Reconsidering the evidence from discontinuous phrases

: The occurrence of discontinuous phrases is an important feature of so-called non-configurational languages, which are characterized by relatively free word order in clauses and phrases. Partly based on this feature, it has been argued that Biblical Hebrew is a non-configurational language. However, after reviewing broader linguistic literature on this topic, the present article shows that the seemingly discontinuous phrases in Biblical Hebrew are not comparable to the typically non-configurational types of discontinuous phrases, and thus need a different explanation. The suggested alternative analyses are afterthought (or right dislocation) and extraposition. Since these constructions occur in languages at various positions on the configurationality continuum, they should not be interpreted as evidence for the assumed non-configurational nature of Biblical Hebrew.


‫ס‬ ִֽ
‫ח‬ ‫ינְ‬ ִֽ ‫ּופ‬ ‫ִ֖י‬ ‫נ‬ ‫פְ‬ ‫ח‬ ‫תּו‬ ֵ֔ ‫מ‬ ‫י‬ ‫ל‬ ‫ִֽי-ע‬ ‫נ‬ ‫בְ‬ ‫ֵ֤י‬ ‫נ‬ ְ ‫ּושׁ‬ This feature, along with other evidence, has led some of them to argue that Biblical Hebrew could be considered a "non-configurational" language, characterized by relatively free order of words and constituents (cf.Andersen and Forbes 2012:87-93).The present article investigates this claim, focusing on the analysis of discontinuous phrases while incorporating insights from research on other languages, for it could be questioned whether the seemingly discontinuous phrases in Biblical Hebrew are in fact comparable to the discontinuous constructions in typical non-configurational languages or whether they merit a different analysis.
In an attempt to answer this question, the article first reviews some general linguistic and typological literature to get a better understanding of non-configurationality and, especially, the interpretation of apparent and real discontinuous phrases (section 2.1).Alternative analyses for seemingly discontinuous phrases are also considered (section 2.2), resulting in a more specific definition of true discontinuous phrases (section 2.3).Subsequently, this is applied to Biblical Hebrew by summarizing and assessing the arguments from the literature pro and contra its nonconfigurational nature (section 3.1 and 3.2), and comparing the Biblical Hebrew data to the discontinuous constructions described in the broader literature.It is contended that the seemingly discontinuous phrases in Biblical Hebrew should not be interpreted as an obvious sign of non-configurationality, because these phrases are generally different from the appositional type of discontinuous "phrases" that occurs in prototypical non-configurational languages as well as from the "true" type of discontinuous phrases according to the previously established narrow definition (section 3.2).Instead, they could be analysed as either some sort of afterthought (or right dislocation), i.e., an independent phrase that occurs in extra-clausal position to denote specific information structure values, or as extraposition, which places a long or complex constituent at the end of the sentence to facilitate processing (section 3.3).Finally, the conclusion (section 4) mitigates the claim for non-configurationality in Biblical Hebrew, and looks ahead to a following article that will further explore the different types and possible analyses of seemingly discontinuous phrases in Biblical Hebrew, while also acknowledging the limitations of this research due to the lack of comprehensive prosodic information in the study of a dead language.

Non-configurationality in general and typological linguistics
The occurrence of discontinuous phrases is one of the most prominent characteristics of socalled "non-configurational languages" or "flexible-order languages," of which the most prototypical ones featuring in the literature are certain Australian languages such as Warlpiri (Hale 1983) and Kalkatungu (Blake 1983;cited by Rijkhoff 2002:19,20,255,256).As opposed to "configurational languages" or "rigid-order languages," which could be seen as the more standard type, such languages are said to be characterized by a flatter and less hierarchical structure, which could manifest various properties, like free word order (both within the clause and within the phrase), discontinuous constituents (often accompanied by rich case marking), and frequent use of so-called "null anaphora" or "pronoun dropping" (often coinciding with complex verb forms), as mentioned, e.g., in articles by Hale (1982:86,87;1983:5), to which Gil (1983:146) and Andersen and Forbes (2012:88) refer, applying them to Hebrew 1 .In the discussion of configurationality and how this relates to Biblical Hebrew in the rest of this article, the main focus lies on non-configurational structures as manifested within discontinuous phrases.

The distinction between integral an non-integral phrases and its relation to discontinuity
The feature of discontinuous constituents in combination with the lack of hierarchical phrase structure is related to the distinction between so-called "integral" and "non-integral" phrases (terminology that predominantly occurs in the treatment of NPs, but is probably also applicable to PPs 2 ), such that configurational languages have integral, hierarchical phrases, whereas non-1 Note that the first two of the mentioned properties (i.e., free word order and discontinuous constituents) correspond to the "three parameters relating to ordering" distinguished by Dixon (2010:72), which describe languages according to the opposition between fixed and free order on different levels, viz.: the order of words within a phrase, the order of phrases within a clause, and the order of words within a clause.Dixon asserts the importance of keeping those parameters separate, such that languages that only have free ordering according to the second parameter, should not be called "free word order languages," as is often, confusingly, done, but should rather be referred to as having "free constituent order."The terminology of "free word order" should then be reserved for those languages that, in addition to having free order of words within the phrase and of constituents within the clause, also allow the words of a phrase to occur separately from each other at various positions in the clause and thus have free ordering of words within the clause. 2The abbreviation "NP" is used in the present article, as well as in much of the cited literature, for "noun phrase," while "PP" is used for "prepositional phrase."In some of the cited literature, the abbreviation "DP" for "determiner phrase" is used instead of NP, since generative linguists regard the determiner rather than the noun as head of the phrase.
configurational languages have non-integral, flat phrases, though they may have integral phrases as well 3 .As Rijkhoff (2002:19-23) and Krasnoukhova (2012:168,169) explain, in integral noun phrases, the modifiers are dependent on the head noun both semantically and syntactically and are thus fully integrated within a hierarchically structured phrase, whereas in non-integral noun phrases, the modifiers stand in apposition to the head noun on which they are only semantically but not syntactically dependent and the noun and modifying constituents can be regarded as a set of coreferential NPs consisting of independently referring elements (with separate morphological marking) 4 .
The appositional relation between these elements of a non-integral phrase can hold either within a larger phrasal structure or at sentence level.America, Krasnoukhova (2012:169,257) notes that "the possibility to form integral NPs may also vary within languages," and she argues that languages with and without integral NPs cannot be clearly distinguished, because various types of modifiers could behave differently, such that "Some languages show evidence for integral NPs with some categories of modifiers but not with others." 4 For some of the non-configurational languages, it may even be debated whether nouns and adjectives can be distinguished as separate grammatical categories (cf., e.g., Siewierska 1988:169). 5Cf., e.g., the "miniature noun phrases" in an appositional structure in Nama, that are each separately marked for person, number, and gender (whereas such marking is absent for modifiers in hierarchical NPs), and are able to occur in isolation, but together form one phrase to which a case marker may be attached (J.R. Payne 2006:717), and the "fractured phrases" in Gooniyandi, that consist of contiguous parts which are broken up by the occurrence of more than one postposition (McGregor 1989).For the Gooniyandi data as described by McGregor (1989), however, one may perhaps wonder whether this phenomenon could not be alternatively analysed as involving extra-clausal constituents that are adjacent to the co-referent clause-internal constituent, but do not form one phrase with it. 6The examples from other sources are reproduced in simplified form.While the original language data and translation are included, the gloss is only roughly rendered without detailed information on grammatical markers, etc.If not in the original, bold marking has been added to indicate the (seemingly) discontinuous phrase. 7Cf.also the constructions in various other Australian languages, e.g., the "discontinuous phrases" in Gooniyandi, which consist of non-contiguous, coreferential pieces (McGregor 1989), and the combination of nominals in Kalkatungu that are associated with one argument role, but function as individual constituents at clause level, thus resulting in various possible word orders and allowing for intervening words in between, as reported by Blake (1983:145) cited by Rijkhoff (2002:19,20,256).Furthermore, Krasnoukhova (2012:48,49,177-179) mentions several Cariban languages (Tiriyó, Hixkaryana, and Panare) in which non-integral noun phrases consisting of syntactically independent NPs in apposition, which may occur in various orders and even non-contiguously and may each have its own postposition attached to it as phrasal clitic, are possible with specific types of modifiers, viz., with (2) Wawirri kapirna pantirni yalumpu.kangaroo will spear that 'I will spear that kangaroo.'However, Rijkhoff (2002:255-257) does not consider these seemingly discontinuous NPs in non-configurational languages as real discontinuity, since, even though the discontinuous constituents are semantically related, they are individually referring expressions in apposition that do not form one integral phrase and thus, per definition, cannot be called a discontinuous phrase 8 .Similarly, according to Krasnoukhova (2012:169,182), such syntactically independent, appositional modifiers are not to be regarded as discontinuous modifiers, but rather as separate NPs in apposition to their semantic head.Another case of apparent, but not real, discontinuity, that both Rijkhoff (2002:22,255-257) and Krasnoukhova (2012:180,182) mention, consists of constituents that, judging by their meaning, seem to be discontinuously realized nominal modifiers, but that are better analysed as functioning adverbially or predicatively, and thus should, in fact, be considered as distinct, non-apposed phrases at clause level 9 .
