DEVELOPMENT OF WRITTEN TEXT PRODUCTION...
Transcript of DEVELOPMENT OF WRITTEN TEXT PRODUCTION...
Tel-Aviv University
The Lester & Sally Entin Faculty of Humanities
The Shirley & Leslie Porter School of Cultural Studies
DEVELOPMENT OF WRITTEN TEXT
PRODUCTION OF NATIVE ISRAELI AND
ETHIOPIAN IMMIGRANT SCHOOLCHILDREN
AND ADOLESCENTS: LINGUISTIC AND SOCIO-
CULTURAL PERSPECTIVES
THESIS SUBMITTED FOR THE DEGREE
"DOCTOR OF PHILOSOPHY"
by
MICHAL SCHLEIFER
SUBMITTED TO THE SENATE OF TEL-AVIV UNIVERSITY
AUGUST 2003
This work was carried out under the supervision of
Prof. Ruth Berman and Dr. Gadi Ben-Ezer
Acknowledgments
I could not have completed this work without the intellectual, emotional and
financial support of so many people. I am grateful for the opportunity to acknowledge
them.
I would like to thank first and foremost the supervisors of my dissertation. Prof.
Ruth Berman, who challenged me to do my best work, and taught me all I know
about conducting high-quality research and how to examine one of the most
remarkable human achievements - the written text. Dr. Gadi Ben-Ezer, a gifted
teacher, a skillful advisor and a good friend, whose expertise in the field,
encouragement, and multicultural sensitivity in building "mutual creative spaces"
allowed me to construct new insights.
I thank my Ethiopian colleagues and friends Yirga Erate, Embeat Melesa, Elimelech
Yitzhak and Dr. Anbessa Teferra, who opened the door for me into their ancient and
magnificent culture. Their help in the data collection and its interpretation was
invaluable. I am grateful for their special way of showing me the humbleness of those
who really know.
I am also grateful to my interviewers for their hard work and commitment: Tali Elbaz,
Orit Gaon-Dekkers, Vered Gavish, Irit Mero, Ruth Abramov, Tahel Hason and Ilan
Bloom.
I appreciate the goodwill of the school inspectors, principals, counselors and teachers
in the process of data collection. I am indebted to my dear subjects who willingly
participated in this study.
I wish to express my deep gratitude to my colleagues in the crosslinguistic study for
their help in analyzing the data: Bracha Nir, Nurit Assaiag, Dr. Irit Katzenberger, Dr.
Dalia Cahana-Amitay, Dr. Tsila Shalom, Elisheva Baruch and Sheva Salmon.
Special thanks to my friends Ofra Barak, Tehila Grunwald and Tsafrit Grinberg for
their participation in the interpretation of the data and for their confidence in me.
I wish to acknowledge the support and trust of my friends and superiors at the Centre
for Educational Technology in giving me the opportunity of working with the
Ethiopian children.
I wish to express my sincere appreciation to Dr. Dorit Ravid who taught me about late
acquisition and how to examine it.
I thank Gila Batori for sharing with me her expertise in statistics and also her
friendship and generosity.
I thank Jayne Robinson for typing this study and for her patience and professional
work.
I thank Lisa Amdur and Yael Heffer for their proof-reading and encouragement.
I thank Lea Godelman from the School of Cultural Science for her moral support and
sincere caring.
Finally, my greatest indebtedness is reserved for my family.
I thank my older brother Ram Asher, who has constantly supported and stood by me.
My love and boundless appreciation to my best friend and life companion, Rick, for
being who he is, and for his unfailing love and encouragement not only in my
academic venture.
Many hugs and cheers for my special winning team: my children, Ido, Elad and Yael
for their wisdom, patience and love. Without them I could not have done it at all.
Table of Contents Abstract
Chapter 1: Theoretical Background Introduction 1
1.1 Development of Written Language 2
1.2 Socio-Cultural Perspectives on Language Use 8
Chapter 2: Research Design and Methodology 2.1 General background 20
2.1.1 The original project and pilot research study 20
2.1.2 Overall conceptual framework 22
2.2 Research Population 23
2.2.1 The Ethiopian group 25
2.2.2 The non-Ethiopian group 26
2.2.3 The external control group for the literacy questionnaire 27
2.3 Language teaching in the research populations schools 27
2.4 Data collection procedures 29
2.4.1 Introduction to the task 29
2.4.2 Personal data sheet 30
2.4.3 Text production 31
2.4.4 Literacy questionnaire 31
2.4.5 Interviewer Questionnaire 33
2.4.6 Teacher's evaluation sheet 33
2.4.7 Teachers' reports on the teaching of writing 33
2.5 Transcription 34
2.6 Categories of Analysis and Associated Predictions 35
2.6.1 Text length 36
2.6.1.1 Text length as measured by number of words per text 37
2.6.1.2 Text length as measured by number of clauses 38
2.6.2 Syntactic complexity 39
2.6.2.1 Connectivity or clause-combining as measured by
clause packages 40
2.6.2.2 Linking devices in clause packaging 41
2.6.3 Lexical richness 43
2.6.4 Thematic content 45
2.6.4.1 Thematic content with regard to content words
expressing physical aggression 46
2.6.4.2 Thematic content with regard to family members 47
2.6.4.3 Thematic content with regard to reliance on the video
clip 48
2.6.4.4 Thematic content with regard to affronts to honor 49
2.6.5 Violations of linguistic norms 50
2.6.5.1 Grammatical errors 50
2.6.5.2 Lexical infelicities 52
2.6.5.3 Register mixing 58
2.6.6 Global text construction 58
Chapter 3: Results
Part 1: Results of the written texts
Introduction 63
3.1 Text length 63
3.1.1 Text length in words 63
3.1.2 Text length in clauses 65
3.2 Syntactic Complexity 67
3.2.1 Syntactic complexity in terms of number of words per clause 67
3.2.2 Syntactic complexity in terms of interclausal packaging 69
3.2.2.1 Number of clauses per clause package 70
3.2.2.2 Types of linking devices in clause packages 70
3.3 Lexical Richness in terms of Lexical Density 72
3.4 Analysis of Thematic Content 78
3.4.1 Reference to the video 79
3.4.2 Reference to physical aggression 79
3.4.3 Reference to the location of conflicts 79
3.4.4 Reference to affronts to honor 80
3.5 Violation of Linguistic Norms 80
3.5.1 Grammatical errors 80
3.5.2 Lexical infelicities 82
3.5.3 Register mixing 84
3.6 Global Text Construction 86
3.6.1 Text Components 86
3.6.2 Text component length 88
3.6.3 Levels of organization of information 89
3.7 Summary of Results of the Written Texts 91
Part 2: Results of Literacy Questionnaire
3.8 Introduction 93
3.9 Success in Hebrew Language Studies 95
3.9.1 Teacher evaluations of language achievements 95
3.10 Writing Proficiency 96
3.10.1 Teacher evaluations of narrative writing 96
3.10.2 Teacher evaluations of expository text writing 96
3.10.3 Teacher evaluations of questionnaire writing 96
3.10.4 Student evaluation of narrative writing 97
3.10.5 Student evaluation of expository writing 97
3.10.6 Student evaluation of questionnaire writing 97
3.11 Writing Abilities and Attitudes 98
3.11.1 Attitudes to and command of writing activities 99
3.12 Level of Difficulty of the Various Writing Activities 99
3.13 Proficiency in Amharic of the Ethiopian Students 100
3.14 Attitudes to Writing 100
3.14.1 Why writing is important 100
3.14.2 What makes a good writer? 101
3.15 Language Attitudes 103
3.16 Interviewer Information on the Questionnaire as a Task 104
3.16.1 Time required to fill out the questionnaire 104
3.16.2 Need for help 105
3.16.3 Task difficulty 106
3.17 Parental Background 106
3.17.1 Parents’ country of birth 106
3.17.2 Parents’ level of formal education 107
3.18 Home Literacy 108
3.18.1 Languages of reading and writing 108
3.18.2 Home-based reading and writing activities 108
3.18.3 Sources of assistance 110
3.19 Summary of the Results 110
Chapter 4: Discussion 4.1 Overview of the Research Findings 114
4.1.1 Patterns shared by the Ethiopian and non-Ethiopian subjects 114
4.1.2 Patterns shared by low and middle-class populations 115
4.1.3 Patterns shared by low SES subjects from poor educational
systems and low SES native-speaking subjects in
well-establishedschools (comparison to Salmon, 2003) 117
4.1.4 Effect of schooling in immigrant low SES subjects from
poor educational systems and immigrant subjects attending
well-established schools (compared to Rabukhin, 2003) 118
4.1.5 Different language skills of the Ethiopian and
non-Ethiopian subjects 118
4.1.6 Different patterns in the three populations 120
4.1.6.1 Thematic themes 120
4.1.6.2 Avoidance of talk 121
4.1.6.3 Cultural conventions of structures 123
4.1.7 Home literacy 124
4.1.8 Attitudes towards writing 125
4.2 The Four Themes 127
4.2.1 More is not necessarily better 127
4.2.2 Early distinctness late command of genres 132
4.2.3 Research involvement of community representatives 133
4.2.4 What makes writing communicative? 135
References 137
Appendix 165
Appendix 167
List of Tables Table 1: Breakdown of research population by origin,
sex and grade level 24
Table 2: Means and standard deviations number of words per text,
by age, genre and origin 64
Table 3: Means and standard deviations of clauses per text,
by age, genre and origin 66
Table 4: Means and standard deviations of words per clause,
by age, genre and origin 68
Table5: Means and standard deviations of clauses per CPs,
by age, genre and origin 70
Table 6: Means and standard deviations of the three types
of inter-clausal links, by age, genre and origin 72
Table 7: Mean number of content words and standard deviations
by age, genre and origin 73
Table 8: Mean number of content words per clauses
and standard deviations by age, genre and origin 75
Table 9: Means and standard deviations of lexical density by age,
genre and origin 77
Table 10: Total amount of grammatical errors by age and origin 81
Table 11: Percentages of the three common grammatical errors
by age and origin 81
Table 12: Number of lexical infelicities by age,
genre and origin 83
Table 13: Number of lexical infelicities in the original study
by age and genre 84
Table 14: Number of the sub-categorizations of the lexical
deviations by population 84
Table 15: Number of cases of register mixing by age,
genre and origin 85
Table 16: Number of cases of register mixing in the original study
by age and genre 85
Table 17: Numbers of texts containing each of three text
components by age, genre and origin 88
Table 18: Means and standard deviations of CPs
in each text component by age, genre and origin 89
Table 19: Numbers of texts including Linear (L), Partially
Hierarchical (PH), and Fully Hierarchical (FH) levels of
information organization by age, genre, and origin 91
Table 20: Numbers of texts of the original data including Linear (L),
Partially Hierarchical (PH), and Fully Hierarchical (FH) levels
of information organization by genre and age 91
List of Figures Figure 1: Comparison of mean number of words
by genre and population 65
Figure 2: Length of texts in number of clauses
by genre and population 67
Figure 3: Comparison of mean number of words per clause
by genre and population 69
Figure 4: Comparison of mean number of content words
by genre and population 74
Figure 5: Comparison of means of content words per clause
by genre and population 76
Figure 6: Comparison of means of lexical density
by genre and population 78
" I ask of G-d only one thing: Do not let me become
indifferent to the wilting of those who had to part and
live a different culture"
Leon Gieco
Abstract
The forces that affect the linguistic behavior of members of an immigrant group,
who are in contact with other ethnic groups became an important subject of research
due to contemporary demographic changes and to a great increase of immigration in the
modern world. The major question of this study is what is involved in becoming
"literate" in different environments? This study focuses on socio-cultural effects on the
development of written discourse abilities of students whose background cultures and
home languages differ from the receiving society and its school system. The overall
goal of the study is to understand how schoolchildren from different ages and from
different cultural and socio-economic backgrounds write monologic texts.
The study analyzes written texts produced by two groups of Hebrew-speaking
schoolchildren, who share similar low socio-economic status (SES). The first group
consists of students of recent-immigrant Ethiopian parentage, while the second consists
of students of non-recent-immigrant and non-Ethiopian background. These two groups
were further divided into three levels of schooling: grade-school (Grade IV), junior high
school (Grade VII), and high school (Grade XI). Subjects were asked to write two texts,
a personal-experience narrative and an expository discussion on the topic of
interpersonal conflict. They were also asked to fill out an informative questionnaire
probing writing experiences, attitudes to language and writing and home literacy. The
focus of this study is thus the development of literacy and language use in relation to
socio-cultural background, by examining the ability of schoolchildren to produce
different kinds of written texts. The study aims to tease apart the variables of SES
background, origin, age and genre, in terms of three types of interactions: 1) between
linguistic forms and discourse functions, 2) between top-down global text organization
and appropriate use of linguistic forms at a local bottom-up level of text construction,
and 3) between background literacy and text production abilities.
This study does not support the popular view that equates economic poverty and
minority or immigrant background with low literacy interests and achievements. Rather,
its underlying assumption is that the differences in the language use of students from
different SES backgrounds are confined to specific literacy-related activities rather than
to general language proficiency. Thus, the different populations would demonstrate
heterogeneous literacy skills, and that variety in performance would be both within-
group as well as across-group. I assumed that more literate writers would reveal greater
sensitivity to differences across text types and that this would be expressed in the
linguistic forms they select in each case. On the one hand, subjects across populations
should rely on shared developmental patterns, expressing the common socio-cognitive,
maturational and experiential base underlying text construction, like strategies for
global text construction, information organization, and distinguishing between narrative
and expository texts. On the other hand, these populations were expected to display
different levels of proficiency in their use of morpho-syntactic and lexical
constructions, and particularly in the thematic content of the text they write. In addition,
this study supports the idea that each writer is an exponent of his/her own culture.
Hence, the different populations were also expected to differ in their conventional
linguistic norms, cultural values and attitudes that characterize their unique literate
usages.
The present study applied a multidisciplinary approach aiming at integrating a
range of linguistic, socio-cultural, and educational perspectives. The study adopts both
quantitative measures of the kind associated with research in education and the social
sciences combined with more qualitative, culturally motivated analyses, on the one
hand, and structural analyses of linguistic forms, on the other. This study also uses a
multilayered approach on the levels of a word, a clause, a clause-package, and the text.
Such an approach suits the linguistic research of the written text since it combines
bottom-up analysis of lexical items with top-down analysis of the overall structure of a
text.
Results of this study derive from two kinds of analyses. The first is a detailed
linguistic analysis of 192 written texts of three age groups of Israeli low SES
Ethiopian and non-Ethiopian schoolchildren. The written texts, about the same topics,
are compared with a corresponding set produced by an external control group of
Israeli schoolchildren of the same age and levels of schooling from middle-class
“mainstream” backgrounds, elicited under similar conditions in the framework of a
large-scale crosslinguistic project (Berman & Verhoeven, 2002). Following the
elaborately devised and carefully tested crosslinguistic study in overall conception
and research design allowed us to achieve a maximal comparability of similarities and
differences in language proficiency across the different populations. The second set of
analyses concerns similarities and differences in language proficiency, in school and
home literacy-related activities, and in attitudes towards writing that may affect the
language use of students from different backgrounds. Information on background
factors was gathered by means of a specially devised literacy questionnaire. These
socio-cultural and demographic variables were compared to Israeli-born high
achievers of high SES background in the same age groups and school levels.
The analyses are the means for characterizing what is involved in "becoming
literate" in these environments, and for constructing "a profile" of linguistic literacy
(Ravid & Tolchinsky, 2002) for each of the research populations. Such insight into
the children's writing development is indispensable, especially in the case of the
recent immigrant Ethiopian children, since there are no carefully designed,
linguistically based studies, which have considered the ways in which these children
express themselves in writing. To the best of my knowledge, this research is the first
developmental study on their acquisition of writing and home literacy. Moreover, this
study offers research in socio- linguistics and education with the first available set of
computerized authentic texts written by Israeli schoolchildren of Ethiopian parentage.
The computerized data on the written texts of the non-Ethiopian students enlarges the
scope of non-computerized sets on the language use of Israeli students from
disadvantaged backgrounds (Erlich, 2001; Salmon, 2003).
The results indicate that immigrant and native-speaking Israeli students from
similar low SES backgrounds manifest similar writing behaviors across age and text
types (personal experience narrative and expository discourse). Nor does analysis of
their home and school literacy-related activities, their language achievements, and
attitudes towards writing reveal significant difference. The comparison of these two
groups with middle class students shows that varied domains including canonical
narrative construction, global expository structure, organization of information, and
lexical density were found to be similar across populations. It seems that all these
abilities represent shared developmental patterns, which underlie text production.
In comparison, when text length is neutralized, middle class students produce
more words per clause across age (indication of greater syntactic complexity, Ravid,
2003). The low SES students produce longer texts containing lots of verbal output,
indicating that they might suffer from a pragmatic failure caused by their lack of
confidence in their ability to get their meaning across. Thus, "verbosity" is not a sign
of linguistic proficiency of low SES students, but rather of lack of confidence in their
communicative competence. SES was also found to be sensitive to lexical richness,
since low SES students not only used fewer words per clause and fewer content words
per clause across genres, but also displayed less proficient usage in lexical selection,
in discourse conventions and in register consistency compared with the middle class
students of the original study. Thus, the difference between low and high SES
populations should not be attributed to the relative scopes of their lexicons, but rather
to the level of their communicative competence, and/or to their awareness for
regulative norms of writing.
The results also show that culture is an important factor in constructing the
underpinnings of linguistic development, and that cultural knowledge serves as a rich
context for interpretation and understanding of specific elements of both form and
content in the texts produced by schoolchildren and adolescents. All the three
populations differ with regard to thematic content. It seems that students choose to
express the unique cultural norms of their own community in their writing. For
example, half of the Ethiopian students chose to write on affronts to their honor in
their narratives; this is a theme that reflects an important norm in Ethiopian society.
With regard to home literacy and attitudes towards writing, the high SES
students reported on more varied and complex literacy activities in their homes, and
they define writing as an abstract process, aimed at self expression and cognitive
enrichment. In contrast, low SES students observe mainly letter-writing in their
homes and define writing as a notational system, aimed at achieving practical goals.
This study concerns the general relationship between socio-cultural
background factors and the developing literacy. It shows that in any environment
extensive experience with the production of a variety of types of texts, especially with
expository academic type of writing, is crucial for the development of linguistic
literacy of schoolchildren. Thus, a clear implication for the language educators is that
explicit instruction of how to construct well-organized expository texts, starting in the
beginning of high school might encourage the development of such abilities, not yet
mastered by most of adolescents. Students of less literate, less educated backgrounds
were found to have more difficulties in developing their communicative proficiency.
Thus, register and lexical selections could serve as criteria for evaluating the language
use of low SES students, and as good diagnostic tools of text maturity.
More researches on the language use of various populations are needed in
order to find out what it means to become literate in different environments, and how
the written texts reflect the unique norms, attitudes and literacy experiences of the
members of these different cultural socio-economic groups.
1
Chapter 1: Introduction and Research Introduction
This study is concerned with the development of written discourse abilities of
schoolchildren from different backgrounds and age groups. The overall goal of the
study is to understand how children of different ages and from different cultural and
socio-economic backgrounds write monologic texts. The study analyzes written texts
produced by two groups of Hebrew-speaking schoolchildren who share similar low
socio-economic status (SES). The first group consists of children of recent-immigrant
Ethiopian parentage, and the second of children of nonrecent-immigrant and non-
Ethiopian background were at three levels of schooling: grade-school (Grade IV),
junior high school (Grade VII), and high school (Grade XI). Subjects were asked to
write two texts, a personal-experience narrative and an expository discussion, both on
the topic of interpersonal conflict. They were also asked to fill out an informative
questionnaire probing writing experiences, attitudes to writing, and home literacy.
The focus of this study is thus the development of literacy and language use in
relation to socio-cultural background, by examining schoolchildren's ability to
produce three different kinds of written texts. From a developmental perspective, the
study is in the domain of “late acquisition”, since it examines the language production
of schoolchildren at ages ranging from nine to seventeen years, rather than of pre-
schoolers. Schoolchildren at these ages are engaged in a cumulative process of getting
familiarized with the written language in addition to further developing the
complexity and sophistication of their oral language use. They acquire a large lexicon
and a complex grammar, and produce different types of texts by flexibly using
different morpho-syntactic resources for diverse communicative purposes (Berman,
1997a; Blum-Kulka, 1997; Gillis & Ravid, in press; Nippold, 1998; Ravid &
Tolchinsky, 2002). From a socio-cultural perspective, the study compares the texts
produced by two groups of children of Ethiopian and non-Ethiopian background
(henceforth the "research" groups) with a corresponding set of texts produced by a
group of Israeli schoolchildren of the same age levels and of schooling from middle-
class, well educated home backgrounds (the "original study" comparison group,
Berman & Verhoeven, 2002).
The study yields two sets of analyses: a careful linguistic analysis of the form
and content of 192 narrative and expository texts, and of 114 literacy questionnaires.
2
The research and the comparison groups wrote about the same topics, and were elicited
by the same set of procedures and the same overall research design as in a large-scale
cross-linguistic study of text production abilities of schoolchildren and adults (Berman,
1996; Berman & Verhoeven, 2002). The main goal of the elaborately devised and
carefully tested crosslinguistic study was to understand how schoolchildren and
adolescents construct monologic narrative and expository texts in writing and in speech
in comparison to highly literate well-educated adults. The present study considers only
the written texts of the research groups compared with the original study in which
spoken texts were also examined (for reasons described below in Section 2.1.1).
Analysis of the parallel sets of materials made it possible to examine the similarities and
differences in language proficiency of the different populations.
The second set of analyses concerns school and home literacy-related activities
and attitudes towards writing yielding a description of the socio-linguistic variables
affecting text-production abilities. The two analyses provide the means for
characterizing what is involved in "becoming literate" in different environments. As yet,
there are no carefully designed, linguistically based studies concerning the ways low
SES students in general and recently immigrated Ethiopian children in particular,
express themselves in writing.
Relevant literature is reviewed here along the two main axes of this study: Research
on the development of writing abilities and linguistic literacy (Section 1.1) and on
socio-cultural aspects of literacy (1.2).
1.1 Development of Written Language This study, as noted, is concerned with the development of written discourse
abilities of schoolchildren from different backgrounds. Regarding modality or the mode
of language production, a wide range of research shows that even young schoolchildren
are able to distinguish between written and spoken language. Recent research
comparing the written and spoken texts produced by schoolchildren compared with
adults reveals an interesting developmental trend. Among grade schoolchildren, what
they write is anchored strongly on how they speak, as shown by a Hebrew-language
study of the narrative and argumentative texts produced by subjects in the same age-
groups as the present study (Berman in press, Berman & Ravid, 1999). On the other
3
hand, more in-depth study of gradeschoolers (aged 9 to 10 years old) producing similar
texts in different languages shows that their written language is already differentiated
from those of their spoken usage when producing narratives and expository discussions
on parallel topics (Danielewicz, 1984). For example, in writing they avoid use of
interactive discourse markers (Berman & Ravid, in press), and they use more complex
syntax and nominal expressions (Ravid & Cahana-Amitay, in press).
With development, this trend for differentiation between the spoken and written
language grows more marked by junior high school, although it is consolidated and
consistently evidenced only from high school (age 15 and above), while among
educated, highly literate adults, the earlier pattern is reversed, so that mastery of the
written language comes to have an increased impact on how people speak (Strömquvist,
1996). Thus, it seems that only in high school students start to use thinking for writing
in their written discourse. Strömquvist and Wengelin (2003) found that only 15 year
olds spent more time on discourse-initial planning, and began using more frequent and
differentiated intersentential connectivity. Strömquvist (1996) showed that high school
students avoided a characteristic of the spoken language in comparison to the younger
children who included this characteristic in their written texts. Gayraud (1999) showed
that French-speaking students, mainly in high school, consistently relied on and marked
lexical, morphological and syntactic features typical of more formal styles in their
written texts (Gayraud, Jisa & Viguié, 1999). Ravid and Zilberbuch (2003) showed that
the written texts of Israeli highschoolers were denser than their spoken texts, lexically
and syntactically, as measured by number of novel N-N compounds (prixat^ha tapuzim-
smixut- 'orange- blossoms- adjacency'), and denominal adjectives (xashmal-i, 'electr-ic')
per clause. In comparison, adults' writing is distinct even from that of highschoolers,
and is characterized by long discourse initial planning pauses, extensive content editing,
great control of large chunks of discourse, and embedded free direct speech designed to
facilitate inferences about the identity of the speaker (Strömquvist et al., 2003). Berman
and Ravid (in press) showed that across age groups, spoken texts are "less informative",
while the written texts contain more text oriented, extra-contentive material.
With regard to genre, this study is concerned with the abilities of schoolchildren
to write three different types of texts. The narrative and expository genres are defined as
relatively stable text regularities, which accomplish different social or communicative
functions (Kress, 1993; Nicolopoulou, 1997) and reflect two distinct “ modes of
4
thought” (Bruner, 1986). Narration focuses on recounting events from fictive
frameworks or personal experiences, and so relies on both episodic and semantic
memory (Tulving, 1972). Expository texts, in comparison, rely heavily on semantic
memory for the expression of thoughts and ideas, and so require academic-type
language proficiency (Nelson, 1988; Cummins, 1981a). Expository texts lack the single
temporal sequencing factor as the major principle of organization of information.
Moreover, content and structure are inextricably interwoven in expository type of texts.
As a result, narration and exposition differ developmentally, since the narrative use of
language emerges well before the expository (Katzenberger, submitted a). The two
types of texts also have different basic principles of discourse organization (Berman &
Katzenberger, submitted). The function of temporal sequencing is crucial for narrative
structure (Hatav, 1975; Labov, 1972), whereas informative and other non-sequential
elements form the core of the expository structure (Giora, 1990). The linear hierarchical
structure of narratives consists of a setting plus episode(s) plus coda. In comparison, in
expository texts the segments are ordered as introduction to middle to ending. The
higher level narrative “action structure” consists of an initiating event(s), attempt and
resolution (Shen, 1985). In comparison, expository texts are characterized as having at
least two types of discourse elements: core or nuclear propositions, and satellite or
supplementary material (Fox, 1978; Matthiessen & Thompson, 1988). These elements
compare broadly to Britton’s “move-on”, which introduces a new topic, “expand”
which develops the already introduced topic, and “unitize” which summarizes
information previously mentioned (Britton, 1995). Discourse stance is also used in
order to describe the differences between the types of texts along three interacting
dimensions of "discourse stance": 1) orientation- sender, text, recipient; 2) generality of
reference and 3) attitude-epistemic, deontic, affective (Berman, Ragnarsdףttir &
Strömquvist, 2002). For example, an oral narrative recounting personal-experience
events expresses a more communicatively motivated sender/receiver orientation, more
specific reference to people and objects and more affective attitudes. In comparison, a
written expository text is more text-based in orientation, more general in reference,
more cognitively epistemic, and more objective in attitude. These different perspectives
of the speaker-writer are expressed in the selection of the thematic content, the
rhetorical expressions, lexical items and grammatical categories (Berman, 2003).
