Maschler, Yael and Miller Shapiro, Carmit. (2016). The role of prosody in the grammaticization of Hebrew naxon (‘right/true’): Synchronic and diachronic aspects. Journal of Pragmatics

Available online at www.sciencedirect.com

ScienceDirect

The role of prosody in the grammaticization of Hebrew naxon(‘right/true’): Synchronic and diachronic aspects§

www.elsevier.com/locate/pragmaJournal of Pragmatics 92 (2016) 43--73

Yael Maschler a,*, Carmit Miller Shapiro b

aDepartment of Hebrew Language, University of Haifa, IsraelbDepartment of Hebrew Language, Oranim College, Israel

Received 15 June 2015; received in revised form 18 November 2015; accepted 22 November 2015

Abstract

Basedonasynchronicanalysisofall naxon (‘right/true’) tokens found throughoutacorpusof casual spokenHebrewdiscourse,weoutlinetwo continua of synchronic usage suggesting two functional itineraries for naxon. We show that naxon employed in non-appeal (Du Boiset al., 1992) intonation contours first evolved fromaverb toanadjectiveand then toaprototypical discoursemarkeremployed toagreewith orconfirm an interlocutor’s utterance while expressing epistemic stance of certainty. Taking another grammaticization (Hopper, 1987) path,independent of the former one, naxon employed in appeal intonation contours evolved beyond the adjective into the widely-debated xagam(‘lacking person, gender, and number’, Rosén, 1963) fossilized impersonal form. It then continued to evolve into a ‘quasi-xagam’ formconstituting a projecting (Auer, 2005) construction, the function of which is to foreshadow an assertion requiring agreement/confirmation.Diachronicevidence fromBiblical,Mishnaic,Medieval, andearlyModernHebrew is thenaddedas further support for thesegrammaticizationpaths.The twodifferent pathsunderscore the importanceofconsideringprosody indetermining the functional itineraryofa linguistic form.Weprovide counter-evidence from a Semitic language for the asymmetric hypothesis concerning the left and right peripheries (Degand andFagard, 2011; Beeching and Detges, 2014). Finally, we contribute a discourse-functional perspective on the xagam debate, showing themechanisms by which a particular syntactic category may have come about and how it continues to evolve in the language.© 2015 Elsevier B.V. All rights reserved.

Keywords: Agreement/confirmation tokens; Epistemicity; Discourse markers; Left/right peripheries; Grammaticalization and prosody; Hebrewxagam

1. Introduction

Hebrew naxon, roughly equivalent to English right, is a present tense form of a word which functioned in BiblicalHebrew as a verb1 but in Modern Hebrew is generally classified in dictionaries as adjective. Even-Shoshan’s dictionary(1986), e.g., provides three meanings: (1) ‘true, which is exactly this way’; (2) ‘appropriate, aimed at the true intention’; (3)‘ready, prepared’. Like other Hebrew adjectives, naxon is declined for gender and number but not for person or tense, as

§ The authors have contributed equally to this study. Yael Maschler would like to acknowledge Grant # 887/12 from the Israel ScienceFoundation and a Visiting Professorship at the Finnish Center of Excellence in Research on Intersubjectivity in Interaction at the Department ofFinnish, Finno-Ugrian, and Scandinavian Studies, University of Helsinki, which have both enabled completion of this study. The study has beensupported by the project (15BYY045) about the pragmatics of media discourse in the context of crisis funded by the National Planning Office ofPhilosophy and Social Sciences in P.R. China.* Corresponding author at: Department of Hebrew Language, University of Haifa, Mount Carmel, 31905 Haifa, Israel. Tel.: +972 4 811 0941.E-mail addresses: [email protected] (Y. Maschler), [email protected] (C. Miller Shapiro).

1 The verb comes from the root √k.v.n. in the nifˈal verb morphological pattern (binyan).

http://dx.doi.org/10.1016/j.pragma.2015.11.0070378-2166/© 2015 Elsevier B.V. All rights reserved.

http://crossmark.crossref.org/dialog/?doi=10.1016/j.pragma.2015.11.007&domain=pdf

http://crossmark.crossref.org/dialog/?doi=10.1016/j.pragma.2015.11.007&domain=pdf

http://www.sciencedirect.com/science/journal/03782166

http://dx.doi.org/10.1016/j.pragma.2015.11.007

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.pragma.2015.11.007

Y. Maschler, C. Miller Shapiro / Journal of Pragmatics 92 (2016) 43--7344

are verbs. However, upon examination of naxon in mundane Hebrew conversation, we will see that most of its tokensfunction as prototypical discourse markers (Maschler, 2009:17).

In this study, in order to be considered a prototypical discourse marker, an utterance must fulfill two requirements:

(1) S

2

requ3

HopgrammeaMas

4

con5

bett(Ku

emantically, the utterance must have ametalingual interpretation in the context in which it occurs. In other words,rather than referring to the extralingual world, it must refer metalingually (a) to the realm of the text, in which case weare concerned with a ‘textual discourse marker’, (b) to the interaction among its participants (this includes markersreferring to relations of speaker to utterance (Becker, 1979)), in which case we are concerned with an ‘interpersonaldiscourse marker’, and/or (c) to the participants’ cognitive processes occurring during verbalization (in which casewe are concerned with a ‘cognitive discourse marker’).

(2) S
tructurally, the utterance must occur at intonation-unit initial position, either at a point of speaker change,2 or, insame-speaker talk, immediately following any intonation contour other than continuing intonation, unless it occurs aspart of a discourse marker cluster.
In Maschler’s study of Hebrew discourse markers, the semantic and structural requirements for prototypicaldiscourse markerhood coincided for 94% of all 613 discourse markers employed. Tokens fulfilling only the firstrequirement are considered non-prototypical discourse markers (see Maschler, 2002, 2009; Maschler and Schiffrin,2015).

In the present study, we investigate all tokens of naxon found throughout a corpus of casual spoken Hebrew discourse.Based on a synchronic analysis considering syntax, semantics, pragmatics, as well as prosody, we outline two continua ofsynchronic usage suggesting two grammaticization3 (Hopper, 1987) paths for naxon. Diachronic evidence from Biblical,Mishnaic, Medieval, and early Modern Hebrew is then added as further support for these paths. We thus present strongsynchronic as well as diachronic evidence suggesting that naxon first evolved from a verb to an adjective and then to aprototypical discourse marker. Taking another grammaticization path, independent of the former one, naxon continued toevolve beyond the adjective into a fossilized form which some Hebrew linguists classify as a syntactic category (‘part-of-speech’) termed xagam (‘lacking person, gender, and number’, in other words, ‘indeclinable’, Rosén, 1963, see below). Itthen continued to evolve into what will be termed here a ‘quasi-xagam’ form (see below) functioning as a projectingconstruction (Auer, 2005; Günthner, 2011).

Modern Hebrew, generally an SV(O) language, exhibits amarked predicate-first construction (VS) in complex clauses,e.g.4:

tov she'asita halixa hayomgood MASC SG that you did [a] walk todayit’s good that you walked today

In most approaches, the subject of this construction is considered to be the nominalized clause (finite or non-finite) (herethe finite clause: ˈasita halixa hayom ‘you walked today’) and its predicate (here: tov ‘good’) may be an adjective, a verb, anoun, a prepositional phrase, or a word belonging in a group of about 10 fossilized forms, which historically have comefrom various sources.5 Rosén (1963:220) coined a new part-of-speech for these forms, which he termed xagam -- aHebrew acronym for ‘lacking person, gender, and number’, i.e., ‘indeclinables’. Such [Predicate + Subject Clause]constructions are termed xagam constructions, and they are always impersonal. If the predicate happens to be a verb, it isalways in the unmarked 3rd person masculine singular present form. Rosén, and later Kuzar, showed that xagamconstructions have mainly modal and existential functions (Rosén, 1963, 1977a,b; Kuzar, 1992, 2002, 2007, 2012). For

A stand-alone token is, of course, also verbalized within an intonation-unit. It occupies the entire intonation unit and therefore satisfies theirement for occurring at intonation-unit initial position at a point of speaker change.Following Hopper (1987), we employ this term rather than the more common one, ‘grammaticalization’ (e.g., Meillet, 1912; Kurylowicz, 1964;per and Traugott, 2003), in order not to evoke any mistaken assumptions concerning the evolution of some ungrammatical form into amatical one. Furthermore, our approach is first and foremost a synchronic one, focusing on the continually changing categories andnings from a discourse-pragmatic (e.g., Hopper and Thompson, 2008; Hopper, 2011), rather than a historical-diachronic perspective (seechler, 2009:33--34).Unless otherwise specified, all examples come from The Haifa Corpus of Spoken Hebrew (Maschler, 2015a) (see below). Transcriptionventions and abbreviations can be found in the Appendix.For example, the preposition + interjection keday ‘worthwhile’ from ke-day (‘as enough’), the passive participle mutav (‘had better’, lit. ‘madeer’), the adverb of possible Persian origin ˈefshar (‘possible’), the interjections xaval (‘too bad’), day (‘enough’), ˈoy vaˈavoy (‘woe and alas’)zar, 2012:104--105).

Y. Maschler, C. Miller Shapiro / Journal of Pragmatics 92 (2016) 43--73 45

Table 1Categories of intonation contour.

Continuing Form Mid level, mid rise, mid fallMeaning ‘I’m not finished with my current discourse activity’

Final Form Low fallMeaning ‘I’m finished with my current discourse activity’

Appeal Form High riseMeaning ‘I’m seeking an (implicit) response from you’

(Du Bois, ms, Chapter 5, Section 5.3)

instance, the example above carries a modal flavor in that it expresses the speaker’s stance toward the proposition madein the following subject clause (that ‘it’s good’).

Rosén observed that the proliferation of xagam constructions in Modern Hebrew was strongly influenced byimpersonal constructions in the contact languages Yiddish, Russian, German, and English (cf., e.g., English it’s good/possible/interesting to/that. . .) at the time of revival of spoken Hebrew in the late 19th century. However, as Mor and Pat-El(in press) show, the construction can be found -- mostly in direct speech -- also in Biblical Hebrew (10th century B.C.--1st

century A.D.).6 Mor and Pat-El show that the xagam construction is even found in much earlier periods of the language, i.e., in the El Amarna Canaanite tablets dating back to the 14th century B.C.7

The xagam construction has been widely debated in Hebrew linguistics over the years (e.g., Rosén, 1963, 1966,1977a,b; Rubinstein, 1968; Stern, 1983; Kuzar, 1992, 2002, 2007, 2012; Azar, 1995; Abramson, 1998; Trummer, 2000;Dubnov, 2008; Mor and Pat-El, in press). While some consider the predicate in a xagam construction as belonging to apart-of-speech unto itself (e.g., Rosén,8 1963, 1966, 1977a,b; Stern, 1983), others view it as a syntactic positionrestricted to the predicate in the [Predicate + Subject Clause] construction, whichmay be realized by words belonging to avariety of parts-of-speech (e.g., verb, adjective, noun, prepositional phrase) (e.g., Rubinstein, 1968; Azar, 1995). AsDubnov notes (2008:34), the issue basically depends on whether the researcher favors morphology or syntax in his/hercategorization into parts-of-speech. The latter approach is problematic, as Kuzar observes, because it leaves the 10fossilized forms, which historically have come from a variety of different sources (see footnote 5), unclassified as to theirsyntactic (part-of-speech) category. On the other hand, the former approach is also problematic because there are nocommon morphological features to these 10 words that would justify classifying them as belonging in the same syntacticcategory other than their being indeclinable. Furthermore, it is difficult to imagine a syntactic category of content wordswith only 10 members (Kuzar, 1992:247, see also 2012:115).

Our study sheds new light on the xagam construction by examining it from a discourse-functional perspective. We takea dynamic approach, viewing syntactic categories as emerging from the use peoplemake of utterances in discourse.We explore the mechanisms by which a particular xagam construction may have come about in Hebrew and how itcontinues to evolve in the language.

The study is based on 105 conversations (approximately 5 h of talk) taken from the Haifa Corpus of Spoken IsraeliHebrew (Maschler, 2015a), fully transcribed and segmented into intonation units (Chafe, 1994). These natural andspontaneous conversations took place over the years 1994--2007 among Israeli students, their friends and relatives, with2--5 participants per interaction; the total number of participants is 290. Altogether, 160 tokens of naxon were employedthroughout the corpus.