As opposed to these two cases of "apparent discontinuity," Rijkhoff (2002:257-259) defines true discontinuous phrases as instances where "a constituent of an integral NP occurs outside its proper domain," and he cites, among others, the following examples of elements that can be realized discontinuously from the noun phrase to which they relate: contrastive adjectival modifiers in Polish (Siewierska 1984:58-60), highly emphasized numerals in Ngiti (Kutsch Lojenga 1994:336-342,354,355), and extraposed relative clauses in Dutch.He also refers to Vilkuna (1998:223-226), who describes NP discontinuity in various Uralic languages, especially providing examples of "possessor extraction."The following example from Siewierska (1984:60) shows discontinuous phrases as they occur in Polish in the specific context of double contrastiveness.It is noteworthy that the discontinuous elements often occur at marked positions, either at the beginning or at the end of the clause, which seems to be in line with the remark by Rijkhoff (2002:257) that "instances of true discontinuity can typically be attributed to some pragmatic factor, complexity, or both," where "complexity" applies to the cases of clause-final occurrence of heavy NP constituents.Similar statements are made by other authors, who also attach importance to pragmatic or discourse factors in explaining various forms of discontinuity.For example, Krasnoukhova (2012:169) mentions both pragmatic factors (e.g., focus on constituents) and structural factors (e.g., avoiding long strings of modifiers) as possible motivations for the use of non-integral noun phrases, while McGregor (1989:219,222) claims that discourse considerations must be taken into account to explain phrase discontinuity (viz., assuming that the discontinuous elements fulfil distinct textual roles), and Siewierska (1988:170) notices for several languages that the noun phrase discontinuities "appear to be motivated by discourse factors."With respect to discontinuous adjectives in spoken Russian, Miller and Weinert (1998:168) speak about a word order that "is not free but controlled by discourse factors," while according to Schultze-Berndt and Simard (2012:1016,1017,1039,1047,1051) information structure values play an important role in the occurrence of discontinuous noun phrases, and Fanselow and Féry (2006:75) even call information structure "the motor of splitting."Furthermore, Rijkhoff (2002:259) suggests that notwithstanding its relatively rare occurrence in written language and, consequently, in linguistic literature, discontinuity is probably a more common phenomenon in spoken language10 .It should be noted that phrases that are truly discontinuous according to Rijkhoff thus typically occur in non-configurational languages with integral noun phrases.Krasnoukhova (2012:168,169,177) also describes discontinuous phrases as forming one unit, thus contrasting with appositional phrases, although she regards both discontinuous phrases and appositional phrases as subtypes of non-integral phrases that are not always easily distinguishable in practical terms.This contrasts with the view of Rijkhoff, who seems to consider discontinuous phrases as a special type of integral phrases or perhaps as a separate category, but not belonging to the non-integral phrases.Krasnoukhova further clarifies that, while the parts of an appositional NP may exhibit individual phrasal marking, the pieces of a discontinuous NP that are separated by an intervening constituent do not constitute independently referring expressions, since they show some sort of dependency relation between them, which can be indicated by morphological agreement marking assigned by the head noun to the dependent modifier, e.g., for gender in the South-American language Mosetén.

Alternative explanations for alleged discontinuous phrases
Apart from the often mentioned "non-configurational" Australian languages, there is a wide variety of other languages that are described by certain authors as having discontinuous noun phrases and that are sometimes deemed non-configurational as well, partly based on that characteristic.However, this is denied by other authors, who occasionally question the evidence and/or favour a stricter application of the term "discontinuous phrases" because many of the constructions that at first sight resemble discontinuous phrases could actually be accounted for by alternative explanations and are thus not true discontinuous phrases 11 .The main arguments to reject an analysis as a single discontinuous phrase in these cases seem to be that either the separate parts do not truly belong to one phrase but are rather independently generated units within the clause, or that one of the elements is actually filling an extra-clausal position for discourse-pragmatic reasons.This leaves only a rather small amount of structures to be considered as true discontinuous phrases (cf.section 2.3).Generally, the various reasonings found in the literature regarding apparent discontinuous noun phrases could thus be subsumed under the following two recurrent types of analysis, even though the terminology used in different articles is not always in agreement12 .
First, there are the constructions in typical non-configurational languages (cf.example 2 above).As has already been mentioned with reference to Rijkhoff (2002:255-257) and Krasnoukhova (2012:169,182), the main reason to disregard discontinuous noun phrases in these languages is the explanation that the discontinuously occurring noun and modifiers actually all constitute independent phrases, which is sometimes supported by the assumption that such languages completely lack the structural level of the noun phrase (cf.Siewierska 1988:168,169;Schultze-Berndt 2008:1;Schultze-Berndt andSimard 2012:1016).There are several variants of this theory (cf.Schultze-Berndt andSimard 2012:1026;Fanselow and Féry 2006:51, and references therein), e.g., describing the discontinuous elements with referential overlap as standing in an appositional relationship (e.g., Krasnoukhova 2012:169;Rijkhoff 2002:19,22,23,256,257), or analysing them as predicative adjuncts which are modifying the possibly empty argument position of the verb and are said to be "base-generated" in terms of generative grammar (cf., e.g., Hale 1983:33) 13 .Besides the Australian languages for which these analyses are proposed, there are also languages (e.g., Asiatic languages like Korean) for which the discontinuous parts could be analysed as independently generated "free topics," which is similar in the sense that these also do not directly function as argument of the verb and are thus less limited in their number and position of occurrence (cf.Fanselow and Féry 2006:10,51,64).
Second, there are other languages that are not commonly considered prototypical examples of non-configurationality, but that nevertheless seem to exhibit (noun) phrase constructions who claims that there are no free word order languages because "A true free word order language would be a language in which, for a given set of constituents, all possible permutations of these constituents would not only be grammatical, but also communicatively equivalent to each other."Obviously, this is not true for any language since all languages exclude at least some sequences of constituents and there is usually some significant difference between coexisting sequences.Besides, in functional grammar, constituent ordering is not a deep property of languages, and, therefore, there is no deep difference between languages with relatively fixed or "free" word order, but only a difference in the amount to which placement rules carry the functional load of expressing underlying relations.
involving discontinuity, which for some authors even forms a reason to use the term "nonconfigurational," e.g., Miller and Weinert (1998:183) for spontaneous spoken Russian, and Gil (1983:150) tentatively for Modern Hebrew 14 .However, these languages predominantly feature continuous noun phrases while the appearance of phrasal discontinuity is more restricted 15 .Therefore, they cannot be claimed to lack the level of an NP constituent as in the analysis that is posited for some of the Australian languages, but an explanation for the discontinuous constituent structure in these languages should rather be sought in functional or discourse factors like, e.g., different information structure values for the separate but coreferential elements (Schultze-Berndt 2008:1;Siewierska 1988:170).Moreover, in many of these cases, the seemingly discontinuous noun phrases could still be denied the status of true discontinuous noun phrase partly because of these discourse-pragmatic reasons, for the postulation of different information structure values for the discontinuous elements is actually an argument to consider them as separate constituents, since it is impossible for a single constituent to have two distinct information structure values according to Schultze-Berndt and Simard (2012:1039) 16 .Depending on the precise characteristics of the construction, various alternative analyses are available for the discarded discontinuous phrases, of which the most important ones seem to be "afterthought," sometimes considered a type of "right dislocation" (cf.Ott and De Vries 2016:642,643), and "split topicalization" as determined by, respectively, the clause-final and the clause-initial occurrence of the element that deviates from the regular structure.In addition to the fact that one part of the construction appears in a special position with respect to information structure, there are several other observations that support an analysis of the separate parts as individual phrasal constituents.
