<<

1

Syllabic and the Speaking Culture of Literate, Professional Israeli Hebrew (IH) – Speaking Adults

Thesis submitted in partial fulfillment of the requirements for the degree of “DOCTOR OF PHILOSOPHY” by

Daphna Cohen Ben Shaul

Submitted to the Senate of Ben-Gurion University of the Negev

April 2013

Beer-Sheva

2

Syllabic Orthography and the Speaking Culture of Literate, Professional Israeli Hebrew (IH) – Speaking Adults

Thesis submitted in partial fulfillment of the requirements for the degree of “DOCTOR OF PHILOSOPHY” by

Daphna Cohen Ben Shaul

Submitted to the Senate of Ben-Gurion University

of the Negev

Approved by the advisor: Yishai Tobin, Ph.D.

Approved by the Dean of the Kreitman School of Advanced Graduate Studies______

April 2013

Beer-Sheva

3

This work was carried out under the supervision of

Yishai Tobin, Ph.D.

Professor of Linguistics

In the Department of Foreign Literatures and Linguistics

Faculty of Humanities and Social Sciences

4

Research-Student's Affidavit when Submitting the Doctoral Thesis for Judgment

I, Daphna Cohen Ben Shaul, whose signature appears below, hereby declare that

(Please mark the appropriate statements):

X_ I have written this Thesis by myself, except for the help and guidance offered by my Thesis Advisors.

X_ The scientific materials included in this Thesis are products of my own research, culled from the period during which I was a research student.

___ This Thesis incorporates research materials produced in cooperation with others, excluding the technical help commonly received during experimental work. Therefore, I am attaching another affidavit stating the contributions made by myself and the other participants in this research, which has been approved by them and submitted with their approval.

Date: April 2013 Daphna Cohen Ben Shaul

5

To professor Yishai Tobin,

I thank you deeply for your knowledgeable guidance, support, encouragement, wide open heart and smile, and your freedom of thought which provided the solid ground for my enjoyable and fruitful work.

I will cherish all these gifts forever!

To my beloved husband Nitzan:

I thank you with all my heart and love for your constant encouragement to undertake research, your intellectual and emotional support, and your proficient editing in which you invested your keen thought, time and heart.

6

Table of Contents

Abbreviations 9

Tables 10

Figures 12

Abstract 13

I. Introduction 17

I.1 The Israeli Hebrew (IH) Revival 19

I.2. The Evolution of IH Orthography 20

I.2.1 From IH Institutionalization to Spoken Practice 21

I.2.2 From Spoken Practice to Orthographic Change 26

I.3 Mapping IH Orthography 31

I.3.1 Hebrew’s Syllabic Orthography 31

I.4 Orthographic Effects on IH Speech Production 34

I.4.1 Orthographic Influences on Speech Perception 34

I.4.2 Orthographic Influences on Speech Production Processes 37

I.4.3 Orthographic Influences on Diction 38

I.5 The Research Question 40

I.6 The Hypothesis 40

I.7 The Study Objectives 40

II Methodology 41

II.1 The Pilot Experiment 43

II.1.2 Material 43

II.1.3 Experiment Procedure 47 7

II.1.4 Tools 47

II.1.5 Measurements 47

II.1.6 Subjects 48

II.1.7 Results 49

II.2 Discussion: 58

II.2.1 Vowel Letters in IH 58

II.2.2 Pseudo Words 60

II.2.3 Vowel Letters and Additions 60

II.2.4 Multi-Syllable Words and Sentence Structure 61

ALEF – Ambiguity Issues 62 /א/ II.2.5 The

ALEF) 64) /א/ AYIN) and) /ע/ II.2.6 The Homophones

III. The Main Test 66

III.1 Test Material 66

III.1.1 Isolated Words 66

III.1.2 Sentences 67

III.1.3 Free Speech 68

III.2 Subjects 68

III.3 Experiment Procedure 69

III.4 Tools 70

III.5 Measurements 70

III.6 Results 71

III.6.1 Comparison Between IH Words in IHS and HEBRISH Scripts 71

III.6.2 Comparison Between IH Words in IHVL and HEBRISH Scripts 72

III.6.3 Comparison of IHS Words in 3 Different Scripts (IHS, IHVL, HEBRISH)74 8

III.6.4 Comparison of IH Words containing 1 VL in 3 Different Scripts (IH, IHVL, HEBRISH) 76 ALEF/AYIN ) 78) א/ע III.6.5 Words Beginning with

III.6.6 Comparison of Sentences in IH, IHD and HEBRISH 83

III.6.7 (IHVL) Full sentences compared to IHD and HEBRISH Orthographies 84 83

III.6.8 Free Speech Analysis 86

IV. Discussion 88

IV.1 Orthographic Effects on Vowel Length Variation 88

IV.2 Vowel Length Variation within Different IH Scripts 89

IV.3 Effects of Internalized Visual Templates on Speech Production 90

IV.4 The Interplay Between Orthographic and Prosodic Effects 91

IV.4.1 Word Length and Syllabic Stress Factors / Hesitation as a Factor

in IHVL vs. HEBRISH Scripts 91

IV.4.2 Hesitation Factor in Established and non Established IHVL Scripts 92

IV.4.3 The Stress-in-Words Dynamics Factor 93

IV.4.4 Orthographic Factors Influencing ALEF/AYIN 93

IV.4.5 Sentences in Script Variations 95

V Conclusions 97

Bibliography 100

Appendix I: Pilot Test Materials 108

Appendix II: Main Test Materials 115

Hebrew Abstract 129

9

Abbreviations

VL - Vowel

IH - Israeli Hebrew

IHS - Israeli Hebrew Syllabic orthography

IHD - Israeli Hebrew orthography with marks

IHVL - Israeli Hebrew orthography with Vowel Letters

LPVU - Literate Professional Voice Users

CVC - Consonant-Vowel-Consonant syllable

10

Tables

Table 1: Tiberian NIQUD according to IH vowel 23

Table 2: Sepharadi Hebrew pronunciation for 10 phonemes and their respective 26

Table 3: Sephardic Hebrew pronunciation for 10 phonemes and their respective graphemes compared to their pronunciation in present day IH 29

Table 4: Consonant/vowel graphemes 33 Table 5: Test’s demonstration in order of presentation 46 Table 6: Descriptive statistics as well as comparison of vowel length in IH and HEBRISH scripts 52 Table 7: Descriptive statistics and comparison of vowel length in IHS and transparent orthographies (IHVL /HEBRISH) 52

ALEF, HEI, VAV, YOD) 63) /א,ה,ו,י/ Table 8: The dual function of AYIN, ALEF, HEI, VAV, YOD) 65) /ע,א,ה,ו,י/ Table 9: The dual function of

Table 10: Subjects of the main experiment 69 Table 11: Comparison of words in IHS and in HEBRISH scripts 72 Table 12: Comparison of words in IHVL and in HEBRISH Scripts 73 Table 13: Comparison of IHS words in IHVL and in HEBRISH scripts 75 Table 14: Comparison of IHS words with 1 VL in IHVL (reaching full phonemic presentation) and HEBRISH scripts 77 Table 15: Words beginning with ALEF/AYIN in IH, compared to the HEBRISH script 78 Table 16: Interaction effect measured for mono-syllable words initiating with ALEF/AYIN in two scripts, compared to their minimal-pairs scripts 80 Table 17: Interaction effect measured for two-syllable words initiating with ALEF/AYIN in two scripts, compared to their minimal-pairs scripts 80 Table18: Interaction effect measured for the vowel length difference between words initiating with ALEF/AYIN and their HEBRISH script, compared to the vowel length difference measured between their minimal pairs and their matched HEBRISH script 81 11

Table 19: Interaction effect measured for the vowel length difference between words initiating with ALEF/AYIN followed by a vowel letter and their HEBRISH script, compared to the vowel length difference measured between their minimal pairs and their matched HEBRISH script 82 Table 20: Descriptive statistics of 8 sentences presented in three different scripts 83 Table 21: Pair wise comparative analysis of 8 sentences in three different scripts 84 Table 22: Descriptive statistics of 2 sentences containing IHVL words (i.e. full script) presented in 3 different orthographic scripts 85 Table 23: Pair wise comparative analysis of 2 sentences containing IHVL words (i.e. full script) presented in 3 different orthographic scripts 85 Table 24: CVC stressed syllables in free speech 86

12

Figures

Figure 1: Vowel length comparison of 10 IH monosyllable words sentences in IH and in HEBRISH scripts for subject 1 (actress) 50 Figure 2: Vowel length comparison of 8 IH monosyllable pseudo words sentences in IH and in HEBRISH scripts for subject 1 (actress) 50 Figure 3: Vowel length comparison of 4 IH multi-syllable words sentences in established IH script and in IHVL script (i.e. full script) for subject 1 (actress) 51 Figure 4: Mean vowel length for sentences lacking VL presentation and sentences with overt VL presentation, measured for each subject, as well as overall mean vowel length for all 4 subjects 53 Figure 5: 2 spectrograms of a 2 monosyllable pseudo words sentence (subject 3) 54

Figure 6: 2 spectrograms of a 3 monosyllable pseudo words sentence (subject 4) 56

Figure 7: 2 spectrograms of a 4 IH multi syllable words sentence (subject 1) 57

Figure 8: CVC stressed syllables in free speech 87

Figure 9: Words in IHS and IHVL scripts compared to their HEBRISH script 89

13

Abstract

This study is concerned with how the orthographic characteristics of a given language affect speech. It focuses upon how Israeli Hebrew (IH) orthography affects IH speech. In IH this influence might be particularly strong given that it was revived as a spoken language in the late 19th century out of a written text tradition, having lost most of its currency as a spoken language after the Roman exile of Jews from the land of Israel in the first century A.D.

This relatively recent approach to the relation of orthography to speech reverses the traditional trend which was primarily concerned with how orthography evolves from speech. There is a growing interest in this reversed focus, stemming from the expansion of literacy and more so following the recent expansive use of electronic platforms for communication such as the internet or the cellular phone. This has resulted in a growing body of research concerned with the bi- directional relation between orthography and speech. I hereby maintain that the consonant syllabic nature of Hebrew orthography influences IH speech.

Throughout history Hebrew has had a "deep orthography" stemming from its consonant based morphologic root system, where letters represent consonants with an overall lack of overt vowel signs (except for four consonants that occasionally function linearly as vowel letters - (ALEF, HEI, ,Hebrew’s consonantal script, where letters often denote a syllable unit .(/א/ /ה/ /ו/ /י/ VAV, YOD may also be defined as a consonantal- based . Developments and modifications of written Hebrew over time reflect the tension between spoken and written Hebrew that mostly stems from this lack of vowel presentation. This has engendered methods to represent these “missing vowels” in order to fill the gap between spoken and written forms of the language.

The most significant development occurred in the middle Ages when spoken Hebrew risked disappearance, leading to the creation of the Tiberian NIKUD – a notation based upon the additional placement of diacritical points vertically to the consonant letters in order to represent vowels. In such manner, the NIKUD was used to preserve earlier features of spoken Hebrew with a more complex orthographic system and has become the dominant system for vocalizing all Jewish varieties of Hebrew to this day. 14

In IH however, this complex post-biblical system of diacritical marks, present at the beginnings of IH’s revival and easing Israeli Jews’ vocalization of the language, was later discarded (save for its specialized use in religious texts, children's book and poetry), with IH orthography reverting to the ancient Hebrew script characterized by its overall vowel-bare consonant-based syllabary. This was probably due to the tendency of language users to fill in the vowels as means towards “minimal effort to achieve maximal communication” (Tobin 1990a, 1990b,1997), so that once the spoken language was revived, there was no need for a complex and cumbersome system of vertically placed vowel diacritics on a more simple primarily consonantal syllabic orthography.

I argue in this study that IH’s historical and traditional consonant based syllabic orthography, its rich morphological root system along with an overall lack of vowel visualization affects IH speech production. This is found, mostly in respect of a spoken emphasis on consonants and an attendant diminishing of the length of vowel expression within an utterance. This seems to relate in particular to literate Israelis, for whom IH is a native and all engulfing language (Ornan 2009), the primary language used for all their written and spoken needs. I presume this would not be the case for Jewish communities for whom Hebrew is not the primary language. Support for this argument can be found in psycholinguistic research from the past two decades showing, that the mind language-templates of literate Israelis prime consonants over vowel presentations (Katz & Frost 1992, Frost, 1995, 1998, 2006).

In light of these assumptions and findings, I conducted a pilot study based on the hypothesis that IH syllabic orthography (IHS) has effects on speech. I focused my research upon Literate Professional Voice Users (LPVU) given that the written text comprises the principal stimulus in their professional speech production, thus distinguishing this group as homogeneous in respect of this study’s dependent variable, namely the primarily consonantal syllabic orthography’s effect on speech production.

I hypothesized that the primarily consonant-based syllabic nature of IH orthography engenders -- at least for LPVU -- a consonant-emphasized Hebrew diction, diminishing the length of vowel expression within an utterance. Following the encouraging primary results obtained in the pilot study, I conducted a more comprehensive study where I recorded 8 LPVU Israeli native IH 15

speaking adults (teachers, singers, actors) aged 20-55, who were asked to read different orthographic scripts representing the same words and sentences: the common IH syllabic (IHS) orthography with no vowel visualization; IH with vertical diacritical points visualizing vowels (IHD) and reaching full phonemic presentation; IH with linearly added vowel letters approximating full phonemic presentation (IHVL); and HEBRISH script (IH presented in Latin letters enabling full phonemic presentation of all consonants and vowels). In order to reduce the impact of the reading screening process and enhance the effect of stored orthographic templates to approximate free speech, I chose words that are highly frequent in daily speech and whose reading is rapid and automatic. Spectrogram analysis of the recorded material, based upon a comparison of the measured vowel temporal portion within an utterance produced across the different orthographic scripts, shows that there is a significant difference in vowel/consonant ratio between an utterance consisting of visualized vowel letters versus the same utterance having no vowel visualization. Moreover, a comparison of Hebrew written words to their HEBRISH presentation, shows that the more vowel signs are present in the Hebrew orthography of the same specific words, the smaller the difference in vowel/consonant temporal ratio. In a follow-up experiment, based upon recent studies examining latency responses that found the same orthographic effects when reading aloud and in free speech (e.g. Perre et al 2009, Desroches et al 2010), I tested whether the free speech of literate subjects also exhibits the same effects of orthography on the quality of their free speech as those found in my controlled read-aloud experiment. I asked the 8 LPVU subjects that participated in my previous experiment to freely respond to the same questions. I then compared their articulation of IH CVC (Consonant-Vowel- Consonant) stressed syllables written in established IH orthography with or without vowel representation. The findings replicated those of the controlled reading aloud task, namely showing significantly longer vowel production in syllables with vowel representation than in those lacking vowel representation. These findings, taken together, corroborate my hypothesis that Hebrew’s primarily consonant based syllabic orthography, at least in respect of orthographic vowel visualization or lack thereof, affects the diction of IH speakers.

The findings of this study support the findings of psycholinguists in respect to the effects of consonant based Hebrew orthographic mental templates upon speech and reading perception (Frost 1995, 2006). Furthermore, my study expands their findings to include not only speech perception but also speech production. The results obtained indicate that in the automated 16

reading (i.e. template guided) of the established Israeli Hebrew Syllabic script (IHS) with no vowel visualization, the production of speech by highly literate native speakers presents, in terms of the variability of duration in their diction, a vowel-shortened and consonant-emphasized spoken Hebrew.

The additional research methodology used to corroborate the hypothesis, namely its emphasis upon consonant/vowel temporal ratio in speech production, opens inroads into the study of the influence of orthography on the qualitative aspects of diction such as variability of duration, intensity and pitch. Furthermore, the ways, in which the gap between spoken and written forms of IH has been analyzed, can help examine the effects of orthography upon speech in other . It would be particularly interesting to examine other syllabic orthographic languages such as those based on logographic syllabary (e.g. Japanese or Korean). Moreover, this approach could also enhance new perspectives on rehabilitation programs in respect of a range of clinical speech pathologies, particularly voice pathologies, various learning disabilities, diction, and fluency, as well as on the development of speech and voice skills.

Finally, this study may be also beneficial for studying the reactive influence of spoken IH on IH's as evidenced in the adding of vowel letters when writing in electronic platforms for communication such as the internet or the cellular phone. This reactive trend may stem from IH users' wish to disambiguate IH orthography.

Key words: Israeli Hebrew (IH), Syllabic Orthography, consonantal scrip Literate Professional Voice Users (LPVU),Vowel length, Vowel Duration.

17

I. Introduction

The acquiring primacy of spoken over written language has traditionally led researchers to assume that the acquisition of reading/ abilities capitalizes on the cognitive structures that are specialized for spoken language (e.g. Liberman, 1992). Furthermore, observations based on this assumption have converged into showing that children are highly dependent on during the acquisition of literacy (e.g. Wagner & Torgesen 1987; Gelder & Vroomen 1991; Perfetti & Marron 1998; Palmer 2000; Antony & Francis 2005; Kovelman et al. 2011).

This widespread assumption whereby speech influences orthography1 rather than the opposite has been challenged in the past three decades where research is providing growing evidence suggesting a bi-directional influence. The notion that orthographic information has an impact on spoken word processing has been demonstrated using a range of different auditory tasks in different cultures and writing systems, relating orthography to the perception and production processing of the oral language (e.g. Seidenberg and Tanenhaus 1979; Ziegler and Ferrand 1998; Damian et al 2003; Bolger et al 2008; Brewer 2008; Pattamadilok et al, 2008; Taft 2011).

A major precedent to this latter approach can already be found in de Saussure’s writings from over 100 years ago, where he claimed that “By imposing itself upon the masses, spelling influences and modifies language. This happens only in highly literate languages where written texts play an important role. Then visual images lead to wrong pronunciations; such mistakes are really pathological.” (de Saussure 1966: 31). He then went on to describe what he called "pronunciation mistakes" such as the tendency of some Parisians at his time to pronounce the silent /p/ in sept femmes, (similar to the pronunciation of the /t/ in the word often in English) blaming such phonemic deformities on external influences. As a purist, de Saussure suggested

1 Orthography is a comprising a structure along with the method and rules of its usage. Graphemes are the smallest units or minimal signifiers in the orthographic system of a specific language. In alphabetical orthographies as Hebrew or English, graphemes refer to the letters of the .

18

that these orthographic "monstrosities" be grouped in a special compartment as “teratological cases” (de Saussure, 1966: 31-2).

While de Saussure's purist disdain of orthography found its effect on speech to be deplorable, he also admitted thereby that there is a bi-directional relationship between speech and writing. However, it took over seventy years after de Saussure for this line of research to take hold. This renewed interest was probably due to the growing role played by literacy in the daily life of the majority of the Western world's population, with education becoming the basic routine for all children. Furthermore, the growing interest in Learning Disabilities as well as the recent extensive use of electronic communication has increased research into the influence of orthography on speech.

The present study shares the notion that there is a bi-directional influence between orthography and speech, and focuses upon how Israeli Hebrew's specific orthographic depth2 influences the speech production of native literate Israelis.

The process of Hebrew revival in Israel as a modernized updated living and developing language is unique because the revival of IH was mostly based upon the continuous history of Hebrew orthography rather than upon spoken Hebrew, largely extinct at the time of the language’s revival (Azar 2009). Before turning to present IH orthographic depth and how it affects speech production, it is instrumental to survey the process of the IH revival since it exemplifies the bi- directional relation and tension between spoken IH and IH’s deep orthography.

2 Writing and speaking are distinct systems within a given language. A major outcome of the various relations between spoken language and its written representation is the degree of correspondence between the phonetic unit and the written unit. This correspondence is described in terms of shallowness or depth of orthography. In alphabetic orthographies, when a written language is characterized by transparency, i.e. a one-to-one approximate and consistent relation between the and its grapheme, it is defined as having a shallow orthography (as in Spanish or Italian). Conversely, the more the correspondences between spelling and pronunciation are opaque, in respect of completeness or consistency (as in Hebrew or English), the deeper the orthography of the language (e.g. Katz & Frost 1992).

19

I.1The Israeli Hebrew (IH) Revival

There is a debate in dating the revival of Hebrew (Ben Haim 1963). The revival of modern Hebrew at the turn of the century, and the special controversial role played by Eliezer Ben Yehuda, referred to as the man who was responsible for the revival, have been the subject of much research (Bar Adon, 1975; Fellman, 1973; Harshav, 1993). It has been claimed that Hebrew was 'revived' several times as a literary language, first during medieval times, when educated Jewish philosophers and poets turned to written Hebrew, but more significantly from the late 18th century to the mid-19th century by the “Haskalah” (Enlightenment) movement,3 who turned to written Hebrew for literary and narrative uses. The major result in respect of the Hebrew language of these two periods’ literary work was a lexical modernization of Hebrew (Ornan 2009). Lexical modernization however, resulting in an enlarged lexicon, is a phenomenon that characterizes any diglossic4 community and does not indicate a living language. Ornan claims that Israeli Hebrew’s revival should be considered from when it became an encompassing and engulfing language that is the primary language serving all of the written and spoken needs of the community. In this respect, Ornan argues that the revival of the Hebrew language can be said to have initiated in the late 19th century, mostly due to the efforts of Eliezer Ben-Yehuda, who joined the Jewish national movement and immigrated to Ottoman ruled Palestine in 1881. Ben-Yehuda, motivated by emergent Zionist ideals of Jewish life renovation and the rejection of the Diaspora

,Enlightenment) refers to a Jewish movement in the 18-19th centuries, originating in Germany/השכלה) The Haskala 3 and aimed at encouraging Jews to integrate into European society by adapting the values of enlightenment shared by many Europeans at the time.

4 Diglossia refers to the coexistence of two varieties of the same language for a community of speakers. Often, one form is literary or a prestigious and the other is a common dialect spoken by most of the community. Sociolinguists may also use the term Diglossia to denote bilingualism, the speaking of two or more languages by the members of the same community (e.g. Shiffman 1996).

20

"shtetl”5 lifestyle, set out to develop tools for turning literary and liturgical Hebrew into a vibrant living language, to be exploited for all the daily spoken and written needs of the language community. His further organizational efforts, including the establishment of the first “Israeli Hebrew Committee” together with David Yellin in 1890 in order to institutionally regulate the language,6 his involvement in the establishment of Hebrew-speaking schools, and his writing of Hebrew language textbooks, propelled an intense effort to turn Hebrew into the vernacular in the growing Jewish settlements in Palestine. From Jewish children that started their education in Palestine at Hebrew-speaking schools, to the victory of Hebrew in the “language war”,7 this period can be seen as the turning point of a new era, in which Hebrew was revived as an all engulfing language (Ornan 2009).