Our transcription systemwas developed based on the Santa Barbara transcription system (Chafe, 1994; Du Bois et al.,1992; Du Bois, ms). One important feature here concerns the contour of intonation units. Du Bois distinguishes three basictypes of contour: continuing, final, and appeal. These intonational categories can be thought of as intonational signs, eachwith a signifier (form) and a signified (meaning) (Table 1).

In our study of Hebrew naxon, we have found it useful to separate out those tokens uttered in appeal intonationcontours from the rest of the tokens, which occur in what we term ‘non-appeal intonation’ (i.e., continuing and finalcontours). Our findings strongly suggest that naxon employed in appeal-intonation contours has followed a differentgrammaticization path compared to naxon employed in continuing and final intonation contours. This underscores theimportance of considering prosody in determining the functional itinerary of a linguistic form.

6 For example, tov lagever ki yisa ˈol bineˈurav (‘It’s good for a man that he bear the yoke in his youth’), Lamentations 3:27.7 For example, pa-si-ih a-na su-nu ‘It is tranquil for them’ (El Amarna 362:57), Mor and Pat-El, in press.8 Rosén further argues that when an adjectival form appears in a xagam construction, it does not constitute an adjective (a form declinable in

four forms in Hebrew, according to the categories of gender and number) but rather a xagam grammatical category (an indeclinable) (1963).


We begin in Section 2 with a classification of all non-appeal naxon tokens in our database according to syntacticcategory, exploring for each category the syntactic scope, semantics, pragmatic functions, and prosodic properties ofnaxon. In Section 3, we similarly explore naxon tokens occurring in appeal intonation contours. Based on this synchronicanalysis, we propose two different grammaticization paths for naxon. Following a summary of our findings (Section 4), inSection 5 we discuss implications of our study for issues of left and right periphery in the study of discourse markers (e.g.,Degand and Fagard, 2011; Traugott, 2012; Beeching and Detges, 2014). In Section 6, we support our synchronic analysiswith a diachronic analysis of Biblical, Mishnaic, Medieval, and early Modern Hebrew and discuss implications of ourresearch for the study of the Hebrew xagam construction and for the role of prosody in grammaticization studies. Section 7concludes the study.

2. Analysis of naxon in non-appeal intonation contours: form and function

2.1. naxon as adjective

2.1.1. Attributive adjectiveDespite the dictionary’s classification of naxon as adjective (Even-Shoshan, 1986), only 11 (or 7% of) naxon tokens

throughout the corpus belong in this syntactic category. They are all employed in masculine singular, as in the followingexcerpt from a conversation between two male students about what makes people laugh when watching the ‘CandidCamera’ show:

Excerpt 1 (‘Candid Camera’):

72 Boaz: .... 'ata tsoxek mima shemaftia,you laugh from what that surprisingyou laugh from what is surprising,

73 shehuthat itthat is

74 ..ba baheksher hanaxon.in in the context the rightin in the right context.

We see that naxon is verbalized within the prepositional phrase baheksher hanaxon (‘in the right context’, line 74) andfunctions as an attributive adjective modifying the noun heksher (‘context’) agreeing with it in gender, number, anddefiniteness.9 Its scope is thus a noun referring to an entity in the extralingual world -- ‘the context’, and it appears as part ofthe relative nominal clause hu baheksher hanaxon (‘it [is] in the right context’, lines 73--74), itself modifying thenominalized clause ma shemaftia (‘what is surprising’, line 72) and combined with it via the complementizer she- (‘that’,line 73).

As can be seen in Fig. 1, none of the prosodic cues indicating an intonation unit boundary10 are found between the finalsyllable of baheksher (‘in the context’) and the first syllable of hanaxon (‘the right’): there are no pauses, no anacrusis, nospecial voice quality phenomena characterizing intonation unit boundaries, and there is no pitch reset (the pitch remainsaround 166 Hz). In terms of structure, then, we see that naxon is employed intonation-unit internally, and its scope is anoun within a clause.

9 A Hebrew adjective always follows the noun it modifies and agrees with it in gender, number, and definiteness. The definite article consists ofthe ha- morpheme prefixed both to the noun and to the adjective. In the case of a prepositional phrase, the preposition often merges with thedefinite article, e.g.: be- (‘in’) + ha- (‘the’) + heksher (‘context’) becomes baheksher (‘in the context’).10 Following the Santa Barbara approach to the transcription of spoken discourse (e.g., Chafe, 1987, 1994; Du Bois et al., 1992; Du Bois, ms),each line denotes an intonation unit. Intonation unit boundaries are characterized by a combination of a variety of prosodic cues, such that themore cues present -- the clearer the boundary between the two intonation units. These prosodic cues include: pitch reset, pause, change induration or ‘‘anacrusis’’ (Cruttenden, 1986:24, 39) (accelerating at beginning of intonation unit and decelerating toward its end), voice quality(initial in-breath, creaky voice toward the end), and the coherence of intonation contour (‘‘one displaying overall gestalt unity’’, Du Bois et al.,1992:100) (for more on this, see Chafe, 1994:53--61; Du Bois et al., 1992:100--114). Each text in the present corpus was transcribed by at least 4transcribers. Only segments on which there was agreement among all transcribers were included in the study.

Y. Maschler, C. Miller Shapiro / Journal of Pragmatics 92 (2016) 43--73 47[(Fig._1)TD$FIG]

Fig. 1. PRAAT spectrogram of excerpt 1, lines 72--74.

2.1.2. Predicative adjectiveThe adjective naxon can also function predicatively, as in the following excerpt, in which Yifaˈt reports to a friend a

conversation she had with another friend:

[TD$INLINE]

Excerpt 2 (‘Drunk on Campus’):

54 Yif'at: ... 'asiti laxavera sheli xishuv.I did to the friend my calculationI calculated for my friend.

55 .. 'amarti la,I said to her,

56 shatit shloshim birot be'erev.you drank thirty beers in [one] night.

57 ..shelexi ted'i 'im ze naxon,that go know if this rightthat who knows whether it’s right,

58 hamispar she'at notenet li.

[TD$INLINE]

the number that you are giving methe number you’re giving me.

We see that naxon functions here as an adjectival predicate in the nominal clause ze naxon (‘it [is] right’), itself acomplement of the preceding clause lexi tedˈi (‘go know’) connected to it via the conjunction ˈim (‘whether’). Thedemonstrative subject pronoun ze (‘it’, line 57) refers anaphorically to ‘thirty beers’ of the previous intonation unit and isthen expanded in the following apposition to the subject hamispar sheˈat notenet li (‘the number you’re giving me’). Thus,naxon evaluates the accuracy of the number of beers the woman claims she drank -- an entity in the extralingual world.

As can be seen in Fig. 2, none of the prosodic cues indicating an intonation unit boundary are found preceding naxon.Again we see that naxon is employed intonation-unit internally, as part of a clause.

Thus, naxon can function as either attributive (excerpt 1) or predicative (excerpt 2) adjective. The predicativeadjective may also function to evaluate an interlocutor’s utterance. In the following family interaction between a middle-aged couple, their son and his girlfriend, the mother is trying to recall the name of someone seen on TV, thinking it is DaliaRabikovitch:

Y. Maschler, C. Miller Shapiro / Journal of Pragmatics 92 (2016) 43--7348[(Fig._2)TD$FIG]

Fig. 2. PRAAT spectrogram of excerpt 2, line 57.

[TD$INLINE]

Excerpt 3 (‘Political Argument’):

407 Mother: da--lia,Dalia,

408 ..hazame--retthe singer[FEM],

409 ..dalia rabikovitch /hazo,Dalia Rabikovitch /the this/[FEM]/this/ Dalia Rabikovitch,

410 nu.self-urging PART

411 Sharon: hi-- meshoreret.she[’s a] poet.

412 Mother: ..lo,no,

413 ..'az lo.so no.

414 ..'az 'ani mitbalbelet. so I[’m] confused.

......

453 Mother: ..Dalia--,

454 ze--,it that’s,


[TD$INLINE]

458 /'aval 'aval/but but

459 ...'aval ze naxonbut it[’s] true

460 Sharon: @ze @lo @meshane.it doesn’t matter.

461 Mother: ze naxon, it [’s] true,

462 sheze meshorer,that it[‘s a] poet,

463 /veze//and it/

464 ze--..bilbel 'oti. it confused me.

455 ..kim'at batuax,almost certain,

456 Father: ...lo.no.

457 Mother: ve--and

At line 411, Sharon corrects her boyfriend’s mother, saying that Dalia Rabikovitch is actually a poet, not a singer, and themother concedes her confusion (lines 412--414). Following an argument between the mother and her husband over thename of this woman (not transcribed here), at lines 453--455 the mother re-attempts to recall the name. At line 459, shereturns to Sharon’s correction that the woman is actually a poet, not a singer, and agrees with it via the clause ze naxon (‘it[is] right’, or, more idiomatically here, ‘it’s true’), composed of the demonstrative subject ze (‘it’) and the adjectival predicatenaxon. Again, no intonation unit boundary is involved between the subject and the predicate. Perhaps because she isinterrupted by Sharon, themother repeats this clause at 461 and then elaborates what the demonstrative subject ze at 461refers to: ze meshorer11 (‘it’s a poet’, line 462). Since the utterance agreed with is a clause rather than a noun phrase, it isconnected to themain clause via the complementizer she- (‘that’): ze naxon,shezemeshorer (‘it’s true, that it’s a poet’). Asit is found in apposition to the subject ze of 461, one might consider sheze meshorer a subject clause.

In this study, ‘epistemic stance’ or ‘epistemic modality’ refer to a speaker’s commitment toward an utterance (Biberet al., 1999; Kärkkäinen, 2003:19). While this token of naxon is still literal in the sense that it refers to a situation in theextralingual world -- that Dalia Rabikovitch is a poet, it is also epistemic because it relates to the speaker’s commitmenttoward her utterance.12 Furthermore, it is also intersubjective (exhibiting ‘‘the explicit expression of speaker’s/writer’sattention to the ‘self’ of addressee/reader’’ (Traugott, 2003:128)), because it reaffirms a statement made by theinterlocutor. Thus, this naxon functions both in the realm of speaker-text relations as well as in the realm of the interactionbetween discourse participants. Over 50% of adjectival naxon tokens in our data (6 out of 11 tokens) are epistemic.

2.2. naxon as adverb

Two additional tokens of naxon (1%) function as adverbs. For example, in excerpt 4, Limor, a student, is telling Ofraabout a conversation she had with another student, who had responded to her ad concerning switching residence halls:

11 Note that althoughDaliaRabikovitz is female, the speaker uses themasculine formmeshorer (‘poet’) andnot the femininemeshoreret (‘poetess’).12 A reviewer asks why can epistemic not be ‘literal’. The epistemic meaning of naxon is obviously related to one of its literal meanings, namely,‘true, which is exactly this way’ (Even-Shoshan, 1986) because the epistemic meaning involves the speaker’s committed stance toward the extentto which an utterance holds true (e.g., whether Dalia Rabikovitch is in fact a poet). However, there is a difference between the literal and theepistemic meanings. While the former relates to a situation in the extralingual world, the latter relates to the realm of the interaction, specifically, tothe relation of speaker to text (cf. Keevallik, 2006).

Excerpt 4 (‘Reznik Dorms’):

18 Limor: ...mitkasheret yom 'exad mishehi,calls day one somebody (FEM)one day some woman calls,

19 'omeret li,says to me,

20 bekesher lamoda'a.[it’s] about the ad.

21 Limor: .. 'az 'ani 'omeret la,so I say to her,

22 'at betuxa she.. 'at hevant naxon ta’moda'a?{---laughing---}

you sure that.. you understood correctly the ad?are you sure you understood the ad correctly?

The adverb naxon at line 22 refers to the extralingual world -- to the manner by which the caller understood Limor’s ad. Ascan be seen in Fig. 3, none of the prosodic cues indicating an intonation unit boundary are found immediately beforenaxon. Structurally, this token, too, is part of a clause and appears intonation-unit internally. Its scope is the predicate ofthe clause, the verb hevant (‘you (FEM) understood’).

2.3. naxon as prototypical discourse marker

All adjectival and adverbial naxon tokens in our corpus have scope within the clause and refer to an entity in theextralingual realm (some of them, as we have seen, are also epistemic). Furthermore, they are always verbalized as partof a larger intonation unit. We now turn to naxon tokens occupying the entire intonation unit. We will argue that thesetokens constitute discourse markers.


[(Fig._3)TD$FIG]



Let us first investigate naxon tokens constituting prototypical discourse markers (Maschler, 2009, see Section 1). Withvery few exceptions (see Tables 2 and 3 in Section 4 below), such tokens serve one of two functions in our data,depending on whether naxon is uttered in non-appeal intonation (Section 2.3.1 below) or in appeal intonation (explored inSection 3.1).