Prosodic information appears to be an important criterion, since many accounts of seemingly discontinuous phrases notice that the dislocated part is separated from the rest of the clause by a pause, implying that the two discontinuous parts belong to separate intonation units.For example, McGregor (1997:86,107) distinguishes two types of discontinuous phrases in Gooniyandi, which he refers to as Type A and Type B, that are distinct not only semantically, i.e., with respect to function, but also formally, i.e., with respect to supra-segmental structure, thus stressing the need to include intonation in the analysis 17 .The second piece of Type B discontinuous phrases is described as constituting its own intonation unit as well as information unit, and is characterized as a type of afterthought, which is "predicated of the first [piece], and either attributes a quality of it or identifies it" functioning "to provide further information concerning something mentioned in the previous unit, and simultaneously give prominence to 14 Several of these references are restricted to certain genres or registers of the language.Besides Miller and Weinert (1998), cf. also Siewierska (1988:170), who mentions instances of NP discontinuity in Russian impromptu speech, Polish impromptu speech and literary prose, Latin poetry and prose, and colloquial Hebrew (based on Gil 1983). 15Even though Miller and Weinert (1998:182) remark about spoken Russian that "the occurrence of split NPs is not a peripheral construction that can be discounted." 16"Assigning discrete information structure values like focus and topic to the subconstituents of a single phrase … is inherently problematic … As several authors have pointed out, a modifier within a noun phrase cannot constitute a focus domain on its own."Note that complex contrastive focus constructions may seem to contradict this, but for these cases, the solution lies in distinguishing between information structure (focus or topic) and accessibility (new or given), as is illustrated by some of the true discontinuous phrases that will be discussed below (Schultze-Berndt and Simard 2012:1039,1040). 17These Type A and Type B discontinuous phrases correspond to true discontinuous phrases and afterthoughts (considered separate phrases), respectively, in the terminology of Schultze-Berndt and Simard (2012:1028,1032).that additional information" (McGregor 1997:83,86,94,101,102) 18 .The following example from Gooniyandi is analysed like this by McGregor (1997:101). (4) Wayandi jardjidi nyamani.fire we.lit.itbig 'We lit a fire, a big one.' Similarly, Schultze-Berndt and Simard (2012:1025,1026,1028) and Schultze-Berndt (2008:3) use prosodic information to distinguish afterthoughts or dislocated structures from true discontinuous phrases in the Australian language Jaminjung.Moreover, they explicitly adduce the prosodic detachment of afterthoughts as an argument to analyse them as independent NPs rather than as belonging to a (single) discontinuous noun phrase, defining them as focal nominal constituents that overlap in reference with an NP in the preceding intonation unit from which they are separated by a boundary intonation contour and/or a pause and that disambiguate, elaborate on, or correct the previous description of the referent.
The presence of a prosodic break can also be used as criterion to distinguish true discontinuous noun phrases from split topicalization, which occurs, e.g., in German and Slavic languages, and is illustrated by the following German example from Fanselow and Féry (2006:59).
( 5) Zeitungen liest er nur eine.newspapers reads he only one 'He reads only one newspaper.' According to Schultze-Berndt and Simard (2012:1047-1049), split topicalization is a construction in which the first element is a contrastive, clause-external topic, while the second element is focal and could be either discontinuous or contiguous to the first, and which could probably best be analysed as involving two semantically linked but independent NPs because of the separate intonation contours and the different information structure values 19 .This pragmatic pattern of contrastive topic on one part and focus on the next part is also described by Fanselow and Féry (2006:2,4,5,51,75).Although they treat it as a type of discontinuous phrase, they nevertheless seem to mainly support an analysis in which the two parts are generated independently, but they mention some contradicting evidence as well.Moreover, they attach much weight to prosodic differences between this type, which they call "non-cohesive split," since the discontinuous parts are realized as two intonation phrases, and another type, which they refer to as "cohesive split," since the discontinuous parts are integrated into a single intonation unit (cf.true discontinuous phrases).Furthermore, they notice that the prosodic distinction often correlates with a hierarchical one, such that non-cohesive discontinuous noun phrases tend to show inverted word order compared to the corresponding continuous noun phrase, whereas cohesive ones tend to be simple splits that preserve the regular word order 20 .
Another important criterion that could be applied to distinguish split topicalizations as well as afterthoughts from true discontinuous noun phrases, consists in the ability of the separated parts to appear as independent phrases.Schultze-Berndt and Simard (2012:1026,1027,1047) adduce this as an argument to analyse both afterthoughts and split topicalization as involving two autonomous phrases rather than one discontinuous phrase, remarking that the combination of the discontinuous parts in these constructions "would not necessarily result in a [grammatically] well-formed noun phrase," which they illustrate with an example of afterthought in Jaminjung that includes repetition of an element from the preceding coreferential NP, and an example of split topicalization in German that shows disagreement in number between the two parts (cf.example 5 above).Fanselow and Féry (2006:54-61) describe this characteristic more specifically for inverted non-cohesive discontinuous noun phrases (i.e., split topicalization) as they mention various morphological adjustments in comparison with continuous noun phrases, e.g., case marking on adjectives (in Hungarian, Georgian, Warlpiri, and Quechua), different declension of a quantifier or determiner (in German and Romanian), addition of a nominalizer (in Yucatec Maya and Cantonese), or doubling of a preposition (in Russian), all of which seem to be aimed at allowing the discontinuous parts to appear as autonomous phrases, which leads them, too, to the conclusion: "That both parts have to fulfill the constraints on independent DPs suggests they are base-generated independently of each other." 21

"True" discontinuous phrases
After filtering out the above-explained types of seemingly discontinuous phrases, there are considerably less constructions that remain to be regarded as true discontinuous noun phrases.These phrases consist of elements which have the distinguishing characteristics that, even though they are separated by one or more other constituents, they belong to a single intonation 20 While Schultze-Berndt and Simard (2012:1049,1050) indeed assume that the terms "inverted splits" and "simple splits," as used by Fanselow and Féry (2006), correspond to the "contrastive focus type of discontinuous NP" and the "split topicalization construction," respectively, in their own terminology, they deem the attention paid to this distinction in word order unnecessary because the difference between the two types could equally well, and even better, be captured on the basis of purely functional grounds. 21For Russian, Miller and Weinert (1998:164-169,176-189) similarly reach the conclusion that the apparently discontinuous phrases actually consist of independently generated phrases, but the solution they propose is slightly different.They reject an interpretation as afterthoughts because they refer to data without a pause in intonation before the presumed afterthought, and they rather posit an analysis in which the non-adjacent noun and modifier form separate phrases that are linked semantically and inflectionally and that stand in apposition, as supported by the deictic origin of long adjectives in Russian and their ability to occur as referring expressions.This terminology seems comparable to the first type of so-called discontinuous noun phrases explained above, which also complies with the fact that Miller and Weinert (1998:180-183) consider spontaneous spoken Russian to be non-configurational and compare it to Australian languages.However, it is perhaps more likely that the various Russian examples mentioned by Miller and Weinert (1998) are possibly congruous with an alternative analysis as one of the various types of seemingly or truly discontinuous phrases discussed by Schultze-Berndt and Simard (2012) and Fanselow and Féry (2006), since Schultze-Berndt and Simard (2012:1038) mention occurrence of true discontinuous NPs in contrastive discourse context in Russian, while Fanselow and Féry (2006:2,58,60,70) provide Russian examples for cohesive simple splits, i.e., true discontinuous phrases, as well as for non-cohesive inverted splits, i.e., split topicalization, but for the present work, it goes too far to sort this out in more detail.
unit, they would not normally be able to occur as independent phrases for either semantic or morphological reasons, they typically-though not necessarily-preserve the word order of the corresponding continuous noun phrase, and they have a so-called "merged" interpretation equivalent to a contiguous phrase that describes a single referent (cf.Fanselow and Féry 2006:4,5,72,75;Schultze-Berndt andSimard 2012:1032;McGregor 1997:94-100).Furthermore, these true discontinuous phrases could generally be defined not only by their prosodic and morphosyntactic properties but also by the specific discourse functions or information structure categories that they express, which could perhaps even be understood as iconically motivating the discontinuity (Schultze-Berndt and Simard 2012: 1015,1040,1049,1051;McGregor 1997:94,98,99;Schultze-Berndt 2008:8).