I.2.The Evolution of IH Orthography

The process of Hebrew revival in Israel, mostly reconstructed from Hebrew's written tradition, was initially artificial to the speaking community it addressed, consisting of an institutionalized orthographic reconstruction of Hebrew and of directives for a standardized pronunciation and intonation. The historical, political, sociological and religious circumstances that brought Jewish communities from around the world to mandatory Palestine in the end of the 19th century and the beginning of the 20th century, particularly their desire for a shared national home and identity, helped the artificial and institutionalized reconstruction of the Hebrew language to become

5 Shtetl (a Yiddish language word) refers to a small town with a large Jewish community in Eastern and Central Europe.

6 This regulatory body continues its activities to this day though it changed its name to “The Hebrew Language Academy” in 1963.

7 A widespread campaign led by youngsters and teachers in 1913-1914, demanding of all Jews in Palestine to speak Hebrew and to force the use of the Hebrew language in all educational institutes. It was grounded on the conviction that a shared Hebrew language will form the cornerstone of the emergent Israeli national identity. The campaign succeeded: Jewish schools turned Hebrew into their official language, also officially acknowledged by the Othman authorities ruling Palestine at the time. 21

established, propelling its vocalization. However, the spoken IH that eventually resulted affected in turn IH orthography. The dynamics of this process suggest the bi-directional influence between IH orthography and speech in respect to IH’s deep orthography, and form the background for my original hypothesis and the empirical experiments of the effects of present day IH orthography on IH speech production.

I.2.1From IH Institutionalization to Spoken Practice

Eliezer Ben-Yehuda’s IH revival was based on the ancient Hebrew alphabet dating back to the end of the second millennium B.C. Like IH, this alphabet was written from right to left, and included 22 letters representing mainly consonant phonemes with an overall lack of overt vowel signs (e.g. Goldwasser 2010).8 In his attempt at resuscitating a standardized Hebrew vocalization, pronunciation and accent from this alphabet’s consonant based syllabic orthography, Ben-Yehuda turned to the continuous historical development of Hebrew orthography.

As previously mentioned Hebrew primarily lost currency as a spoken language but persevered and developed in its written form. This dichotomy between spoken and written Hebrew can be said to have begun after the Roman Empire exiled most of the Jewish population following the Bar Kochba revolt (132-135 A.C) (Rabin 1972). This first exile and the second one carried out by the Roman Empire in 70 A.C, created an ongoing diglossic language characterizing most Jewish communities dispersed around the world. This was due to the Jewish communities’ need to adapt to the different cultures and languages of the societies in which they lived. Hence, in the different Jewish communities’ diglossic language to this day, excluding Israeli Jews, the primary language used is the local language, leading to a variety of intonations and pronunciations of an almost extinct spoken Hebrew, along with the perseverance of written Hebrew, used mostly for religious

8 Goldwasser argues that this first Ancient Hebrew Alphabet, Ancient Hebrew-Phoenician, was derived from the so- called Phoenician or Old Semitic letters, to which almost all systems of letters now in use, even the Roman, can be traced. He also claims that the Hebrew Alphabet was the first Alphabet in the History of writing and can be probably attributed to the Canaanites working for the Egyptian Kingdom in Serabit-Sinai (Goldwasser 2010). 22

purposes. In fact, Hebrew as a spoken language has vanished from non-Israeli Jews’ daily life (Rabin 1972: ch.6). Hebrew writing however, persevered and proliferated and it was this continuously evolving written tradition rather than its spoken expression that was particularly functional for Ben-Yehuda in his project to revive Hebrew as a living language. The proliferation of written Hebrew is already evidenced in the evolution of various regional literary Hebrew during medieval times. Among these writing dialects, The Tiberian Hebrew or Masoretic Hebrew, a local dialect of Tiberias in the Galilee, stands out because it developed the Tiberian “NIKUD” system approximately in the seventh century A.C. The Tiberian vocalization system introduced a revised Hebrew orthographic script based on the vertical addition of vowel diacritics to the primarily syllabic alphabet’s consonants. Devised by the Masoretes to add to the consonantal Masoretic text of the Hebrew Bible, this system was widely accepted and became the dominant system for vocalizing all Jewish varieties of Hebrew to the present day (Gesenius 1909; Rosen 1967). The fact that this written dialect was adopted by Jewish communities worldwide, testifies to the ongoing need for filling the gap between written Hebrew and its pronunciation. This, given that the consonant based syllabic orthography could not preserve the crucial features of a spoken Hebrew that was hardly used or practiced, but which was particularly needed for public praying, and occasionally needed for vocalizing other types of Hebrew texts in correspondences and personal meetings between Jews from different communities (Rabin 1972). Another important development in written Hebrew, which helped Ben-Yehuda revive the language, occurred during the “Golden age” of Jewish culture in Spain in the (8th to 12th centuries). During this period, Jewish intellectuals and poets turned to written Hebrew to express scientific and philosophical concepts taken from Classical Greek and Medieval Arabic in order to communicate these to other well educated Jews in all parts of the world, through a mutually intelligible language or lingua franca (Rabin 1972). This motivated Medieval Hebrew to borrow terminology from these other languages or create equivalent terms, and to adapt these languages’ grammar to the tri-consonant syllabary of Hebrew and its inflections, expanding the lexicon (Rosén 1967). This process gave rise to a distinct “philosophical” style of Hebrew (Rosén 1967), allowing Jews in all parts of the world to exchange written poetry, philosophy, science, medicine, commerce, daily correspondence and the drafting of contracts (Rabin 1972). A similar phenomenon propelling lexical expansion evolved from the late 18th century to the mid-19th century due to the “Haskalah” (Enlightenment) movement and developed a rich literary tradition, mostly in Europe (Rabin 1972, Tene 1969, Rosén 1967, Ornan 2009). 23

Ben-Yehuda’s research of the continuous historical development of Hebrew orthography led him to use the 7th century created Tiberian orthographic script to renew vocalization. This vocalization system consists of a diacritical vertical addition of marks to Hebrew consonants, marking vowels, finer distinctions of consonant quality and length, punctuation, and iambic stress. Table 1 /ד/ demonstrates diacritic marks with respect of IH vowel distinctive phonemes, appended to the

IH grapheme ([d]).

Table 1: Tiberian NIQUD according to IH vowel phonemes

Phoneme Tiberian NIQUD ד ד ד A ** ד ד ד ד E די ד I דֳ דוֹ ד O דּו דֻ U ד Ø* * signifying no vowel. Originally was [ɔ]. **eventually pronounced as [ɛ /e] when beginning a syllable in some consonants.

The NIKUD system as applied to IH is totally inefficient. This is exemplified by the fact that Israeli high-school students are forced to memorize it by heart for their Hebrew Language Matriculation exam and immediately forget it. The reason for this inefficiency stems from the fact that the NIKUD system was based on a more complex vowel system consisting of approximately 14 vowel phonemes where vowel length was a distinctive phonemic feature, while IH consists of approximately 5 vowel phonemes without vowel length as a distinctive feature. In addition, the mid front vowel in spoken IH is closer in articulation to [ɛ] rather than [e]. The reader should be aware that in texts on Hebrew Phonology, the [e] is used primarily even though the phonetic articulation in IH is much closer to [ɛ].

The different variations under the graphemes, respectively pertaining to vowel phonemes, describe different phonetic variations. The “” in IH([ə]), signifies either the absence of a 24

vowel (Ø) or the phoneme [ɛ], depending on its placement within the syllable or upon the consonant it marks. It should be pointed out that diacritics appear vertically, mainly under the consonants (except for some cases of [o] and [u]), although acoustically pronounced after the consonant marked.

Ben-Yehuda also turned to the methodologies developed during the Spanish “Golden Age” and the Haskalah to modernize the language (Rosén 1967), by updating, adapting and expanding the Hebrew lexicon with an array of new words for modern utilities and concepts. However, the extensive borrowing of terminology, primarily from European Latin , and its incorporation into the Semitic based alphabet of Hebrew, generated Latin-Semitic compounds. This in turn, engendered an array of pronunciation conflicts embedded in the consonant/vowel lettering ambiguities. This can be observed for example in the abundance of consonant clusters – itstrubal]; Hostel] אצטרובל – afarsek]; Pine nut] אפרסק – ambatya] ; Peach]אמבטיה – e.g. Bath) .([axsanya] אכסניה

This ambiguity in pronunciation was enhanced by the variety of pronunciations and intonations characteristic of the new-coming Jewish communities. While these communities, constituting the emergent Israeli society in Palestine towards the end of the 19th century, brought with them (from a traditional Hebrew education) knowledge of the Hebrew graphemes according to the Tiberian vocalization, all members of the same linguistic community also brought with them the phonemic background of their particular vernacular. Therefore, the Hebrew pronunciation of each different group established itself as a result of two factors (Blanc 1968): 1. a set of spelling-pronunciation rules that established grapheme-to-phoneme equivalences; 2. the allophonic and distributional mechanisms of the spoken vernacular. None of the spoken vernaculars had a phonemic pull equivalent to the graphemes of the Biblical Masoretic text. Communities differed in the actual establishment of the phonemic distinctions equivalent to the graphemic ones.

This is what probably led Ben-Yehuda to try and impose a standardized pronunciation and intonation along with his conviction that IH should be pronounced as in biblical times, before Hebrew lost its currency as an all engulfing language. Following extensive historical research he 25

concluded that IH pronunciation, intonation and syllabic stress9 should be conducted according to that characterizing “Mizrahi” or “Sepharadi”10 Jews. In an enlightening article, Ben Yehuda (1978) claims to have proven that the Sepharadi pronunciation has been well preserved from ancient times11 in the three main aspects of the debate. First, in respect of intonation, Ben Yehuda embraced the Sepharadi overruling iambic foot (ultima stress), which he maintained in his use of the Tiberian script. Second, he encouraged the reduction of vowels to five [a, e, i, o, u] after arguing that the Septuagint translation’s conflation of the various diacritic vowel marks into the five vowels of the reflected the pronunciation of Hebrew in Biblical times. Despite this position, Ben Yehuda did not modify the Tiberian script to reflect 5 vowels, creating thereby the inconsistence of vowel marks and resulting in vowel homophones12 (see table 1). Thirdly, in respect of consonant pronunciation, Ben Yehuda supported the pronunciation of some of these as pronounced by the Sepharadi Jews (see table 2).

9 Syllabic stress: a stress placed on a given syllable in a word of 2 or more syllables, resulting in the change of acoustic parameters (pitch, intensity, duration)

10 “Mizrahi” (literally Oriental) and “Sepharadi” (literally Spanish) refers to the pronunciation and intonation used in prayers and in reading the Bible by Jews coming from, or originating in Arab lands. David Yellin, who was nominated as president of the “Hebrew Language Committee” in 1904, supported Ben-Yehuda’s position and called from 1905 for the introduction of the Sephardi pronunciation. In 1913 Yellin convinced the Committee to adopt this pronunciation and it was henceforth mandatory for Hebrew teachers in Palestine (Tene 1969).

Ben Yehuda ,("המבטא בלשון העברית" /"In his seminal article on the subject ("Pronunciation in the Hebrew language 11 (1978) mostly based his argument in support of the Sepharadi pronunciation upon a Greek-Hebrew comparative analysis of the third Century B.C. Septuagint translation of the Bible into Greek. The Septuagin was intended for Greek speaking Jews living in Alexandria, Egypt, under Hellenic rule.

12 Homophone – the same phoneme represented by 2 or more different graphemes.

26

Table 2: Sepharadi Hebrew pronunciation for 10 phonemes and their respective graphemes

ק כ ת ט כ ח ו ב ע א Sepharadi Hebrew [q]=ק [k]=כ [t]=ת [ t]=ט [χ]=כ [ħ]=ח [و]w=ו [v]=ב [ʕ]=ע [ʔ]=א pronunciation*

While the above described Institutionalization of IH by Ben-Yehuda and his followers was highly instrumental and influential in the widespread revival of IH speech, some of its components countered the language user’s need for “minimal effort and maximal communication” (Tobin, 1997). We turn now to this embedded conflict to discuss how IH speech, as it consolidated after its triggering, had effected in turn on IH orthography.

I.2.2 From Spoken Practice to Orthographic Change

As mentioned, the Committee for Israeli Hebrew upheld Ben-Yehuda's use of the Tiberian script and took his position that the Sepharadi pronunciation was closest to the pronunciation of Hebrew before it ceased to be an engulfing spoken language. However, their assumption that these positions and their imposed inculcation through education will be sufficient for this institutionalized IH to materialize were wrong (Tene 1969). Ben-Yehuda himself, while calling for the institutionalization of the Sepharadi pronunciation, inconsistently supported the scientific approach whereby there is no such thing as an a-priori “real” or “unreal” pronunciation, as the only real and acceptable pronunciation is the one practiced by the majority of a community of language users, and that like in other aspects of language, pronunciation is dynamic by its nature and therefore prone to change in time (Ben Yehuda, 1978). 27

In reality, Modern Hebrew emerged from the dialectics between its institutionalization and its practice. As mentioned above, while the Tiberian orthographic script and the imposition of the Sepharadi pronunciation were highly instrumental in sparking and reviving spoken Hebrew as an all engulfing language, these often countered the intended user’s need for “minimal effort and maximal communication” (i.e. Tobin, 1990a,1990b,1997,). Thus, the Tiberian vocalization system’s diacritical vertical addition of marks to Hebrew consonants is counter-intuitive. Its markings upon or under a consonant do not conform to the linear acoustic characteristic of an utterance.13 Furthermore, given that in this system, vowel marks shift their phonetic pronunciation according to placement within the syllable or the consonant marked, an inconsistency ensues when the same vowel marking indicates different phonetic pronunciations (e.g., ָ , representing

both [o] & [a]), or when different vowel markings indicate homophones (e.g., ָ ָ , both signs representing the phoneme [a]). Finally, the system overloads the sequence of graphemes with different diacritical markings (for vowels, finer distinctions of consonant quality and length, punctuation, and stress), thereby arresting the flow of reading, certainly for those who are not practiced enough. Hence, while the Tiberian orthographic script approximates a shallow orthography it is inefficient because it does not accurately reflect the simpler IH vocalic system.14 This counterintuitive nature, marking overload, and its complicated rules engendering inconsistency and phonologically unjustified, turned the diacritics into a most demanding system for daily usage. This is what probably led, once vocalization was internalized, to the gradual

13 Relevant in this respect are some aspects of the (the dialect that for centuries served as an alternative vernacular language among the Diaspora Jewish communities in, or originating from Eastern and central Europe). Based on the Hebrew alphabet, Yiddish orthography does not use the Tiberian diacritical system to signal vowel pronunciation, but rather developed an extensive vowel sign orthography that renders a linear, shallow as a / ע/ (orthographic representation more attuned to the sequence of the utterance (e.g. using the letter (AYIN to exclusively represent the vowel [a]. Diacritic marks are rarely /א/(vowel grapheme representing [e] and the (ALEF

For further information .(,/א/ (used as in the representation of the vowel [o], where ָ is added under the letter (ALEF see Weinreich Uriel, Modern Yiddish-English English-Yiddish Dictionary, YIVO Institute for Jewish Research, New York, 1968.

14 This will be further discussed in respect of the methodology as well as the results of the present study. 28

abandonment of this very complex and cumbersome vertical diacritical marking system: As IH became an engulfing language used for all speaking and writing activities, its orthography gradually reverted to the fundamental tri-consonant root syllabary of ancient Hebrew.15 In fact, today, diacritics remain mostly in religious texts, poetry, children’s books, and, rarely, in other IH texts when extreme ambiguities need clarification such as those present in homographs or heteronyms.

Also, counter to the intentions of the Committee, Israeli Hebrew pronunciation developed from a mixture of the different Jewish traditions, generally tending towards simplification.16 This tendency resulted in an overall process of vowel and consonant phoneme reduction in accordance with the respective reductions of Ashkenazi and Sepharadi (Blanc 1968, Hoffman, 2004).

While the institutionally imposed Sepharadi pronunciation of consonants was in accordance with

15 Root system refers to a sequence of consonants that are used in the formation of word inflections, a peculiarity to Semitic languages. Such abstract sequences of consonants are used in the formation of actual words by adding the vowels and non-root consonants according to morphological rules. While there are some roots with more consonants, the majority of roots are tri-consonantal.

16 In the census carried out in May, 1961, 2,200,000 persons gave Israel as their country of permanent residence. Linguistically, three groups emerged. Group A consisted mainly of 200,000 Arabs, who declared that Palestinian Arabic was their first and main language. Group B, its primary language neither Hebrew nor Palestinian Arabic, numbered 1,500,000 (74.7% of the Jewish population). Group C was made up of about 500,000 inhabitants who gave Hebrew as their first and main language, constituting 25.3% of the Jewish population (Tene 1969). Among the group of those who claimed Hebrew as their native language two varieties emerged, depending on the primary language of their parents and/or grandparents: One group belonged to parents of speakers that originally came from countries, which, before World War I, had belonged to the Ottoman Empire, including Palestine. Their primary language was some dialect of spoken Arabic and/or Ladino (Judeo-Spanish). The second group belonged to speakers whose parents immigrated from Eastern or Central Europe and their primary language was generally some dialect of Yiddish and/or a European language pronounced with a tinge of Yiddish. As suggested by Tene (1969), "This classification of varieties within native Hebrew speech and its correlation to these primary languages were, of course, a schematic generalization” (p.52).Still, it serves to describe and understand the evolution of spoken Israeli Hebrew. 29

the various consonant graphemes of the Hebrew alphabet, it conflicted with Ashkenazi pronunciation,17 which created homophones out of closely pronounced consonants, particularly /כּ/-[and [k/ק/-[out of differing pharyngeal and uvular consonants (e.g. Ashkenazi homophone [q into [ʔ]. Hence, the tendency towards /ע/-[and [ʕ /א/-[into []; [ʔ /ח/-[and [ʕ /כ/-[into [k]; [x simplification may be said to have driven most IH speakers to adopt the Ashkenazy reduction of consonant pronunciation (Hoffman 2004) towards “loss of phonetic distinction” (Bolozky 1978, p.11-12) (see table 3).18

Table 3: Sephardic Hebrew pronunciation for 10 phonemes and their respective graphemes compared to their pronunciation in present day IH

ק-כ ת-ט כ-ח ו-ב ע-א Graphemes

[q]=ק ;[k]=כ [t]=ת ;[ t]= ט [χ]= כ ;[ħ]= ח [و] = ו ;[v] = ב [ʕ]=ע ;[ʔ]=א Sepharadi Hebrew pronunciation*

[k]= ק + כ [t]= ת + ט [χ]= כ + ח [v]= ו + ב [ʔ]= ע + א Pronunciation in present days **

17 Ashkenazi refers to the Jewish communities originating in Eastern and central Europe. Their original pronunciation differed from the Sephardic pronunciation mainly in pharyngeal consonants, in their richer vowel variation, and in their mainly trochaic accent, where stress is put upon the first or penultimate syllable in a word as opposed to the Sephardic iambic accent, stressing the ultimate syllable in a word.

18 The following citation I translated from Ben-Yehuda (1978), presents the approach of the Academy at that time: “Our language, a Semitic Language, was put counter to its favor into European pronunciation tools. These pronunciation tools were shaped by European languages that lack pharyngeal phonemes, and according to the rule that Human nature chooses the easy way out, Hebrew speakers from Europe abandoned many phonemes of the Semitic Hebrew language, including grammar/language Professors in the University” (Leshonenu 1963: p.98). 30

Likewise, spoken IH rejected the variation in vowel pronunciation implied in the Tiberian script vowel markings and pronounced by Ashkenazi, and adopted the Sepharadi vowel pronunciation into [e]),19 / ד /-[and [e / ד /-[which conflated the Tiberian script's vowel-marking variations (e.g. [ei into the five vowels: [a, e, i, o, u] (see table 1).

The bi-directional influences between the emergent written and spoken IH, driven by Hebrew’s specific deep orthography, resulted in a modern IH characterized in speech by the reduction of neighboring consonants and vowels into homophones and in a written Hebrew stripped of its initial complex vowel marking diacritics.

As my research will show, IH orthographic reversion to the its consonant based syllabary has effects in turn on IH speech production, particularly in respect to the characteristic emphasis upon consonant voicing on account of vowel voicing.

19 While Ben-Yehuda insisted that the Sephardic pronunciation of vowels was the one intended by the Tiberian script, his arguable argument was flawed since it was based on the Greek conflation of these vowels in the Septuagint translation. As opposed to his argument in favor of Sephardic pronunciation in respect of consonant variation as stemming from the Hebrew alphabet, he did not provide a satisfactory explanation for the vowel variations implied in the Tiberian script marking, arguing instead that the pronunciation practiced by the Sephardic retained for Historical reasons the Biblical pronunciation. 31

I.3 Mapping IH Orthography

The Hebrew writing system is an alphabetic based orthography, namely, it belongs to the family of languages based on the phoneme,20 consonant or vowel, as minimal grapheme (as in all Latin or Semitic languages).21

Within languages with alphabetic orthography, Hebrew belongs to the Semitic languages, characterized by a consonant based morphologic root system, where letters represent consonants with an overall lack of overt vowel signs. Nevertheless, Hebrew has four consonants Hebrew also .(/א, ה, ו, י /that occasionally function linearly as vowel letters (ALEF, HEI, VAV, YOD has an orthographic script where vowels are presented through the vertical addition of diacritical points to consonant letters (an orthographic script presently used only in religious literature, poetry and children’s books).

I.3.1 Hebrew’s Syllabic Orthography

The specific depth of Hebrew orthography results from its fundamental consonant-based syllabic orthography, where there is missing phonemic information in the written representation of

20 Phoneme: the smallest/minimal sound unit to define and change meaning in a specific language: the total sounds of a language that define meaning become its phonological system.

21 Alphabetic orthography differs from logographic orthography (as in Chinese), which is based on a word or as minimal grapheme.