2.3.1. Agreeing/confirming while expressing epistemic stance of certaintyThe great majority of naxon tokens in our data (106, 66%of all naxon tokens) occur in a separate intonation unit, are not

part of a clause, and function to agree with an assertion made by the interlocutor while expressing epistemic stance ofcertainty concerning it.13 Some of these tokens function also to construct the speaker’s epistemic independence and/orauthority (Heritage, 2002; Heritage and Raymond, 2005) concerning the matter discussed. Examine, for instance, thefollowing excerpt from a conversation in which three young Israelis discuss an unusual text from theGemara14 concerningJesus:

[TD$INLINE]

Excerpt 5 (‘Lions’):

36 Yuval: ....nixnas yeshu,entered JesusJesus entered,

37 lekodesh hakodashim,to the holy of holy placesthe Holy of Holies,

38 .... 'axshav,now,

39 ..kol hashem shelo,all the name hishis name,

40 ..harey ze shem yehudi,after all is name Jewishafter all is a Jewish name,

[TD$INLINE]

41 ..hashem shelo ha..'amiti ze yeshuá.the name his the..real is Yeshua

his..real name is Yeshua.

42 Omri: yeshua.Yeshua.

43 Yuval: ... 'im 'áyin.with [the letter] 'ayin.

44 Eynav: naxon.right.

45 Yuval: ...vehafxu 'et ze leyeshu,and they turned DEF OBJ it into Jesus,

46 Eynav: yeshu.Jesus.

13 In 8 of these cases, naxon combines with the negation lo to form the fixed construction lo naxon (‘wrong’, ‘not true’) constructing disagreementwhile expressing epistemic stance of certainty; in 2 additional cases, naxon combines with the discourse marker ’ah (‘oh’). See Miller (2010).14 The Talmud is comprised of the Mishnah, the first major redaction into written form of Jewish oral traditions, and the Gemara, which containsrabbinical commentaries and analysis of the Mishnah.


At lines 39--41, Yuval asserts that Yeshu, the Modern Hebrew name for Jesus, is actually the Jewish name Yeshua in theNew Testament. To emphasize the difference between the two names, at line 43, he adds that Yeshua is spelled with theletter ‘ˈayin’. He employs the Hebrew particle harey (‘after all’, line 40), which, as Ariel (1998) has shown, is employed tosignal the speaker’s stance that the utterance under its scope is not new information for the recipient. Indeed, bothrecipients indicate their prior knowledge concerning the name Yeshua: Omri co-constructs the name Yeshua at line 42,overlapping the final vowel of Yuval’s Yeshua at line 41, and Eynav latches on with naxon (‘right’, line 44). This expressesher agreement and epistemic stance of certainty with respect to Yuval’s assertion, based on her prior knowledgeconcerning the name Yeshua. This naxon, then, also constructs her epistemic independence with regard to the matter inthe interaction.

In comparison to the previous tokens we have seen so far, this token of naxon is not part of a clause. It is employed in aseparate intonation unit, by another speaker, in stand-alone position, in fact, and its scope is the interlocutor’s entireutterance (lines 39--43) -- a discourse segment much larger than a clause. Rather than referring to some entity in theextralingual world, it functions metalingually -- in the realm of the interaction between discourse participants.

The token of naxon at line 44 fulfills both requirements in the definition of prototypical discourse markers: (1)semantically, it functions metalingually in the realm of the interaction between discourse participants in two senses: (a) itconstructs agreement between participants, and (b) it constructs relations of speaker to text (epistemic certainty andindependence). In this sense it may be termed a stance discourse marker. (2) Structurally, naxon is employed intonation-unit initially, at a point of speaker change. We thus classify it as a prototypical discourse marker. Of the 135 naxondiscourse marker tokens in our data, the great majority -- 87% -- are prototypical discourse markers (for non-prototypicaldiscourse markers, see Section 3).

Just as speakers can relate to truth or accuracy in the extralingual world (excerpts 1--4, e.g.), so can they relatemetalingually to the truth or accuracy of an interlocutor’s utterance. This may be accomplished via literal ze naxonfollowing the interlocutor’s utterance, with the masculine singular demonstrative ze (‘this’) anaphorically referring tothis utterance, as we have seen in excerpt 3 above. However, far more often, semantically empty ze is left out, as it isin excerpt 5. In this sense, one might say that discourse marker naxon carries the longer metalingual utterance zenaxon ‘this (/what you just said) is true’ in the background. In the case of epistemic independence, the metalingualutterance in the background would be ‘I know that this is true’; and in the case of epistemic authority -- ‘I confirm thatthis is true’. As in Traugott’s work on the rise of epistemic meanings in English (1989), the change in meaning fromliteral naxon to epistemic naxon can be explained by the conventionalizing of these conversational implicatures(Grice, 1975).

2.4. A grammaticization path for naxon in non-appeal intonation contours

On the basis of our synchronic study thus far, we suggest the following functional itinerary for naxon in non-appealintonation contours:

PredicativeAdj>D:M:

We have argued that metalingual utterances such as ze naxon (‘this [is] right’), in which naxon functions as predicateadjective and the demonstrative refers anaphorically to the interlocutor’s previous discourse, plays a pivotal role in thisgrammaticization process. Interestingly, Méndez-Naya (2007) has found diachronic support for a similar grammaticizationpath for the English discourse marker right from utterances such as that’s right. In Section 6, we present diachronicevidence supporting this path for Hebrew naxon.

3. Analysis of naxon in appeal intonation contours: form and function

All remaining naxon tokens in our data (35, 22%) are verbalized in appeal intonation contours, either occupying theentire intonation unit, or at the beginning of a longer intonation unit. We begin with the tokens occupying the entireintonation unit (Sections 3.1 and 3.2).

3.1. naxon? as prototypical discourse marker: seeking agreement

Seven prototypical naxon discourse markers (4%), all uttered in sentence-final rising intonation contour, do notexpress epistemic stance but rather function to seek the interlocutor’s agreement with the speaker, as in the immediatecontinuation of excerpt 1:


Excerpt 6 (‘Candid Camera’) (continuation of excerpt 1):

75 Boaz: ..bdixa harey tamid hi harbe yoter tova,joke after all always is much more gooda joke after all is always much better,

76 kshehi baheksher.when it[‘s] in the [right] context.

77 ..naxon?right?

78 (6.0) 'axshav,now,

79 Nahum: ... 'ani 'asaper lexa bdixa.I[‘ll] tell you [a] joke.

At lines 75--6, Boaz asserts that a joke is always better when it is verbalized in the right context. His use of harey (‘after all’)manifests his certainty concerning the statement, as well as his expectation that this utterance is not new to theinterlocutor (Ariel, 1998). However, no response is made by the interlocutor, and so at line 77, Boaz seeks agreement withnaxon? (‘right?’) in sentence-final rising intonation. We see that no epistemicity of uncertainty is involved in this token ofnaxon?. Still no positive alignment (Du Bois, 2007) is forthcoming. Following a pause of 6 s, and what looks like an attemptto begin perhaps the next stage of Boaz’ argument (line 78), Nahum announces that he will tell a joke.

The spectrogram (Fig. 4) clearly shows the sentence-final falling intonation contour at the end of baheksher (‘in the[right] context’, line 76) and the sentence-final rising intonation of naxon?. It also shows a significant pitch reset from 105.3at the end of baheksher (line 76) to 267 Hz at the beginning of naxon? (line 77), indicating a clear boundary between thetwo intonation units.

We see that this token of naxon, too, functions metalingually in the realm of the interaction between discourseparticipants. It is employed intonation-unit initially following a sentence-final intonation contour (line 76), and thus fulfillsboth requirements for prototypical discourse-markerhood.

[(Fig._4)TD$FIG]



3.2. naxon? as a non-prototypical discourse marker

We now turn to non-prototypical naxon tokens. Fourteen tokens in appeal intonation (9% of all naxon tokens) fulfill thesemantic requirement for discourse markerhood but not the structural one and are therefore considered non-prototypicaldiscourse markers (Maschler, 2009).15

3.2.1. Seeking confirmation while expressing epistemic stance of uncertaintyThe majority of tokens in this category (11 tokens) function to request the interlocutor’s confirmation, thus expressing

the speaker’s epistemic stance of uncertainty. Examine, for instance, the following excerpt, in which two young men arediscussing patent registration. Gal has just invented some patent, which he intends to register:

Excerpt 7 (‘Patent Registration’):

22 Shay: ..yaxol lihiyot matsav te'oreti,could be situation theoreticalcould there be a theoretical situation,

23 ..she'ata roshem ta’patent,that you register the patentin which you register the patent,

24 vehu kvar kayam?and it already exists [as a registered patent]?

25 .. 'en kaze davar,there’s no such thing,

26 naxon?right?

27 'im 'ata ba lirshom patent kayam,if you come to register patent existingif you come to register an existing patent,

28 Gal: 'im 'ani roshem ta’patent,if I register the patent

if I’m the one registering the patent,

29 ... 'az hu lo kayam.then it not existthen [this means] it doesn’t exist.

At lines 22--24, Shay asks Gal whether in theory it would be possible to register a patent which turns out to already be inexistence as a patent (in another country, for instance). He follows this immediately with the assertion ˈen kaze davar,(‘there’s no such thing’, line 25), followed by the tag question naxon? (‘right?’, line 26) in appeal intonation, with which herequests Gal’s confirmation, as someone experienced in the matter. This naxon, then, functions to request audienceconfirmation. Since the speaker has just manifested his lack of knowledge concerning the matter in his previous question,this naxon also signals the speaker’s epistemic stance of uncertainty, thereby constructing the audience’s epistemicauthority on the matter -- i.e., it is intersubjective. It thus functions metalingually in the realm of the interaction betweendiscourse participants, as well as in the realm of speaker-text relations, and fulfills the semantic requirement for discoursemarkerhood. However, as can be seen from Fig. 5, it does not fulfill the structural requirement for discourse markerhood.The reason for this is the slightly rising continuing intonation contour with which davar (‘thing’) ends (unlike the clearsentence-final falling intonation contour of excerpt 6, line 76 (Fig. 4), e.g.).

In our corpus, the scope of tag-naxon is always the immediately preceding clause uttered by the same speaker in theprevious intonation unit -- in excerpt 7, the clause ˈen kaze davar (‘there’s no such thing’, line 25). Since this previous

15 Three additional non-prototypical discourse markers are employed in non-appeal intonation throughout the data, see Table 2 below and Miller(2010).



intonation unit is verbalized in a continuing intonation contour, we see that the structural requirement for discoursemarkerhood is not fulfilled. We thus classify such tokens of naxon as non-prototypical discourse markers.

The change in meaning from literal naxon to epistemic tag-naxon? can be explained by the conventionalizing ofconversational implicatures again (Traugott, 1989): when a speaker seeks confirmation concerning the truth of anutterance -- in other words, when the longer metalingual utterance ze naxon? ‘is this/what I just said true?’ is in thebackground -- the speaker’s epistemic stance of uncertainty toward it is often implied. Over repetitive usage, this newmeaning may become conventionalized.

3.2.2. Seeking agreementThree tag-naxon tokens in our data (2% of all tokens) function to seek agreement without being epistemic. In the

following interaction, Yafit has just described to three of her friends a social chain she has recently participated in,according to which, upon receiving a particular starter dough from a friend, one lets it rise, then uses a fourth of it to bakebread, which apparently brings one good luck, and gives the other three fourths of the starter dough to three other friends.Yafit is telling this story enthusiastically to her interlocutors, one of whom asks how long this bread chain has been goingon, and Yafit answers:


Excerpt 8 (‘Jerusalem Bread’):

160 Yafit: ...bisvivot hakama 'asrot shanim,around the few tens years roughly a few decades,

161 ..mistovev halexem hayerushalmi haze.circulates the bread the Jerusalem the thisthis Jerusalem bread has been passing around.

162 Lihi: ..'ani behelem.I in shockI’m amazed.

163 Yafit: ..magniv,cool,

164 naxon?right?