There is a rich body of research on children's text production abilities in the
domain of oral narration. This is perhaps because the narrative is universally familiar,
5
type of discourse to which even young pre-schoolchildren are exposed in oral and
literate cultures alike. A range of studies examining different kinds of narratives show
that even children as young as three years already have scriptlike event-structure
knowledge, on which they base their story production (Nelson, 1986). A large-scale
study of personal-experience narratives of English-speaking children aged 3-10
(Peterson & McCabe, 1983) and of make-believe narratives of children from age 4
years (Nicolopoulou, 1996) showed that by late preschool age, children had some
command of global narrative structure. Other research based on descriptions of picture
series produced by Hebrew speaking children aged 4-6 and 10 years old (Katzenberger,
1994), and of children aged 7, 10 and 15 in English, French, German, Dutch and
Chinese (Hickmann, 1995) showed that a narrative schema was well-established by age
9-10 years. This is confirmed by a crosslinguistic comparisons of children's developing
text production ability (Berman & Slobin, 1994), which compared the oral, picture book
based narratives of children aged 3 to 9 with those of adults in five different languages
including Hebrew. Thus, these studies show that by late preschool, children show some
command of narrative schema, and construct a well-formed oral narrative text and that
this ability concordats across different narrative contexts and elicitation processes by
age 9 to 10, the age of the younger group in the present study (Berman, 1995). All
preschoolers begin to use the typical devices available in their native language for the
construction of cohesive discourse, but they do not master adult use of such devices and
structures at discourse level until late, despite their frequent early uses of "correct"
forms. The development of children's oral narrative activity constrained by factors such
as sociocultural contexts, the dynamics of every individual's cognitive structures and
communicative purposes.
Less research is available on children’s written narratives. Written narratives
both differ from and share properties with oral narratives and other types of written
texts (Freedman, 1987; Purcell-Gates, 1988; Shuman, 1986). Current research
indicates that once children have established a narrative schema, they will transpose
story production abilities from their spoken usage to their writing as was shown in the
crosslinguistic "original" study (Tolchinsky et al., 2002)
Far less research is available on children's expository texts in general and on
their ability to differentiate between the two genres in particular. Analyses of children’s
text production in Hebrew show that by early school age children differentiate between
narrative and expository texts (Berman & Nir, submitted; Berman & Ravid, 1999;
6
Peled, 1996; Peled & Blum-Kulka, 1992). Findings from the crosslinguistic study show
that although even 9 to10-year old grade-schoolers distinguish between narrative and
expository texts in both thematic content and linguistic use, it still takes them until high
school to master the global construction of the expository text and to use a rich range of
linguistic domains in grammar and the lexicon appropriately and flexibly (see papers in
special volume of the Journal of Written Language and Literacy, edited by Berman and
Verhoeven 2002, and also for Hebrew Ravid & Cahana-Amitay, in press). Thus, while
children begin to differentiate between genres at an early age, a lengthy developmental
course is needed for them to reach mature control over the distinct forms of expression
used by proficient language users for inter-genre differentiation (Berman & Nir,
submitted).
A third, highly specific type of text, which differs markedly from both narration
and exposition is the questionnaire. While narratives concern events ordered along a
timeline and expository texts deal with ideas, the essential components of a
questionnaire are data and facts. This type of discourse is an important part of modern
bureaucratic life. It was used in this study as a means of providing information about
socio-linguistic variation in the different research populations. Probing the different
linguistic, cultural and literacy-related backgrounds was done by means of a literacy
questionnaire. To the best of my knowledge how people perform this task has not been
the subject of special research, certainly not from a developmental point of view.
Following Ravid and Tolchinsky (2002) in conception of linguistic literacy,
"literate language users" are defined here as people who possess a rich and varied
linguistic repertoire, encompassing a wide range of registers and genres, together with a
rhetorical flexibility to adapt this repertoire to meet different communicative contexts
and goals. Thus, being literate means being able to access all kinds of written texts
constrained by different communicative purposes, from narrative to expository
discourse, from newspaper reports to questionnaires, from technical manuals to literary
reviews, from Internet chats to poetry (Paltridge, 1997). This familiarity with the
written language promotes the awareness of literate language users to the implicit
structure of their language, and the construction of an internal model of thinking about
their own spoken and written systems (Olson, 1994). This meta-linguistic level of
language awareness leads to the perception of the written language as both a notational
system and as a special discourse style (Bialystock, 1993; Ong, 1992). Moreover, this
meta-linguistic development enables linguistic knowledge to become more complex,
7
dense, and accessible with age (Karmiloff-Smith, 1992). Thus, literate language users
adequately and flexibly to employ a large repertoire of lexical and morpho-syntactic
structures taking into consideration discourse type and goal (Gillis & Ravid, in press;
Saltzman & Reilly, 1999). In conclusion, being literate across modularity and across
genres means being familiarized with writing and the written language, and making a
clear distinction between the types of texts by controlling the thematic content and the
linguistic means of expression suited to different discourse contexts.
Underlying this study is a form/function approach to text analysis (Slobin,
1996). The term “form” here refers to linguistic devices such as bound morphemes,
lexical expressions, and syntactic constructions, while "function" refers to the role
played by these devices in the course of text production (Berman & Slobin, 1994).
Research on children’s language acquisition indicates that a large range of linguistic
forms emerge early in development. Nevertheless, a lengthy developmental route is
required until children acquire mastery of both a full repertoire of linguistic forms and
the ability to use them appropriately and flexibly in extended discourse (Berman,
1997b; 1999a; Berman & Slobin, 1994; Peled, 1996; Levin, Ravid & Rappaport, 1999;
2001). With age, the range of forms used for any particular function becomes not only
more elaborate but also more consistent and more appropriate to a specific
communicative setting (Berman, 1996b; Berman & Slobin, 1994; Hickmann, 1995). In
the present context, “text production abilities” refers not only to the interaction between
linguistic forms and discourse functions, but also to the interaction between top-down
global text organization and appropriate use of linguistic forms and rhetorical devices at
a local bottom-up level of text construction. Developmental studies in the domain of
“late acquisition” (Karmiloff–Smith, 1986; Nippold, 1988, 1998; Ravid & Avidor,
1998) show that it takes until well beyond early school age before children deploy a full
repertoire of linguistic forms, as appropriate rhetorical options, to meet a range of
narrative, expository and informative functions.
The goal of this study is to examine the process of becoming literate.
Consequently, an important focus of this study is the extent to which, and by what
means, schoolchildren distinguish between different types of texts as a function of age.
Their texts, both in narration and exposition are expected to be more differentiated with
age, and to become increasingly more “written-like” as a function of increased literacy.
8
1.2 Socio-Cultural Perspectives on Language Use
The accepted view equates economic poverty and minority immigrant
background with low interests and achievements. The present study does not support
this view. The perspective adopted here is that parents and children from low SES or
immigrant backgrounds do not constitute uniform or homogenous entities, but that
they reveal considerable within-group variation. The present study aims to show that
the language of less-advantaged children is not necessarily "worse" but it is different
to that of their better-educated peers from well-established backgrounds.
The research literature on this topic reveals contradictory approaches and,
consequently also findings. The most extreme orientation is exemplified by the early
work of the British sociolingustic Basil Bernstein (1960). Many other studies appear
to confirm the lower level of language used by children of less-advantaged SES
backgrounds. For example, Hart and Risley (1995) found widening gaps between high
and low SES children in vocabulary size over the first three years of life. Anglin
(1993) also noted that low SES children had smaller vocabularies on the average, but
he pointed out that they were not significantly less capable of morphological analysis
and composition than upper-SES children. Chall and Snow (1988) showed that low-
income children scored less on fourth grade vocabulary tests in reading compared to
their higher SES peers. In other domains, using standardized national tests,
Whitehurst (1997) showed that children from low-income backgrounds began formal
schooling behind their peers with regard to general language ability. The U.S.
National Center for Statistics (1999) showed that around two thirds of fourth graders
of Hispanic and Afro-American low SES backgrounds were below the basic level of
language achievements. Relatedly, Dubow and Ippolito (1994) showed that low SES
children were at risk for later reading difficulties and academic achievements.
There are also Israeli researchers who claim that low SES children attain lower
scores in reading and writing tests in comparison to high SES children both in pre-
school (Smilansky & Shefatia, 1979), and at school (Eshet, 1980; Minkovich, Davis
& Bashi, 1977). Other Israeli studies suggest that appropriate use of formal register
of high-educated language is a major source of difference between children from
different social backgrounds (Davis, 1978; Kais, 1979; Shtal, 1977). Register is used
here as “a general cover term for situationally defined varieties” (Biber, 1995), while
to be register-appropriate means revealing linguistic awareness and sensitivity to
9
communicative setting and social circumstances when producing a piece of discourse.
The Israeli-based studies claim that children of more educated backgrounds use high-
level language in situations which require a more formal style or register, while
children from less educated backgrounds mix high register language with colloquial,
everyday speech in such circumstances. For example, Kemp (1984) showed that a
major difference between Hebrew speaking seventh-graders from disadvantaged
compared with well-established backgrounds was the inappropriate application of
register in an error recognition test.
Other studies attempt to explain different literacy achievements of low SES
children by examining a range of different background factors of these children in and
of themselves. For example, some researchers showed that differences in the
children's reading and writing ability or school readiness are a function of the
economic level of their parents (Alexander & Entwisle, 1988; National Assessment of
Educational Progress, 1998; White, 1982; Wigfield & Asher, 1984). A ten-year
longitudinal study which examined SES, level of parental education, status and level
of family income of forty families from low and middle SES showed that low SES
children had poorer achievements on reading, writing, spelling and mathematics than
their higher SES peers (Walker, Greewood, Hart & Carta, 1994). Holden (1997)
reviewed studies that showed that children of affluent parents attained higher
academic achievements than low SES children.
In contrast to these "child-output" based studies is a broad range of research
that focuses on "child-input" factors such as parental attitudes and home literacy-
related activities. Thus, Dickinson and Tabors (1991; 2001) claimed that different
home and pre-school environment variables, such as adult-child verbal interactions,
home support for literacy, exposure to rare words and extended teacher discourse in
pre-schools could explain the difference in the receptive vocabularies or
decontextualized language skills of children from different socio-economic
backgrounds. Along similar lines, Pflaum (1986) also emphasized the relation
between the different styles of parent-child communication in different SES
backgrounds and linguistic achievements. She added that low SES parents gave less
encouragement to their children’s linguistic initiatives compared with high SES
parents. Tough (1982) described adult-child linguistic interactions in low SES
backgrounds as consisting of short and less complex sentences, having few
explanations, and less use of the language for alternative solutions, for descriptions
10
and for expression of imagination. Walker, et al (1994) showed that low SES parents
play less language games with their children and demand less linguistic productions.
Another study documented pre-schoolers' exposure to infrequent vocabulary, and
found that 3- to 4-year olds from low-income families, who heard and used such
vocabulary at home tended to score high on a standard vocabulary test at age five
(Beals & Tabors, 1993). Storch & Whitehurst (2001) claim that literacy environment
and parental characteristics such as parental expectations for their child’s school
success and parental reading behaviors account for approximately 40% of the
variance in pre-school "outside-in" skills, such as vocabulary about the printed
language, understanding of narrative story structure, and conceptual knowledge. Other
studies emphasize the importance of parental (especially maternal) intelligence,
education and native language on the child’s language and literacy development
(Bowey, 1995; Eshel, 1979; Payne, Whitehurst & Angell, 1994; Plomin, DeFries &
Fulker, 1988; Wigfield & Asher, 1984). Other researchers observe that children,
whose home literacy environments are lacking with regard to shared reading activities
and printed materials are at risk for later reading difficulties (Allen, Cipielewski &
Stanovich, 1992; Scarborough & Dobrich, 1994). Heath (1983) showed that less
literate mothers from low SES expose their children to fewer reading and writing
situations aimed for learning and pleasure, or even aimed at achieving instrumental
everyday needs or work necessities. Snow, Barnes, Chandler, Goodman and Hemphill
(1991) emphasize the relationship between home and school factors for language and
literacy development of children from varying socio-economic backgrounds. As home
factors, they point to the mother's expectation for the child and her level of education.
The contact between parents and teachers regarding academic issues also affects the
child’s language. Gregory (1999) showed that parents of successful literate adults
raised in a poor neighborhood in East London passed down literacy-related
knowledge and encouraged their children to learn from other mediators of literacy.
The research on the effect of school resources on low SES students' outcomes
has also yielded conflicting findings. On the one hand, McCarty (1989) reports that
instability of the educational staff and limited success of the Navajo long-term
educational program were due to the poverty of the area and the dependence on
fluctuating federal money. In comparison, Hanushek (1990) claims that there is no
systematic relation between school resources (e.g., school administration, facilities,
teacher education, teacher-student ratios) and educational outcomes.
11
Less research is available on the writing development of children from
different SES. In Israel, Levin, Share and Shatil (1996) built a developmental scale for
the evaluation of writing of pre-schoolchildren. They showed that low SES children
write at a lower level than high SES children. In Argentina, Ferreiro and Teberosky
(1982) found that middle-class preschoolchildren had an advantage over low SES
preschoolers with regard to first name identification and writing.
Other researchers focused on the mediation of literacy as a factor in their
children’s literacy development. Aram (1998), who examined the nature of low SES
mothers’ mediation of writing, their level of literacy and the children’s literacy-
related environments, showed that the first factor contributed most strongly to the
differences in independent functioning of children regarding literacy in the
kindergarten.
Along very similar lines, research on the second language (the language of the
"dominant" " receiving" society) of immigrant children shows that minority children
attain less on language tests than their native-speaking peers. A study of Canadian
immigrants, schooled entirely in English since arrival in Canada, shows it took them
approximately 5-7 years to achieve grade norms on achievement tests comparable to
their native-speaking peers (Cummins, 1981b). Other studies show that it took
immigrant children 4-10 years, depending on their age on arrival, and level of
previous schooling in their country of origin, to achieve norms comparable to native
English speakers on standardized achievement tests (Collier, 1987; Collier & Thomas,
1988; Thomas & Collier,1997). Other studies carried out in Sweden and the U.S.A.
suggest that immigrant students with interrupted educational backgrounds and limited
world knowledge may experience academic difficulty at the post-secondary level of
language (Bosher & Rowekamp, 1992; Skutnabb-Kangas & Toukoman, 1976). Other
studies have tried to explain the gap in academic achievements by the immigrant
children's lack of exposure to, or lack of understanding of, the vocabulary and
context-specific language needed to perform the more demanding tasks required in
academic courses (Short & Spanos, 1989). In research that is more related to the
present study, Akinחi, Jisa, and Kern (2001) found a delay in the acquisition of L2
linguistic devices for encoding macro-structures in the narratives of school-aged
Turkish children compared to monolingual French children. They suggest that this
relative delay is due to differences in the amount of exposure to literacy-related
activities. The background research to this study pointed to differences in register in
12
terms of use of past tense typical of written versus oral French, in comparing the oral
narratives of 7-year old monolingual French children with Turkish-French bilingual
children (Akinחi, 1999), which they explained by the different quantitative and
qualitative exposures to written French of the two populations. Verhoheven (in press)
reviews studies of "frog story" picturebook narratives of Arabic-Dutch and Turkish-
Dutch children compared to their native Dutch peers (Aarssen,1996; Bos, 1997). The
subjects, minority children aged 4-8 years old of low SES background had a lower
level of narrative production skills than their native Dutch peers. Arabic-Dutch
children also used a more restricted register in use of temporal adverbials in
comparison to monolingual Dutch children, who showed a more complex use of
adverbials, conjunctions and particles. Obviously, the Arab-Dutch children were
exposed to far more restricted channels of literate Dutch input, which might account
for their less developed knowledge of the rules of subordination and the inversion of
basic word order.
Verhoeven (1991) also examined socio-cultural factors in an attempt to find
those which effect L2 proficiency. He identified eight positive factors that predict the
level of Dutch grammatical skills of Turkish children at the age of six. Some relate to
home environment such as the parents' attitude towards the Dutch culture, the parents'
cultural behavior, the extent of L2 literacy within the family. Others relate to
interactions with the immediate surrounding such as the extent of peer interaction in
L2 and of caretaker interaction in L2. Another factor concerns the contact between
parents and the educational institution such as the extent of involvement of the parents
in day-care activities. Leseman and De Jong's study (2001) of low SES minority
children compared to middle-class children in the Netherlands revealed an impact of
home literacy practices, measured by opportunities for literacy related interactions, on
vocabulary and word decoding development.
Incongruence between immigrant cultural practices at home and in school is
another line of explanation for the children’s limited skills in the second language.
Boyd and Nauclיr (2001) showed that 5-6 year old Turkish children are generally not
invited to contribute much to the co-construction of a story with their mothers. The
Turkish mothers generally ask fewer questions and are more dominating compared to
Swedish mothers. On the other hand, the same study proposes that the source of the
problem is at least equally centered in the pre-school practices. The researchers
propose that lack of experience with typical ways of displaying knowledge in the
13
receiving society and classroom interaction have consequences for immigrant
children, especially in schools in which homogenous teaching-learning procedures are
common. They showed that the Swedish pre-school teachers used different strategies
when interacting with children from different origins. The teachers gave the Turkish
children less opportunity to express their knowledge, whereas with the Swedish
children they used a cooperative strategy with much turn-taking on the part of the
children.
Such teachers might very likely hold stereotypical views of ethnic minorities.
Philips (1983) proposed that cultural differences affected teachers’ attitudes towards
American-Indian children and their assessments of the children’s capabilities. She
observed that children who did not participate in classroom discussions were
perceived as not paying attention, lacking motivation, or as less intelligent than the
students who met the teachers’ expectations of behaviors. These teachers failed to
understand that the children of American-Indian origin are taught that it was improper
to draw attention to themselves as individuals, to display knowledge, or to appear to
know more than others. It seems that in this study, many of the problems of
American-Indian children noted by their teachers resulted from incompatibilities
between Indian and Anglo systems for expressing one's knowledge orally.
Thus, devaluation of immigrant children's abilities, talents and potential may
result in fewer learning opportunities for these students. Fewer opportunities means
fewer learning successes, and thus fewer opportunities for positive feedback. If
teachers feel a student is incapable of accomplishing a task, they are less likely to
assign challenging tasks to these students, and are likely to see such challenges as a
burden on both their teaching time and on student's ability. Suבrez-Orozco (1989)
reports that many Central American refugee children may be inappropriately tracked
into vocational classes because counselors assume they are not "college material." In
Israel 50% of the immigrant students from the former USSR reported that the school's
attitude towards them is not good enough. Their teachers noted that insufficient and
unsatisfactory school resources in addition to a continuous influx of immigrant
students contributed to severe staff erosion. The typical pattern between immigrant
and veteran students in the elementary school was harassment, while in high school it
took the form of estrangement or alienation (Tatar, Kfir, Sever, Adler & Regev,
1994). Discriminative attitudes is also the explanation used by Au (1998) to explain
educational inequalities in school dealing with immigrant children in the U.S. She
14
claims that the different cultural norms and linguistic behaviors of minority groups
may be in conflict with the norms and educational processes of the receiving society.
Thomas and Collier (1997) examined 700,000 language minority students'
records collected in five urban and suburban districts in the U.S. over a period of 10
years to examine the effect of school programs on English language proficiency. They
showed that English learners who experience well-implemented versions of the most
common education programs of English studies finish their school at average
achievement level between 10th and 30th national percentiles. In comparison, whereas
native English-speaking students typically finish school at 50th percentile nationwide.
English L2 learners who received bilingual enrichment education finish their
schooling with average scores that reach or exceed the 50th national percentile.
Contrary to the above studies, other recent studies, especially on second-
generation children in the U.S., claim that the education of immigrants is of equal
validity and scope as of their non-immigrant peers. Portes and Rumbaut (2001) report
that immigrant children have higher grades, and sharply lower dropout rates than
native-born American children. The immigrant children work harder and aspire to
greater achievements. At the end of high school the vast majority (98%) speak and
understand English well. Vernez and Abrahams' (1996) study of a national sample of
more than 21,000 tenth and twelfth graders showed that in general, immigrant youth
and parents have higher educational aspirations than do natives of the same racial or
ethnic group. Individual and family factors associated with high school graduation,
college-going and college continuity are generally the same for immigrants and
natives, as well as across racial or ethnic groups. Immigrant students most likely to
attend college have parents with higher income and education levels, and higher
educational expectations for their children. Suבrez-Orozco and Suבrez-Orozco (1996)
examined the attitudes of Latino-adolescents in comparison to non-Latino whites.
They showed that immigrant students hold much more positive views of school and of
school authorities, and are less bored and alienated than their U.S.A. born peers.
An Israeli study that examined issues concerning the integration of
schoolchildren who had emigrated from the former USSR found that teachers
reported no significant differences in academic achievements between the immigrant
children and their Israeli-born classmates (Tatar, et al., 1994). It is important to note
here that the Russian immigrants are on average highly educated as compared with
the receiving population. From all immigrants aged 15 and upwards, over half (56%)
15
completed 13 or more years of schooling, compared to 28% in Israeli society in 1989
(Donitsa-Schmidt, 1999). Another study, with similar elicitation procedures as the
present study, showed that bilingual immigrant students of highly literate Russian
origin, who attained well-established schools, and had been speaking Hebrew for 8
years prior to data collection, write well-organized narratives and expository texts,
and use literate lexicon and register accordingly (Rabukhin, 2003).
Very little research is available on the Hebrew language used by the Ethiopian
population who started immigrating to Israel in 1978 (Shuster, 1997). To the best of my
knowledge, there have been no studies undertaken from a developmental perspective
among grade school to high school students. A study on the absorption of Ethiopian
immigrant youth reported that in two thirds of the families, most of the parents lacked
proficiency in basic Hebrew language skills, while 20% of the youth reported having
difficulties with writing papers in Hebrew, and in understanding and answering
questions on examinations (Lifshitz, Noam & Habib, 1998). Another study showed that
a large percentage (about 45%) of Ethiopian immigrant parents are unable to hold a
simple conversation in Hebrew and most (about 75%) are unable to read or write simple
Hebrew. Another study also claims that adult immigrants' success in learning Hebrew
has been limited (Habib, 2001). A study of teachers’ evaluations of their Ethiopian
gradeschool students indicates that most of their difficulties were in the domain of
language proficiency (Golan–Kuk, Horowitz & Shefatia, 1987). Zehavi (1992) points
to difficulties in acquisition of reading by recently immigrated Ethiopian adolescents,
and notes that these students do not distinguish clearly between written and spoken
modalities, rather they usually write as they speak. Nor do they seem to be aware that
the written text should produce exactly the same meaning for all who encounter it.
Consequently, they do not exploit their formal knowledge of grammar and syntax
properly while writing. About a quarter of Ethiopian immigrant students and their
teachers report the students as having major problems related to study habits, bringing
required materials to class, preparing homework and active participation in class.
(Lifshitz, et al., 1998).
A national normative assessment research (Ministry of Education, 2001) showed
large gaps in academic achievement between Ethiopian immigrant students and non -
Ethiopian students in Language, Mathematics, Science and English both in elementary
and junior high schools. Another national sample study in the 5th, 9th and 11th grades,
comparing the achievements in language and in mathematics between immigrant
16
students and their non-immigrant Israeli peers showed that Ethiopian students
(including second generation) scored significally less in both disciplines across age.
Moreover, it takes 5-11 years for the immigrant student of Ethiopian origin to reach
native proficiency in the Hebrew language (Levin, Shohamy & Spolsky, 2003). Another
study held in only seven elementary schools of low SES children showed that in almost
all the schools the Ethiopian students’ scores in language were lower than the non-
Ethiopians (Mero, 2003).
Dropout rates among Ethiopian students in Israeli high schools (5.7%) are
higher than among Israeli born students (3.9%) (Ministry of Education, 2002). The
percentage of Ethiopian students entitled to a full matriculation certificate reveals only a
third (34%) compared to 44% of the entire population, and less than 2,000 Ethiopian
immigrants (out of 12,400 between the ages 20-29) study in institutions of higher
education (The Student Authority 2000; The Association for Advancement in Education
2000).
These problems in academic achievements are probably linked to the Ethiopian
students' low economic status and inability of their parents to play a more prominent
role in their education. In addition, there is a growing concentration of Ethiopian
children in weaker, less well-established and less "successful" schools, and these
schools are further weakened as high-achieving students leave them in great numbers
(Habib, 2001). Nevertheless, attitudes to education, research and professionals' reports
indicate that Ethiopian children and their parents are highly aware that succeeding in the
educational system is of great importance and one of the major means of social
mobility, and they have high motivation for their children’s success in school (Lifshitz
et al., 1998). Levin et al (2003) show that Ethiopian students who regard their efforts to
acquire Hebrew positively, use Amharic and support the maintenance of their home
language, have better academic achievements.
Although the above studies point to the difficulties of the Ethiopian students in
the study of language, other studies showed that narratives, fables, legends and the use
of proverbs are very important among Ethiopian Jews, and are used in everyday
communication in Amharic (Ben Ezer, 1992; Nezer & Polani, 1988; Rozen, 1999).
These findings are consistent with the results of other studies on oral cultures which
display restricted use of reading and writing along with a rich variety of poetic and
narrative genres of expression (Derive, 1994; Finnegan, 1970; Levin, 1965).
17
From a socio-cultural perspective, some researchers have tried to explain
children’s narratives either by presenting the “speech genre” as the specialized
discourse form of the child’s group or community (Bakhtin, 1986) or by their previous
exposure to narratives (Michaels & Collins, 1984; Pontecorvo & Orsolini, 1996).
Researchers have recently begun to consider social and affective factors in children’s
speaking and writing (Daiute & Griffin, 1993; Nicolopoulou, 1996). The assumption in
all these studies is that cultural values influence what a speaker-writer determines as
important when telling or writing a story. Blum-Kulka (1993) explored cultural
diversity in dinner table conversation narratives of Jewish-American families and Israeli
ones. She found differences between the narratives of the two communities in the
content of story telling, in child involvement and in spatial and temporal distance in
telling stories. Bavin (2003) showed that adult speakers of Warlpiri (an Australian
Aboriginal language) use a lot of path marking with a locative noun, a case marker, a
verbal affix or a proverb in their stories, since in their culture, location, paths and
directions are highly valued. A study by Invernizzi and Abouzeiad (1995) suggests that
narrative is a primary means by which children develop the voice of their own culture.
