From Predication to Prediction the Grammaticalization of the Hebrew Verb ‘Haya’

Tel Aviv University

The Lester and Sally Entin Faculty of Humanities

Linguistics Department

From Predication to Prediction The grammaticalization of the Hebrew verb ‘haya’

M.A. Thesis Submitted

by

Elior Elkayam

Prepared under the supervision of

Prof. Mira Ariel

August 2018

Acknowledgments

This thesis is the result of six years of education in linguistics – three years as a Bachelor’s student and another three years as a Master’s student at Tel Aviv University. I would like to thank the people who supported me during this long and fascinating process and provided me with educational and emotional guidance. First, I would like to thank my advisor – Prof. Mira Ariel who introduced me to the world of pragmatics and grammaticalization back when I was a Bachelor’s student. Thanks to Prof. Ariel I have learned a lot about the pragmatic and discoursal factors that shape grammar. Her guidance in writing this thesis and other conference presentations has been invaluable.

I also thank Prof. Hava Bat-Zeev Shyldkrot who taught me the foundations of grammaticalization theory. In her seminars, I was introduced to the patterns that guide language change and to the constructional perspective of these processes. It was in her seminar when the idea to investigate the grammaticalization of the ‘haya + participle’ construction arose. Thanks to the participants of these seminars, I was able to improve my understanding in particular facets of grammar. In particular, I would like to thank Dr.

Danny Kalev who opened my eyes to the world of aspect and modality. Thanks to his guidance I learned about the foundations of aspect and the verbal system of Biblical

Hebrew.

Furthermore, I thank my friend Jonathan Avidan who has been my companion in the education of linguistics since the early days of our B.A until our M.A. graduation.

Jonathan has taught me a great deal about the diversity of linguistic theories as well as about the history of linguistics. His friendship was always supportive and encouraging.

Also my boyfriend, Daniel Hansen, deserves a large gratitude for his help in the process of writing this thesis. Danny and I have had hours of talks and discussions into the foundations of language and grammar. Together we have educated ourselves and cultivated our curiosity. Even in rough times, Danny provided me with emotional support that filled me with motivation to keep going. I would also like to thank Alex Liban who always believed in me. Finally, I thank my parents who provided me with financial support during all these years.

Transcription Conventions

Phonetic Transcription Used here IPA symbol IPA description S /ʃ/ postalveolar fricative c /ts/ alveolar affricate x /x/ velar fricative ‘ /ʔ/ glottal stop ħ /ħ/ voiceless pharyngeal fricative ʕ /ʕ/ pharyngeal approximate

Other Symbols Symbol Meaning * semantic anomaly # pragmatic anomaly

Table of Contents 1. INTRODUCTION (ABSTRACT)...... 1 2. RESEARCH QUESTIONS ...... 3 3. DIACHRONY AND HISTORY: THE THREE STAGES OF HEBREW ...... 4 4. THEORETICAL FRAMEWORK ...... 5 4.1 DISCOURSE FUNCTIONS ...... 8 4.2 THE VERBAL SYSTEM OF HEBREW ...... 12 5. THE PAST HABITUAL ...... 14 5.1 BIBLICAL HEBREW ...... 15 5.2 MISHNAIC HEBREW ...... 25 5.3 MODERN HEBREW ...... 31 5.3.1 Introduction to the Hebrew Habituals ...... 33 5.3.2 Research Questions ...... 40 5.3.3 The Corpus ...... 41 5.3.4 Methodology ...... 42 5.3.5 Results and Discussion ...... 43 5.3.6 The Function of the Habituals in Discourse ...... 49 5.3.7 Conclusions ...... 53 5.4 SUMMARY ...... 54 6. THE HYPOTHETICAL CONDITIONAL ...... 55 6.1 BIBLICAL HEBREW ...... 57 6.2 MISHNAIC HEBREW ...... 59 6.3 MODERN HEBREW ...... 66 6.4 SUMMARY ...... 69 7. THE HYPOTHETICAL PREDICTIVE...... 70 7.1 HYPOTHESIS AND PREDICTION ...... 73 7.2 METHODOLOGY ...... 74 7.3 RESULTS...... 75 7.4 SUMMARY ...... 78 8. DISCUSSION ...... 79 9. CONSTRUCTIONAL POLYSEMY ...... 89 10. CONCLUSIONS ...... 90 REFERENCES ...... 92

List of Figures

Figure 1. The three communicative layers of language: form, meaning and function ...... 7

Figure 2. Synchrony and diachrony – two theoretical approaches ...... 8

Figure 3. An illustration of semantic categories...... 11

Figure 4. Percentage of discourse function within construction type ...... 43

Figure 5. Percentage of clause dependence within construction type ...... 47

Figure 6. Percentage of clause dependence within discourse function ...... 47

Figure 7. Percentage of adverbial frame type within construction type ...... 48

Figure 8. Illustration of the Habitual functions ...... 50

Figure 9. The Past Habitual and the Hypothetical Conditional in Modern Hebrew ...... 55

Figure 10. Demonstration of the relationship between Counterfactuality and Hypotheticality ...... 56

Figure 11. Interaction between clause type and modality type for ‘haya + participle’...... 75

Figure 12. Interaction between clause type and subject person for ‘haya + participle’ ...... 78

Figure 13. The grammaticalization path that leads to the Hypothetical Predicitve ...... 80

Figure 14. The grammaticalization of the scope of haya ...... 82

Figure 15. Summary of the grammaticalization path from predication to prediction ...... 89

List of Tables

Table 1. The paradigms of verbs, participles and nouns in Hebrew (Gordon 1985:124) ...... 13

Table 2. The yiqtol verbal inflection ...... 14

Table 3. The meaning and function of the participle with and without the aux. haya ...... 25

Table 4. The changes in the verbal system from Biblical to Mishnaic ...... 25

Table 5. Summary of the Hebrew Habituals ...... 40

Table 6. Summary of the discourse profiles of the Hebrew Habituals...... 53

Table 7. The differences between Counterfactuality and Hypotheticality ...... 57

Table 8. The hypothetical conditional constructions of Hebrew ...... 69

Table 9. The discourse profiles of the two readings of the ‘haya + participle’ construction ...... 90

1. INTRODUCTION (ABSTRACT)

In Modern Hebrew, haya ‘was’ is a verb which participates in various constructions to convey different meanings. From a grammatical point of view, haya functions as an existential verb (e.g. haya po xatul ‘there was a cat here’) or a copula (e.g. hu haya more ‘he was a teacher’). However, in this thesis I will focus on three constructions1:

1. The Past Habitual (haya + participle) hu haya holex le-Sam he was going to-there ‘he used to go there’ 2. The Hypothetical Conditional (if haya + participle, haya + participle) im hayita omer li, hayiti mevi od if were.you telling me was.I bringing more ‘if you had told me, I would have brought more’ 3. The Hypothetical Predictive (haya + participle) hayiti ma’adif la-lexet la-kolnoa were.I preferring to-go to.the-cinema ‘I would rather go to the cinema’

The question that arises by observing these constructions is – what is special about haya that allows concepts such as habituality and hypotheticality to arise? What connects these two concepts? And what does the collocation ‘haya + participle’ contribute to each of these constructions?

There are two ways to answer these questions – synchronically and diachronically.

The synchronic approach aims to view the way these constructions interact with each other. The idea is that by revealing the formal, semantic, and discourse properties of each

1 Names of constructions will be capitalized. The meaning they convey will appear in lowercase letters. For example, the Past Imperfective refers to a construction that conveys past imperfective tense/aspect. 1 of these constructions, we will be able to see the gradience between them (Hopper &

Traugott 2003). Moreover, it will enable us to hypothesize about how they grammaticalized.

The purpose of the diachronic approach is to reveal the particular contexts that bridge between each of these constructions. By doing so, we will be able to see what connects the haya constructions from formal, semantic, and discoursal perspectives. This evidence will serve us while evaluating the synchronic hypothesis regarding the path that these constructions followed to grammaticalize.

To begin, I will show that the ‘haya + participle’ construction emerged in Biblical

Hebrew with the meaning of the past imperfective (section 5). The discourse function of this construction was to introduce a new episode to the discourse (i.e. new subject, tense, or place). In Mishnaic Hebrew, this construction, while still conveying the past imperfective, acquired the function of describing an event that occurs within a larger episode. In Modern Hebrew, only the past habitual reading of the ‘haya + participle’ construction persisted. A corpus-based study reveals that the function of the modern Past

Habitual is eventive, like in Mishnaic Hebrew. This study provides new evidence in favor of Rosén’s (1985) hypothesis that the modern Past Habitual was adopted from Mishnaic

Hebrew and not Biblical Hebrew (conra Doron 2006).

Furthermore, I will trace the Hypothetical Conditional to its Biblical and Mishnaic sources and will show the bridging contexts that connect between the Past Imperfective and the Hypothetical Conditional (section 6). Compared with the Past Habitual, the

2 grammaticalization path of the Hypothetical Conditional is characterized by more formal changes than semantic changes.

The next grammatical development is the emergence of the Hypothetical Predictive

(section 7). I claim that in Modern Hebrew the Hypothetical Predictive emerged as an independent construction from the consequent of the Hypothetical Conditional. In order to support my claim, I will analyse a corpus of colloquial written Hebrew and show that the Hypothetical Predictive emerged as an independent clause in subjective contexts in which the construction conveyed the attitude of the speaker towards the occurrence of the hypothetical event.

Consequently, I will discuss and analyse the patterns of grammaticalization that the haya constructions followed (section 8). In section 9, I will summarize how the speaker differentiates between the two meanings of the ‘haya + participle’ construction (Past

Habitual and Hypothetical Predictive). Section 10 briefly concludes this thesis.

2. RESEARCH QUESTIONS

My research questions are as follows:

a) How did the copula haya develop the auxiliary function? That is, how did haya join

the Past Habitual?

b) Does the Past Habitual in Modern Hebrew retain any of its older meanings in

Biblical or Mishnaic Hebrew? Can a synchronic corpus reflect the

3 grammaticalization path that gave rise to the discourse function of the Past

Habitual?

c) Why does the ‘haya + participle’ construction participate in both the Past Habitual

and the Hypothetical constructions? What connects habituality and

hypotheticality?

d) What are the formal, semantic and discoursal properties that connect between the

Hypothetical Conditional and the Hypothetical Predictive? Can a synchronic

corpus reflect the grammaticalization path that gave rise to the emergence of the

latter?

e) What does the overall diachronic development of haya tell us about the patterns of

grammaticalization that shape language change?

3. DIACHRONY AND HISTORY: THE THREE STAGES OF HEBREW

Hebrew is a Northwest Semitic language of the Afroasiatic family. Its earliest known form is Biblical Hebrew. At this stage, Hebrew was spoken in ancient Israel between the 11th and

5th centuries BCE. Early Biblical Hebrew was spoken between the 11th and 6th centuries BCE, and late Biblical Hebrew was spoken between the 6th and 5th centuries BCE (Gordon 1982).

Biblical Hebrew was the written language of the Hebrew Bible.

Mishnaic Hebrew was spoken from the 1st to the 4th century CE. It is mainly represented by the Mishnah and Midrash, two central texts of Rabbinic Judaism. Around

400 CE, Hebrew ceased to be an everyday spoken language. However, it survived into the

4 medieval period as the language of Jewish liturgy, rabbinic literature, intra-Jewish commerce and poetry. Then, in the 19th century, Hebrew was revived as a spoken and literary language. Hence, I shall refer to this stage of the language as Modern Hebrew. As of 2014, it comprises 9 million native speakers worldwide, most of whom are Israeli.

In this paper, I account for the grammaticalization of haya based on diachronic data cited from the Bible, Mishnah and other Rabbinic texts. By doing so, I rely on the evolutionary trait of retention (Bybee and Pagliuca 1987) or persistence (Hopper 1991) which states that forms from older stages of development coexist with their newer developments in one synchronic stage of the language.

In addition, I provide examples from Modern Hebrew which are taken from the

Corpus of Spoken Israeli Hebrew (henceforth CoSIH)2 as well as the HeTenTen corpus, which is based on online written Hebrew from all genres (news, literature, blogs, forum discussion etc.). This corpus was tagged morphologically by Adler (2007), with a contribution of Noam Ordan.

4. THEORETICAL FRAMEWORK

In order to investigate the grammaticalization path of the Past Habitual, the Hypothetical

Conditional and the Hypothetical Predictive I will employ several theoretical perspectives.

To begin, the theory that will guide the syntactic and semantic perspective will be

Construction Grammar (Goldberg 1995). The main premise behind Construction Grammar

2 Corpus of Spoken Israeli Hebrew (http://cosih.com). 5 is that words make up larger structures called constructions which carry a meaning of their own. In other words, constructions are linguistic units that carry meanings which cannot be predicted based on the sum of their parts. For example, the di-transitive construction

(John gave Jill a ring) carries a meaning of transfer, even if it is used with a non-transfer verb (e.g. John slid the present to Susan).

Since constructions are not fully compositional, they allow us to view language as it really is – emergent. Words emerge to convey meaning, and when several words are arranged into a particular structure, that structure gets meaning as well. In addition to form and meaning, I will observe another communicative layer – the discourse. Within the discourse layer, we are concerned with the purpose of conveying a certain meaning, i.e. we ask the question – for what purpose did the speaker convey X? We also ask the pragmatic question – what did the speaker intend the addressee to infer by expressing X? These kinds of questions will be relevant for the functional layer of language.

Another way to distinguish between meaning and function is by relying on

Blakemore’s (1987) distinction between conceptual and procedural linguistic items.

Conceptual linguistic markers map onto cognitive concepts (e.g. table, run) and relate to the propositional content of the utterance. Procedural linguistic items regulate linguistic processes. They contain instructions from the speaker to the hearer as to how to interpret the utterance (e.g. pronouns, discourse markers). The following figure demonstrates the theoretical perspective that will be employed in this thesis. The eye symbolizes the scope of analysis and each circle represents a communicative layer of language.

6 Figure 1. The three communicative layers of language: form, meaning and function

Function

Meaning

Form

These three communicative layers will be observed from both a synchronic and a diachronic perspective. Following what was mentioned in the introduction, the purpose of the synchronic approach is to view the objects of our analysis (in our case the three haya constructions) as living phenomena that interact with each other. This allows us to see the interplay between form, meaning and function in greater detail. The idea is that by employing a corpus-based study, we can infer linguistic patterns based on real-world data.

This will allow us to account better both for the synchronic analysis of the three haya constructions, as well as for inferring the grammaticalization path that connects them.

This is where the diachronic approach is employed. The purpose of the diachronic approach in this thesis will be to reveal the contexts that bridge between the haya constructions. Once we do that, we will be able to see more clearly the graduality of form, meaning and function which support the emergence of these constructions.

The following figure illustrates the synchronic and diachronic approaches. The eye on the right side represents the diachronic approach and the eye above represents the synchronic approach. The three rings symbolize each stage of the Hebrew language

(Biblical, Mishnaic and Modern). Within each ring there are linguistic phenomena

7 (grammatical markers, constructions, etc.). In Biblical Hebrew there is phenomenon A which persisted into Mishnaic Hebrew, albeit with minor changes. In Modern Hebrew, phenomenon A split into two different phenomena: A and B.

Figure 2. Synchrony and diachrony – two theoretical approaches

A B Modern Hebrew A Mishnaic Hebrew A Biblical Hebrew

Hence, the definition of grammaticalization that will be adopted here relies on Hopper and

Traugott (2003: 231): “a change whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions and, once grammaticalized, continue to develop new grammatical functions”.

4.1 Discourse Functions

This study will be conducted within a discourse-oriented framework. In particular, this means that I will observe the discourse functions of each of the haya constructions in order to understand the formation of their grammatical uses. The idea is that salient discourse patterns that are highly accessible and more easily available for use are the ones that shape the grammaticalization of their parts (Ariel 2008).

8 In order to deduce the semantic and discourse profiles of the Habituals, I will employ a corpus-based approach. The advantage of the corpus approach is that it confronts the researcher with real-world uses which they did not consider before. It can also challenge the acceptability judgments of the researcher – what sounds bad for the researcher could be naturally uttered by another speaker.

Moreover, I will employ a functional-cognitive view of semantic representations. This view, inspired by Hengeveld’s Functional-Discourse Grammar (2008), sheds light on the hierarchical and conceptual nature of semantics. The notions that will be relevant for us are proposition, episode and event. These three semantic categories are organized in relation to each other: a proposition may contain an episode which contains one or more events.

A proposition is the part of the clause which expresses the idea the speaker intends to convey. This idea remains constant regardless of other pragmatic or discourse considerations. Some propositions contain an event without an explicit episode. In such propositions it is most likely the case that the episode is inferred from the discourse.

An episode is defined as the part of discourse which frames the subsequent events3.

This framing is done by introducing a new subject, a new tense, a new place, etc. (see Givón

2015: 163). The episode is thus not an accessible piece of information (see generally Ariel

1990). Finally, an event is the conceptual part of the discourse which describes a particular

3 Notice that the term episode is used differently here than in Boneh & Doron (2008, 2010). While they use the term to refer to a particular occasion within a habit, I use it to refer to the event or the habit as a whole. My use of the term episode is backed up not only by functionalist theory but also by the Oxford Dictionary definition: “an event or a group of events occurring as part of a sequence; an incident or period considered in isolation”.

9 situation which the speaker refers to. An event contains actions (e.g. verbs), objects (e.g. nouns) and sometimes modifiers (e.g. adjectives). To exemplify let us observe the following quote:

(1) a. Well, there was this movie I seen one time b. about a man riding ’cross the desert and it starred Gregory Peck c. He was shot down by a hungry kid trying to make a name for himself d. The townspeople wanted to crush that kid down and string him up by the neck (Bob Dylan ‘Brownsville Girl’)

In the above example, sentence (a) and its continuation in (b) are the episode – this is the part which presents the new subject (‘this movie’), new participants (‘I’, ‘a man riding…’) and a new tense (‘there was’) to the poem. Sentence (c) corresponds to the event – it presents a particular scene from the movie. It does not introduce a new subject nor a new tense or place. In fact, the pronoun ‘he’ refers to an entity already introduced in the episode.