32

syllables due to a lack of vowel representation. In other words, in Hebrew, overall, the vowel within a syllable is not represented but latent in the consonants constituting it.22

This phenomenon, most likely stems from Hebrew’s characteristic root system23 and its rich, consonant based morphological inflections. This results in a situation whereby, at least in the reading process, where only consonants are visualized, vowels are basically assumed by readers according to their lexical knowledge with regards to the possible word pattern for full phonological retrieval (Frost 2006). This situation is succinctly emblematized in the outstanding phenomenon of Hebrew homographs, where the same sequence of consonants and their respective phonology express different meanings, or of heteronyms, where the same consonant sequence expresses different meanings through a different phonology. In such cases, when no semantic redundancy is given due to context, one would not know the phonemic breakdown or the relevant meaning.

grapheme sequence can be a homograph or a (ספר) For example, the tri-consonant root s-f/p-r [represents two [p פ heteronym generating 11 different meanings: (the same letter and [f] which are now phonemic). 24

22 Written Hebrew’s vowel-lack as generating Hebrew’s deep orthography differs, for example, from whose depth is due to inconsistencies in the grapheme representation of phonemes. In other words, English, as opposed to Hebrew, uses letter vowels in syllables, but their representation of the phonemic utterance varies in different letter clusters (such as enough/In—f/, duck/d—k/). In fact, when Hebrew is written using the Tiberian orthographic script where vowels are presented through the vertical addition of diacritical points to consonant letters, it approximates a shallow orthography in its completeness of representation of the phonetic articulation. Yishai Tobin reports that when growing up he saw in a New York subway the following advertisement for a secretarial school; “If u cn rd ths u cn bcm a sctry” reflecting the nature of the Hebrew deep orthography.

24 The Sephardic pronunciation, established by Ben Yehuda, maintained only three of the alternations originally KAF) [k/x], while the Ashkenazic had kept) /כ/ ,[PEI) [p/f) /פ/ ,[BET) [b/v) /ב/ :reflected in the begged kefet alternations DALET) (Hoffman) /ד/ GIMAL) and) /ג/TAF) [t/s], and both systems leaving out the original alternations of) /ת/ the 2004). 33

1. SEFER=book {noun}

2. SAPAR= a barber {noun}

3. SIPER= he has cut the hair {[verb}

4. SIPER= he told {verb}

5. SUPAR= something was told {verb}

6. SUPAR (homograph to No. 5) hair was cut {verb}

7. SAPER= tell! {verb}

8. SAPER = (homograph to No. 7) cut! {verb}

9. SAFAR = he counted {verb}

10. SFAR = fringe {adjective}

11. SPER (adopted from the English spare) = an extra object (e.g. a spare wheel) [adjective}

Notwithstanding the overall lack of vowel representation, as emblematized in the recurrence of homographs and heteronyms, the Hebrew alphabet does include four graphemes (ALEF, HEI, which when articulated are considered to be consonants, but when not (/ א, ה, ו, י / VAV, YOD articulated, i.e. silent, they serve as a vowel sign. However, when used to signify a vowel, these graphemes only signify that a vowel is represented without specifying which because each such grapheme functions to represent several vowels (see table 4).

Table 4: Consonant/vowel graphemes

The Grapheme Pronounced as Pronounced as consonant vowel ,(ALEF) [ʔ] [a (most frequent) א e, i, o, u] [HEI) [h] [a, e) ה [VAV) [v] [o, u) ו

[YOD) [y] [i (most frequent), e) י

34

The occasional representation of vowels by graphemes that also function as consonants,25 as well as the opaque specification of the vowel represented by these graphemes, testifies to the subservient nature of vowels within Hebrew’s consonant dominated syllabary.

It should be pointed out however, that despite the occasional ambiguity in pronunciation engendered by homographs, heteronyms and the unspecified vowel value of consonants-turned- vowels, in Hebrew “more often than not, the connections and relations between lexical root and the fixed morphophonemic patterns are rather transparent” (Tobin 1990b, p.138). This transparency however, does not result from a complete grapheme representation of utterances (because vowels are either latent or unspecified), but rather from the fixed rich morphological inflections of the tri-consonant root system. These inflections, themselves signaled by consonants, establish the necessary context indicating the type of latent or unspecified vowel to be pronounced, but do not turn IH orthography into a shallow one in respect to the graphemes' completeness of phonemic representation.

I.4 Orthographic Effects on IH Speech Production

A survey of the growing body of psycholinguistic and brain imaging research gathered during the past decade on the influence of orthography on spoken language, offers a meaningful background and supports my novel approach and related innovative methodology.

I.4.1 Orthographic Influence on Speech Perception

Seidenberg and Tanenhaus (1979) were among the first to research the influence of orthography on speech perception. Through an auditory rhyming judgment task subjects were asked to

presents a special case. This exception, discussed in (א) Notwithstanding this overall characterization, the Alef 25 detail further on, is relevant when focusing on the vowel-consonant relation between orthography and speech. 35

determine whether pairs of words that varied in their orthographic overlap, rhymed. They found out that subjects were faster to judge as rhyming those pairs that shared an orthographic representation of the rhyme (like pie-tie) over those that only rhymed phonologically but were orthographically different (like rye-tie).

This type of research, focused on measuring the decision latencies of response to words with consistent and inconsistent spelled syllables, has become an overruling paradigm when examining the influence of orthography on speech processing.

For example, Ziegler and Ferrand (1998) observed a consistency effect of orthographic representation in an auditory and visual lexical decision task based on phonological processing. Defining a consistent syllable as one that can be spelled only in one manner (e.g. the ending [ɪŋ] such as in ring, going or singing), and an inconsistent syllable as one whose sound has more than one written representation (e.g. [ʌp] may be scripted as cup or stop and [ɛɪn] forms the rhyming part of sane and rain), they found that subjects were faster to respond to words whose rhymes can only be spelled one way (consistent) and slower in their response to words whose rhymes can be represented in multiple ways (inconsistent). The existence of an orthographic consistency effect in spoken or visualized word recognition has been demonstrated in different degrees for diverse languages (English: Miller & Swick, 2003; Portuguese: Ventura et al 2004; French: Pattamadilok et al, 2007).

The above findings were further corroborated by recent brain imaging studies. Perre & Ziegler (2008) for example, aimed to find on-line activation of orthographic information during spoken word recognition. Using ERP (Event Related brain Potential) they provided spoken words with consistent and inconsistent orthography and tracked the online activation of the brain in respect of lexical decision. In both types of words they noticed that activation was not post-lexical but rather pre-lexical or upon lexical decision, with activation occurring earlier in orthographic consistent words. These findings imply that the decision between a word and a non-word computes orthographic information on-line, as we listen to the spoken words.

The above-mentioned research has examined how orthography influences speech processing in phonological as well as lexical decision tasks. However, the orthographic consistency effects detected were specific process dependent and the question arose as to whether orthographic 36

influences in spoken word recognition also emerge in tasks requiring word processing beyond simple lexical access. This has been undertaken in a study conducted by Peereman et al (2009), where they showed that orthographic consistency also influences semantic and gender categorization tasks.26

Another relevant precedent for my research is the experiment carried out by Ziegler & Munreaux (2007) were they showed that orthographic information influences in particular advanced readers' spoken word recognition. Based on findings by Ziegler et al (2004) suggesting that orthographic neighbors27 facilitate processing of spoken word recognition, they conducted a study to examine the effect both of orthographic neighborhood (ON) and Phonologic neighborhood28 (PN) within 3 groups: Beginning readers; advanced readers, and dyslexics that were matched chronologically to the advanced readers. Results confirmed predictions that ON influenced the experienced readers strongly in contrast to the beginners as well as the matched age Dyslexics, whereas PN was not influenced by reading level. This skill based hypothesis was further corroborated in a study by Desroches et al (2010).29 Using fMRI to examine the differences in brain activation between children with reading difficulties (9-15) and age-matched children with typical achievement (as they responded to an auditory rhyming task), both groups showed similar activation in regions associated with phonological processing, but only the typical achievers showed activation in the cortex region implicated in orthographic processing.30

26 Further evidence of orthographic consistency effects in semantic tasks was provided by Pattamadilok et al (2008).

27 Orthographic neighborhood: a word having orthographic similarity with other words in the mental lexicon.

28 Phonological neighborhood: a word having phonological similarity with other words in the mental lexicon

29 Desroches provides in this article a comprehensive and in-depth survey of recent neuro-imaging research supporting the effects of orthography on speech perception processing without the presence of external visual stimuli.

30 Of particular interest is a follow-up study by Perre et al (2009) to the above mentioned ERP study, where he used standardized low-resolution electromagnetic tomography (sLORETA) in order to determine the possible underlying cortical generators of the orthographic consistency effect. The results showed that the effect was clearly localized in a classic phonological area with no evidence found for activation in the posterior cortical areas coding orthographic 37

The findings of these studies provide support for the online effect of orthography on the auditory and visual phonological perception processing of literates. Moreover, these studies indicate that literates, in the process of phonological input processing (e.g. spoken words) are affected by mentally stored orthographic schemes (e.g. letters, words) when there is no external visual stimuli as in reading. In these respects, the studies surveyed above form a meaningful background to, and support of my research. However, these findings are limited to phonological or visual perception processing. My own research also concerns speech production as well. A survey of recent literature supporting my contention on the orthographic effects on speech production will follow.

I.4.2 Orthographic Influences on Speech Production Processes

Recent research provides evidence that orthographic effects on speech perception are replicated in speech production. Using a new experiment involving both perception and production processing in respect of orthographic effects, Rastle et al (2011) designed a study in which participants were trained on the first day on a set of associations between invented words and photos of unidentified objects. Spelling-sound consistent or spelling-sound inconsistent writing of the invented words was introduced on the 2nd day (e.g. [kisp] spelled KISP or CHISP, [hɒf] spelled WHOFF or HOFF, based on existing English spelling – sound relationships). The influence of these on speech production processing was assessed on the 3rd day. Results showed significant similar orthographic effects on speech perception and speech production in respect of latency responses concerning lexical decision and naming tasks. Their results robustly indicated that both in speech perception and production, there is a rapid and automatic activation of orthography.

information, such as the visual word form area (VWFA). These findings further support the hypothesis that phonological representations are “contaminated” by orthographic knowledge without recourse to external or internalized visual stimuli.

38

In an earlier groundbreaking study, Damian and Bowers (2003) conducted four experiments in which they asked participants to memorize three types of word pairs: Homogenous word pairs, that is words which overlapped in their initial letter and its sounding (e.g., coffee–camel); heterogeneous word pairs with no overlap (e.g., gypsy–camel), and inconsistent word pairs overlapping in their initial sound but not in its lettering (e.g., kidney–camel). Participants were required to produce a response word to a prompt word in the sets of words belonging to the same category, by reading the response word to the presented prompt word, but more revealing, by saying aloud their memorized response word following their listening to the prompt word in a total auditory task without any word visualization. Results showed that in respect of the inconsistent word-pairs there was a similar slower response in both the reading and auditory conditions. This study strongly suggests that orthography similarly affects the speech production of literate speakers both during reading aloud and when responding aloud to an auditory stimulus without visual representation.

Taken together, the above studies indicate that orthography not only affects speech perception but also speech production, both when reading aloud and when producing speech without an external visual representation of the words. However, all of the studies discussed so far are limited to measuring latency of response whereas I am concerned with the qualitative effects that orthography has on speech production - that is, on diction.

I.4.3 Orthographic Influences on Diction

Relative to studies on latency of response as indicating the effects of orthography on speech perception and production, I have found only one meaningful study that addresses the effects of orthography on diction. Jordan Brewer, in her dissertation (Brewer 2008) conducted a series of experiments that showed how the orthographic characteristics of word-final spelling and of the spelling of the words themselves (specifically the amount of letters in final position of words) influence the duration of the spoken production of these words by native American-English literate adults. The results showed an unambiguous effect of orthography on speech production. This unambiguous effect of orthography on speech production was found both in a controlled test using a reading aloud elicitation task of words with different orthographic representations of 39

specific sounds (i.e. rich/switch, wait/weight) and during the spontaneous generation of words within a sociolinguistic interview format. Brewer's experiments strongly support the hypotheses underlying my thesis insofar as these measured the effects of orthography on diction in respect of letter amount as affecting length of utterance (rather than reaction latencies of response as in the studies dealt with above, which is irrelevant for assessing orthographic influences on diction). However, Brewer's experiments measured a very limited aspect of diction without considering the possible effects of orthography on prosodic features or on the quality of the speech-sound.

Differing from Brewer's study, this dissertation focuses on the specific temporal relation between consonants and vowels in speech production as it is influenced by orthography. Thus this research offers an innovative methodology and provides a new tool that will allow the examination of how the orthography specific to a given language influences the quality of diction. While the present research is conducted upon speakers of IH, a language that offers a unique example of the relation and tension between orthography and speech production, the approach used may provide a new way to evaluate the relation of orthography to diction in different orthographic systems and their peculiar orthographic depth.

Specific research that offers an initial but very important support for this study's concern with the influence of Hebrew orthography on the diction of IH literates can be found in the “Orthographic Depth Hypothesis” (ODH) presented by Katz & Frost (1992). Based on cross languages research, they contend that the reading process is different for different orthographies. Thus, among alphabetic orthographies, where depth is the result of differences between the languages' phonology and , the reading process differs according to orthographic depth (that is, the degree of correspondence between the phoneme and the grapheme). Thus, in languages with a shallow orthography (like Spanish), the phonology encoding in a reading process is closer to the phonology created and stored during the process of language acquisition. However, in languages with deeper orthographies (like Hebrew), the reader is encouraged to process printed words by referring to morphology via the printed word’s visual-orthographic structure along the mind language templates created and stored during literacy acquisition. Specifically in respect to IH, which as detailed above, is a fundamentally consonant-based syllabary, it has been consistently revealed that reading IH involves the sequential interplay of two computational processes: The first process is characterized by the conversion of units of single letters into consonantal information, and the second process involves the use of morphological information, 40

mainly word-pattern , to provide the vowels missing in the written representation (Frost 1995, 2006). Thus, psycholinguistic research shows that the mind language templates of literate Israelis take mostly the form of consonants, primed over vowels in the process of reading (Ben Dror et al 1995, Frost 1995, 2006).

These findings, indicating the primacy of consonants in the processing of IH by literate IH speakers, supports my hypothesis that IH orthographic depth plays an important and influential role in the diction of IH speakers.

I.5 The research question:

How IH orthography -- a fundamentally consonant-based syllabary --influences the diction of Literate IH speaking adults?

I.6 The Hypothesis:

In light of the characteristics of IH's deep consonant based syllabic orthography, and following psycholinguistic research showing that the mind language templates of literate Israelis take the form of consonants having priming over vowel representation, it is hypothesized that IH orthography engenders in IH literates a consonant-based diction. That is, in the diction of IH speaking literates, extra emphasis is put on consonant production on account of vowel production.

I.7 The study objectives:

The study’s main objective was to find a direct influence of IH orthographic depth on IH diction during reading aloud and during a free speech response to an oral question.

41

II. Methodology

My hypothesis suggests that in the diction of IH speaking literates, extra emphasis is put on consonant production at the expense of vowel production due to the overall lack of vowel representation in IH orthography. In order to test my hypothesis I focused upon a group of IH native speakers that were Literate Professional Voice Users (LPVU) of IH. This group was chosen because it is homogeneous in respect to the independent variables of the study: LPVUs share a high level of literacy, a daily use of orthography as stimulus for speech production, and awareness to diction.31 Hence, it is reasonable to assume that LPVUs of IH are an exemplary sample demonstrating a more general tendency in other literate experienced native IH speakers.

I hypothesized that LPVUs, consciously concerned with prominence and accurate diction for clarity of speech, will put extra emphasis in their diction upon: (a) symbols that they are visually exposed to in reading aloud tasks, or, (b) when in free speech, where no visual representation is given, emphasis in diction will be upon symbols that have the innate primacy in retrieval form their phonological templates (Ben Dror et al 1995, Frost 1995; 2006). In both cases I hypothesized that the LPVUs, when reading aloud or when spontaneously speaking IH, will emphasize the articulation of consonants and de-emphasize the articulation of vowels in accordance with IH's deep orthography. I then set out to measure through spectrograms the temporal ratio between vowels and consonants in the LPVUs utterances.32 Out of the three acoustic parameters characterizing diction emphasis, namely length, intensity and pitch, such

31 LPVUs are concerned with promoting the clarity and comprehension of speech, which is dependent on the full presence of the consonants with minimal co-articulation and processes. Also, they are concerned with balancing voice projection and clear diction, dependent to a large degree upon balancing the emphasis on consonants and vowels through length (i.e. Linklater 1976)

32 The consonant/vowel ratio usually referred to in the literature, measures the average quantity of vowels vs. consonants in a language. This ratio is then calculated simply by dividing the number of consonants (C) by the number of vowel qualities (VQ) and is referred to as the C/VQ ratio in a language system (Haspelmath M. et al, 2008). Such ratio measurement is irrelevant for my study which measures the proportional time allocated to the speech production of vowels vs. consonants. 42

measurement mostly provides information upon emphasis through length.33 This is a novel and simple method, rendering clear cut results on the variety in emphasis through length. It is also particularly useful for studying orthographic influences on IH diction (in IH or in other Semitic languages), in that it allows to test the relation of emphasis through length to IH orthography's representation of consonants with an overall lack of vowel representation. In order to test this relation I manipulated orthographic scripts in a pilot experiment and in a revised extended experiment following the pilot's results. In both controlled experiments I measured through spectrograms whether vowel length is or can be manipulated by the nature of a written script. It was predicted that when presenting the participants with the same words or sentences, once written in the established IH consonant based syllabic orthography, namely with the absence of most overt vowel presentations, and once when written in a linear consonant-vowel letter orthography, a significant difference in the ratio of the duration of vowels versus consonants will be observed. Subjects were asked to read aloud these orthographic scripts and the recorded readings were then measured in respect to their consonant to vowel temporal ratio. This allowed me to test the relation between orthographic representation and diction variability concerning emphasis through length. Following the robust results obtained in these controlled tests, I conducted a third experiment where subjects were asked to spontaneously respond in speech to an identical question. I then measured through spectrograms their articulation of orthographically established CVC stressed syllables, comparing those with orthographic vowel representation to those without it. The findings replicated the results of the controlled reading aloud tasks, in that emphasis through length varied according to the innate orthographic primacy in retrieval of the different internalized orthographic scripts.

33 Measuring varieties of emphasis through length also indicates varieties in emphasis through intensity and pitch as will be discussed further below. 43

II.1 The Pilot Experiment

Given that the test was the first of its kind, specifically developed for this study and used for the first time, I wanted to evaluate the quality and problems of the test material, procedure and the subjects’ performance, in order to improve the performance and validity of results in a broader and more controlled experiment. Since some of the pilot's results could carry on to the more elaborated experiment and since the weaknesses discovered in other aspects of the Pilot experiment form the rational of the changes entered into the material and procedure of the elaborate experiment, it is worthwhile to fully present the pilot experiment.

II.1.2 Material:

We presented the subjects with a test, which consisted of different orthographic patterns representing the same words with vowels present, partially present or implied in order to observe whether this would lead to an immediate difference in the temporal vowel to consonant ratio. In order to be sure that observed vowel length variation exclusively stems from the change in orthography, I devised the reading material presented to the subjects in an attempt to control stress factors that determine vowel length in IH (beyond the orthographic factor I aim to prove).34

The test consisted of 3 groups of sentences:

1) 10 mono-syllable word sentences of 2-5 words: I chose IH monosyllables in order to exclude the syllabic stress factor influencing vowel length,35 which can only come forth in words with two

34 Another possible interference in terms of vowel length could have been variations in distinctive phonemes. However, while this is the case in Japanese for example, and may have been the case in ancient Hebrew (as implied in the differing vowel marks of the Tiberian script - see above, in the section dealing with IH revival), this is no longer the case in IH (as a result of the vowel conflations discussed ibid.). In IH there are no distinctive phonemes based on vowel or consonant length (see Becker 2003, Bolozky 1978).

35 Phonetic measurements show that vowels in stressed syllables are twice as long as vowels in unstressed syllables, regardless of syllable structure. 44

or more syllables. In order to control vowel length variation stemming from sporadic or pragmatic prosodic stress36 or caused by intonation marks (i.e. question, imperative or full stop marks) I varied prosodic stress (e.g. by using different intonation marks for each sentence ending) and used a mean vowel length calculation for each sentence to see whether different prosodic stresses meaningfully affect the overall ratio between vowels and consonants. In order to assess if vowel length is affected by orthography, the sentences were presented before the subjects in IH orthography where vowel representations are absent or partially absent, and compared their reading aloud to the reading aloud of the same sentences written in a “HEBRISH” script (i.e. phonemic IH rendered orthographically in a and offering a complete and consistent shallow orthography), which allows offering a linear and full representation of the vowels, more than any IH vowel additions to established IH orthography (for the latter, see group 3 below).

2) 10 monosyllable pseudo word sentences of 2-5 words each conforming to Hebrew phonology37. were presented in IH orthography and in HEBRISH. The use of pseudo words was chosen in order to neutralize possible influences on speech production stemming from automatic retrieval from the stored word lexicon (as in whole sentence content redundancy). It also lays bare any observable online influence of orthography on the phonemic expression of the written script while reading aloud. This strategy of using pseudo-words is widely used in this respect. It follows well established evidence that orthographic influences on speech production arise indirectly from the stored phonological lexicon, and that the nature of the stored phonological lexicon is modified by orthographic knowledge (Taft 2011).

3) As multi-syllable sentences are more likely to reflect connected speech, also when read from written material, 10 multi-syllable sentences of 4-5 words each where presented. These sentences were presented in an established IH script and in an IH script with an extended linear

36 Prosodic stress: stress patterns within a prosodic unit that may involve the placing of emphasis expressed on particular words which may add or change the meaning of a sentence. Emphasis is expressed by change of part or all the acoustical parameters of pitch, intensity and duration of a stressed word.

37 Most of these words were taken from a list of words constructed by Most & Adi Ben Said (2001). 45

representation of vowel letters. In IH, although often omitted, it is feasible to overtly and linearly present most vowels, a task made easier in multi-syllable words. 38 In some cases the addition of vowel letters exceeded the known and commonly used Hebrew Academy modifications.39 This was done tendentiously, and according to the test’s objective. As with the use of HEBRISH in the two previous groups, this was done in order to assess if vowel length is affected by orthography. I did not present this group with a HEBRISH script as earlier because I wanted to test the influence of actual and linear representation of vowels within IH orthography. I also estimated that more complex material involving sentences composed of multi-syllable words would demand longer practice in HEBRISH for the newly presented script, at least for part of the subjects who might have been unfamiliar with Hebrew written in Latin letters, causing hesitation that might affect the consonant/vowel ratio irrespective of orthographic effects.

Examples of the test material are presented in table 5.