Some studies showed that children produce narratives that reflect their cultural
linguistic norms and values. Children from two different cultural backgrounds (Ponam
Island and American) differed significantly in their structural components while re-
telling stories in English. The Ponam Island children wrote detailed factual recalls, but
omitted affective and moral elements, evidently, because their culture entails ritualistic
recitation of important events and regards morals as superfluous and tricks as
unexpected. The American children, on the other hand, condensed their recalls to the
gist of the story and always included the “main idea”, resolution, and consequence
supported by general impressions and evaluations. Kuntay and Nakamura (1993)
examined the use of narrative evaluative devices. They showed that Turkish and
Japanese narrators avoided making explicit evaluative comments. Relatedly, differences
were also found between Japanese and English-speaking children telling a personal
narrative: the Japanese children spoke succinctly about a collection of experiences
rather than elaborating on any one experience in particular, a finding which the authors
attribute to the high value of short narratives in Japanese society (Minami & McCabe,
1991; 1995). Clearly, the older members of the community socialize the children into
producing culturally specific narrative discourse by teaching and modeling the standard
norms and appropriate language forms for this discourse
18
Like these studies, the present study regards cultural background as a crucial
contextual factor in text production. Cultural values play an important role not only in
what people include in their texts, but also in whether they are willing to produce a
particular kind of text. A pilot study revealed that Ethiopian children were reluctant to
tell a personal-experience narrative. This reluctance occurred whether the researcher
was a non-Ethiopian or an Ethiopian adult. This could be explained as an expression of
the “honor code”, a cultural pattern which inhibits speaking in front of one’s elders or
strangers as an expression of respect (Ben Ezer, 1992; 1999). Therefore, I assume that
the differences between the various SES backgrounds will be mainly connected to the
thematic content in the children's' narratives. As the above studies show, narrators of
different cultural backgrounds are the exponents of norms, attitudes, and behaviors of
their own specific communities. This study recognizes the multiplicity of ways in which
meaning is made in a variety of cultural contexts. Such recognition emphasizes the
symbolic representation of the worth of the cultural group far beyond the particular
written texts.
This study is based on the conviction that factors conventionally associated with
low SES such as ethnic background, family income, or parents’ level of education are
not the sole factors in determining children’s language potential. Accordingly, the study
will probe, by means of a specially constructed questionnaire, a range of other
background factors in the subjects’ home and school environments, including attitudes
towards writing and towards their native language, their writing experiences and home
literacy activities. Coupled with this questionnaire designed specifically to address these
issues, a detailed linguistic analysis of the written text production of students from two
different low SES backgrounds (Ethiopian and non-Ethiopian) should provide us with
the means for characterizing what it takes to become a proficient writer in these
environments.
A further assumption underlying this study is that both immigrant and native-
speaking children from similar SES backgrounds will manifest similar behaviors in
their school-based language activities. This idea is forcefully argued by Laparra (1999),
but to the best of my knowledge, the proposed study is the first to address these issues
directly by means of carefully controlled linguistic analysis of the texts composed by
children from similarly low SES, but culturally distinct backgrounds.
Following this hypothesis, I further assume that there will be no differences
between children from different SES backgrounds with regard to general cognitive
19
development, as manifested in global text structure. Thus, on the one hand, aspects of
language use such as overall discourse organization in construction of narrative
compared with expository texts and in differentiating between them, should be similar
across population. On the other hand, there will most likely be differences in low SES
use of linguistical forms in both lexicon and morpho-syntax. Such differences can be
explained by environmental factors, which have an impact on literacy skills, such as
home background, cultural norms and school settings in which these children
function.
20
Chapter Two: Research Design and Methodology Introduction
This chapter describes the methodology and research design used in collecting
and analyzing the database of the study. The chapter starts with a general background
to the study (Section 2.1) followed by a description of the research populations (2.2),
a review of language teaching in the schools (2.3), procedures for data collection
(2.4), methods of transcription (2.5), and categories of analysis and research
predictions (2.6).
2.1 General Background This section describes the original crosslinguistic project from which the
present study derived and the procedures and findings of a pilot study, which
preceded the final research study (2.1.1). The second part delineates and motivates the
overall conceptual framework within which the present study was undertaken (2.1.2)
2.1.1 The original project and pilot research study
The overall design and procedures of data elicitation for this study were adapted
from a large-scale international research project on developing literacy and text
production abilities ("the original project")1. The main goal of the original project was
to understand how schoolchildren and adolescents construct monologic narratives and
expository texts in writing and in speech in comparison with highly literate adults.
The project investigated how monolingual, native speakers of seven different
languages including Israeli Hebrew, from mainstream, well-educated, middle- to
upper-middle class backgrounds, and attending mainstream schools, deploy linguistic,
cognitive and communicative resources to adapt their texts to different circumstances
1 As noted in Chapter 1, the study, entitled "Developing Literacy in Different Contexts and Different Languages" was funded by a Spencer Foundation major grant to Ruth Berman.
21
(narrative versus expository texts and written versus spoken discourse). Preliminary
results of the original project are presented in two issues of a special volume of the
Journal of Written Language and Literacy, edited by Berman and Verhoeven (2002).
Its aims, procedures and major crosslinguistic trends are described in detail in the
introductory article to the first issue (Berman & Verhoeven, 2002, pp.1-44). In
addition, findings for each of the seven languages in the study are analyzed in relation
to the shared theme of expressing discourse stance in a special issue of Journal of
Pragmatics (Berman, in press). A series of studies on the Hebrew data-base appear in
Aisenman, (Ed), 1999, while others have been published and/or are in press (Berman,
2001b; Berman & Nir, submitted; Berman & Katzenberger, submitted; Katzenberger
& Cahana-Amitay, 2002). In addition, and of particular relevance to the present study,
are two recently completed M.A. theses that applied similar methods of data-
elicitation on different populations. One investigated low SES children in the same
age-groups as mine who attend schools that are known for their rigorous academic
demands and high scholastic achievements (Salmon, 2002). The other examined
bilingual immigrant students of highly literate Russian origin who attend well-
established schools and had been speaking Hebrew for at least 8 years prior to data
collection (Rabukhin, 2003).
The main goal of the present research was to examine how low SES children
of different ages and from different ethnic backgrounds (Ethiopian and non-
Ethiopian) write monologic narrative and expository texts. The present study followed
the devised and carefully tested crosslinguistic project in overall conception and
research design in order to ensure maximum comparability of results from diverse
populations. My study was, however, modified in several important ways to meet its
underlying purpose and the particular research goals, which it pursued.
The design of the present study was adapted from the original project as
follows: (1) It included only schoolchildren, without a comparable group of adults
(for reasons detailed in section 2.3. below). (2) In contrast to participants in the
original study, the children were from relatively uneducated, low SES backgrounds –
one of Amharic-speaking Ethiopian parentage and the other consisting of native
speakers of Hebrew from similarly low SES backgrounds. (3) The study elicited only
written and non-oral texts, again for reasons specified in the section dealing with the
research population (2.3); (4) And it extends the original design by means of a
detailed and specially devised literacy questionnaire aimed at providing relevant
22
information about the research groups. As noted, these modifications suited the
overall goals of the study, and were supported by extensive piloting among
schoolchildren from backgrounds similar to the target populations.
The pilot study was conducted with 16 schoolchildren of Ethiopian and non-
Ethiopian parentage living in the same poor neighborhood and attending the same
schools. All participants in the pilot study were of low SES backgrounds and the
schools they attended were defined (as further detailed below) as of relatively lower
standards and academic achievements compared with the participants in the original
project both in Israel and abroad. As in the original project, the participants in my
pilot study were each required to produce four different texts: a written and a spoken
personal experience narrative and a written expository essay and a spoken class talk
on the same topic of interpersonal conflict. However, most of the Ethiopian children
proved reluctant to speak, or to produce any kind of oral text, especially when asked
to do so by a non-Ethiopian adult. In this, they differed markedly not only from the
middle-class subjects of the original project, but also from their non-Ethiopian peers
in the pilot study.
Piloting further revealed that the literacy questionnaire that had been used with
the Israeli subjects of the original project proved very difficult for all the children
from low SES background. Unlike the children of the original project, all the students
participating in my pilot study needed considerable help and direction in order to
answer the questions on the original literacy questionnaire, which required subjects to
describe their reading, writing, and other literacy-related activities both at school and
after school hours. As a result, it was decided to construct a separate questionnaire,
specially designed for the present study, which would take into account the
difficulties encountered in our pilot study (See, further, Section 2.7)
2.1.2 Overall conceptual framework
A piece of discourse reflects a piece of life. Thus, an analysis of texts that
relate to a complex human phenomenon such as interpersonal conflict cannot be
confined to the methods of a single discipline. Rather, in approaching such pieces of
discourse, a multidisciplinary approach is called for. Such a methodology is
particularly appropriate and even necessary in the case of texts written by grade
school children and adolescents from different backgrounds and ethnic origins. An
interpretation of immigrants' texts requires a contextualized framework for
23
understanding of crosscultural encounters. In other words, a multifaceted description
is needed to shed light on such complicated human experience (Denzin & Lincoln,
1994). Such a description is anchored in insights deriving from a range of disciplines
including linguistics, literacy studies, sociology, anthropology, education, and
psychology. In the present study, an integrated research approach of this kind was
applied by combining linguistic, socio-cultural, and educational perspectives. These
different perspectives provide a basis for an interconnected clustering of
interpretations and so allow fresh and richer insights into the research findings. In
keeping with this orientation, the study combines quantitative measures of the kind
associated with research in education and the social sciences and qualitative,
culturally motivated analyses together with structural analyses of linguistic forms and
constructions associated with the domains of general linguistics and Hebrew language
studies.
2.2 Research Population
The study investigates two different, though related research populations:
children of Amharic-speaking Ethiopian parentage (henceforth “the Ethiopian”
participants) and their native-Hebrew-speaking peers (“the non-Ethiopian subjects”),
both of similarly low SES background. Participants were selected so as to allow for
maximum comparability with the subjects of the original project, and particularly with
the Hebrew-speaking Israeli sample of that project. However, in contrast to the
original, Hebrew-speaking population, all the subjects in the present study come from
the same disadvantaged neighborhood in the outskirts of a large city in central Israel.
Schoolchildren were taken from the same three age groups as in the original project:
4th grade elementary school children, aged 9 to 10 years (labeled G), 7th grade junior-
high students aged 12 to 13 years (labeled J), and 11th grade high-school students
aged 16 to 17 years (labeled H). This yielded a breakdown of 16 Ethiopian children
[ETH] and 16 non-Ethiopian [NETH] in each age group. The two research groups
had an even number of boys and girls over all, but we were not able to balance each
group for sex, since it was impossible to find three such groups of Ethiopian children
at the same grade levels and residing in the same neighborhood throughout the
country. This did not seem too important to me, since sex did not turn out to be a
significant factor across the variables and the different countries in the original project
24
(Berman & Verhoeven, 2002, pp. 27). Students rated by their class and language
teachers as under-achievers in language studies were excluded from the study, as were
children suffering from known developmental or learning disabilities as reported by
the school counselors.
Table 1 shows the breakdown of the 96 participants in the study, in terms of
background (Ethiopian /non-Ethiopian), sex, and age and level of schooling.
Table 1 Breakdown of research population by origin, sex, and grade level
Ethiopian [N=16] Non- Ethiopian [ N=16]
Grade level Girls Boys Girls Boys
G (grade school) 9 7 10 6
J (junior-high) 7 9 7 9
H (high-school) 12 4 11 5
Total 28 20 28 20
All the subjects from both the Ethiopian and non-Ethiopian groups were taken
from the same grade level at the same school, in order to minimize possible effects of
school ideology or of specific methods for teaching writing and other language-
related skills between different schools and across different school systems. The three
schools (grade-school, junior-high, and high-school) attended by the subjects all
belong to the religious sector of the Israeli state school system. Evidently, the
Ethiopian children and adolescents were directed to the religious state education
system, which caters for students of religious or what is regarded as “traditional”
background (Weil, 1988, pp.125-126). To gain relevant background information, the
inspectors, principals, and counselors of each of the schools were asked to define their
schools on a three-level scale: high, average, and low, according to the following four
factors: (1) socio-economic level of the students’ home background; (2) academic
achievements of the students measured by standardized national language tests in
relation to national norms; (3) availability and use of adequate learning materials,
current programs and innovative educational projects; and (4) level of resources
available from the local authorities. All the educators that were consulted without
exception defined their students as members of low-income families, with high
25
percentages of unemployment, and relatively limited schooling, whose parents had a
maximum formal education of nine years of school. They ranked school resources and
facilities as low, while academic achievements and resources were defined as low to
average. The breakdown of student population in the three schools that participated in
the study was as follows: half of the grade-school students were of Ethiopian origin,
and in junior-high and high-school, only 30% were of Ethiopian origin.
The present study was confined to three age groups, or three levels of
schooling, in contrast to the original project, which included university graduate-level
adults. The reason for this was that it was almost impossible to find a sufficient
number of adults of Ethiopian background, who met the two criteria of speaking
Hebrew for eight years prior to the time of data collection and of having graduated
from university in Israel or elsewhere.
2.2.1 The Ethiopian group
All the Ethiopian children are of recent-immigrant origins and are raised in
multilingual settings. Two-thirds of the children in this group (35) are Israeli-born,
while one third (13) were born in Ethiopia. The vast majority of these subjects
reported that they speak at least some Amharic, but most of them (83%) reported that
they could not read or write in Amharic. Participants in the study had all spoken
Hebrew for at least the past eight years. This criterion was based on a study showing
that accommodation towards native-like linguistic behavior is reached after eight
years of stay in the host or target-language community (Blum-Kulka & Olshtain,
1984). Thus, the issue of “language” is not listed as an independent background
variable for the following reason. Hebrew, as the language of the Israeli school
system in general and the schools attended by members of both research populations
in particular, is defined in the present context as the shared “first language” or “L1” of
both the Ethiopian and non-Ethiopian subjects. In the present context, following
Polinsky (1997) and Romaine (1995), we define this not as the first language acquired
by children, typically the “home language”, but rather their “primary” or dominant
language. Thus, Hebrew is the main language of the country, and hence of the
subjects’ surroundings in the neighborhood and, most importantly, in the school.
All the parents of the Ethiopian children were born in Ethiopia, with Amharic
as their native language. School files revealed that parents of the Ethiopian subjects
26
averaged less than two years of formal schooling in their country of origin. In
contrast, in filling out the literacy questionnaires (Section 3.18), about two thirds of
their children reported that their parents could read and write Amharic, and one third
reported that they were able to read and write in Hebrew.
Teachers reported that the Ethiopian children take less part in class
discussions than the other students in the same class. As noted, in the Ethiopian
tradition, young people are inhibited from talking in the presence of their elders as a
signal of respect (Ben Ezer, 1992). An Ethiopian educational counselor explained the
children's disinclination for speaking in front of adults as yielding to a cultural
dictum. He told me that when an Ethiopian father sends his son to school, he says to
him: "Remember, my son, when you attend school, you go to ‘buy’ and not to ‘sell’".
This implies that a "good" student listens obediently and silently to the teacher
without expressing his or her knowledge by speaking in public. Their teachers
confirmed that they are also familiar with the Ethiopian children's reluctance to speak
in class. As a result, they prefer to evaluate their Ethiopian students' school
achievements mainly through their written work, rather than on the basis of their rare
oral output. The very short oral texts produced by the Ethiopian children in the pilot
phase of this study, coupled with their reluctance to speak with the young Ethiopian
man and woman, who conducted the pilot study, supported the teachers’ perceptions.
Consequently, it was decided for purposes of the present study to compare only
written texts of both the Ethiopian and the non-Ethiopian groups.
2.2.2 The non-Ethiopian group
Nearly all (92%) of the non-Ethiopian subjects were Israeli-born. Two were
born in Eastern Europe, but came to Israel as babies. At least half of their parents
were also Israeli-born. The rest emigrated from North Africa, with a few from Eastern
Europe. In marked contrast to the Ethiopian parents, information from school files
revealed the parents of this group as having an average of nine years of formal
education. Their children reported that all of their parents could read and write
Hebrew. The vast majority of the non-Ethiopian subjects (92%) also reported that
they spoke Hebrew in their homes. This breakdown is typical of the make-up of the
Israeli population in general, particularly in less advantaged neighborhoods. It is also
typical of the schools that service them. And it contrasts markedly with the Israeli
27
population of the original project, who were all third-generation and monolingual
speakers of Hebrew both of whose parents had both completed at least high-school
and often had further education as well.
2.2.3 The external control group for the literacy questionnaire
In addition to the two research groups, a third group of schoolchildren was
also asked to fill out the literacy questionnaire. This was done because the
comparison-population of the Israeli participants in the original project was given a
much more limited literacy questionnaire than the subjects of our study. The
“controls” for the literacy questionnaire were Israeli-born children of high SES
background in the same age groups and at the same levels of schooling as the research
groups, all from Hebrew-speaking, middle-to upper-class home backgrounds. In
addition, they were all above average, high achievers in school language studies, with
an average grade score of 95 (equivalent to A+ in the American school system). Their
school files revealed their parents as highly literate in Hebrew, with an average of
fifteen years of formal schooling.
This group was included in order to compare the responses to the literacy
questionnaires by children from varied backgrounds and with differential language
skills. This comparison provided a broad characterization of socio-cultural and
demographic variables, attitudes and literacy-based activities of children of low
compared with high-SES backgrounds.
2.3 Language Teaching in the Research Populations Schools
Israel as an immigrant-absorbing country has traditionally adopted a
monolingual ideology in relation to language teaching. Immigrants' Hebrew
proficiency was considered a key for their acculturation and integration into Israeli
society. Moreover, they were expected to forget their own languages as a sign of
loyalty to the country and of their assimilation to the Israeli community (Spolsky,
1995; Spolsky and Shohamy, 1999). Minority groups were even indoctrinated by
societal institutions to use no other language apart from Hebrew (Harshav, 1993).
This pattern of learning Hebrew at the expense of immigrant languages resulted in a
process of language shift, in the sense of the loss of most home languages of
28
immigrants arriving in Israel during the last hundred years (Ben Rafael, 1994, 2002).
In recent years, the Ministry of Education has supported a gradual change in
this ideology, expressing a new, less monolithic ideology. In the Language Education
Policy published by the Israel Ministry of Education in 1996, immigrants (especially
those from the two major groups of Russians and Ethiopians) are to be encouraged to
maintain their home languages in the course of their acquiring Hebrew. However, this
shift in policy towards a more open acceptance of multilingualism and language
maintenance has not been implemented in many educational institutions (Spolsky &
Shohamy, 1999).
Responses to the teachers’ questionnaires (Section 2.4.7) revealed that the
language teachers (who teach Hebrew literature, composition, and grammar) in the
three schools still believe in the hegemony of Hebrew. They also do not make use of
texts (in Hebrew) relating to Ethiopian Jewish culture or narratives typical of this
community. The teachers do not encourage the Ethiopian children to maintain
Amharic and they discourage them from using the language in oral interactions with
their friends in school. Some teachers rated Amharic as a low prestige language
compared to Hebrew or even to the home languages of other minority groups. The
language teachers also did not use any written texts in Amharic and no ethnic music
or even songs in Amharic were broadcast during school breaks. None of the teachers
knew more than three words or expressions in Amharic, although some of them had
been teaching Ethiopian children for as long as ten years.
With regard to the teaching of writing, the language teachers were revealed as
using conservative methods of instruction and relying on old, well-established
learning materials. In grade school, none of the teachers had adopted recently
developed, newer learning materials for teaching of writing. Nor had the different
language teachers in the school developed an integrated language program with
common goals for all students. Similar findings emerged for the junior-high-school,
where the language teachers could not define common goals for the teaching of
writing in their school. Moreover, none of them used the recently published language
book recommended by the Ministry of Education. The high-school language teachers
made wide use of teaching materials that they had developed on their own, aimed
mainly at rote practice of grammar. Various teachers used different books in the same
age-grade, and they did not share the same norms for defining language success. Only
two teachers had taken part in an in-service language-training course during the year
29
prior to data collection.
2.4 Data Collection Procedures
The research design involved six components: four were conducted with
students in class or individually; the fifth was a structured questionnaire for
interviewers; and the sixth, a questionnaire for teachers. Three of the four student
components were held in class, introduction to the task; personal questionnaire;
writing of the two texts; while the literacy questionnaire was conducted individually
by an adult interviewer. In all four parts of the study, the original procedures and
instructions were adapted as noted, from the original project, and geared especially to
the cultural needs and norms of Ethiopian subjects.
2.4.1 Introduction to the task
To start with, the overall aims and procedures of the study were explained in
detail to each class in order to overcome the typical reluctance of Ethiopian
immigrants to cooperate with research about themselves and their community. In the
Ethiopian community, the qualities of doubt, mistrust, suspicion and caution are
highly regarded. Children are warned against the danger of being naןve in this respect.
Ethiopians are also expected not to share personal details of their life with others.
Negative feelings such as seeking vengeance, anger, and envy are kept to oneself
(Ben-Ezer, 1992; 2002). Given these major features of Ethiopian social interaction,
there was a risk that the theme of “interpersonal conflict”, which formed the basis for
eliciting personal-experience narratives in the original project, would constitute an
insurmountable source of difficulty for the participants in the present study 2 and the
entire research design could have been undermined. In order to overcome the issue of
trust, the researcher presented herself as a “friend”, where a friend in this context is
2 In fact, earlier versions of the original crosslinguistic project showed that this topic failed to elicit personal-experience narratives of interpersonal conflict from participants from the northern countries such as the Netherlands and even more markedly Sweden. Accordingly, in the final version of the crosslinguistic project, participants in these countries were asked to tell and write a story about a situation where someone had helped extricate them from a difficult situation (Berman & Verhoeven, 2002).
30
defined as somebody who first and foremost can keep a secret (Ben Ezer, 2002b).
Accordingly, all subjects were assured of complete confidentiality, and told that no
information concerning their participation in the study would be given to their
teachers.
In order to overcome the Ethiopian children’s reluctance to share negative
personal feelings, another important Ethiopian behavioral code was activated. This
cultural code requires every member of this community not to refuse an authority
figure, even when one disagrees with or does not accept what this figure says (Ben
Ezer, 1999). The school and student counselors were asked to encourage collaboration
with the researcher by motivating their students to do their best in performing the
task. They did so by being present during data collection and also by stressing the
prestige of the participation as the “Israeli representatives in an important
international research project”. They made it clear that the school authorities expected
the students to cooperate with the researchers, but that their performance would not
constitute a factor in teacher evaluation or school grades.
To ensure that subjects understood the key terms relating to the variables of
the study, the introduction to the task, which used the same instructions as in the
original project, was supplemented by a short discussion concerning the differences
between a narrative and an expository text -- a procedure that had not been found
necessary in the original project. Subjects were then encouraged to ask questions,
most of which related to the confidentiality of the data. The researcher ended the
introduction by thanking them for their cooperation. Subjects were then presented
with the same short video clip without words as was used in the original project,
showing scenes of different kinds of unresolved conflicts of a moral, social, and
physical nature in a typical culturally neutralized school setting
2.4.2 Personal data sheet
Following the introduction to the task, each student was given a short personal
data-sheet to fill out in writing – stating name, sex, age, home-address, class, and
school. The aim of this short questionnaire was to verify details given by the school.
31
2.4.3 Text production:
All participants produced two texts, in two different orders, following the
original design: half the subjects (group A) first produced an expository text and then
a narrative text (order A), and the other half (group B) first produced a narrative and
then an expository text (order B). Both texts were produced during the same class
session, but elicitation was conducted in separate classrooms - one for group A and
the other for group B.
In order to elicit narratives, subjects were asked to write a story describing an
incident that had happened to them personally, in which they had experienced
problems with somebody else. In order to elicit expository texts, subjects were asked
to write a composition in which they discussed the topic of problems between people.
This method intentionally mirrored the class-like atmosphere of a school-based task
[See Appendix II]. Adult investigators, Ethiopian and non-Ethiopian, who were
introduced as “helpers” to the students during the introduction to the task, were
available for each row in each classroom, in order to answer questions, and to ensure
that students were not confused or upset.
2.4.4 Literacy questionnaire
The literacy questionnaire of the original project was expanded considerably
to meet the particular goals of the present study and the general nature of the research
populations. Its main goal was to shed light on school and home variables and
attitudes towards language and identity, which may affect language proficiency. For
example, the questionnaire probes the subjects’ attitudes towards Hebrew, towards
writing, and towards their native language (following the examples in Akinçi, 1999;
Ben Raphael, Olshtain & Gist 1994; Donitsa-Schmidt, 1999; Molokandov, 1999).
Second, the questionnaire investigates the writing behaviors of the target populations
based on guidelines published by the Israeli Ministry of Education (1999) for the
required literacy skills in first grade. Third, it surveys home literacy activities, along
lines proposed by Lahire (1995) in working with lower SES populations in Britain. In
addition to relying on personal experience with the school system and administration,
I also consulted with ten Israeli researchers and educators in relevant domains,
including: developmental psychology, education, educational evaluation, Ethiopian
culture, linguistics, literacy, and sociology. This was done in order to include
questions concerning socio-linguistic, educational, and psychological variables
affecting text-production abilities in the new version of the questionnaire. The first
32
version of the revised questionnaire was piloted among several children of different
ages from the same backgrounds as the research populations. It proved too long and
tedious for them, and a shorter third version was constructed. This was then
successfully tried out on several children at the same ages as the final research groups.
The final version of the research questionnaire contains 24 open-ended and multiple-
choice items relating to attitudes to language, writing behaviors and home literacy.
Each student was required to fill out a questionnaire in writing. In order to
lighten the load on the children, the questionnaires were administered on a one-to-one
basis during a meeting with a supporting investigator, held one to three days after text
production. Subjects were allowed to choose which one of eight available adults they
wished to work with. The eight interviewers were divided by sex (four women and
four men), by age (four in their twenties and four in their forties), and by origin (four
Ethiopians and four non-Ethiopians). This was done in order to provide positive
origin and sex-role models, especially for the Ethiopian children, who are inhibited
from talking in the presence of strangers as a mark of respect (Ben Ezer, 1992; 1999).
I believed the Ethiopian children would feel more at ease if they were able to share
personal details concerning their homes with a familiar adult from their own
community and sex. Ascher (1991) proposes the participation of race and sex role-
models in grade schools. His review of over twenty programs for African-American
boy students led him to recommend those which included male African-American
classroom teachers, adults from the community and also older students to help young
students of the same race and sex. Along these lines, subjects in the present study
were free to choose which interviewer to work with out of eight adults, divided
equally by race and by sex. Each subject sat opposite the interviewer at a small table
where he or she filled out a questionnaire in writing. Subjects were allowed to ask the
interviewer to read out and/or to explain the questions to them.
The literacy questionnaire was analyzed according to the following domains:
1) level of success in Hebrew language studies, 2) proficiency in writing, 3) attitudes to
writing activities, 4) level of proficiency in Amharic, 5) attitudes to writing and
definition of the proficient writer, 6) language attitudes, 7) interviewer information on
the questionnaire as a task, 8) parental background, and 9) home literacy. Each domain
was evaluated according to age and origin and in terms most relevant to the general
purpose of the study – the impact of literacy and literacy-based experiences on
students’ perceptions and attitudes.