Sentence (a-b) corresponds to a proposition which includes an episode (‘…there was…’) and events (c) and (d)4. The following is a (simplified) illustration of the relations between these semantic categories:

4 ‘Well’ in (a) is in fact not part of the (semantic) proposition but of the (discoursal) utterance. 10 Figure 3. An illustration of semantic categories

‘…there was this movie I seen one time…’

‘…there was…’

‘He was shot down…’ ‘The townspeople…’

event event

episode proposition

Additionally, I will rely on Suh’s (1992) findings about the English Habituals. Suh conducted a study with a similar approach – she collected instances of the English

Habituals, particularly the would, used to and simple past Habituals, and analyzed their interactions with each other within discourse. Her findings show that when used to and would occur sequentially, used to marks an episode boundary or sets up a rhetorical frame for a past habitual episode, while would or the simple past tense, on the other hand, describe the details of the habit or elaborate on the topic. The following piece of discourse exemplifies this:

(2) a. I'd go barefoot to school. b. The bad thing was they used to laugh at us, the Anglo kids. c. They would laugh because we'd bring tortillas and frijoles to lunch. d. They would have their nice little compact lunch boxes with cold milk in their e. thermos and they'd laugh at us because all we had was dried tortillas. f. Not only would they laugh at us, but the kids would pick fights. g. My older brother used to do most of the fighting for us and h. he'd come home with black eyes all the time. (Suh 1992: 31-32)

11 Up until line (a), the speaker had been talking about the poverty of Mexican-Americans. In line (b) he shifts the focus of the discourse by using a cleft sentence (“the bad thing was…”), hence putting the current utterance (b) in contrast to the previous one in (a). In other words, used to in (b) serves the function of initiating a new episode. From line (c) to (f), the speaker uses would to elaborate on the habit in (b), namely on the way the Anglo kids treated the

Mexican kids. Notice that all the events expressed by would (c-f) refer to the same subject

(‘they’) introduced in episode (b). In line (g), the speaker shifts the topic to a new episode with used to – he introduces a subject (‘my older brother’) in a new scenario. Sentence (h) elaborates on this new habit (‘do most of the fighting for us…’) with would.

4.2 The Verbal System of Hebrew

Verbs in Hebrew are classified according to seven paradigms, each called a binyan. Each paradigm is a phonological construction combining root consonants (lexico-semantic material) and vowels (paradigmatic material). Verbs inflect for number, person, gender and tense/aspect (see below), while nouns and adjectives inflect only for number and gender. The participle can function as a noun, an adjective or a verb. When it is a verb, it conveys imperfective aspect (Givón 1977). The following table exemplifies the verbal, participial and nominal inflections with the root g-n-v which means ‘steal’ (as a verb) and

‘thief’ (as a noun):

12 Table 1. The paradigms of verbs, participles and nouns in Hebrew (Gordon 1985:124)

VERB in qatal singular plural I ganav-ti we ganav-nu you m. ganav-ta you m. ganav-tem you f. ganav-t you f. ganav-ten he ganav they ganv-u she ganv-a

PARTICIPLE NOUN masc.sg gonev masc.sg ganav fem.sg gonev-et fem.sg ganev-et masc.pl gonv-im masc.pl ganav-im fem.pl gonv-ot fem.pl ganav-ot

Tense in Biblical Hebrew is relative (Polak 2009). Suffixed verbs (referred to as qatal; as in the 3rd person masculine singular inflection in the table above) encode the anterior tense and serve the function of marking thematic discontinuity (Givón 1977, Polak 2009). In contrast, the prefixed verbs (with the 3rd person masculine singular form of yiqtol) encode the posterior tense and function to mark thematic continuity (more on this in section 5.1). 5

The following table demonstrates the yiqtol verbal inflection.

5 This view is contrasted with the traditional association of yiqtol verbs with the imperfect (past imperfective) and qatal verbs with the perfect. The root cause for the difference is that the traditional view (see Hatav 1997 for review) assumes Biblical Hebrew employs an absolute-tense system (like in Modern Hebrew and most Indo-European languages) while it actually employs a relative-tense system (Polak 2009). 13 Table 2. The yiqtol verbal inflection

VERB in yiqtol singular plural I e-gnov we ni-gnov you m. ti-gnov you (m/f) ti-gnevu you f. ti-gnevi they yi-gnevu he yi-gnov she ti-gnov

5. THE PAST HABITUAL

In this section I will reveal the diachronic development of the ‘haya + participle’ construction in Hebrew. I will begin by introducing the construction and its parts, i.e. the verb haya and the participle. Then I will analyze the meaning and function of the ‘haya + participle’ construction in Biblical Hebrew. I will claim that this construction conveys a various range of meanings, all of which are durative and occur in the past. The unique function of this construction in Biblical Hebrew is that it is used to introduce a discontinuous episode to the discourse. In Mishnaic Hebrew the ‘haya + participle’ construction lost most of its discontinuous and episodic functions and gathered an eventive function. In other words, instead of introducing a discontinuous episode it relied on some previously mentioned episode in order to express past imperfective meaning.

I hypothesize that in Modern Hebrew, the function of the ‘haya + participle’ construction is eventive, similar to Mishnaic Hebrew. In order to test my hypothesis, I will analyze the discourse functions of the ‘haya + participle’ construction in Modern Hebrew and compare it to the habitual qatal. The results show that the ‘haya + participle’ construction is mostly eventive (elaborative) whereas the habitual qatal is episodic –

14 contrary to Biblical Hebrew. Hence, the newly-found evidence supports the claim that the

‘haya + participle’ construction in Modern Hebrew was adopted from the Mishnaic ‘haya + participle’ and not the Biblical one (Rosén 1985; contra Doron 2006).

5.1 Biblical Hebrew

Before we turn to the ‘haya + participle’ construction, I would like to dedicate a brief discussion to the parts of the construction, namely haya ‘was’ and the participle. First, we shall talk about haya.

haya ‘was’ is an existential verb inflected in the qatal paradigm which encodes anterior tense (Polak 2009). Hence, the past meaning of haya is always relative to a particular reference point. Following Reichenbach’s (1947) terminology, when the reference point is time of speech the event is in the past (i.e. the simple past), when the reference point is in the past (as marked by another verb) the event occurs before it (i.e. the pluperfect). From a discoursal point of view, haya (like every qatal verb) conveys thematic discontinuity

(Givón 2015). Three kinds of thematic discontinuity will be mentioned here: a) Beginning: an introductory event set apart from the main storyline. Polak (2009) refers

to this function of the qatal paradigm as “anchor”. b) Episode shift: an event that shifts the theme of the discourse/narrative by introducing a

new subject, tense and sometimes even place (see section 4.1). c) Background (Hopper and Thompson 1980): a side event that does not progress the

narrative or the discourse. It is often a side comment of the speaker/narrator.

15 Notice that both (a) and (b) are episodic but (c) is not necessarily so. A qatal verb that functions to present a background event does not necessarily introduce a new subject (let alone tense and place). Now, let us consider some examples of haya. The following example demonstrates the existential/locative meaning of haya.

(3) wa-yehi ka’aSer patar la-nu ken haya and-was.Y as interpreted.Q to-us so was.Q6

‘And it came to pass, as he interpreted to us, so it was’ (Genesis 41:13)

What (3) shows us is that haya predicates the existence of the event which Joseph (“he”) predicted would happen based on his interpretation of a dream. Hence, the interpretation event expressed by patar occurred before the event expressed by haya. In other words, patar

‘interpreted’ expresses the pluperfect, which could be rephrased as ‘had interpreted’.

Moreover, haya occurs before time of speech, hence it expresses simple past tense.

Therefore, example (3) exemplifies the anterior tense of the qatal verbal inflection and the grammatical function of haya – an existential copula, for it predicates about existence. Now we shall observe some more grammaticalized functions of the copula haya:

6 Each gloss of an inflected verb will be marked as Q for the qatal inflection and Y for the yiqtol inflection. Additionally, only parts that are relevant for the linguistic analysis will be transliterated. 16 (4) a. ki amar ger hayiti bə-‘erec noxriyya for said.Q stranger was.I.Q in-land foreign ‘for he said, “I have been a stranger in a foreign land.”’ (Exodus 2:22) b. ki la-hem haya ha-goral riSona for theirs was.Q the-lot first ‘for theirs was the first lot’ (Joshua 21:10) c. we- ha- naħaS haya ʕarum and-the-serpent was.Q wily ‘And the serpent was wily’ (Genesis 3:1) d. we-eloh-e av-i haya imad-i and-gods-of father-my was.Q with-me ‘but the God of my father has been with me’ (Genesis 31:5)

All the examples in (4) exemplify the copular uses of haya. In (a) haya predicates the classification of the subject, namely that the subject was a foreigner (as expressed by the noun ger ‘stranger’). In (b) haya predicates about the possessor (i.e. the winner) of the lot

(as expressed by the preposition phrase la-hem ‘theirs’). In (c), haya predicates an attribute of the serpent (as expressed by the adjective ʕarum ‘wily’). Finally in (d), haya predicates the subject’s relation to God (as expressed by the preposition phrase imad-i ‘with me’).

As we can tell from the representative list in (4), all of the uses of haya function to express thematic discontinuity. This is evidenced by the subordinate clause they occur in, as expressed by ki ‘for’ in (a-b) and we ‘and/but’ in (c-d). However, the examples also show us differences between the uses of haya – while examples (a-b) are eventive (i.e. they depict an event which depends on a previous episode), (c-d) are episodic (as we can tell by the explicit mention of the subjects). To sum up, haya is a thematically discontinuous copula that has some episodic uses.

17 Now, we shall move on to the next element of the construction – the participle. The meaning of the participle paradigm is the imperfective aspect (Givón 2015, see also Hatav

1997, Joüon 1923)7. In order for the imperfective event to be interpreted as taking place within a certain tense (past, present or future), the participle relies on some contextual cue

(verb, adverb, noun, etc.) (Gordon 1982). The following example shows how the participle can express a past imperfective event without the need for an auxiliary haya:

(5) a. wa-yelex wa-yaʕas ke-devar yawweh ‘So he went and did according to the word of the Lord.’

b. wa-yelex wa-yeSev be-naħal krit aSer ʕal pney ha-yarden ‘he went and lived by the Wadi Cherith, which is east of the Jordan.’

c. we-ha-ʕorvim mevi’im lo lexem u-basar ba-boker and-the-ravens bringing to.him bread and-meat in.the-morning ‘The ravens brought him bread and meat in the morning […]’

(1 Kings 17:6-7)

In the above example, we see the participle in (c) mevi’im (‘bringing’). It is categorized as imperfective since it participates in the depiction of a habit (bringing bread and meat every morning). Importantly, it takes place in the past and it does so without relying on haya ‘was’ as a past reference point. Instead, it relies on the previously mentioned yiqtol verb wa-yeSev

(‘and he lived’). Hence, the interpretation is that the ravens brought bread and meat to the

7 Following Comrie (1976), imperfectivity is defined in this paper as a durative event, i.e. an event which takes place under a certain period of time and can overlap with other events (both the progressive and the habitual are imperfective). Perfectivity is defined according to Croft (2012) as a temporally bounded event, i.e. an event which ended before time of speech. Perfective events are usually viewed as a “whole” while imperfective events are viewed with regards to their “internal structure” (Smith 1997). 18 subject when he lived by Wadi Cherith. In other words, the participle (‘bringing’) gets the past tense by relying on a yiqtol verb (‘and he lived’).

In biblical narrative, the yiqtol verbal inflection was specialized to present the storyline (a.k.a. gist) of the narrative, i.e. the main actions that build up the foreground of the narrative (Polak 2009). Givón (2015) refers to the yiqtol verbal inflection as thematically continuous, for it functions to progress the theme of the narrative. Hence, the yiqtol past event (‘lived by the Wadi Cherith’) and the participle event that follows it (‘the ravens bringing him bread and meat…’) are interpreted as not only contemporaneous but also of equal discoursal status, i.e. thematically continuous. Lastly, we can tell that a participial verb (mevi’im ‘bringing’) can introduce a new episode, as evidenced by the new subject ‘the ravens’. The reader is referred to Doron (2006:265) for more examples of past tense interpretations for the participle.

Now, let us turn to the ‘haya + participle’ construction. According to Rosén (1977,

1985) and Doron (2006), this construction conveyed the past habitual. Rosén claimed that

Modern Hebrew acquired the ‘haya + participle’ construction from Mishnaic Hebrew since in Biblical Hebrew this construction is scarce and follows “different rules” (Rosén 1985:

292). Doron, on the other hand, claims that the Modern ‘haya + participle’ construction was adopted directly from Biblical Hebrew. She explains it this way: in Mishnaic Hebrew the

‘haya + participle’ construction conveyed the past progressive in addition to the past habitual. Since both Biblical and Modern Hebrew lack the past progressive reading, the

‘haya + participle’ construction must have been adopted from Biblical Hebrew along with the biblical verbal system (Doron 2006: 266). 19 However, I argue that Rosén’s and particularly Doron’s assumptions regarding the meaning of the ‘haya + participle’ construction are mistaken. First, I will show that this construction conveyed a wide range of meanings far beyond the past habitual. Secondly, I will show that the ‘haya + participle’ construction in Modern Hebrew is more similar to its

Mishnaic equivalent than to its Biblical one in terms of both meaning and function. Thus, in agreement with Rosén, I claim that the ‘haya + participle’ construction was adopted from

Mishnaic Hebrew and not Biblical Hebrew (contra Doron 2006).

To begin, let us observe examples in which the ‘haya + participle’ construction expresses a wide range of temporal relations:

(6) josef ben Səva ʕesre Sana haya roʕe et eħaw Joseph son seven teen year was.Q shepherd with brothers.his

ba-con in.the-sheep

‘Josef, being seventeen years old used to shepherd [habitual] / was a shepherd [noun] with his brothers among the sheep’ (Genesis 37:2) (7) Sibʕ-im melax-im […] hayu melaqqt-im taħat Sulħa-ni seventy kings […] were.Q gathering under table-my ‘”Seventy kings [...] used to gather / have been gathering [their food/scraps] under my table”’ (Judges 1:7) (8) ki ʕad ha-yam-im ha-hemma hayu beney yisrael meqattrim for until the-days the-those were.Q children-of Israel burning lo to.him ‘for up to that time the Israelites had been burning incense to him’ (2 Kings 18:4)

In (6) the ‘haya + participle’ construction conveys either the past habitual or a copula + noun. Joseph is described either to be a shepherd or to be involved in the habit of

20 shepherding. Both meanings are equally possible due to the ambiguous nature of the participle. In (7), the ‘haya + participle’ construction is ambiguous between a perfect progressive reading in which the kings had been gathering food up to the time of speech and a past habitual meaning (the kings were simply involved in the habit of gathering food under his table). Finally, the ‘haya + participle’ construction in example (8) clearly conveys the past perfect progressive. It is explicitly said that up to a certain time in the past

(‘these days’) the event denoted by the ‘haya + participle’ construction (burning incense) occurred. Therefore, the temporal meaning that the ‘haya + participle’ construction conveys is of a durative event that occurred in the past. Hence, I shall refer to it as the past imperfective (a.k.a. imperfect).

The reason both Rosén (1985) and Doron (2006) claim that the ‘haya + participle’ construction in the above examples conveyed the past habitual is that they implicitly assume that haya had the same function and meaning in Biblical Hebrew as in Modern

Hebrew. However, as I showed earlier, the discourse function of haya is to express thematic discontinuity. Hence, I claim that in Biblical Hebrew, the ‘haya + participle’ construction conveyed a thematically discontinuous past imperfective, whereas the participle alone can convey a thematically continuous past imperfective. To see this, let us re-examine the function of the ‘haya + participle’ construction of example (6) repeated here as (9) with its full context:

(9) a. ele toldot yaʕakov these history.of Jacob ‘This is the history of Jacob’

21 b. josef ben Səva ʕesre Sana haya roʕe et eħaw Joseph son seven teen year was.Q shepherd with brothers.his

ba-con in.the-sheep ‘Joseph, being seventeen years old used to herd / was a shepherd with his brothers among the sheep’

c. we-hu naʕar et bney bilha we-et bney zilpa and-he lad with sons.of Bilhah and-with sons.of Zilpah

neSey aviv women.of his.father ‘And the lad [was] with the sons of Bilhah and the sons of Zilpah, his father’s wives’

d. wa-yave josef et diba-tam raʕa el avi-hem and-brought.Y Joseph ACC slander.their bad to father.their ‘and Joseph brought a bad report of them to his father’

(Genesis 37:2)

As we can tell from (9), the verse begins with an introduction to the history of Jacob.

Sentence (b) presents a new construal – Joseph is the subject, the situation is “with his brothers among the sheep” and the tense is past, as expressed by the qatal verb haya. Here, the verb haya is used as an anchor, it introduces new participants, place and tense to the discourse, hence it is both thematically discontinuous and episodic – it provides new background information that is required for the rest of the narrative. Sentence (c) begins with we (‘and’) which marks contemporaneity with sentence (b). This is also indicated by the lack of an explicit verb. Sentence (d) begins with a posterior wa-yiqtol verb (wa-yave ‘and

[he] brought’) which begins with a new action that progresses the narrative – Joseph tells his father the bad things his brothers were talking about. Hence, example (9) shows us that

22 the ‘haya + participle’ construction functions to mark a discontinuous episode. The same function can be observed from example (7) repeated here as (10) with its full context:

(10) a. wa-yomer adoni-bezeq and-said.Y Adoni-Bezeq ‘And Adoni-Bezeq said’

b. Sibʕ-im melax-im […] hayu melaqqt-im taħat Sulħa-ni seventy kings […] were.Q gathering under table-my ‘”Seventy kings […] used to gather / have been gathering scraps under my table”’

c. ka’aSer ʕasiti ken Silem li elohim as did.I.Q so repaid.Q me God ‘”as I have done, so God has repaid me”’

d. wa-yaviuhu YeruSalem and-brought.they.Y Jerusalem ‘Then they brought him to Jerusalem’

e. wa-yamat Sam and-died.he.Y there ‘and there he died’

(Judges 1:7)

In (a), we begin with the posterior verb wa-yomer ‘and [he] said’ which continues the narrative with the content of the utterance of Adoni-Bezeq in sentence (b). There, the narrator uses the ‘haya + participle’ construction in order to convey a past habitual (‘used to gather’) or a perfect progressive event (‘have been gathering’). Importantly, in (b) we witness an episode shift: a new subject (‘seventy kings’), new tense (haya ‘was’) and a new place (‘under my table’) are introduced. Moreover, since this is a direct quote, the ‘haya +

23 participle’ construction is interpreted as thematically discontinuous. In other words, the direct quote provides some background information that does not directly progress the narrative. Hence, here too the ‘haya + participle’ construction functions to introduce a discontinuous episode. The foreground of the narrative continues in sentences (d-e) in which Adoni-Bezeq is being taken to Jerusalem, where he died.