165 Lihi: ..mamash.totally.

Lihi is highly impressed by the length of time this social chain has been going on (ˈani behelem ‘I’mamazed’, line 162), andYafit shares her enthusiasm with magniv, naxon? (‘cool, right?). This tag-naxon has scope over the preceding clause(comprised of the predicate magniv (‘cool’)), and functions to seek Lihi’s agreement that this bread chain is ‘cool’. Indeed,Lihi’s enthusiastic agreement is soon to follow with mamash (‘totally’, line 165). Since Yafit has been enthusiasticallydescribing this custom to her friends in the immediately preceding discourse, there is no uncertainty involved in her tag-naxon. This token, then, is not employed epistemically, but rather functions metalingually in the realm of interpersonalrelations among speakers. Because of the continuing intonation of line 163, it does not fulfill the structural requirement fordiscourse markerhood and is therefore considered a non-prototypical discourse marker.

3.3. naxon in xagam constructions: seeking agreement

Three naxon tokens in our corpus (2%, all from the same stretch of discourse) occur in xagam constructions, i.e., in the[Predicate + Subject Clause] construction introduced in Section 1. According to traditional grammar, when a finite subjectclause is involved, it must be connected to the main predicate via a complementizer (see Nir, 2013), as in our firstexample:

tov she'asita halixa hayomgood MASC SG that you did [a] walk todayit’s good that you walked today

Indeed, we find the complementizer she- (‘that’) connecting the subject finite clause ˈasita halixa hayom (‘you walkedtoday’) to the predicate tov (‘good’).

When naxon appears in this construction in our corpus, it is always in appeal intonation and functions to seekagreement from the interlocutor. In the following interaction between three girlfriends, Vicky is debating whether or not toreturn a pair of pants she had bought. On the one hand, it was very expensive; on the other, she finds the pants really‘cool’:


Excerpt 9 (‘Pants Dilemma’):

65 Vicky: 'az madadeti 'oto,so I tried on itso I tried it on,

66 vehu mamash hidlik 'oti.and it really turned on meand it really turned me on.

67 Lily: lo xoshevet shekeday lax.not think that it’s worthwhile to you[I] don’t think you should [return it].

68 Maya: ...@

69 Vicky: ....naxon shehuright that it

70 Maya: hidlik 'otax?turned on youturned you on?

71 Vicky: ...naxon shehu madlik?right that it cool[isn’t it] true that it [the pair of pants] is cool?

72 Lily: 'asa lax 'et ze?did for you ACC itdid it for you?

73 Vicky: ... 'azvi kama hu 'ole.let go how much it costsforget how much it costs.

74 ...naxon shehu madlik?right that it cool[isn’t it] true that it’s cool?

75 Maya: ...hu yafe--,it[‘s] pretty,

76 ..ke--n,yes,

77 'aval hu mamash lo shave me'a 'esrim shekel.but it really not worth hundred twenty shekelbut it’s really not worth a hundred and twenty shekels.

Seeking agreement from her two interlocutors that the pants are ‘really cool’, Vicky employs naxon in a xagamconstruction three times here (the first two in a case of self-repair (cf. Fox et al., 2010) at lines 69, 71, possibly becauseof the overlap at 69--70): naxon shehu madlik? (lit. [isn’t it] true that it’s cool?’, or, more idiomatically, ‘isn’t it cool?’). Thesubject clause hu madlik ‘it’s cool’ is connected to the predicate naxon via she- (‘that’). Indeed, in response, Maya, whohad been teasing Vicky at line 70, agrees that ‘it’s pretty’ (lines 75--76), but claims that it’s not worth the price (lines 77). Noepistemics of uncertainty are involved in Vicky’s seeking agreement via these naxon xagam constructions, as in line 66she clearly asserts her liking the pants.

The function of this xagam naxon token bears a striking resemblance to the tag-naxon of excerpt 8. Because of thegreat similarity between the two predicates magniv and madlik, both of which are slang expressions for English ‘cool’ andshare verbal morphology patterning (binyan), these utterances form almost a minimal pair syntactically:

Y. Maschler, C. Miller Shapiro / Journal of Pragmatics 92 (2016) 43--7358[(Fig._6)TD$FIG]


Excerpt 8: magniv, naxon? (‘cool, right?)Excerpt 9: naxon shehu madlik? (‘right that it’s cool’?)

Whereas the token in Excerpt 8 has scope over the preceding clause, this naxon token has scope over the followingclause. However, here the predicate and the clause it has scope over appear within the same intonation unit. As can beseen in Fig. 6, naxon occurs intonation-unit initially, as part of a longer intonation unit which ends with rising intonation onthe final syllable of madlik (‘cool’).

3.4. naxon in quasi-xagam constructions

Nine more tokens of naxon (6%), all in appeal intonation, appear as predicates in constructions resembling the xagamconstruction, except that they are connected to the following subject clause asyndetically and will therefore be termed‘quasi-xagam naxon’ tokens. They often carry a function identical to the function we have seen in Section 3.2.1 for



tag-naxon; namely, seeking confirmation while expressing epistemic stance of uncertainty, but they may also functionnon-epistemically. We examine them in more detail below.

3.4.1. Seeking confirmation while expressing epistemic stance of uncertaintyExamine the following excerpt from a conversation in which Shira reports sitting in the back seat of a very expensive

jeep of a friend, while her partner Shaxar was sitting next to the owner and driving:

[TD$INLINE]

Excerpt 10 (‘Jeep’):

97 Shira: 'etmol 'amarti lo ba'erev,yesterday I said to him in the eveningyesterday evening I said to him,

98 ..Shaxar,

99 naxon nora hitragashta she--..n..nasata?right extremely you were nervous that ..y..you drove

[isn’t it] true [that] you were extremely nervous when you drove?

100 hu 'omer,he says,

101 lama,why,

[TD$INLINE]

102 ra'u 'alay?why see 3 MASC PL on mewas it noticeable?

At lines 98--99, Shira recounts her question to Shaxar later that evening after the ride: naxon nora hitragashta she--.. n..nasata? (‘isn’t it true [that] you were extremely nervous when you drove?, or, more idiomatically, ‘weren’t you extremelynervous when you drove?’). Shaxar obviously has greater access to his feelings than Shira, and Shira therefore turns tohim for confirmation. This move simultaneously lowers Shira’s epistemic authority on the matter while raising Shaxar’s -- i.e., this naxon, too, is intersubjective. Syntactically, this utterance begins with naxon functioning as predicate of theasyndetic subject clause nora hitragashta she--.. n.. nasata (‘you were extremely nervous when you drove?’). Fig. 7shows that naxon is employed intonation-unit initially, as part of a larger intonation unit. There is self-repair in the midst ofthis intonation unit, with recycling of the first consonant of the final word (n..nasata, ‘you drove’), but it is still possible todetect the rising intonation contour of this unit:

Here we see naxon having scope over a clause (itself embedding an adverbial clause, albeit ungrammatically,according to traditional grammar (see Maschler, forthcoming)), with a function identical to that of tag-naxon -- seeking theinterlocutor’s confirmation while expressing epistemic stance of uncertainty. In both cases, naxon has syntactic scopeover the entire clause and occurs at one of its ends -- either initially (quasi-xagam naxon) or finally (tag-naxon). This isunlike the literal cases, which occur clause-internally and have scope over a clause component, and unlike the discourse


marker cases, which occur in a separate intonation unit, are not part of any clause, and generally have scope over adiscourse segment larger than a clause.16

3.4.2. Seeking confirmation of referent identificationNot all quasi-xagam naxon tokens are epistemic. They can also be employed to seek agreement (cf. excerpts 6 and 8),

or, as in the following example, to seek confirmation concerning identification of a referent which will become topical in theimmediately ensuing discourse. For instance, in the following excerpt, a couple, Avi and Tami, co-tell Ortal about Tami’sfather’s unusual manner of reading the Haggadah, the text read at the Passover meal around the dinner table:

[TD$INLINE]

Excerpt 11 (‘Passover Haggadah’):

135 Avi: .....gam matsxik,[it's] also funny,

136 ..'aba sh’la kore ta’hagada,father her reads the Haggadahher father reads the Haggadah,

137 ...venaxon tamid yesh.. bentayim,and right always there are..meanwhile,

and [isn’t it] true [that] there are always meanwhile,

[TD$INLINE]

138 ..beyn hashurot,between the lines,

139 Tami: he--'arot ka'ele.comments like thesecomments kinda.

140 Avi: she'omrim ma that mention what

tsarix la'asot 'axrey ze? need to do after this needs to be done next?

141 Ortal: ken.yes.

142 Avi: ...hu kore 'et ze,he reads it,

143 ke'ilu ze hemshex.as if it continuation. as if it’s the continuation [of the actual text].

144 Tami/Ortal?: ...(laughter)

145 Avi: mevina?y’get it?

146 Ortal: ..gadol!big[that’s] hysterical!

147 All?: {laughter}

16 The confirmation/agreement-seeking function of xagam-naxon has even crossed typologically-unrelated language boundaries. In YaelMaschler’s home, where Hebrew-English bilingual children were raised, one often hears English constructions such as ‘right we’re going out fordinner?’, also when these children have become young adults.


The Passover Haggadah consists of excerpts to be read at the dinner table, interspersed with comments detailing certainactions to be performed in association with the reading (such as pointing to or raising certain foods while reading particularexcerpts, getting up to wash one’s hands before continuing the reading, etc.). By employing quasi-xagam naxon at lines137--140, Avi first draws his audience’s attention to the existence of these comments: naxon tamid yesh [...], beynhashurot, he--ˈarot kaˈele. sheˈomrim ma tsarix laˈasotˈaxrey ze? (‘[isn’t it] true [that] there are always, between the lines,comments kinda. that mention what needs to be done next?’). (He is aided by Tami, who co-constructs (Lerner, 1991) line139 for him). A more idiomatic English translation would be ‘there are those comments between the lines, that mentionwhat needs to be done next, right?’ (the English equivalent of tag-naxon!), or ‘you know those comments between thelines, that mention what needs to be done next?’. Once Avi has obtained Ortal’s confirmation with her ken (‘yes’, line 141)indicating that she knows which comments he is referring to, he is able to make his assertion concerning those comments-- that Tami’s father reads them as if they were part of the text (lines 142--3). This will ensure her appreciation of his point,which is indeed forthcoming with Ortal’s gadol! (‘[that’s] hysterical!’, line 146, and the laughter at lines 144, 147). Thisnaxon, then, draws attention and requests the audience’s confirmation concerning identification of a referent which willbecome topical in the immediately ensuing discourse. It ‘fronts’ this discourse topic in order to then make an assertionabout it. No epistemicity is involved

3.4.3. The status of quasi-xagam naxon as predicateWe have seen that quasi-xagam naxon occurs clause-initially as part of a larger intonation unit and has scope over a

clause rather than a larger discourse unit. We are not concerned with a discourse marker (prototypical or not) here. Somemight consider quasi-xagam naxon a predicate of an immediately following subject clause. However, the lack of thecomplementizer she- preceding the subject clause calls into question the status of such naxon tokens as predicates,because the asyndetic combining diminishes the connection between this ‘predicate’ and its ‘subject clause’ (cf.Thompson and Mulac, 1991a,b; Thompson, 2002 for English ‘that’).

Rather, our data suggest that naxon in these quasi-xagam constructions functions less as a predicate and more as aprojecting element (Auer, 2005; Günthner, 2011) external to the clause, the function of which is to foreshadow anassertion requiring confirmation/agreement. Recent studies of clause-combining in English, German, and French(Hopper and Thompson, 2008; Günthner, 2011; Pekarek Doehler, 2011) show that part A of what is traditionallyconsidered a bi-clausal construction (e.g., wh-clefts, extrapositions, if clauses, pseudo-clefts, French il y apresentational clefts) is often better understood as a ‘projecting construction’ -- a ‘prefab’ that projects certain typesof actions to come, which occupy a syntactically independent stretch of discourse. Similarly, quasi-xagam naxon isbetter understood not as part A (the predicate of the matrix clause) of a bi-clausal construction in which one clause isembedded as the subject of another, but rather, as a ‘prefab’ external to the clause, projecting an assertion requiringconfirmation/agreement.

This syntactic status of quasi-xagam naxon in our data raises the question of whether discourse marker naxon, whichis always external to the clause, has perhaps developed from quasi-xagam naxon. Since, as we have seen, quasi-xagamnaxon tokens in our corpus all involve seeking functions which are significantly distinct from the asserting functions ofdiscourse marker naxon, our synchronic analysis does not provide support for such a functional itinerary. In Section 6, weconsider this possible functional itinerary in light of diachronic data.