33
2.4.5 Interviewer questionnaire
Each of the eight interviewers filled out a short questionnaire for each subject
that he or she accompanied regarding the subject’s conduct in the course of
performing the task. The questionnaire contained information about how much and
what kind of help each student needed, the time it took to fill out the questionnaire,
and how difficult it was for him or her. Each interviewer was also asked to write his
or her own general impression of how the subject handled the task. The interviewers'
protocols describing the contents of the interview provided essential cultural context
regarding the children's behavior while filling out the literacy questionnaire.
2.4.6 Teacher’s evaluation sheet
The class language teachers (that is, teachers of Hebrew language, grammar,
composition, and/or literature) filled out an evaluation form for each student. They
were asked to give each student an average grade on his or her level of general
achievement in Hebrew language studies, including composition and literature.
Grades were ranked on a scale from 50 to 100, as accepted in the Israeli system, as
follows: 50 (fail), 60 (fair), 70 (average), 80 (good), 90 (very good), 100 (excellent).
This is roughly equivalent to the scale used in the U.S. school system:
50 = F, 60 = D, 70 = C, 80 = B, 90 = A. This was done in order to examine whether
the teachers evaluate the Ethiopians' and the non-Ethiopians' language proficiency in
similar ways.
2.4.7 Teachers' reports on the teaching of writing
The language teachers were also asked to report on the specific learning
materials they used in class for teaching writing. In addition, they were asked to
specify three topics or goals they focus on in the course of their teaching (e.g.,
introducing composition genre differentiation, writing abstracts and summaries,
underlying key concepts etc.).
2.5 Transcription
34
The present study adopted the methods for text transcription used in the
Hebrew sample of the original project, with the aim of achieving maximum
comparability across the two data sets, and in order to describe general developmental
tendencies in different research populations (Berman & Verhoeven, 2002). Following
the basic CHAT format designed for the original project of child language data in the
Child Language Data Exchange Systems (CHILDES), as described in MacWhinney
(1995; 2000), the CHAT method of broad phonetic transcription serves as a base for
the CLAN system of language analysis programs, and so was adopted in the present
study. As in the Hebrew sample of the original study, here too, the transcription was
adapted to suit extended texts produced by schoolchildren and adults, to represent
written as well as spoken texts, and to accommodate a language with a non-Latin
system of orthography. The method of transcription applied in the original study for
these purposes proved fully appropriate to the present study (for an example of a
transcribed text see Appendix II).
Two different versions for each text were produced in the original project: a
replica or “mirror” version representing the text as closely as possible in the form
produced by the subject and a standardized or "stripped" version that omitted or
corrected deviations such as pauses, hesitations, filler syllables and repetitions in speech
and spelling errors and deletions or corrections in writing. The present study used only
the standardized version for each of the two texts, since its focus was on the written texts
as products of the text construction process. It was therefore decided in advance not to
include subjects' spelling as a variable of analysis. On the other hand, the original
handwritten texts (that is, the “mirror” versions) were used for further examination of
"writing as a notational system" (Ravid & Tolchinsky, 2002) to compare the spelling of
the subjects of this study with those of the subjects in the Hebrew sample of the original
project (Nir-Sagiv, 2002).
35
2.6. Categories of Analysis and Associated Predictions This section describes and motivates the categories established for analyzing
the data-base, supplemented by relevant research hypotheses relating to the major
independent variables of the study: (1) Development -- three levels of age and
schooling; (2) Genre or text type -- narrative or expository; and (3) Origin --
Ethiopian or non-Ethiopian children of low SES background -- in many cases
compared with their mainstream peers. The predictions relating to each variable in
each category are based on several sources. First, they rely heavily on prior research
regarding language and discourse abilities of such populations in Israel and abroad –
as discussed in the preceding chapter and elaborated here below. Second, they take
into account findings of the original project for subjects in the same age-groups but
from middle-to upper-class, well-educated backgrounds – mainly although not only
from the Hebrew sample of the original project. Third, they rely on extensive piloting
conducted as background to this study. As a fourth source of input for my predictions,
I take advantage of my own close familiarity and lengthy professional experience with
the Israeli school system, curriculum construction and educational practices in
general, and work with children of disadvantaged backgrounds and of Ethiopian
extraction in particular.
This study involves different categories and domains of analysis which have
been shown to be relevant to and even diagnostic of developing literacy and text-
writing abilities. These include both "bottom-up" analyses at the level of lexical items
and internal clause structure as well as more global "top-down" analyses of overall
text length and discourse structure. The domains analyzed include the following: text
length as measured in terms of number of words and clauses per text (2.6.1);
syntactic complexity - measured in terms of number of words per clause, interclausal
syntactic packaging, and types of linking devices in clause packaging (2.6.2); lexical
richness - as defined in terms in terms of lexical density, or proportion of content
words per text (2.6.3); thematic content - what topics subjects chose to refer to as
expressed by the lexical content of their texts (2.6.4); violations of linguistic norms -
as measured by (a) grammatical errors, (b) lexical infelicities, and (c) register mixing
(2.6.5); and global text construction - in terms of overall discourse structure, text
components and organization of information in the text (2.6.6).
36
2.6.1 Text length
Text length was measured in terms of number of words and clauses per text.
In line with a key motif of the original project, which highlighted the critical role of
the lexicon in developing deployment of morpho-syntactic structures and text-
production abilities (Strömqvist, Johansson, Ragnarsdóttir, Aisenman & Ravid, 2002;
Cahana-Amitay & Sandbank, 2000; Ravid & Cahana-Amitay, in press), this study
used the word as a basic unit of analysis. Any lexical analysis needs to start out by
specifying a basic unit of demarcation in the form of a “lexical item” and by defining
a “word”, generally recognized as the basic building block of any language. Yet both
are extremely difficult to define (Anderson, 1983; Berman, 2001). In view of
problems of principle in demarcating and defining linguistic lexical elements and
difficulties encountered in this respect in the original project (Berman, 2002;
Strömqvist et al., 2002), it was decided to adopt a conservative traditional
orthographic convention in specifying what constitutes a word, corresponding to what
is termed in Hebrew teyva literally, ‘box, crate’, used to refer to a written word. This
ensured consistency in transcription with written conventions and also yielded
comparability between our data base and the Hebrew sample of the original project.
Following the method of segmentation in the original project, a lexical item
was defined as any element separated from the preceding and following elements by a
space in the transcription of both narrative and expository texts (Berman, 2001).
Special account was taken of the fact that in Hebrew, the definite article ha- ‘the’, the
prepositions le- ‘to’, be- ‘in’, mi- ‘from’, ke- ‘as’, and the conjunctions ve- ‘and’, še-
‘that’ are orthographically bound to the following word. For example, the complex
string vekšehaish (ba) = ve-k-še-ha- ish ' and-as-that-the-man (came)' = ‘and when
the man(came)’ was coded as one word. These orthographically bound morphemes
were specially marked with the symbol ^ in order to allow for listing them as distinct
lexical items, corresponding to their European counterparts, which are written as
separate words.
The term "lexical item" was extended from “a word” to include multilexemic
strings consisting of more than a single transcribed word in the case of set, formulaic
or idiomatic expressions of the kind that are typically semantically unanalyzed, and
which often have monolexemic counterparts. Examples of such lexicalized
expressions include bound compounds like bet sefer ‘house-of book = school’, baal
bayit ‘master-of house = landlord’ and other multi-word expressions like af~pa'am
37
‘no time = never’, af al pi xen ‘but on mouth-of so = even so’ (Aisenman & Berman,
2000; Berman, 2001). Appendix II illustrates how texts were divided into words.
Analysis was also conducted on the clause as the basic unit for syntactic
analysis on two levels of syntactic organization - within and across clauses.
Following Labov (1972), the clause was chosen as a basic unit of text segmentation in
two large scale crosslinguistic and developmental studies – one on oral narratives of
children aged 3 to 9 years compared with adults (Berman & Slobin, 1994) and also in
the original project dealing with oral and written narratives and expository texts
(Berman & Verhoeven, 2002). In both these studies, the clause -- defined as "a
unified predicate describing single situation (an activity, event or state)" (Berman &
Slobin, 1994, p.660) -- proved to be a reliable, developmentally diagnostic, and
psychologically real unit of analysis across languages, modalities, and genres. In the
present study, like in the original project, the clause served in place of the behavioral
unit of an “utterance” as the boundary for the "textline" as defined in the CHILES
transcription conventions (MacWhinney, 1995, 2000). Based on the conventions of
text segmentation applied in the original project, texts in the present study were
divided into clauses. Three native speakers of Hebrew with training in linguistics
divided all the texts into clauses. All cases of disagreements were discussed and
resolved. Inter-judge agreement was 99.5%.3
The present study followed the CHAT conventions as described in the
CHILDES procedures for language production analysis for coding each unit of
analysis on a separate coding tier.
2.6.1.1 Text length as measured by number of words per text
Text length, as noted, measured in terms of number of words and clauses per
text increases with age. Scott and Windsor (2000) showed that even school-age
children with language learning disabilities write longer texts with age. The following
predictions were specified with respect to text length as measured by number of
words:
• Text length was expected to increase with age, especially between 7th to
11th grade in both research groups, in line with what was found for the
3 I am indebted to Nurit Assayag and Bracha Nir for their help in this connection.
38
crosslinguistic subjects (and see, too, for oral narratives, Berman & Slobin,
1994, page 31).
• The narratives of the research groups were expected to include more words
than the expository texts across age-groups, again in line with findings for the
crosslinguistic population.
• Country of origin was not expected to make a difference in this respect,
since the two research groups share similar literacy environments, both at
home and at school, of the kind that are critical to vocabulary development.
For this reason, too, country of origin was not expected to be a differentiating
factor between the two research groups, across all measures of vocabulary
development (number of words, number of content words, lexical density,
etc.).
• With regard to population, I predicted that the texts of the crosslinguistic
subjects would in general include more words – and more different words --
than those of the low SES children, since upper SES gradeschool children are
known to have larger vocabularies than their lower-class counterparts (Anglin,
1993; Hart & Risley, 1995). Moreover, high SES children have the advantage
of more varied written language experiences and richer literacy environments
both in their schools and their homes (Dickinson & Tabors, 1991, 2001; Duke,
2001).
2.6.1.2 Text length as measured by number of clauses
Text length was measured in a range of related studies by number of clauses
per texts, and this measurement was found to be sensitive to age and genre in Hebrew
as in other languages (Berman & Ravid, 1999; Berman & Slobin, 1994; Berman &
Verhoeven, 2002). I therefore predicted for the present study that (1) average number
of clauses would also increase with age and (2) average text length would be higher in
the expository texts, (3) I predicted that there would be no differences between the
two research groups, who were expected to reveal the same general school-related
abilities, since they attend the same school system, they share the same classes in
language and the same language teachers, and they have similar backgrounds of low
literacy. (4) With regard to the different populations, following studies showing that
the narrative development of minority children may be hampered – particularly in the
39
host language -- due to restricted channels of input (Akinçi, 1999; Akinçi et al 2001;
Verhoeven in press), I predicted that text length as measured by average number of
clauses would be higher in the original project than in the research populations of this
study.
In sum, with regard to overall text size in my sample I expected the two
measures of text length, in words and clauses, to show the same patterns across the
populations.
2.6.2 Syntactic complexity
This study lies in the domain of "later language acquisition", since it examines
the language production of children well beyond the preschool years, from age 9 years
up. Language development can be characterized in terms of increasing complexity
and explicitness in general and greater syntactic complexity and coherency in
particular. Later syntactic development is attested by qualitative and quantitative
increases in appropriate syntactic constructions for a variety of genres (Berman &
Verhoheven, 2002; Nippold, 1998; Ravid, 2003). This increase in amount and kind of
syntactic constructions is manifested in longer and more complex text length (Berman
& Ravid, 1999; Berman & Verhoheven, 2002; Salmons, 2002). Even though
preschool children have established grammars with a varied range of syntactic
constructions including in Hebrew (Berman & Dromi, 1986), complex syntax
continues to develop during schooling and into adulthood (Scott, 1995). During these
years language users become more and more familiarized with linguistic features
typical of the written language of various text types (Gillis & Ravid, in press), and
learn how to deploy these structures appropriately in the context of extended
discourse (Berman, 1996, 1998; Berman & Slobin, 1994). In written language, clause
length increases from approximately five words per clause in fourth grade to eight
words per clause by twelfth grade (Hunt, 1965). Thus, the syntactic development of
proficient writers is expressed in the scope and quality of their particular choice of
syntactic means from the available repertoire of linguistic forms and constructions. In
a functional, discourse-oriented perspective, the hierarchical organization of
information is reflected by means of the skillful and appropriate deployment of
complex syntax. Consequently, proficient writers organize their ideas not only in the
form of linear chaining, but as hierarchically packaged constructions.
40
Along these lines, syntactic complexity in this study is measured both in terms
of number of words per clause, interclausal syntactic packaging and types of linking
devices in clause combining. The number of words (as defined above) per clause has
been shown to reflect increasing density of packaging of more information inside a
given syntactic unit in the form of more words per package and more phrases per
clause (Saltzman & Reilly, 1999). With regard to syntactic complexity measured by
number of words per clause, I predicted an increase with age, and in expository texts
compared with narration I also predicted that this measure would not be sensitive to
origin, but would be sensitive to population. Thus, it would be lower in both the low
SES research population in my study compared with the subjects of the original study.
2.6.2.1 Connectivity or clause-combining as measured by clause packages
In an attempt to further examine later syntactic development as attested by
diverse and complex syntactic architectures, each text was also segmented into text
units termed Clause Packages- a unit of analysis established in the original project,
and termed originally "L(onger)units" (Aparici, Tolchinsky & Rosado, 2000; Cahana-
Amitay, submitted; Cahana-Amitay & Berman, 1999; Cahana- Amitay & Sandbank,
(2000); Katzenberger, 1999; Kriz, Reilly & Saltzman, 2000; Verhoeven et al, 2002).
L(onger) units, as defined in the original project, take into consideration not only
syntactic, but also thematic and discourse properties in defining clause linkage or
inter-clausal connectivity. They extend what is known in the linguistic and discourse
literature as clause linkage (Haiman & Thompson, 1988), syntactic packaging
(Berman & Slobin, 1994: pp. 538-554), or connectivity (Berman, 1998).
Segmentation of texts into clause packages was based on the criteria proposed in the
above studies for Hebrew.
Thus, clause packaging is taken to throw light on how speaker-writers
organize or package information in texts. The ability to use more complex packaging
reflects the attempt of more proficient writers to create more tightly organized and
integrated combinations of information within the boundaries of a single "sentence",
and enables speaker-writers to present more information as background or secondary
compared with others by means of "the subordination of event component to a high
point and the conflation of event-complex" (Berman & Slobin, 1994, p.538).
41
In the earlier crosslinguistic study of oral narrative development, the amount of
interclause linkage per text, defined as "syntactic packaging", increased
quantitatively, and also qualitatively in terms of scope and type of packaging devices
with age (Berman, 1998). Berman showed a consistent age-related rise in overall
amount of subordinate clauses by children in different discourse languages. Cahana-
Amitay (2003) found that the scope of clause packages in the narratives of the
Hebrew sample of the original study, measured by clause/clause package ratio, was
not sensitive to development since it remains constant across ages and modalities. On
the one hand, these findings match Kess' claim that combinations of over three
clauses are difficult to process (Kess, 1993). Rather, on the other hand, the internal
structure and organization of the clause inside the packages - so called "syntactic
architecture" - change markedly with age and literature.
Against this background, as in the original study, I predicted that clause
packaging, measured by number of clauses per clause package would remain constant
with age in both research groups. On the other hand, with respect to the variable of
genre, I predicted that, clause packages would be longer in the expository texts since
these in general reflect greater linguistic complexity (Ravid, in press). As before, with
regard to population, I expected to find shorter clause packages in both low SES
research groups across genres.
2.6.2.2 Linking devices in clause packaging
As a second measure of inter-clausal syntax (in conjunction with lexical
usage), I also examined the linking devices that is, the kind of connection elements
used to link one clause to another. "Linking devices" in clause packaging were ranked
along a continuum in descending order of explicitness and transparency: syntactic <
syntactic-discursive < discursive . Syntactic links include all forms of subordination,
for example in an expository text of a junior-high non-Ethiopian girl: im xaver shelxa
loh ohev] she (subordination complement) ^ata pogea bo o kol miney dvarim] az
(juxtaposition) al ta' asu zot] ki az (subordination adverbial) ze yaxol ligrom
le^meriva alimut recax ve^xu ' if a friend of yours does not like that you hurt him and
such things, so don't do it, because then it might cause a quarrel, violence, murder
etc.'.4 Syntactic links also include coordinating conjunctions that occur between
4 In transcribing Hebrew examples, as before, I used a broad phonemic transcription. The subordinated markers are underlined. The examples are followed this time by only a standardized free translation.
42
consecutive clauses with a shared subject, and serve for grouping clauses together in a
package. e.g: in a narrative of high-school Ethiopian boy- leaxar miken hem hitvakxu
al kax] ve-basof hem hitxilu lariv al kax, 'later they argued about it and finally they
began fighting about it', Shay kam al raglav ve-רrac be-mehirot le-beyto, ' Shay got
up on his feet and ran home quickly', in an expository text of the same boy- lefi da'ati
carix lemager et tofa'at ha-alimut ve-lifol le-shipor ha-xevra, ' in my opinion there is
a need to demolish the phenomenon of violence and to act for the improvement of
society'.
Syntactic-discursive links included coordination of different-subject clauses,
e.g. in a narrative of an Ethiopian high-school girl - aval besofo shel davar aviha
halax l-tipol psixologi ve hem xazru le-shigra normalit, ' but finally her father went
for psychological therapy and they returned to the normal routine'.
Purely discursive links included discursively juxtaposed clauses; e.g. in a
narrative of an Ethiopian high-school girl- kvar shalaxnu axat la-shniya katavnu
mixtavei hitnaclot ve-slixa, 'already (we) sent each other (we) wrote letters of
apologies and forgiveness'.
Cahana-Amitay (2003) found a clear difference in the transparency of clause
packages in the narratives of the Israeli children compared with those produced by
adults. The adults used more syntactic devices alongside clearly identifiable discourse
relations. The links between clauses were primarily syntactic across the age groups.
I predicted the research groups would follow the subjects of the original study and
also primarily use syntactic links, which would increase with age.
With regard to genre, I predicted that the expository texts would include more
varied types of links. This is because the hierarchical discursive relations expressed in
expository texts are reflected in tight syntactic packaging in terms of embeddedness
and dependence. (Matthiessen & Thompson, 1988; Cahana-Amitay & Sandbank,
2000). Regarding the different populations, I also expected the middle class subjects
to include more varied types of links especially in their expository texts. This
prediction is based on the fact that upper-class children have a more developed
repertoire of syntactic means, which they properly deploy while producing different
types of texts. In comparison, the expository development of low SES children is
stagnated due to restricted channels of input.
In the present study a specialist in linguistics and two native speakers with
training in linguistics divided all the texts into clause packages independently. All
43
cases of disagreements were discussed and resolved. Reliability as measured by inter-
judge agreement reached as high as 99% 5. All clause packages were marked and
counted.
2.6.3 Lexical richness
Later-language lexicon is characterized by a more complex and larger lexical
inventory (Anglin, 1993; Gillis & Ravid, 2003). In this study lexical richness is
measured by lexical density - in the sense of relative proportion of content words per
text. This measure was used in research contrasting the modalities of speech and
writing (Halliday, 1985; Ure, 1971). Written discourse was characterized as having a
higher "lexical density", as measured in terms of higher percentages of content words
– items in the major lexical classes of noun, verb, adjective -- per total number of
word tokens in a text. It seems that lexical information is more important in written
discourse, since meaning has to be inferred only from the words of the text itself.
Lexical density was applied to examine the effect of modality in the English, Hebrew,
Icelandic, and Swedish data of the original project (Strömqvist et al, 2002). Lexical
density was defined in their study as “the proportion of content words to total number
of words, where a content word is a noun, a verb (excluding auxiliaries and copula),
an adjective, or a word derived from one of these” (p. 48). Results showed that the
written texts consistently scored higher on lexical density with age. In addition,
lexical density was used as a measure of text comprehensibility or text explicitness.
De Vries (1998) used lexical density as an accurate readability measure, and
demonstrated that there is a correlation between low lexical density and
comprehension test scores. Thus, less densely packed texts are more easily
comprehended, particularly among non-proficient readers.
Following the definition in the study of Strömqvist et al (2002), I decided to
measure the proportion of content words or “open-class items”, in the (written)
database of my research, since these items are known to be the primary bearers of
semantic or referential content in a text (Berman, 1988, 2001; Landau & Gleitman,
1985; Maratsos, 1988). As such, content words constituted the basic units for analysis
of lexical density in the present study.
5 I am indebted to Dalia Cahana-Amitay and Bracha Nir for their help in this connection.
44
In order to ensure reliability of the definition of “content words” in the present
study, judgements of the researcher of this study were compared with those of a
specialist in Hebrew language and linguistics and another native speaker of Hebrew
with training in linguistics. All three independently divided all lexical items in the
database of the two research populations into the following four lexical classes:
nouns, verbs, adjectives, and adverbs derived from adjectives (e.g., be-yesodiyut
‘with-thoroughness = thoroughly”, following the conventions used in the original
project. Inter-judge agreement reached 100% 6
The following predictions were specified with respect to lexical density as
measured by proportion of content words to total words per text.
Lexical density with regard to development
The growth rate of these content words was expected to increase with age.
This prediction is based on Verhoeven’s (1994) analysis of spontaneous speech and
picture descriptions of Turkish children living in the Netherlands, between ages 6 to 8
years, which revealed an age-related advance in the use of content words in Dutch as
a second language. And on the original project conducted by Strömqvist et al. (2002),
which also found significant differences in relative proportion of content words as a
function of age. Text length was expected to increase with age, especially between 7th
to 11th grade in both research groups, in line with what was found for the
crosslinguistic subjects (and see, too, for oral narratives, Berman & Slobin, 1994,
p. 31).
Lexical density with regard to genre
With regard to genre, lexical density was expected to be higher in the
expository texts, which tend to a greater density than narratives (Berman &
Verhoeven, 2002; Ravid & Cahana-Amitay, in press).
Lexical density with regard to origin
Again, lexical density was not expected to be sensitive to country of origin, for
the reasons noted earlier with respect to text length.
Lexical density with regard to population
With respect to population, I expected to find a higher lexical density in the
texts of the middle-class subjects of the original project compared with both the
6 I am indebted to Marit Sternau and Bracha Nir for their help in this connection.
45
research populations, and that this would apply across both the narrative and
expository genres. This prediction is based on research showing that low SES children
lag behind middle-class students in command of vocabulary, as measured in different
ways. For example, low-income fourth-grade children in the United States were
shown to slip below expected achievements in reading, and this is particularly marked
in tests of vocabulary (Chall & Snow, 1988). Hart and Risely’s (1995) longitudinal
analysis of early preschool children’s vocabulary from 42 families of different SES
backgrounds showed that at three years of age, the child with the smallest vocabulary
of the children from professional families had a significantly larger vocabulary
compared with children from welfare families. Weizman (1995) examined the
vocabulary of 53 low-income pre-schoolers in the Boston area from the point of view
of vocabulary knowledge needed for literacy success at school in interaction with
their mothers from preschool into early schoolage. Her findings suggest that even
though some low-income children are exposed to quite sophisticated and varied
maternal vocabulary input, their generally limited exposure to a wide and advanced-
level word stock is likely to condemn them to what she terms “literacy failure”. In
Israel, Kemp (1984) also found gaps between junior-high-school students from
different SES backgrounds, in terms of vocabulary and concept knowledge measured
by synonyms, exceptions, opposites, idioms and definitions.
2.6.4 Thematic content
An underlying assumption of this study is that culture is a significant source of
meaning construction and of self-identity. Many researchers considered the
importance of personal story-telling for the development of self in different cultural
contexts (Bruner, 1986, 1990; Miller, Fung & Mintz, 1996; Sperry & Sperry, 1995),
since the “narrative is acknowledged to be a primary means by which children
develop the voice of their own culture” (Invernizzi & Abouzeid, 1995, p. 3). In a
personal-experience narrative, the speaker-writer's identity and behavior respond to
socially established expectations. These expectations reflect how belonging to a
certain ethnic group may affect the speaker-writer's action in particular domains of
life (De Fina, 2000). Consequently the texts constructed by children from different
cultures and backgrounds can be expected to differ mainly in the thematic content
they choose to write about, and the content of their narratives will reflect the norms
and accepted behaviors of their community.
46
The common belief is that the main function of the home languages of
immigrants is to serve for inter-group communication and for expressing their
ethnicity (Extra & Verhoeven, 1993). This study adopts the view of scholars who
assume that the narratives of immigrants that are told in the non-home language (in
this case, Hebrew), may also serve as a means for communication between members
of the shared ethnic group, on the one hand, and the general society (Israel), on the
other. In this encounter between communities and cultures, the narratives in the non-
home language may function as a potential means for self and/or collective
identifications (De Fina, 2000).
In order to specify “thematic content”, I decided to rely on the use of content,
open-class words as defined in the preceding section, as a tool for analysis of
narrative text content. This represents (to the best of my knowledge,) an original
attempt to provide a quantitative measure to describe the content of the topics referred
to in the narrative texts -- although a partly similar procedure was adopted in a recent
study of nominalized content of the predicates in the English and Hebrew texts of the
crosslinguistic sample (Ravid & Cahana-Amitay, in press).
Thematic content of the narrative and/or expository texts in our sample were
analyzed for thematic content – as measured by use of content words – along several
dimensions, including reference to physical aggression and violence, family and home
situations, reliance on the school-based scenes shown in the video clips, and so on (as
outlined in Sections 2.6.4.1 to 2.6.4.4 below).
2..6.4.1 Thematic content with regard to content words expressing physical
aggression
The first prediction with regard to thematic content was that children from
low SES backgrounds would use more nouns, verbs, and adjectives expressing
physical aggression in comparison to their high SES peers. Reports show that low
SES children are more exposed to situations of violence and to physical and mental
abuse in their families and in their immediate surroundings than are their higher-class
peers (Dileonardi, 1993; Sherman & The Children's Defense Fund, 1994). Lev-
Weisel (2001) also used thematic-content analysis in comparing Israeli first graders
from poor backgrounds with children from wealthy families, with regard to their
perception of socio-economic status and future careers. She found that children from
poor families reported more about violent behavior of their parents, and even used
more expressions of physical aggression in negative terms (such as “they do not hit
47
me”) when describing positive relations in their families. And the highly relevant
study of Walton and Brewer (2002) of 364 narratives of inner-city 4th and 6th graders
about a personal experience with interpersonal conflict shows that, with age, children
in the high-risk neighborhood were more likely to write about physical aggression and
acts of violence, in contrast to children attending a moderate- to low-risk school,
whose narratives showed a decline in reference to such topics from 4th to 6th grades.