Thus, according to Schultze-Berndt and Simard (2012:1015,1036,1038,1039,1049), true discontinuous noun phrases, with split interrogative phrases (cf.example 6 below) as a subtype, could commonly be related to "contrastive argument focus," which is reported in the literature for a wide variety of languages, often formulated as a combination of a focal and a topical or backgrounded element, e.g., by Siewierska (1984:65-69) for Polish (cf.example 3 above).However, Schultze-Berndt and Simard (2012:1034,1039,1040,1049) argue that it is more appropriate to consider the entire discontinuous phrase as focal but including a given element; e.g., in Jaminjung, it is the modifier that contributes the contrastive semantics and has a focal prosodic contour, while the noun is given, but the entire discontinuous NP should be regarded as focal because it is needed in its completeness to identify the referent 22 .This is probably congruent with the pragmatic pattern of "narrow focus on the left part with givenness on the right part" mentioned by Fanselow and Féry (2006:5) for cohesive simple discontinuous noun phrases 23 , which they illustrate, e.g., by the following example (Fanselow and Féry 2006:70) of a discontinuous interrogative construction embedded in a prepositional phrase in Russian. (6) V kakoj on poedet gorod? to which he will.gotown 'To which town will he go?' Perhaps it is also compatible with the analysis by McGregor (1989:219;1997:83,86,94,96) for "Type A" discontinuous phrases in Gooniyandi as simultaneously fulfilling the distinct textual roles of theme (especially the first part) and unmarked focus, although it is not clear whether this completely overlaps with the contrastive discourse contexts reported for Jaminjung and other languages (cf.Schultze-Berndt and Simard 2012:1039), since McGregor 22 The crucial point in the analysis of Schultze-Berndt and Simard (2012:1033,1034,1040) is thus the distinction between "information structure role (topical or focal), on the one hand, and accessibility ("given" or "new") on the other hand," which allows for the less typical combination of "given" information with "focal status," viz., in these contrastive focus constructions that identify an accessible referent among a set of alternatives. 23Although this distribution of focus and givenness over the left and the right part is commonly found, it is not obligatory, as is demonstrated by example (3) above in which the given part precedes the contrastive-focus part.This is consistent with the comment by Fanselow and Féry (2006:4,5) about the imperfect correlation between order and type of (seemingly) discontinuous phrase, while Schultze-Berndt and Simard (2012:1049,1050) explain that the order of elements in the contrastive-focus type of discontinuous NPs may vary depending on the language-specific features of basic word order (if there is a fixed word order) and primary focus position (cf.also footnote 20 above).Furthermore, which part of the NP occupies the focus position may be variable, too, even within the same language, since, although it is usually the modifier that contributes the contrastive focus while the nominal head is given, the reserve seems possible as well (Schultze-Berndt and Simard 2012:1038,1041,1050,1053n11), as can also be observed in example (3).
( 1997:98,99,107,111n10) remarks that these discontinuous NPs do not reinforce but rather minimize contrasts, while he interprets them as a strategy to introduce new entities of intermediate and local significance or, in a broader sense, representing topic discontinuity.
Besides the more widely attested contrastive focus type, Schultze-Berndt ( 2008) and Schultze-Berndt and Simard (2012:1015,1041,1046,1050) recognize still another type of true discontinuous phrases with distinct prosodic, syntactic, and functional properties, viz., discontinuous phrases that mark a subtype of "sentence focus" or "thetic" clauses and that are used in "annuntiative" contexts to introduce a new entity with a particular quality into the discourse, of which they provide, e.g., the following example in Jaminjung (Schultze-Berndt and Simard 2012:1043).
(7) Burdaj garamngardi gujugu!wind is.coming big 'A big wind is coming!'They specifically describe this construction for Jaminjung and claim that it is not commonly attested in other literature on noun phrase discontinuity, at least not as explicitly associated with sentence focus.Thus, although McGregor (1997:96-100) describes Gooniyandi discontinuous phrases which introduce new entities into the discourse, and Siewierska (1984:60,66-68) mentions Polish discontinuous phrases which occur in first-utterance clauses, Schultze-Berndt and Simard (2012:1046,1053n14) do not accept these analyses as compatible with presentative sentence focus, because, according to them, a sentence-focus structure cannot have a topic or theme, contrary to the explanations of both of these authors.However, this seems possibly to be a matter of disagreement on interpretation and/or terminology, while the phenomenon described in the different languages may be similar, since even though McGregor and Siewierska assign a topic or theme function to the whole discontinuous NP or part of it, they maintain that all constituents of the clause convey new information.
Based on functional grounds, Schultze-Berndt and Simard (2012:1040,1047,1050) defend an analysis of true discontinuous phrases as single phrases that are initially generated in split form to mirror the specific information structure configuration, with the coherence between the separate parts being accounted for by semantic dependency (cf.also McGregor (1997:86,94,107), who provides a functional explanation of discontinuity as conveying meaning without the theoretical requirement that constituents always be continuous).Fanselow and Féry (2006:72,75) adhere to a stricter syntactic framework, which requires phrases to be originally contiguous, and, therefore, suggest to explain the discontinuity as involving a movement process based on the formal features of the cohesive, simple discontinuous noun phrases, contrasting with the assumed base generation as independent phrases for the noncohesive inverted splits.Although such true discontinuous phrases are less common than the other constructions from which they should be distinguished, like afterthought and split topicalization, they are reported to occur in a wide variety of configurational as well as assumed non-configurational languages (with considerable and not yet understood cross-linguistic variation), in which they often exist besides the earlier discussed types of seemingly discontinuous noun phrases (Fanselow and Féry 2006:74,75;Schultze-Berndt and Simard 2012:1049,1050).They are described, for example, for the Australian languages Jaminjung (Schultze-Berndt 2008; Schultze-Berndt and Simard 2012) and Gooniyandi (McGregor 1989;1997), while Schultze-Berndt and Simard (2012:1038) and Fanselow and Féry (2006:70-72) furthermore provide literature references and/or examples for languages as diverse as Bulgarian (Slavic), Lak (Caucasian), Nogai (Turkic), Swampy Cree (Algonquian), and Finnish (Uralic).

Non-configurationality in Biblical Hebrew
As becomes clear from the literature, a rather broad range of opinions can be found about what languages and which constructions may be called non-configurational. Hebrew appears on both sides of the line, since various authors describe characteristics of the Hebrew language, both Biblical and Modern 24 , that could be interpreted as arguments for an understanding as either non-configurational or configurational, even though some of them do not explicitly use this terminology.

The assumption of non-configurationality and discontinuous phrases in Hebrew
At the non-configurational side, for example, Andersen and Forbes (2012:87-93) adduce various examples 25 to demonstrate that Biblical Hebrew fulfils the requirements to be categorized as a non-configurational language (referring to Hale 1983:5), viz., free word order (or, rather, "partially free phrase order," as they express themselves in more moderate terms), syntactically discontinuous expressions, and null anaphora.Regarding the second characteristic, it could be remarked that, since they do not avoid representations involving discontinuity, Andersen and Forbes (2012:89-91,294-300) actually analyse quite a variety of constructions as discontinuous phrases with several different "semantic licensing relations," e.g., discontinuous apposition (typically involving relative clauses, as in example 8 below), discontinuous superset constituents (in which the second part is a coordinate phrase consisting of a pronoun and an expansion that together form a superset of the referent of the first part, as in example 9 below), discontinuous union, and discontinuous modification.Similarly, Gil (1983:146-150) tentatively applies the label "non-configurational" to Modern Hebrew based on evaluation of features suggested by Hale (1982:86), especially focusing on free word order within phrases (although the alternative orders he mentions seem to be restricted to specific registers), and the occurrence of discontinuous phrases consisting of a noun with a non-adjacent modifier (although it could be questioned how representative the scarce and possibly marked examples that he provides are).Of the latter, he provides, e.g., the following example from a Hanukkah song.