3.5. A grammaticization path for naxon in appeal intonation contours

On the basis of our synchronic study thus far, we suggest the following functional itinerary for naxon uttered in appealintonation contours:

xagam> quasi-xagam

We support this grammaticization path with diachronic evidence in Section 6.When seeking confirmation of an assertion one is uncertain about, or when seeking agreement for an assertion, a

Hebrew speaker has two possible strategies which are distinguished in terms of their temporality of production (Auer,2009; Hopper, 2011; Depperman and Günthner, 2015). One can first make the assertion requiring confirmation/agreement and then ask for it via a metalingual utterance such as ze naxon? (‘is this true?’), where the demonstrative ze(‘this’) refers anaphorically to the assertion. However, as is the case with ze naxon in non-appeal intonation (e.g., excerpt5), far more often, semantically empty ze is left out. Thus, just as the non-appeal assertion ze naxon plays a pivotal role inthe grammaticization of discourse marker naxon (Section 2), we would like to suggest that the anaphoric question zenaxon?, in which naxon functions as adjectival predicate, plays a pivotal role in the grammaticization of the non-prototypical discourse marker tag-naxon.


An alternative strategy available to a Hebrew speaker for seeking confirmation/agreement of an assertion is to projectthe assertion requiring confirmation/agreement by beginning the utterance with a (quasi-)xagam naxon. The differenttemporality of production natures of the two strategies strongly suggests that (quasi-)xagam naxon is an outcome of analtogether different functional itinerary than that of discourse marker naxon. This will be supported diachronically inSection 6.

4. Summary: form and function of naxon

The analysis presented in Sections 2 and 3 shows that there is no one-to-one correspondence between form andfunction of naxon in our corpus, but some tendencies can be discerned. Table 2 summarizes the distribution of uses ofnon-appeal naxon tokens as a function of syntactic category,17 while Table 3 does so for appeal naxon tokens:

Table 3naxon uses as a function of syntactic category (appeal intonation contours).

naxon in xagamconstructions

naxon in quasi-xagamconstructions

Non-prototypical discoursemarkers (tag-naxon)

Prototypicaldiscourse markers

Seeking agreement 3 3 3 7Seeking confirmation + epistemicity 2 11 1Seeking confirmation of referent identification 4 1

Total 3 (8%) 9 (26%) 14 (40%) 9 (26%)

Grand total = 35 (100%)

Table 2naxon uses as a function of syntactic category (non-appeal intonation contours).

Adjectives/Adverbs Non-prototypical discourse markers Prototypical discourse markers

Literal 6Agreement/Confirmation + epistemicity 7 3 (non-tag) 106Other 3

Total 13 (10%) 3 (2%) 109 (88%)

Grand total = 125 (100%)

These two tables show, first of all, that there are almost four times as many naxon tokens in non-appeal intonationthroughout the data (125/160, 78%) as compared with those in appeal intonation (35/160, 22%). Of these tokens(Table 1), by far the great majority (106/125, 85%) constitute prototypical discourse markers whose function isgenerally confirming or agreeing with the addressee while expressing epistemic stance of certainty.

Second, no adjectival or adverbial naxon tokens are uttered in appeal intonation (Table 3). On the other hand, allxagam-naxon and quasi-xagam naxon tokens in the corpus are verbalized in appeal intonation contours.

Third, as can be seen in Table 3, all naxon tokens in appeal intonation contours seek agreement and/or confirmationof some kind from the interlocutor. Epistemicity (of uncertainty) is involved in 40% of these cases (second row of Table 3:2 + 11 + 1 = 14; i.e., 14 out of 35, 40%), and this is accomplished most often (11 tokens) via a non-prototypicaldiscoursemarker consisting of tag-naxon. Prototypical discourse markers in appeal intonation are employed most often(7 out of 9 tokens) to seek agreement in cases not involving epistemicity.

Apart from 6 literal tokens, the functions of naxon in both tables all have to do with agreement and/or confirmation --either asserting it, in which case epistemic stance of certainty is almost always involved (often accompanied byepistemic independence and/or authority), or seeking it, in which case epistemic stance of uncertainty may beinvolved (in 40% of the cases). Thus, whether agreement/confirmation is asserted or sought, and what type ofepistemicity is constructed (certainty vs. uncertainty), is strongly linked to the intonation contour in which naxon isverbalized. In this particular corpus of casual everyday conversation, naxon is employed roughly four times more

17 The category of ‘Other’ in Table 2 consists of 3 tokens of the lo naxon (‘not right’) construction, see Miller (2010).


Table 4Prosodic, syntactic, semantic, and pragmatic properties of naxon.

Position within I.U. Syntacticscope

Reference (Inter-)subjective

Functions

Adj, Adv Internal(non-separate I.U.)

Clausecomponent

a. Extralingualb. Extralingual,metalingual

a. �b. +

a. Literalb. Agreeing/confirming + epistemicity

naxon in xagamconstructions

Initial(non-separate I.U.)

Followingsubjectclause

Metalingual + Seeking agreement

naxon in quasi-xagamconstructions

Initial(non-separate I.U.)

Followingsubjectclause

Metalingual + a. Seeking agreementb. Seeking confirmation + epistemicityc. Seeking confirmation of referentidentification

Non-Prototypical DM(tag-naxon?)

Initial (separate I.U.),following continuingintonation insame-speaker talk

Precedingclause

Metalingual + a. Seeking agreementb. Seeking confirmation + epistemicity

Prototypical DM Initial (separate I.U.) Discourse Metalingual + a. Agreeing/confirming + epistemicityb. Seeking agreement

often (78%) in order to assert, rather than seek agreement/confirmation, thereby constructing an epistemic stance ofcertainty.

We see that the prosody with which naxon is verbalized has implications for its uses, and consequently, it is argued, forthe functional itinerary followed by it. In a functionalist approach to the emergence of grammar from discourse patterns, it isunlikely that a linguistic element functioningmostly in non-appeal intonation would evolve from one hardly ever functioningin non-appeal intonation. This underscores the utmost importance of taking into account intonation patterns whenstudying grammaticization paths.

5. Issues of left and right periphery in the study of discourse markers

Traugott’s seminal work on the role of pragmatics in semantic change (e.g., Traugott, 1982, 1989, 1995a,b, 2003;Traugott and Dasher, 2002) has shown that as utterances become grammaticized, they increase in metatextuality,subjectivity, and intersubjectivity. In other words, they move from focusing on the extralingual world to focusing on therealm of the interaction, with its metatextual structure, procedures for organizing that structure, and interactional aspectssuch as subjectivity and intersubjectivity. In the process, these utterances undergo an increase in scope, from an elementwhose scope is within the clause, to one whose scope is over the entire clause, and finally to one whose scope is over adiscourse segment. Both grammaticization paths argued for in the present study support Traugott’s argument. Note that inthe dialogic approach (Du Bois, 2007) taken in the present study, the scope of the grammaticized element need notnecessarily be the discourse of a single speaker. Furthermore, the scope can be entirely the discourse of the interlocutor,as in stand-alone tokens such as in excerpt 5.18

Table 4 summarizes the prosodic, syntactic, semantic, and pragmatic properties of all naxon tokens in both appeal andnon-appeal intonation contours, in relation to Traugott’s claims:

Table 4 shows that adjectival or adverbial naxon (first row of Table 4) is always employed intonation-unit internally aspart of a larger intonation unit. Its syntactic scope is over a clause component and it refers literally to the truth or accuracyof some entity in the extralingual world (excerpts 1, 2, 4). However, when it appears as predicative adjective in theconstruction ze naxon (‘this [is] true’) following the talk of an interlocutor, it begins to acquire epistemic uses, functioning

18 Such stand-alone grammaticized tokens have been explored in other synchronic studies of grammaticization. See, e.g., the grammaticizationof the Hebrew adverb beˈemet (‘in truth’) into a discourse marker (‘really, actually, indeed’) (Maschler and Estlein, 2008), or the grammaticizationof the Proto-Germanic temporal adverb *nū (‘now’), into a discourse marker (NU/NÅ) in a variety of European and -- through language contact --non-European languages (e.g., English now, German nu(n) and na, Dutch nou, Swedish/Norwegian nå, Icelandic nù, Russian/Yiddish/Hebrewnu, Polish/Czech/Slovak no, Finnish no, Estonian no(o)(h)), (See Maschler and Dori-Hacohen, in press and the collection of studies in Auer andMaschler, in press).


metalingually, subjectively, and intersubjectively (excerpt 3). The origins of the non-literal functions of naxon can thusbe detected already in its adjectival uses.

In the prototypical discourse marker uses (fifth row of Table 4), naxon occurs intonation unit initially, almost alwaysoccupying the entire intonation unit,19 has scope over the discourse, and functions (a) as an agreeing/confirmingepistemic token expressing certainty (excerpt 5) or, (b) in appeal-intonation, as an agreement-seeking token20 (excerpt 6).In both of these functions, naxon refersmetalingually to the interaction between participants, i.e., it is intersubjective. In theformer function, it refers also to the relation of speaker to text (expressing certainty), i.e., it is subjective.

Moving now to the second, third, and fourth rows of Table 4, we have seen that xagam-naxon, quasi-xagam naxon andtag-naxon tokens all have scope over an entire clause (either preceding or following naxon). They all refer metalinguallyrather than extralingually and may function subjectively (expressing the speaker’s epistemic stance of uncertainty whileseeking confirmation). Because they always seek response from the addressee, these tokens, too, always functionintersubjectively. Tokens in (quasi-)xagam constructions appear at the left periphery of the clause, whereas tag-naxontokens appear at the right periphery.21

Recent studies building on Traugott’s work argue for a robust asymmetric tendency in a variety of languages,according to which elements recruited in this process from clause-medial position to the left and right peripheries of theclause tend to fulfill different functions: left periphery elements usually serve metatextual and subjective functions,whereas right periphery elements tend to serve intersubjective ones (e.g., Degand and Fagard, 2011, the collection ofstudies in Beeching and Detges, 2014).

However, our study shows that Hebrew naxon may carry intersubjective functions at the two clause peripheries: (a)seeking agreement (e.g., excerpts 8 and 9) and (b) seeking confirmation while expressing epistemic stance of uncertainty(e.g., excerpts 7 and 10). Since seeking a response from the interlocutor is always intersubjective, naxon at bothperipheries is intersubjective. Furthermore, when expressing epistemic stance of uncertainty, naxon can also functionsubjectively at both of these positions (e.g., excerpts 7, 10). In conclusion, left and right periphery naxon are bothalways intersubjective, and they can both also be subjective. Our study thus provides counter-evidence from aSemitic language for the asymmetric hypothesis concerning the left and right peripheries (see also Traugott, 2012 forcounter-evidence from Present Day English no doubt and surely, Van der Wal, 2013 for counter-evidence from the Bantulanguage Makhuwa).

6. Discussion: grammaticization of naxon

Hopper and Traugott define grammaticization -- or rather grammaticalization, a term which we have avoided in thisstudy (see footnote 3) -- as ‘‘the change whereby lexical items and constructions come in certain linguistic contexts toserve grammatical functions and, once grammaticalized, continue to develop new grammatical functions’’ (2003:xv). LikeTraugott (2007), we construe grammaticization as the development of grammatical material among which is included ‘‘thematerial that signals speakers’ perspective on the relationship [...] of utterances to each other and the beliefs of speakersand hearers (modal and discourse markers)’’ (Traugott, 2007:151).

A typical path of grammaticization follows a cline of de-categorialization, whereby, starting out as belonging to a majorgrammatical category such as noun or verb, a lexical itemmight lose themorphological and syntactic features identifying itas a full member of that particular grammatical category (Hopper and Traugott, 2003:106--109). ‘‘In its most extreme formsuch a change is manifested as a cline of categoriality, statable as:

19 Ap20 Onidentifi21 Thiconcepclause

major category (> adjective/adverb) > minor category’’ (Hopper and Traugott, 2003:107).

Adjective and adverb thus constitute an intermediate category, whereas the term ‘minor category’ includesgrammatical categories such as preposition, conjunction, as well as discourse marker (as shown in Ferrara, 1997).

Wewould like to suggest that naxon started out in Biblical Hebrew as a lexical item in themajor grammatical category ofverb, and has gradually acquired grammatical functions, so that nowadays it functions first and foremost as a discoursemarker. In the process, naxon underwent de-categorialization from verb to prototypical discourse marker: naxon is rarelyemployed as a verb in Modern Hebrew (only in written discourse of very high register) -- no tokens were found in our

art from a few tokens appearing with either lo (‘not’) or ’ah (‘oh’) in the same intonation unit, see footnote 13.e token of a prototypical discourse marker seeking agreement also expresses uncertainty, and another one seeks confirmation of referentcation (cf. excerpt 11), see Table 3.s terminology is, of course, highly biased toward written Western languages (Hebrew, e.g., is written right to left). A more accuratetualization would relate to the termporal nature of production of an utterance, using labels such as ‘preceding/following verbalization of the’ rather than ‘left/right’.


corpus. Following another functional itinerary, naxon underwent de-categorialization to a xagam-form, and then to aquasi-xagam form.