With regard to the Ethiopian group, around one-fifth (21%) of Ethiopian
children are known to the social services as being in direct risk situations, that is, as
suffering from neglect by their parents, as exhibiting behavioral or functional
problems at school or at home, and suffering from violence perpetrated by their
parents or witnessing domestic violence. This rate of children at risk in the Ethiopian
community is three times higher than among the overall population of Israeli children
(Habib, 2001). I further expected that low SES children would write more about their
experiences in violent situations, since they are more accustomed to talking about
them either to their friends or to their school counselors. This prediction is based on
personal communication with the school counselors of the children of the study, who
confirmed my intuitions, noting that these students usually initiate more meetings
with them in which they expose more such information than high SES children
attending other schools where they worked.
To test this prediction, all the open class items (content words) in the lexicon
of the present study and all those in the Hebrew sample of the original project were
classified as “plus or minus aggression”. This was done by myself in consultation
with several educated outsiders, who generally agreed with my evaluations. The
relevant measure which this yielded was a comparison of + aggressive vocabulary.
2.6.4.2 Thematic content with regard to family members
Another prediction relating to thematic content was that the narratives of low
SES subjects would include more references to conflict events which had occurred in
their homes involving family members than subjects in the middle-class Hebrew
sample. This prediction holds for both Ethiopian and non-Ethiopian low SES children
since these children, as noted, are more exposed to violent situations in their
immediate surrounding. With regard to the Ethiopian subjects it is well known that
they have high respect for their enlarged multi-generational families, and usually feel
obligated to its members’ well-being and to comply with the traditions and
requirements of their culture (Ben Ezer, 1992; Erlich, Kaplan & Salmon, 2002). As a
48
result, they tend to identify themselves above all with their families, making it likely
that they will refer to their home and family in their self-experience narratives. In
addition, as noted, low SES children are more accustomed to sharing information
about conflicts in their homes with their school counselors. In this they differ from
children from high SES background who regard this kind of information as more
private and as a source of possible shame.
2.6.4.3 Thematic content with regard to reliance on the video clip
Vocabulary was also used in order to ascertain to what extent the expository
texts of low SES children include descriptions from the scenes shown in the video
clip, which depicted different conflict situations in a school setting. As noted, the
video served as a trigger for producing texts with a shared thematic content
(Section 2.4.1). Subjects were explicitly instructed not to rely directly on the contents
of the video clips in their writing. The prediction was that, as in the original project
(Berman, 2000; Tolschinsky et al, 2002), the younger children would rely more
directly on the video in their expository texts than the older ones.
I also wanted to find out whether background origin would make a difference
in this respect. Here the prediction was that the Ethiopian subjects would regard the
video scenes as if they were the set saying, culturally accepted axioms, or proverbs
which often precede a story with moral typical of oral cultures. African proverbs, as a
technique of verbal expression of abstract ideas through compressed and allusive
phraseology, are interwoven with other aspects of linguistic and literacy behavior
(Finnegan, 1970). Proverbs as an anecdote in a nutshell must be seen as arising from
the context, thus in these cultures there is no proverb without a situation. As noted,
proverbs have a distinct structural and concise form, and are distinguished by the
popular acceptance of the truth tersely expressed in them. In some cases proverbs are
also used to express group identity in an almost explicit manner. Hazan-Rokem
(1993) showed that proverbs are often used when a particular ethnic group of relative
newcomers, in possession of a distinct cultural heritage and well-defined cultural
repertoire, actively and directly directs communication to the receiving society.
With respect to narrative content as a function of country of origin, stories that
begin with proverbs are often told for internal community communication in the
Ethiopian culture. These stories are aimed at teaching moral obligations, norms and
standards to the members of the community. Such proverbs are highly important and
represent the main ideas of the story (Ben Ezer, 1992; Noy, 1988; Rozen, 1999). I
49
therefore predicted that the Ethiopian subjects would refer more to the specific
contents of the video scenes in constructing their narratives than would their lower
SES non-Ethiopian peers. These specific contents, like the proverbs, might represent
significant normative ideas or authoritarian prescriptions for them.
2.6.4.4 Thematic content with regard to affronts to honor
Vocabulary was also used to find out whether the two research groups differ
with regard to reference to affronts to honor. It was predicted that low SES children,
especially of Ethiopian parentage, would refer more in their narratives to remarks or
behavior offensive to their self-respect than would the middle-class peers in the
original project. This prediction is based on research comparing Israeli children from
high-income families with children from poor backgrounds, which shows that the
latter are more exposed to mockery and humiliation in their immediate school
surroundings and in the community at large (Lev-Wizel, 2001). Moreover, reports on
children from disadvantaged backgrounds indicate that they are subject to neglect,
maltreatment, and mental abuse on the part of their parents (Dileonardi, 1993;
Sherman & The Children's Defense Fund, 1994).
In addition, some studies showed that belonging to a certain ethnic group may
affect the content of the stories people tell. For example, the content of personal
experiences of Mexican workers in the U.S.A. indicates that these immigrants use
ethnic identity as a central identification category of self and others in their stories
(De Fina, 2000). This study supplied an additional support to my prediction that the
Ethiopian children would make more reference to affronts to their own or their family
honor, since in the Ethiopian community offences to someone's dignity, impertinence,
insults or humiliations are highly unacceptable and are reprehensible types of
behaviors (Ben Ezer, 1999).
2.6.5 Violations of linguistic norms
In view of the generally low level of literacy and formal schooling that
characterizes the backgrounds of both research populations and the fact that the
Ethiopian subjects are of Amharic, non-Hebrew speaking backgrounds, the present
study undertook an analysis of deviations from standard Hebrew usage. Following
Berman’s (1999) model of level of non-native language proficiency as defined along a
50
continuum from obligatory rules of morpho-syntactic structure (“core grammar”) via
lexical convention to discourse-motivated and culturally-determined factors of
language use, three types of violations were specified: morpho-syntactic constructions
(Section 2.7.1), lexical usage (2.7.2), and register mixing (2.7.3).
2.6.5.1 Grammatical errors
Three types of morpho-syntactic errors were defined: in grammatical
agreement (marking of gender, number and person concord between subject and
predicate and between noun and adjective); choice of prepositions; and overuse of the
generic subordinating conjunction še- ‘that’, for example, in place of the more
specific form kše- ‘as-that = when’.
Children acquiring Hebrew as a first language gain basic command of
grammatical agreement marking of gender, number and person, and they make correct
use of a range of different prepositions by early preschool age. Moreover, their
proficiency in these domains increases with age and level of literacy (Berman, 1997b;
Dromi, 2002; Kaplan, 1983; Ravid, 2002).
Based on these studies, my first prediction regarding grammatical accuracy
was that grammatical errors would decrease with age and with development of the
lexicons of the non-Ethiopian subjects, for whom Hebrew is a first language.
In contrast, with regard to background origin, I expected grammatical errors
to be more pervasive in the Ethiopian grown-ups than among the non-Ethiopians,
since the Ethiopian children are less exposed to literate and formal Hebrew in their
homes, and most of their parents do not read or write Hebrew. Moreover, most of the
Ethiopian subjects acquired Hebrew as a second language, since they were born in
Ethiopia (Section 2.2.1). Thus, this kind of comparison between the two research
groups might shed light on the role of home backgrounds in linguistic proficiency.
With regard to the different populations, I assumed that grammatical errors
would be higher in both of the two research groups than in the original project. This
prediction is based on research which showed that school age children from
advantaged homes do consistently better on language-based tasks, in the sense of
being closer to the norms of educated Hebrew usage in a range of different
morphological and morphophonological features of written Hebrew than those from
disadvantaged backgrounds (Schwarzwald, 1978, 1981). Further, environment
51
variables in the preschool educational systems of high SES children (nursery schools
and kindergartens) were shown to have a positive effect on the development of the
vocabularies and decontextualized language skills of these children in comparison to
their peers from lower socio-economic backgrounds (Dickinson & Tabors, 1991;
2001).
In an attempt to look at the extent of grammatical error in low SES children, I
also compared results with another group of comparison. The goal of this comparison
was to shed light on the schools' success in teaching basic grammar as an essential
component of proficient writing and academic achievement in the humanistic subjects
studied at school in general. This group of comparison consisted of 42 low SES,
Israeli born students of the same age groups who were all native speakers of Hebrew.
This research, like the present study, also adapted the overall design and methods of
data elicitation from the original project (Salmon, 2002). The students in Salmon's
study, unlike the research study, attend schools of high scholastic achievements and
academic demands. Their schools were defined as high on all the following seven
factors, except SES and class size: 1) socio-economic level of the students
2) percentage of teachers with academic education 3) academic achievements of the
students measured by national norms 4) level of the municipality resources in the
schools 5) application of adequate learning materials, current programs and
innovative educational projects 6) class size measured by 13-17 students or 22-30
students per class 7) counseling system which uses professional teachers and
programs for students with special needs. Research on learning environments of
schools serving low-income and those serving high-income children showed
considerable differences between them. Duke (2000) characterized the differences of
print environments in these populations not only by means of amount of literate
activities, but also by means of their quality and variety. She visited first grade
classrooms in some very high- and very low-SES schools in the Boston area looking
at the kinds and amounts of print experiences. She found classroom libraries in low-
SES schools were about 40% smaller than those in high SES schools. Moreover, in
the low SES classes fewer books were added during the year. The children in low-
SES classes used books less often and in fewer ways. In comparison, high-SES
classrooms displayed more and more varied types of printed material. The teachers in
these classes more frequently called attention to books, and also integrated them with
topics of study. In these more advantaged school environments, students were also
52
given more choice in their reading. The children were more likely to have a high
degree of authorship in their writing and were more likely to write for audiences
beyond the teacher. Hycock (2001) estimated that less effective teaching is one of the
main causes of the achievement gap between high and low SES children.
Based on these studies, I predicted that with regard to schools' success in
teaching language, low SES children in relatively disadvantaged school systems
would make more errors in grammar and lexicon than children of low SES
backgrounds attending well-established schools that are known to provide a high level
of education.
As a further check on language usage at this level, the heads of the language-
teaching teams in the research schools were asked to report what they considered
common grammatical errors or deviations in the Hebrew used by teachers in the
schools. This was done in an attempt to find out whether grammatical misusages and
non-standard language of the students can be related to their linguistic input not only
at home but also at school.
2.6.5.2 Lexical infelicities
Deviations from conventional lexical usage in terms of preciseness and felicity
of vocabulary selection were defined here as deviations from collocational
appropriateness (single words or multilexemic expressions) in a given clause in the
form of combination of words that strike the reader-hearer as deviating from accepted
or conventionalized native usage. The term “collocations” is used mainly in British
linguistics as referring to “habitual word accompaniments”. Thus, Firth (1957) defines
collocations as representing lexical relations along the syntagmatic axis, in the sense
of the ability of a given word to combine with other words. In the same spirit, Lyons
(1977) devotes an entire section headed “Syntagmatic lexical relations” to the topic of
“collocational restrictions”, which he illustrates at some length by comparing the
English verb ride with its apparent German counterpart reiten. Collocations are
defined in a dictionary of linguistics as “two or more words, considered as individual
lexical items used in habitual association with one another in a given language”
(Hartmann & Stork, 1972, p. 41). The researchers further point out that “every
individual word in a language has its range of collocations which limits its meaningful
usage, and equivalent words in different languages rarely, if ever, have the same range
of collocations”. Along similar lines, Crystal (1985) defines a collocation as “a term
53
used in lexicology to refer to the habitual co-occurrence of individual lexical items”
(1985, p. 55).
The criteria proposed by these characterizations of collocations note that they
are non-compositional, non-substitutable, and non-modifiable. As a result,
collocations cannot be translated from one language to another, nor can they alternate
words within or across set expressions. This is shown by the illformedness of such
possible sequences as raise war or wage peace in place of conventionalized wage war
or by an analogous expression used by a Hebrew-speaking high-school student in a
comparable sample to the present one (Berman & Ravid, 1999) who wrote le-hashrot
shalom ‘to-immerse (in) = to-inspire peace’ in place of the conventional collocation
le-hashkin shalom ‘to-instil peace’. Such usages strike a discordant or “creaky” note
(from the Hebrew term xarika ‘creaking, grating’), in the syntactic or lexical context
in which they occur, and so yield a sense of bizarreness, even though the speaker-
writer’s intention and the (intended) semantics of the expression are usually clearly
understandable.
Research into command of collocational usage among second language
learners shows that they have difficulty in acquiring collocations in the target
language. One explanation for this is related to interference or negative transfer from
the first language, due to inappropriate reliance on L1 lexical usages, particularly in
translating tasks (Baths & Eldaw, 1993; Biskup, 1992; Farghal & Obiedat, 1995).
Another reason given for collocational errors of second language users is attributed to
lack of culturally related knowledge of collocations (Teiliya Bragina, Oparina &
Sandomirskaya, 1998). One of the goals of the second language researchers was to
help teachers to identify effective an way of promoting phraseological competence in
English. Such studies are aimed at improving second language learners’ proficiency in
lexical expression. To the best of my knowledge, the present study is the first to look
at collocational “discordances” or inappropriate usages in L1 speaker-writers, where,
following Polinsky (1997) and Romaine (1995) the term “L1” refers not necessarily
to the first or home language, but to the dominant language in the sense of the primary
language of use - typically the school language as in the present study. The decision to
examine typical collocational violations in my database arose from my sense that the
language of texts produced by low SES speaker-writers in their main language was
not exactly “ill-formed” in the sense of violating target language grammar. Rather, it
54
contained numerous deviations from linguistic appropriateness, and that seemed to be
largely a matter of lexical usage.
I therefore examined collocational violations of students at different levels of
schooling in the process of becoming proficient writers in their dominant language.
Again, as far as I know, this topic has not been studied to date from a developmental
perspective with a discourse-anchored methodology. Underlying this analysis is the
assumption that, as pointed out by Crystal (1995), appropriate use in collocations of
particular lexemes is dependent on native-like intuitions, the sense of or feeling for a
language, an ear for what is idiomatically correct or usage-appropriate known as
sprachgefühl. Haller (1988) defines sprachgefühl as an intuitive certainty or sureness
of touch resulting from talent, experience and analogizing. This is revealed when a
speaker-writer deals with language both in linguistic production and in the meta-
linguistic evaluation of what is linguistically right and proper. Schulte (1988) suggests
that people can develop their sprachgefühl by extending their lexical repertoire, which
implies that choosing the appropriate word is itself a kind of strategy which can to
some extent at least be taught or learned. Native speakers gifted with a well-
developed sprachgefühl typically have command of an extensive vocabulary and rich
experience with language use in diverse contexts, which might quite generally reflects
a certain level of “linguistic literacy” (Ravid & Tolchinsky, 2002). An interesting
feature of deviations from collocational appropriateness, which presents a challenge
to any analysis, is that while native speakers judges sense that something is amiss
when they encounter such a deviation, they have a hard time specifying exactly where
the infelicity lies or what is its source.
Against this background, I conducted a pilot study on a sub-sample of the texts
collected for the present study, by presenting examples of “discordant” usages to a
group of graduate students and researchers in Hebrew language and general
linguistics, selecting instances of what appeared to me intuitively “misusages” but not
necessarily “errors”. Following this, together with another researcher, an expert in
Hebrew language and linguistics 7, we marked up all expressions in the database that
each of us independently considered deviating from conventional usage, discounting
errors in basic grammar and anaphora. Rejecting all instances that one or both of us
did not agree on, we reached a total of nearly 100 such expressions.
7 I am indebted to Nurit Assayag for her insight and help in this connection.
55
Since lexical usage and collocational appropriateness are among the most
highly language-particular and conventionalized features of a language, I did not even
attempt to translate the expressions under discussion here. Instead, I hope to provide
some feeling for what is involved by analyzing selected examples, which seem to be
particularly indicative of the phenomenon. An example is the expression used in the
narrative of an Ethiopian high-school girl: še-yiheye ba-tinok eyze-še-hu pega 'that-
will-be in-the-baby some-kind-of stigma’ = ‘, that the baby will-have some kind of
defect'. In standard Hebrew, the correct restricted collocation is lelo pega 'with no
defect', and not še-yiye pega, the way it was used in this example8; a second example
is from a the same narrative of the above Ethiopian high-school girl who described a
woman’s behavior as: isha ha-sotיmet et pane-ha ‘woman who-shuts ACC face-her
= a woman who shuts her face’ (cf. conventional li-stom et ha-pe ‘to-shut
(someone’s) mouth’, with the higher-register bound form paneha used for ‘her-face’
rather than everyday ha-panim shela ‘the-face of-her’, but in a totally non-normative,
unconventional collocation, one which mixes this high register usage with the low
level, even slang expression li-stom et ha-pe). In the Ethiopian culture, emotional
restraint is considered a virtue, while open expression of emotions is considered taboo
in human relations, as noted by Ben Ezer, 1999. Thus, a woman who “shuts her face”
is one who complies with the norms of her culture. A third example from the narrative
of a non-Ethiopian high-school girl writing about a quarrel between friends is: še-ota
yalda adayin pgu’a gam mimeni lamrot ha-slixot ‘that-same girl (is) still offended by
me too, in spite-of the-pardonings’, where the last word is a misuse of the term for
religious penitential prayers said on a Jewish holiday, in place of conventional
hitnatsluyot ‘apologies’ or the related, more colloquial and yet more appropriate še-
bikashti mimena slixa ‘I asked from-her (a) pardon = I said I was sorry’.
These kinds of lexical infelicities were divided into six sub-categories: (1) use
of a single inappropriate word, as in this last example in the previous paragraph;
8 In transcribing Hebrew examples, I use a broad phonemic transcription representing standard, but colloquial rather than normative Hebrew pronunciation; words are stressed on the final syllable unless marked by an acute accent as having non-final stress; hyphens in the Hebrew transcription stand for morphemes that form part of the next word which are separated by spaces in European languages, but form part of the next word in standard Hebrew orthography; the relevant violation of lexical or collocational convention is underlined; and examples are first given a literal, morpheme-by-morpheme gloss (the label ACC stands for the accusative direct-object marker et); and this is followed by a standardized free translation.
56
(2) inappropriate free collocations as in use of the term for a religious ritual in the
second example; (3) inappropriate combinations of words in set, restricted
collocations, as in shutting her face instead of her mouth; (4) conventional
collocations in an appropriate context (5) over-collocation, in cases where an entire
expression is used rather than an appropriate part of it; and (6) under-collocation, the
converse, where only part of the a collocation is used in place of the entire expression.
In order to apply a restricted collocation correctly and appropriately, the
speaker-writer must have encountered and internalized these more complex lexical
forms at some time in his or her past experience, since it is not usually possible to
figure out the meaning of their separate components without prior learning. As a
result, people who misuse such set expressions can be assumed to have access to a
relatively sophisticated, high-level, and extensive lexical repertoire from which to
select or invent items that occur in collocation with one another, but they may be
unable to use them in a way that precisely encodes the meaning they wish to convey.
For example, use of an expression like raise war indicates that the high-school girl
who wrote this has some vague or partial knowledge of the conventional expression
wage war, or at least some familiarity with the relevant semantic notion she wished to
express. This type of linguistic usage involves a special kind of creativity or lexical
innovation and indicates that the writers are in fact in the process of developing their
sprachgefühl. In comparison, at this advanced stage of developing language
proficiency, learners may not be sure whether they know something and how well
they know it. In such cases, speaker-writers do not themselves appear to be aware of
the fact that they are violating conventional usage, since there is no evidence that they
attempt to correct or change these discordant or inappropriate expressions. Besides,
the accuracy with which learners assess their knowledge and use of language could be
a function of lexical knowledge and/or cultural background.
Laufer and Yano (2001) showed that university students in Israel, China, and
Japan who demonstrate better command of the lexicon and a larger vocabulary are
able to assess their own performance more accurately when they encounter unfamiliar
words in English as a second language than are students with smaller vocabularies
and a less advanced lexical repertoire. Moreover, the Japanese learners in their study
exhibited the lowest mismatch between perceived and objective lexical understanding,
while the Israeli students, coming from a culture where unwillingness to admit
ignorance or failure is common, exhibited the largest mismatch.
57
Against this background, I predicted that the Ethiopian students, who belong
to a culture similar to that of the Japanese, where caution and modesty are virtues,
would carefully select their words and would adhere to familiar lexical items. This
might lead them to commit fewer lexical infelicities than their Israeli-born non-
Ethiopian peers. Besides, the Ethiopian students might have less developed lexical
repertoires to select from, since they are less exposed to more literate or formal
Hebrew in their immediate surroundings. It should be noted, however, that an
important component in Amharic in a conversation dealing with serious or important
issues is the use of expressions aimed at disguising the real sense of what speakers are
saying in order to highlight their underlying or hidden intentions. As a result, the use
of hints and double meanings or innuendoes and ambiguities are highly respected
devices in the Ethiopian community (Levine, 1965). An Amharic speaker-writer who
makes rich use of idiomatic expressions, proverbs, set collocations, and a variety of
related rhetorical devices tends to be highly regarded (personal communication with
Dr. Anbessa Teferra, a specialist in Amharic languages). The Ethiopian subjects in
this study might thus be expected to transfer this marked preference for richly varied
use of a highly flowery vocabulary from Amharic to Hebrew. This could be the
source of a relatively large number of lexical infelicities in their texts, since these
subjects have not fully mastered high-level literate Hebrew.
Against this information, I expected that the middle-class subjects from the
original project would reveal fewer lexical infelicities, since students of more
educated, higher-class backgrounds are known to score higher on vocabulary tests
than their low SES peers (Chall & Snow, 1988), and also on innovative language use
as measured by lexical coinages (Kemp, 1984). Further, as noted earlier, children
from higher-SES backgrounds also demonstrate more familiarity with rare vocabulary
items than their low SES peers (Hart & Risley, 1995).
2.6.5.3 Register mixing
All lexical infelicities were further specified for level of language usage or
register, where “linguistic register” is used in the sense of level of linguistic usage,
ranging from everyday casual or colloquial style to highly formal, carefully monitored
forms of expression (Biber, 1995; Ferguson, 1994). Specifically, all instances of
“register mixing” were noted, where this refers to use of colloquial, everyday
language (single words or multilexemic expressions) in the same context -- within the
58
boundaries of a single clause – as high-level, formal usages. This procedure was
adopted as part of the analysis of general lexical development and proficiency, to
evaluate the socio-communicative skills of low SES children. The aim was to
ascertain whether the subjects in this study are able to use a variety of linguistic
patterns suitable to different contexts and social situations in constructing monologic
texts, since consistently appropriate and accurate register selection is a very late-
developing and highly sophisticated type of language knowledge (Berman, 1999).
The prediction was that the Ethiopian subjects would produce more register mixing
than the non-Ethiopians, for reasons related to their possibly more marked tendency to
violate norms of lexical appropriateness in general. Similarly, I predicted that low
SES children would produce more register mixing than their middle-class peers in the
original project, since Israeli schoolchildren of low SES background were found to
show more difficulties in adopting different style of usage in the appropriate situations
than middle- to high-SES children (Davis, 1978; Kais, 1979; Kemp, 1984; Shtal,
1977).
2.6.6 Global text construction
The top-down analysis of discourse refers to the whole text as its unit of
analysis as a whole. In the present study the analysis focuses on the global text
structure and its function. The aim of this analysis is to identify developmental
differences interacting with genre across the different populations. This top-down
analysis is based on the assumption that there is an internalized narrative schema
characterized by a canonic structure having a setting, an episode(s) and a coda
(Berman, 1997c; Berman, 2001; Labov,1972, 1998; Van Dijk, 1980). In comparison,
the expository text does not have a universal schema. Moreover, the structure of this
type of text is not as well defined as the structure of the narrative. Although there is
no canonical structure for the expository text, some researchers describe it in terms of
introductory “core” propositions, which are elaborated by illustrative or delimiting
“satellite” discourse elements (Fox, 1978; Matthiessen & Thompson, 1988). Britton
(1994) describes the structure of the expository text as “move on” statements, which
introduce a new topic, “expand” which develops the already introduced topic, and
“unitize” which summarizes information previously mentioned.
An adequate examination of narratives and expository texts requires
comparable units of analysis. The original project used the opening and closing
59
elements of the two types of texts as comparable units of analysis (Berman &
katzenberger, submitted; Tolchinsky, Johannssen & Zamura, 2002). Tolchinsky et al
found that the analysis of opening and closing elements of texts produced by the
English, Spanish and Swedish participants in the crosslinguistic project showed that
the youngest subjects in the study already had the ability to open narratives in a
canonical way. In addition, only the more mature subjects could open expository texts
with a direct reference to the topic, as a means of introduction. Berman and
Katzenberger (submitted) used the opening of the narratives and expository texts as a
window on the text as a whole. They examined the openings in terms of discourse
functions (background information in narratives and introducing the topic in
expository texts), organizational pivot (temporality in narratives and generality in
expository texts), and linguistic forms (verb tense and predicate semantics in
narratives and nominal structure and content in expository texts). They found that the
openings to narrative texts emerge as better constructed at an earlier age than
openings to expository texts. In comparison, fully proficient openings are produced
later in development for both types of texts.
I wanted to refer to the texts as wholes, so I followed Katzenberger's model
(Katzenberger, submitted b). She divides the two types of texts into three obligatory
analogous components representing functional segments of discourse. The three
segments are beginning, middle and end. In narrative, the beginning is the setting,
which typically presents the temporal, spatial and motivational framing of events, the
middle narrates the events, and the end provides a resolution (with or without a coda).
In expository text, the beginning serves as the introduction, which typically presents
the text topic, the middle expresses the idea(s) on this topic, and the end contributes a
conclusion.
In this study, I consider the ability to produce text global structures a universal
cognitive one, which does not depend on population and origin. This cognitive ability
is a part of every human being’s competence. Following this proposition and based on
the results on the Hebrew sample in the crosslinguistic study, I predicted that the
ability to construct global text structures would not be sensitive to population. I also
predicted this ability to increase with age and level of literacy and schooling. I further
predicted that the ability to produce global structures is more accessible to the
younger children in the narrative texts. With regard to genre, I predicted the ability to
produce global structures would develop latter in the expository texts.
60
In the present study, all the texts were divided individually into three
components by two judges: an expert in discourse analysis and myself. Inter-judge
agreement was about 92%9. The scope of each component of global text structure was
measured in terms of number of clause packages. I predicted that text components'
length would increase with age. With regard to genre, I predicted that the number of
clause packages per each component would be higher in the expository texts.
Regarding the different populations, I predicted the text component's length would not
be sensitive to origin and population.
Katzenberger (submitted a) also suggests three levels for the organization of
the flow of information in both types of text. The first level is characterized as linear
or a minimal level, a level in which the text is pair-wise ordered. In narrative texts this
level contains at least two chronologically ordered events, while a linear expository
text contains a nucleus followed by at least one complement. A partially hierarchical
level is defined as a text, which contains global and local-level elements (content
and/or structure). In a narrative, the global level introduces the problem and/or
presents its resolution. In an expository text the global-level introduces and/or sums
up the main idea(s) in a form of an explicit abstract generalization. The third fully
hierarchical level is characterized by a bi-directional interplay of flow of information
between generalizations and specifics, which is explicitly marked by linguistic means.