VAV for [o] or [u]; [a] & [e] are often /ו/ YOD for [i] and / י/ i] [o] [u] are occasionally signaled by the vowel letters] 38 [HEI, where distinction between the two vowels [a /ה/ALEF or /א/ signaled in final word position by the vowel letters .[can be read as [nir-a] as well as [nir-e נראה .and [e] is often ambiguous (e.g

39 For further information, see http://hebrew-academy.huji.ac.il/hahlatot/MissingVocalizationSpelling/Pages/ikkar.aspx (also: "כללי הכתיב חסר הניקוד ,"לשוננו לעם ,טבת התשס"ב(.

46

Table 5: Test’s demonstration in order40 of presentation*

Order of Sentences Orthography Example Vowel Presentation Presentation Groups יש לו לב רע! Monosyllable IHS /YESH LO 1 sentences LEV RA/ (He has a bad heart) I

Weak Vowel ספרתי ספור מאד מענין. Multi syllable IHS 2 sentences /SIPARTI SIPUR Presentation MEOD MEANYEN./(I told a very interesting story. פש הל מיט? Pseudo IHS /PESH 3 monosyllable HEL MIT? / words sentences 4 Pseudo “HEBRISH” PESH HEL MIT? monosyllable (phonemic words orthography) II sentences Strong Vowel 5 Monosyllable “HEBRISH” YESH LO LEV RA! Presentation sentences (phonemic orthography)

סיפרתי סיפור מאוד מעניין. Multi syllable IH extended 6 sentences vowel letters /SIPARTI SIPUR presentation MEOD MEANYEN. (I told a very interesting

(IHVL) story.)

* For the full test see Appendix I.

40The test was presented continuously. The order of presentation was aimed at neutralizing influences of word memorization from one test to the other. I therefore separated the IH written tests (1) from its HEBRISH counterpart (5) and the IH established multi-syllable test (2) from its version with linear vowel representation (6). The IH and HEBRISH pseudo words counterparts (3, 4) were not separated because no known word memorization is involved). 47

II.1.3 Experiment Procedure

No coaching and guidance of the participants concerning the rationale of the pilot experiment was provided. A pre-test tutoring and practice was carried to facilitate reading a HEBRISH script in order to avoid an English phonological influence, as well as to assure the right interpretation of phonemes exclusive to Hebrew that do not have an English equivalent. Tutoring was repeated once more before the first encounter with the actual HEBRISH script of the test. No further information about the nature of the test was given to the participants. It should be noted that in Israel, the English alphabet is highly familiar to IH literates given that English is the default foreign Language and the first foreign language Israeli children learn from primary school onwards.

Each participant was separately presented with the six groups of sentences in the pre-planned order (see table 5) and was asked to read these aloud. The reading was recorded for later analysis.

II.1.4 Tools

Recording and spectrographic analysis were carried out with the Kay Pentax Multi Speech Lab, Model 3700, version 3.2.

II.1.5 Measurements

Each phoneme was measured on the spectrogram in order to obtain the respective length of each consonant and each vowel produced. Spectrogram measuring was simultaneously supported by the audio input for maximum accuracy: The transit point for each phoneme was decided according to the co-articulation and assimilation processes characterizing each subject's 48

speech production.41 In this respect transition was determined when the majority of consonant characteristics faded towards the vowel’s formants and vice-versa. The transition point in consonant clusters in words alone was irrelevant for the ratio calculation which systematically included the overall length of consonants with respect of vowels in each word or sentence. Consistency was kept by immediately marking the same point in each of the same vowel/consonant transition in the two recorded utterances of the parallel orthographic versions. Whenever the visual spectrogram was ambiguous with respect of the transition from the consonant to the vowel or vice versa, audio input was used to disambiguate. Likewise, in all stop consonants, whenever followed by obstruent, non-phonemic bursts, these bursts were measured as part of the consonants. These border definitions were consistently implemented throughout the entire analysis.

The vowel to consonant ratio was calculated by extracting the overall vowels' length in each sentence from the overall length of the sentence utterance (as a result, the consonants' duration is the difference between the vowels' length and the length of the overall production).

II.1.6 Subjects

Four LPVU participants were recorded and their spectrograms analyzed:

1) an actress aged 26

2) a male light music singer aged 28

3) a female classical singer aged 26

(4) a female teacher aged 52.

41 Co-articulation in connected speech is defined as "The variation in the phonetic manifestation of a given sound due to taking on some of the features of nearby sounds" (Ohala 1993). Assimilation refers to the phonological phenomenon whereby a segment changes to resemble its neighbours more closely. 49

Beyond attempting to get an indication on whether orthography overall affects LPVUs speech production, their different disciplines as well as their age and gender differences were taken into account as initial indication on whether these variants are meaningful in respect of the study's objective.

II.1.7 Results

For the first subject (actress) the analyzed material included: 1) all 10 IH sentences as well as their HEBRISH counterparts; 2) 8 out of 10 Hebrew pseudo words sentences and their HEBRISH counterparts (the last 2 sentences and their counterparts were dropped due to recording quality); 3) 4 IH multi-syllable-words sentences with few vowel letters along with the counterparts that included extended linearly represented vowel letters. No further sentences in this group were analyzed as the data showed speech production consistency in statistically sufficient data (given that the words were multi-syllabic). The sentences chosen for analysis (1, 4, 6, 8 - see appendix I) were those with the most significant difference between the 2 versions (i.e. with poor versus extended presentation of vowel letters). The full results of subject one are presented in figures 1, 2, 3 showing that this subject's consistency was evident in all tests. Given this exemplary consistency, the same four sentences of group 3 were analyzed for the other three subjects and only 5 sentences and their counterparts from groups 1 and 2. To get maximum variation in the selection of sentences within group 1, I chose sentences that were phonemically balanced, and with varying amount of syllable types (CVC, VC, CV), words and prosodic ending signs (see in appendix I, group 1, sentences 3, 4, 6, 8, 10).

For all measurements and results presented, the dependent variable was the overall vowel length in a sentence, expressed in percentage, in relation to the whole sentence itself as measured on the spectrograms.

50

Figure 1: Vowel length comparison of 10 IH monosyllable words sentences in IH and in HEBRISH scripts for subject 1 (actress).

80% 70% 60% 50% 40% 30% 20% 10% 0% total 1.1- 1.10- 1.2-6.1 1.3-6.3 1.4-6.7 1.5-6.4 1.6-6.5 1.7-6.6 1.8-6.8 1.9-6.2 averag 6.10 6.9 e Hebrew 45% 43% 41% 36% 55% 32% 37% 39% 37% 51% 41.60 Hebrish 71% 47% 50% 36% 56% 45% 41% 37% 38% 54% 47.50

Hebrew Hebrish

Figure 2: Vowel length comparison of 8 IH monosyllable pseudo words sentences in IH and in HEBRISH scripts for subject 1 (actress). 60% 50% 40% 30% 20% 10% 0% total 3.1-5.9 3.2-5.2 3.3-5.1 3.4-5.5 3.5-5.6 3.6-5.7 3.7-5.8 3.8-5.3 averag e Hebrew 28% 25% 33% 34% 38% 30% 29% 35% 31.50% Hebrish 34% 31% 33% 39% 48% 34% 37% 36% 36.50%

51

Figure 3: Vowel length comparison of 4 IH multi-syllable words sentences in established IH script and in IHVL script(i.e. full script) for subject 1 (actress). 70% 60% 50% 40% 30% 20% 10% 0% 2.1-7.1 2.4-7.4 2.6-7.6 2.8-7.8 average Hebrew 35% 40% 44% 37% 39.00% Hebrew+ 37% 52% 58% 48% 48.75%

Results for each of the 4 subjects as well as the overall averages are presented in tables 6 and 7. Paired t-tests were used to compare IH orthography to its HEBRISH version for all subjects.

52

Table 6: Descriptive statistics as well as comparison of vowel length in IH and in HEBRISH scripts

. Subject Sentences IH mono- syllable Mono-syllable Pseudo IH Multi syllable type words words Script IH HEBRISH IH HEBRISH IHS IHVL

1 41.60% 47.50% 31.50% 36.50% 39.00% 48.75%

2 36.80% 40.40% 26.80% 31.40% 36.25% 42.25% 3 41.00% 53.00% 29.00% 39.40% 42.80% 46.00%

4 41.00% 48.60% 30.00% 36.80% 38.42% 43.11%

Average for all 4 40.10% 47.38% 29.33% 36.03% 39.12% 45.03% subjects Paired t-test for all 4 0.02* 0.007** 0.013* subjects

* p<0.05 ** p<0.01

Table 7 Descriptive statistics as well as comparison of vowel length in IH and HEBRISH scripts

Subject Lack of vowel letters with vowel presentation letters presentation 1 37.37% 44.25% 2 33.28% 38.02% 3 37.60% 46.13% 4 36.42% 43.13%

Total average for all 4 subjects 36.17% 42.88% SD 1.99% 3.47% Paired t-test: 0.008**

** p < 0.01

Tables 6 and 7 show a significant difference of vowel length percentage in the comparisons of parallel scripts of the 3 sentences’ groups. All evidence the consistent significant longer vowels 53

for the group of sentences with overt vowels ( p< 0.05 for the mono & multi syllable word sentences, p<0.01 for the pseudo words sentences). Table 7 adds the overall significant difference between syllabic script lacking overt vowel letters and their parallel sentences with overt vowel letters (both in HEBRISH and in IHVL). This is also clearly demonstrated in figure 4.

It should be pointed out that pseudo words maintained the gap between the parallel scripts but with higher significance than the one found in the two other groups of sentences (p<0.01versus p<0.05, respectively, see table 3). Also, there are lower mean values in pseudo words both in the IH script as well as in the HEBRISH one when compared to the two other groups.

Figure 4: Mean vowel length for sentences lacking VL presentation and sentences with overt VL presentation, measured for each subject, as well as overall mean vowel length for all 4 subjects

50.00% 45.00% 40.00% 35.00% 30.00% 25.00% lack VL 20.00% with VL 15.00% 10.00% 5.00% 0.00% actress c-singer l-singer teacher average

Following are spectrogram samples visualizing the speech production of compared orthographies.

54

Figure 5: 2 spectrograms of a 2 monosyllable pseudo words sentence (subject 3).

ויר טם - a: IH.5 [v------i------r------t------a------m------]

5.b: HEBRISH - VIR TEM

[v------i------r------t------e------m------]

In the IH script pertaining to figure 5.a., the vowel of the second word is not visualized. Because it is a non-recognized pseudo word lacking stored phonological information, it allows for 2 options according to Hebrew orthography, namely [e/ɛ] or [a]. All four subjects chose the [a] as default as 55

can be discerned in the spectrogram.42 Still, in spite of the fact that the vowel [a] is longer and higher in sonority than [e], figure 5.b demonstrates that the vowel [e /ɛ ] in the second word in HEBRISH (in the word /TEM/) is longer. Also, note that the stopping of the [t] is significantly shorter in HEBRISH. These lead to the HEBRISH version having more overtones across the sentence, enabled by its longer vowel.

Likewise, when comparing spectrograms 6.a. and 6.b., it can be clearly seen that 6.b. (HEBRISH version) is overall richer with overtones than 6.a. (IH version) due to longer vowels and shorter [t] stop burst in its 3 monosyllable pseudo words (in the sentence /PESH HEL MIT/).

42 The phoneme [a] is the most prominent vowel in Israeli Hebrew, having the highest sonority and frequency, and is the least marked phoneme in the five-vowel system. Consequently, [a] usually functions as the default choice (Bolozky, 1990, 1997). Also, [a] is the vowel which is both the simplest to produce as well as the most common across the world’s languages (Stemberger,1992). 56

Figure 6: 2 spectrograms of a 3 monosyllable pseudo words sentence (subject 4).

פש הל מיט - a. IH.6

6.b. HEBRISH – PESH HEL MIT

[p---a------B ------h--a------l------m------i------t------]

[p--e------B ------h----e------l------m------i------t------]

57

Figure 7: 2 spectrograms of a 4 IH multi syllable words sentence (subject 1) כל אב קבל משכרת - a: IH established syllabic orthography.7 כול איש קיבל משכורת - b: IHVL.7 Note: The second word is different in the two versions.43

k-----i--b—e-----l----m----a--s-----k------o-----r-----e-----t---

k---i----b—-e------l----m---a—---s------k------o------r----e------t---

representing the vowel [a], it could not be manipulated by /א/ AV = father) has the letter ALEF) אב As the word 43 adding an extra vowel letter. Therefore, it was replaced by another monosyllable word initiating with the letter ALEF ISH = person). Thus, the first version reads KOL AV KIBEL MASKORET (every) איש – that carries a similar meaning father got a salary) and the second reads KOL ISH KIBEL MASKORET (every man got a salary). 58

Figure 7 shows similar differences as in figures 5 and 6: as apposed to no vowel presentation in the IH version presented in 7.a, vowel letters presentation for most syllables in 7.b (i.e. 4 vowel letters out of 7 existing vowels were added ) result in longer vowels, shorter stops and more overtones reflecting the sonority of longer vowels.

II.2 Discussion

The primary results of the pilot study gave favorable results supporting the hypothesis underlying this dissertation, that the primarily consonant-based nature of IH orthography engenders, at least for LPVU, a consonant-emphasized IH diction, diminishing the length of vowel expression within an utterance. When an overall comparison is conducted between the two types of scripts within each group, all 4 subjects, regardless of age, gender and specific discipline, clearly shorten the vowel length in speech production when reading IH aloud, and enlarge it when reading aloud the HEBRISH script and the IH script with the presentation of expanded vowel letters.

Overall, the pilot study aimed at evaluating the quality of results and to identify problems in this first application of a new methodology, tool and innovative procedure, showed that presenting the same text with several scripts is very useful in showing online orthographic effects in terms of vowel to consonant ratio in the diction of PLVUs reading it aloud. However, the pilot results aroused the need for a more controlled address of some pertinent issues on which the results were inconclusive, and the controlled exploration of new issues that arose:

II.2.1 Vowel Letters in IH

Quite surprisingly, results showed that the effect of the linearly overt addition of Hebrew vowel letters to the IH multi-syllable sentences, i.e., IHVL script, replicated the effect found in the comparison between IH and HEBRISH scripts in the monosyllable sentences. However, while I assumed that compared to the established IH script, both the addition of Hebrew vowel letters to established IH, and the Latin vowel representation in the HEBRISH script will engender longer 59

vowel length in speech production, I expected the IHVL effect to be of less significance. This implicit assumption was based upon the fact that the HEBRISH script was complete and consistent in respect of vowel presentation, whereas the IHVL was still incomplete and inconsistent in this respect (though more than in the established historical IH syllabic script). While adding Hebrew letter vowels to an IH established script reduces the orthographic depth in reading aloud, Hebrew vowels in themselves are incomplete and inconsistent. This is because, as already mentioned, most Hebrew vowel graphemes also function as consonants, and because these vowel graphemes are used to denote more than 1 vowel (see table 4). Moreover, there is /חמוד/ ,tipeʃ], meaning- fool] /טיפש/,.no way to linearly represent the vowels [e] and [a] (e.g /ה/, /א/ xamud] meaning- cute). This excludes words ending with the silent graphemes] (more] for male and [mora] for female meaning- teacher] /מורה/ ,meaning- reads /קורא/ [e.g.,[kore) ʃma] meaning- listen!). However, in the latter case one] /שמע/) as well /ע/ and in some cases by, cannot differentiate between [a] and [e] without semantic redundancy. Finally, while the visual representation of HEBRISH Latin vowel letters is as prominent as that of consonants in terms of VAV representing [o] & [u]) are / ו/ YOD representing [i] and /י/) width, 2 vowel letters in Hebrew thinner and less prominent than the graphemes for the other Hebrew consonants.

Given the surprising results whereby there was no significant difference between the HEBRISH and IHVL scripts, a fact that may have implications on the evolution of IH orthography and IH diction in respect to vowel length (given the growing use of linear vowel additions in computer and cellular phone IH scripts),44 I decided to further explore this phenomenon in the main experiment that followed the Pilot.45

44 This may indicate the reverse influence of speech on orthography, since the growing addition of vowels to established IH in computer and cellular phone usage may stem from the writer's need to disambiguate the orthography. I will discuss this further in my concluding remarks.

45 I think that the reason for an overall expanded vowel length in the speech production of IH with added vowel representation, an expansion that goes even beyond the reduced inconsistency and incompleteness incurred by the addition of vowels and equaling that of HEBRISH, results from an impetus effect propelled by the further but still partial vowel representation. 60

II.2.2 Pseudo Words

Interestingly, the comparison of the two different script versions of the same sentences when using pseudo-words, showed a higher significance in respect of the vowel to consonant ratio than in the two other groups of sentences compared (p<0.01versus p<0.05, respectively, see table 6). This finding further supports my hypothesis that there is a strong online effect of orthographic characteristics in online speech processing when reading aloud, since in pseudo words there is no possibility of retrieval from stored word lexicons. Also, compared to other groups of sentences, the overall values of vowel length percentage are lower in the reading aloud when compared to the mono & multi-syllables' script versions (the IH and the HEBRISH script). I assume that this is due to the online screening and decoding process needed when reading unknown or unrecognized lexical material. Thus, when no online facilitation of retrieval from the lexicon is available (such as in real words), there is more hesitation, causing interruption of the speech airflow which mostly affects vowel expression, since vowels are the carriers of the airflow in a word or sentence. In any case, the pseudo word sentences analysis was conclusive in regards of the orthographic effect in the phonological process. Thus, given the redundancy that will most probably be found in any further experimentation with pseudo words, pseudo words were not used in the revised test.

II.2.3 Vowel Letters and Diacritics Additions

In the third part of the pilot study I used an IH script with added Hebrew vowels to assess the effect of the linearly overt addition of these vowel letters on vowel length speech production. As we have seen, the results showed that when such vowel letters are added, overall vowel elongation equals that of the elongation engendered by the HEBRISH script. However, it was unclear from this test whether the results stemmed solely from the addition of vowel letters, since prosodic and syllabic stress attendant to the sentence flow within which the words figured, could also affect vowel length. This issue was complicated by the fact that the letter vowels added still rendered an incomplete representation of the phonemic utterance (see the discussion in section II.2.1 in this respect). I therefore reverted again to the use of single words to neutralize possible influences of prosodic and syllabic stress, so as to better control the orthographic effect 61

stemming from vowel letter additions. I also addressed the complicating issue of completeness, by selecting words that offered a full phonemic representation through vowel additions to established IH, or some of the few Hebrew words that also in established IH offer full phonemic representation. Furthermore, concerning completeness of representation, I introduced alongside the IH and HEBRISH scripts, the diacritical marked IH script (table 1), which allows for a full phonemic representation of many words that cannot be fully represented by linear vowel letter peace - cannot be fully represented without diacritical marks - שׁלוׄם /additions (e.g. /SHALOM since there is no way to represent the [a] vowel by a Hebrew letter vowel addition).46

II.2.4 Multi-Syllable Words and Sentence Structure

In all parts of the pilot I used sentences in order to approximate connected speech. In connected speech however, syllabic stress leads to vowel length elongation,47 often complemented by vowel length shortening in the resulting non-stressed syllables of the word.48 This means that syllabic stress strongly influences vowel length in speech irrespective of orthography. This is why in 2 of the 3 pilot test parts I used only sentences composed of mono-syllable words (IH words as well as pseudo words) neutralizing syllabic stress. However, this selection was too distant from IH practiced connected speech because it left out the major portion of the Hebrew lexicon - multi- syllable words, where syllabic stress resides. Although in the pilot's third part I used some multi- syllable words, these were used again within sentences, to further approximate connected

46 The addition of the diacritically marked Hebrew script was also compared to the HEBRISH and established IH scripts, rendering interesting results in respect of spoken vowel length alteration.

47 Quantitative data shows the high frequency of final stress in Hebrew. Final stress refers to the right-aligned iambic foot to the ultimate syllable in a word, differing from the right-aligned trochaic foot to words with penultimate stress. The wide distribution of the iambic stress evidenced in the quantitative data led researchers to the conclusion that the iambic foot is the dominant foot in Hebrew (e.g. Adam & Bat El 2009; Graf & Ussishkin 2002; Bat El 1993). This results in the characteristic IH elongation of vowel length in ultimate syllables.

48 As shown by Bolozky & Schwarzwald (1990), in casual IH speech production the vowel length is shortened by assimilation or omitted altogether in unstressed syllables. However, they did not discuss the correlation of this phenomenon to the vowel length elongation in stressed syllables. 62

speech. But here again, the placing of multi-syllabic words within the flow of a sentence made it difficult to control the various aspects of stress in these words due to a mix of prosodic stress with syllabic stress. I therefore decided to include in the main test lists of representative multi-syllable words without placing these within sentences so as to neutralize prosodic stress and get a better control on the effects of syllabic stress. I also included new sentences containing a large amount of the multi-syllabic words controlled in the word list, so that I could better control syllabic stress within sentences including multi-syllable words.

ALEF Grapheme – Ambiguity Issues /א/ II.2.5 The

(ALEF) /א/ Throughout the pilot, in measuring vowel length, I found a recurring problem with the grapheme, defined by the Hebrew Language Academy as a dual functioning vowel/consonant grapheme, forming part of a distinct group of such graphemes that also includes the dual YOD).These graphemes, when articulated are considered) /י/ VAV), and) /ו/ ,(HEI) /ה/ functioning to be consonants, but when not articulated, i.e., silent, they serve as vowel signs. However, while as consonants are clearly distinct from the vowels they represent ,/י/ VAV), and) /ו/ ,(HEI) /ה/ as a consonant voiceless stop with the respective vowel, is only slightly ,/א/ when silent, the ALEF different, if at all, from the vowel it denotes (see table 8).49

49 ALEF, in fact, mostly functions as a vowel, figuring in any place of the written word and representing all IH vowels: ;(here) /כאן/ read),KAN) /קרא/ a] as in KARA].- ;(reads) /קורא/ e] as in KORE] - ;(first) /ראשון/ i] as in RISHON] - ;(head) /ראש/ o] as in ROSH] - a vowel letter representing [u] as well as [o]), in words such as) /ו/ u] when following the Hebrew vowel letter] - .(hated) /שנוא/ he) or SANU) /הוא/ HU ALEF also functions as a prostatic vowel changing the beginning of Hebrew words that previously began with a cluster of consonants, so as to ease their pronunciation. This addition occurred already in the Bible, where words ,/אכסניה/ etmol]. It is also found in borrowed words such as]/אתמול/ tmol] (yesterday) often figure as] /תמול/ such as .[itstrubal] /אצטרובל/ ,[askola] /אסכולה/ [axsania]

63

(ALEF, HEI, VAV, YOD) /א,ה,ו,י/ Table 8: The dual function of

The letter As consonant As vowel ,(ʔ a (most frequent א e/”, i, o, u ”/h a, e ה v o, u ו y I י

This ALEF vowel/consonant ambiguity results in an escalating inconsistency in its IH speech production.50 When articulated, the ALEF grapheme in fact sounds like vowels in other languages that are characterized by allophonic rather than phonemic glottal stops51, and therefore form part of the vowel length production irrespective of its orthographic representation (e.g. the German allophonic [ɂI] in "ich”, meaning I, or the French allophonic [ɂa] in “amour”, meaning love. Occasionally this also occurs in English as in I am [ɂaɘ ɂᴁm]). The above discussion strongly indicates the possible ambiguity in the way ALEF is perceived and therefore pronounced, an issue which is most relevant when focusing on the relation between orthography and speech, and specifically the vowel consonant relation in speech production with respect of orthography’s impact, the topic of this dissertation.