As for the first grammaticization path from verb to prototypical discourse maker, we have seen evidence for thePredicative Adj > D.M. stage of this functional itinerary in the synchronic analysis above (Section 2). In what follows,we first bring evidence for the earlier stage of this grammaticization path (V > Adj) and also support our claimsconcerning the Adj > D.M. stage with data extracted from (1) the Judaic Responsa Project (including the Bible (10th

century B.C. -- 1st century A.D.), all written sources of Mishnaic Hebrew (1st -- 4th century A.D.), and Maimonides’writings in Medieval Hebrew (1170--1180 A.D.)) and (2) the Historical Dictionary compiled by the Academy of theHebrew Language (a variety of Hebrew texts from 200 B.C. until 2000 A.D.). We then turn to the secondgrammaticization path.

In Biblical Hebrew, we find naxon in the grammatical categories of both adjective and verb. As a verb, Biblical naxonmay be employed in four meanings: ‘to be accurate and true’, ‘to be based, established’, ‘to be strong, exist’, ‘to beprepared, ready’ (Even-Shoshan, 1986). Like other Hebrew verbs, it conjugates according to tense, person, number, andgender; e.g.:

Excerpt 12:

vehamamlaxa naxona beyad shlomoand-the-kingdom.FEM be based.3 FEM SG PAST in-hand Solomon ‘and the kingdom was firmly based in the hands of Solomon.’ (Kings 1,2, 46)

In this case, the verb naxona (‘was based’) is declined for tense, person, gender, and number, agreeing with the subject ofthe clause mamlaxa (‘kingdom’).

As an adjective, Biblical naxon comprises the present tense of the verb and may be employed in three meanings: ‘true,which is exactly this way’, ‘based, established’, and ‘ready, prepared’. Like other Hebrew adjectives, it declines accordingto gender and number, e.g.:

Excerpt 13:

vehayu nexonim layom hashlishiand-be.2 MASC PL IMPERT ready.MASC PL for-the-day the-third‘and be ready for the third day’ (Exodus 19, 11).

Here the adjective nexonim (‘ready’) appears in masculine plural as predicate complement of the verb ‘to be’ in theimperative.

The most productive way of forming Hebrew adjectives is to employ the present tense of a verb. We see thatnaxon, like all Modern Hebrew adjectives derived from verbs, has lost its morphological properties that would identifyit as a full member of the major grammatical category of verb (in particular, the possibility of declining according totense) and has thus undergone decategorialization first to the intermediate category (Hopper and Traugott, 2003:107)of adjective.

Recall that in our corpus, only the masculine singular form of this adjective is employed. This is evidence for further de-categorialization, since in our corpus, even as an adjective, naxon does not manifest its full declension.

De-categorialization of naxon did not stop here. We now turn to the second functional itinerary followed by naxon.Further loss of morphosyntactic properties occurred as the adjective naxon fossilized into an indeclinable functioning aspredicate of the clause in a xagam construction (e.g., excerpt 9). Furthermore, as we have seen in excerpts 10 and 11,naxon continued to undergo de-categorialization and lost the syntactic property of being connected to the subject clausevia a complementizer, thus becoming an asyndetic quasi-xagam form.22

We have seen synchronic evidence for the functional itinerary xagam > quasi-xagam (Section 3). We find supportfor this stage of grammaticization of naxon, as well as for the stage Predicative Adj > xagam, also in the diachrony ofnaxon. Searching the Judaic Responsa Project, we found no xagam-naxon tokens in Biblical or Mishnaic Hebrew.

22 A reviewer points out that according to Norde (2009), an increase in syntactic freedom is considered a case of degrammaticalization. We donot follow Norde, 2009 in this view, but rather the many studies which have shown that in the process of grammaticization, forms may becomesyntactically detached and ‘move’ to a position which is external to the clause (Thompson and Mulac, 1991a,b; Thompson, 2002; Traugott andDasher, 2002; Kärkkäinen, 2003; Günthner, 2011, to mention just very few). It is now fairly widely accepted that the development of discoursemarkers often involves grammaticization. In the process, such markers have, of course, decreased in their syntactic dependency on clause-internal elements.


According to Abramson (1998:62), xagam-naxon appeared in the language only in Modern Hebrew; i.e., beginning inthe late 19th century. However, a search in the Historical Dictionary shows that it appeared close to a thousand yearsearlier. The earliest unambiguous token in this database comes from a Cairo Genizah fragment from 922 A.D. ofRabbi Saadia Gaon’s polemic against another rabbi concerning when in the Jewish calendar a particular holidayshould be set for:

Excerpt 14:

uve'exad beshabat 'eyn naxon lekab'o. and-in-one in-Sabbath not right to-set-ACC MASC SGand for a Sunday it is not right to set it.

lefixax yidaxe lesheni beshabat.therefore be postponed.3 MASC SG FUT to-second in-Sabbaththerefore it will be postponed to a Monday.

(Rabbi Saadia Gaon 922: 548).

We see that naxon (preceded by the negator ˈeyn) functions here as predicate of the non-finite subject clause lekabˈo beˈexad beshabat (‘to set it for a Sunday’). Recall that no complementizer is necessary in the case of non-finite subjectclauses23 (Section 3.3).

The earliest token found in the Historical Dictionary with a finite verb comes from a Biblical commentary written around1050 A.D.:

Excerpt 15:

lo yihiye naxon she-yexpats hakravato.not be.3 MASC SG FUT right that-want.3 MASC SG FUT offering-his ‘it will not be right that he will have an inclination for his offering.’

(Ben Moshe 1050: 5, 16).

Here naxon (preceded by the negator lo) functions as predicate of the following subject clause yexpats hakravato (‘he willhave an inclination for his offering’). The predicate is connected to its subject clause via the complementizer she- (‘that’).Recall that all xagam forms are indeclinables. Thus, in order to express the predication in future tense, a periphrastic formis used with the 3rd person MASC SG FUT form of the verb ‘to be’ (yihiye) preceding xagam-naxon, as would be the case alsoin Modern Hebrew.

Not a single asyndetic quasi-xagam naxon token with a finite verb was found in any of the historical databases, noteven in the Modern Hebrew texts included in the Historical Dictionary. This form, which we have argued is a projectingconstruction external to the clause (rather than a predicate in a xagam construction), is a fairly recent Modern Hebrewdevelopment. We thus have also diachronic evidence for the xagam > quasi-xagam stage of this functional itinerary. Justlike other syntactic constructions, the xagam construction continues to evolve in the language as a result of the usepeople make of it in interaction. We have argued that from predicate of a matrix clause in a bi-clausal construction(xagam-naxon), naxon has evolved into a projecting construction the function of which is to foreshadow an assertionrequiring agreement/confirmation. The fact that the subordinator she- is no longer verbalized loosens the nexus betweenthis ‘predicate’ and its ‘subject clause’.

Both xagam-naxon and quasi-xagam naxon have scope over the following clause. We have seen that discoursemarker naxon (both prototypical and non-prototypical) no longer functions as part of a clause, but rather has scope over alarger discourse segment. Therefore, following Traugott’s argument about an increase in scope in the process ofgrammaticization of discourse markers (Section 5), discourse marker naxon might seem to be a further development ofthe quasi-xagam form. However, we cannot argue for a grammaticization path from quasi-xagam naxon to discoursemarker naxon on the basis of either synchronic or diachronic evidence.

With respect to synchrony, we have seen strong evidence against such a grammaticization path in the two differentfunctional itineraries presented in Sections 2 and 3. With respect to diachrony, not a single case of discourse markernaxon is found in the Bible. Neither have we found any in all written sources of Mishnaic Hebrew or in Maimonides’

23 Note that part of the clause that this naxon has scope over precedes the predicate -- the adverbial complement beˈexad beshabat (‘for aSunday’).


source of Medieval Hebrew. A later edition of Even-Shoshan’s dictionary (2003) adds two discourse marker24 uses fornaxon, which he did not include in the 1986 edition of his dictionary: (1) milat haskama (‘a word of agreement’) (2)milat xiyuv bitshuva lisheˈela(‘an affirmative answer to a question’). Even-Shoshan dates these two uses to ModernHebrew.

Indeed, the earliest discourse marker use documented in the Historical Dictionary comes from a Hebrew playpublished in 1911 by the Ukrainian-born writer Brenner (1881--1921) in the dialogue of the lady of the house to her servant,Fritz:

Excerpt 16:

<Lisa Bensch>: frits! lex na 'el xadar hamevashlot, frits!Fritz! go please to room the cooks, Fritz!Fritz, please go to the cooks’ room, Fritz!

nitkablu kosot xadashot bishvil yeyn shampanya;were received glasses new for wine Champagne new glasses for Champagne wine were received;

'ani xoshevet sheha'adonim yishtu hayom yeyn shampanya.I think that the gentlemen will drink today wine ChampagneI think the gentlemen will drink Champagne wine today.

<Fritz>: naxon! be'oneg, ha'alma bensh.right! with pleasure, Miss Bensch.

(Brenner 1911).

Employing naxon in exclamatory intonation (as indicated by Brenner’s punctuation), Fritz responds affirmatively to MissBensch’s request to go fetch the Champagne glasses, adding beˈoneg, haˈalma bensh (‘with pleasure, Miss Bensch’).Modern Hebrew does not employ naxon in this manner, and so it seems that the author, writing at a time in which Hebrewis just beginning to be revived as a spoken language, is translating here English ‘right’ (or its ‘equivalent’ in Ukrainian orYiddish) which agrees to perform an action. In any event, this token of naxon clearly appears outside the clause and fulfillsboth requirements for prototypical discourse markerhood (Maschler, 2009).

Since xagam-naxon is documented already from at least 922 A.D., relying on these historical data, onemight concludethat it took over a thousand years (during which Hebrew was not used much as a spoken language) for naxon to emergeas a discoursemarker. However, we would like to argue that there is insufficient evidence tomake such a claim. First of all,we must point out the extremely different natures of our spoken and written databases. One cannot compare casualspoken discourse with written discourse of genres such as holy scriptures, commentaries on holy scriptures, oral Jewishlaw, literature, or even plays. We have no way of ascertaining that naxon was never employed in order to agree with aninterlocutor in earlier periods in which Hebrew was spoken. The historical databases are in fact inadequate when it comesto describing the way naxon evolved as a discourse marker: excerpt 16 is one of very few discourse marker naxon tokensin the Historical Dictionary, even though the database continues well into the twentieth century. This is in clearcontradistinction to the fact that in the spoken corpus, 74% of all naxon tokens constitute prototypical discourse markers.Indeed, excerpt 16 comes from a play, a dialogic written genre somewhat reminiscent of conversation, but the greatmajority of sources in the historical databases do not constitute plays.

Furthermore, there is another major difference between the historical data concerning xagam-naxon and the (quasi-)xagam naxon tokens found in our corpus. Recall that in the spoken corpus all (quasi-)xagam-naxon tokens were uttered inappeal intonation contours, seeking agreement/confirmation from the interlocutor. None of the xagam-naxon tokens in thehistorical databases appear in questions; they are all found in assertions.

We thus have no evidence -- synchronic or diachronic -- that discourse marker naxon developed from quasi-xagamnaxon. On the contrary, we have seen strong synchronic evidence suggesting the different origins of the two, based ontheir different prosodic properties.

There is also some diachronic evidence in support of two different functional itineraries. However, the following accountis partial, because it applies only to naxon tokens uttered in non-appeal intonation.

We have seen that the earliest clearly discourse marker use of naxon in the written corpora is found in a 1911 play.However, already in Mishanic Hebrew we find metalingual ze naxon (‘this [is] true’) from which, we have argued, the

24 Even-Shoshan, however, does not use this term.


discourse marker has evolved. In a Mishnaic text (around 400 A.D.) providing the rules for the cross-examination ofwitnesses, we find:

Excerpt 17:

'amar rav xisda.said rabbi XisdaRabbi Xisda said:

'exad 'omer. basayif harago.one says with-sword killed-himone says: he killed him with a sword,

ve'exad 'omer. ba/'aridan/ harago.and-one says with-/dagger/ killed-him,and another says: he killed him with a dagger.

ve'eyn ze naxon.and NEG this naxonand this is not admissible [evidence].