(10) Ner li dakik.candle to.me slender 'I have a slender candle.' 24 Although the focus of the present article is on assumed non-configurationality and discontinuous phrases in Biblical Hebrew, some relevant information from Modern Hebrew bearing on these topics is mentioned as well.Despite the differences between Biblical and Modern Hebrew, the two are considered to be sufficiently related to justify this as an interesting addition from a typological linguistics view. 25The specific types of examples by which the various authors mentioned in this paragraph illustrate the alleged non-configurational nature of Hebrew, especially those regarding discontinuous phrases, will be discussed further below as to their appropriateness and as to possible alternative explanations.
Throughout his article, he argues for a characterization of non-configurational languages as favouring syntactic structures of lesser hierarchic depth, which would also explain the semantics of stacked adjective constructions in Hebrew in which the order of the adjectives does not contribute to different meanings (thus contrasting with stacked adjectives in English) because, according to his hypothesis, they involve a flat instead of a recursive hierarchic structure (Gil 1983:144,150-152).
Furthermore, discontinuous constructions in Biblical Hebrew are also discussed by Zewi (1999;2007:22) and Gottstein (1949), who do not use the term "non-configurational," but who mention various types of "interrupted syntactical structures," which they recognize as genuine part of Biblical Hebrew syntax functioning to contribute to the information structure or stylistics of the sentence.Thus, Zewi (1999:88-91) notices, e.g., attributes, appositions, and attributive clauses that are separated from their head, which she regards as intentional syntactical variants, suggesting that they are used for "presenting an additional less important piece of information in a clause."26Similarly, Gottstein (1949:35-41) describes several types of "Afterthought," for which he assumes explanations as "pure" afterthought (adding further information on a detail after completing the sentence), stylistic means of emphasis (emphatic extraposition or apposition), emphatic coordination, or afterthought-relative clause, and which, according to him, should thus not be dismissed as anacoluthon or the result of incorrect textual transmission.The verse in (11) below is mentioned by both Gottstein (1949:37) and Zewi (1999:91n12) as an example of apposition of proper names (also illustrated by example 1 above), while, according to Gottstein (1949:37) Before continuing with the authors who deny Hebrew a non-configurational status, it should be remarked that even the above-mentioned authors who call Hebrew non-configurational actually speak about a certain degree of non-configurationality.Thus, Gil (1983:155-157) mentions a "configurationality scale" on which English and Hebrew exhibit a contrast but to a lesser degree than the extreme types of configurational and non-configurational languages represented by, e.g., Maricopa and Warlpiri, respectively27 .Furthermore, Andersen and Forbes (2012:88,93) take a moderate stance towards non-configurationality and conclude that "Biblical Hebrew is, to some yet-to-be-determined extent, a nonconfigurational language" of which the exact position on the "nonconfigurationality gradient" needs to be further researched.This conforms to the recurrent view in the broader linguistic literature that there exists a continuum rather than a dichotomy between configurational and non-configurational languages.For example, Frank (2003:220), who is cited by Andersen and Forbes (2012:93), states that "languages exhibit different mixtures of morphological and/or structural marking of functional information, yielding a continuous scale along the dimension of (non-)configurationality."Similarly, Van Valin (1999:513) posits a continuum of languages with respect to rigidity vs. flexibility of word order (cf.also Jones 2015:92), while Givón (2001a:271) asserts that "word-order flexibility is a matter of degree."28Even Hale as early as 1982 seems to assume the existence of a spectrum between configurational and non-configurational languages (1982:86), although the rest of his article appears to maintain a binary distinction, whereas Hale (1989:294,299) more clearly suggests the possibility of languages displaying various degrees of (non-)configurationality, arguing that (non-)configurationality is a property of constructions instead of of languages such that one language can possess constructions of both types, and that the various properties associated with non-configurationality are not caused by a single parameter.

Arguments against a non-configurational analysis of Hebrew and the lack of true discontinuous phrases
Given the assumption of a continuum of (non-)configurationality, it is perhaps not surprising to find other authors who treat Hebrew as a configurational language and/or as lacking typical non-configurational features.Thus, Jones (2015:93,101,113n125) remarks that ancient Hebrew "is not a non-configurational, free order language."This should not, however, be taken as an affirmation of a rigidly configurational position at the continuum, since he comments that "[m]ost scholars treat B[iblical] H[ebrew] as a middling language with respect to the configurationality of its syntax," while he himself claims concerning the specific corpus of Hellenistic Period Hebrew, which forms the topic of his study, that it "is mainly nonconfigurational."Since Jones is especially interested in how word order relates to syntax and information structure, his discussion of configurationality focuses on the order of constituents within the clause, paying no attention to discontinuity and word order within phrases.With respect to Modern Hebrew, the lack of non-configurationality is perhaps most clearly reflected by Fanselow and Féry (2006:11,37,46), who claim repeatedly that Hebrew does not show discontinuous noun phrases.Furthermore, according to Gil (1983:149n6), it could be deduced from the analysis of pro-drop phenomena in Modern Hebrew by Borer (1981) that "she assumes Hebrew to be configurational" (as opposed to Gil himself).In addition to that, the argument for non-configurational structure in Hebrew that Gil (1983:142-146) advances based on the free order-and concomitant lack of semantic contrasts-in stacked adjective constructions, is countered by Shlonsky (2004, who considers it to be untenable when more data are taken into account. Focusing on the characteristic of discontinuous phrases, there are probably two main types of reasoning that could be adduced against an analysis of Hebrew as a non-configurational language.On the one hand, Hebrew lacks some specific types of discontinuous phrases that seem to be typical for non-configurational languages, while, on the other hand, the types of discontinuous phrases that are reported for Hebrew by some authors (e.g., Andersen and Forbes 2012) would be analysed differently and not be considered true discontinuous phrases by other authors (cf. the alternative explanations in section 3.3).
Regarding the first argument, Biblical Hebrew does not generally exhibit discontinuous constructions of the same sort that are commonly found in prototypical non-configurational languages in which various types of single-word modifiers, like adjectives, demonstratives, and numerals, appear separated from the noun they modify and actually constitute individual phrases that stand in apposition at sentence level 29 .Thus, comparison of examples ( 1) and ( 2) above shows that the Hebrew construction is different from the non-configurational one, for its second part contains more than one word and does not consist of a modifier that is necessary to identify or restrict the reference of the first part.Besides, in addition to the possibility of discontinuous occurrence of appositional noun prototypical non-configurational languages are expected to show two other, related features, which appear to be absent from Hebrew.First, the appositional phrases in non-configurational languages are usually concomitant with extensive morphological marking of grammatical functions (cf., e.g., Frank 2003:220), which is not clearly present in Hebrew.Second, non-configurational languages typically display free order of elements within the phrase (cf. McGregor 1997:88;Rijkhoff 2002:20), whereas in Hebrew, the word order within the phrase is relatively fixed with the noun usually preceding modifiers like adjective or demonstrative 30 .The only significant exception may be posed by cardinal numbers, but even the order of noun and numeral is not completely free, since there appear to be certain tendencies, e.g., in relation to definiteness (cf. Shlonsky 2004Lettinga 2012:75,76 §40f,k), which is in line with Krasnoukhova (2012:191) according to whom the more flexible order of numerals and adjectives with respect to the noun may be caused by dependency on the definiteness or specificity of the referent, whereas these categories are irrelevant in the case of demonstrative and possessive modifiers.Thus, the rather fixed word order in Hebrew phrases suggests a hierarchical structure, which is in accordance with the configurational form of the Semitic noun phrase assumed by Shlonsky (2004), who, even though he acknowledges dialectal variation in the order of numeral and noun in Hebrew, still accounts for that within a hierarchical analysis (Shlonsky 2004(Shlonsky :1487(Shlonsky ,1490(Shlonsky ,1497)).