In a narrative text, the problem and its resolutions are presented and the retrospective
story-teller’s point of view is marked. In comparison, in an expository text the main
idea(s) and its summary are presented using an explicit abstract generalization.
Transition from core to satellite elements is usually explicitly marked by linguistic
means, like the expression bederex~agav ‘by the way’ to signal a digression from the
topic.
Katzenberger and I tried this analysis on the written texts subjects of the
original study with the aim of finding effects of age and genre on global text structure
(Schleifer & Katzenberger, 2001). We found that all the narratives of the subjects
from the original study, even those of the 4th graders, had partially hierarchical
structures. In comparison, expository text construction lagged behind that of
narratives across all subjects. Therefore, I predicted that organization of information
would increase with age and that the young subjects of the research study would use
9 I am indebted to Irit Katzenberger for her insight and help in this connection.
61
more complicated levels of organization of information in their narratives before
using them in their expository texts. I also predicted that organization of information
would not be sensitive to origin and population.
In this study, a special coding was used for global text structure for both text
types. The opening component of the narrative (setting), which contains information
about the place, time, opponents (the characters in a state of conflict), state of affairs,
and motivation for story telling and elements of the story script, was coded as
OPEN:SET. The middle component of the narrative (episode/s), which contains at
least two chronologically ordered events, was coded BOD:N (narrative). The closing
component (resolution, and or coda), which contains resolution to the problem and
sometimes also refers to the beginning and to the motivation for story-telling, was
coded in the same method as CLO:RES. Expository texts in this study were also
divided into three analogous components. The opening component is “introduction”,
which contains information about the main idea/s and was coded OPEN:INT by the
same rational. The middle component (core proposition) contains supplementary
elaborations and illustrations to the main idea, and was coded BOD:E (expository).
The closing component is characterized by conclusions, summaries and occasionally
connections to the main idea/s and was coded CLO:CON (See Appendix II).
62
Chapter 3: Results
The overall goal of this study was to understand how schoolchildren of
different ages and from different cultural and socio-economic backgrounds write
monologic narratives and expository texts. Two sets of data were elicited: 1) Written
narratives and expository texts, and 2) questionnaires examining school and home
literacy and attitudes towards language. The written texts yielded a corpus of 192
texts: 2 texts x 32 subjects (two research groups - 16 Ethiopians and 16 non-
Ethiopians in each age group) x 3 age groups. The literacy questionnaires consisted of
114 texts (96 questionnaires from the two low SES research groups and 18
questionnaires from a middle-high SES control group). Results of the analysis of the
written texts are presented in Part 1 of this chapter and of the literacy questionnaires
are presented in Part 2. Each part begins with a description of the measures in each of
the two sets of results.
Part 1: Results of the Written Texts
Introduction
The database of this study includes 192 texts: 96 narratives and 96 expository
texts. Results are presented below for the database of 192 texts (half narrative, half
expository) following the order in which categories of analysis and predictions were
described in Chapter 2. I start by presenting an analysis along three dimensions:
measures of text length (Section 3.1); syntactic complexity (3.2) in terms of number
of words per clause, (2) number of clauses per clause packages and (3) number of
types of linking devices; and lexical richness (3.3) in terms of lexical density
measured by number of content words per text.
A more qualitative measure is thematic content (3.4) as measured by four
indicators: (1) percentages of expository texts with references to the scenes from the
video, (2) percentages of content words (Ocs) expressing physical aggression, (3)
percentages of narratives describing conflict in the subjects’ homes and (4)
percentages of the narratives with references to affronts to honor.
The next analysis (Section 3.5) presents three types of violation of linguistic
norms: grammatical errors, lexical infelicities and register mixing. Global text
construction (3.6) is presented in terms of overall discourse structure (an opening, a
63
plot and a closing), by text component length (measured by number of clause
packages per text component), and also in terms of organization of information
(measured by three levels of organization of information). The last section includes an
overall summary of the results (3.7).
Each section begins with the data concerning the relevant research predictions
followed by a comparison between the two low SES research groups: the Ethiopian
and the non-Ethiopian subjects across age and genre. This part is followed by a
comparison between the original subjects of middle-class background, and the
Ethiopian and non-Ethiopian subjects as one low SES group. Wherever possible, a
third comparison of all the low SES research subjects with another group is provided
relating to low SES subjects who attend well-established schools (Salmon, 2002).
Multivariate analyses of variance were calculated whenever possible across
these measures, relating to factors of age and grade level (G= Grade school, 4th
grade; J= Junior high school, 7th grade; H= high school, 11th grade), genre (Narr=
narrative, Exp= expository); origin (Eth= Ethiopian, NE= non-Ethiopian) and
population (research study, original study).
All the data was divided into clauses, clause packages, clause types, global
text components, and types of violation of linguistic norms by at least two judges.
Interrater reliability as measured by degree of agreement between judges ranged
between 92.3%-100% across these variables.
3.1 Text Length In this study text length was measured by number of words, and by number of
clauses (Section 2.61).
3.1.1 Text length in words
Table 2 describes text length by means of number of words per text across age,
genre and origin. As expected, text length increases with age. For means of words in
the narratives, a between group ANOVA yields significant effect of age
F(2,90)=58.97, p<.001. Contrast analysis between 4th grade and 11th grade yields
significant effect F(1,90)=99.30, p <.001, and also between 7th grade to 11th grade
F(1,90)=76.05, p <.001. For means of words in the expository texts, ANOVA yields
significant effect of age F(2,90)=94.65, p<.001. Contrast analysis between 4th and 7th
grade yields significant effect F(1,90)=6.40,p <.05, and between 4th grade and 11th
64
grade F(1,90)=168.39, p <.001, and also between 7th grade and 11th grade
F(1,90)=109.15, p <.001. For means of words ANOVA with repeated measures yields
significant effect of genre F(1,93)=19.43, p<.001.
There is a clear developmental pattern for text length for all types of texts such
as mean number of words increases with age across the two groups. Moreover,
narratives are longer than expository texts across the board. Narrative texts average
125 words per text in the Ethiopian and 109 in the non-Ethiopian compared with 78
and 88 respectively in the expository texts. Moreover, this inter-genre difference in
number of words is more marked in the Ethiopian narratives and in the expository
texts of the non-Ethiopians. The development of changes in number of words also
differ in the two research groups: the high school group of Ethiopians shows a greater
increase in length in their narratives from 7th grade to the 11th grade, whereas the non-
Ethiopian group shows a more rapid shift in length in both their narratives and
expository texts from gradeschool to junior high. There is no significant difference
between the two research groups across origin with regard to text length.
Table 2 Means and standard deviations number of words per text, by age,
genre and origin
Ethiopian non-Ethiopian Age N=16 Narrative Expository Narrative Expository G SD
57.8 (32.9)
33.8 (17.4)
50 (34.4)
25.1 (17.9)
J SD
66.2 (31.6)
54.7 (22.1)
83.9 (55)
57.2 (32)
H SD
253.3 (113.8)
147.3 (65.3)
193.9 (92.1)
183.5 (64.2)
Compared with the data of the original study (henceforth "crosslinguistic " in
figure 1 etc.) for mean number of words in the narratives, ANOVA yields a
significant effect of population F (1,149)=24.01, p<.001, but not in the expository
texts. Unexpectedly, the mean number of words in the narratives of the research group
(109.1) is significantly higher than that of the original study (67.3). Moreover, the
65
mean number of words in the expository texts of this research group (88.64) is higher
than the one of the original group (74.6). In both studies, there is a major difference in
mean scores between junior high and high school, suggesting that the gross amount of
text producing increases sharply at this age range. Further, the mean number of words
in the narratives is higher than in the expository texts in the research groups, while the
reverse is true in the original group.
Figure 1 Comparison of mean number of words by genre and population
3.1.2 Text length in clauses
Text length was also measured by number of clauses. Table 3 gives mean
number of clauses per text across age, genre and origin. As expected, for mean
number of clauses in the narratives ANOVA yields significant effects of age
F(2,90)=52.55, p<.001. Contrast analysis between 4th grade and 11th grade yields a
significant effect F(1,90)=89.17, p <.001, and also between 7th grade and 11th grade
F(1,90)=66.87,p <.001. For mean clauses in the expository texts ANOVA also yields
a significant effect of age F(2,90)=63.72, p<.001. Contrast analysis between 4th grade
and 7th grade yields significant effect of age F(1,90)=6.75, p <.05, and between 4th
Text length: Number of words
89
109
67
75
0
20
40
60
80
100
120
Expository
Narrative
Research Crosslinguistic
66
grade and 11th grade F(1,90)=116.93, p <.001, and also between 7th grade and 11th
grade F(1,90)=67.49, p <.001. For means of clauses ANOVA with repeated measures
yields a significant effect of genre F (1,93)=40.28, p<.001. As expected, mean
number of clauses increases with age; narratives (24 in the Ethiopian group and 21 in
the non-Ethiopians) across the board have more clauses per text than expository texts
(14 in both groups). There is no significant difference between the Ethiopian and the
non-Ethiopian group with regard to number of clauses per text.
Table 3 Means and standard deviations of clauses per text, by age, genre and
origin
Ethiopian Non-Ethiopian Age N=16
Narrative Expository Narrative Expository
G SD
12.1 (6.6)
7.1 (3.7)
10.1 (6.8)
4.7 (2.9)
J
SD
13.2 (96.8)
10.6 (4.5)
17.3 (11.4)
10.9 (6.1)
H
SD
47.8 (21.9)
24.5 (11.7)
36 (16.4)
27.6 (10.6)
In comparing the original data with these data for mean number of clauses in
the narratives (in figure 2), an ANOVA yields a significant effect of population
F(1,149)=33.19, p<.001.Again, the mean number of clauses in the narratives of the
research groups (22.5) is significantly higher than the one in the original study (11.7).
In comparison, again unexpectedly, text length in the expository texts proved to be
insensitive to population. The mean number of clauses in the research study (14.4)
was not significantly higher than in the original study (12.2). Thus, the length of the
expository texts, in terms of number of clauses is similar across low SES and middle
class populations, whereas the length of the narratives is higher in low SES groups
and not as expected in higher SES groups.
67
Figure 2 Length of texts in mean number of clauses by genre and population
3.2 Syntactic Complexity Syntactic complexity was measured by number of words per clause, by
number of clauses per clause packages, and by number of types of linking devices in
each package classified into syntactic, syntactic-discursive, and discursive links (See
Section 2.6.2).
3.2.1 Syntactic complexity in terms of number of words per clause
Table 4 shows the mean number of words per clause for the independent
variables of age, genre and origin. Unexpectedly, syntactic complexity measured by
number of words per clause increases with age only in the expository texts. There is
no significant effect for age in the narratives. As expected, this measure of syntactic
complexity in the expository texts, ANOVA yields significant effect of age
F(2,90)=6.35, p<.01. For mean number of words per clause, ANOVA with repeated
measures yields significant effect of genre F(1,90)=10.95, p<.01, around 5 words per
clause in expository texts in the 5th grade, and 7 words per clause in high school.
In sum, there is no significant effect of genre in the non-Ethiopian group
regarding this measurement of syntactic complexity. As expected, mean number of
words per clause increases with age, but only in the expository texts. In the course of
Text length: Number of clauses
14.4
12.2
22.5
11.7
0
5
10
15
20
25
ExpositoryNarrative
Research Crosslinguistic
68
schooling, especially between junior high and high school (5.3 compared with 6.8),
there is some increase in children's clause construction, mainly in expository texts,
while there is almost no development in their narratives, in this respect. The Ethiopian
highschoolers show greater range of means of words per clause in their expository
texts than the non-Ethiopians. There is no significant difference between the
Ethiopian group and the non-Ethiopian group with regard to syntactic complexity
measured by number of words per clause.
Table 4 Means and standard deviations of words per clause, by age,
genre and origin
Ethiopian Non-Ethiopian Age N=16
Narrative Expository Narrative Expository G
SD
4.8 (1.1)
4.9 (1.1)
4.8 (.8)
5.5 (2)
J
SD
5.1 (1)
5.3 (.9)
5 (.9)
5.4 (1.5)
H
SD
5.3 (.75)
6.8 (3)
5.4 (.8)
6.7 (1.1)
In comparison with the original data for mean number of words per clause in
the narratives and in the expository texts, ANOVA yields a significant effect of
population F(1,149)=10.00, p<.01. As expected, the mean number of words per clause
in both the narratives and in the expository texts in the original group (5.8 and 6.3) is
higher than in the research group (5 and 5.8). Mean number of words per clause
seems to be highly diagnostic for age, genre and population. Not only does it
neutralizes the effect of text length, it also reflects increased clause-internal density
and syntactic complexity.
69
Figure 3 Comparison of mean number of words per clause by genre and population
Similar analyses were conducted with populations of low SES children, who
attend well-established schools (Salmon 2002). A comparison of means of number of
words per clause measured by the same method of word counting (as in the Salmon's
study) showed that low SES children in poor school systems averaged fewer words
per clause (Narr. 3.8, Exp. 4.2) than low SES children in good schools (Narr. 4.2,
Exp.5) across genre.
3.2.2 Syntactic complexity in terms of interclausal packaging
Syntactic complexity was also measured by number of clauses per clause
packages, and by the types of linking devices (syntactic, syntactic-discursive, and
discursive) in each package.
Syntactic complexity: Number of words per clauses
5.8
6.3
5
5.8
0
1
2
3
4
5
6
7
Expository
Narrative
Research Crosslinguistic
70
3.2.2.1 Number of clauses per clause package
Syntactic complexity is measured by number of clauses per clause package. Table
5 gives the mean number of clauses per clause package (CP) by age, genre and origin. As
expected, for means of clauses per CPs in the narratives, ANOVA yields a main effect of
age F(2,90)=29.75, p<.001.Contrast analysis between 4th grade and 7th yields a significant
effect F(1,90)=16.63, p <.001, between 4th grade and 11th grade F(1,90)=59.43, p <.001,
and also between 7th grade and11th grade F(1,90)=13.18, p <.001. For means of CPs
length, ANOVA with repeated measures yields significant effect of Genre F(1,93)=10.90,
p<.001. As expected, mean number of clauses per clause package generally increases with
age across genre. Narratives (2.1 in the Ethiopian group versus 2.2 in the non-Ethiopian
group) across the board have less clauses per CP than expository texts (2.7 in the
Ethiopian group versus 2.8 in the non-Ethiopian group).
There is no significant difference between the Ethiopian group and the non-Ethiopian
group with regard to number of clauses per clause package.
When the variable of text length is neutralized the prototypical clause
packaging in text production in both the original study and in this research lie within
the scope of two or three clauses.
Table 5 Means and standard deviations of clauses per CPs, by age, genre and origin
Ethiopian Non-Ethiopian Age N=16
Narrative Expository Narrative Expository
G SD
1.7 (.3)
2.1 (1.4)
1.5 (.3)
2.2 (.7)
J
SD
2.2 (.5)
2.4 (.7)
2.1 (.5)
3 (1.7)
H
SD
2.4 (.3)
3.7 (2.9)
3 (1)
3.2 (1)
3.2.2.2 Types of linking devices in clause packages
The types of linking devices in each package were counted and classified into
three categories: syntactic (SY), syntactic-discursive (SYD), and discursive (D).
Syntactic links included all forms of subordinators and coordinators of same-subject
71
clauses with and without ellipsis; syntactic-discursive links included coordinators of
different-subject clauses; purely discursive links were those that initiated discursively
juxtaposed clauses, and packages of a single clause in length (See Section 2.6.2.2 for
details and examples). The proportion of each of the three groups of links was
calculated out of the total number of links.
Table 6 gives types of links as a function of age, genre and origin. As
expected, syntactic links increase with age. For mean syntactic links in the narratives
ANOVA yields a significant effect for age F (2,90)=12.19, p<.001. Contrast analysis
between 4th grade and 7th grade yields a significant effect F(1,90)=14, p <.001, and
also between 4th grade and 11th grade F(1,90)=21.73, p <.001. For mean syntactic
links in the expository texts ANOVA yields a significant effect for age F (2,90)=3.73,
p<.05. Contrast analysis between 4th grade and 7th grade yields a significant effect
F(1,90)=7.41, p <.01. For mean syntactic-discursive links in the narratives there is no
significant effect for age. For mean syntactic-discursive links in the expository texts,
ANOVA yields a significant effect for age F (2,90)=3.56, p<.05. Contrast analysis
between 4th grade and 11th grade yields significant effect F(1,90)=7.08, p <.01. For
mean discursive links in the narratives, ANOVA yields a significant effect for age F
(2,90)=18.56, p<.001. Contrast analysis between 4th grade and 7th grade yields a
significant effect F(1,90)=22.78, p <.001, and also between 4th grade and 11th grade
F(1,90)=32.09, p <.001. For mean discursive links in the expository texts there is no
significant effect for age. For means of syntactic-discursive links, ANOVA with
repeated measures yields significant effect of genre F (1,93)=14.49, p<.O01. For
means of discursive links, ANOVA with repeated measures yields significant effect of
genre F(1,93)= 41.81, p<.001.
The distribution of clause links indicates that with age there is a development
in overt lexico-syntactic marking of inter-clausal connections across genre. In the
narratives, there is also a development of discourse-markers relations with age and
level of schooling indicating children's increasing abilities to employ a range of
different expressions. Syntactic-discursive and discursive links are also sensitive to
genre and occur more in the expository texts than in the narratives.
72
Table 6 Means and standard deviations of the three types of inter-clausal links
(out of the total number of links), by age, genre and origin
Ethiopian Non-Ethiopian
Narrative Expository Narrative Expository
Age N=16
SY SYD D SY SYD D SY SYD D SY SYD D G
SD .3
(.1) .02
(.06) .02 (.1)
.3 (.1)
.01 (.03)
.7 (.7)
.2 (.1)
.1 (.1)
.3 (.2)
.03 (.01)
.01 (.04)
.3 (.4)
J
SD
.4
(.1)
.07
(.07)
.1
(.09)
.3
(.1)
.02
(.04)
.7
(.7)
.4
(.1)
.06
(.06)
.1
(.1)
.4
(.1)
.03
(.07)
.4
(.4)
H SD
.4
(.08)
.05
(.04)
.1
(.08)
.4
(.1)
.03
(.03)
.6
(.4)
.5
(.1)
.07
(.04)
.1
(.08)
.4
(.1)
.05
(.03)
.8
(.6)
In comparison to the narratives of the original study the same trend is observed
with regard to syntactic links, which increased with age. On the other hand, unlike in
the original study the syntactic-discursive links in the narratives do not decrease with
age, but stay almost the same. With regard to the discursive links in the narratives,
they increase with age in this study, but stay almost the same in the original study.
3.3 Lexical Richness in terms of Lexical Density Lexical richness was measured in terms of lexical density calculated as
number of content words out of the total number of words.
The lexicon was analyzed into the lexical category of content words called Open
Class items (OC), which include nouns, verbs and adjectives (See Section 2.6.3).
Table 7 shows mean number of content words across age, genre and origin.
As expected, the number of content words increases with age. For mean number of
content words in the narratives ANOVA yields a significant effect for age
F(2,90)=55.76, p<.001. Contrast analysis between4th grade and 11th yields a
significant effect F(1,90)=91.83, p <.001, and between 7th and 11th grade
73
F(1,90)=74.55, p <.001. As expected, for mean number of content words in the
expository texts, ANOVA yields a significant effect for age F(2,)=98.88, p<.001, and
for the interaction Age x origin F(2,90)=3.23, p <.05. Contrast analysis between 4th
grade and 7th grade yields a significant effect F(1,90)=4.09, p <.05, between 4th and
7th grade F(1,90)=170.66, p <.001 and between 7th and 11th grade F(1,90)=121.88, p
<.001.
A clear developmental pattern can be observed for all types of texts. ANOVA
with repeated measures yields a significant effect for age F(2,90)=108.86, p<.001.
Contrast analysis between 4th grade and 11th grade yields a significant effect
F(1,90)=183.37, p <.001 and between 7th and 11th grade F(1,90)=140.35, p <.001,
suggesting that the gross amount of content words increases sharply with age.
For mean number of content words ANOVA with repeated measures yields a
significant effect for genre F(1,90)=17.01, p<.001, for the interaction genre x origin
F(1,90)=4.64, p <.05 and for the interaction genre x age x origin F(2,90)=6.18,
p <.01.
In the non-Ethiopian group, content words are sensitive to age and genre. In
comparison, in the Ethiopian group content words are also sensitive to age, but
unexpectedly, unlike the non-Ethiopian group, their expository texts contain fewer
content words than their narratives. There is no significant difference between the
Ethiopian and the non-Ethiopian group regarding the overall number of content
words.
Table 7 Mean number of content words and standard deviations by age,
genre and origin
Ethiopian non-Ethiopian Age N=16 Narrative Expository Narrative Expository
G SD
25.12 (13.53)
15.12 (8.66)
21.68 (15.50)
11.81 (8.60)
J SD
27 (12.79)
21.37 (8.39)
16.68 (20.75)
23.25 (12.27)
H SD
99.75 (43.41)
61.56 (24.57)
77.12 (38.95)
79.56 (29.31)
74
Like the original study, this study found that content words are sensitive to
age. Content words occur more in the expository texts. Unexpectedly, when
comparing means of content words across genre in both studies, (figure 4) it shows
that the means in this study are higher (44 and 35) than the original study (24 and 32).
Again unexpectedly, the means of the content words in the expository texts are almost
the same across populations.
Figure 4 Comparison of mean number of content words by genre and population
Counting of the content words per clause was also done to measure lexical
complexity and density. Table 8 shows the number of content words per clauses
across age, genre and origin. Unexpectedly, content words in the narratives seem to be
insensitive to age since ANOVA does not yield a significant effect.
As expected, for means of content words per clause in the expository texts
ANOVA yields a significant effect for age F(2,95)=4.45, p<.05. Contrast analysis
between 7th and 11th grade yields a significant effect F(1,90)=8.61, p<.01. As
expected, the content words in the expository texts are sensitive for age and level of
schooling. For means of content words per clause ANOVA with repeated measures
Number of content words
35
32
44
24
0
5
10
15
20
25
30
35
40
45
50
Expository
Narrative
Research Crosslinguistic
75
yields a significant effect of genre F(1,90)=16.92, p<.001 and a significant effect for
the interaction between genre x age F(2,90)=3.42, p<.05.
As expected, content words are sensitive to genre; expository texts are more
dense than the narratives. My assumption that content words per clause would
increase with age was confirmed only in the case of the expository texts.
Table 8 Mean number of content words per clauses and standard deviations by age,
genre and origin
Ethiopian non-Ethiopian Age N=16
Narrative Expository Narrative Expository
G SD
2.18 (0.62)
2.24 (1.01)
2.10 (0.43)
2.62 (1.27)
J SD
2.11 (0.45)
2.10 (0.45)
1.94 (0.36)
2.27 (0.77)
H SD
2.12 (0.30)
2.86 (1.30)
2.16 (0.44)
2.95 (0.73)
A comparison to the original study shows that, as expected, the number of
content words per clause is sensitive to genre: F(1,149)=28.01, p<.001. It is higher
both in the narratives and in the expository texts of original study ( Narr. 2.18,
Exp.2.79) than in the research study ( Narr. 2.1, Exp.2.51).
Figure 5 shows the development of number of content words per clause across genre
in the two populations
76
Num. of content words per clauses
2.51
2.79
2.12.18
0
0.5
1
1.5
2
2.5
3
Expository
Narrative
Research Crosslinguistic
Figure 5 Comparison of means of content words per clause by genre and population
Lexical density was calculated as the percentage of content words or "open
class" (OC) out of total number of words. Table 9 shows lexical density by age, genre
and origin. Unexpectedly, for means of lexical density, an ANOVA with repeated
measures yields a significant effect of genre only in the non-Ethiopian group,
F (1,90)=4.31, p<.05.
Mean lexical density in the expository texts (0.43) is higher than in the narratives
(0.4). On the other hand, the mean lexical density of the narratives of the Ethiopian
group equals that of their expository texts (0.42).
There is no significant difference between the Ethiopian and the non-Ethiopian group
regarding lexical density.
77
Unexpectedly, with regard to lexical density, there is no significant difference
between the research study and the original study. In both the non-Ethiopian group
and in the original study, lexical density is higher in the expository texts. ANOVA for
means of lexical density in each age group was also performed in order to compare
the two studies in each grade. For lexical density in the narratives in the 4th grade it
yields a significant effect for population, F (1,5 0)=4.95, p<.05. The mean for lexical
density in the research study (0.44) is higher than in the original study (0.38).
ANOVA with repeated measures yields a significant effect for genre, F (1,5 0)=4.98,
p<.05 . The mean for lexical density in the expository texts of the research study
(0.45) is higher than in the original study (0.43).
For lexical density in the narratives in the 7th grade, the same pattern is found.
ANOVA yields a significant effect for population, F (1,5 0)=9.60, p<.05. The mean
for lexical density in the research study (NM=0.40) is higher than in the original study
(NM=0.36). ANOVA with repeated measures yields a significant effect for genre,
F (1,5 0)=14.59, p<.001. A reverse pattern occurs with regard to the expository texts.
Table 9 Means and standard deviations of lexical density by age,
genre and origin
Ethiopian non-Ethiopian Narrative Expository Narrative Expository
Age N=16
OC OC OC OC G
SD
.45 (.06)
.4 (.1)
.43 (.04)
.4 (.07)
J SD
.4 (.05)
.3 (.05)
.3 (.03)
.4 (.05)
H SD
.3 (.04)
.4 (.04)
.3 (.05)
.4 (.04)
The mean for lexical density in the expository texts in the research study
(0.40) is smaller than in the original study (0.42). A significant effect for the
interaction of population x genre is also found, F (1,5 0)=10.33, p<.01.
In the 11th grade, lexical density across genre an ANOVA did not yield a significant
effect for population. On the other hand, ANOVA with repeated measures yields a
78
significant effect for genre, F (1,5 0)=15.97, p<.001. As in the 7th grade, the mean for
lexical density in the expository texts of the research study (0.42) is smaller than the
one in the original study (0.44).
In sum, although there is no significant difference between the two studies
with regard to lexical density, a closer examination shows that in the 4th grade, the
two types of texts are denser in the research study, while with age and level of
schooling the expository texts of the middle-class subjects of the original study are
denser, and reveal greater lexical density than those of the two research groups.
Figure 6 shows the development of lexical density across genre in the two
populations.
Figure 6 Comparison of means of lexical density by genre and population
3.4 Analysis of Thematic Content
Thematic content examines what children choose to write on the topic of
interpersonal conflicts (See Section 2.6.4).