'exad 'omer. keylav shxorim.one says clothes-his blackone says: his clothes were black,

ve'exad 'omer. keylav levanim.and-one says: clothes-this whiteand another one says: his clothes were white.

harey ze naxon.PART this naxonnow this is admissible [evidence].

“Hisda said: If one testified that he [the accused] slew him with a sword, and another, that he slew him with a dagger, it [the evidence] is inadmissible. If one says, His clothes were black, and the other, His clothes were white; the evidence is admissible.”

(Babylonian Talmud, Sanhedrin Tractate 41a).

One can see the path from this use of ze naxon to confirm the admissibility of some piece of evidence to discourse markernaxon employed to confirm a statement made by the interlocutor. The clause ze naxon manifests SV word order, whichhad begun to develop in Hebrew toward the late Biblical period (Givon, 1977).

However, Biblical Hebrew VSword order continues to live in the language also inmodern times (cf. Auer andMaschler,2013; Maschler, 2015b). We will now argue that it is utterances employing naxon in VS word order that have led to thedevelopment of the xagam construction.

Hebrew is a language abundant with nominal clauses. A common clausal construction in Biblical Hebrew is PredicativeAdjective + Nominal Subject, as in Joseph’s reply to Paraoh concerning the interpretation of his two dreams:


Excerpt 18:

ve'al hishanut haxalom 'el par'o pa'amayimand-about repeating the-dream to Pharaoh twice‘and concerning the fact that Pharaoh’s dream occurred twice,’

ki naxon hadavar me'im ha'elohim,because established the-thing from the-god‘[it is] because established [is] the thing by God,’

umemaher ha'elohim la'asoto.and rushes the-god to-do-it‘and God will shortly do it.’

‘And for that the dream was doubled unto Pharaoh twice; it is because the thing is established by God, and God will shortly bring it to pass.’

(Genesis 41, 32 (King James version)).

We see here the nominal clause naxon hadavar (‘established [is] the thing’) with the adjectival predicate naxon precedingthe subject hadavar (‘the thing’).

Asmentioned, VS word order continues throughout the history of the language. The following excerpt from an essay byMoses Mendelssohn comes from the beginning of the 19th century:

Excerpt 19:

'al ken naxon hadavar, ki kalomba gila kvar ha'aretsso true the-thing SUBORD Columbus discovered already the-land ‘So true [is] the thing, that Columbus already discovered the land’.

Or, more idiomatically:

‘So it is true, that Columbus had already discovered the land’.

(Mendelssohn 1807: 100).

Here we find the noun hadavar (‘the thing’) modified by the clause ‘Columbus had already discovered the land’. Since themodifier of this noun is a clause, it must be connected to its nucleus by a subordinator, which in this register is ki (‘that’) butin less formal registers would be she- (‘that’).

From here to omitting semantically void davar ‘thing’, the way is short. Indeed, in a text from 1876, we find:

Excerpt 20:

vehaya naxon be'eynav shetihiye 'ishti gidemet 'o xigeretand-was right in-his-eyes COMP-will be wife-my missing a limb or lame‘and it was right in his view that my wife would be missing a limb or lame’.

(Lilienblum 1876 : 53).

This, of course, is already a full-fledged xagam construction with the clause ‘my wife would be missing a limb or lame’functioning as subject clause of the predicate naxon and connected to it syndetically via she-.

Thus we see that xagam-naxon, too, originates in a predicate adjective, but in one occurring in a predicate-firstconstruction, rather than in the subject-first ze naxon construction. While the semantically void subject ze (‘this’) is nolonger verbalized in discourse marker naxon, the semantically void subject hadavar (‘the thing’) is no longer verbalized inxagam-naxon. However, the clause modifying hadavar has not been ‘dropped’ along with its nucleus. Rather, it hasbecome reanalyzed as the subject clause of the predicate naxon. The fact that Hebrew she- (just like English ‘that’) canfunction either as a complementizer or as a subordinator connecting a noun with its modifying clause enables thisreanalysis.


On the basis of our synchronic and diachronic analyses, we thus propose two functional itineraries for naxon, sharing acommon beginning (what Craig (1991) has termed ‘polygrammaticalization’):

Discourse Marker

Verb > Predicative Adjective

xagam> quasi-xagam

However, since the diachronic analysis above concerns only naxon tokens uttered in non-appeal intonation, diachrony inthis case does not provide the full picture.

7. Conclusion

A summary of our synchronic findings concerning the forms and functions of Hebrew naxon was presented in Section4. This synchronic analysis provides counter-evidence from aSemitic language for the asymmetric hypothesis concerningthe left and right peripheries in the study of discourse markers (e.g., Degand and Fagard, 2011; Traugott, 2012; Beechingand Detges, 2014). Our study also sheds new light on the xagam debate in Hebrew linguistics, examining the xagamconstruction from a discourse-functional perspective and illustrating the emergence of syntactic categories from the usepeople make of utterances in their talk. We have shown how this construction has continued to evolve into an asyndeticconstruction, in which naxon functions less as a predicate and more as a projecting element external to the clause, thefunction of which is to foreshadow an assertion requiring confirmation/agreement.

Finally, to previous studies on the role of prosody in morphosyntactic change, our study contributes a different angle.Unlike previous studies, which focus mostly on the loss of prosodic prominence (stress) of the grammaticized element (e.g., Bertrand and Chanet, 2005; Wichmann, 2009, 2011; Wichmann et al., 2010), we have correlated grammaticizationwith the type of contour of the entire intonation unit in which the grammaticized element is employed. We hope to haveshown the essential role of prosody -- specifically, the different intonation contours with which naxon is verbalized,uncovered through a synchronic analysis of language in its natural habitat of everyday conversation -- for elucidating thegrammaticization of Hebrew naxon.

Appendix. Transcription conventions and abbreviations used

Transcription basically follows Chafe, 1994, as sometimes adapted by Du Bois et al. (1992) and Du Bois (ms), with afew additions:

. . . -- half second pause (each extra dot = another 1/2 s)

.. -- perceptible pause of less than half a second(3.56) -- measured pause of 3.56 s, -- comma at end of line -- continuing intonation (‘more to come’). -- period at end of line -- sentence final falling intonation? -- question mark at end of line -- sentence final ‘appeal intonation’?, -- question mark followed by comma -- ‘continuing appeal’ intonation! -- exclamation mark at end of line -- sentence final exclamatory intonationø -- lack of punctuation at end of line -- a fragmentary intonation unit, one which

never reached completion.-- two hyphens -- elongation of preceding vowel sound�square bracket to the left of two consecutive lines indicatesbeginning of overlapping speech, two speakers talking at once

alignment such that the right of the top lineis placed over the left of the bottom lineindicates latching, no interturn pause

pp -- pianissimo (spoken very softly)ff -- fortissimo (spoken very loudly)/??????/ -- transcription impossible/words within slashes/ -- uncertain transcription{in curly brackets} -- transcriber’s comments


ˈ -- uninverted quotation mark in the middle of a transliterated word indicates the glottal stop phoneme.’ -- inverted quotation mark in the middle of a transliterated word indicates an elided form (e.g., ts’xa instead of tsrixa(‘needs’, FEM, SG)).- one hyphen -- bound-morpheme boundaryMASC-- masculineFEM -- feminineSG -- singularPL -- plural1ST/2ND/3RD -- first/second/third personACC -- definite direct object markerPAST -- past tensePRES -- present tenseFUT -- future tenseIMPERT -- imperativePART -- particleEXIST -- existentialNEG -- negatorCOMP -- complementizerSUBORD -- subordinator

References

Abramson, Pnina, 1998. hanasu xasar guf umispar (haxagam): degem taxbiri mitraxev lehabaˈat hamodaliyut (‘The impersonal innumeral Hebrewpredicate: an expanding syntactic pattern for expressing modality’) (M.A. thesis). Department of Hebrew Language, University of Haifa.

Ariel, Mira, 1998. Discoursemarkers and form-function correlations. In: Andreas, H. Jucker, Yael Ziv, (Eds.), DiscourseMarkers: Descriptions andTheory. John Benjamins, Amsterdam, pp. 223--259.

Auer, Peter, 2005. Projection in interaction and projection in grammar. Text 25, 7--36.Auer, Peter, 2009. On-line syntax: thoughts on the temporality of spoken language. Lang. Sci. 31, 1--13.Auer, Peter, Maschler, Yael., 2013. Discourse orGrammar? VSpatterns in spokenHebrew and spokenGermannarratives. Lang. Sci. 37, 147--181.Auer, Peter, Maschler, Yael (in press). NU and NÅ: A Family of Discourse Markers Across the Languages of Europe and Beyond. Walter de

Gruyter, Berlin.Azar, Moshe, 1995. taxbir leshon hamishna (‘The Syntax of Mishnaic Hebrew’). The Academy of the Hebrew Language, Jerusalem.Becker, Alton L., 1979. Text-building, epistemology, and esthetics in Javanese shadow theater. In: Becker, Alton L., Yengoyan, Aram (Eds.), The

Imagination of Reality. Ablex, Norwood, N. J., pp. 211--243.Beeching, Kate, Detges, Ulrich (Eds.), 2014. Discourse functions at the left and right periphery. Brill, Leiden.Bertrand, Roxane, Chanet, Catherine, 2005. The pragmatic functions and prosody of enfin in spontaneousFrench. Rev. Semant. Pragmat. 17, 1--41.Biber, Douglas, Johansson, Stig, Leech, Geoffrey, Conrad, Susan, Finegan, Edward, Quirk, Randolph, 1999. Longman Grammar of Spoken and

Written English. Longman, London.Chafe, Wallace, 1987. Cognitive constraints on information flow. In: Tomlin, Rusell (Ed.), Coherence and Grounding in Discourse. John

Benjamins, Amsterdam/Philadelphia, pp. 21--51.Chafe, Wallace, 1994. Discourse, Consciousness, and Time: The Flow and Displacement of Conscious Experience in Speaking and Writing.

University of Chicago Press, Chicago.Craig, Colette G., 1991. Ways to go in Rama: a case study of polygrammaticalization. In: Elizabeth Closs Traugott, Bernd Heine, (Eds.),

Approaches to Grammaticalization, vol. II. John Benjamins, Amsterdam, pp. 455--492.Cruttenden, Alan, 1986. Intonation. Cambridge University Press, Cambridge.Degand, Liesbeth, Fagard, Benjamin, 2011. Alors between discourse and grammar: the role of syntactic position. Funct. Lang. 18 (1), 29--56.Depperman, Arnulf, Günthner, Susanne (Eds.), 2015. Temporality in Interaction, Studies in Language and Social Interaction Series. John

Benjamins, Amsterdam.Du Bois, John W., 2007. The stance triangle. In: Robert Englebretson (Eds.), Stancetaking in Discourse. John Benjamins, Amsterdam/

Philadelphia, pp. 139--182.Du Bois, John W. (ms), 2012. Representing Discourse Linguistics Department, University of California at Santa Barbara (Fall 2012 version).Du Bois, John W., Cumming, Susanna, Schuetze-Coburn, Stephan, Paolino, Danae, 1992. Discourse Transcription: Santa Barbara Papers in

Linguistics, vol. 4. Department of Linguistics, University of California, Santa Barbara, Santa Barbara.Dubnov, Keren, 2008. Adjectives functioning as impersonals (xagam) in early modern Hebrew. Igud 3, 31--40.Even-Shoshan, Avraham, 1986. Hamilon Hexadash (‘The New Dictionary’). Kiryat Sefer, Jerusalem.Even-Shoshan, Avraham, 2003. Milon Even-Shoshan (‘The Even-Shoshan Dictionary’). Hamilon Hexadash Ltd., Jerusalem.Ferrara, Kathleen, 1997. Form and function of the discourse marker anyway: implications for discourse analysis. Linguistics 35, 343--378.Fox, Barbara, Maschler, Yael, Uhmann, Susanne, 2010. A cross-linguistic study of self-repair: evidence from English, German, and Hebrew. J.

Pragmat. 42, 2487--2505.Givon, Talmy, 1977. The drift from VSO to SVO in Biblical Hebrew: the pragmatics of tense-aspect. In: Li, C.N. (Ed.), Mechanisms of Syntactic

Change. University of Texas Press, Austin, pp. 181--254.

http://refhub.elsevier.com/S0378-2166(15)00330-6/sbref0005






































Grice, H. Paul., 1975. Logic and conversation. In: Peter Cole, Jerry, L., Morgan, (Eds.), Syntax and Semantics: Speech Acts, vol. 3. AcademicPress, New York, pp. 41--58.