Finally, let us see how also in example (8) (repeated here as (11)) the ‘haya + participle’ construction functions as a discontinuous episode.

(11) ki ʕad ha-yam-im ha-hemma hayu bene yisrael meqattrim for until the-days the-those were.Q children-of Israel burning lo to.him ‘for up to that time the Israelites had been burning incense to him’ (2 Kings 18:4)

In the above example, the ‘haya + participle’ construction is found within another discontinuous environment, namely a subordinate clause (marked by ki ‘for’). Moreover, a new episode is introduced – the subject “the Israelites” and the past tense hayu (‘were’).

Therefore, we can conclude that the Past Imperfective (expressed by ‘haya + participle’) functions to introduce a discontinuous episode. To sum up, the following table compares between the meaning (tense and aspect) and function of the participle and the ‘haya + participle’ construction:

24 Table 3. The meaning and function of the participle with and without the aux. haya

participle haya + participle Aspect imperfective (durative) imperfective (durative) Tense relative (past/present/future) past Discourse Function continuous episode discontinuous episode

5.2 Mishnaic Hebrew

In a process that began in late Biblical Hebrew and continued more prominently in

Mishnaic Hebrew, the verbal system gradually changed towards a less relative and more absolute tense system in which the predominant reference time was time of speech/narration (Givón 2015, Gordon 1982, Polak 2009). As a consequence, the discourse functions of the verbal inflections (qatal and yiqtol) were bleached. Hence, discourse continuity and episode shifts were instead inferred based on the general organization of discourse. As we shall see later (in section 6.2), the irrealis meaning of the verbs was also bleached. The following table summarizes the main changes of the verbal system:

Table 4. The changes in the verbal system from Biblical to Mishnaic

Biblical Hebrew Mishnaic Hebrew qatal inflection anterior tense anterior tense (with more absolute uses) participle imperfective aspect imperfective aspect (with more present tense uses) yiqtol inflection posterior tense (usually perfective) future tense

The indeterminacy between a present tense reading and an imperfective reading for the participle facilitated the obligatorification of haya for the expression of the past imperfective

– hence, the two readings are now differentiated morphologically (Gordon 1982).

25 Rosén (1985) claimed that the ‘haya + participle’ construction was borrowed from

Aramaic, since in Aramaic the auxiliary hawa ‘was’ (the equivalent of haya) was used productively for tense inflections. Gordon (1985), on the other hand, claims that Aramaic did not cause this development but at most facilitated it, since this construction is but one development of a much larger change in the tense/aspect system of Hebrew (as explained earlier).

I do not intend to determine whether Aramaic had or did not have an influence on the development of the Mishnaic verbal system since any change that a language undergoes affects its inner structure, regardless of the source. Oftentimes linguists tend to disregard linguistic innovations that were borrowed from other languages, assuming that these innovations do not reflect the “real” grammar of the language. However, living languages are constantly in contact with other languages and each language changes gradually and slowly based on its own configuration and capability (Hopper & Traugott

2003). If language A borrows a certain structure or word from language B, that means that language A was able to integrate it into its grammar. We do not simply borrow everything that comes our way, we only borrow things that our language can contain and integrate into its system. Therefore, I see the grammaticalization of ‘haya + participle’ as an innovative process, regardless of external influence.

Having clarified that, let us now turn to some examples of ‘haya + participle’ in

Mishnaic Hebrew. In (12), we see two examples that express the past progressive:

26 (12) a. paʕam ‘aħat hayi.nu yoSvim […] we-naSva ha-ruaħ time one were.we.Q sitting […] and-blew.Q the-wind

‘once, we were sitting […] and the wind blew’

(Sifra Behuqqotay 2:5)

b. […] we-mac-‘u-hu Se-haya yoSev roS-o […] and-found-they-him.Q that-was.Q sitting head-his

we-rub-o ba-suka we-Sulxan-o betox ha-bait and-most-his in.the-sukkah and-table-his inside the-house

we-lo amr-u lo davar and-no said.they.Q to.him thing

‘[…] and they found him sitting with his head and the majority of his body inside the sukkah, while his table was in the house, and they did not say a thing to [stop] him.’

(Mishnah Sukkah 2:7)

Since the past progressive reading was not retained in Modern Hebrew, I shall focus my analysis on the past habitual reading, as exemplified in (13):

(13) a. Simon ha-cadik haya me-Sayarey kneset ha-gdola ‘Shimon the Righteous was from the remnants of the Great Assembly’

b. hu haya omer al SloSa dvarim ha-olam omed he was.Q saying on three things the-world stands ‘He would say, "On three things the world stands:

c. al ha-avoda al ha-tora we-al-gemilut xasadim ‘on the Torah, on the service and on acts of lovingkindness.”’

(Pirkei Avot 1:2)

27 (13) is a classic example of the Mishnaic past habitual. The text begins in (a) with an introduction about Shimon the Righteous. In the following sentence (b), ‘haya + participle’ is used to convey the past habitual (‘he would say’). Notice that here in sentence (b), there is no introduction of a new subject (Shimon was introduced in (a)), nor a new tense or place

(haya and ‘the great assembly’ were already mentioned in (a) as well). Hence, this use of the past habitual is clearly not episodic as in Biblical Hebrew. In fact, this innovative use is eventive as it depicts a particular event within a larger episode which was introduced in

(a). The following is another example of the same sort:

(14) a. raban yoxanan ben zakai kibel me-hilel u-mi-Samay ‘Rabban Yochanan ben Zakkai received [the tradition] from Hillel and Shammai’

b. hu haya omer im lamade-ta tora harbe al taxzik he was.Q saying if learned-you Tora much do-not hold

tova le-acmexa virtue for-yourself ‘He used to say: “If you have learned a lot of Torah, do not credit it favorably for yourself’

c. ki le-kax nocarta ‘because for this you were created.”’

(Pirkei Avot 2:8)

In (14), we begin with an introduction – Raban Yochanan, the subject, received the tradition from Hillel and Shammai. Here too, the introductory verb kibel ‘received’ is of the qatal inflection. In (b) the ‘haya + participle’ construction conveys the past habitual (‘he used to say’). Therefore, this past habitual is eventive.

28 Examples (13) and (14) represent an innovation that began in Mishnaic Hebrew – the ‘haya + participle’ construction began to be used in contexts where the past habit was attributed to a subject that was already introduced in the discourse. In other words, the past habitual construction was not limited to episodes as in Biblical Hebrew. Nonetheless, the following is an example for one case in which ‘haya + participle’ is episodic:

(15) a. rabi nexunya ben ha-kana haya mitpalel be-knisa-to rabbi nechunya ben ha-kana was.Q praying in-entrance-his

le-bet ha-mikdaS to-house the-hall ‘Rabbi Nechunya ben HaKanah would offer a brief prayer when he entered the study hall’

b. u-beyecia-to tefila kcara ‘and when he left’

c. amru lo ma makom le-tefila kazo? ‘They said to him: What is the nature of this prayer?’

(Mishna Berakhot 4:2)

In this example, the habit of praying before and after entering the study hall is introduced together with a subject (Rabbi Nechunya), a past tense (haya ‘was’) and a place (the study hall). Hence, here we see evidence for the retention of the Biblical episodic past habitual.

The narrative continues in sentence (c) with a qatal verb amru ‘they said’. This is evidence for the trend that was mentioned earlier regarding the reanalysis of qatal verbs as absolute (i.e. simple past). Since late Biblical Hebrew, qatal verbs were evaluated more based on time of speech/narrative and less based on other time references in the discourse.

29 Hence, the qatal inflection inherited the thematically continuous function that was restricted to (wa-)yiqtol verbs in early Biblical Hebrew (Givón 2015).

In the examples we saw so far, the reported habit does not take place during time of speech/narrative. All the things Shimon, Raban Yochanan and Rabbi Nechunya said are understood to be restricted to the past. Hence, I shall refer to such past habitual uses as non-current, for they do not take place during the current state of affairs.

The next grammaticalization stage of the Past Habitual in Mishnaic Hebrew regards its modal meaning. This construction was being used in fables in which the subject is a generic person who fits certain hypothetical circumstances. The subject is usually not real

(although it might be based on a real person), and the time and place of the situation is also undefined. In other words, in its next stage, the ‘haya + participle’ construction began to be used in hypothetical contexts. Consider the following example:

(16) a. ha-Sole’ax get le-iS-to we-hegia ba-Saliax […] the-sender.of divorce to-wife-his and-arrived to.the-messanger ‘[With regard to] one who sends a bill of divorce to his wife [via a messenger], and then catches up with the messenger […]’

b. ba-riSona haya ose bet din be-makom axer in.the-first was.Q doing house.of court in-place other

u-mevatl-o and-cancel-it ‘At first, a man [who had already sent his wife a bill of divorce by means of a messenger] would set up a religious court in a different place [from where the wife lived] and cancel [the bill of divorce]’

(Mishna Gittin 4:1-2)

30 Example (16) is situated in a fable which details the unfolding of a story revolving around a person who fits certain circumstances. The purpose of the fable is to propose a moral decision or command based on the man’s actions. In (a) the person is introduced (‘one who sends a bill of divorce’) and in (b) a certain habit is attributed to him (setting up a religious court). Since this person is generic, and there is no particular place or time in which it takes place, the habit is interpreted as hypothetical. Unlike examples (12)-(15) which describe a real situation in the past that no longer takes place, example (16) depicts a habit that never occurred. Thus, I shall refer to such uses of the ‘haya + participle’ construction as hypothetical past habitual.

To summarize this subsection, we saw that the ‘haya + participle’ construction in

Mishnaic Hebrew continued to convey the past imperfective – the past habitual and the past progressive meanings were retained from Biblical Hebrew. However, I could not find examples that involve a “perfect” reading. This is due to the fact that the qatal inflection was slowly restricted to the simple past or the “perfective” (Givón 2015). Moreover, we saw that the ‘haya + participle’ construction gathered eventive uses, although it retained its episodic function (15). Furthermore, we saw that the Past Habitual in Mishnaic Hebrew gathered a hypothetical reading (16). This reading will be crucial to the development of the

Counterfactual Conditional (section 6.2).

5.3 Modern Hebrew

In Modern Hebrew, the ‘haya + participle’ construction retained mostly its past habitual sense. Its past progressive reading is rare and dialectical (but see example (18)). The 31 following are examples of the modern Past Habitual. Notice that the Past Habitual can be translated to English either as “used to” or “would”.

(17) a. hu ra’a et ha-universita kol pa’am Se-hu haya holex he saw ACC the university every time that-he was.Q going la-gan.xayot to.the-zoo ‘he saw the university every time would go to the zoo’ (CoSIH: C412ND)

b. kulam hayu kor’im le-xa zaken everyone were.Q calling to-you old

‘everyone used to call you “old”’ (CoSIH: C212_2ND)

c. le-kol makom el-av haya holex, to-every place to-which were.Q going,

haya nose ito tik xum gadol were.Q carrying with-him bag brown big

‘he used to carry a big brown bag to every place he would go’ (HeTenTen) (18) kSe-ba-ti el-av hu haya kore sefer when-I-came.Q to-him he was.Q reading book

‘When I came to him, he was reading a book’ (Gordon 1982: 46)

The purpose of this section is to produce an answer to Rosén (1985) and Doron’s (2006) conundrum regarding the origin of the modern Past Habitual. To remind you, Rosén claimed that the modern Past Habitual was borrowed from Mishnaic Hebrew, whereas

Doron claimed that it was borrowed from Biblical Hebrew. In order to settle this question,

I will analyse not only the semantic meaning of the ‘haya + participle’ construction, but also its discourse function.

We have seen that in Biblical Hebrew the ‘haya + participle’ construction functioned to present a discontinuous episode (section 5.1). We have also witnessed that in Mishnaic

32 Hebrew this construction acquired eventive uses, which do not present a new construal but depict a particular occasion of a larger episode which was already introduced earlier in the discourse. Therefore, in this section I will conduct a synchronic corpus-based study in order to reveal the discourse function of the modern Past Habitual.

As we shall see, the results show that the Past Habitual has a predominantly eventive function rather than an episodic function. Hence, we will have evidence to support Rosén’s (1985) hypothesis that this construction was borrowed from Mishnaic

Hebrew (contra Doron 2006).

In order to make the analysis more interesting and reliable, I will compare the ‘haya

+ participle’ habitual (henceforth Periphrastic Habitual) to the habitual expressed by the anterior inflection qatal (henceforth Simple Habitual) (see Kalev 2017 for the analysis of this verbal inflection in Modern Hebrew). The aim of the analysis is to reveal the discourse function of each of the Habituals. From a pragmatic perspective, we ask the question – which discoursal or pragmatic factors predict the use of one habitual construction over the other? Once we have the answers for this question we will be able to answer the diachronic question regarding the origin of the ‘haya + participle’ construction in Modern Hebrew.

5.3.1 Introduction to the Hebrew Habituals

The Past Habitual is used to express a situation which occurs customarily on different occasions and is used to characterize a period of time in the (often) remote past (Comrie

1976). Modern Hebrew, like English, Spanish and other languages, has more than one form to express habituality. In this study, I will focus on two of the colloquial forms: The 33 Periphrastic Habitual and the Simple Habitual. The following are examples of the two

Habituals, quoted from Boneh & Doron (2008:2):

(19) The Simple Habitual:

yael nas’a la-avoda ba-otobus yael drove.Q to.the-work in.the-bus ‘Yael went to work by bus’

(20) The Periphrastic Habitual:

yael hayta nosa’at la-avoda ba-otobus yael was.Q going to.the-work in.the-bus ‘Yael used to go to work by bus’

Rosén (1977, 1985) notes that the Simple Habitual (19) is ambiguous, i.e. it can be interpreted as punctual (in which case Yael went to work by bus on a particular occasion) or as habitual (in which case Yael was accustomed to go to work by bus during a certain period of time). The Periphrastic Habitual, on the other hand, is restricted to the habitual interpretation, i.e. it has no punctual interpretation whatsoever.

In agreement with the traditional view of habituality (e.g. Comrie 1976, Bybee &

Pagliuca 1994), Boneh & Doron (2008) view the Periphrastic Habitual as an imperfective event. As evidence, they provide an example in which it overlaps with another (punctual) event:

(21) kSe-higati la-arec, yael hayta nosa’at ba-otobus when-arrived.I.Q to.the-country, Yael was.Q going in.the-bus ‘When I arrived in the country, Yael used to go to work by bus’ (Boneh & Doron 2008)

34 In the above example, Yael’s habit to go to work by bus was already taking place by the time the speaker arrived in the country. In other words, the speaker’s arrival in the country is within the period of time during which Yael went to work by bus. In comparison, the

Simple Habitual is ambiguous between two readings, an overlap reading and a consecutive reading:

(22) kSe-higati la-arec, yael nas’a ba-otobus when-arrived.I.Q to.the-country, Yael went.Q in.the-bus ‘When I arrived in the country, Yael went by bus’ (Boneh & Doron 2008)

In the above example, there is an additional possible reading in which Yael went to work by bus after the speaker arrived in the country. Thus, the Periphrastic Habitual is more imperfective than the Simple Habitual.

However, Boneh & Doron (2010) justly assert that the aspectual property of imperfectivity is not sufficient to account for the full semantic meaning of the Periphrastic

Habitual. In particular, the perfective/imperfective distinction cannot account for two crucial properties of the Hebrew Habituals: actualization and relevance for time of speech.

Actualization is the property which pertains to whether the denoted event occurred or did not occur. While the Simple Habitual is not necessarily actualized, the Periphrastic

Habitual is necessarily actualized. The following examples demonstrate this difference:

(23) a. dani limed ba-universita Danny taught.Q in.the-university ‘Danny taught at the university’

35 b. dani haya melamed ba-universita Danny was.Q teaching in.the-university ‘Danny used to teach at the university’

(Boneh & Doron 2008)

In (a), there is a marked interpretation in which Danny was a professor at the university but did not actually teach due to a lack of registered students. In (b), Danny is understood to have taught actual courses at the university – he was not necessarily a professor, but he surely did teach something at the university. Boneh & Doron (2008, 2010) support their claim with example (24). The (possible) context for this example is a situation in which the speaker (an employer) details the responsibilities of each of his employees, of which Meri is one.

(24) meri tipla ba-doar me-antarktika Meri handled.Q in.the-mail from-Antarctica ‘Meri handled the mail from Antarctica’

Based on the abovementioned context, tipla ‘handled’ is understood not as an actualized habit but as a habit which primarily characterizes the subject. Under such an interpretation, sentence (24) could be true even if Meri never received a single letter from Antarctica. This habitual function will be referred to as dispositional.

Both Habituals refer to the past tense. However, the relevance of the past event to time of speech differs between the Simple and the Periphrastic Habitual. The past habit expressed by the Simple Habitual is not necessarily limited to the past. In fact, the habit can still hold at time of speech. In (21a), it is possible that Danny is still teaching at the university

36 and in (24) it is possible that Meri still handles the mail from Antarctica. All the speaker intends to convey by using the Simple Habitual is that the habit took place in the past.

The Periphrastic Habitual, on the contrary, necessarily ends before time of speech.

Thus, Yael’s (20) habit to go to work by bus and Danny’s (21b) habit of teaching at the university cannot be understood as occurring at time of speech. This is further exemplified by the incompatibility of the Periphrastic Habitual with adverbs that include time of speech

(Boneh & Doron 2008):

(25) #mi-Snat 1981, hu haya me’aSen golwaz #since-year.of 1981, he was.Q smoking golwaz ‘#Since 1981, he used to smoke Gauloises’

The adverb mi-Snat ‘since’ expresses the point at which the habit began. The lack of an explicit end-point implies that the event still holds at time of speech. However, this sentence creates a pragmatic anomaly. The anomaly is caused by the incompatibility of

‘since’ and the Periphrastic Habitual. Thus, we can deduce that the Periphrastic Habitual must assume an end-point. Example (26) shows that when we use an adverb that expresses an end-point (ad ‘until’), the sentence is acceptable:

(26) ‘ad šnat 1987, hu haya me’ašen golwaz until-year.of 1987, he was.Q smoking golwaz ‘until 1987, he used to smoke Gauloises’

Based on this information, one can conclude that the Periphrastic Habitual encodes perfectivity. However, Boneh & Doron (2010) show that this is not the case. The perfective interpretation in the following example is cancelled with the expression ‘already then’:

37 (27) be-1990 le-ruti haya oto, in-1990 to-Ruti was.Q car ‘Ruti had a car in 1990’

aval kvar az hi hayta nosa’at la-avoda ba-otobus but already then she was.Q driving to.the-work in.the-bus ‘but already back then she would go to work by bus.’