Lexical density: OCs per total tokens
0.38
0.43
0.36
0.37
0.32
0.34
0.36
0.38
0.4
0.42
0.44
ExpositoryNarrative
Research Crosslinguistic
79
3.4.1 Reference to the video
The subjects were shown a wordless video clip depicting different conflict
situations in a school setting, after which they were requested to produce a narrative
and an expository text in writing. The video served as a trigger for producing texts
with a shared thematic content. This analysis was conducted on expository texts only
(for reasons see Section 2).
As expected, Ethiopian children referred more to the video scenes (in their
expository texts) than the non-Ethiopians - 39% of the Ethiopian subjects, in
comparison to 20% of the non-Ethiopian subjects. In the non-Ethiopian group, most
of the subjects who referred to the video scenes were in junior high (60%). In the
Ethiopian group the reference to the video scenes was the same in all three age groups
(33%).
3.4.2 Reference to physical aggression
The category of content words "open class" (OC) was used to examine
reference to physical aggression, like dakru ' they stabbed'. They served as a measure
of thematic content. All the OCs in the lexicons of the three groups were marked
either + or - in reference to physical aggression, in an attempt to find the proportion of
words expressing physical aggression from the total number of OCs of each group.
The Ethiopian group had 15.9% OCs marked as + aggression in their lexicon,
similarly to the non-Ethiopian group (16.4%), compared with 11.2% of the OCs
marked as such in the lexicon of the original study. Thus, the prediction that subjects
from low SES would include more expressions of physical aggression in their texts
was confirmed.
3.4.3 Reference to the location of conflicts Content words were also used to identify references to the subjects' homes as
the location of the conflicts between people. Nine of the non-Ethiopian subjects wrote
about conflicts in their homes, as compared to only two of the Ethiopian subjects,
whereas none of the subjects in the original study wrote about conflicts in his/her
family. My prediction that low SES would refer more to conflicts inside the family
was partly confirmed. It seems that the low SES non-Ethiopian subjects (most of them
80
in high-school) find it acceptable to write about conflicts in their families, while most
of the Ethiopian subjects do not.
3.4.4 Reference to affronts to honor Content words were also used for measuring references to an important norm
of the Ethiopian community - avoidance of affronts to honor. Fifty-two percent of the
Ethiopian adolescents, compared to only 10% of the non-Ethiopians, wrote personal-
experience narratives telling about affronts to one's honor: ha^ rocxim meabdim et
ha^kavod shel ha^mishpaxa shelahem 'the murderers loose the honor of their families'
3.5 Violation of Linguistic Norms
Berman’s (1999) model defines the level of non-native language proficiency
along a continuum from obligatory rules of morpho-syntactic structure (“core
grammar”) via lexical convention to discourse-motivated and culturally-determined
factors of language use. Following this model, three types of violations were
specified: expressing morpho-syntax, lexical usage, and register mixing. All
violations of linguistic expressions were marked and analyzed (See Section 2.6.5).
3.5.1 Grammatical errors
Three types of morpho-syntactic errors were defined: in grammatical
agreement marking of gender, number and person, choice of prepositions and overuse
of the generic subordination še- 'that' for example in place of the more specific form
kše- ‘as-that’ = when. All the grammatical errors were categorized and counted.
Table 11 shows the total amount of grammatical errors across age and origin.
Unexpectedly, the number of grammatical errors increases with age in the non-
Ethiopian group. The same trend occurs in the Ethiopian group, where more
grammatical errors occur in grade 7 and in grade 11 compared to the non-Ethiopian
group. All the Ethiopian 11th graders and also 75% of the 7th graders are not Israeli-
born, and so, although they are considered native speakers of Hebrew, they have not
been using Hebrew since birth, and therefore have fewer years of exposure to the
language.
In the non-Ethiopian group, grammatical errors increased between the 7th and
the 11th grade. In the 4th grade, where all the children including those of Ethiopian
81
origin are Israeli-born, the non-Ethiopian group has almost twice the number of
grammatical errors as the Ethiopian group.
Table 10 Total amount of grammatical errors by age and origin
Age N=32
Ethiopian non-Ethiopian
G 23 40 J 54 36 H 130 71
Total 207 147
Table 11 shows the three most common types of grammatical errors across age
and origin. The first type refers to agreement marking of gender, number and person
(Agr), the second to choice of prepositions (Prep), and the third to overuse of the
generic subordinator še- ‘that’, for example in place of the more specific form kše-
‘as-that’ = 'when'.
Table 11 Percentages of the three common grammatical errors by age and origin
Ethiopian Non-Ethiopian Age N=32 Agr Prep še Agr Prep še
G
8 26 26 38 15 25
J
52 22 11 33 27 9
H
40 15 8 46 28 7
With regard to grammatical errors, the most frequent are in verb number,
gender and person agreement in both research groups. In comparison, in Salmon’s
(2002) study the only significant deviation was in overuse of the generic subordinator
še, which is similar in all three populations. The Ethiopian subjects err more in choice
of prepositions, while the non-Ethiopians err more in the marking of agreement.
A comparison of means of grammatical errors per clause between the two populations
showed no difference (an average of 0.1). Thus, it seems that low SES children from
82
good schools show similar grammatical accuracy to low SES children from poor
educational systems.
The first two types of grammatical errors in the research study also point at an
interaction between syntax and lexicon, since they can be attributed to either or both
the grammatical and lexical levels of language structure.
The heads of the language teams in the schools report that the commoner type of
grammatical error in teachers' speech is in marking of agreement.
3.5.2 Lexical infelicities
These types of violations of lexical expressions refer to deviations in the
lexicon in terms of precision and felicity of vocabulary selection. Lexical infelicities
refer to deviations from collocational appropriateness (single words or expressions) in
a given clause, which strike the reader-hearer as deviating from accepted or
conventionalized native usage. For example: še-ota yalda adayin pgu’a gam mimeni
lamrot ha-slixot ‘that-same girl (is) still offended by me too, in spite of the
pardonings' where the last word is a misuse of the term for religious penitential
prayers said on a Jewish holiday, in place of conventional hitnacluyot ‘apologies’ or
the related, more colloquial and yet more appropriate še-bikashti mimena slixa ‘I
asked from her (a) pardon = I said I was sorry'.
All lexical infelicities were marked, counted, and categorized into six sub-
categories: (1) use of a single inappropriate word, (2) inappropriate free collocations,
(3) inappropriate restricted collocations, (4) appropriate collocations in an inadequate
context, (5) over-collocation - use of a collocation where only a part of it is needed
and (6) under-collocation - use of only a part of the appropriate collocation. All
lexical infelicities were further marked only for high literate register, in an attempt to
describe the context in which such deviations, are produced.
Table 12 shows the number of lexical infelicities across age, genre and origin.
As expected, a clear developmental pattern occurs with regard to lexical infelicities.
Moreover, it seems that this phenomenon does not occur in the early years of
schooling, but rather is typical of high-school students. Lexical infelicities seem to be
sensitive to genre, since the non-Ethiopian group has about three times more lexical
infelicities in their expository texts than in their narratives. A lesser but similar pattern
is found in the Ethiopian group.
83
The lexical infelicities marked as high literate register (see Section 2) were
higher in the expository texts of the Ethiopian group (52%), while only 42% of them
were marked as such in the narratives. An opposite trend is found in the non-
Ethiopian group, where 78% of the lexical infelicities marked as high register
occurred in the narratives, while only 32% in the expository texts.
Table 12 Number of lexical infelicities by age, genre and origin
Ethiopian Non-Ethiopian Age N=16 Narrative Expository Narrative Expository
G
_ _ _
_
J
3 _ 1 3
H
16 21 14 37
Total = 94 19 21 15 40
In order to construct a comparable sample of relative measure of complexity,
16 texts of the original study were matched to those of this study by mean number of
words per clauses. Table 13 shows the number of lexical infelicities in the original
study across age and genre. As expected, a comparison of lexical infelicities with the
original study shows that this phenomenon is far more pervasive (94 compared with
14) among low SES subjects. Two subjects of the original study in the 4th grade
produced lexical infelicities in their expository texts, in comparison to none of the low
SES subjects. These infelicities increase with age and with the writer’s level of
literacy and proficiency across populations, and occur mostly in the 11th grade. Half
of the lexical infelicities in the expository texts of the original study, and a third in the
research study seem to accompany a deliberate attempt of the writers to raise the
register.
84
Table 13 Number of lexical infelicities in the original study by age and genre
Original study Age Narrative Expository
G N=16
_ 2
J N=16
1 2
H N=16
5 4
Total= 14 6 8
Table 14 shows the sub-categorizations of the lexical infelicities in the research study
and in the original study (See Section 2.6.5.2). Most of the lexical infelicities in both
studies occurred in restricted collocations, which can be characterized by the high
literate register.
Table 14 Number of the sub-categorizations of the lexical deviations by population
Population Single
inappropriate word
Inappropriate free collocation
Inappropriate restricted collocation
Appropriate collocation inadequate context
0ver- collocation
Under- collocation
Total
Research study
26 21 26 11 3 1 94
Original study
2 4 8 14
3.5.3 Register mixing
Register mixing refers to usage of colloquial everyday language with high
style expressions (single words or multilexemic expressions) in a given clause. All
such deviations were marked and counted. Table 15 shows the number of register
mixing across age, genre and origin. Unexpectedly, the total number of register
mixing in the non-Ethiopian group is equal across genre. In comparison, the Ethiopian
85
subjects have more register mixing in their narratives. As expected, register mixing is
more common among the Ethiopian subjects, who have more such deviations than the
non-Ethiopians.
Table 15 Number of cases of register mixing by age, genre and origin
Ethiopian Non-Ethiopian Age N=16 Narrative Expository Narrative Expository
G
13 10 7 4
J
13 15 14 11
H
14 10 4 10
Total = 125 40 35 25 25
Table 16 shows the number of register mixings in the original study across age
and genre. As expected, register mixing seems to be sensitive to population. The
original group has less of such register mixing (M=7) than the research groups
(M=10.2). It seems that the original group is more aware of register appropriateness
and consistency across genre, since they have fewer such deviations in their
expository texts (2) and fewer in their narratives ( 0.27), compared to the research
groups (Exp 9.5, and Narr. 0.67) respectively.
Table 16 Number of cases of register mixing in the original study by age and genre
Original study Age N=16 Narrative Expository
G 3 2 J 1 1 H 9 5
Total=21 13 8
In sum, the scope of lexical infelicities and register mixing is rather small,
especially in the original data. Nevertheless, in terms of amount of deviations, both
groups have 1.5 times more deviations in register mixing than in lexical infelicities
across genre. In terms of deviance from native norms, low SES subjects have 3 times
86
more lexical infelicities and register mixing than the subjects of the original study. It
seems that inconsistencies of register are more frequent among low SES subjects,
while middle-high SES subjects appear to have better mastered the task of precision
and felicity of lexical selection. In terms of register, the production of lexical
infelicities seems to be accompanied by a deliberate attempt to raise the level of
language use. Moreover, the native speaker-writer (middle-high SES, and of
Ethiopian origin) seems to produce more register mixing when writing a personal
experience narrative. The original subjects seem to have better abilities to produce
well-structured texts in writing, with conventionalized lexical items, and also better
native speaker-writer socio-communicative skills, which ensure the appropriate level
and type of language usage.
3.6 Global Text Construction
A “top down” approach was used to examine genre, origin and population
differences and also developmental changes in the production of global text structure
in terms of overall discourse structure, text component length and organization of
information in the text (See Section 2.6.6).
3.6.1 Text Components
The analysis of global text structure uses three obligatory analogous text
components for both text types. With regard to the narrative about interpersonal
conflicts, the setting contains information about the place, time, protagonists (the
characters in a state of conflict), state of affairs, motivation for story telling and
elements of the story script. The middle component of the narrative (the events or
plot, episodes) contains at least two chronologically ordered events, while the closing
component (resolution) contains a resolution to the problem and sometimes also a
coda that refers back to the beginning and to the motivation for story telling.
In comparison, in expository texts the opening (the introduction) contains
information about the main idea(s) to be discussed. The middle contains
supplementary elaborations and illustrations to the ideational core, while the closing
87
component (conclusion) is characterized by conclusions, summaries and occasionally
refers back to the main idea(s)
Table 17 shows the global text structure in three components: openings (OP),
text center (TC) and closings (Cl) by age, genre and origin. For the openings,
ANOVA with repeated measures yields significant effects for genre F(1, 90)=25.058,
p< .001, for age F(2,90)=22.807, p<.001, and for the interaction between genre x age
F(2, 90)=9.561, p< .001, genre x origin F(1, 90)=4.451, p<.05, and genre x age x
origin F(2, 90)=3.712, p< .05. For the core propositions ANOVA with repeated
measures yields significant effects for genre F(1,90)= 32.946, p<.001, for age
F(2,90)=21.655, p< .001, for the interaction between genre x age F(2,90)=4.092,
p<.05 and for the interaction between genre x age x origin F(2,90)=7.450, p=.001.
For the closings ANOVA with repeated measures yields significant effects for genre
F(1,90)=12.049, p=.001 and age F(2,90)=32.783, p< .001.
As expected, the ability to construct narratives and expository texts including
all three text components increases with age. Moreover, the global text structure for
the narrative is more accessible to younger children than for the expository text,
which seems to be well-constructed only in high school. Openings are constructed at
an earlier stage of development than closings; openings and closings emerge earlier in
narratives than in expository texts. Most of the subjects in junior high produce
openings in their narratives, while about half of them still find it difficult to construct
openings in their expository texts. It seems to be a difficulty in constructing closings
across genre even among the high-school students. The non-Ethiopian subjects begin
to produce closings across genre earlier and construct more of them across age in
comparison to their Ethiopian peers. Thus, genre seems to be a relevant factor in
relation to age differences in global text construction. There is no significant
difference between the Ethiopian group and the non-Ethiopian group with regard to
global text structure.
88
Table 17 Numbers of texts containing each of three text components
by age, genre and origin
Ethiopian non-Ethiopian Narrative Expository Narrative Expository
Age N=16
OP CN CL OP CN CL OP CN CL OP CN CL G 14
16
8
9
16
0 16
16
7
6
16
0
J 16
16
7
9
16
2
15
16
14
7
16
7
H
Total
16
46
16
48
13
28
15
23
16
48
13
15
16
47
16
48
12
33
16
19
16
48
13
20
3.6.2 Text component length
Text component length was measured in terms of number of clause packages
per text component. Table 18 shows mean number of CPs in each text component.
(opening, core proposition and closing) by age, genre and origin. As expected, for the
openings in the narrative texts a t test shows a significant effect for age
F(2,93)=16.938, p<.001. For the openings in the expository texts a t test yields a
significant effect for age F(2,93)=12.04, p<.001. For the core propositions in the
narrative texts a t test shows a significant effect for age (2,93)=13.032, p<.001. For
the core propositions in the expository texts a t test yields a significant effect for age
F(2,93)=12.04, p<.001.For the closings in the narrative texts a t test shows a
significant effect for age F(2,93)=11.160, p<.001. For the closings in the expository
texts a t test yields a significant effect for age F(2,93)=35.350, p<.001. As expected,
text component length, measured by number of CPs per text component, is sensitive
to age across genre.
89
Table 18 Means and standard deviations of CPs in each text component
by age, genre and origin
Ethiopian non-Ethiopian Narrative Expository Narrative Expository
Age N=16
OP CN CL OP CN CL OP CN CL OP CN CL G
SD 1.3 (1)
5.1 (2.6)
.6 (.8)
.6 (.6)
3 (1.6)
.06 (.25)
1 (.6)
4.1 (3.1)
.6 (.8)
.5. (.5)
1.8 (1)
0
J SD
1.2 (.5)
3.8 (2.5)
.5 (.6)
.8 (1)
3.6 (1.6)
.1 (.5)
1 (.3)
5.8 (4)
1.1 (.8)
.6 (1)
3.3 (2.1)
.5 (.7)
H
SD 5.6 (3)
12.3 (7.6)
2.1 (1.6)
2.1 (1.6)
4.7 (3)
1.7 (1.1)
3 (2.3)
7.2 (3.8)
1.8 (2.1)
1.8 (2.1)
5.8 (3.4)
1.3 (1.1)
On the whole, there is no significant difference between the Ethiopian group and
the non-Ethiopian group with regard to text component length. Nevertheless, a
comparison of the narratives of the two groups (especially in the 11th grade) shows
that the Ethiopian subjects have more CPs per text component, especially in their
openings and core propositions. Consequently, their narratives intuitively seem to be
denser.
Text construction has a clear developmental pattern. Global text constructions
are learned and constructed throughout the school years especially between junior-
high and high school. Canonical narrative structure is acquired earlier than global
expository structure. Construction of well formed openings and core propositions in
the narratives, which contain more CPs than those of the expository text, occur during
the early years of schooling. Closings seem to be the most difficult and the last to be
constructed across genre. Even during the final years of schooling some subjects have
difficulties in constructing well formed closings in their expository texts.
3.6.3 Levels of organization of information
Text structure was also examined in terms of a more qualitative analysis of
three levels of information organization: the linear (L), partially hierarchical (PH),
and fully hierarchical (FH) levels based on Katzenberger's model (submitted a). At the
linear level adjunct clauses or pieces of information are organized in a pair-wise
fashion; at the partially hierarchical level the text contains both global and local-level
90
a
a
elements of content and/or structure; and at the fully proficient level there is a graded
flow of information between general and specific, and also global text organization
and flow of information explicitly are marked by linguistic means. These levels were
examined in both types of texts.
Table 19 shows global text structure in terms of three levels of information
organization: linear (L), partially hierarchical (PH), and fully hierarchical (FH) by
Age, Genre and Origin. As expected, for the linear level, ANOVA with repeated
measures yields significant effects of genre F(1,90)=12.273, p=.001, of age
F(2,93)=13.444, p<.001 and for the interaction between genre x age F(2,90)=12.273,
p<.001.
For the partially hierarchical level, ANOVA with repeated measures yields significant
effects of genre F(1,90)= 735.423, p<.001, of age F(2,93)=6.007, p<.01 and for the
interaction between genre x age F(2,90)=224.577, p<.001.
For the fully hierarchical level, ANOVA with repeated measures yields a significant
effect of genre F(1,90)=12.273, p=.001, of age F(2,93)=13.444, p<.01 and for the
interaction between genre x age F(2,90)=12.273, p<.001. As expected, the levels of
information organization are sensitive to age and genre. The ability to construct a
partially hierarchical ordered expository text, which contains both global and local-
level elements, develops between junior high and high school. Almost none of the 7th
graders construct even a partially hierarchical expository text (except two Ethiopian
students and one non-Ethiopian). Moreover, almost none of the subjects in this study
constructed a fully hierarchical and explicitly marked expository text (except one non-
Ethiopian high-school student.). In comparison, a third of the population in high
school constructed fully hierarchical narratives; and the non-Ethiopian group
constructed more such texts than the Ethiopian group. All 4th graders constructed
partially hierarchical narratives. There is no significant difference between the
Ethiopian and the non-Ethiopian group with regard to levels of information
organization.
91
Table 19 Numbers of texts including Linear (L), Partially Hierarchical (PH),
and Fully Hierarchical (FH) levels of information organization by age, genre, and
origin
Ethiopian non-Ethiopian Narrative Expository Narrative Expository
Age N=16
L PH FH L PH FH L PH FH L PH FH G
- 16
- 16
- - - 16
- 16
- -
J
- 16
- 14
2
- - 16
- 15
1
-
H Total
- -
13 45
3 3
3 33
13 15
- -
- -
9 32
7 7
4 35
11 12
1 1
The openings of the original texts across genre were also examined by the
same method as above and by the same judges in terms of the three levels of
information organization: the linear (L), partially hierarchical (PH), and fully
hierarchical (FH) levels. Table 20 shows global text construction of the original data
in terms of three levels of information organization by genre and age. With regard to
population, there is no significant difference between the original subjects and both of
the two research groups with regard to global text structure. All fourth graders
constructed partially hierarchical narratives, while only about a third of the high-
school children constructed fully hierarchical narratives. Only two high-school
subjects constructed a fully proficient rhetoric expository text.
Table 20 Numbers of texts of the original data including Linear (L), Partially
Hierarchical (PH), and Fully Hierarchical (FH) levels of information organization by
genre and age
Narrative Expository Age N=20
L PH FH L PH FH
G
-
-
-
20
-
-
J
- 20
- 17
3
-
H
Total
- 13
33
7 7
-
37
18
21
2 2
In sum, all populations show similar production abilities of text structure across age
and genre.
92
3.7 Summary of Results of the Written Texts
All the above findings suggest that the measures of text construction applied
here: text length, clause packaging and syntactical links are sensitive to age. Thus, the
results of this study confirm the results of the original study with regard to
development. For text length, there is a clear developmental pattern for both types of
text. Text increases in length with age and level of schooling as measured by number
of words, and number of clauses per text. Development of text length is also sensitive
to genre. Narratives are generally longer than texts in the expository genre in number
of words and of clauses. Syntactic complexity measured by number of words per
clause and number of clauses per clause package also increases with age. With regard
to genre, a reverse pattern emerges for mean length of clause and for mean length of
clause packages compared with mean length of texts: Expository texts have more
words per clause, and more clauses per clause package.
On the one hand, origin is not significant regarding text length, syntactic
complexity, lexical richness, and global text construction. Both research groups
exhibit similar behavior regarding these measures. On the other hand, regarding
grammatical errors, the Ethiopian subjects produce more such errors than their non-
Ethiopian classmates
With regard to population, low SES students use a lower number of words per
clause and fewer content words per clause across genre, when text size is neutralized.
They also display less proficient usage in lexical selection and in discourse
conventions than the middle-class children of the original study. Consistencies of
register seem to be especially difficult to access for low SES students.
Similar results between students from different socio-economic backgrounds
were found with regard to clause packaging, lexical density, and global text
construction. In addition, no differences were found in terms of grammatical accuracy
between low SES children from poor educational systems and low SES children who
attend well-established schools.
Thematic content also emerges as sensitive to population. Students seem to
include in their personal-experience narratives themes that refer to their immediate
surrounding and to common norms and behaviors of their community.
93
Part 2: Results of Literacy Questionnaire
3.8 Introduction As noted, the second database consists of 114 questionnaires (96 of the two
low SES research groups and 18 questionnaires of the middle-high SES control
group). This “literacy questionnaire” was designed specifically for this study,
extended and elaborated from a preliminary design used in the original study (see
Appendix I). The original design, which served as the basis of the literacy
questionnaire, included only open-ended questions. The literacy questionnaire used in
this study consists of 24 items (both open-ended and multiple-choice). It was
expanded considerably to meet the particular goals of the present study, and to
provide an informative profile of the research population. The main goal of the
questionnaire was to elicit responses that would shed light on literacy-related
activities at home and in school, and on attitudes towards language of children from
different backgrounds. All these variables may affect language proficiency.
This individually tailored background and literacy related questionnaire is a
special type of discourse, which forms an important part of modern bureaucratic life.
Immigrant children are familiar with this type of text since they often function as go-
betweens for their parents in contact with bureaucratic institutions, so they have some
experience with filling out questionnaires.
The questionnaire was administered to each student individually. In addition to
the two research groups, a third group of 18 schoolchildren was also asked to fill out
the literacy questionnaire. These were Israeli-born children of high SES background
in the same age groups and school levels as the research groups. All the children in
this control group spoke Hebrew at home. They were all high achievers in school
language studies. The parents of the children in the control group were highly literate
in Hebrew, with an average of fifteen years of formal schooling.
An adult interviewer (investigator/facilitator) was present throughout the
proceedings. Once subjects had completed the task, the adult interviewer fulfilled a
short questionnaire for each subject about his/her conduct in the course of performing
the task (Section 3.16). Information provided by students in response to the
questionnaire was supplemented by two additional sources of information: 1) an
evaluation sheet filled out by class language teachers for each student, and 2) data
94
entered on students’ personal files by their parents when they enrolled in the school.
These tools were implemented in order to further verify and check the data from the
literacy questionnaire.
The data derived by these means (student questionnaire, supplementary details
provided by interviewer, teacher evaluation sheets, and school files) were analyzed
according to criteria relating to the two main independent variables of the study: age
and origin. In addition, where relevant, teacher evaluations were compared with
student self-perceptions in each domain.
A qualitative descriptive analysis was chosen to gain a ‘holistic’ overview of
the studied phenomena. This analysis was used in order to obtain an insight into the
students’ particular attitudes towards writing, their perceptions of home literacy
activities and their day-to-day writing behaviors in school. This kind of data, about
the students’ general, linguistic and familial background, is qualitative in nature and
cannot be quantified nor measured. Moreover, it represents how some actors in some
contexts deal with some issues (Miles & Huberman, 1994, p. 22). Most analyses of
such phenomena are done using words rather than numbers (pp. 6-7). The
questionnaire deliberately contained a relatively large number of open-ended
questions. The analysis, therefore, is essentially qualitative rather than statistical in
nature. This is in spite of the fact that the analysis sometimes refers to quantitative
distributional data. Each domain is evaluated in terms most relevant to the general
purpose of the study - the impact of literacy and literacy-based experiences on
students’ perceptions and attitudes. Each of the nine following topics representing the
variables investigated by means of the literacy questionnaire is discussed in a separate
section:
1) Level of success in Hebrew language studies is assessed by teacher evaluations of
language achievements (Section 3.9).
2) Proficiency in writing is assessed by both teacher and student evaluations of
narrative, expository and questionnaire writing (Section 3.10).
3) Attitudes to writing activities are assessed by the students’ evaluations of their
command and liking of a specific list of activities, and by their ranking of the
difficulty of these activities (Section 3.11).
4) Level of difficulty of the various writing activities is evaluated by the students'
choice of the easiest and the most difficult writing activities out of a list of ten
(Section 3.12).
95
5) Level of proficiency in Amharic is evaluated by the students’ evaluations of their
speaking, reading and writing of the language (Section 3.13).
6) Self-perception of what characterizes the proficient writer is evaluated by context
analysis of the students’ responses into eight categories (Section 3.14).
7) Attitudes towards Hebrew and the children’s home language were evaluated by
students’ acceptance or rejection of statements describing this issue (Section 3.15).
8) Interviewer information on the questionnaire as a task was assessed by the
interviewer’s general impression, time required to fill out the questionnaire, student’s
need for help and interviewer’s evaluation of task difficulty (Section 3.16).
9) Parental background was assessed by parents’ reports on their years of formal
education and country of birth (Section 3.17).
10) Home literacy was assessed by parental command of reading and writing in their
native language and in Hebrew, home-based literacy activities, and sources of
assistance while encountering difficulties in writing (Section 3.18).
A summary of the results of the literacy questionnaires ends the second part of the
chapter describing the results (Section 3.19).
3.9 Success in Hebrew Language Studies
Success in Hebrew language studies was assessed by teacher evaluations of
language achievements.