Likewise, phrases that resemble the true discontinuous phrases found in both nonconfigurational and configurational languages (described in section 2.3 above) also occur only to an extremely limited extent in Hebrew.The examples mentioned by Gil (1983:148) for Modern Hebrew possibly fall into this category (cf.example 10 above) 31 .Additionally, the 29 At first sight, the frequent, and even possibly non-continuous, occurrence of Hebrew appositional constructions in which two nouns or noun phrases are juxtaposed without the first one being in construct state (Livnat 2013), might be interpreted as indicating that Hebrew fills a position towards the nonconfigurational end of the continuum.However, this is still not really comparable with the apposition of individual nouns or noun phrases in prototypical non-configurational languages because the appositional items in Hebrew are usually continuous and could be interpreted as one complex phrase with various possible internal semantic relations, while the instances of discontinuous apposition in Hebrew seem to be more marginal and of a different type, not commonly involving single-word modifiers and usually expressing different semantic relations between the two parts.One specific phenomenon in Biblical Hebrew that might perhaps be considered a bit similar to the occurrence of separate appositional phrases linked to one grammatical function in nonconfigurational languages, consists of multiple "stacked" phrases that independently specify the same type of element of the valency pattern of a verb without referring to each other, viz., the second object of double object constructions with the verb ‫עׂשה‬ 'to make' in building instructions or the source or goal with motion verbs, as discussed by Dyk, Glanz, and Oosting (2014:52-54) and Oosting and Dyk (2017:65,66,69,75).
Biblical Hebrew data used for the present research, as extracted from the ETCBC database 32 , include a few potential candidates for such an analysis as well.While the majority of the seemingly discontinuous phrases in Biblical Hebrew consist of separate parts that would be able to occur as independent nominal or prepositional phrases (disregarding for the moment compound relations as specified by the conjunction ‫,)ו-‬ there are a few rare cases in which the second part contains a modifier that would apparently need to be interpreted in combination with the noun in the first part.This is illustrated by the following examples with non-adjacent adjectives33 .At first sight, these examples (except example 14) might correspond well with the type of true discontinuous phrases that mark "theticity" while introducing a new entity with a particular property as discussed by Schultze-Berndt ( 2008) and Schultze-Berndt and Simard (2012:1041,1042,1046), or perhaps with the "Type A" discontinuous phrases that introduce new participants of intermediate importance as described by McGregor (1997:96-100).In the examples from Genesis 41, this annuntiative interpretation could even be related to the occurrence of ‫ֵּה‬ ‫נ‬ ‫הִ‬ 'behold,' whereas ‫עלה‬ 'to go up' as verb of appearance in Gen 41:3,5,19,22, and the possessive construction with a predicate of existence in Song 8:8 are both compatible with such a presentative reading as well 35 .
Furthermore, especially the construction in Song 8:8 seems to fit well with the "merged" or "restrictive" (argumental) interpretation that is required for true discontinuous phrases, which widespread in Hebrew," which seems a rather strong statement if it is really based on these three examples only. 32This Hebrew Bible database of the Eep Talstra Centre for Bible and Computer (ETCBC), also known as the BHSA, is archived at https://doi.org/10.17026/dans-z6y-skyh.The query searching for all cases that potentially qualify as true discontinuous phrase, of which the results are mentioned in the present section, can be found at https://shebanq.ancient-data.org/hebrew/query?version=2017&id=3512.
typically consist of two single words, viz., a noun and a non-adjacent modifier, that must be interpreted as together identifying a single referent and thus forming one phrase (cf., e.g., Schultze-Berndt and Simard 2012:1026,1032).However, for the examples from Gen 41 and Ezek 23:32, it is less clear whether the separate, multi-word elements really belong to the same phrase because they do not fit the usual pattern of true discontinuous phrases comprising only two words in total (cf.Siewierska 1984:62;McGregor 1997:93;Schultze-Berndt andSimard 2012:1034), and because the referent in these cases is already sufficiently introduced by the first part, while the adjectives in the second part only provide additional qualifying information, without changing the identification of the referent as determined by the adjective ֲ ‫א‬ ‫ת‬ ‫רוֺ‬ ‫חֵּ‬ 'other' in Gen 41 and by ‫ך‬ ‫ֲחֹותֵּ‬ ‫א‬ 'of your sister' and the mention in the previous verse in Ezek 23.Therefore, the construction in these verses could perhaps better be analysed as involving some sort of afterthought, thus representing the "unmerged" or "non-restrictive" (predicative) interpretation (cf., e.g., McGregor 1997:83,101,102;Schultze-Berndt andSimard 2012:1026).If the adjectives in the second part should indeed be considered an independent phrase, they could obviously be understood as nominalized (especially the ones preceded by the article in Ezek 23:32), although this is not strictly necessary for the indefinite adjectives in Gen 41, for which analysis as independent adjective phrase (with predicative function and morphologically congruent with the previous noun phrase) would be possible as well 36 .Even though the adjective in Song 8:8 seems to function attributively with respect to the noun, the alternative analysis with a separate, predicative and possibly nominalized adjective phrase may be a valid option here as well.To decide on this matter, it would be helpful to include prosodic information in the argument, but the Masoretic accents could only be taken as a tentative indication in this respect.Especially the stronger disjunctive accents in the cases of Gen 41 and Ezek 23:32 might point in the direction of an analysis as afterthought, while the smaller disjunctive accent in Song 8:8 could perhaps still favour an analysis as true discontinuous phrase.
The present data contain no clear examples of possibly true discontinuous phrases with modifiers other than adjectives, except for the somewhat peculiar construction in Deut 15:9 37 , in which the noun ‫ַל‬ ‫ַע‬ ‫יּ‬ ‫לִ‬ ‫בְ‬ 'wickedness' seems to attributively modify the preceding, non-adjacent noun ‫ר‬ ‫בָ‬ ‫דָ‬ 'word,' to which it may have a specification or appositional relation (the existential verb ‫היה‬ 'to be' in this clause would perhaps fit the presentative reading of discontinuous phrases) 38 .All in all, Biblical Hebrew might feature a few true discontinuous phrases of the 36 A nominalized interpretation of the adjectives constituting the second part would be comparable to the analysis that can be found in the ETCBC data for ‫ָר‬ ‫ָכ‬ ‫ז‬ ‫אוֺ‬ ‫ה‬ ‫בָ‬ ‫ְקֵּ‬ ‫נ‬ 'male or female' in Lev 3:6 (note that the nominalization of the adjective ‫ָר‬ ‫ָכ‬ ‫ז‬ 'male' is obligatory here due to the nominal form of the noun ‫ה‬ ‫בָ‬ ‫קֵּ‬ ְ ‫נ‬ 'female' with which it forms a compound phrase) and ‫ֳשים‬ ‫ֲד‬ ‫ח‬ 'new' in Deut 32:17.Thanks are due to Constantijn Sikkel (p.c. 24-4-2018) for the discussion on these cases. 37The analysis as discontinuous phrase in the ETCBC database presupposes a relation between ‫ַל‬ ‫ַע‬ ‫יּ‬ ‫לִ‬ ‫בְ‬ and ‫ר‬ ‫בָ‬ ‫דָ‬ as 'wicked thought' which is also found in many translations, rendering ‫ֶ֣ה‬ ‫יֶׁ‬ ‫ִהְ‬ ‫ֶׁן-י‬ ‫פ‬ ‫ר‬ ‫בָ‬ ‫דָ‬ ‫ך‬ ‫בְ‬ ‫בָ‬ ‫ם-לְ‬ ‫עִ‬ ‫ַל‬ ‫ע‬ ַַּ֜ ‫יּ‬ ‫לִ‬ ‫בְ‬ along the lines of 'that there be not a wicked thought in your heart.'However, the King James Version instead translates ‫ַל‬ ‫ַע‬ ‫יּ‬ ‫לִ‬ ‫בְ‬ in relation to ‫ך‬ ‫בְ‬ ‫בָ‬ ‫לְ‬ as 'thy wicked heart', while Janet Dyk (p.c. 5-7-2018) suggests another alternative analysis with the phrase ‫ַל‬ ‫ַע‬ ‫יּ‬ ‫לִ‬ ‫בְ‬ as predicate complement and the phrase ‫ך‬ ‫בְ‬ ‫בָ‬ ‫לְ‬ ‫בָ‬ ‫דָ‬ ‫ר‬ ‫ם-‬ ‫עִ‬ as subject, leading to the interpretation 'let no thing in your heart be wicked.' 38Albeit the examples are scarce and debatable, the fact that they practically only involve adjective or adjective-like modifiers could perhaps be regarded as confirming the scale of degree of integration for modifiers suggested by Krasnoukhova (2012:181,184), according to which adjectives ("property words" in her terminology) and numerals, which commonly have a specifying and describing function, are more likely to occur discontinuously than lexical possessors and demonstratives, which commonly have an identifying and referring function.
presentative type, but it is also well possible that these are exceptional cases, which could not be used to support a non-configurational view on Hebrew.