(Boneh & Doron 2010:8)

In the above example, the periphrastic habit ‘would go to work by bus’ necessarily occurs at time of speech since it is followed by ‘already then’. Boneh & Doron (2010) conclude that the perfective interpretation8 is cancelled and thus it is not part of the (encoded) semantic meaning of the Periphrastic Habitual, but rather a pragmatic inference, or more specifically

– a conversational implicature (Grice 1975).

According to the classic Gricean meaning of the term, this is indeed an implicature. But following Ariel (2016), I would like to restrict the meaning of a conversational implicature to an implicated conclusion (see below). I claim that the perfective interpretation of the

Periphrastic Habitual in (27) is thus an implicated premise. The difference between an implicated conclusion and implicated premise is crucial not only for understanding the discourse profiles of the Habituals but also for revealing their encoded semantic meaning.

I will elaborate on this point in section 5.3.6. For now, let us see the difference between the two inferences.

8 I will use the term perfective interpretation to refer to the inference that the expressed habit ended before time of speech. Boneh & Doron (2010) refer to it as a “disjointness implicature”. However, since I claim it is not necessarily an implicature I use the general term interpretation. Moreover, as mentioned earlier, the term perfective is used in the narrow sense of ending before time of speech. The additional connotations of perfectivity (i.e. a punctual event which is viewed as a whole) are excluded from this definition. 38 A conversational implicature is part of the intended meaning of the utterance. It is thus an implicated conclusion that the speaker intends the addressee to infer (Ariel 2016).

Crucially, a conversational implicature can be cancelled (Grice 1975) and when it does, it must be done explicitly with words such as ‘not’ or ‘if’ (Ariel 2008: 29), or with a contradiction. Both methods of cancellation usually come after the utterance. If the contradiction does not trigger a pragmatic anomaly with the utterance, we can deduce that the pragmatic inference triggered by the utterance is a conversational implicature.

A background assumption (a term introduced by Searle 1980) is an implicit interpretation which the speaker does not necessarily mean to communicate. Thus, the speaker is not likely to entertain the interpretation or stay committed to it. Therefore, background assumptions, unlike conversational implicatures, are not intended by the speaker but rather assumed by them. For this reason, they are called implicated premises

(Ariel 2016). Premises or background assumptions cannot be cancelled in the same way conclusions are cancelled. Since premises are not intended by the speaker, the only way to cancel them is by expressing it implicitly before the utterance.

The phrase ‘already then’ in example (27) is not an explicit cancelling device, but rather an implicit cancelling device. Karttunen (1973) referred to such presupposition- cancelling devices as “plugs” for they block the formation of subsequent presuppositions.

In our example, ‘already then’ is a plug which blocks the perfective interpretation.

However, in other contexts, the perfective interpretation can be a conversational implicature. In the following example, a speaker who utters (a) can possibly utter (b) which contradicts the perfective interpretation. 39 (28) a. be-1990 ruti hayta nosa’at la-avoda ba-otobus ‘In 1990, Ruti used to go to work by bus’

b. hi adain nosa’at la-avoda ba-otobus ‘She still goes to work by bus’

The perfective interpretation of the Simple Habitual, on the other hand, is not a conversational implicature nor a background assumption. I claim that the speaker who utters the Simple Habitual does not intend the speaker to infer nor to take into consideration the end of the habit, since it has no relevance in the given context. If an addressee infers that the habit denoted by the Simple Habitual ended before time of speech, their inference is a truth-compatible inference (Ariel 2016). The following table summarizes the differences between the two Habituals:

Table 5. Summary of the Hebrew Habituals

Simple Habitual Periphrastic Habitual Form qatal inflection haya + participle Semantic meaning a situation which occurs customarily on different occasions and is used to characterize a period of time in the past Actualization undetermined actualized Aspect undetermined durative Modality dispositional undetermined The factual status of the Truth-compatible inference * Conversational implicature, or perfective interpretation * Background assumption

5.3.2 Research Questions

The purpose of the corpus analysis is to reveal the discourse function of the Habituals.

From a synchronic perspective, it will provide us with predictions as to when a speaker opts for one Habitual over the other. From a diachronic perspective, it will provide us

40 evidence for the origin of the Periphrastic Habitual. If its discourse function in Modern

Hebrew is eventive, we will have evidence to support Rosén’s hypothesis regarding its origin in Mishnaic Hebrew. If, on the other hand, the Periphrastic Habitual has more episodic functions, then we will have evidence to support Doron’s hypothesis which states it was adopted from Biblical Hebrew.

The following are the research questions that will guide this study:

1. Which discoursal factors predict the use of one Habitual over the other? i.e. what is

the discourse function of each of the Habituals?

2. Why can the Simple Habitual trigger dispositional readings while the Periphrastic

Habitual cannot?

3. What determines the factual status of the perfective interpretation?

5.3.3 The Corpus

The data of this study was gathered from two corpora of online written Hebrew. The first corpus is the “parallel corpus” which gathers text from articles on news, politics, sports, economics, and literature. The second corpus is HeTenTen. Similar to the former, it is based on online articles, but it also includes colloquial language taken from blogs and private conversations. It contains nearly 900 million words. Both of the corpora were tagged morphologically by Adler (2007), with a contribution of Noam Ordan. The following analysis is based on 93 Habituals – 41 Periphrastic Habituals and 52 Simple Habituals.

41 5.3.4 Methodology

Each Habitual expression was tagged for five parameters: a. Construction type: periphrastic/simple b. Transitivity: transitive/intransitive c. Clause dependence: independent/subordinate d. Discourse function: episode shift/elaboration/background

i. Episode shift: an event which introduces a new piece of discourse which includes

a new subject, a new tense/aspect and possibly more. The episode shift is also

used to provide a temporal or propositional frame for subsequent events.

ii. Elaboration: an event which provides more details to the previously mentioned

event. Elaboration is manifested by providing circumstances or details to the

discussed topic or event.

iii. Background: an event which is not directly related to the topic of the discourse

and does not progress it. This event usually expresses the speaker’s/narrator’s

comment on a particular event. e. Adverbial frame: iterative/period/circumstance/periodically

i. Iterative: the adverb expresses the repetitive nature of the event with words such

as kol/bexol ‘every’9.

ii. Period: the adverb expresses the period of time which the habit characterizes.

9 Unlike Boneh & Doron (2008), I do not consider adverbs such as pa’am/pa’amaim ‘once/twice’ as iterative, since they are in fact multiplicative – they express how many times the event happened (hence they are telic) and not the iterativity of the event (which is atelic). 42 iii. Circumstance: the adverb expresses the occasion or reason for the occurrence of

the habit. This is expressed with terms such as ka’aSer ‘whenever’, merov

‘because of too many/for all’ or with a proposition.

iv. Periodically: the adverb conveys that the denoted event happened periodically

or sporadically with terms such as lifamim ‘sometimes’.

5.3.5 Results and Discussion

We begin with observing the interaction between construction type and discourse function.

The data shows that the episode-shift function is almost exclusive to the Simple Habitual

– it is responsible for 55% of the Simple Habitual uses and only 2% of the Periphrastic

Habitual uses. The background function is exclusive to the Simple Habitual (37% of its uses). The elaboration function is used significantly more with the Periphrastic Habitual

(98% of its uses) than with the Simple Habitual (8% of its uses).

Figure 4. Percentage of discourse function within construction type

120%

100% 98% 80%

60% 55% 40% 37% 20% 8% 2% 0% 0% Episode Shift Background Elaboration

Simple Habitual Periphrastic Habitual

43 Let us observe the following example10:

(29) a. Christopher Robin, whose parents called him (Se-horav kar’u lo) Billy Moon

b. spent most of his time (bila et rov zmano), as was customary then, in the company of his nanny. c. During the weekends and holidays, [they] used to spend time (hayu mevalim) in Ashdown Forest. d. He would climb on the trees (haya metapes), observing (mitbonen) the animals and plants e. and the stories which his father used to tell (siper), when he came back (haya xozer) from his trips in the woods, provided (sipku) inspiration to the stories which his father wrote (katav) in Winnie-the-Pooh.

In (a-b), the narrator begins by introducing new participants to the discourse (Christopher

Robin [the subject] and his nanny), a new space-construal (the company of his nanny) and a new tense (past tense). The narrator uses the qatal inflection for this purpose (bila ‘spent’).

Hence, the Simple Habitual in sentence (b) marks an episode shift. Moreover, the qatal verb in sentence (a) fulfils the function of backgrounding – the narrator provides background information about the subject. This piece of information (that he was also called Billy

Moon) is not crucial for the purpose of the narrative but it does provide some background information that could be relevant for the story as a whole. Notice that both verbs in (a) and (b) characterize their subjects – they do not depict an accidental habit but a habit which characterizes the dispositions of the subject (in this case Christopher Robin).

10 For ease of transcription, I translated the surrounding context of the Habituals to English but preserved the Hebrew transliteration for the Habituals themselves. All Hebrew transliterations are marked in italics. The Simple Habitual is marked in bold and the Periphrastic Habitual is underlined. An English word or phrase in square brackets signals the lack of this form in the Hebrew version. 44 In (c) the narrator uses the Periphrastic Habitual to elaborate on the previously mentioned habit (i.e. Christopher’s spending time with his nanny) by expressing a particular event that fits this circumstance (i.e. spending time in Ashdown Forest). The narrator continues elaborating about the particular ocurrences of this habit in sentence (d).

In (e) we witness another episode shift. The narrator uses the qatal inflection because the story-telling event (siper) and the following consequences (‘providing inspiration’ and ‘writing Winnie-the-Pooh’) are not conceptualized as part of the elaborated habit. Instead, the argument is built upon the elaborated habit expressed by the

Periphrastic Habitual (haya xozer). The following is another example of the same sort:

(30) a. Indeed, it was a risk to let him run free.

b. But on the other hand, it was worth it to take the risk.

c. Because seconds after [he] entered (nixnas) the woods, [his] screams dissipated (namogu), until [they] disappeared (ne’elmu) completely. d. And he would find (haya moce) himself a clearing in the forest and run (ma’avir) his fingers through the sand or look (mabit) at a piece of wood e. that interested him (Se-inyena oto).

In the above example, although the Simple Habituals in (c) do not introduce a new subject

(‘he’ is inferred), they do introduce new habitual events that describe the subject and the situation he used to be immersed in – i.e. entering the woods and stopping to scream. In

(d) we are provided with particular occasions that elaborate on the general habit of visiting the woods (‘finding a clearing’, ‘running his fingers’ and ‘looking’). And finally, in (e) we witness another background use of the Simple Habitual.

45 The previous examples showed us the prototypical functions of the Habituals. The following is an example of a Periphrastic Habitual with an episodic function. Notice that this example accounts for only 2% of the uses of the Periphrastic Habitual:

(31) When I was little, my grandfather used to take (haya loke’ax) me to see the model of the second temple in the Holy Land Hotel.

What we see in (31) is an introduction of a habit at the beginning of a narrative. This function is what Polak (2009) refers to as an “anchor”. However, here the anchor is not a qatal verb but the whole ‘haya + participle’ construction. The following is an example of a

Simple Habitual with the elaboration function:

(32) He seemed (nira) so emotionally connected. He turned (hiSir) his gaze to our eyes.

In (32), the second qatal verb functions as the Simple Habitual (‘turned’ [his gaze]) which elaborates on the topic introduced by the previous event (i.e. that he seemed to be emotionally connected).

Next, the interaction between construction type and clause dependence reveals a significant difference between the two Habituals (p<0.03) – while the Simple Habitual is used almost equally (p>0.87) in both independent and subordinate clauses, the Periphrastic

Habitual is used significantly more in independent clauses (p<0.005).

46 Figure 5. Percentage of clause dependence within construction type

80% 69% 70% 60% 51% 49% 50% 40% 31% 30% 20% 10% 0% Simple Habitual Periphrastic Habitual

Subordinate Independent

When we look deeper into the various functions of each of the Habituals, the picture becomes clearer. Within the Simple Habituals the episode shift function is used significantly more in independent clauses (p<0.007). The background function, on the contrary, is used significantly more in subordinate clauses (p<0.000). Within the

Periphrastic Habituals, the elaboration function is used significantly more in independent clauses (p<0.000).

Figure 6. Percentage of clause dependence within discourse function

100%

80% 93% 76% 79% 60%

40%

20% 24% 7% 21% 0% Episode Shift Background Elaboration

Independent Subordinate

Moving on to the interaction between construction type and adverbial frame, we observe no significant difference in the prevalence of adverbial frames within each Habitual: 41%

47 of the Simple Habituals and 45% of the Periphrastic Habituals have an adverbial frame.

When we observe the type of adverbial frame within each construction type, we reveal one significant difference – the Periphrastic Habitual is accompanied by a circumstantial adverb significantly more than the Simple Habitual (p<0.01). Moreover, iterative adverbs tend to be used more with the Simple Habitual than with the Periphrastic Habitual. This is a near-significant finding (p=0.058).

Figure 7. Percentage of adverbial frame type within construction type

70% 65% 60% 50% 40% 40% 33% 30% 18% 20% 13% 13% 12% 10% 6% 0% Iterative Period Circumstance Periodically

Simple Periphrastic

Consider the following examples:

(33) […] every evening I explained to him (hisbarti lo) that one does not simply become deaf all of a sudden. (34) When I was little, my grandfather used to take (haya loke’ax) me to see the model of the second temple in the Holy Land Hotel. (35) Whenever we called his name (kSe-karanu be-Smo), he would not lift (lo haya merim) his gaze. (36) Sometimes, he would not calm down (haya nirga) for hours.

In (33), the Simple Habitual (‘explained to him’) is accompanied by the iterative adverb

‘every evening’. In (34), the Periphrastic Habitual (‘used to take me’) is accompanied by a period habit (‘when I was little’). In (35), the Periphrastic Habitual is under the scope of the 48 circumstantial habit expressed by the Simple Habitual ‘whenever we called his name’. In

(36), the Periphrastic Habitual ‘would not calm down’ is accompanied by the “periodical” adverb ‘sometimes’.

Finally, the interaction between construction type and transitivity reveals no significant differences (p=0.73).

5.3.6 The Function of the Habituals in Discourse

Based on the corpus analysis I conclude that the Simple Habitual expresses an episodic habit. Semantically speaking, the Simple Habitual conceptualizes an unmarked habit which is prototypically viewed as a whole (unless it overlaps with another event). This habit often provides a temporal frame for other habits expressed by the Periphrastic

Habitual (examples (29), (35)). The discourse function of the Simple Habitual is a topical one – to introduce new information to the discourse, i.e. a new subject, a new tense/aspect

(past imperfective), or a new construal.

Another function of the Simple Habitual concerns the addition of background information which the speaker/narrator finds relevant for the discourse/narrative as a whole (as in example (29)). This function is similar to the episode shift function in that it shifts the focus of the discourse. In this case, the focus is shifted from the expressed habit to a parenthetical comment of the speaker/narrator. Moreover, the functions of the Simple

Habitual differ in their information status: the backgrounding function is used more in subordinate clauses and the episode shift is used more in independent clauses. Hence, the episode-shift function is more salient in discourse than the backgrounding function. 49 Furthermore, The Periphrastic Habitual expresses an eventive habit. The eventive habit expresses a particular habitual occasion. For this reason, the Periphrastic Habitual shows more imperfective properties than the Simple Habitual (e.g. it has an overlap reading but not a consecutive reading; see section 5.3.1). The Periphrastic Habitual thus functions to elaborate on an episodic habit by expressing a particular occasion which characterizes it (as in example (29)). This function is used predominantly in independent clauses which demonstrates their saliency in discourse.

The following illustration demonstrates the interaction between the episode shift, backgrounding and elaboration functions based on example (29). Black arrows represent elaboration. Red arrows represent elaboration achieved by characterization and green arrows represent the backgrounding function. SH stands for Simple Habitual and PH stands for Periphrastic Habitual.

Figure 8. Illustration of the Habitual functions

‘it was worth it to take the risk’

SH: ‘…he entered the woods…’

PH: ‘He would find PH: ‘run his fingers PH: ‘look at a piece of

himself a blind spot’ through the sand’ wood’ event SH: ‘…which he found interesting…’ episode proposition

Based on these findings, we can understand better the factual status of the perfective interpretation. When the Periphrastic Habitual has the function of presenting an episode,

50 the perfective interpretation is a conversational implicature – the speaker intends the addressee to infer that the spoken habit ended before time of speech and no longer takes place. When the Periphrastic Habitual has the function of elaboration, the perfective interpretation is a background assumption since the particular occasion expressed by the

Periphrastic Habitual already takes place within an explicit temporal frame. In other words, the perfective interpretation is the implicated premise which the Periphrastic

Habitual relies on when elaborating.

After answering two of my research questions (questions 1 and 3 regarding the

Habitual functions and the perfective interpretation respectively), I will now turn to question number 2, repeated here: why can the Simple Habitual trigger dispositional readings (as in example (24)) while the Periphrastic Habitual cannot? To answer this question let us look at example (24) repeated here as (37) with its continuation:

(37) meri tipla ba-doar me-antarktika Meri handled.Q in.the-mail from-Antarctica ‘Meri handled the mail from Antarctica.’

mikevan Se-lo haya doar kaze, haya la harbe zman panuy ‘Since there was no such mail, she had a lot of free time.’ (Boneh & Doron 2010: 12-13)

What the above example shows us is that unlike the Periphrastic Habitual (which would not be acceptable in this context) the Simple Habitual does not require actualization.

However, such interpretations are marked as they require a very specific context (I did not find such examples in my corpus). I claim that example (37) is acceptable because of the very context that facilitates it. When the topic of the discourse is people’s responsibilities, 51 the event expressed by the qatal inflection does not require actualization precisely because it is not being referred to but rather attributed to the subject.