Thus, in the pilot experiment I decided in respect of this ambiguity, to include the unvoiced ALEF length (i.e. voiceless glottal stop), separating it from the vowel that followed, and calculated it as

50 Arabic, a related Semitic language that has lived continuously since ancient times may contribute to our /ﺎ/ or /ﺍ/ understanding of the ALEF phenomenon. In the past, the consonant syllabary of Arabic used the ALIF letter with a double function similar to the Hebrew ALEF. Due to confusion, in modern Arabic, the "hemzah" diacritical mark to signify a glottal stop, while the bear ALIF exclusively functions to signify an /ﺂ/ or /ﺃ/ was added to the ALIF elongated vowel. Given the IH user's confusion with the ALEF, the Arabic precedent may indicate a comparative change in IH orthography.

51 Allophonic glottal stop [Ɂ]: a distinct phonetic variant produced by obstructing airflow in the glottis – the contact area of the vocal cords. 64

part of the overall consonant length calculation. 52 However, the inconsistency in its voicing irrespective of its function as vowel or consonant, shed doubt on whether the ALEF as consonant produces a consistent effect pertaining to its orthography, or whether its occasional voicing should be controlled since it forms part of the allophonic effects upon speech production. In the main study I set out to further explore this phenomenon.

(ALEF) /א/ AYIN) and) /ע/ II.2.6 The Homophones

While ALEF traditionally and historically has been an ambiguous letter as discussed above, AYIN, historically, has been a pharyngeal voiced fricative consonant (whose articulation is perceived to be stop-like). Still, in IH speech production ALEF and AYIN have become homophones.53 Therefore, when arguing the ambiguity of vowel letters, the question rises whether the ambiguity characterizing the letter ALEF is also applicable to the letter AYIN. A possible solution to this question can be found in Yiddish orthography that for centuries served as an alternative vernacular language among western European Jews. Based on the Hebrew alphabet, the Yiddish written system strives for a transparent, shallow orthography. Therefore Yiddish developed extensive vowel sign orthography. For our concern, it uses the letter AYIN to represent the vowel [e] while ALEF represents the vowel [a]. This precedent, along with the observable homophony of ALEF and AYIN, led me to conclude that AYIN is ambiguous in the same sense as ALEF is. Consequently, I decided to treat AYIN as ALEF in the forming of the main test, resulting in respect to their function as vowels, in a revised, descriptive and non grammatical classification (see table 9).

52 The Hebrew Academy defines consonants as those letters forming part of a word root. In this respect the ALEF is to collect), and is) /אסף/ ,(to say) /אמר/ ,(to love) /אהב/ a consonant since it forms part of the root in words like henceforth defined as a glottal stop consonant. However, as my study suggests, it seems to be the case that due to the ALEF's dual function, its glottal stop characteristic may occasionally be allophonic rather than phonemic. Moreover, according to its functioning in the Bible (e.g. as prostatic vowel), it may have been that ALEF was considered originally as a vowel.

53 For the suggested historical reasons leading to this and other homophonies see the section on IH revival. 65

(AYIN, ALEF, HEI, VAV, YOD) /ע,א,ה,ו,י/ Table 9: The dual function of

The letter As consonant As vowel ʔ a, e/ɛ, i, o, u ע

,ʔ a (most frequent), e/ɛ א i, o, u H a, e/ɛ ה V o, u ו Y I י

66

III. The Main Test

The main test included refinements and adjustments stemming from the results of the pilot test discussed above. Also, several rounds of subjects’ recordings to evaluate the validity of various issues were conducted.

III.1 Test Material

The main test was designed as follows:

III.1.1 Isolated Words

The test material included 278 words. All words were chosen in light of the revisions required following the pilot test results and are all used by IH speakers with high-frequency. These words were presented in three parts:

a. The first part of the read-aloud words consisted of 115 IH syllabic orthography (IHS) words presented in 12 columns. Each column was read and recorded separately. An unrelated word (that was not analyzed) was added at the end of each column to neutralize participant's tendency towards an energy drop in the last word of a column, to get a full stop intonation pattern. The words included mono-syllable, two-syllable and three-syllable words. In order to avoid ambiguity in the case of words without vowel presentation, an accent mark was added when necessary so that all words could be read in their relevant intonation (with iambic or trochaic foot).

b. The second part consisted of 46 IH orthography words with added Vowel Letters (VL) that reached full phonemic presentation (IHVL). These words’ lists repeated those syllabic words of the first part that could be extended by vowel letter additions and were presented in 6 columns. Once more, and for the same reason (see above), at the end of each column an unrelated word was added. In this section there was no need for accent marks, as the added VL created a transparency and cancelled any possible ambiguity in intonation.

67

c. The third part presented 115 HEBRISH orthography words of the same list of words in IHS presented in part a (i.e. phonemic orthography in the Latin alphabet rendering a complete and consistent shallow orthography). These words were presented in the same order, same column distribution and with the addition of irrelevant words as in part a.

III.1.2 Sentences

11 sentences were presented in 3 scripts. The repeated sentences in their different scripts were separated from each other in order to avoid effects of familiarization or memorizing in the production of the different scripts. The sentences included 60% of the words used in the words lists. The repetition of words from the isolated word tests in the sentences tests was done in order to better observe whether there are inconsistencies in orthographic effects between the two types of tests. These sentences were presented in three parts:

a. Sentences in IH orthography (presented after the first part of words lists). 9 of the 11 sentences were overall syllabic and the remaining 2 sentences were presented in IHVL (sentences 6 and 9 - see appendix II, p.120). This was done in order to see whether the 2 IHVL sentences differ in vowel length production between the 3 scripts (see below) compared to IH sentences.

b. The same sentences were presented with the addition of diacritic signs to the IH script (IHD).54 As has been mentioned in the pilot discussion, I introduced alongside the IH and HEBRISH scripts the IHD script (table 1) because it allows for a full phonemic representation of many words that cannot be fully represented by linear vowel letter additions. Furthermore, the use of the IHD script, presenting a shallow orthography through vertically added diacritical marks, was used to observe whether it has different orthographic effects in respect of vowel length variation than the IHVL script which presents a shallow orthography through presentation of vowels linearly within the word.

c. The same sentences were presented in a HEBRISH script.

54 In the standardized IH with diacritics that is known to every IH LPVU speaker. 68

III.1.3 Free Speech

Free speech was recorded at the end of the test. The elicitation of free speech was through an answer to a question requesting reflection on the test. The same question enabled elicited free speech from a similar mental and emotional state, as well as some similarity in lexical choice. This was done examine whether vowel length variations could be observed with no external orthographic stimuli, which may indicate the stored orthographic templates55 activated in spontaneous speech production.

The full test appears in Appendix II.

III.2 Subjects

8 subjects participated in this study (table 10): To assess gender or age influences, 4 females and 4 males aged 20-55 were selected. All were IH LPVU native speakers: 2 teachers, 2 singers, 3 actors, 1 choir conductor. All subjects had at least 5 years of experience in their respective LPVU discipline.

55 Perre et al (2009) also conducted a speech-based experiment with no external orthographic stimuli to show that phonological representations are influenced by internalized orthographic effects. 69

Table 10: Subjects of the main experiment

No. Gender Age Profession 1 F 20 Actress 2 M 23 Singer 3 F 28 Singer 4 M 35 Actor 5 F 52 Actress 6 M 54 Teacher 7 F 55 Teacher 8 M 55 Singer & choir conductor

III.3 Experiment Procedure:

I presented the participants with the test material. Each subject was asked to read aloud the presented text in a moderate rhythm. Also, in the isolated words columns, each subject was asked to pause at the end of each column. This was done to enable short rests.

A pre-test tutoring and practice was carried to facilitate the reading of the HEBRISH script. This was done in order to avoid an English phonological influence, as well as to assure the right interpretation of phonemes in IH. This was repeated once more before the first encounter with the HEBRISH script in the test itself. No further information about the nature of the test was given to the participants.

70

III.4 Tools:

Recording of each subject's readings and spectrographic analysis of the recordings were carried on a Kay Pentax Multi Speech Lab. Model 3700 version 3.2.

III.5 Measurements:

Each phoneme was measured on the spectrogram in order to obtain the consonant vowel proportion in speech production. As in the pilot test, each phoneme was measured on the spectrogram in order to obtain the respective length of each consonant and each vowel produced. Spectrogram measuring was simultaneously supported by the audio input for maximum accuracy: The transit point for each phoneme was decided according to the co- articulation and assimilation processes characterizing each subject's speech production. In this respect transition was determined when the majority of consonant characteristics transitioned into the vowel’s formants and vice-versa. The transition point in consonants clusters in words alone was irrelevant for the ratio calculation which systematically included the overall length of consonants with respect of vowels in each word or sentence. Consistency was kept by immediately marking the same point in each of the same vowel/consonant transition in the two recorded utterances of the parallel orthographic versions. Whenever the visual spectrogram was ambiguous with respect of the transition from the consonant to the vowel or vice versa, audio input was used to disambiguate. Likewise, in all stop consonants, whenever followed by obstruent, non-phonemic bursts, these bursts were measured as part of the consonants. These border definitions were consistently implemented throughout the entire analysis.

The vowel to consonant ratio was calculated by extracting the overall vowels' length in each sentence from the overall length of the sentence utterance (as a result, the consonants' duration is the difference between the vowels' length and the length of the overall production).

71

III.6 Results:

For all measurements and results presented, the dependent variable was the overall vowel length portion in a word, in relation to the whole word itself as measured on spectrograms.

III.6.1 Comparison between IH words in IHS and HEBRISH scripts:

Descriptive statistics of established IHS words and their HEBRISH counterparts (1, 2, 3 syllables in a word), as well as paired t-tests comparing the respective groups are presented in table 11. Word examples are presented in List A.

List A: 4 word examples for Table 11

SYLLABIC HEBRISH HEBREW KOL כל 1

SE FER ס̂פר 2

MISKEN מסכן 3

TILBOSHET תלבשת 4

MISTAKEL מסתכל 5

72

Table 11: Comparison of words in IHS and in HEBRISH scripts

List A HIS HEBRISH T value Mean SD Mean SD

1, 2, 3, 4, Overall syllabic words .35 .03 .39 .02 9.73*** 5

1 Mono syllable words .35 .03 .38 .03 3.72*** 2 2 syllable words with .35 .05 .39 .03 5.36*** trochaic foot 3 2 syllable words with .34 .03 .37 .02 9.50*** iambic foot 4 3 syllable words with ..40 .05 .43 .05 3.65** trochaic foot in 2nd syllable 5 3 syllable words with .35 .04 .35 .02 0.95 iambic foot in 2nd syllable

** p=0.001 *** p<0.001

Note that the groups of 2 and 3 syllables words are divided into words with trochaic foot (Mil’el) and words with iambic foot (Milra).

As can be seen in table 11, overall vowel length of words in HEBRISH is significantly longer than in IHS orthography, suggesting the immediate effect of the visualized vowels. This excludes the 3 syllable words with iambic foot which showed no difference of vowel length in both scripts.

III.6.2 Comparison Between IH Words in IHVL and HEBRISH Scripts.

Descriptive statistics of IH words in IHVL script, reaching full phonemic presentation, and the same words in HEBRISH script (1 or 2 syllables in a word) as well as paired t-tests comparing the respective groups are presented in table 12. Note in table 12 that there is a sub-group (1) of ALEF within a word as silent VL. Also, the two syllables words are/ א words containing the letter divided into the sub groups: a) Iambic foot; b) Iambic foot with the 2nd syllable CV (open) (i.e. ALEF, HEI); and c) Trochaic foot, with the 2nd / א, ה/ when presented orthographically by an syllable open as in b. Examples of words used are presented in List B.

73

List B: 5 word examples for Table 12

HEBREW HEBRISH

KAN כאן 1

KOL כול 2 SHO T שוברות 3 SIML שימלה 4 Y FI יופי 5

Table 12: Comparison of words in IHVL and in HEBRISH scripts

List B IHVL HEBRISH T value

Mean SD Mean SD 1, 2, 3, Overall .40 .04 .40 .03 0.56

4, 5

*ALEF) within the word .35 .05 .34 .04 2.20) א 1 2 Mono syllable words .35 .04 .35 .02 0.21

3 2 syllable words with iambic .38 .05 .37 .05 1.92* foot 4 2 syllable words with iambic .48 .04 .49 .04 0.92 foot, 2nd syllable CV (open) 5 2 syllable words with trochaic .54 .05 .53 .03 0.16 foot, 2nd syllable CV (open)

* P<0.05

Table 12 shows significant differences (p<0.05) between the Hebrew and the HEBRISH in words ALEF) as a silent VL (1), as well as in the 2 syllables words with iambic foot (3). In) א// containing these measures the vowel length in IH is slightly longer than in HEBRISH. No significant differences were found with respect of the overall comparison, as well as in groups 2, 4 and 5. 74

III.6.3 Comparison of IHS Words in 3 different scripts (IHS, IHVL, HEBRISH)

This test (table 13) compared between: a) IH syllabic words that can be manipulated by adding vowel letters [i, o, u] with either iambic or trochaic foot; b) the same words with the actual addition of VL on stressed/non stressed syllables (rendering in monosyllable words full phonemic presentation (G=P) and in words with 2 or 3 syllables a lesser phonemic presentation); and c) a HEBRISH script of the same words. Note that groups 1, 2 and 3 contain words of VL addition in stressed syllables, whereas group 4 includes words of VL addition in non stressed syllables.

Word examples are presented in list C.

List C: Examples for Table 13:

IHS IH+VL HEBRISH K L כול כל 1 SH RESH שורש שרש 2 MASK RET משכורת משכרת 3 MU X מוכרח מכרח 4

One way analysis of variance with repeated measures was used to compare the three scripts. Descriptive statistics and results of the comparison are presented in table 13.

75

Table 13: Comparison of IHS words IHVL and in HEBRISH scripts

List C Measure Mean SD F(2,14) Comparison

1, 2, 3, Overall 4

I IHS with possibility for VL .35 .03 5.97** I < II, III addition II IHS with addition of VL .38 .04 III HEBRISH parallels to I .38 .02 1, 2, 3 VL addition in stressed syllable

I IHS of 1, 2, 3 syllables .37 .03 4.11* I < II, III with possibility for VL addition in stressed syllable II IHS of 1, 2, 3 syllables .39 .04 with addition of VL in stressed syllable III HEBRISH parallels to I .39 .03 4 VL addition in non stressed syllable I IHS words of 2 syllables .33 .03 2.68 I=II=III with possibility of VL addition in non stressed syllable II IHS words of 2 syllables .36 .05 with addition of VL in non stressed syllable III HEBRISH parallels to I .36 .02

* P<0.05 ** p<0.01

Table 13 shows the overall significance of the difference between vowel length in IHS and the 2 other, more transparent script versions. Note that that both transparent versions have significantly higher vowel length values compared to the IHS. This is in accordance with the results presented separately in tables 11 &12. VL additions in non stressed syllables show a 76

mean difference between scripts but with no significance. On the other hand, the contribution to vowel length of an added VL to vowel length in a stressed syllable is significant.(these results in respect of stress factor will be discussed in the next chapter).

III.6.4 Comparison of IH Words Containing 1 VL in 3 different scripts (IH, IHVL and HEBRISH)

This test compared between: a) partial syllabic words (containing one VL in a stressed syllable as part of their established script) that can be manipulated by adding VL; b) words with an additional VL that reach a full phonemic presentation with a 1/1 G=P relation for the 2 syllable words; and c) the same list of words in a HEBRISH script version.

One-way analysis of variance with repeated measures was used for the comparison. The results of the overall words comparison (2 & 3 syllable words), as well as of the sub groups, namely, 2 syllable words and 3 syllable words, are presented in table 14. Examples are presented in list D.

List D: Examples of words used for Table 14

Hebrew (with 1 Hebrew (with 1 HEBRISH VL) VL)+1VL SIML שימלה שמלה 1 IKA N זיכרון זכרון 2

77

Table 14: Comparison of IHS words with 1 VL to IHVL (reaching full phonemic presentation) and HEBRISH scripts. Measure Mean SD F (2, 14) Comparison

1,2 I Overall 2 & 3 syllable words .40 .05 5.45* I < II, III that contain 1 VL with possibility of addition of 1 VL

II Overall 2 & 3 syllable words .42 .05 that contain 1 VL with addition of 1 VL III The HEBRISH parallel for I .42 .05

1 2 syllable words

I 2 syllable words that contain .39 .05 5.83* I < II, III 1 VL with possibility of addition of 1 VL

II 2 syllable words that contain .41 .05 1 VL with addition of 1 VL

III The HEBRISH parallel for I .41 .05

2 3 syllable words

I 3 syllable words that contain .43 .07 0.74 I=II=III 1 VL with possibility of addition of 1 VL

II 3 syllable words that contain .44225 .07 1 VL with addition of 1 VL

III The HEBRISH parallel for I .44313 .05

* P<0.05 78

Table 14 shows the overall significant difference of mean vowel length between words in partial syllabic script and the other two versions of transparent orthography. Furthermore, when a VL is added to 3 syllable words (the word still remaining partially syllabic in the IHVL version), there is no mean vowel length significance between the groups.

( ALEF/AYIN) א/ע III.6.5 Words Beginning with

Paired t-tests were used to compare the IH orthography and the HEBRISH script in words ALEF/AYIN), presumed to represent a vowel letter. Results are presented in) ע/א beginning with table 15. Examples of words used are presented in List E.

List E: 3 Words beginning with ALEF/AYIN and their HEBRISH written counterparts

Hebrew HEBRISH ER ער 1 BA אבא 2 ISH אשה 3

Table 15: Words beginning with ALEF/AYIN in IH compared to the HEBRISH script.

Hebrew HEBRISH T value ALEF/AYIN Mean SD Mean SD

presumed to represent VL 1, 2 1 & 2 penultimate .59 .04 .59 .04 .954 stressed syllable words

3 2 ultimate .55 .05 .53 .04 .106 stressed syllable words 1, 2, 3 Overall words .57 .04 .55 .03 .069 79

As can be seen in table 15, no significant differences were found between IH and HEBRISH orthographies in the overall list of words as well as in the subgroups, with a slight non significant higher value for the IH script. This supports my assumption that in these words, when no VL is added, ALEF or AYIN seems to be perceived as VL.

In order to further test this assumption, a two way analysis of variance with repeated measures was used: The difference between words beginning with ALEF/AYIN and their parallel HEBRISH orthography, were compared to the difference between their minimal pairs56 (MP) in IHS words and in a matched HEBRISH script. This is because comparing the target words to their minimal pairs may add to our understanding of whether ALEF or AYIN are perceived as vowels when not succeeded by an additional VL: If the comparison to the HEBRISH script differs in such words from their minimal pairs with respect of vowel length, their “role” as a vowel could be confirmed. As one way analysis and paired t tests of these groups were already presented above, only the results of the interaction effect are presented in table 16 (mono syllable words), table 17 (2 syllable words), and table 18 (overall words). Examples of words used for the latter 3 tables are presented in List F.

List F: Examples of words used that begin with ALEF/AYIN, compared to their minimal pairs, each with matched HEBRISH script.

Hebrew HEBRISH Hebrew HEBRISH (MP) (MP) NER נר ER ער 1 S BA סבא BA אבא 2

in Hebrew (קל/טל) Two words differing in one phonemic feature only, such as buy/guy in English or Tal/kal 56

80

Table 16: Interaction effect measured for mono-syllable words initiating with ALEF/AYIN in two scripts, compared to their minimal-pairs scripts.

Mean SD Mean SD F(1,7) 1 1syllable words .50 .04 Mono Syllable .35 .04 intiating with Minimal pair ALEF/AYIN *.015 HEBRISH Orth. .48 .04 HEBRISH Orth. .38 .03

* P<0.05

Table 17: Interaction effect measured for two-syllable words initiating with ALEF/AYIN in two scripts, compared to their minimal-pairs scripts.

Mean SD Mean SD F(1,7)

2 2 syllable .61 .05 2 syllable .40 .07 words initiating minimal pairs

with ALEF/AYIN .054 HEBRISH .61 .05 HEBRISH .44 .04 Orth. Orth.

81

Table18: Interaction effect measured for the vowel length difference between words initializing with ALEF/AYIN and their HEBRISH script compared to the vowel length difference measured between their minimal pairs and their matched HEBRISH script.

Mean SD Mean SD F(1,7) 1, 2 1 & 2 syllable .56 .03 1 & 2 syllable .38 .05 words minimal pairs initializing with ALEF/AYIN **.009 HEBRISH .55 .02 HEBRISH .42 .03 Orth. Orth.

** p<0.01

As can be seen from the 3 tables above, all show that there is a significant interaction effect measured by the difference between the 2 scripts of the target words (i.e. words initializing with ALEF or AYIN) compared to the difference between their minimal pairs presented in their 2 scripts: Whereas vowel length of target words are almost equal in the two scripts, their minimal pairs show a significant higher vowel length in HEBRISH compared to IH (syllabic or partially syllabic).

Also note that the subgroup of two-syllable words (table 17) demonstrates a clear tendency towards significant difference, and yet, the interaction effect significance is borderline (p=0.054).

are perceived differently when followed by a /ע/ or AYIN /א/ Finally, in order to see whether ALEF vowel letter, a two way analysis of variance with repeated measures was used to compare words beginning with ALEF or AYIN followed by a VL, to their minimal IHVL paired words reaching full phonemic representation. Once again, only the results of the interaction effects are presented in table 19. Examples of words used are presented in List G. 82

List G: Examples of words used, that begin with ALEF/AYIN followed by a vowel letter, and their minimal pairs, each with matched HEBRISH script

IH HEBRISH IH (MP) HEBRISH (MP) BOR בור OR אור 1 GISH גישה ISH אישה 2 Y FI יופי FI אופי 3 SHO T שוברות O T עוברות 4

Table 19: Interaction effect measured for the vowel length difference between words initializing with ALEF/AYIN followed by a vowel letter and their HEBRISH script compared to the vowel length difference measured between their minimal pairs and their matched HEBRISH script.