Alternative explanations for seemingly discontinuous phrases in Biblical Hebrew
Regarding the second argument against the non-configurationality of Hebrew, it should be noted that the types of seemingly discontinuous phrases that lead some authors to believe that Biblical Hebrew may be non-configurational could actually be accounted for by different syntactic processes, as is demonstrated by other authors.This might thus, in fact, amount to an issue of terminology regarding the definition of discontinuous phrases.A broader understanding of the term "discontinuous phrase" would also include, e.g., extraposed relative clauses, extraposed appositives, afterthoughts, and split or emphatic coordination, all of which are reported for Biblical Hebrew, while there is some divergence, and consequently overlap, in the terminology used within different treatments, and not all authors are equally explicit about whether these constructions should be considered to constitute a single, discontinuous phrase (cf.Andersen and Forbes 2012:89-91,294-300;Gottstein 1949:36-41;Zewi 1999:90-92;Michel 1997a;b;Thorion-Vardi 1987;Holmstedt 2014:132-140;Floor 2004:93-95,290-292).Most authors who describe these structures agree that the use of the discontinuous variants may be motivated by pragmatic reasons related to information structure and/or by an attempt to reduce complexity and thus facilitate ease of processing by letting heavy constituents occur clause-finally.A more narrow definition, however, excludes such constructions from the discussion on discontinuous phrases, and rather analyses them differently.Disregarding the differences in use of terminology, which will be further discussed in a following article, the collection of seemingly discontinuous phrases in Biblical Hebrew could roughly be divided into two groups.
On one hand, there are the constructions that could probably best be analysed along the lines of the second category of alternative explanations mentioned above in section 2.2, i.e., as two partly or completely coreferential but independent phrases, of which the second one is extraclausal and elaborates on the first one.A strong case for such an analysis could especially be made when the supposed extra-clausal constituent involves some sort of repetition of an element from the (partly) coreferential phrase that occurs earlier in the clause.This alternative explanation, which is commonly called "afterthought" or "right dislocation," probably pertains to the examples (1), ( 9), ( 11), and perhaps also ( 14) and (15), above.Besides in the earliermentioned work of Gottstein (1949), it is applied to Biblical Hebrew by, e.g., Floor (2004:38,39,76,93-95,290-292), who refers to it as "tail topic," providing as examples, among others, Gen 2:25, which is similar to the apposition in ( 1) and ( 11), and Gen 13:1, which is similar to the compound phrase that expands the referent of the initial pronoun in (9).Furthermore, much of what Thorion-Vardi (1987) calls "Ultraposition" could probably also be subsumed under this category, for her examples include (1) above, as well as (9), which she describes more specifically as "Gleitkonstruktion," while she also mentions constructions with It might be objected that the ETCBC database includes a considerable number of demonstratives that constitute one element of a discontinuously analysed phrase, but this concerns a different use of the demonstrative, viz., as independent pronoun and not as attributively modifying a noun.In these cases, the second part should probably be regarded as a further specification when it has the form of, e.g., a phrase starting with ‫ן‬ ‫מִ‬ 'from,' or as some sort of appositional phrase when it constitutes a list of items which is coreferential with the demonstrative (cf.also Livnat 2013), although the latter type is also often analysed as ellipsis in the ETCBC database.lexical repetition under the subtype "Repetitionskonstruktion," as illustrated by the following example (Thorion-Vardi 1987:15,75-77,86,87) On the other hand, there are the constructions that can be defined as "extraposition from a noun phrase," like, e.g., extraposed relative clauses or prepositional phrases.Extraposition was not yet discussed in section 2.2 as one of the alternatives for constructions resembling discontinuous phrases, because it is not elaborately encountered in the literature that aims to narrow down the definition of discontinuous phrases in the context of non-configurationality.The reason for that might be that extraposition also occurs in clearly configurational languages like English (cf.Givón 2001a:281), and, thus, is not a defining characteristic non-configurationality.
Although some authors would include this phenomenon in their interpretation of discontinuous phrases (cf.Siewierska 1988:221;Rijkhoff 2002:257-259;Andersen and Forbes 2012:294), others describe extraposition as a separate construction (cf., e.g., Fanselow and Féry 2006:3;Holmstedt 2014:132-136), which is generally explained in syntactic theory as rightward movement of a dependent element.Nevertheless, Fanselow and Féry (2006:3n4) remark that this may be more similar to discontinuous noun phrases than to the leftward movement of a dependent element called extraction 39 .The main difference between extraposition and true discontinuous phrases seems to lie in the type of element that occurs separately: in a discontinuous phrase, this concerns a single-word modifier of the head noun (e.g., an adjective, quantifier, or demonstrative), which cannot form a complete phrase without the noun, whereas in extraposition, it is a whole constituent that is thematically dependent on the noun (either as argument or adjunct), which is structurally embedded in the noun phrase, but which could theoretically also form an independent constituent without the noun (cf.Fanselow and Féry 2006:3;De Vries 2009:66).Related to the type of element, there is usually a difference in length as well, such that extraposition involves longer elements than discontinuous phrases, since an important reason for extraposition, apart from possibly pragmatic reasons, is found in the principle that "heavy" (i.e., long or grammatically complex, e.g., embedded) constituents are moved towards the end of the sentence to facilitate processing (cf., e.g., Holmstedt 2014:140,150;Rijkhoff 2002:259;Wasow 1997:82,83;Dik 1989:374,375;1997:341,342).
Extraposition is relevant for Biblical Hebrew 40 , since it may account for various instances of what are called discontinuous phrases by, e.g., Andersen and Forbes (2012), such as example (8) above.The term "extraposition" is prominently employed in the analysis of Biblical Hebrew, by Holmstedt (2014:132-140,150;2001:9-13), who provides many examples of extraposed relative clauses (but see the next paragraph for comments on his rather broad application of the concept of extraposition), as well as the following example of an extraposed PP (Holmstedt 2014:133n47), which clearly illustrates the tendency to place more complex constituents at the end of the clause.
39 Furthermore, in the Holmstedt-Abegg database (as available in Accordance), extraposed appositives are in fact visualized as discontinuous constituents, but Cook (2016) explained that this is a choice of representation, while the syntactic analysis actually involves movement. 40In the present article, the term "extraposition" is not used as in some works on Biblical Hebrew grammar (e.g., Khan 1988:xxvi;Lettinga 2012:134,135 §70), in which it refers to left (and possibly right) dislocated elements outside the clause, but as in the modern, general linguistics works mentioned in this section, in which it refers to elements that are moved from their normal position towards the end of the clause (as is also explained by Holmstedt 2014:132n44 Furthermore, even though Michel (1997a;1997b) does not explicitly speak about "extraposition," it may nevertheless be applicable to part of the Biblical Hebrew constructions that he treats as "gespaltene Koordination" (cf.also example 12 above).