Moreover, I claim that the attribution of a habit to a subject requires more than the qatal inflection and an appropriate context. It also requires a particular verb constellation, namely a transitive verb that depicts a process. The following examples demonstrate this:

(38) a. #meri mac’a pitriyot ba-ya’ar (Transitive Achievement) ‘Meri found mushrooms in the forest.’

b. #meri halxa ba-ya’ar (Intransitive Activity) ‘Meri walked in the forest’ c. #meri yaSva al ha-safsal (Intransitive State) ‘Meri sat on the bench’ d. meri kisxa et ha-deSe (Transitive ‘Meri mowed the lawn’ Accomplishment) e. mikevan Se-lo haya (#pitriyot/#ya’ar/#safsal/deSe), haya la harbe zman panuy ‘Since there were no (#mushrooms/#forest/#bench/lawn), she had a lot of free time.’

The unacceptability of (a) and (e) shows that a telic event without a process (finding mushrooms in the forest) is not compatible with its contradiction in (e). Similarly, an event with a process which is intransitive (b) is also contradicted by (e). (c) shows that an intransitive state also cannot be understood as non-actualized, given its contradiction with

(e). However, (d) which depicts a transitive verb with a process is compatible with (e).

Therefore, I conclude that the dispositional reading of the Simple Habituals is achieved under the following conditions: (1) a supportive context with a topic that discusses people’s responsibilities, (2) a transitive verb, (3) a verb that depicts a process.

Thus, I refer to the habit in (37) as an attributive habit as it is being attributed to the subject

52 rather than being referred to. For this reason, the Simple Habitual allows non-actualized habits. For comparison, the Periphrastic Habitual depicts a referential habit since only an actualized habit can be referred to.

This is why the prominent adverbial frame of the Periphrastic Habitual is a circumstance. While the Simple Habitual specializes in depicting more permanent and independent states of affairs, the Periphrastic Habitual specializes in elaboration which is circumstantial in nature. The following table summarizes the discourse profiles of the

Hebrew Habituals:

Table 6. Summary of the discourse profiles of the Hebrew Habituals

Simple Habitual Periphrastic Habitual Semantic representation Episode (the habit as a whole) Event (a particular occasion of the Proposition (the attribution of habit) the habit to the subject) Discourse function Episodic Eventive (elaboration) (episode shift / backgrounding) Prominent adverbial frame undetermined Circumstance Clause dependence Independent (episode shift) Independent (elaboration) Subordinate (background) Actualization undetermined Actualized Aspect undetermined Durative The factual status of the Truth-compatible inference Conversational implicature, or perfective interpretation Background assumption

5.3.7 Conclusions

In this section I have determined the discourse functions of the Hebrew Habituals based on a corpus of colloquial and literary written Hebrew. I have shown that the Periphrastic

Habitual refers to an eventive habit. This habit functions in discourse prominently to

53 elaborate on a previously mentioned habit by expressing a particular occurrence of it.

These discourse profiles explain better the imperfectivity and the actualization of the habits expressed by the Periphrastic Habitual. The Simple Habitual was shown to refer to an episodic habit which is presented as a whole. In discourse it shifts the focus of attention to present a new subject, tense and sometimes place. Additionally, it provides a temporal frame for the occasions of the habit expressed by the Periphrastic Habitual. This study shows that semantic meanings are inseparable from their discourse functions. In order to understand why a certain from has its particular semantic meaning, one has to observe how this form is used in discourse.

5.4 Summary

In this section I have shown the grammaticalization path of the ‘haya + participle’ construction in Hebrew. I have shown that in Biblical Hebrew this construction conveyed the past imperfective and functioned to introduce a discontinuous episode. In Mishnaic

Hebrew, the meaning of the construction remained the past imperfective, however it lost its distinctive discourse function – it was no longer restricted to episodes or thematically discontinuous contexts; hence it became eventive. Furthermore, I have shown that in

Modern Hebrew the ‘haya + participle’ construction was narrowed to the past habitual while being prominently eventive – the past habitual event is used to elaborate on an episodic habit expressed by the qatal inflection (i.e. the Simple Habitual).

Hence, I have provided evidence to support the hypothesis that the Modern ‘haya + participle’ construction was borrowed from Mishnaic Hebrew rather than Biblical Hebrew. 54 Therefore, the data supports Rosén’s hypothesis and contradicts Doron’s hypothesis regarding the origin of the construction. Moreover, I’ve shown that the Simple Habitual, expressed by the qatal inflection, functions as an episodic habitual – a function that in

Biblical Hebrew was reserved to the yiqtol inflection.

6. THE HYPOTHETICAL CONDITIONAL

The element that links between the Past Habitual (section 5 above) and the Hypothetical

Conditional in Modern Hebrew is of course the ‘haya + participle’ construction. The following figure demonstrates this:

Figure 9. The Past Habitual and the Hypothetical Conditional in Modern Hebrew

Past Habitual haya + participle Hypothetical Conditional im haya + participle, haya + participle

The question that arises by observing these constructions is – what is the function of ‘haya

+ participle’ in each of these constructions, and why did the Hypothetical Conditional employ a construction that expresses the past habitual? To answer these questions, I will begin by introducing the relationship between counterfactuality and hypotheticality. Later on, I will show the Biblical Counterfactual Conditional and claim that in Mishnaic Hebrew the Past Habitual began to carry hypothetical and counterfactual readings, which were eventually adopted by the Counterfactual and Hypothetical Conditional constructions.

A conditional sentence has three parts: an ‘if’ participle, an antecedent (a.k.a. protasis) and a consequent (a.k.a. apodosis). Following Ziegeler (2000), a hypothetical message is defined here as a message which the speaker intends the addressee to construe as unreal.

55 For example, the sentences “I would choose B” or “If only I could fly” express hypotheticality – the message of these sentences is to construe an imaginary situation. The construal of a hypothetical situation could serve various functions – such as expressing wish and desire.

A subtype of hypotheticality is counterfactuality. A counterfactual message is a hypothetical message which is intended by the speaker to counter a real past situation.

Hence, semantically, two equally important concepts are conveyed – the real and the hypothetical (Givón 1994). For example, by uttering the sentence “I should have bought bread”, the speaker conveys both a realis assertion (a) “I didn’t buy bread” and a hypothetical message (b) “I would have bought bread”. The dissonance (i.e. the lack of relevance) between the two possibilities gives rise to implicature (c) “I wish I had bought bread”. Therefore, as we shall see in this section, the main functions of expressing a counterfactual message is (a) to explain the past, and (b) to support moral judgments

(Byrne 2016). To summarize, the following figure demonstrates the relationship between counterfactuality and hypotheticality:

Figure 10. Demonstration of the relationship between Counterfactuality and Hypotheticality

Counterfactuality

Hypotheticality

The following table summarizes the conceptual and discoursal differences between counterfactuality and hypotheticality: 56 Table 7. The differences between Counterfactuality and Hypotheticality

Hypotheticality Counterfactuality Construal 1. Hypothetical 1. Real in the past 2. Hypothetical alternative to (1) Function 1. Make a moral evaluation 1. Explain the past 2. Express personal stance 2. Make a moral evaluation 3. Express personal stance

6.1 Biblical Hebrew

The Counterfactual Conditional in Biblical Hebrew carries the form: ‘lu qatal, verb’. lu is a counterfactual if-particle – it indicates that the hypothetical situation expressed by the qatal verb in the antecedent is opposite to the situation that really happened in the past. In the consequent there is a verb which functions to indicate the hypothetical consequence of the hypothetical circumstance expressed by the antecedent. In the following example, this verb is of the qatal form as it carries a performative (speech-act) utterance (see Polak 2009 for this reading of the qatal inflection).

(39) a. wa-yomar axay bney imi hem ħay yawweh ‘And he replied, “They were my brothers, the sons of my mother; as the Lord lives,’

b. lu hexye-tem otam, lo harag-ti etxem if revive-you.pl.Q them no kill-I.Q you.pl ‘if you had saved them alive, I would not kill you.”‘

(Judges 8:19)

The context of the above examples is the scene where Gideon is informed about the killing of his brothers from two of his soldiers. It is then that Gideon tells the soldiers that if they had not killed his brothers he would have spared them from death. In the verse that

57 follows, Gideon indeed kills the soldiers Zebah and Zalmunna. Hence, the antecedent

(harag-ti etxem; lit. “I kill you”) is interpreted as a speech act. Let us now turn to another example in which the consequent does not express a speech act, but a hypothetical consequence. This is done with the yiqtol verb form:

(40) a. I thought to scatter them and blot out the memory of them from humankind;

b. but I feared provocation by the enemy, for their adversaries might misunderstand and say, “Our hand is triumphant; it was not the Lord who did all this.”

c. They are a nation void of sense; there is no understanding in them.

d. lu xaxm-u, yaskil-u zot if wise-they.Q, understand-they.Y this ‘If they were wise, they would understand this;’ e. yavin-u le-axarit-am understand-they.Y to-end-their they would discern what the end would be.’

(Deuteronomy 32:26-29)

In (a-b), God (Yawweh) states the fear that if he had scattered the people as he intended to, the result would be that other nations would credit this victory to themselves and not to

God. In (c), he goes on to justify his argument by claiming that these nations are senseless.

Sentence (d) continues this justification with the Counterfactual Conditional. It begins with the antecedent lu xaxmu ‘if they were wise’ (which they are not) and continues with the consequent yaskilu zot ‘would understand this’ (which is improbable). Hence, the purpose of this Counterfactual Conditional is to convey the message that other nations are not capable of understanding God’s actions and intentions.

58 Referring to the grammaticalization of hypothetical markers in English, Bybee

(1992:26) claims that “[...] It is not the past tense alone that is contributing the hypothetical meaning, but rather the past in combination with a modal verb, a subjunctive mood, a hypothetical marker (such as ‘if’)”. And indeed, in Hebrew as well, the counterfactual if- particle lu is the element that puts the realis assertion expressed by the qatal verb under the scope of hypotheticality. In the consequent, there is another verb which could be of the qatal or yiqtol inflection, depending on the function. In our examples we saw that a qatal consequent expresses a speech act and a yiqtol consequent expresses a hypothetical consequence.

6.2 Mishnaic Hebrew

Having analysed the Biblical Counterfactual Conditional, we can now observe the

Mishnaic Counterfactual Conditional which for the first time replaces the verb in the consequent with ‘haya + participle’:

(41) ilu zaxu israel, le-Yod”Alef yom hayu nixnasim la-arec if won.Q Israel, for-eleven day were.Q entering to.the-country ‘If Israel had been meritorious, then in eleven days they would have entered Eretz Yisrael’

(Sifrei Devarim 2:2)

59 To begin, the counterfactual if-particle lu had changed its form to ilu, which most probably derived from the expression im lu ‘if if’ through assimilation11. The antecedent verb was still of the qatal form (zaxu ‘won’), but the most dramatic change is in the consequent, which now consists of ‘haya + participle’ (hayu nixnasim ‘[they] would enter’).

The question that comes to mind is: what explains the use of ‘haya + participle’ in the consequent instead of an inflected verb (qatal or yiqtol) like in Biblical Hebrew? The answer to this question relies on the interpretation of the ‘haya + participle’ construction as the Past

Habitual. As I mentioned in section 5.2, the Past Habitual in Mishnaic Hebrew had two modal readings – a non-current and a hypothetical. Let us be reminded with example (13), repeated here as (42).

(42) a. Simon ha-cadik haya me-Sayarey kneset ha-gdola ‘Shimon the Righteous was from the remnants of the Great Assembly’

b. hu haya omer al SloSa dvarim ha-olam omed he was.Q saying on three things the-world stands ‘He would say, "On three things the world stands:

c. al ha-avoda al ha-tora we-al-gemilut xasadim ‘on the Torah, on the service and on acts of lovingkindness.”’

(Pirkei Avot 1:2)

11 This is according to the Wiktionary website. I speculate that this transition occurred due to the bleaching of the meaning of lu to counterfactuality alone (leaving out conditionality) (c.f. Lehmann 2002 for semantic bleaching). At the same time im, which expresses conditionality, underwent the process of obligatorification (Traugott 2010), hence im was required by the grammar to appear in any conditional clause, even if it is counterfactual. Thus, im emerged to precede lu, which over time was phonologically reduced to ilu. 60 The above is an example of a non-current Past Habitual – Shimon used to utter his saying

(b-c) but the habit of uttering it does not take place during the time of the narrative. Let us now be reminded with the hypothetical Past Habitual of example (16) repeated here as (43):

(43) a. ha-Sole’ax get le-iS-to we-hegia ba-Saliax […] the-sender.of divorce to-wife-his and-arrived.Q to.the-messanger ‘[With regard to] one who sends a bill of divorce to his wife [via a messenger], and then catches up with the messenger […]’

b. ba-riSona haya ose bet din be-makom axer in.the-first was.Q doing house.of court in-place other

u-mevatl-o and-cancel-it ‘At first, a man [who had already sent his wife a bill of divorce by means of a messenger] would set up a religious court in a different place [from where the wife lived] and cancel [the bill of divorce]’

(Mishna Gittin 4:1-2)

Here, the ‘haya + participle’ construction conveys a hypothetical habit which does not counter any real event. The purpose of this text is to portray a fable from which a moral law would be concluded. One of the indicators for this is the generic identity of the subject

(“one who sends a bill of divorce”). Hence, here the hypothetical Past Habitual refers to a completely (past) hypothetical situation which has no real alternatives. I claim that it is this hypothetical ‘haya + participle’ which manifested itself in the Counterfactual Conditional

(ilu qatal, haya + participle). To understand exactly how the transition occurred in the

Counterfactual Conditional consequent from a verb to ‘haya + participle’, we need to have a larger perspective on the transition of the Hebrew verbal system from Biblical to

Mishnaic.

61 According to Givón (1977, 2015) and Gordon (1985), in Biblical Hebrew both the qatal and the yiqtol verbal inflections expressed various irrealis meanings. An irrealis qatal expressed low-probability, consequences, speech acts and more. An irrealis yiqtol expressed future tense, commands, and customs (Polak 2009). The transition to Mishnaic

Hebrew eliminated most of the irrealis readings of these verbs. Gordon (1985) claims that this is a result of the growing uses of the participle in the present tense, which led to the absolute tense system we know today in Modern Hebrew.

Hence, I claim that now (in Mishnaic Hebrew) that the irrealis senses of the verbs were bleached, the new hypothetical Past Habitual became the optimal and the most salient candidate for the expression of hypotheticality. Not only because it expresses a hypothetical event, but also because it relies on one. As the Past Habitual it relies on the premise that the story is an imaginary fable, and as the consequent it relies on the antecedent which expresses a counterfactual (ultimately hypothetical) situation. Therefore, the salient discourse profile (Ariel 2008) of the ‘haya + participle’ construction as the Past

Habitual paved the way for its grammaticalization as a hypothetical consequent in conditional sentences.

However, in some contexts even the antecedent carried the ‘haya + participle’ construction (instead of a qatal verb):

(44) a. amar lo ilu hayiti yodea Se-hu ken, said.Q to.him if was.I.Q knowing that-he such,

lo hayiti noder no was.I.Q vowing He said to him, “If I had known that it is so, I would not have vowed.”

62 b. we-hetira rabi akiva ‘And Rabbi Akiva released his vow.’

(Mishna Nedarim 9:5)

In this example, the reason for a ‘haya + participle’ in the antecedent instead of a qatal verb is the discoursal status of this Counterfactual Conditional – it is a thematically discontinuous piece of information, particularly a direct quote. Hence, although the tense system of Mishnaic Hebrew was advancing towards the absolute tense system, example

(44) is proof for the retention of the discontinuous function of haya. As we saw in section

5.1, when the Past Habitual was used in a direct quote the participle required the verb haya.

This was exemplified with example (10) repeated here as (45):

(45) a. wa-yomer adoni-bezeq and-said.Q Adoni-Bezeq ‘And Adoni-Bezeq said’

b. Sibʕ-im melax-im […] hayu melaqqt-im taħat Sulħa-ni seventy kings […] were.Q gathering under table-my ‘”Seventy kings […] used to gather / have been gathering scraps under my table”’

(Judges 1:7)

So far, we’ve seen two variants of the Counterfactual Conditional in Mishnaic Hebrew:

i. ilu qatal, haya + participle (thematically continuous)

ii. ilu haya + participle, haya + participle (thematically discontinuous)

In addition, the ‘haya + participle’ construction participates in Hypothetical Conditional sentences (which are set completely within the irrealis domain – no real alternative exists).

63 Hypothetical Conditionals in Mishnaic Hebrew begin with the if-particle im, since im expresses conditionality but not counterfactuality. Consider the following example:

(46) im haya meSamər-an, tehor-im if was.Q guarding-them, pure.pl ‘If he was guarding them, they are pure’

(Mishna Tahorot 8:3)

The above Hypothetical Conditional is of the form: im haya + participle, adjective. Here, the

‘haya + participle’ construction appears in the antecedent and not in the consequent as in the first Counterfactual Conditional (lu qatal, haya + participle). I claim that this is further evidence for the hypothetical meaning of the ‘haya + participle’ construction.

In the (thematically continuous) Counterfactual Conditional, the counterfactual if- particle ilu makes redundant the introduction of hypotheticality (in the form of haya + participle in the antecedent), since ilu already encodes counterfactuality. In the

Hypothetical Conditional, on the other hand, the if-particle im expresses only conditionality. This is why ‘haya + participle’ is required for the expression of a hypothetical condition.

The consequent of the Hypothetical Conditional of the above example contains only an adjective. This is because the purpose of the consequent is to express a moral decision that is universally true. Therefore, the purpose of the above Hypothetical Conditional is to convey under what (hypothetical) circumstances a moral decision (regarding purity) is valid. The consequent of the Hypothetical Conditional can also contain a verb:

64 (47) haya roxev al ha-ħamor, yered was.Q riding on the-donkey, dismount.Y ’If one was riding a donkey, he should dismount from it [while he prays]’

(Mishna Berakhot 4:5)

In the above example, the function of the verb in the consequent (yered ‘should dismount’) is to express the moral command that applies to a hypothetical situation in which a man rides a donkey while praying on it. Notice that here there is no if-particle. There are numerous examples of this sort in the Mishna, which shows that conditionality is often understood from the context to the extent that the if-particle im can be omitted.