Mean SD Mean SD F(1,7) 1, 2, 3 syllable .56 .04 1, 2, 3 syllable minimal .42 .04 words initiating pairs 1/1 grapheme- with ALEF/AYIN phoneme

.782 HEBRISH Orth. .56 .03 HEBRISH Orth. .41 .03

Results presented in table 19 show no interaction effect between groups. There is no difference in vowel length between orthographies of words having full phonemic presentation (as has been also shown in table 11 for the overall words within the test). Also, target words, namely those beginning with ALEF or AYIN, also show no vowel length difference between orthographies, even when more graphemes form the IH script than the number of graphemes in the counterpart HEBRISH script. 83

III.6.6 Comparisons of sentences in IH, IHD and in HEBRISH Orthographies

11 sentences in 3 different orthographies interspersed within words lists were read aloud by the subjects. Due to material quality problems in one of the sentences for som of the subjects, 10 out of 11 sentences were included in the analysis. Results for the descriptive statistics and pair wise comparisons analysis for 8 IHS sentences in the 3 different versions used, are presented in tables 20 and 21. Mean and SD for the words’ vowel length is presented in table 20, whereas the pair wise comparison analysis is presented in table 21. An example of 1 sentence in the different scripts is presented in List H.

List H: 1 IHS sentence in three scripts

IHS = כל אב קבל משכרת. IHD = כֺּל ָאב ׅק בּל מ שֽׂכֺּרת. (every father KOL A KIB L MASK RET = HEBRISH received a salary)

Table 20: Descriptive statistics of 8 sentences presented in three different scripts:

Orthography Mean SD IHS .38 .04

IHD .40 .04

HEBRISH .42 .03

84

Table 21: Pair wise comparisons analysis of 8 sentences in three different scripts.

IHD HEBRISH

IHS ***.001 **.003

IHD .216

** p<0.01 *** p<0.001

As can be seen in tables 20 and 21, there is a significant difference in vowel length between sentences written in IHS and the same sentences when written in IHD (with diacritic marks) when compared to the HEBRISH script. No significant difference in overall vowel length was found between both transparent orthographies, IHD and HEBRISH. These results support the overall results obtained from the words production analysis.

III.6.7 (IHVL) Full Script Sentences Compared to IHD and HEBRISH Orthographies

This comparison was carried out in order to see whether the same tendency can be observed in the comparison of 3 scripts containing full phonemic representation, namely, IHVL, IHVL + diacritics and HEBRISH as well as to estimate the effect of diacritics addition in those sentences compared to this addition in the IHS sentences.

Results for the descriptive statistics and pair wise comparison analysis of 2 sentences written in full phonemic representation in three different versions (IHVL, IHVL+ Diacritics and HEBRISH) are presented in table 22 and 23. Mean and SD of the vowel length are presented in table 22. The pair wise comparison analysis is presented in table 23. Example of one sentence in its three versions is presented in list I.

85

List I: IHVL sentence in three scripts

IHVL = היא לובשת שימלה ארוכה. ׅ = IHVL+ Diacriticsהיא לוֹב שׁ ת שׂ ימֽלה א רּוכּה. HI LO SHET SIML A UK = HEBRISH (She is wearing a long dress)

Table 22: Descriptive statistics of 2 sentences containing IHVL words (i.e. full script) presented in 3 different orthographic scripts

Orthography Mean SD IHVL .54 .05

IHVL+Diacritics .52 .04

HEBRISH .53 .06

Table 23: Pair wise comparisons analysis of 2 sentences containing IHVL words (i.e. full script) presented in 3 different orthographic scripts

IHVL+Diacritics HEBRISH

IHVL .338 .1.000

IHVL+Diacritics 1.000

Results show no siginifcant difference in overall vowel length in the 2 sentences presented in 3 different transparent orthographies. These results support the results obtained for IHVL stand alone words. Furthermore, when comparing the mean vowel length (table 23), there is no contribution of diacritics to IHVL orthography while the IHVL reaches the highst value of vowel length compared to the 2 other scripts.

86

III.6.8 Free Speech Analysis

In order to verify whether vowel length variations could be observed with no external orthographic stimuli, indicating the stored orthographic templates activated in spontaneous speech production, free speech analysis was conducted.

CVC syllables were obtained from an excerpt/sample of 1 minute free speech of the subjects when responding to the same question. CVC IHS syllables according to IH established orthography were compared to CVC IH syllables with vowel letters according to established IH57 (thus with transparent/full phonemic representation). Both types of CVC used are habitually stressed by IH speakers whether due to syllabic or prosodic stress. The syllables types chosen were equal in number and were the most frequently used in all speech samples. The stressed factor assured the highly equal effort invested in their voicing in respect of the acoustic parameters of duration, intensity and pitch.

Results of the free speech analysis are presented in table 24 and figure 8:

Table 24: CVC stressed syllables in Free Speech

CVC Syllabic CVC with VL Mean Vowel length 0.334 0.419

SD 0.008 0.013

T-test P<0.001

.etc ,(נכון) x-o-n in the word NAXON ,(קיר) Example for an originally CVC syllable with vowel letter: k-i-r 57 87

Figure 8: CVC stressed syllables in free speech

Free Speech

50.00% 42% 40.00% 33%

30.00%

20.00% Vowel length

10.00%

0.00%

CVC SYL CVC VL

As can be seen in t table 24 and figure 8, there is a significant difference (p<0.01) in vowel length between the CVC IHS syllables and the CVC syllables containing a vowel letter (VL). These results conform in terms of orthographic influence on vowel length production to the results obtained in the reading aloud test, and further imply the option that stored orthographic templates are activated with no external orthographic stimuli.

88

IV. Discussion

The main concern of this study was to examine how Hebrew orthography -- a fundamentally consonant-based syllabary -- influences the speech production of Literate IH speaking adults. This was based on the hypothesis that the result of the primarily consonant-based nature of written IH has effects on the diction of literates in terms of vowel and consonant length variation. This may also suggest that at least for highly literates, the consonant-based nature of spoken IH leads them to put extra emphasis on the consonants at the expense of vowel production.

To my knowledge, this is the first study of its kind in the field of the bi-directional connection between spoken and written Hebrew. Therefore, one of the objectives underlying this study was to design an appropriate methodology to enable the examination of this new approach.

The test designed aimed at evaluating the online orthographic effects in terms of vowel to consonant ratio in the diction of LPVUs. This was done through three different tests: a) Each subject was asked to read aloud the same stand alone words in three different orthographies: IHS, IHVL, and HEBRISH. b) Each subject was asked to read aloud the same sentences in three different orthographies: IHS, IHD and HEBRISH. c) Each subject was asked to respond to the same question in free speech.

In the words alone test and in the sentences test eliciting connected speech production, analysis of the subjects' speech validated the hypothesis, which was further confirmed in the free speech test. Moreover, these results were obtained despite the possible influence of recall from one part of the test to successive ones rendered in different scripts.

IV.1 Orthographic Effects on Vowel Length Variation

The results presented in tables 11 & 12, along with figure 9 show a significant online influence of the reading material on speech production. When the reading aloud of the HEBRISH script and of the IHS script are compared, the vowels’ length within the words themselves is significantly 89

longer in the reading aloud of the HEBRISH script (p<0.001). These results support the main hypothesis that the overall primarily consonant-based nature of written IH engenders, at least in highly literate adults, a consonant-based spoken IH, resulting from diminished vowel length production.

IV.2 Vowel Length Variation within Different IH Scripts

While IHS words showed a significant lower mean vowel length when compared to their HEBRISH counterparts, when IHVL words are presented, rendering a full phonemic presentation, the difference in vowel length between these words and their HEBRISH counterparts disappears (Fig. 9). This indicates that vowel length in IH production also varies according to the presence or absence of IH vowel letters, with vowel length lengthened when vowel letters are presented.

Figure 9: Words in IHS and IHVL scripts compared to their HEBRISH script.

HEBREW HEBRISH

40.09% 39.94%

38.50%

35.00%

EQ. REP. SYL

EQ. REP = equal full phonemic presentation,. SYL= syllabic words in IHS

It should be noted that in established IH orthography there are fewer words containing vowels כתב YOFi=beauty), than there are IHS words such as) יופי presented linearly (IHVL), such as (KATAV=wrote). This is due to the rich morphological consonant root-system of IH. Nevertheless, given the potential for IH to approximate a full phonemic representation in IHS words by linearly adding vowel letters, and given the growing use of these vowel letter additions in computer 90

written IH, internet and SMS evidencing the reactive influence of spoken IH on IH's orthographic depth. This gradual change, particularly given the gradual approval by the,58 probably indicates the language user's need to overcome IH deep orthography's ambiguity by closing the gap between spoken and written IH. Further research in this respect might provide insights into the process of the bi-directional influence between written and spoken language in IH and in other languages.

IV.3 Effects of Internalized Visual Templates on Speech Production

The results obtained from the reading aloud tests, and particularly those obtained from the free speech analysis, may be taken to support the hypothesis presented by Katz & Frost (1992). Their prediction, tested through varying response latencies to manipulations of auditory and written input, was that deep orthography encourages readers to process printed words by referring to morphology via the printed word’s visual-orthographic structure. This addresses them to a different mental lexicon than the phonological one acquired with language at an early pre-literate age. The experienced reader gradually becomes more dependent on an internalized visual input consisting of mentally stored orthographic templates acquired with written language acquisition. Moreover, highly frequent orthographic words become primed visual templates. Frost, in a complementary study, argued that IH readers, due to the ambiguity stemming from IH's deep orthography, process isolated words by priming the computing of morphological word roots which in IH mostly consist of consonant-only clusters, subsequently adjusting the vowels into this consonantal root (Frost 1995, 2006). When taken together, these studies suggest that IH experienced readers prime mentally stored consonants because of frequent exposure to IH's consonant-based syllabary, and based on the finding that they compute consonants first and vowels second when reading IH. My results above support these findings concerning IH perception and expand them to include the priming of consonants in speech production. Furthermore, the results obtained through my analysis of the subjects' free speech (table 24 and

58 For further information, see http://hebrew- academy.huji.ac.il/hahlatot/MissingVocalizationSpelling/Pages/ikkar.aspx (also: "כללי הכתיב חסר הניקוד ,"לשוננו לעם ,טבת התשס"ב(.

91

figure 9), support the preliminary contention that experienced readers have orthographic mind templates. This is because the comparison between orthographically established CVC stressed syllables with or without vowel presentation (table 24), indicating that vowel length varies according to orthographic primacy, could only be attained due to stored orthographic templates for there was no external orthographic stimuli.59 The nature of the influence of literacy on free speech should be further researched by comparing the speech of highly literate typical native speakers to the speech of others such as dyslectics, immigrants, bilinguals, and the hearing impaired.

IV.4 The Interplay Between Orthographic and Prosodic Effects

Beyond the overall supporting results obtained, focusing on the results' breakdown in the different sub-groupings of the various tests provides insights into the interplay between orthographic and prosodic effects in respect of vowel length variations.

IV.4.1 Word Length and Syllabic Stress Factors / Hesitation as a Factor in IHVL vs. HEBRISH Scripts

In respect to the possible effects on vowel length production stemming from word length variations (words with 1, 2, or 3 syllables) or stress variations (iambic or trochaic), results show that overall, these do not have effects on vowel length variation. This is because in most instances vowels were lengthened when reading the HEBRISH script relative to the IHS script. That is, vowel production was lengthier in reading the HEBRISH script than when reading the IHS script irrespective of whether the word used consisted of 1, 2, or 3 syllables. Likewise, vowel production was lengthier in reading the HEBRISH script than when reading the IHS script irrespective of whether the words carry an iambic or a trochaic stress. Had the case been that

59 Perre et al (2009) also conducted a speech-based experiment with no external orthographic stimuli to show that phonological representations are influenced by internalized orthographic effects. 92

one of the type of words used behaved differently than the others (e.g. a two syllable word with trochaic stress would render an anomalous equal vowel length production from the reading of HEBRISH or IHS scripts), then we could have said that vowel length or stress effect vowel length production and overrule the effect of vowel presence or absence from the orthographic script. Since no such difference was observed, it can be said that vowel length varied according to the vowel's orthographic presentation or non-presentation, irrespective of word length or stress (see table 11). An exception is the vowel production of 3 syllable words: While those with trochaic stress showed a significant difference in vowel length, elongated in HEBRISH when compared to its IHS (p<0.01), the 3 syllable words with Iambic stress showed no difference of vowel length in both scripts. This might be explained by the possible hesitation of readers when decoding a longer word in the less familiar HEBRISH script. In the HEBRISH words presentation, stress was marked to facilitate the decoding process of the IH words (this was done in the IHS words only for those words that might be interpreted in more than 1 way, which was not the case in this specific group of words). The stress mark appearing in the second syllable of the words with trochaic foot (i.e. penultimate stress) may have enabled an easier decoding process of these words as opposed to words with iambic foot in which stress was marked on the final syllable (i.e. ultimate stress). As has been shown (e.g. Katarina et al, 2001, Munson et al, 2004; Munson, 2007), hesitation effects (disturbs prosody), that may result in vowel characteristics and duration such as the shortening of vowel length production since prosody is carried by the airstream provided in speech by vowel production. This suggested explanation for the described phenomenon requires further research.

IV.4.2 Hesitation as a Factor in Established and non Established IHVL Scripts

As mentioned, results overall show the same vowel length elongation when IHVL and HEBRISH scripts of the same words are compared to the IHS script of these words. Interestingly, in words and in 2 syllable words with ultimate stress (i.e. SHO T (כאן / with ALEF within them (i.e. KAN results show a little more vowel elongation in the IHVL script than in the HEBRISH ,(שוברות / script, than the vowel elongation found in the other IHVL words used. I presume this is because the presented vowels in these specific words were not added for manipulation, but are present in established IH orthography and are always written with the vowels present. Beyond their habitual appearance, it seems that the mental templates of the specific established IHVL words used, 93

have been internalized as opposed to the lesser familiarity with the vowel additions to syllabic words. Therefore I suggest that non-established IHVL words are more hesitantly read in respect of these vowel additions. This hesitancy, as explained above, may shorten vowel length.

IV.4.3 The Stress-in-Words Dynamics Factor

Further support to the overruling effect of transparent orthography (IHVL and HEBRISH) on vowel length within a word, is well demonstrated in table 13 (p<0.01 for IHVL and p<0.05 for HEBRISH - in comparison to IHS). Once again, when vowel letters are added to an original IHS word, or when vowels are originally present in an established IH word, a significant change is observed in the mean vowel length in comparison to IHS words (i.e. with no vowel presentation). However, an exceptional result in this respect concerns the subgroup of words with a non- stressed syllable, where the presence or absence of a vowel letter is of no consequence to vowel length production (table 13/4). This may result from the dynamics of prosody, which in this respect are generated by the dynamics of stress in a word. That is, it seems that the shortening of vowel length in a non-stressed syllable (irrespective of whether the vowel is presented or not) is enhanced by the stress derived elongation of vowel length in an adjacent syllable (and the latter's overall prominence in prosody). This hypothesis requires further research.

IV.4.4 Orthographic Factors Influencing ALEF/AYIN

As has been discussed previously, within the special IH group of letters with a dual the ALEF, along with its ,(/י/ YOD ,/ו/ VAV ,/ה/ HEY ,/א/ consonant/vowel function (ALEF presented inconclusive results in the Pilot test as to whether they ,/ע/ homophonous letter AYIN were produced in speech as a vowel or as a consonant. In the main test I decided to look deeper into this apparent inconsistency, asking myself whether there is a consistency in some uses of the glottal stop (often indicating a consonant) that stems from orthographic effects, or whether glottal stops are invariably allophonic in these letters, therefore requiring a better control than the one used in the pilot test. I did this by comparing words beginning with ALEF and AYIN to their rav]). Taken together, tables 15, 16, 17]-רב/ɂ)av] compared to RAV)]-אב/minimal pairs (i.e. AV and 18 show a consistent orthographic effect: the first test (table 15) shows that all the target 94

words (beginning with ALEF/AYIN) are vowels (with the occasional glottal stop stemming from an allophonic influence60). It results from a significant (p<0.01) difference between IH target words compared to their HEBRISH version. Given that a) in words beginning with ALEF that hold the ɂ)av]) the vowel length)] אב/same amount of graphemes as their HEBRISH counterpart (e.g. AV is equal, and b) given that the same phonemic vowel presentation as that of the 2 grapheme word beginning with ALEF and that of a 2 grapheme IHS word beginning with a consonant (e.g. renders a different vowel length when the IHS 2 grapheme word is compared to (רב/RAV אב/AV .c) it turns out that in case (a) the ALEF (e.g ,(רב/its 3 grapheme HEBRISH word (e.g. RAV is a vowel (for if it were a consonant it would have had a different vowel length as is the (אב/AV case with the described 2 grapheme IHS word when compared to its HEBRISH 3 grapheme counterpart). Furthermore, the comparison of target words to their minimal pairs (tables 16, 17, 18) replicated these findings: while there are no differences in vowel length between the target words that begin with ALEF and their equal grapheme number HEBRISH counterparts, the minimal pairs, consisting of more vowels and graphemes in the HEBRISH version, show the expected significant difference between IH syllabic or partially syllabic scripts and their parallel HEBRISH script. This may suggest again that the letter ALEF was processed as a vowel in this words’ context (i.e. an ALEF letter followed by a consonant).

However, the following test (table 19) offers a different interpretation of the ALEF/AYIN in suggesting that whenever ALEF/AYIN were followed by a vowel letter, its function is that of a consonant. In this test, the difference between the IH and the HEBRISH script of words beginning with ALEF/AYIN and followed by a vowel letter (in this case only the phonemes /i, o and u/ can be used), was compared to the difference obtained from the parallel scripts of IHVL minimal pairs .(BOR/בור - OR/אור .e.g)

60 When measuring vowel length and excluding the glottal stop duration in occasional cases, no significant difference was observed in the comparison between scripts. Because in the HEBRISH script it was considered as an (since not preceded by a consonant), it was decided to keep consistency in measurements and whenever this was the case, it was measured as part of the vowel duration. 95

Results show no interaction effect between groups: target words, namely words initiating with ALEF or AYIN, show no difference between orthographies even when having one less grapheme OR) as do the minimal pairs which have an equal number of graphemes in/אור .in HEBRISH (e.g BOR). The production of target words should have shown a higher mean/בור .both scripts (e.g vowel length if target letters took the “role” of a vowel, resulting in two succsesive vowels (e.g. OR) but this was not the case. Consequently there was no interaction effect. Therefore, these/אור results suggest that the target letters function as consonants when followed by a vowel. This requires further research however, given that all the cases of ALEF/AYIN followed by a vowel that can be studied, always appear in non-stresed syllables. Since, as has already been shown (table 13), non-stressed syllables with vowel inclusion do not lengthen the vowel, the reason for the vowel shortening in the ALEF/AYIN followed by a vowel cases may not be due to the ALEF/AYIN being a consonant but rather to the fact that the syllable is non-stressed.

IV.4.5 Sentences in Script Variations

To see whether the results found in isolated words recur in read aloud sentences (presenting connected speech) I included in the sentences used 60% of the words from the stand alone words tests, and presented the sentences in three scripts (IH, IHD and HEBRISH). I used an IHD script to see whether its shallow orthography differs in terms of vowel-length variation from that of its HEBRISH script counterpart, given that IHD's shallowness is presented through vertically added diacritical marks whereas the HEBRISH script's shallow representation is through a linear consonant vowel sequence in words. Results (tables 20, 21, 22, 23) confirmed that overall, a transparent/shallow orthography, whether IHD, IHVL or HEBRISH, elongate vowel length when compared to an IHS script. Furthermore, this occurs in both isolated words tests and sentences tests.

This can be observed in the gradual growing mean value of vowel length within sentences in the three different groups, IHS, IHD and HEBRISH (38%, 40%, 42% respectively - see table 20). The pair wise comparison showed significance in respect of mean vowel length between IHS and the two transparent orthographies (IHD and HEBRISH) whereas the mean vowel length difference between IHD and HEBRISH was found to be insignificant in respect of vowel length (table 21). This countered my expectation since I predicted that an IHD script would present a significantly 96

shorter vowel length compared to the HEBRISH script, given their differing forms of presenting a shallow orthography. Likewise, In the two sentences written in IHVL and in IHVL with added diacritics (tables 22, 23) there was no significant effect on vowel length. This confirmed my prediction that IHVL, although offering a slightly lesser phonemic presentation than when it includes diacritics, is perceived as shallow as when diacritics are included. This supports the results obtained in the Pilot study (tables 5, 6, 7) in respect of IHVL.

Notwithstanding slight and insignificant variations, the overall results obtained strongly confirm the hypothesis concerning the influence of orthography on diction.

97

V Conclusions

The current study demonstrated that the qualities of diction, particularly the vowel temporal portion in speech production in IH, can be manipulated by changing the orthographic presentation of the text employed as a stimulus for voiced reading. The results also indicate a direct impact of IHS orthographic characteristics on the diction of IH literate speakers. In this script, characterized by the lack of representation of vowels, the diction of the literate native speaker, and particularly that of the professional speech and voice user, is characterized by a consonant-based spoken Hebrew.

The group of Professional Voice Users (PVU) was chosen for this study particularly for being a homogenous group with regards to the dependent variable, namely literacy in relation to speech or singing, with the written text as the principal stimulus. For this group in particular, IHS effected diction may have further implications.