Extraposition differs not only from discontinuous phrases, but also from afterthought or right dislocation, although, in practice, this distinction might be difficult to establish (cf.also Holmstedt and Jones 2017: §5;Ott and De Vries 2016:644;De Vries 2009:80).Besides, the matter is further complicated because there exists variation in the interpretation of terminology, and, hence, similar structures may be referred to by various terms in the literature.For example, the frequently occurring construction in Biblical Hebrew in which a clause-final phrase elaborates on or specifies a coreferential phrase earlier in the clause appears to be analysed as "extraposed appositives" by Holmstedt (cf. Holmstedt 2014:133,134; Holmstedt and Jones 2017: §5; with reference to the term "Ultraposition" by Thorion-Vardi 1987), whereas, according to the terminology followed in the present article, this construction would rather be categorized as right-dislocated afterthought.Holmstedt (2014), however, employs a narrow definition of right dislocation, only referring to dislocated phrases that resume a pronominal reference in the clause (called the "backgrounding" variety of right dislocation by, e.g., Ott and De Vries 2016), and, consequently, a rather broad application of extraposition, also including phrases that resume an NP or PP earlier in the clause (called the "afterthought" variety of right dislocation by, e.g., Ott and De Vries 2016).In principle, though, while extraposition is traditionally viewed as a movement process within the clause, afterthought is regarded as a base-generated independent phrase outside the clause, and this difference could be recognized in various characteristics.Thus, an extraposed phrase is integrated in the sentence with respect to focus assignment and prosody (attracting the sentence accent), whereas afterthought or right dislocation shows disintegrated behaviour regarding the preceding clause, which forms a complete sentence with its own intonation contour and information structure (cf.Frey and Truckenbrodt 2015:96-99;Ott and De Vries 2016:644).However, the prosodic criterion, which could aid in determining the right clause boundary and thus proving the extra-clausal position of afterthoughts, is, of course, not of much help for the analysis of Biblical Hebrew.Furthermore, afterthought being coreferential with an element in the preceding clause usually has the same form as this element (e.g., both NP or both PP), contrary to extraposition as it is typically treated in the literature (cf.Guéron 1980;Huck and Na 1990;Takami 1999;Vries 2009:66), which often has a different form being itself originally embedded in another phrase (e.g., PP embedded in NP as in 17 above) with which it is thus not coreferential.Related to this, extraposition, generally, is restrictively modifying a preceding element in the clause and as such is included in the propositional content of the sentence, while afterthought has a non-restrictive interpretation, being parenthetically construed with respect to the coreferential element in the preceding clause, and thus forms a secondary message which does not contribute to the primary proposition of the sentence (cf.Ziv 2015:134;De Vries 2009:62,67-78,85).With respect to this last feature, it should also be noted that most of the non-adjacent relative clauses in Biblical Hebrew, in fact, appear to be non-restrictive (cf. the remarks by Gottstein 1949:39n10;Zewi 1999:91;Stackert 2010:174), whereas general linguistic literature on extraposition often mentions restrictive relative clauses, at least for English (e.g., Givón 2001b:207,208;Huck and Na 1990:60).This may suggest that these non-restrictive relative clauses in Biblical Hebrew could perhaps better be considered as an extra-clausal, afterthought-like construction (cf.also Frey and Truckenbrodt 2015:98 for the similarity of non-restrictive relatives with right dislocation and afterthought), leaving only the non-adjacent restrictive relative clauses (like example 8) to be analysed as extraposition, contrary to Holmstedt (2001;2014:132-134;2016:186-188), who includes many examples of non-restrictive relative clauses (like, e.g., Gen 24:15) in his treatment of extraposition in Biblical Hebrew (even though Holmstedt 2016:196 mentions the rare occurrence of extraposition as distinguishing feature of non-restrictive relatives, referring to English-based syntactic literature).
Summarizing, the figure below contrasts the alternative analyses that are relevant for Biblical Hebrew with true discontinuous phrases, which, in the narrow definition, only occur in highly specific pragmatic contexts (cf.Schultze-Berndt andSimard 2012:1040;Siewierska 1984:57,69).The oppositions between extraposition, afterthought (or right dislocation), and true discontinuous phrases are listed, according to the characteristics that were previously described in section 2.2, 2.3, 3.2, and 3.3 for these seemingly discontinuous phrases which comprise a clause-final element, thus excluding clause-initial constructions, like left dislocation, which are not commonly misunderstood as discontinuous phrases in Biblical Hebrew.For each category, the table also provides the corresponding examples from the present article, with the Biblical Hebrew ones marked in bold.
Extraposition examples 8, 12, 17 -movement of embedded, multi-word constituent (e.g., relative clause, PP) -within clause -same intonation unit -same information unit -restrictive part of primary predication -(usually) not coreferential -often different form from related element -combination with related element leads to well-formed phrase -often due to processing reasons, viz., placing "heavy" constituents at the end (possibly also pragmatic reasons) Afterthought examples 1, 4, 9, 11, 14?, 15?, 16 -separate generation of independent constituent -extra-clausal -separate intonation unit -separate information unit -non-restrictive, secondary predication -(partly) coreferential -usually same form as related element -combination with related element does not necessarily lead to well-formed phrase (e.g., repeating lexeme or pronoun) -due to pragmatic reasons, viz., different information structure value True discontinuous phrase examples 3, 6, 7, 10, 13(?), 14?, 15? -separation of single-word modifier (e.g., adjective, quantifier, demonstrative), which does not normally form an independent constituent -due to pragmatic reasons, viz., contrastive argument focus (incl.given element) or presentative sentence focus -generation as single constituent (initially in split form or followed by movement) -within clause -same intonation unit -restrictive interpretation -not coreferential, not independently referring -combination with related element leads to well-formed phrase

Conclusion
Even though Biblical Hebrew exhibits grammatical possibilities that are not attested in some other, more strongly configurational languages-which means that it is at least not at the extreme configurational end of the continuum-this seems insufficient evidence to warrant categorization as a truly non-configurational language.As regards the seemingly discontinuous phrases in Biblical Hebrew, these cannot simply be used to support the claim of non-configurationality, since investigation of general and typological linguistic literature has shown that they are not comparable to the discontinuous constructions formed by appositional, singleword NPs in typical non-configurational languages.Furthermore, in the light of the same literature, it seems defensible to apply a stricter definition of true discontinuous phrases which may occur in languages at various positions on the configurationality continuum, and Biblical Hebrew exhibits only a marginal number of constructions that might fit this narrow description.This means that, instead of ascribing the occurrence of seemingly discontinuous phrases to the presumed non-configurational nature of Biblical Hebrew or resorting to an interpretation as true discontinuity, a different explanation should be sought.
The most plausible remaining possibilities for analysis of these discontinuous constructions, which usually contain a clause-final element, seem to be afterthought and extraposition, but further research is needed to obtain a better understanding of the varieties and intricacies involved.Therefore, a following article dedicated to this topic will elaborate on the various possible analyses for these seemingly discontinuous phrases in Biblical Hebrew, looking for patterns that will provide more insight into their role within the syntactic and pragmatic structure of the sentence, while also taking into account processing considerations.
In any case, a complete analysis of all instances is not feasible due to the lack of prosodic information.For example, to distinguish pragmatically marked, extra-clausal, independent phrases from real discontinuous phrases, or from extraposed phrases, prosodic features should be taken into consideration as well, which is, however, more difficult in the case of Biblical Hebrew as a dead language 41 .The Masoretic cantillation signs may be a topic for further research in this respect, but even these are not expected to provide decisive evidence, since they form a later addition to the text and only represent one possible reading, thus leaving the original prosodic information largely inaccessible (cf.Zewi 2007:8;Holmstedt 2016:28;Floor 2004:11).
Podobno mają piękny dom.apparently have beautiful house 'Apparently they have a beautiful house.'Nie! Piękny mają ogród.Dom mają kiepski.no beautiful have garden house have crummy 'No!They have a beautiful garden.Their house is crummy.'

(
The category of appositional phrases can accordingly be further divided into what Krasnoukhova (2012:169) calls "close apposition" (with juxtaposed constituents) and "loose apposition" (with separately occurring constituents), or whatRijkhoff (2002:19-21)describes as apposition of minor NPs within a larger phrasal unit 5 and apposition at sentence level.Although in the first type of apposition, the appositive elements must still be adjacent, the relatively independent nature of the elements in the second type accounts for the possibility of discontinuous occurrence at clause level(cf.Rijkhoff  2002:23), as is illustrated by the following example 6 from the Australian language Warlpiri(Hale 1983:6)  7 .
(McGregor 1989)rarchical phrases in Nama (J.R.Payne 2006:717)and "non-fractured phrases" in Gooniyandi(McGregor 1989), both of which are languages that appear in the literature as providing examples of non-integral phrases. Smilarly, based on the study of a sample of 55 languages of South , the construction in (12) exemplifies emphatic coordination.
'And David's two wives were taken captive: Ahinoam the Jezreelitess, and Abigail the wife of Nabal the Carmelite.'(1Sam 30:5) 'And our God brought on us all this evil, and on this city.'(Neh13:18) . ).