To summarize, in Mishnaic Hebrew there are two conditional constructions which employ the ‘haya + participle’ construction – the Counterfactual Conditional and the

Hypothetical Conditional. The Counterfactual Conditional (ilu qatal, haya + participle OR ilu haya + participle, haya + participle) functions to conceptualize an alternative to a particular event that took place in the past. The antecedent construes a counterfactual condition (e.g. Israel being meritorious [they were not]) and the consequent portrays its hypothetical outcome (e.g. entering Israel in eleven days [instead of forty years]).

In comparison, the Hypothetical Conditional conceptualizes hypothetical situations in order to assist in the evaluation of moral issues. The antecedent depicts a completely hypothetical condition (e.g. riding a donkey while praying) and the consequent determines the universal moral command that applies to it (e.g. dismounting from it).

Importantly, I have shown that the ‘haya + participle’ construction, which emerged to express the past imperfective, developed a hypothetical meaning which later was

65 recruited by the Counterfactual Conditional and the Hypothetical Conditional to express a hypothetical condition and consequence.

6.3 Modern Hebrew

The most dramatic change in the expression of the Counterfactual Conditional in Modern

Colloquial Hebrew is its merger with the Hypothetical Conditional – now it carries the form of ‘im haya + participle, haya + participle’. For this reason, from a formal perspective I shall refer to it as the Hypothetical Conditional, for it supports both counterfactual and purely hypothetical readings.

As we can tell, the reason for the merge is the if-particle im. In Modern Colloquial

Hebrew, the (Biblical) counterfactual marker lu and the (Mishnaic) counterfactual conditional particle ilu became restricted to literary and formal registers. As a consequence, im, which expresses mere conditionality, took over the place of the Mishnaic counterfactual conditional particle ilu in the expression of counterfactual conditionals.

Furthermore, both the antecedent and the consequent of the Hypothetical

Conditional contain the ‘haya + participle’ construction. We saw a similar pattern with one of the Mishnaic Counterfactual Conditionals (44) (ilu haya + participle, haya + participle) which appears in thematically discontinuous contexts. However, in Modern Hebrew haya

(by itself) lost its discontinuous function12. Therefore, the modern Hypothetical Conditional

12 As we saw in section 5.3, the modern Past Habitual construction (haya + participle) retains its discontinuous function as an elaborator to a larger episode. Nonetheless, the data shows us that this discontinuous discourse 66 employs the auxiliary haya regardless of the thematic structure of the discourse. The following example (48) is evidence – the speaker utters the conditional construction in an independent sentence as part of the topic of the discourse:

(48) a. im hayiti matxil la’asot et ha-balagan ha-ze if was.I.Q beginning to.do ACC the-mess the-this

im ha-horim Seli with the-parents my ‘If I had started that whole mess with my parents’

b. az hi lo hayta mityaxeset be-emun then she no was.she.Q treating in-trust ‘then she wouldn’t have treated [it] trustfully’

(CoSIH: P423aND_sp)

In the antecedent, the speaker conveys a real situation (“I did not start that mess with my parents”) and a parallel hypothetical situation (“I did start that mess with my parents”).

The pragmatic purpose for expressing the two messages is resolved in the consequent (“she wouldn’t have treated it trustfully”). This means that by uttering (48) the speaker intends to explain the past and by doing so to modulate emotional experience – two of the main pragmatic functions of counterfactual conditionals (Byrne 2016). Therefore, the above example shows us that counterfactuality lost its explicit status to become an (implicit) implicature. The following is an example for a hypothetical (non-counterfactual)

function was retained for the Past Habitual construction alone and not for any other use of the ‘haya + participle’ construction, such as in the modern Counterfactual Conditional. 67 conditional. The speaker is a vegan woman and her message was posted in a Paleo diet

Facebook group.

(49) a. im hayiti sama laxem al ha-Sulxan netax basar xay if was.I.Q putting for.you on the-table chunk meat raw ‘If I were to put a chunk of raw meat on the table for you’

b. hayitem mesugalim le-exol oto ve-leakel bli le-haki? were.you.Q able to-eat it and-to-digest without to-puke? ‘would you be able to eat and digest it without puking?’

(Facebook Paleo group)

As we can tell, the speaker intends the Paleo group members to envision a situation in which she puts in front of them a raw chunk of meat. Obviously, she does not actually do it, nor does she compare it with some other real situation. Hence, the purpose of uttering antecedent (a) is to construe a particular hypothetical situation in the minds of the addressees. The pragmatic function of this Hypothetical Conditional is revealed in the consequent (b) – the speaker asks the participants whether they would eat this chunk of raw meat. Therefore, the speaker intends to evaluate the moral values of the Paleo dieters.

To sum up, the Modern Hypothetical Conditional carries the form ‘im haya + participle, haya + participle’. It expresses all kinds of hypothetical conditionals – both counterfactual and non-counterfactual. Moreover, unlike in Mishnaic Hebrew, it is not bounded to a particular strand of discourse coherence, i.e. it appears in both thematically continuous and discontinuous contexts.

68 6.4 Summary

In this section I have shown the diachronic grammaticalization path of the Hypothetical

Conditional. I began by introducing the Biblical Counterfactual Conditional (lu qatal, verb).

Then, I claimed that in Mishnaic Hebrew the hypothetical Past Habitual (haya + participle) replaced the verb in the consequent. This happened due to (1) the bleaching of irrealis meanings of verbs (qatal/yiqtol) and (2) the growing saliency of the hypothetical meaning of the ‘haya + participle’ construction.

Finally, I have described the Hypothetical Conditional in Modern Colloquial

Hebrew. We have seen that the if-particle changed from the explicit counterfactual conditional marker ilu to im which expresses general conditionality. Hence, in Modern

Colloquial Hebrew there is one construction which expresses both counterfactual and non- counterfactual conditionals. Therefore, it was named the Hypothetical Conditional, since the form of this construction maps onto the higher abstract category of hypotheticality. The following table summarizes the grammaticalization path of the Hypothetical Conditional:

Table 8. The hypothetical conditional constructions of Hebrew if-particle Antecedent Consequent Construction name Stage of Hebrew lu qatal verb Counterfactual Biblical Conditional ilu qatal haya + participle Counterfactual Mishnaic Conditional ilu haya + participle haya + participle Counterfactual Mishnaic Conditional im haya + participle haya + participle Hypothetical Modern Conditional

69 7. THE HYPOTHETICAL PREDICTIVE

Now we turn to the last development of the Hypothetical Conditional – the Hypothetical

Predictive. I define the Hypothetical Predictive as a construction of the form ‘haya + participle’ which conveys a hypothetical situation that the speaker predicts would happen under certain implicit circumstances. As we shall see, the Hypothetical Predictive falls into two modal categories:

i. Deontic Hypothetical Predictive is defined as a prediction which the speaker

implicitly wishes to occur or not to occur. This kind of modality is also known as

volitive modality (Narrog 2005).

ii. Epistemic Hypothetical Predictive is defined as a prediction which the speaker has

no interest as occurring or not occurring. Instead, the prediction is based on the

probability of the hypothesized situation. This kind of modality is also known as

non-volitive modality (Narrog 2005).

From a pragmatic point of view, the two modal categories of the Hypothetical Predictive serve different functions. Deontic Hypothetical Predictives function to evaluate the hypothetical event by expressing a stance towards it (see generally Du Bois 2007).

Epistemic hypothetical predictions, on the other hand, function to express the certainty of the speaker regarding the occurrence of the hypothetical event (see also evidential modality Narrog 2015).

70 Having defined the semantic and discoursal properties of the Hypothetical Predictive, we shall now present some examples. To begin, the following is a deontic Hypothetical

Predictive:

(50) a. lo amart li ‘you didn’t tell me’ b. hayiti xofefet et ha-se’ar al ha-boker was.I.Q washing ACC the-hair on the-morning’ ‘I would have washed my hair first thing in the morning’ (CoSIH: P423aND_sp2)

In the above example, the speaker utters the deontic hypothetical prediction (b) “I would have washed my hair…”. We know it is deontic because her previous utterance (a) “you didn’t tell me” implies that the speaker wished the addressee had told her (that she’s about to come visit) so that she would have washed her hair and been prepared. In other words, the speaker’s utterance in (a) is an implicit condition for the hypothetical prediction she utters in (b). And the Hypothetical Predictive in (b) expresses the speaker’s aversion of the past situation (not washing her hair) and a preference for a better situation (having washed her hair). Now let’s consider the following epistemic Hypothetical Predictive:

(51) taxles ba-arec hayu moc’im et ha- Sita le-gale’ax in.fact in.the-country were.Q finding ACC the-system to-shave le-xa et ha- kesef to-you ACC t he-money ‘In fact, in Israel (they) would find a way to take your money’ (HeTenTen)

Unlike the speaker in (50), this speaker does not express his stance towards the occurrence of the hypothetical event (“they would find a way to take your money”). Instead, the 71 speaker expresses his high certainty regarding the occurrence of the hypothetical event, i.e. that in Israel they would find a way to take your money. This interpretation is based on two main cues – first, the hypothesized prediction “would find a way to take your money” is attributed to a 3rd person subject (‘they’), meaning the speaker does not take responsibility for the occurrence of the hypothetical event. Secondly, the speaker begins his utterance with the discourse marker taxles which means roughly ‘in fact’ or ‘practically’, thus marking that the evaluation of the speaker is based on his knowledge of the world.

Therefore, (51) is an example of an epistemic Hypothetical Predictive.

It is important to stress that the particular kind of modality (deontic or epistemic) of the Hypothetical Predictive is not encoded in the meaning of the construction, but rather it is inferred based on contextual cues (as shown above) such as the person of the subject, whether there are implicit conditions and how implicit they are.

In order to help the linguist to differentiate between the two modal interpretations

I have created an acceptability test. The premise behind the test is that by explicating the deontic interpretation (of preference or dispreference) we can judge whether the

Hypothetical Predictive coheres with the explication (in which case it is deontic) or does not cohere (in which case it is epistemic).

The test is as follows: If we can embed the ‘haya + participle’ construction in a clause that opens with: haya ma’adif im (‘would prefer if’) and create an acceptable sentence, then the Hypothetical Predictive is deontic. If the embedding results in a pragmatic anomaly, then the Hypothetical Predictive is epistemic. For example, the deontic Hypothetical

Predictive in (50) is repeated here as (52) with the test: 72 (52) a. lo amart li ‘you didn’t tell me’ b. hayiti ma’adifa im [hayiti xofefet et ha-se’ar al ha-boker] was.I.Q prefer if [was.I.Q washing ACC the-hair on the-morning]’ ‘I would prefer if [I would have washed my hair first thing in the morning]’

The test creates an acceptable sentence, hence (52) is deontic. The following is example (51) repeated here as (53) with the test:

(53) a. #hayiti ma’dif im ‘I would rathar if… b. #taxles ba-arec hayu moc’im et ha- Sita le-gale’ax in.fact in.the-country were.Q finding ACC the-system to-shave le-xa et ha-kesef to-you ACC the-money ‘In fact, in Israel (they) would find a way to take your money’

The test fails in (53) – there is a clash between the explication (“I would rather if”) and the hypothetical prediction (“they would find a way to take your money”), meaning that the speaker did not intend to convey whether or not he likes it that in Israel they would take your money. Hence, the test supports the categorization of (53) as epistemic.

7.1 Hypothesis and Prediction

I hypothesize that the Hypothetical Predictive emerged into the grammar from the consequent of the Hypothetical Conditional. This is based on Givón’s (1994) hypothesis of syntactic liberation. The idea is that the antecedent or consequent of a

Hypothetical/Counterfactual Conditional often emerge as an independent construction to convey a various range of modal expressions (e.g. the past subjunctive in Spanish).

73 I speculate that this development occurred in contexts where the consequent was more relevant to the discourse than the antecedent. Once the speaker does not indicate the conditions for the realization of the hypothetical situation, she is abstracting away from the external circumstances relevant for her prediction. Hence, I claim that this abstraction from external circumstances paves the way for focusing on her internal subjective stance

(preference/dispreference) towards the hypothetical situation she expresses. In other words, it is deontic predictions that pave the way for the grammaticalization of the

Hypothetical Predictive.

If my hypothesis is right, I expect deontic predictions to be more prevalent in

Hypothetical Predictives in independent sentences than in the consequents of Hypothetical

Conditionals. Likewise, I expect epistemic predictions to be more prevalent in the consequents of Counterfactual Conditionals than in Hypothetical Predictives.

7.2 Methodology

In order to evaluate my hypothesis, I conduct a corpus-based study on the Hypothetical

Conditional and the Hypothetical Predictive. If these two constructions indeed pattern in a complementary distribution in which the former is used for epistemic predictions and the latter for deontic predictions, then we can support the hypothesis that the Hypothetical

Predictive emerged from the consequent of the Hypothetical Conditional to express a subjective attitude of the speaker.

i. The data is based on 321 haya clauses from three different corpora – CoSIH,

HeTenTen and Facebook groups. 74 ii. Each clause was marked as conditional or simple.

iii. Each ‘haya + participle’ construction was marked for

a. Modality (epistemic/deontic)

b. Clause type (independent/conditional consequent)

c. Person inflection (1/2/3)

iv. The results were calculated and compared based on a chi-square test.

7.3 Results

The interaction between the modality of the ‘haya + participle’ construction and clause type reveals significant differences. Deontic predictions are significantly more prevalent in independent clauses (as the Hypothetical Predictive) (64%) than in the consequent of

Hypothetical Conditionals (36%) (p<0.000). The opposite pattern is evident for epistemic predictions – they appear significantly more in the consequents of Hypothetical Predictives

(87%) than in independent clauses (13%) (p<0.000). This is demonstrated in the following figure:

Figure 11. Interaction between clause type and modality type for ‘haya + participle’

100% *

80% * 87%

60% 64% Independent 40% Consequent 36% 20% 13% 0% Deontic Epistemic

75 Now consider an example from each category. We have already seen an example for a

Hypothetical Predictive in an independent sentence in (50). But now I would like to show an example of the same modality and clause type but with an expression of dispreference instead of preference:

(54) iSit lo hayiti soxer mime-na afilu ofanaim personally no was.I.Q renting from-her even bicycle axrey ha-post ha-ze after the-post the-this ‘Personally, I would not event rent a bicycle from her after this post’ (Facebook real estate group)

The above example is a reply to a woman who made an angry post in a Facebook real- estate group blaming people who complain about rent prices in the Tel Aviv area. The speaker in (54) expresses his personal stance against that woman by claiming he would not even rent a bicycle from her (let alone an apartment). Hence, the above is an example for an expression of dispreference (with the negator lo ‘no’) using the Hypothetical Predictive.

Such examples represent 64% of Hypothetical Predictives in independent clauses. We have also seen an example for an epistemic Hypothetical Predictive in (51). But for the sake of diversity, I provide here another example:

(55) a. mazal Se-ha-erev rak hitxil kan ve-ani be-karov yoce liStot ‘Fortunately, the evening just began here and I’m about to go out drinking’ b. axeret lo hayiti sored et bilbuley ha-sexel Selxa otherwise no was.I.Q survive ACC confusions.of the-mid yours ‘Otherwise, I wouldn’t have survived your nonsensical bullshit’ (Facebook veganism group)

76 In (55), the speaker envisions an alternative hypothetical situation (i.e. not going out and listening to the addressee) in order to express his certainty that he would not survive the addressee’s talking. It is true that a personal stance is expressed here – the speaker explicitly says he will not tolerate the addressee’s talking. However, this stance does not emanate from the Hypothetical Predictive directly. Instead, the negative stance of the speaker is based on his certainty of the occurrence of the hypothetical event. Now let us turn to examples of the Hypothetical Conditional. The following is an example for a Hypothetical

Conditional with an epistemic consequent:

(56) a. im lo hayinu mamci’im xec va-keSet if no were.we.Q inventing arrow and-bow ‘If we hadn’t invented the bow and arrow’ b. lo hayinu cadim klum no were.we.Q hunting nothing ‘we wouldn’t have hunted anything’ (Facebook veganism group)

In (56), the speaker expresses a Hypothetical Conditional. The consequent (b) expresses the epistemic prediction of the speaker. Hypothetical Conditionals with epistemic consequents are the majority of the Hypothetical Conditionals (87%). The following is a Hypothetical

Conditional with a deontic consequent. Such examples comprise only 13% of the

Hypothetical Conditionals:

(57) a. gam im hayiti dati o ma’amin also if was.I.Q religious or believer ‘Even if I was religious or a believer’ b. lo hayiti roce le-hizdahot im Sum zerem Selahem no was.I.Q wanting to-identify with no sect theirs

77 ‘I wouldn’t want to identify with any of their sects’

(Facebook atheism group)

In addition to testing the interaction between modality and clause type I have also checked the interaction between clause type (independent/consequent) and the person of the subject (1st/2nd/3rd). The results show that when the Hypothetical Predictive appears in independent clauses, the subject is mostly in the 1st person (in 70% of the cases) – a statistically significant result (p<0.001). In comparison, within consequents there are no significant differences – 1st and 3rd person have equal prevalence. The following figure summarized these results:

Figure 12. Interaction between clause type and subject person for ‘haya + participle’

80% 70% 60% 70% 50% 1 40% 42% 43% 2 30% 3 20% 27% 10% 2% 15% 0% Independent Consequent

7.4 Summary

To summarize this section, I have presented the Hypothetical Predictive construction and traced its origin to the consequent of the Hypothetical Conditional. First, I introduced the two modal types of the Hypothetical Predictive – epistemic and deontic. According to

Givón’s (1994) syntactic liberation, I claimed that the Hypothetical Predictive emerged when

78 a consequent of a Hypothetical Conditional surfaced as an independent sentence.

Furthermore, following Ariel (2008) I hypothesized that this grammaticalization path occurred due to a salient discourse profile of the ‘haya + participle’ construction. Since explicit subjective messages are salient in discourse, I further hypothesized that it is the deontic sense of the Hypothetical Predictive that triggered the emergence of the

Hypothetical Predictive in independent clauses.

I predicted that if my claims are true, we should find more deontic senses of the

Hypothetical Predictive in independent clauses than in consequents of Hypothetical

Conditionals. To test it, I conducted a corpus-based study in which I collected examples of both constructions from corpora of colloquial written Hebrew. The results showed that indeed there are significantly more deontic Hypothetical Predictives (in independent clauses) than deontic consequents of Hypothetical Conditionals. Furthermore, there were significantly more epistemic senses in the consequents of Hypothetical Conditionals than in Hypothetical Predictives in independent clauses. The results support my hypothesis regarding the emergence of the Hypothetical Predictive into the grammar. Deontic subjective uses of the Hypothetical Conditional are those that triggered the syntactic liberation of the consequent. This grammaticalization path is characterized by the general trend of subjectification (Traugott 2010).