Given my long and extensive professional clinical experience in the guidance and treatment of PVUs, I knowledgably and from first-hand experience can argue that the primarily consonant- based syllabic orthography of written Hebrew, contributes to the "harshness" and the lack of natural IH speech flow in literates in general and particularly in IH speaking adults whose profession depends on the articulation of written texts (e.g. teachers, actors, lecturers, singers). In a research carried out by Carreiras & Price (2008), where they followed brain activations for consonants and vowels, they based their experiment on the reading aloud of, and lexical decisions on visually presented pseudo-words. These pseudo-words were created by transporting or replacing consonants or vowels within the real words. Using fMRI they showed that in the reading aloud test, when the pseudo words were obtained by the erroneous replacement of vowels within a word, activation occurred in the area associated with prosodic processing of speech input. They then argued that prosodic processing of vowel changes arises during the self-monitoring of the speech output. In the case of PVUs who apply this self- monitoring processing to an IHS text, lack of vowel presentation leads to a particular difficulty for voice production in the self-monitoring processing of prosodic changes (since based on vague vowels). Therefore, IH orthography can be a risk factor that occasionally may lead to voice abuse 98

and voice impairments.61 These findings may contribute to rehabilitation programs for IH PVUs and with further research along the method outlined here, may also be adapted to PVUs in other languages. While speech and voice training for actors, singers and narrators is part and parcel of their study curriculum (regrettably excluding overall teachers’ seminars from such training), the specific and unique orthographical characteristics of a language are not considered. This exclusion affects the attainment of the required balance between consonant and vowel production, often resulting in an unhealthy voice. Pending further research of the unique orthographic effects characteristics of IH and of other syllabic based orthographies like Arabic or Japanese, the inclusion of training aimed at maintaining a healthy voice based on orthographic manipulations may be very beneficial. As this study has shown orthographic based online changes in vowel length production, it suggests that consistent training based on script variation may lead to a growing capacity to manipulate vowel changes in a more salubrious manner.

Moreover, this approach could also enhance new perspectives on rehabilitation programs in respect of a range of clinical speech pathologies, particularly voice pathologies, various learning disabilities, diction, and fluency, as well as skills developments which are influenced by the bi- directional relation of orthography and speech.

Finally, this approach might help us understand the process whereby written communication via all electronic platforms, adapting to IH speakers' need for disambiguation by the insertion of vowel letters, seems to gradually influence IH orthography and consequently its speech production.

61 In literate IH-speaking speakers, the motivation to self-monitor prosody based upon vague vowels leads to a speech that concentrates most of its energy on consonant production, resulting in a speech deprived of the flowing sound and overtones that generally emerge from the vowel - the nucleus of syllable structure. Therefore, this tendency causes extensive force to be imposed on the consonants, in turn promoting glottal attacks. The relation between articulation of vowels and consonants to glottal attacks has been researched (e.g. Aronson 1990, Watterson & McFarlane 1993). The self-monitoring processing of prosody is known to be a risk factor that may lead to voice abuse (e.g. Boone 1983, Morrison& Rammage 1993, Hogikyan 1999, and Buckmire 2008). .

99

A Final note: "Speaking Culture" refers to the various cultural characteristics that affect speech, such as cultural codes of communication, supra segmental characteristics, age, gender, fashion, institutional and national attributes. As stated by de Saussure, “The culture of a nation exerts an influence on its language and the language is largely responsible for the nation” (de Saussure, 1966: p.20). Further research of the fundamental effect of language characteristics, specifically the effect of orthographic characteristics of a given language on the culture of speech in respect of diction, may be highly interesting and carries much promise. In examining the cultural aspect and the human factor in speech production, this approach was inspired by the theory of “Phonology as Human Behavior” (e.g., Diver 1979, Davis 1984/1987, Tobin 1995, 1997, 2000, and 2009).

100

Bibliography

 Adam, G. & Bat-El, O. (2009). When Do Universal Preferences Emerge in Language Development? The Acquisition of Hebrew Stress. Brill’s Annual of Afro-Asiatic Languages and Linguistics 1, 255–282.  Anthony, J.L. & Francis, D.J. (2005). Development of Phonological awareness. Current Directions in Psychological Science, 14(5), 255-258.  Aronson, A.E. (1990). Clinical Examination of Voice Disorders. In: Aronson A.E., Clinical Voice Disorders. Thieme Inc., New York.  Becker, M. (2003). Hebrew Stress: Can’t you hear those trochees? In: Kaiser, E. and Arunachalam, S. (eds.) Proceedings of PLC 26(9.1), 45–58.  Azar M. (2009). “Ancient Language in a New eality”- Reassessment. In Cohen, H.A. (ed.) : Two hundred fifty Years of Modern Hebrew. Hebrew Language Academy 1,17-32 [in Hebrew]; אזר, מ. )9002(."לשון עתיקה במציאות חדשה" - בחינה מחודשת. מתוך: ח.א. כהן )עורך(. מאתיים וחמישים שנות עברית מחודשת. הוצאת האקדמיה ללשון העברית, 1 ,11-29.  Baker, C.I., Liu, J., Wald, L.L., Kwong, K.K., Benner, T. & Kanwisher, N., (2007). Visual word processing and experiential origins of functional selectivity in human extrastriate cortex. Proc. Natl. Acad. Sci. U.S.A. 104, 9087–9092.  Bar Adon, A. (1975). The Rise and Decline of a Dialect. The Hague: Mouton.  Bat-El, O. (1993). Parasitic metrification in the Modern Hebrew Stress System. The Linguistic Review 10, 189 - 210.  Ben Dror, I., Frost R. &Bentin S. (1995). Orthographic representation and phonetic segmentation in skilled readers. Psychological Science 6(3), 176-181.  Ben-Shachar, M., Dougherty, R.F., Deutsch, G.K., Wandell, B.A., 2007. Differential sensitivity to words and shapes in ventral Occipito-temporal cortex. Cereb. Cortex 17, 1604–1611.  Ben Haim, Z. (1292). Ancient Language in New Reality. Leshonenu La’am d Vol. 15.(in (ז. בן חיים. לשון עתיקה במציאות חדשה. לשוננו לעם ד' :Hebrew  Ben Yehuda, E. (1978). Diction in the Hebrew Language. In: The Unfulfilled Dream a Selection of Essays in Language , R. Sivan ed. Mossad Bialik.[ in Hebrew]; 101

בן יהודה א. )1211(. המבטא בלשון העברית. מתוך: החלום ושברו: מבחר כתבים בעינייני לשון, ההדיר ר. סיון, מוסד ביאליק. /http://benyehuda.org/by  Blanc, H. (1968). The Israeli Koine as an Emergent National Standard. In: Fishman J.A. & Fergusson C. (eds.), Language Problems of Developing Nations. John Wiley, New York: 237-251.  Bolger, D.J., Hornickel J., Cone N.E.,1 Douglas D. Burman D.D., & Booth J.R.(2008). Neural Correlates of Orthographic and Phonological Consistency Effects in Children. Human Brain Mapping 29 (12), 1416-1429.  Bolozky S. (1978).Some Aspects of Modern Hebrew Phonology. In: Aronson Berman, R. (eds.), Modern Hebrew Structure, University Publishing Projects, Ltd.Tel Aviv.  Bolozky S. (1990). My suggestion: Marking the Vowels [e] & [a] and Marking the owels’ Absence in Modern Hebrew Orthography. Journal of Linguistics, Languages and Hebrew Language, Sept.-Oct., 34-37.]in Hebrew]; בולוצקי, ש. )1220(. יש לי הצעה: על סימון התנועות e ו a ועל סימון העדר התנועות בכתיב של העברית החדשה. כתב עת לעינייני לשון, שפות והלשון העברית, ספט.-אוקט. , 23-21.  Bolozky, S. & Scwarzwald, O. (1990). On vowel Assimilation and Deletion in Casual Modern Hebrew. Hebrew Annual Review 12, 23-48.  Bolozky, S. (1997). Israeli Hebrew phonology. In: Kaye, A. S. & Daniels, P. (eds.), of Asia and Africa. Winona Lake: Eisenbrauns. Vol. 1. Ch. 17, 287-311.  Boone, D.R. (1983).The Voice and Voice Therapy. Englewood Cliffs, New Jersey p. 206.  Braet, W., Wagemans, J. & OP de Beek, H.P. (2012). The visual word Form Area is Organized According to Orthography. Neuroimage, 59, 2751-2759.  Brem, S., Bucher, K., Halder, P., Summers, P., Dietrich, T., Martin, E., & Brandeis, D. (2006). Evidence for developmental changes in the visual word form processing network beyond adolescence. Neuroimage, 29, 822–837.  Brewer, J. (2008). Phonetic Reflexes of Orthographic Characteristics in Lexical Representation. PhD dissertation, University of Arizona, Department of Linguistics.  Buckmire, A. (2008). Vocal Polyps and Nodules. eMedicine Specialists/ Otolaryngology and Facial Plastic Surgery/ Laryngology. 102

 Carreiras, M. & Price, C.J.(2008). Brain Activation for Consonants and Vowels. Cerebral Cortex 18(7), 1727-1735.  Cohen, L., & Dehaene, S. (2004). Specialization within the ventral stream: the case for the visual word form area. Neuroimage, 22, 466-476.  Damian, M., & Bowers, J. (2003). Effects of Orthography on Speech Production in a Form Preparation Paradigm. Journal of Memory and Language, 49, 119-132.  Davis, Joseph. (1984/1987). A Combinatory Phonology of Italian. Columbia University Working Papers in Linguistics (CUWPL) 8,1-99.  Dehaene, S., Pegado, F., Braga, L.C., Ventura, P., Filho, G.N., Jobert, A., Dehaene-Lambertz, G., Kolinsky, R., Morais, J., & Cohen, L. (2010). How learning to read changes the cortical networks for vision and language. Science 330 (6009), 1359–1364.  De Saussure F. (1966). Course in General Linguistics. The Philosophical Library, Inc. N.Y.  Desorches, A.S., Cone, N.E., Bolger, D.J., Bitan, T., Burman, D.D. & Booth, J.R., (2010). Children with Reading Difficulties Show Differences in Brain Regions Associated with Orthographic Processing During Spoken Language Processing. Brain Research 1356, 73-84.  Diver, William. 1979. “Phonology as human behavior.” In: Aaronson, D. & Reiber, P. (eds.). Psycholinguistic Research: Implications and Applications, Hillside NJ: L. Erlbaum, 161-186.  Fellman, J. (1973). The Revival of a Classic tongue: Eliezer Ben Yehuda and the Modern Hebrew Language. The Hague: Mouton.  Frost, R., Repp, B.H. & Katz, L.(1988). Can Speech Perception be influenced by Simultaneous Presentation of Print? Journal of Memory & Language, 27, 741-755.  Frost, R. (1995). Phonological Computation and Missing Vowels: Mapping Lexical Involvement in Reading. Journal of Experimental Psychology: Learning, Memory and Cognition, 21, 398–408.

 Frost, R. (1998). Toward a Strong Phonological Theory of Visual Word Recognition: True Issues and False Trails. Psychological Bulletin, 123(1), 71-99.  Frost, R. (2006). Becoming Literate in Hebrew: The Grain-Size Hypothesis and 103

Semitic Orthographic Systems. Developmental Science, 9 (5), 439-440.  Gelder, B. & Vroomen, J. (1991). Phonological Deficits: Beneath the Surface of Reading Acquisition Problems. Psychological Research, 55, 88-97.  Gesenius, W. (1909). The Vowels in General, Vowel Letters and Vowel Signs. Gesenius' Hebrew Grammar Ch. 7.

 Goldwasser, O. (2010). How the Alphabet Was Born from Hieroglyphs. Biblical

Archaeology Review, 36(02), 1-27.  Graf, D. & Ussishkin, A. (2002). Emergent Iambs: Stress in Modern Hebrew. Lingua 113, 239 – 270.  Haley, K.L. & Overton H.B. (2001). Word Length and Vowel Duration in Apraxia of

Speech: The Use of Relative Measures. Brain and Language 79, 397–406.  Harshav, B. (1993). Language in Time of Revolution. Berkeley and Los Angeles: University of California Press.  Haspelmath M., Dryer, M.S., Gil, D. and Comrie, B. (eds.) (2008). The World Atlas of Language Structures Online. Max Plank Digital Library, Munich, Ch.3.  Hoffman, J.M. (2004). Modern Hebrew. In: In the Beginning: A Short History of the Hebrew Language. New York University Press, Ch.10, 187-213.  Hogikyan, N.D., Appel, S. & Guinn, L.W. (1999). Vocal fold nodules in adult singers: Regional Opinions about Etiologic Factors, Career impact, and Treatment. A Survey of Otolaryngologists, Speech Pathologists, and Teachers of singing. J Voice. 13(1), 128-42.  Katz, L., & Frost, R. (1992). Reading in Different Orthographies: The Orthographic Depth Hypothesis. In: Frost, R. &. Katz, L. (Eds.). Orthography, Phonology, Morphology, and Meaning Amsterdam, North-Holland, 67-84.

 Kovelman, L., Norton, E.S., Christodoulou, J.A, Gaab, N., Lieberman, D.A., Triantafyllou, C., Wolf, M., Whitfield-Gabrieli, S. & Gabrieli, J.D.E. (2011). Brain Basis of Phonological Awareness for Spoken Language in Children and Its Disruption in . Cerebral Cortex Advance Access, 21,1-11. Downloaded

from http://cercor.oxfordjournals.org/ at Tufts University.  Lieberman, P. (1992). On the Evolution of Language. In: Hawkins, J. A. & Gell- Mann, M. (eds). The Evolution of Human Languages. Reading, MA, Addison 104

Wesley, 21-48.  Linklater, L. (1976). Freeing the Natural Voice. Drama Book Publishers, New York, Ch. 16.  Miller, K. M., and Swick, D. (2003). Orthography Influences the Perception of Speech in Alexic patients. Journal of Cognitive Neuroscience, 15, 981-990.  Morais, J., Content, A., Cary, L., Mehler, J. & Segui, J. (1989). Syllabic Segmentation and Literacy. Language and Cognitive Processes, 4, 57-67.  Most, T. & Adi Ben Said, L. (2001). The Influence of Contextual Information on Speech by Postlingually and Prelingually Profoundly Hearing-Impaired Hebrew- Speaking Adolescents and Adults. Ear &Hearing, 22(3), 252-263.  Munson, B. & Solomon, N.P. (2004). The influence of Phonological Neighborhood Density on Vowel Articulation. Journal of Speech, Language, and Hearing Research, 47, 1048-1058.  Munson, B. (2007.) Lexical Access, Lexical Representation, and Vowel Production. Papers in Laboratory Phonology, 10, 1-24.  Morrison, M. D. and Rammage, L. A. (1993) Muscle misuse voice disorders: Description and classification. Acta Oto Rhino laryngol Stockh. V. 113, 428-434.  Ohala, J.J. (1993). Coarticulation and Phonology. Language and Speech, 36(2,3), 155- 170.  Ornan, U. (2009). When was Modern Hebrew Born? In: Two hundred fifty Years of Modern Hebrew. Cohen H.A. ed. Hebrew Language Academy, 1,3-16 [in Hebrew]; אורנן, ע. )9002(."מתי נולדה העברית החדשה?"-בחינה מחודשת. מתוך: מאתיים וחמישים שנות עברית מחודשת. ח.א. כהן עורך. הוצאת האקדמיה ללשון העברית, 1 , 2-19.  Palmer, S. (2000). Development of Phonological Recoding and Literacy Acquisition: A Four-Year Cross-Sequential Study. British Journal of Developmental Psychology, 18, 533–555.

 Pattamadilok, C., Perre, L., Dufau, S. & Ziegler,. J.C. (2008). On-line Orthographic Influences on Spoken Language in a Semantic Task. Journal of Cognitive Neuroscience 21(1), 169–179. 105

 Pattamadilok, C., Morais, J., Ventura, P., & Kolinsky, R. (2007).The locus of the Orthographic Consistency Effect in Auditory Word Recognition: Further Evidence from French. Language and Cognitive Processes, 22, 1–27.  Peereman, R., Dufour, S. & Burt, J.(2009). Orthographic Influences in Spoken Word Recognition: The consistency Effect in Semantic and Gender Categorization Tasks. Psychonomic Bulletin & Review 16 (2), 363-368.  Perre, L. and Ziegler, J. C. (2008). On-line activation of orthography in spoken word recognition. Brain Research, 1188, 132–138.  Perre, L., Pattamadilok, C., Montant, M., & Ziegler, J. C. (2009). Orthographic Effects in Spoken Language: On-line Activation or Phonological Restructuring? Brain Res. 1275, 73–80.  Perfetti, C.A.& Marron M.A. (1998). Learning to Read: Literacy Acquisition by Children and Adults. In Wagner, D.A. (ed.).Advances in Adult Literacy Research and Development. Hampton Press.  Rabin, C. (1972). A Short History of Hebrew Language. The Jewish agency.  Rastle, K. McCormick, S.F., Bayliss, L. & Davis, C.J.(2011). Orthography Influences the Perception and Production of Speech. J. of Experimental Psychology: Learning, Memory and Cognition, 37(6), 1588-1594.  Rosen, H. (1967). Our Hebrew: In Light of Linguistic Methods. Am Oved pub.[In Hebrew]; רוזן, ח. )1291(. העברית שלנו: דמותה באור שיטות הבלשנות. הוצאת עם עובד.  Seidenberg, M.S. & Tanenhaus, M.K. (1979). Orthographic Effects on Rhyme Monitoring. Journal of Experimental Psychology: Human Learning and Memory, 5, 546- 554.  Shiffman, H. (1996). Diglossia as a Sociolinguistic Situation. In Coulmas, F. (ed.). The Handbook of Sociolinguistics. Oxford: Blackwell.  Steinberg, D. (2012). Modern or Israeli Hebrew. In: History of the Ancient and Modern Hebrew Language, E-book, 2.5.aft, M. (2011). Orthographic Influences when Processing Spoken Pseudo Words: Theoretical Implications. Frontiers in Psychology / Language Sciences, 2(140),1-7.  Stemberger, J. (1992). Vocalic under Specification in English Language Production. 106

Language 68(3), 492–524.  Tene, D. (1969). Israeli Hebrew. In: Ariel Vol. 25, 48-63 (Reprinted by Steinberg, D.)  Tobin, Y. (1990a). Combinatory Phonology of the Hebrew Triconsonantal (CCC) Root System. La Linguistique 26(1), 99-114.

 Tobin, Y. (1990b)."Principles for a Contrastive Phonotactics: The Hebrew Triconsonantal

(CCC) Root System a Case in Point". Papers and Studies in Contrastive Linguistics 26,

137-153.  Tobin, Y. (1997). Phonetics and Phonology. In Tobin, Y. (ed). Phonology as Human Behavior: Theoretical Implications and Clinical Applications. Durham, NC/London: Duke University Press, I, 1-49.  Tobin, Y. (2000). Comparing and Contrasting Optimality Theory with the Theory of Phonology as Human Behavior. The Linguistic Review 17(2-4), 291-301.  Tobin, Y. (2004). Between Phonology and Lexicon: The Hebrew Triconsonantal (CCC) Root System Revolving around /r/ (C-r-C). In: Kirsner, R., Contini-Morava, E. & Rodriguez- Bachiller, B. (eds.), Cognitive and Communicative Approaches to Linguistic Analysis. Amsterdam and Philadelphia, John Benjamins, 289-323.  Tobin, Y. (ed). 2009. Special Theme Issue: Phonology as Human Behavior. Asia-Pacific Journal of Speech, Language and Hearing. 12(2).  Ventura, P., Morais, J., Pattamadilok, C., & Kolinsky, R. (2004). The Locus of the Orthographic Consistency Effect in Auditory Word Recognition. Language and Cognitive Processes, 19, 57–95.  Wagner, R. & Torgesen, J. (1987). The Nature of Phonological Processing and its Causal Role in the Acquisition of Reading Skills. Psychological Bullatin, 101, 192-212.  Watterson, T. & McFarlane, S.C. (1993). Phoneme Effects on Vocal Effort and Vocal Quality. American Journal of Speech-Language Pathology, 2, 74-78.  Ziegler, J. C. & Ferrand, L. (1998). Orthography Shapes the Perception of Speech: The Consistency Effect in Auditory Word Recognition. Psychonomic Bulletin & Review, 5, 683-689.  Ziegler, J.C., Ferrand, L. & Montant M., (2004). Visual Phonology: The Effects of Orthographic Consistency on Different Auditory Word Recognition Tasks. Memory & cognition, 32(5), 732-41. 107

 Ziegler, J. C., & Muneaux, M. (2007). Orthographic Facilitation and Phonological Inhibition in Spoken Word Recognition: A Developmental Study. Psychonomic Bulletin & Review, 14, 75–80.

108

Appendix I

Pilot Test Materials

לפניך מספר סדרות של משפטים אשר יוצגו ב3 אופנים:

1. בעברית 2. ב HEBRISH: עברית הכתובה באותיות לטיניות 3. מילות טפל באותיות עבריות ולטיניות

הסבר ותרגול מקדים לקריאת העברית ומילות הטפל באותיות לטיניות::

התנועות:

A – יבוטא כמו X

E – יבוטא כמו X

I - יבוטא כמו Xי

O – יבוטא כמו Xו

U – יבוטא כמו Xּו

וכעת לתרגול קצר:

KAL, GEN, TIL, SUR, NOL, BUL

העיצורים:

ישנם עיצורים אשר ייצוגם ב HEBRISH דורש הבהרה:

CH = ח/כ

TS= צ

Z = ז

SH = ש

Y = י 109

וכעת לתרגול קצר:

מילים ב HEBRISH:

TSAM, TSLIL, ZER, SHIR, YOM, CHAG, YESH, ZACH, HAR, GASH

מילות טפל:

TOL , KESH, SEZ, CHUV, TSUN, ROTS, NISH, SER, POL, VOZ

משפטי חימום: 1. היום יש לי חום. 9. יש לך תיק חדש? 3. בוא אלי, תן לי יד! 3. זה צל של קוף עם בננה. 2. אם אין אני לי, מי לי?

HEBRISH:

1. HAY M YESH LI CHOM. 2. YESH LACH TIK CHAD SH? 3. BO ELAI, TEN LI YAD! 4. ZE TSEL SHEL KOF IM BAN NA. 5. IM EIN ANI LI, MI LI?

110

עברית

1. יש לו לב רע! 9. אין לי כל יום חג. 2. תן לי כוס תה! 3. סיר זה קצת חם. 5. דוב לא בא. 9. גם הוא כלל לא שב? 1. קר לו כאן? 1. טל רץ. 2. ניב קם. 10. טוב לי פה.

משפטים עם מילות טפל:

1. פיג שוף 9. ויר טם. 2. פש הל מיט? 3. לז קג סוץ רי. 5. תוץ נג בוט איז רל! 9. דיק לוג וש דך? 1. אוק פיש ווד קיז לוס. 1. ביט צו זוך . 2. שוף אוי זם נק, פץ יל.

10. יוס הח, אג מון פון שוי.

111

אנא קרא/י את המשפט לעצמך לפני הקריאה בקול רם.

1. כל אב קבל משכרת.

9. כלם לומדים חבור וחסור מספרים בכתה א'.

2. אין לי כן ציור.