Günthner, Susanne, 2011. Between emergence and sedimentation: projecting constructions in German interactions. In: Peter Auer, StephanPfänder, (Eds.), Constructions: Emerging and Emergent. Walter de Gruyter, Berlin, pp. 156--185.

Heritage, John, 2002. Oh-prefaced responses to assessments: a method of modifying agreement/disagreement. In: Cecilia Ford, E., Barbara, A.Fox, Sandra, A. Thompson (Eds.), The Language of Turn and Sequence. Oxford University Press, Oxford, pp. 196--224.

Heritage, John, Raymond, Geoffrey, 2005. The terms of agreement: indexing epistemic authority and subordination in talk-in-interaction. Soc.Psychol. Q. 68, 15--38.

Hopper, Paul J., 1987. Emergent grammar. In: Jon Aske, Natasha Beery, Laura Michaelis, Hana Filip, (Eds.), Proceedings of the ThirteenthAnnual Meeting of the Berkeley Linguistics Society, vol. 13. Berkeley Linguistics Society, Berkeley, pp. 139--157.

Hopper, Paul J., 2011. Emergent grammar and temporality in interactional linguistics. In: Peter Auer, Stefan Pfänder, (Eds.), Constructions --Emerging and Emergent. de Gruyter, Berlin, pp. 22--44.

Hopper, Paul J., Thompson, Sandra A., 2008. Projectability and clause combining in interaction. In: Ritva Laury (Eds.), Crosslinguistic Studies ofClause Combining: The Multifunctionality of Conjunctions. John Benjamins, Amsterdam, pp. 99--123.

Hopper, Paul J., Traugott, Elizabeth Closs, 2003. Grammaticalization, 2nd ed. Cambridge University Press, Cambridge.Kärkkäinen, Elise, 2003. Epistemic Stance in English Conversation: A Description of Interactional Functions, With a Focus on I think. John

Benjamins, Amsterdam.Keevallik, Leelo, 2006. From discourse pattern to epistemic marker: Estonian (ei) tea ‘don’t know’. Nordic J. Linguist. 29, 173--200.Kurylowicz, Jerzy, 1964. The Inflectional Categories of Indo-European. Winter, Heidelberg.Kuzar, Ron., 1992. haxagam: xelek dibur ˈo ˈemda taxbirit? (‘The xagam (lacking person and number): a grammatical category or syntactic

position?’). Leshonenu (‘Our Language’) 56, 241--248.Kuzar, Ron, 2002. tavnit haxagam hapshuta belashon hameyutseget kimeduberet (‘The simple impersonal construction in language presented as

spoken’). In: Izreˈel, Sholmo (Ed.), Teˈuda: The ChaimRosenberg School of Jewish Studies Research Series XVIII: Speaking Hebrew. Tel AvivUniversity Press, Tel Aviv, pp. 329--352.

Kuzar, Ron., 2007. digmey mishpat beˈivrit yisraˈelit ˈal pi rozen (‘Sentence patterns in Israeli Hebrew according to Rosén). haˈivrit veˈaxyoteha(‘Hebrew and Its Sisters’), vol. 6--7. pp. 269--294.

Kuzar, Ron, 2012. Sentence Patterns in English and Hebrew. John Benjamins, Amsterdam, pp. 2012.Lerner, Gene, 1991. On the syntax of sentences-in-progress. Lang. Soc. 20, 441--458.Maschler, Yael, 2002. The role of discourse markers in the construction of multivocality in Israeli Hebrew talk-in-interaction. Res. Lang. Soc.

Interact. 35, 1--38.Maschler, Yael, 2009. Metalanguage in Interaction: Hebrew Discourse Markers. John Benjamins, Amsterdam/Philadelphia.Maschler, Yael, 2015a. The Haifa Corpus of Spoken Hebrew. http://weblx2.haifa.ac.il/�corpus/corpus_website/Maschler, Yael, 2015b. Word order in time: emergent Hebrew (Ns) V/VNs syntax. In: Arnulf Deppermann, Susanne Günthner, (Eds.), Temporality

in Interaction, Studies in Language and Social Interaction Series. John Benjamins, Amsterdam, pp. 201--236.Maschler, Yael (forthcoming). The Emergent Grammar of Hebrew Clause-Combining.Maschler, Yael, Dori-Hacohen, Gonen (in press). Hebrew nu: grammaticization of a borrowed particle from synchronic and diachronic

perspectives. In: Auer, Peter, Maschler, Yael (Eds.), NU and NÅ: A Family of Discourse Markers Across the Languages of Europe andBeyond. Walter de Gruyter, Berlin.

Maschler, Yael, Estlein, Roi, 2008. Stance-taking in Hebrew casual conversation via beˈemet (‘really, actually, indeed’, lit. ‘in truth’). DiscourseStud. 10 (4), 283--316.

Maschler, Yael, Schiffrin, Deborah, 2015. Discourse markers: language, meaning, and context. In: Deborah Tannen, Hamilton, Heidi E., DeborahSchiffrin, (Eds.), The Handbook of Discourse Analysis. 2nd ed. John Wiley & Sons, Ltd., Chichester, UK, pp. 189--221.

Meillet, Antoine, 1912. l’évolution des formes grammaticales. Scientia (Riv. Sci.) 12 (26)Reprinted in Meillet, 1958: 130--148.Meillet, Antoine, 1958. Linguistique historique et linguistique générale. Champion, Paris.Méndez-Naya, Belén, 2007. Adjunct, modifier, discourse marker: on the various functions of right in the history of English. Folia Linguist. 27 (1--2),

141--169.Miller, Carmit, 2010. habaˈat ˈemda ˈepistemit basiax haˈivri hayomyomi hadavur (‘Epistemic Stance in Hebrew Talk-in-Interaction’) (Ph.D. thesis).

Department of Hebrew Language, University of Haifa.Mor, Uri, Pat-El, Na’ama (in press). The development of predicates with prepositional subjects in Hebrew. J. Semit. Stud.Nir, Bracha, 2013. Complementizers in Hebrew. In: Geoffrey Khan (Eds.), Encyclopedia of Hebrew Language and Linguistics. Brill Academic

Publishers, Leiden, pp. 504--511.Norde, Muriel, 2009. Degrammaticalization. Oxford University Press, Oxford.Pekarek Doehler, Simona, 2011. Clause-combining and the sequencing of actions: projector constructions in French talk-in-interaction. In: Ritva

Laury, Ryoko Suzuki, (Eds.), Subordination in Conversation: A Cross-Linguistic Perspective. John Benjamins, Amsterdam.Rosén, Haiim B., 1963. ˈivrit tova: ˈiyunim betaxbir (‘Good Hebrew: studies in syntax’), 2nd ed. Kiryat Sefer, Jerusalem (in Hebrew).Rosén, Haiim B., 1966. A Textbook of Israeli Hebrew, With an Introduction to the Classical Language. Chicago University Press, Chicago.Rosén, Haiim B., 1977a. ˈivrit tova: ˈiyunim betaxbir (‘Good Hebrew: studies in syntax’), 3rd ed. Kiryat Sefer, Jerusalem (in Hebrew).Rosén, Haiim B., 1977b. Contemporary Hebrew. Mouton, The Hague.Rubinstein, Eliezer, 1968. hamishpat hashemani: ˈiyunim betaxbir yameynu (‘The Nominal Sentence: A Study in the Syntax of Contemporary

Hebrew’). Hakibbutz Hameuchad, Tel Aviv.Stern, Naftali, 1983. xagam udmuyey poˈal + shem poˈal baˈivrit hayisraˈelit (‘Impersonals and verboids + infinitive in Israeli Hebrew’). Leshonenu

(‘Our Language’) 47, 248--263.Thompson, Sandra A., 2002. ‘‘Object complements’’ and conversation: towards a realistic account. Stud. Lang. 26, 125--163.Thompson, Sandra A., Mulac, Anthony, 1991a. A quantitative perspective on the grammaticization of epistemic parentheticals in English. In:

Elizabeth, Closs Traugott, Bernd, Heine (Eds.), Approaches to Grammaticalization, vol. II. John Benjamins, Amsterdam, pp. 313--329.
































http://weblx2.haifa.ac.il/~corpus/corpus_website/

http://weblx2.haifa.ac.il/~corpus/corpus_website/




































Thompson, Sandra A., Mulac, Anthony, 1991b. The discourse conditions for the use of the complementizer that in conversational English. J.Pragmat. 15, 237--251.

Traugott, Elizabeth Closs, 1982. From propositional to textual and expressive meanings: some semantic-pragmatic aspects of grammaticaliza-tion. In: Winfred, P. Lehman, Yakov Malkiel, (Eds.), Perspectives on Historical Linguistics. John Benjamins, Amsterdam, pp. 245--271.

Traugott, Elizabeth Closs, 1989. On the rise of epistemic meanings in English: an example of subjectification in semantic change. Language 65,31--55.

Traugott, Elizabeth Closs, 1995a. The role of the development of discourse markers in a theory of grammaticalization. In: 12th InternationalConference on Historical Linguistics. Manchester, August 1995 (version of 11/97) http://www.stanford.edu/�traugott/traugott.html

Traugott, Elizabeth Closs, 1995b. Subjectification in grammaticalisation. In: Dieter Stein, Susan Wright, (Eds.), Subjectivity and Subjectivisation:Linguistic Perspectives. Cambridge University Press, Cambridge, pp. 31--54.

Traugott, Elizabeth Closs, 2003. From subjectification to intersubjectification. In: Raymond, Hickey (Ed.), Motives for Language Change.Cambridge University Press, Cambridge, pp. 124--139.

Traugott, Elizabeth Closs, 2007. Discussion article: discourse markers, modal particles, and contrastive analysis, synchronic and diachronic.Catalan J. Linguist. 6, 139--157.

Traugott, Elizabeth Closs, 2012. Intersubjectification and clause periphery. Engl. Text Constr. 5 (1), 7--28.Traugott, Elizabeth Closs, Dasher, Richard B., 2002. Regularity in Semantic Change. Cambridge University Press, Cambridge.Trummer, Pnina, 2000. tzuzot kategoryaliyot baˈivrit hayisraˈelit: kavim klaliyim lemexkar (‘Categorical shifts in Israeli Hebrew: General principles

for research’). Helkat Lashon 29--32, 222--241.Van der Wal, Jenneke, 2013. (Inter)subjectification in Makhuwa: from demonstrative to pragmatic particle. J. Hist. Pragmat. 14 (1), 1--44.Wichmann, Anne., 2009. Sorry: prosodic evidence for grammaticalisation in progress? Gothenbg. Stud. Engl. 96, 327--337.Wichmann, Anne., 2011. Grammaticalization and prosody. In: Heiko Narrog, Bernd Heine, (Eds.), The Oxford Handbook of Grammatcalization.

Oxford University Press, New York, pp. 331--341.Wichmann, Anne, Anne-Marie, Simon-Vandenbergen, Karen, Aijmer, 2010. How prosody reflects semantic change: a synchronic case study of of

course. In: Kristin Davidse, Lieven Vandelanotte, Hubert Cuyckens, (Eds.), Subjectification, Intersubjectification and Grammaticalization. deGruyter Mouton, Berlin, pp. 103--154.

Further reading

DatabasesThe Database of the Historical Dictionary of the Hebrew Language of the Academy of the Hebrew Language.Online Judaic Responsa Project, retrieved December, 2011 from http://www.biu.ac.il/jh/Responsa/index.html.TextsBabylonian Talmud, Sanhedrin Tractate 41a. English translation by Jacob Shachter. http://www.come-and-hear.com/sanhedrin/index.htmlBen Moshe, Tuvia. 1050. Otsar nexmad (‘A nice treasure’). Oxford: Bodleian Library, 290.Rabbi Saadia Gaon. 922. Tshuvot ˈal Ben Meir beˈinyan haluax (‘Answers to Ben Meir Concerning the Hebrew Calendar’). Oxford: Bodleian

Library, d.74, 28--29.Lilienblum, Moshe Leib. 1876. xatˈot neˈurim (‘The Sins of Youth’).Mendelssohn, Moses. 1807. metsiˈat haˈarets haxadasha (‘The finding of the new land’). Dapey hamikdam.







http://www.stanford.edu/~traugott/traugott.html

http://www.stanford.edu/~traugott/traugott.html


















http://www.biu.ac.il/jh/Responsa/index.html

http://www.come-and-hear.com/sanhedrin/index.html

Documents

Maschler, Yael and Miller Shapiro, Carmit. (2016). The role of prosody in the grammaticization of Hebrew naxon (‘right/true’): Synchronic and diachronic aspects. Journal of Pragmatics