8. DISCUSSION

In this thesis, I have investigated three constructions in Modern Hebrew that carry the auxiliary haya ‘was’ – the Past Habitual (haya + participle), the Hypothetical Conditional 79 (im haya + particple, haya + participle) and the Hypothetical Predictive (haya + participle). In this section I will analyze the diachronic and synchronic processes that led the grammaticalization path of haya all the way from predication (its function as a copula) to prediction (its participation in the Hypothetical Predictive). Hence, this discussion excludes stages of development that do not lead directly to the development of the

Hypothetical Predictive. That includes the Modern Past Habitual and the Biblical

Counterfactual Conditional which were analyzed in previous sections. The following figure maps the discussion to come:

Figure 13. The grammaticalization path that leads to the Hypothetical Predicitve haya ‘was’ (a)

Past Imperfective Past Habitual (b)

(c)

Counterfactual Conditional Hypothetical (d) Conditional

(e) Hypothetical Predictive

Biblical Mishnaic Modern Hebrew Hebrew Hebrew

The purpose of this discussion is to characterize this grammaticalization path from three points of view – formal, semantic and discoursal. This multi-faceted analysis serves three functions: (a) to deduce correlations between form, meaning and function, (b) 80 consequently, to deduce the patterns that govern the grammaticalization of the haya constructions, and (c) to use these patterns to reveal the cognitive factors that shape language change.

We shall begin with the transition of the grammatical function of haya from a copula to an auxiliary in the Past Imperfective in Biblical Hebrew (change (a) in Figure 13). From a grammatical point of view, the change is clear – haya lost its predicative function in order to provide tense to the participial verb in the Past Imperfective. This process is called narrowing, for it narrows down the conceptual and even procedural meaning that the form conveys (Carston 2002). Although haya was narrowed to express the anterior tense alone, its predicative meaning was not completely lost – it was adopted by the construction (haya

+ participle) as a whole. Hence, the Past Imperfective functions as the predicate, with no need for an explicit predicator (i.e. copula), just like ordinary verbs.

Hence, we see a change in the scope of predication. When haya functions as a copula it has a scope over the whole proposition since it links between the subject and the predicate

(i.e. what is asserted about the subject). But when haya functions as an auxiliary its scope is reduced to the level of the construction – its function is reduced to provide the anterior tense to the participle. The relation to the subject is hence mitigated by the construction, which by itself predicates an event. Hence, haya no longer functions on the propositional level but rather on the episodic level in Biblical Hebrew and later on the eventive level in

Mishnaic Hebrew (section 4.1). The following figure demonstrates the grammaticalization of the scope of haya:

81 Figure 14. The grammaticalization of the scope of haya haya (copula) BH

haya (auxiliary) BH

haya (auxiliary) haya (auxiliary) MH MH

event event

episode proposition

The (diachronic) evidence for this transition was exemplified in (6) – an ambiguous use of the ‘haya + participle’ construction in which the participle could have a nominal or a verbal reading (i.e. shepherd or shepherding). Under the nominal reading, haya functions as a copula which predicates the identity of Joseph (as a shepherd), while under the verbal reading haya provides tense to the action of shepherding that Joseph participated in. The latter interpretation reflects the (synchronic) process of reanalysis. Reanalysis is a crucial part of grammaticalization – the addressee “understands a form to have a structure and a meaning that are different from those of the speaker” (Hopper and Traugott 2003: 50).

Hence, at this stage, the past imperfective interpretation arises as a conversational implicature.

Interestingly, we witness two opposing trends of change – on the one hand we observe narrowing of function (from copula to an auxiliary) and on the other hand we see what

Himmelmann (2004) referred to as host-class expansion – haya was expanded to occur with not only nouns and adjectives but also with participles. However, narrowing and

82 expansion are observations of two different levels of analysis. While narrowing observes the change of the grammaticalized item (the auxiliary in our case), host-class expansion concerns the context of the form (the possible contexts of haya) regardless of meaning or function. These two processes occurred simultaneously – the contexts of haya expanded while at the same time it developed a new narrower abstract function (as an auxiliary).

Moreover, the predicative function of haya persisted in all stages of Hebrew.

The next stage of development (change (b) in Figure 13) concerns the conventionalization of the Past Imperfective in Mishnaic Hebrew. Here we witness an additional context expansion – the Past Imperfective is no longer limited to discontinuous episodes. Now it occurs with an eventive function which is thematically dependent on previous episodes (examples (13)(14)). However, here too the episodic function was not completely lost – it was retained (example (15)).

A piece of evidence for the conventionalization and grammaticalization of the Past

Imperfective is also found in the obligatory status of haya within the construction (unlike in Biblical Hebrew). This development represents a fundamental property of grammaticalization which Lehmann (1995) has called obligatorification (see also Traugott

2010). Thus, we see two trends of change in the Past Imperfective – an increased dependency between haya and the participle and at the same time expansion of the functions of the construction. These two processes often go hand in hand as Kiparsky

(2012) has shown.

Crucially, as part of the context expansion, the Past Imperfective occurred in fables which serve to provide a moral lesson. Hence, the past imperfective event itself became 83 hypothetical – as it was completely within an irrealis modality (example (16)). This kind of context expansion is called semantic-pragmatic context expansion (Himmelmann

2004). The context of the wider discourse affects the semantic and pragmatic interpretation of the Past Imperfective – it forces the reader to assume a hypothetical reading for the past imperfective event, for the purpose of understanding the moral lesson.

The hypothetical Past Imperfective became frequent in Mishnaic Hebrew as it conventionalized to convey unreal situations and their consequences. At this intermediate stage, the hypothetical interpretation was a conversational implicature. However, the saliency of this implicature eventually triggered the encoding of hypotheticality in particular contexts. This is evidenced by the integration of the ‘haya + participle’ construction into the consequent (and antecedent) of the Counterfactual Conditional and the antecedent of the Hypothetical Conditional. As we saw (in section 6.2), the hypothetical

Past Imperfective eventually occupied the constructional slots that were previously limited to qatal and yiqtol verbs with an irrealis sense. This development (change (c) in Figure 13) is the result of three processes: (a) the narrowing and semantic bleaching of the irrealis meaning of inflected verbs (section 5.2), (b) the growing saliency of the hypothetical sense of the Past Imperfective and (c), the narrowing of the realis sense of the Past Imperfective.

Thus, the ‘haya + participle’ construction expanded its syntactic context

(Himmelmann 2004) to occur in counterfactual and hypothetical conditionals. Its occurrence within the Counterfactual Conditional shows us the development of the counterfactual sense of the Past Imperfective. When the Past Imperfective is used in independent clauses it refers to a past habit / progressive event which does not take place 84 during time of speech. However, within the Counterfactual Conditional (mostly due to the counterfactual conditional marker ilu ‘if’) it refers to an irrealis event that stands in opposition to the real event.

Later on, the situation gets even more abstract in the Hypothetical Conditional – the

‘haya + participle’ part refers to an irrealis event which does not stand in opposition to another event. In other words, what we see is that hypotheticality became encoded in the

‘haya + participle’ construction within the Mishnaic Hypothetical Conditional. Here again we see the two trends we saw earlier – on the one hand the ‘haya + participle’ construction expands its syntactic and discoursal context and on the other hand it develops a new more abstract meaning of hypotheticality. As Traugott and Dasher (2002) have shown, pragmatically induced invited inferences must first become salient before they entrench into encoded meaning. And indeed, example (46) shows us that hypotheticality became so salient for the ‘haya + participle’ construction that even the if-particle im could be omitted.

At the same time, from a semantic point of view, the ‘haya + participle’ construction loses its imperfective meaning. Hence it is narrowed to a hypothetical / counterfactual event.

The next stage of development occurs in Modern Hebrew (change (d) in Figure 13).

For some reason, during the revival of the Hebrew language, the counterfactual conditional particle ilu became restricted to formal and literary register and the Hypothetical

Conditional construction received the form of im haya + participle, haya + participle. Hence, this construction is no longer limited to purely hypothetical (non-counterfactual) cases (as in Mishnaic Hebrew) nor to counterfactual cases – but to all kinds of hypotheticality

(counterfactual or not). This again can be explained in terms of expansion – due to the loss 85 of the counterfactual marker ilu the discourse context of the Modern Hypothetical

Conditional expanded to allow all kinds of hypotheticality. Hence, here, unlike in previous changes, discourse expansion is correlated with the broadening of hypothetical situations that the construction conveys. However, this would not have been possible if not for the narrowing of the hypothetical scope of the if-particle (from ilu to im).

Finally, we reach the last development – the emergence of the ‘haya + participle’ construction as the Hypothetical Predictive (change (e) in Figure 13). I claimed and have shown (in section 7) that this construction emerged as an independent construction (in independent clauses) by being extracted from the consequent of the Hypothetical

Conditional. This process was previously described as syntactic liberation (Givón 1994; see also Evan 2007 on insubordination). Based on the corpus-study I have found that there are more deontic uses to the Hypothetical Predictive than to the consequent of the

Hypothetical Conditional. Hence, I claim that by liberating itself from the consequent of the Hypothetical Conditional, the ‘haya + participle’ construction adopted a more salient and personal meaning – expressing the speaker’s stance.

I’ve shown that by leaving out the conditions required for the fulfilment of the prediction, the Hypothetical Predictive emerged into the grammar by focusing on the speaker’s attitude towards the occurrence of the hypothetical event. This process is characterized by subjectification (Traugott 2010). Importantly, although the subjective

(deontic) sense of the Hypothetical Predictive is the prominent and salient one, it is not the only one. We saw there are also epistemic senses to the construction. Therefore, the

86 subjective (deontic) interpretation of the Hypothetical Predictive is a conversational implicature.

To sum up, we have observed several patterns of change – context expansion (host- class/syntactic/discoursal), semantic and functional narrowing and broadening, obligatorification, and subjectification. Now, I shall answer the more general question that arises – is there a general pattern to the grammaticalization path as a whole? The answer is yes – context expansion was evident in all stages of grammaticalization, whether it was host-class, syntactic, discoursal or any combination of the above.

Secondly, in all stages of grammaticalization, the new grammaticalized form emerged due to its saliency in discourse. The Past Imperfective conventionalized in

Mishnaic Hebrew due to its expansion to eventive contexts and consequently its increase in frequency. The hypothetical Past Imperfective replaced the inflected verbs in the counterfactual and hypothetical conditionals due to its saliency as hypothetical and bleaching of the irrealis sense of the inflected verbs. Finally, the Hypothetical Predictive emerged in salient contexts in which the message of the prediction carried a highly subjective purpose. Hence, this study provided evidence for Ariel’s (2008) observation that “linguistic change is mediated by discourse pattern, specifically, by the salient discourse pattern” and that “[s]alient discourse patterns, no matter what their source is, are a prerequisite for grammaticization.” (ibid: 189).

Additionally, we witnessed a general direction of the grammaticalization path from conceptual to procedural (see Blakemore 1987). This change was accompanied by an ever- increasing abstraction. First, the formation of the Past Imperfective is characterized by the 87 combination of two different components – the copula haya, which has a procedural

(grammatical) function and the participle, which carries the conceptual part13. Together, they form the Past Imperfective, which by itself is a conceptual construction, for it conceptualizes a particular event. In Mishnaic Hebrew, the hypothetical sense of the construction detaches a little more from the depiction of a concrete concept.

This detachment, or abstraction, continues when the ‘haya + participle’ construction participates in the Hypothetical and Counterfactual Conditional. There, it depicts a hypothetical event which is not necessarily durative nor real. The zenith of this process is manifested in the emergence of the Hypothetical Predictive – the ‘haya + participle’ construction not only depicts a hypothetical prediction but also a message to the addressee as to how to understand this prediction within the pragmatic context. This last stage can also be described as a speech-act oriented progress (Narrog 2015). The following figure summarizes the changes discussed in this section.

13 I use the term “procedural” to refer not only to the (often conscious) process of deriving a pragmatic inference, but also to grammatical items that instruct (often unconsciously) the addressee as to how to interpret the syntactical structure of the utterance. 88 Figure 15. Summary of the grammaticalization path from predication to prediction

9. CONSTRUCTIONAL POLYSEMY

In this brief section, I will talk about the polysemy of the ‘haya + participle’ construction in

Modern Hebrew. As we saw, this construction functions as either the Past Habitual or the

Hypothetical Predictive. In order to know what allows the speaker to distinguish between the two polysemous readings, we need to observe the discourse profile of each of the constructions. That includes both the inner properties of these constructions and their wider context. The following table summarizes exactly this:

89 Table 9. The discourse profiles of the two readings of the ‘haya + participle’ construction

Past Habitual Hypothetical Predictive Person 3rd person (70% of its uses) 1st person (70% of its uses) Tense/Aspect past imperfective The prediction relies either on a (past/present) counterfactual event OR on a hypothetical event Modality Between realis and irrealis Irrealis: mostly deontic (64%) (refers to a past event with no specific time, place or even subject) Discourse Eventive and elaborative Express personal attitude or function perspective

10. CONCLUSIONS

In this thesis I have traced the grammaticalization path that connects three constructions in

Modern Hebrew with the auxiliary haya ‘was’: The Past Habitual (haya + participle), the

Hypothetical Conditional (im haya + particple, haya + participle) and the Hypothetical

Predictive (haya + participle). I traced the origin of the Past Habitual to the Biblical Past

Imperfective. I showed how the Past Imperfective grammaticalized in Mishnaic Hebrew by expanding its function to present an event which elaborates on a previous episode. I claimed and showed based on a corpus analysis that the modern Past Habitual is similar in function to its Mishnaic equivalent.

Moreover, I showed that in Mishnaic Hebrew the Past Imperfective expanded to hypothetical contexts which facilitated its emergence as an antecedent and consequent of the Hypothetical Conditional. Finally, based on a corpus analysis, I provided evidence for my claim that the Hypothetical Predictive emerged into the grammar by the process of syntactic liberation from the consequent of the Hypothetical Conditional.

90 The thesis ended with discussion regarding the grammaticalization patterns that characterize this grammaticalization path. I have shown that in all stages of development, any grammaticalized item is characterized by the following features: (a) it was retained in the language (the copula, Past Habitual, Hypothetical Conditional and Hypothetical

Predictive persisted in Modern Hebrew), (b) it emerged due to its salient discourse function, (c) it emerged due to context expansion and (d) it became more procedural and less conceptual compared with its predecessor.

91 REFERENCES

Ariel, M. (2008). Pragmatics and Grammar. Cambridge: Cambridge University Press.

Ariel, M. (2016). Revisiting the Typology of Pragmatic Interpretations. Intercultural

Pragmatics, 13(1), 1-35.

Bomhard, A. R. (1984). Toward Proto-Nostratic: A New Approach to the Comparison of Proto-

Indo-European and Proto-Afroasiatic. Amsterdam/Philadelphia: John Benjamins

Publishing.

Boneh, N., & Doron, E. (2008). Habituality and the Habitual Aspect. Theoretical and

Crosslinguistic Approaches to the Semantics of Aspect, , 321-347.

Boneh, N., & Doron, E. (2010). Modal and Temporal Aspects of Habituality. In M.

Rappaport-Hovav, E. Doron & I. Sichel (Eds.), Lexical semantics, syntax, and event

structure (pp. 338-404). Oxford: Oxford University Press.

Bybee, J. (1992). The Semantic Development of Past Tense Modals in English and Other

Languages. UNM, Albuquerque:

Bybee, J. (1994). The Grammaticization of Zero: Asymmetries in Tense and Aspect Systems.

In J. Bybee, & W. Pagliuca (Eds.), Perspectives on grammaticalization (pp. 235-254).

Amsterdam: Benjamins.

Bybee, J. L., & Pagliuca, W.The evolution of future meaning. Papers from the 7th International

Conference on Historical Linguistics, 108-122.

Bybee, J., Perkins, R., & Pagliuca, W. (1994). The Evolution of Grammar: Tense, Aspect, and

Modality in the Languages of the World. Chicago: University of Chicago Press.

92 Chafe, W. (1987). Cognitive Constraints on Information Flow. In R. S. Tomlin

(Ed.), Coherence and grounding in discourse: Outcome of a symposium, eugene, oregon, june

1984 (pp. 21-51). Amsterdam/Philadelphia: John Benjamins Publishing Company.

Chomsky, N. (1968). Language and Mind. New York: Harcourt, Brace & World.

Comrie, B. (1976). Aspect. Cambridge: Cambridge University Press.

Comrie, B. (1985). Tense. Cambridge: Cambridge University Press.

Croft, W. (2001). Radical Construction Grammar: Syntactic Theory in Typological Perspective.

New York: Oxford University Press.

Croft, W. (2012). Verbs: Aspect and Causal Structure. New York: Oxford University Press.

Dahl, Ö. (2004). The Growth and Maintenance of Linguistic Complexity.

Amsterdam/Philadelphia: John Benjamins Publishing Company.

DeLancey, S. (1997). Mirativity: The Grammatical Marking of Unexpected

Information. Linguistic Typology, 1, 33-52.

Diewald, G. (2006). Context Types in Grammaticalization as

Constructions. Constructions, 1(9)

Diewald, G., & Smirnova, E. (2010). Paradigmaticity and Obligatoriness of Grammatical

Categories. Acta Linguistica Hafniensia, 41(1), 1-10.

Dik, S. C. (1981). Functional Grammar. Dordrecht/Cinnaminson NJ: Foris.

Doron, E. (2006). Rosén Al Ha-Semantika Šel Maarexet Ha-Poal be-Ivrit [Rosén on the

Semantics of the Hebrew Verb System]. Haivrit Veakhyoteha [Hebrew and its Sisters], the

University of Haifa, 6-7, 249-268.

93 Du Bois, J. W. (2007). The Stance Triangle. Stancetaking in Discourse: Subjectivity, Evaluation,

Interaction, 164, 139-182.

Du Bois, J. W. (2014). Towards a Dialogic Syntax. Cognitive Linguistics, 25(3), 359-410.

Evans, N. (2007). Insubordination and its Uses. Finiteness: Theoretical and empirical

foundations (pp. 366-431). New York: Oxford University Press.

Fitzmaurice, S. (1998). Grammaticalisation, Textuality and Subjectivity: The Progressive

and the Anglo-Saxon Chronicle. In D. Stein, & R. Sornicola (Eds.), The virtues of

language. history in language, linguistics and texts. papers in memory of thomas frank (pp. 21-

49). Amsterdam: John Benjamins.