3.ספרתי ספור מאד מענין.

5. היא לובשת שמלה ארכה.

9.בקר טוב ונפלא לכל אשה.

1. השמים כחלים והלב מתמלא שמחה.

1. השקת התמלאה במים מאד צלולים.

2.התרגשתי כל-כך וקבלתי חם.

10.אם תרצו אין זו אגדה.

112

HEBRISH

A – יבוטא כמו X

E – יבוטא כמו X

I - יבוטא כמו Xי

O – יבוטא כמו Xו

U – יבוטא כמו Xּו

העיצורים:

ישנם עיצורים אשר ייצוגם ב HEBRISH דורש הבהרה:

CH = ח/כ

TS= צ

Z = ז

SH = ש

Y = י

113

משפטים עם מילות טפל Pseudo words sentences

1. PESH HEL MIT? 2. VIR TEM. 3. BIT TSO ZUCH! 4. YOS HACH, EG MUN FON SHOI. 5. LEZ KAG SUTS RI. 6. TUTS NAG BUT IZ REL! 7. DIK LOG VESH DACH? 8. OK PISH VOD KIZ LOS. 9. PIG SHOF! 10. SHUF OI ZEM NAK, PETS YEL.

HEBREW SENTENCES

1. EIN LI KOL YOM CHAG. 2. NIV KAM. 3. TEN LI KOS TE! 4. DOV LO BA. 5. GAM HU KLAL LO SHAV? 6. KAR LO KAN? 7. SIR ZE KTSAT CHAM. 8. TAL RATS. 9. TOV LI PO. 10. YESH LO LEV RA!

114

הערה לקורא: הכתיב המלא ברשימות המילים שלהלן בחלקו ע"פ המקובל62 )אימא לעומת אמא( ובחלקו תוספות לא מקובלות )אין לעומת איין( אשר שמשו את מטרות המבחנים השונים לבדיקת השפעתן של אמות קריאה מוספות לשם ייצג פונמי מלא על אורך התנועה בתנאים השונים.

עברית:

1. כול איש קיבל משכורת 2. כולם לומדים חיבור וחיסור מיספרים בכיתה אלף. 3. איין לי כאן ציור. 4. סיפרתי סיפור מאוד מעניין 5. היא לובשת שימלה ארוכה 6. בוקר טוב וניפלא לכול אישה. 7. השמיים כחולים והלב מיתמלא שימחה. 8. השוקת היתמלאה במיים מאוד צלולים. 9. היתרגשתי כול-כך וקיבלתי חום. 11. אים תירצו איין זו אגדה.

62 For further information, see http://hebrew- academy.huji.ac.il/hahlatot/MissingVocalizationSpelling/Pages/ikkar.aspx (also: "כללי הכתיב חסר הניקוד ,"לשוננו לעם ,טבת התשס"ב(.

115

Appendix II

Main Test Materials

לפניך רשימות של מלים בעברית המוצגות בHEBRISH = עברית המשוכתבת באותיות לועזיות. להלן מקרא ובהמשכו אמון קצר:

התנועות: A – יבוטא כמו X E – יבוטא כמו I - יבוטא כמו י O – יבוטא כמו Xו ׄׄ U – יבוטא כמו Xּו

וכעת לתרגול קצר: KAL, GEN, TIL, SUR, OL, BUL

העיצורים: ישנם עיצורים אשר ייצוגם ב HEBRISH דורש הבהרה : X = ח/כ 63 TZ= צ Z = ז SH = ש Y = י וכעת לתרגול קצר: מילים ב HEBRISH: TZAM, TZLIL, ZER, SHIR, YOM, XAG, YESH, KAX, ZAR, GASH

graphemes was by the letters’ cluster /CH/. This ח/כ 63In the pilot study the phonemic presentation of [x] in HEBRISH for the caused many mistakes by subjects and therefore caused to dismiss subjects’ data . After several pretests the /X/ found to be a better and safer version and therefore was used for the main test. This refers as well to the change in the phonemic .grapheme from /TS/ in the pilot test to /TZ/in the main test /צ/ [presentation of the [ts 116

B KER TOV! SHAB T SHAL M. TOD AB . YESH L NU MA L. AG SAM AX . T DEK T DEK TI D F. T A I X LIKN T AT TZ. MAT TI AH V! SHAL M U A SHANA ZORAXAT VEGAM PORAXAT.

הערה לקורא: הכתיב המלא ברשימות המילים שלהלן בחלקו ע"פ המקובל 64)אימא לעומת אמא( ובחלקו תוספות לא מקובלות )אין לעומת איין( אשר שמשו את מטרות המבחנים השונים לבדיקת השפעתן של אמות קריאה מוספות לשם ייצג פונמי מלא על אורך התנועה בתנאים השונים. הפריטים הוצגו לנבדק כרשימות מעורבות באופן מכוון.

64 For further information, see http://hebrew-academy.huji.ac.il/hahlatot/MissingVocalizationSpelling/Pages/ikkar.aspx (also: "כללי הכתיב חסר הניקוד ,"לשוננו לעם ,טבת התשס"ב(.

117

להלן רשימות מילים בעברית המוצגות בעמודות. אנא קרא כל עמודה מלמעלה למטה . III II I

מזל אומצה אב

כל כסא אולם

ער ִאמון מרץ

נר משרד מלבן

נירה סבא מסרק

אשה כֻּלם מסכן

ח̂שך חשמל שב̂ת

אח תנוק ש̂רש

קבוץ קלמר עוור

פח חבור תלבשת

ז̂בל חסור יפה

118

VI V IV

שמלה שלש מסתכל

אבא קל נשתה

אמא הם אומרת

רב יופי השתדל

ִקבל ש̂קט תומר

בגד שריון שומרת

בור משלש שברה

סיר סכריה ארץ

חבר סמון משקלת

טוב אוגר קשון

חורף בול

119

IX VIII VII

אותות חוברת אורווה

עוורת חורבה סוגר

חוורת מכלת שמרה

מֻ כרח עומר אמרה

אשור ִגשה עוברות

עוברת עברה אור

סֶפר זמון כלא

התקדמה עתונות שומרות

משכ̂רת חוור גשור

ראשון ש̂מש רב

ש̂קל

120

XII XI X

עיר באת חאן

אומרות מוט תן

נצא זאת קוראת

סטונות ראש בֹּקר

אופי שוברות החבאתי

זכרון שמאל מצאת

עקרון אגד כאן

עוני רן אות

שוני מוטות נמאס

חומצה בת

שניים רוצה

1. תן קצת שקט! 9. כל אב קבל משכרת. 2. גם להם יש חג? 3. תמר עוברת כאן כל שבת. 5. יש לְך שם קל. 9. היא לובשת שימלה ארוכה. 1. הכל נגמר! 1. אמא קוראת ספר כל בקר. 2. נירה שברה כיסא. 10. נירה בת שלש. 11. החבאתי משקלת בתוך פח. 121

III II I

שימלה שלוש מיסתכל

שורש נישתה זיכרון

אימא כול עיוורת

כולם יופי הישתדל

קיבוץ אישה חושך

קיבל שיריון חיוורת

מישרד נישתה אימון

חיבור סוכריה גישור

תינוק משולש מישקולת

חשובה חיסור בול

122

VI V IV

כיסא איתות משכורת

מיסכן עיתונות עיקרון

עיוור מכולת סיטונות

מוכרח סימון תודה

אישור גישה

תילבושת קישון

בוקר זימון

היתקדמה חיוור

שלום חידה

1. ִא מּא קוֹ ֵראת ֶס ֶפר ֺכּל בֹּקֶ ר. 9. ֵתּן ְק צת ֶש ֶקט! 2. נִי רה ש ב רה ִכּ י ֵסא. 3. נִי רה בּת שֹלש ֺ .5כּל ָאב ׅק ֵבּל מ שׂ ֺכּרֶ ת. 9. גם ל ֵהם יֵש חג? 1. ֵה ח ֵבּא ִתי ִמ שק ֹ ֶלת בּתוֹך פּח 1. תּ מר עוֹבֶרֶ ת כּאן ֺכּל ש בּת. 2. יֵש לְך ֵשם קל. ׅ .10ה יא לוֹ ֶב ֶשת ִשׂי מ הל אֲרּוכּה. 11. הכּ ֹל נִ ְג מר! 123

לפניך רשימות של מלים בעברית המוצגות בHEBRISH = עברית המשוכתבת באותיות לועזיות. להלן מקרא ובהמשכו אמון קצר:

התנועות: A – יבוטא כמו X E – יבוטא כמו I - יבוטא כמו י O – יבוטא כמו Xו ׄ U – יבוטא כמו Xּו

וכעת לתרגול קצר: KAL, GEN, TIL, SUR, OL, BUL

העיצורים: ישנם עיצורים ייחודיים לשפה העברית: X = ח/כ TZ= צ Z = ז SH = ש Y = י וכעת לתרגול קצר: מילים ב HEBRISH: TZAM, TZLIL, ZER, SHIR, YOM, XAG, YESH, KAX, ZAR, GASH

B KER TOV! SHAB T SHAL M. TOD AB YESH L NU MA L. AG SAM AX . T DEK T DEK TI D F. T A X LIKN T AT TZ. MAT TI AH V! SHAL M U A , SHAN O XAT EG M PO XAT.

124

להלן רשימות מילים בHEBRISH המוצגות בעמודות. אנא קרא כל עמודה מלמעלה למטה .

III II I

AV UMT MA L

UL M KIS KOL

M RETZ IM N ER

MALB N MIS D NER

MAS K S BA N RA

MISK N KUL M ISH

SHAB T ASHM L SHEX

SH RESH TIN K AX

I R KALM R KIB TZ

TILB SHET IB R PAX

K VA IS R BA X

125

VI V IV

MISTAK L SHAL SH SIML

NISHT KAL BA

OM RET HEM MA

HISHTAD L Y FI RAV

T MER SH KET KIB L

SHOM RET SHI Y N B GED

SHA MESHUL SH BOR

RETZ SUKA Y SIR

MISHK LET SIM N A R

KISH N OG R TALM D

TOV SHOM R

126

IX VIII VII

U O RET OT T

SOG R U I RET

SHAM MAK LET I RET

AM MER MU X

O T GISH ISH R

OR A O RET

K LE IM N S FER

SHOM T ITON T HITKADM

GISH R I R MASK RET

D I SH MESH ISH N

SH KEL

127

XII XI X

XAN BAT IR

TAN MOT OM T

KO T ZOT NET

B KER SH SITON T

HE B TI SHO T FI

MAT T SMOL IKA N

KAN GED IKA N

OT RAN NI

SIY M MOT T SH NI

TOD UMT

SHAL M

128

1. N A SHA KIS . 2. N A BAT SHAL SH. 3. KOL A KIB L MASK RET. 4. TAM R O RET KAN KOL SHAB T. 5. YESH LAX SHEM KAL. 6. TEN KT AT SH KET! 7. HE B TI MISHK LET BET X PAX. 8. HI LO SHET SIML A UK . 9. GAM LAH M YESH XAG? 10. HAK L NIGM R! 11. MA KO T S E KOL B KER.

אנא השב: מהי התרשמותך מההתנסות בקריאה?

)Please answer: What is your impression from the reading experience?)

129

תקציר

מחקר זה עוסק בהשפעת האורתוגרפיה )השפה הכתובה( על הדבור ומתמקד בהשפעת השפה העברית הכתובה על הפקת הדבור. בעברית הישראלית שהוחייתה כשפה מדוברת בשלהי המאה ה12 מכתבי הקודש והמסורת, יכולה השפעה זו להימצא ייחודית ובעלת משקל.

בשנים האחרונות עולה וגובר העניין בבדיקת השפעת האורתוגרפיה על תיהלוך השפה המדוברת, גישה שהנה הופכית לגישה הבלשנית המסורתית המתעדת את התפתחות האורתוגרפיה מן השפה המדוברת. בין הגורמים המאיצים את העניין בהשפעת האורתוגרפיה על הדבור היא תפוצתה הגוברת של האוריינות ויתרה מזאת, השימוש העולה וגובר באמצעי התקשורת האלקטרוניים, כמו האינטרנט והטלפון הנייד. עניין זה הוליד מחקרים רבים העוסקים בקשר הדו כווני שבין השפה הכתובה והמדוברת.

לאורך כל ההיסטוריה, השפה העברית הכתובה התאפיינה ב"אורתוגרפיה עמוקה" הנובעת ממערכת אורתוגראפית עיצורית-הברתית, שבבסיסה מערכת שורשים עיצוריים, וייצוג חלקי בלבד של התנועות ע"י אותיות אהו"י. העומק האורתוגרפי, הנובע בעיקרו מהעדר ייצוג התנועות, הוליד תמורות שונות במהלך ההיסטוריה זאת עקב המתח והפער שבין השפה המדוברת והכתובה. בניסיונות השונים למלא פער זה בולטת שיטת הניקוד הטברייני - סימון אנכי שנוסף על מנת לשמר את צלילי השפה העברית הקדומה - שיטה אשר נשתמרה והנה בשימוש עד עצם היום הזה.

מאידך, שיטת הניקוד המורכבת אשר היוותה מרכיב מרכזי בהחייאת השפה העברית, הושמטה מהשימוש היום- יומי בקרב ילדי הארץ למעט בספרי קודש, בספרי שירה ובספרות ילדים. בכך חזרה השפה הכתובה לאפיונה המקורי כאורתוגרפיה עיצורית-הברתית נעדרת ייצוג עקבי של תנועות. ניתן לשער כי מהלך זה נבע מהנטייה של משתמשי השפה לפעול עפ"י הכלל של "תקשורת מרבית במאמץ מינימאלי" ) minimal effort to achieve“ maximal communication” Tobin 1990a, 1990b,1997( היות ומערכת הניקוד המורכבת והעמוסה נמצאה לא יעילה ואף מיותרת. וכך, משהשפה הוחייתה הושמטה שיטת הניקוד מהשימוש השוטף לטובת האורתוגרפיה העיצורית-הברתית.

במחקר זה אני מבקשת לטעון שהאורתוגרפיה העיצורית-הברתית, אשר מבחינה היסטורית מבוססת על מערכת מורפולוגית עשירה של שורשים, והנעדרת ייצוג עקבי של תנועות, משפיעה על הפקת הדבור. הדבר מוצא את ביטויו בדגש על עיצורים והפחתת משכי התנועות במבע. דבר זה מתייחס באופן מיוחד לאורייניים ישראלים, דוברי העברית הישראלית כשפת אם, כ"שפה אופפת כל" )אורנן 9002( המשמשת בכל צרכי הדבור, הקריאה והכתיבה. אני אף מניחה שאין הדבר כולל את כלל המשתמשים בארץ ובתפוצות אשר עבורם אין השפה מהווה שפת אם ו/או נמצאת בשימוש יומיומי. טענתי נסמכת בין היתר על מחקרים פסיכולינגוויסטיים שנעשו בעשורים האחרונים ובדקו 130

ישראלים אורייניים, מחקרים שמצאו אפקט קדימות בייצוג ובזמינות של עיצורים על פני תנועות Katz & Frost) .1992, Frost, 1995, 1998, 2006)

לאור ההנחות והממצאים הנ"ל ערכתי מחקר חלוץ שמטרתו לבדוק את ההשערה כי לאורתוגרפיה ההברתית של העברית הישראלית השפעה על הפקת הדבור. המחקר התמקד בבוגרים דוברי עברית ככלי מקצועי. אוכלוסיה זו נבחרה משום השימוש היומיומי שלה בשפה הכתובה כגירוי מרכזי בעיסוקם המקצועי כדוברים, אפיון המהווה את המשתנה הבלתי תלוי במחקר זה.

שערתי כי הקדימות העיצורית של האורתוגרפיה ההברתית בעברית הישראלית משפיעה על הדבור, בהגייה בה ניתן יתר דגש על עיצורים וזאת על חשבון הפחתת משכי התנועות. לאור תוצאות מחקר החלוץ אשר תמכו בהנחה זו, ערכתי את המחקר המקיף: 1 בוגרים דוברי עברית ישראלית ככלי מקצועי, בני 90-55 הוקלטו בקריאת חומר מבחן שחובר על ידי בהתבסס על לקחי מחקר החלוץ. חומר הקריאה כלל מילים ומשפטים שהוצגו במספר אופני כתיבה: אורתוגרפיה הברתית - ללא ייצוג התנועות; כתיב מנוקד; כתיב מלא; וכן -HEBRISH )עברית הכתובה באותיות לטיניות - כתיב המאפשר כתיבה פונמית הנותנת ייצוג מלא של כל העצורים והתנועות(. על מנת להפחית את השפעת הסריקה במהלך הקריאה ומאידך לתת ביטוי להשפעת התבניות האורתוגרפיות המאוחסנות בזיכרון, בחרתי מילים אשר תפוצתן גבוהה בשימוש היומיומי ומכאן שקריאתן מהירה ואוטומטית. ניתוח הספקטרוגרמות של החומר המוקלט, שהתבסס על השוואת החלק היחסי של משכי התנועות באופני הכתיבה השונים, הראה כי ישנו הבדל משמעותי סטטיסטית ביחסי עיצור תנועה בין האורתוגרפיה ההברתית לבין האורתוגרפיות בעלות ייצוג נראה של התנועות באותם מבעים. יתרה מזאת, השוואת הכתיב העברי ל HEBRISH מראה כי ככל שגדל ייצוג התנועות בכתיב העברי, קטֶ ן הפער בין אופני הכתיבה. בעקבות מחקרים עדכניים, אשר מצאו השפעה של האורתוגרפיה על הפקת הדבור )מחקרים שהתבססו על מדידת מהירויות תגובה בקריאה קולית וזיקתה לדבור ספונטני(, בדקתי האם בניתוח הדבור הספונטני של הנבדקים יימצאו השפעות דומות של האורתוגרפיה על ההגייה כמו אלה שנמצאו במבחן הקריאה המבוקר. בסיומו של תהליך הקריאה שאלתי את הנבדקים שאלה זהה עליה השיבו בדבור חופשי. השוויתי בדיבור החופשי בין היגוי הברות מוטעמות מסוג CVC שהן בעלות ייצוג תנועה בכתיבה המקובלת לכאלו שנעדרות ייצוג תנועה. התוצאות אוששו את ממצאי המבחן המבוקר כאשר נמצא הבדל מובהק בממוצעי אורך התנועה בין הברות בעלות ייצוג ויזואלי של התנועה לבין אלה ההברתיות. ממצאים אלה וכלל ממצאי המחקר מראים כי האורתוגרפיה העיצורית-הברתית של השפה העברית משפיעה על יחסי משכי תנועה/עיצור בהגייתם של דוברים אורייניים ילידי הארץ.

ממצאי המחקר תומכים בממצאי מחקרים פסיכולינגוויסטיים המתייחסים לתבניות האורתוגרפיה העיצורית-הברתית והשפעתם על תיהלוך הקריאה והדבור (Frost 1995, 2006) . זאת ועוד, מחקר זה מרחיב ממצאים אלה לתחום ההגייה בהפקת הדבור. הממצאים מעידים כי בקריאה אוטומטית )מונחית ע"י תבניות בזיכרון( של השפה העברית 131

ההברתית, בהעדר ייצוג עקבי של התנועות, הפקת הדבור של אורייניים ילידי הארץ מאופיינת בהדגשת העיצורים מחד ובקיצור התנועות מאידך.

שיטת המחקר הנוכחית אשר שימשה לבדיקת השערת המחקר, והמתבססת על יחס משך עיצור/תנועה בהפקת הדבור, פותחת צוהר לבדיקת השפעת האורתוגרפיה על אפיוני הגיה שונים כגון משך, עוצמה וגובה טון. כמו כן, אופן ניתוח הפער בין השפה הכתובה והמדוברת יכול לסייע לבחינת השפעות האורתוגרפיה על הדבור בשפות אחרות ובאופן מיוחד, בשפות בעלות אורתוגרפיה הברתית שהן לוגו-גראפיות )למשל יפנית או קוריאנית(. גישה זו יכולה גם להעשיר תוכניות שיקום של אנשים עם בעיות היגוי, שטף, בעיות הכרוכות בהפקת קול, וכן עשויה לתרום לפיתוח מיומנויות בהקשרים אלה. לבסוף, מחקר זה יכול לסייע בבחינת ההשפעה של השפה המדוברת על הכתובה בפרט, כפי שזה בא לידי ביטוי בשפה העברית ע"י הוספת אמות קריאה בכתיבה באמצעי התקשורת האלקטרונים כגון האינטרנט והטלפון הסלולארי. נטייה זו ייתכן ומקורה בשאיפת המשתמשים להפחית את עמימות השפה הכתובה.

132

הצהרת תלמיד המחקר עם הגשת עבודת הדוקטור לשיפוט

אני החתום מטה מצהיר/ה בזאת: )אנא סמן(:

ֹ_X_ חיברתי את חיבורי בעצמי, להוציא עזרת ההדרכה שקיבלתי מאת מנחה/ים.

_X_ החומר המדעי הנכלל בעבודה זו הינו פרי מחקרי מתקופת היותי תלמיד/ת מחקר.

___ בעבודה נכלל חומר מחקרי שהוא פרי שיתוף עם אחרים, למעט עזרה טכנית

הנהוגה בעבודה ניסיונית. לפי כך מצורפת בזאת הצהרה על תרומתי ותרומת שותפי למחקר, שאושרה על ידם ומוגשת בהסכמתם.

אפריל 9012 דפנה כהן בן שאול 133

העבודה נעשתה בהדרכת

Yishai Tobin, Ph.D.

Professor of Linguistics

במחלקה לספרויות זרות ובלשנות

בפקולטה למדעי הרוח וחברה

134

אורתוגרפיה הברתית ותרבות הדבור בקרב בוגרים דוברי עברית ככלי מקצועי

מחקר לשם מילוי חלקי של הדרישות לקבלת תואר "דוקטור לפילוסופיה"

מאת

כהן בן שאול דפנה

הוגש לסינאט אוניברסיטת בן גוריון בנגב

אישור המנחה

אישור דיקן בית הספר ללימודי מחקר מתקדמים ע"ש קרייטמן ______

ניסן תשע"ג אפריל 3102

באר שבע

135

אורתוגרפיה הברתית ותרבות הדבור בקרב בוגרים דוברי עברית ככלי מקצועי

מחקר לשם מילוי חלקי של הדרישות לקבלת תואר "דוקטור לפילוסופיה"

מאת

דפנה כהן בן שאול

הוגש לסנאט אוניברסיטת בן גוריון בנגב

ניסן תשע"ג אפריל 3102

באר שבע