Gesenius, W. (1857). Gesenius's Hebrew and the Chaldee Lexicon to the Old Testament Scriptures.

London: Samuel Bagster and Sons, Paternoster Row.

Givón, T. (1977). The Drift from VSO to SVO in Biblical Hebrew: The Pragmatics of Tense-

Aspect. Mechanisms of Syntactic Change, , 181-254.

Givón, T. (2001). Syntax: An Introduction, Volume 1. Amsterdam and Philadelphia: John

Benjamins.

Goldberg, A. E. (1995). Constructions : A Construction Grammar Approach to Argument

Structure. Chicago: Chicago : University of Chicago Press.

Gordon, A. (1982). The Development of the Participle in Biblical, Mishnaic and Modern

Hebrew. Afroasiatic Linguistics Los Angeles, Cal., 8(3), 121-179.

Grice, H. P. (1975). Logic and Conversation. Syntax and Semantics, 3, 41-58.

Gries, S. T. (2005). Syntactic Priming: A Corpus-Based Approach. Journal of Psycholinguistic

Research, 34(4), 365-399. 94 Halliday, M., & Matthiessen, C. (2004). An Introduction to Functional Grammar (3rd ed.).

London: Hodder Arnold.

Hasan, R. (1987). The Grammarian’s Dream: Lexis as most Delicate Grammar. In M. A. K.

Halliday, & R. P. Fawcett (Eds.), New developments in systemic linguistics: Theory and

description (pp. 184–211). London: Pinter.

Hatav, G. (1997). The Semantics of Aspect and Modality: Evidence from English and Biblical

Hebrew. Amsterdam: John Benjamins.

Heine, B. (1995). Agent-Oriented Vs. Epistemic Modality: Some Observations on German

Modals. In J. Bybee, & Suzanne Fleischman (Eds.), Modality in grammar and

discourse (pp. 17-53). Amsterdam/Philadelphia: Benjamins.

Heltoft, L. (2010). Paradigmatic Structure and Reanalysis: From NPs to DPs in

Scandinavian. Acta Linguistica Hafniensia, 42(1), 11-25.

Hengeveld, K. (1992). Non-Verbal Predication: Theory, Typology, Diachrony. Berlin: Mouton

de Gruyter.

Hengeveld, K. (2017). A Hierarchical Approach to Grammaticalization. The

grammaticalization of tense, aspect, modality and evidentiality from a functional

perspective (pp. 13-37). Berlin, Boston: De Gruyter Mouton.

Hengeveld, K., & Mackenzie, J. L. (2006). Functional Discourse Grammar. Encyclopedia of

Language and Linguistics, 4, 668-676.

Himmelmann, N. P. (2004). Lexicalization and Grammaticization: Opposite Or

Orthogonal? In W. Bisang, & N. P. Himmelmann (Eds.), What makes grammaticalization?

95 A look from its fringes and its components (Trends in Linguistics: Studies and Monographs

158 ed., pp. 19-40). Berlin: Mouton de Gruyter.

Hopper, P. J. (1979). Aspect and Foregrounding in Discourse. In T. Givón (Ed.), Syntax and

semantics: Discourse and syntax (pp. 213-241). New York: Academic Press.

Hopper, P. J. (1991). On some Principles of Grammaticization. In E. C. Traugott, & B. Heine

(Eds.), Approaches to grammaticalization (1st ed., pp. 17). Amsterdam/Philadelphia:

Benjamins.

Hopper, P. J., & Traugott, E. C. (2003). Grammaticalization (2nd ed.). Cambridge: Cambridge

University Press.

Jespersen, O. (1924). The Philosophy of Grammar. London: G. Allen & Unwin.

Joüon, P. (1923). Grammaire De l’Hébreu Biblique. Rome: Pontifical Biblical Institute.

Kalev, D. (2015). The Grammaticization of the Anterior in Modern Hebrew. Proceedings of

the Haiim B. Rosén Israeli Linguistic Society, 20(1), 51-60.

Kalev, D. (2017). Modern Times: New Aspectual and Modal Constructions in Contemporary

Hebrew [in Hebrew]. (Doctoral Dissertation, Tel Aviv University)

Karttunen, L. (1973). Presuppositions of Compound Sentences. Linguistic Inquiry, 4(2), 169-

193.

Katz, A. (1996). Cyclical Grammaticalization and the Cognitive Link between Pronoun and

Copula. Rice University)

Kiparsky, P. (2012). Grammaticalization as Optimization. Grammatical change: Origins,

nature, outcomes (pp. 15-51). Oxford: Oxford University Press.

96 Kousta, S., Vigliocco, G., Vinson, D. P., Andrews, M., & Del Campo, E. (2011). The

Representation of Abstract Words: Why Emotion Matters. Journal of Experimental

Psychology: General, 140(1), 14-34.

Kranich, S. (2008). Subjective Progressives in Seventeenth and Eighteenth Century English.

In M. Gotti, M. Dossena & R. Dury (Eds.), English historical linguistics 2006: Syntax and

morphology (pp. 241–256). Amsterdam: John Benjamins Publishing.

Kranich, S. (2015). The Impact of Input and Output Domains: Towards a Function-Based

Categorization of Types of Grammaticalization. Language Sciences, 47, 172-187.

Kuryłowicz, J. (1965). The Evolution of Grammatical Categories. Esquisses linguistiques (pp.

38-45). Munich: Wilhelm Fink [1975].

Langacker, R. W. (1987). Foundations of Cognitive Grammar: Theoretical Prerequisites.

Stanford, California: Stanford University Press.

Lehmann, C. (2002). Thoughts on Grammaticalization (2nd ed.). Universität Erfurt: Seminar

für Sprachwissenschaft [1995].

Lohndal, T. (2009). The Copula Cycle. In E. v. Gelderen (Ed.), Cyclical change (pp. 209-242).

Amsterdam: John Benjamins Publishing Company.

Meillet, A. (1912). L'Evolution Des Formes Grammaticales. Bologna: Zanichelli.

Michaelis, L. A. (2017). Constructions are Patterns and so are Fixed Expressions.

Narrog, H. (2005). Modality, Mood, and Change of Modal Meanings: A New

Perspective. Cognitive Linguistics, 16(4), 677-731.

97 Narrog, H. (2017). Relationship of Form and Function in Grammaticalization–the Case of

Modality. The Grammaticalization of Tense, Aspect, Modality and Evidentiality: A Functional

Perspective, 311, 75.

Norde, M. (2012). Lehmann’s Parameters Revisited. Grammaticalization and Language

Change: New Reflections, , 73-109.

Nuyts, J. (1993). Epistemic Modal Adverbs and Adjectives and the Layered Representation

of Conceptual and Linguistic Structure. Linguistics, 31, 933–969.

Palmer, F. R. (2001). Mood and Modality. UK: Cambridge University Press.

Polak, F. (2009). The Verbal System of Tense in the Early Biblical Narrative: The Relative

מערכת זמני הפועל בסיפורת המקראית הקדומה: שיטת הזמנים :Tense System [in Hebrew

.223-244 ,11-12 ,מחקרים בלשון: ספר היובל לאבי הורוביץ .[היחסיים

Polak, F. (2011). In W.Th.van P., JW D. (Eds.), Hebrew Hayah: Etymology, Bleaching, and

Discourse Structure. Leiden/Boston: Brill.

Rainey, A. (1988). The Verbal Prefix Inflection in Ancient Hebrew in Light of the Canaanite

נטיית הפועל בעל התחיליות בעברית העתיקה לאור הכנענית של :of Amarna Letters [in Hebrew

.Be'Er Sheva, 3, 149-162 .[מכתבי אל עמרנה

Reichenbach, H. (1947). Elements of Symbolic Logic. New York: Free Press.

Roberts, J. R.The category ‘Irrealis’ in papuan medial verbs. Symposium on Mood and

Modality, UNM, Albuquerque.

Rosén, H. B. (1985). Qavim Letoldot Ma’arexet Hapo’al Be’ivrit [Delineating the History of

the Hebrew Verbal System].Reprinted in East and West: Selected Writings in Linguistics by

Haiim B. Rosén III, 1994, , 410-417. 98 Rosén, H. (1977). Contemporary Hebrew. The Hague: Mouton & Co. B. V. Publishers.

Sasse, H. (2002). Recent Activity in the Theory of Aspect: Accomplishments, Achievements,

Or just Non- Progressive State? Linguistic Typology, 6(2), 199-271.

Saussure, F. (1916). In Bally C., Sechehaye A. (Eds.), Course in General Linguistics. New York,

USA: Philosophical Library [1959].

Searle, J. R. (1980). The Background of Meaning. In J. R. Searle, F. Kiefer & M. Bierwisch

(Eds.), Speech act theory and pragmatics (synthese language library 10) (pp. 221-232).

Boston: Reidel.

Smith, C. S. (1991). The Parameter of Aspect. Dordrecht: Kluwer Academic Publishers.

Smith, C. S. (1997). The Parameter of Aspect. Dordrecht: Kluwer Academic Publishers.

Strong, J. (1890). Strong's Exhaustive Concordance of the Bible. New York: Abingdon-

Cokesbury Press.

Suh, K. (1992). A Discourse Analysis of the English Tense-Aspect-Modality System. UCLA)

Traugott, E., & Dasher, R. (2002). Regularity in Semantic Change. Australia: Cambridge

University Press.

Traugott, E. C. (1978). On the Expression of Spatio-Temporal Relations in Language. In J.

H. Greenberg, C. A. Ferguson & E. A. Moravcsik (Eds.), Universals of human

language (pp. 369-400). Stanford: Stanford University Press.

Traugott, E. C. (1989). On the Rise of Epistemic Meanings in English: An Example of

Subjectification in Semantic Change. Language, 65, 31-55.

99 Traugott, E. C. (2002). From Etymology to Historical Pragmatics. In D. Minkova, & R.

Stockwell (Eds.), Studies in the history of the english language: A millennial perspective (pp.

19-49). Mouton de Gruyter: Berlin.

Traugott, E. C. (2010). (Inter) Subjectivity and (Inter) Subjectification: A Reassessment. In

K. Davidse, L. Vandelanotte & H. Cuyckens (Eds.), Subjectification, intersubjectification

and grammaticalization (pp. 29-74). Berlin: Mouton de Gruyter.

Traugott, E. C., & Trousdale, G. (2010). Gradience, Gradualness and Grammaticalization.

Amsterdam, The Netherlands; Philadelphia, Pa.: J. Benjamins Pub. Co.

Verkuyl, H. (1972). On the Compositional Nature of the Aspects. Foundations of Language

Supplementary Series, 15

Waltereit, R. (2012). On the Origins of Grammaticalization and Other Types of Language

Change in Discourse Strategies. In K. Davidse, T. Breban, L. Brems & T. Mortelmans

(Eds.), Grammaticalization and language change: New reflections (pp. 51-72). Amsterdam:

John Benjamins.

Werner, F. (1980). The Status of Word Formation in the Teaching of the Hebrew Language

(in Hebrew). The Fourth Hebrew European Scientific Conference on Hebrew Language and

Culture (Warsaw 1980),

Yitzhaki, D. (2004). The Semantics of Lexical Aspect in Modern Hebrew. Tel Aviv University)

Zeno Vendler. (1957). Verbs and Times. The Philosophical Review, 66(2), 143-160.

Ziegeler, D. (2000). The Role of Quantity Implicatures in the Grammaticalisation of

Would. Language Sciences, 22(1), 27-61.

100 Zwaan, R. A. (2004). The Immersed Experiencer: Toward an Embodied Theory of

Language Comprehension. Psychology of Learning and Motivation, 44, 35-62.

Zwaan, R. A. (2014). Embodiment and Language Comprehension: Reframing the

Discussion. Trends in Cognitive Sciences, 18(5), 229-234.

101 אוניברסיטת תל-אביב הפקולטה למדעי הרוח ע"ש לסטר וסאלי אנטין החוג לבלשנות

מהשאה לניבוי גרמטיקליזציה של הפועל ״היה״ בעברית

חיבור זה הוגש כעבודת גמר לקראת התואר "מוסמך אוניברסיטה" - M.A. באוניברסיטת ת"א

על ידי אליאור אלקיים

העבודה הוכנה בהדרכת: פרופ׳ מירה אריאל

אוגוסט 2018

תקציר

בעברית מודרנית, ״היה״ הוא פועל המשתתף בכמה קונסטרוקציות המביעות משמעויות שונות. מבחינה דקדוקית, ״היה״ מתפקד בתור פועל קיומי )אקזיסטנציאלי( )לדוג׳: היה פה חתול( או אוגד )לדוג׳: הוא היה מורה(. עבודת גמר זאת תתרכז בשלוש קונסטרוקציות: 1. ההעבר ההרגלי )היה + בינוני( הוא היה הולך לשם 2. התנאי ההיפותטי )אם היה + בינוני, היה + בינוני( אם היית אומר לי, הייתי מביא עוד 3. הניבוי ההיפותטי )היה + בינוני( הייתי מעדיף ללכת לקולנוע השאלה שעולה לגבי הקונסטרוקציות האלה היא – איזה אלמנט ב״היה״ מאפשר את הופעתן של מושגים כמו הרגליות והיפותטיות? מה מחבר בין שני המושגים האלה? ומה הביטוי ״היה + בינוני״ תורם לכל אחד מהקונסטרוקציות האלה? ישנם שני דרכים לענות על השאלות האלה – באופן סינכרוני ובאופן דיאכרוני. מטרת הגישה הסינכרונית היא להציג את האינטראקציה בין הקונסטרוקציות האלה. הרעיון הוא שבאמצעות חשיפת התכונות הצורניות, הסמנטיות והשיחיות של כל אחד מקונסטרוקציות אלה, נוכל לראות את ההדרגתיות ביניהן ) & Hopper Traugott 2003(. כמו כן, דבר זה יאפשר לנו להעלות השערות לגבי תהליך הגרמטיקליזציה שלהן. מטרת הגישה הדיאכרונית היא לחשוף את ההקשרים המגשרים בין כל אחת מהקונסטרוקציות הללו. באמצעות שיטה זו, נוכל לחשוף מה מחבר בין קונסטרוקציות ה״היה״ מבחינה צורנית, סמנטית ושיחית. עזות זו תשמש להערכת ההיפותזה הסינכרונית המתייחסת למסלול הגרמטיקליזציה של הקונסטרוקציות הללו. ראשית, אראה שהקונסטרוקציה ״היה + בינוני״ צמחה בעברית התנכית עם המשמעות של העבר האימפרפקטיבי )פרק 5(. התפקיד השיחי של הקונסטרוקציה הזאת היה להציג אפיזודה )episode( חדשה לשיח )כלומר נושא, זמן ומקום חדשים(. בעברית משנאית, הקונסטרוקציה הזאת אימצה תפקיד חדש – להציג אירוע המשתתף באפיזודה גדולה יותר שהוזכרה בשלב קודם בשיח. בעברית מודרנית, רק המשמעות של ההעבר הרגלי שרדה. מחקר מבוסס קורפוס חשף שהתפקיד השיחי של ההעבר ההרגלי בעברית מודרנית הוא אירועי, בדומה לעברית משנאית. המחקר מספק עדות חדשה התומכת בהיפותזה של רוזן )1985( הטוענת כי ההעבר ההרגלי בעברית מודרנית אומץ מהעברית המשנאית ולא מהעברית התנכית )לעומת דורון 2006(.

בנוסף, אני אראה את המקורות התנכיים והמשנאיים של התנאי ההיפותטי ואציג את ההקשרים המקשרים בין העבר האימפרפקטיבי והתנאי ההיפותטי )פרק 6(. לעומת ההעבר ההרגלי, מסלול הגרמטיקליזציה של התנאי ההיפותטי מאופיין בשינויים יותר צורניים מאשר סמנטיים. השלב הדקדוקי הבא הוא ההתפתחות של הניבוי ההיפותטי )פרק 7(. אני טוען שבעברית מודרנית הניבוי ההיפותטי הופיע בפסוקיות עצמאיות מתוך הפסוקית העוקבת של התנאי ההיפותטי. על מנת לתמוך בטענתי, אני אנתח דוגמאות הלקוחות מקורפוס של עברית כתובה עממית ואראה שהניבוי ההיפותטי הופיע בפסוקיות עצמאיות בהקשרים סובייקטיבים שבהם הקונסטרוקציה ביטאה את העמדה של הדובר לגבי הקיום של האירוע ההיפותטי. בנוסף, אני אדון ואנתח את התבניות של מסלול הגרמטיקליזציה של קונסטרוקציות ה״היה״ )פרק 8(. בפרק 9 אני אסכם איך הדובר מבדיל בין שתי המשמעויות הפוליסמיות של הקונסטרוקציה ״היה + בינוני״ )העבר ההרגלי והתנאי ההיפותטי(. פרק 10 מסכם את התזה.

תוכן עניינים

1. מבוא...... 1 2. שאלות מחקר...... 3 3. דיאכרוניה והיסטוריה: שלושת השלבים של עברית...... 4 4. מסגרת תיאורטית...... 5 4.1. תפקידים שיחיים...... 8 4.2. מערכת הפועל של עברית...... 12 5. העבר ההרגלי...... 14 5.1. עברית תנכית...... 15 5.2. עברית משנאית...... 25 5.3. עברית מודרנית...... 31 5.3.1. מבוא לעבר ההרגלי בעברית...... 33 5.3.2. שאלות מחקר...... 40 5.3.3. הקורפוס...... 41 5.3.4. מתודולוגיה...... 42 5.3.5. תוצאות ודיון...... 43 5.3.6. התפקידים השיחיים של קונסטרוקציות העבר ההרגלי...... 49 5.3.7. מסקנות...... 54 5.4. סיכום...... 54 6. התנאי ההיפוטתי...... 55 6.1. עברית תנכית...... 57 6.2. עברית משנאית...... 59 6.3. עברית מודרנית...... 66 6.4. סיכום...... 69 7. הניבוי ההיפוטתי...... 70 7.1. היפותזה וניבוי...... 73 7.2. מתודולוגיה...... 74 7.3. תוצאות...... 75 7.4. סיכום...... 78

8. דיון...... 79 9. פוליסמיה קונסטרוקציונלית...... 89 10. מסקנות...... 90 ביבליוגרפיה...... 92