3.9.1 Teacher evaluations of language achievements
Language teachers were asked to give each student an average grade on his/her
level of achievement in Hebrew language studies, including composition and
literature together. Grades were ranked on a scale from 0 to 100, as is accepted
practice in the Israeli school system, as follows: 50 (fail), 60 (fair), 70 (average), 80
(good), 90 (very good), 100 (excellent). This is roughly equivalent to the following in
the U.S. school system: 50 = F, 60 = D, 70 = C, 80 = B, 90 = A.
Teachers assigned very similar grades for Hebrew language studies to
Ethiopian and non-Ethiopian students across the three grades (grade school - 4th
grade, junior-high - 7th grade, high-school - 11th grade). All groups rated a score of
around “average” to “good”. This was particularly marked in the youngest and oldest
groups. The mean scores in the 4th grade (73 for Ethiopian students and 70 for non-
Ethiopians), and in the 7th grade (71 for Ethiopian students and 70 for non-Ethiopians)
96
are almost identical. A somewhat greater difference is found in the 11th grade (75 for
the Ethiopians and 83 for the non-Ethiopians). Moreover, the wide range of scores in
both groups is very similar: from 55-95 in the Ethiopian group and from 45-95 in the
non-Ethiopian. That is, both groups showed similar intra-group variability or
individual variation. Their language teachers typically rated both groups of students as
around what is considered “average” or somewhat better than “passable” in the Israeli
school system.
3.10 Writing Proficiency
Teachers and students were asked to rank students’ proficiency in writing,
evaluated in terms of the ability to write narrative and expository texts and to fill out a
questionnaire. This was assessed by a four-point scale where low is described as “find
it very hard”, through “find it hard”, “do well”, and high described as “do very well”.
3.10.1 Teacher evaluations of narrative writing
This section describes the evaluations of the language teachers of the narrative
writing of the Ethiopian and non-Ethiopian students. The teachers of the 4th graders
identically evaluated both groups around “average” between “find it very hard” to “do
well”. The teachers of the7th graders rated the Ethiopian students between “find it
very hard” to “do well”, in comparison to the non-Ethiopians were rated as “do well”.
The teachers of the 11th graders identically evaluated both groups around “average”
between “find it very hard” to “do well”.
3.10.2 Teacher evaluations of expository text writing
The evaluations given by the language teachers are typically lower than for
narratives, and they are the same across the two groups: 4th grade - both groups rated
the same around “find it very hard”. Seventh and 11th grades - between “find it very
hard” to “do well”. The findings show that expository ratings improve with age.
3.10.3 Teacher evaluations of questionnaire writing The language teachers also evaluated the ability of their students to fill out
questionnaires. The results show that scores are typically higher than for both
narratives and expositories, and again they are the same for both Ethiopian and non-
97
Ethiopian in the two lower grades: 4th and 7th grades - both groups rated the same
around “average” between “find it very hard” to “do well”. 11th grade - the Ethiopian
students rated “do well” - slightly higher than the non-Ethiopian students who rated
between “find it very hard” to “do well”.
In sum, with regard to general language achievement as evaluated both by the
students themselves and by their teachers, there is little difference between the level
of writing proficiency that teachers evaluate their Ethiopian and non-Ethiopian
students respectively as evaluated by their teachers. Again, across the three grades and
all three types of writing activities, both groups rated as by and large average, neither
group being close to failing nor close to excellent.
3.10.4 Student evaluation of narrative writing
The students were asked to evaluate their own writing proficiency in narrative,
expository, and questionnaire type texts. The students rated their writing proficiency
on the same scale as was used by their teachers, a four-point scale going from a low
described as “find it very hard”, to “find it hard”, “do well”, “do very well”.
Students consistently rated their narrative writing abilities at a higher level than did
their teachers. Most students rated themselves between “doing well” to “doing very
well”, with non-Ethiopians ranking themselves better on the average than Ethiopian
students in the two lower grades, but with Ethiopian high-schoolers ranking
themselves better than their non-Ethiopian peers.
3.10.5 Student evaluation of expository writing
The students were also asked to assess their ability to write expository texts on
the same scale used for assessing their narrative writing. Ethiopian students in all
three age groups assessed their abilities slightly lower than the non-Ethiopians. Both
groups rated themselves as “doing well” in all age groups. Here too, students
consistently rated their expository writing abilities at a higher level than their
teachers.
3.10.6 Student evaluation of questionnaire writing
With regard to filling out questionnaires, the Ethiopian and the non-Ethiopian
students in all three age groups similarly evaluated their ability; the Ethiopian and the
non-Ethiopian 4th graders rated themselves around “doing well”. Both groups of 7th
98
and 11th graders rated themselves as “doing very well”. Again, students consistently
rated their questionnaire writing abilities as higher than did their teachers.
On the whole, across the three types of writing tasks, the two groups of students
accredited themselves with relatively similar scores between “do well” and “do very
well”, and they typically evaluated themselves higher than their teachers.
Students were also asked to estimate their general writing skills in Hebrew.
This was assessed by a four-point scale going from a low described as “fair” (literally,
in colloquial Hebrew ‘lo kol kach tov’ = ‘not so good’) through to “good”, and on to
“excellent”. With regard to general writing skills in Hebrew, both Ethiopian and non-
Ethiopian students across the three age groups rated themselves quite similarly.
Approximately half of the entire population (24 out of a total of 48 Ethiopian students
and 24 out of the 48 non-Ethiopian students) rated their Hebrew writing as
“excellent”. A third of the Ethiopian students (17 out of 48) and a quarter of the non-
Ethiopians (12 out of 48) reported that they write Hebrew well. In comparison, few
Ethiopian students (7 out of 48) ranked their Hebrew writing as “fair” or “not very
good” compared with rather more non-Ethiopians (12 out of 48).
In all age levels, both groups rated themselves quite similarly. Over half of the
students in the 4th grade in both groups (11 Ethiopian and 9 non-Ethiopian out of 16
each) reported that they write Hebrew well to excellently. The bulk of the 7th graders
in both groups (14 Ethiopian and 13 non-Ethiopians out of 16 each) reported the
same. All the Ethiopian students and nearly all of the non-Ethiopians (14) in the 11th
grade also ranked their writing similarly, as “good” to “excellent”.
On the whole, across the ages both groups evaluated their writing very
similarly. High-school students expressed a better estimation of their writing skills,
and most of them regard themselves as “good” to “excellent” writers.
3.11 Writing Abilities and Attitudes
Students reported on their liking and command of various writing activities.
These activities could be done either in school or at home. They were also asked to
choose the easiest and the most difficult activity from the same list of writing
activities.
99
3.11.1 Attitudes to and command of writing activities
Students were asked to express their attitude towards various writing activities.
The writing activities listed in the questionnaire included: 1) copying from the
blackboard, 2) writing answers to questions, 3) writing a story, 4) writing a summary,
5) writing an expository text, 6) filling out a questionnaire, 7) writing a letter, 8)
writing a poem, 9) writing entries in a diary, and 10) writing personal notes and
messages. Students were asked to indicate whether they like a given activity “very
much”, “quite a lot”, “a fair amount” or “not at all”. These attitudes were then
compared to the same student’s evaluation of his/her command of, or facility with
each of the above writing activities. The students were asked to indicate their
command of these writing activities using the same scale.
A comparison of the students’ reports on their attitudes and command of the
various writing activities shows very considerable variability. A Spearman’s
correlation test was conducted only for activities corresponding to the same types of
text the subjects produced: writing a story, writing an expository text and filling out a
questionnaire. The results show no correlation between how students describe their
attitudes to these three types of writing activities and their actual command of each
activity. Thus, the students do not necessarily like the same writing activities they
perform best.
3.12 Level of Difficulty of the Various Writing Activities Students were asked to mark one activity out of the following as the easiest
and one as the most difficult respectively: 1) copying from the blackboard, 2) writing
answers to questions, 3) writing a story, 4) writing a summary, 5) writing an
expository text, 6) filling out a questionnaire, 7) writing a letter, 8) writing a poem, 9)
writing entries in a diary and 10) writing personal notes and messages. Across the
population, the task that was rated easiest by most students was that of “copying from
the blackboard”: one third of the entire research population (19 out of 48 Ethiopian
students, 13 out of 48 non-Ethiopians, and 6 out of 18 control students alike). Most
students rated “Writing poetry” as the most difficult writing activity across the
different groups; (29 Ethiopian students and 25 non-Ethiopian students out of 48
each), and one third of the control group (5 students out of 18). In other words, all
three groups of students have similar attitudes to what kind of writing activities they
view as relatively more or less difficult.
100
3.13 Proficiency in Amharic of the Ethiopian students
Most of the Ethiopian students (two-thirds) reported that they spoke Amharic
with their parents. Students were also asked to rate how good their knowledge was of
Amharic in speaking, reading, and writing. They were asked to rank it on a scale of
“excellent” to “very good”, “good”, “mediocre”, and “not at all”. Very few of the
Ethiopian students (only 5 out of the total 48) reported that they do not speak Amharic
at all. Moreover, one-quarter (12 out of 48) reported that they speak good to excellent
Amharic (with relatively more of the 4th graders and 11th graders giving themselves
this high evaluation than the 7th graders).
In contrast to the high rating they generally gave to their oral proficiency, the
bulk of the students (40 out of 48) rated their reading and writing in Amharic much
lower, between “none at all” to “mediocre”. This is consistent with a general finding
that two thirds of the Ethiopian population in Israel neither read nor write Amharic
(personal communication with Dr. Anbessa Teferra, a linguist expert in Amharic
languages).
3.14 Attitudes to Writing The questionnaire included items designed to elicit the students’ view of why
writing is important, and what they consider to be “good writing”.
3.14.1 Why writing is important
Students were asked to express their view on the importance of writing, by
writing a response to the following question: “Do you think it is important to learn
how to write? Yes/No. Explain why.” Their responses were analyzed according to
the following categories:
1) Cognitive enrichment: For example, “writing can help you discover things about
yourself”, “you cannot think without it”, “you learn things through it”.
2) Self-expression: “you can pour out things which are in your heart”, “writing is a
means of expression”, “to express feelings”.
3) Instrumental or practical motivations: “to write letters to the bank”, “in order to fill
out a work application form”, “so that when you’re a grown-up and your children ask
you to write something, you won’t have to say ‘I don’t know how’ ”.
101
4) Vague generalities: “necessary in daily life”, “it is useful”, “you would miss out on
things (if you don’t know how to write)”.
The third category, instrumental motivations stating that writing is a means of
achieving certain practical ends, was the most highly favored by the three groups.
More subjects did so among the research groups than the Control group; 29 Ethiopian
students out of 48 and 27 non-Ethiopians out of 48, compared with less than half of
the control group 8 out of 18. In the non-Ethiopian group and the control group there
is an age-related shift from a clear preference for instrumental reasons at the younger
ages to other categories in the highest age group (in 4th grade 12 out of 16 non-
Ethiopian students and 4 out of 6 control students gave an instrumental reason, while
only 4 non-Ethiopians and 2 control students did so in 11th grade). Cognitive
enrichment responses rate three times more in the control group (one-third of their
responses) than in the two research groups (16% and 12% of the Ethiopians and non-
Ethiopians respectively). Reasons of “cognitive enrichment” are not given at all by
4th graders in the control group, while 5 Ethiopian 4th graders and 3 non- Ethiopians
out of 16 in each age group chose this category. Self-expression rated the lowest
proportion of response in all three groups, and was not used by a single one of the
research group students in the two lower grades. Vaguely general statements
accounted for around 10% of the Ethiopian students’ responses and 30% of the non-
Ethiopians’ responses and the controls’. Only the non-Ethiopian students show an
age-related shift from nine vague general statements in 4th grade to only four such
statements in 7th grade. The other populations have an even number of such
statements in all the age groups.
In sum, in attitudes regarding the importance of writing, instrumental reasons
are highly preferred across the population, especially in the younger age groups. Self-
expression ranks higher at the junior-high group. In comparison, cognitive enrichment
rates mainly among the control group students.
3.14.2 What makes a good writer?
The students were asked to define what they considered to make a proficient
writer, as follows: “Think about a person who writes well. What is good in this
person’s writing? What is special about this person’s writing?”. Responses were
divided into the following categories:
1) Communicativeness: “writes for the others”, “enters into the spirit of the reader”.
102
2) Writer-oriented activity: “ [writes] for himself”, “enjoys the act of writing”.
3) Content: “easily understood, you do not need to devote much thought to what the
writer wanted to say”, “writes about interesting topics”.
4) Use of language: “rich vocabulary”, “literary expressions”.
5) Organizational structure: “the writing is well organized”, “the writing has a
beginning, a middle, and an end”.
6) Command of writing conventions: “without spelling mistakes”.
7) Has good/clear handwriting: “has a nice handwriting”.
8) Page layout (paragraphing, line spacing, etc.): “keeps to the lines”, “makes even
spaces between the lines”.
Command of writing conventions in general and “has a good/clear
handwriting” in particular were the favored responses of around half of all the
students in the two research groups (23 Ethiopians, 25 non-Ethiopians out of 48 in
each). In contrast, only a quarter of the control group students (4 out of 18) chose this
category. Again, this changes with age in the control group, where only 4th graders
mention “handwriting” as a factor in their responses. Other aspects of writing
conventions, mainly spelling, account altogether for most of the responses of the 4th
and 7th grade research groups, but of almost none of the control students. Very few
students; only two Ethiopian students, three non-Ethiopians, and one 4th grader in the
control group chose page layout. Language use is mentioned as a major property of
good writing by as high as one third and nearly 20% of the high-school Ethiopian and
non-Ethiopian students respectively. In contrast, only several control group students at
junior as well as high school age gave this response. Only high-schoolers mentioned
"content" as a characteristic of good writers. More control group students (5 out of
18) chose this category in comparison to the Ethiopian students (6 out of 48) and their
non-Ethiopian peers (7 out of 48). Communicativeness is also selected far more by
control students (8 out of 18) than by the other two groups (4 out of 48). Again, the
control students start referring to this category by junior-high school. In comparison,
only high-school students in the two research groups also chose this category. Only a
few high-school students across the three groups referred to text organization.
In sum, high-school students across the population, but most particularly in the
control group, specify “features of writing as a special discourse style” as
characteristics of the proficient writer. Such features are: communicativeness, content
and language use. Whereas features of “writing as a notational system” (Ravid &
103
Tolchinsky, 2002) are typical of evaluations of younger grade-school children. Such
features are mainly handwriting and command of writing conventions.
3.15 Language Attitudes Students were asked to indicate their attitudes to language use in Israel,
comparing Hebrew and other languages used by immigrant groups. They were asked
to express their opinion on the following statements, from “totally agree”, “agree”,
“disagree”, down to “totally disagree”. The following are the statements expressing
the different attitudes towards Hebrew and other language use in Israel: “In everybody
should talk only Hebrew”, “It is important for immigrants to continue speaking their
native language”, “In a public place, new immigrants should not talk in their native
language", “You don’t have to know Hebrew in order to get along in Israel”, "You
want to feel Israeli you have to know Hebrew”, “It is important to speak the language
at home in order to stay in contact with the family”. The opinions “agree” and “totally
agree” were counted together as were “disagree” and “totally disagree” in an attempt
to form a clearer and more dichotomous picture from the various responses.
Preservation of the native language (“it’s important for immigrants to continue
speaking their native language”) is supported by over two-thirds of all the students.
An even higher proportion (around 85%) expressed support for using the language at
home. Between two-thirds to three-quarters of the younger children also expressed
support for the statement “If you want to feel Israeli you have to know Hebrew”.
However, with age, in all three groups, fewer students agree with this. Around half the
students in all three groups across ages agree that the use of Hebrew is needed in
order to get along in Israel.
Relating to Hebrew dominance in schools and use of a language other than
Hebrew in public places, the students of the research group expressed more
conservative attitudes. Two thirds of them favor Hebrew only in schools and around
one half favored only Hebrew in public places. In comparison, 16% to one quarter of
the control group students expressed similar opinions on these two issues.
In sum, it seems that with age there is more tolerance and acceptances of language
variability in immigrant and non-immigrant youth.
104
3.16 Interviewer Information on the Questionnaire as a Task
The information in this section is taken from the reports given by the
interviewers after they had finished interviewing each student. By “interviewers”
here, reference is made to the Ethiopian and non-Ethiopian adults who conducted the
fieldwork and met the students on a one-to-one basis for filling out the questionnaire.
Each of them was asked to fill out personal details and then to provide information on
how their interviewees handled the task. Before filling out the questionnaire, students
in the two research groups were given the option to fill out the questionnaire in the
presence of any one of eight interviewers. These adults were divided by sex (four
women and four men), by age (four in their twenties and four in their forties), and by
origin (four Ethiopians and four non-Ethiopians). This was done in order to make the
Ethiopian children feel more at ease to “share” personal details concerning their
homes with a familiar adult from their own community and sex. (Students in the
control group had no choice, they were all interviewed by the investigator herself, an
Israeli-born, non-Ethiopian woman). Students in the two research groups revealed
almost identical preferences with regard to interviewer’s age and gender. More than
half chose a young interviewer and two thirds chose a female interviewer. The
majority (around two thirds) preferred a non-Ethiopian interviewer, and there was
almost no difference between the Ethiopian and non-Ethiopian students in this
respect.
The data in this section are based on the interviewers’ reports provided in
writing at the end of each interview. They were asked to specify how long it took their
students to fill out the questionnaire, how much help students needed in performing
the task, and how difficult students seemed to have found the task. They were also
asked to write their general impression including any important details or
interpretations. This was done in order to obtain information ‘from the inside’, which
might lead to an insight into the unique adults' perceptions.
3.16.1 Time required to fill out the questionnaire.
The average amount of time required by the Ethiopian students to fill out the
questionnaire came to 28.5 minutes – (ranging from 15 minutes to 55 minutes). The
non-Ethiopians averaged 24.5 minutes (ranging from 22 minutes to 27 minutes). The
control students were the quickest (average time 19.7 minutes). They were also the
105
most homogeneous (ranging from 18 minutes to 21 minutes), taking on average 8 to
10 minutes less time to complete the task than the two research groups at all three age
grade levels. In terms of age, the 4th graders needed the most time in all three groups,
taking 4 to 5 minutes more, on average, to complete the task than the children in both
the two older groups.
3.16.2 Need for help
Interviewers were asked to rate how much help students needed in order to
perform the task by indicating the number of times students addressed them with
questions or queries about the contents of the questionnaire and what they were
supposed to do. Responses were ranked as follows: “did not need any help"; "needed
a little help" (turned to the interviewer 3 times); "needed considerable help" (turned to
the interviewer 4 or more times).
Responses to these items yielded the following picture. Fourth graders across
the three groups required an identical amount of assistance (averaging 3 queries or
requests for help). In contrast, in the two higher age groups, students in the control
group asked for more help (2.6 on average) than did the others (1.8 on average from
the Ethiopian students). Across the three age groups, the mean amount of assistance
(ranked from a high of four down to zero) is almost identical in the three populations
(2.4 for the non-Ethiopian research group, 2.5 for the controls, and 2.6 for the
Ethiopians).
Interviewers further noted in their reports that the Ethiopian children turned to
them not so much because of difficulty they encountered in filling out the
questionnaire, but rather as an attempt on the part of the students to obtain more
details about the context of this mutual activity. For example, “ why did you choose
our school and not another one?", "Are there many Ethiopian researchers in your
university?", "Do you have Ethiopian friends where you work?", "Where are you
going to publish the results?", "Do you think that we (the Ethiopians) are good
students?". In contrast, students in the control group used the investigator as a source
of relevant information concerning the task, in order to perform it more efficiently.
For example, they posed questions of clarification such as “Why are you asking that?,
Do you want to know what kind of writing activities I perform at home?, Are you
interested in specific examples or generalizations?".
106
In conclusion, the amount of assistance, defined as the number of times
students addressed the interviewer with questions or queries, is almost identical in all
the groups. There seems to be a difference in the purpose of the Ethiopians’ and the
non-Ethiopians’ questions. While the non-Ethiopian queries are task-oriented, the
Ethiopians’ are aimed at information about the interviewer and general information
about the context of the research.
3.16.3 Task difficulty
Interviewers were asked to impressionistically rate the level of the difficulty
the questionnaire presented to students, as follows: “What is your impression of the
difficulty the questionnaire involved for the student?” They rated their impression
according to the following evaluation: "The student had no problems filling out the
questionnaire"; "the student filled out the questionnaire with ease"; "the student had
some difficulty in filling out the questionnaire"; " the student had great difficulty in
filling out the questionnaire".
Students in the control group in all three age groups were rated as not having
any difficulty at all in filling out the questionnaire (a mean level of 1.0 on a scale of 1
to 4). They also reported that both research groups at all three ages coped with the
task quite easily, but rated the Ethiopian students as having slightly more difficulty
than their non-Ethiopian peers (2.0 on average for the Ethiopians, 1.8 for the non-
Ethiopians).
In sum, the interviewers reported that according to their impression, filling out the
questionnaire was quite an easy task for all the students. They added that the control
students had less difficulty performing this task than the two research groups.
3.17 Parental Background Data related to parental background refers to the responses given by students
concerning their parents’ country of birth, and level of formal education.
3.17.1 Parents’ country of birth
Almost all the Ethiopian students reported that their fathers were born in
Ethiopia (except for one). Around half the non-Ethiopian fathers are Israeli born and
the other half were born in North Africa or in Europe. In the control group, two thirds
of the fathers are Israeli born and the other third in Eastern Europe or other western
107
countries (only one father was born in North Africa). Countries of origin of mothers
reveal a similar picture: all but one among Ethiopian students were born in Ethiopia;
more than half the non-Ethiopian mothers are Israeli born, with the rest mainly from
North Africa (around a quarter), and a few from Eastern Europe. Almost all the
control group mothers (80%) are Israeli born.
The parental origin of the three groups thus reflects patterns of new versus
established immigrants. The Ethiopian parents are predominantly newcomers who
immigrated to Israel from the late eighties, a third of the non-Ethiopian parents
immigrated to Israel around the fifties and most of the control group children have
parents who are Israeli born. Thus, most of the non-Ethiopian parents and the control
parents can be considered as “veteran” Israelis.
3.17.2 Parents’ level of formal education
Information regarding the parents’ level of formal education was derived from
the number of years that parents reported on themselves as having attended school at
the time that they registered their children in the school system. Results reveal
considerable variability between the groups in this respect. The Ethiopian fathers
report a mean of 1.9 years of schooling while the non-Ethiopian fathers, who live in
the same neighborhood, and whose children attend the same schools as the
Ethiopians, report a mean of 8.8 years. In even more marked contrast, the bulk of the
(highly selective) fathers of the children in the control group all completed at least
high-school, and the bulk report they have some years of formal academic education.
Much the same picture emerged for mean years of schooling of the children’s
mothers in all three groups. Again, there are great inter-group differences in this
respect: (8 years between Ethiopian and non-Ethiopian mothers and 14 years between
the research and the control group mothers on average). Ethiopian mothers report a
mean of 1.6 years of schooling, and none of them have high-school education. The
non-Ethiopian mothers all finished junior-high, and have on average 9.1 years of
school. On the other hand, all the mothers of children in the control group finished
high-school and some have higher education. These differences in parental level of
education also reflect new versus established immigrant patterns. The system of
formal schooling in Ethiopia is least developed, while the Israeli school system is
more advanced than that attended by most Jews in North Africa.
108
3.18 Home Literacy
The information in this section is taken from students’ responses to questions
concerning the languages in which their parents read and write, types and amounts of
reading and writing activities carried out in their homes, and people they turn to for
help when they encounter difficulty with their writing.
3.18.1 Languages of reading and writing
Children were asked to answer the questions “In which languages do your
parents read?” and “In which languages do your parents write?”.About two thirds of
the Ethiopian fathers were reported as reading and writing in Amharic, with only a
half reading and writing in Hebrew. In contrast, the vast majority of non-Ethiopian
fathers (95%), and all the control group fathers read and write in Hebrew. Two
Ethiopian students reported that their fathers read but do not write Amharic. One
Ethiopian student reported that his father reads but cannot write Hebrew. A very
similar picture emerges for mothers’ reading and writing: 60% of the children in the
Ethiopian group reported that their mothers read and write Amharic. About one
quarter reported that their mothers know how to read and write Hebrew. In
comparison, nearly all the non-Ethiopian mothers (95%) and all the control group
mothers read and write Hebrew. Again, as with their fathers, one Ethiopian student
reported that his mother reads Hebrew but does not write the language.
Worth noting is the fact that the Ethiopian students’ reports on their parents’
reading and writing in Amharic are not consistent with a general finding concerning
the Ethiopian Jews' proficiency in Amharic. An expert in Amharic languages, Dr.
Anbessa Teferra, estimated that two thirds of the Ethiopian population in Israel can
not read or write in Amharic (personal communication).
3.18.2 Home-based reading and writing activities
Students were asked to give examples of different types of reading and writing
activities carried out in their homes. To this end, they were given open-ended
questions as follows: “Give some examples of things your family has read at home
over the past week”, “Give examples of what your family has written at home over
the past week”. Responses were divided into the following text types: For reading
activities: letters: regular mail, e-mail and Internet; television subtitles; storybooks;
109
other books, school-related materials and newspapers. For writing activities: notes;
lists; school-related material; work-related material; letters; don’t know and none.
The data in this section is analyzed only by origin because the analysis by age
yields a high variability of the subjects' reports. All the students note reading mainly
newspapers, in nearly all homes of children in the control group homes (90%), in half
the Ethiopian homes and in a third of the non-Ethiopian homes. Reading of books is
the second most common reading activity in all the homes: in three-quarters of the
control group, half the non-Ethiopians, and one third of the Ethiopian homes.
The third most common reading activity is reading subtitles on television (two-thirds
of the control group, half of the non-Ethiopian group and one-third of the Ethiopian).
Reading of study-related material is the fourth most common reading activity in half
of the control group homes, in a fifth of Ethiopian homes and in a tenth of the non-
Ethiopians ones. Only a few Ethiopian students and a few non-Ethiopian students
mentioned reading of storybooks at their homes while around a tenth of control
students reported on it. Reading e-mails and using Internet is reported by a third of the
control group and by only 2 non-Ethiopian students. The Ethiopian students do not
report on reading done on the computer. Only a few Ethiopian and non-Ethiopian
students reported on reading letters in their homes. The students in the control group
did not mention such a reading activity at all.
Students also reported on the types of writing activities conducted in their
homes. The following trend emerges: approximately one tenth of the Ethiopian
students and a few non-Ethiopians could not specify any type of writing activities
conducted in their homes. Students in both research groups noted primarily letter
writing in a third of the Ethiopian and the control students’ homes and in a quarter of
the non-Ethiopian homes. None of the control students and only a few of both
research groups reported on letter-reading in their homes. Writing of study-related
material is the second most common writing activity: in a half of the control group
homes, in a fifth of the non-Ethiopians, and in a tenth of the Ethiopian homes. The
third most common writing activity is writing of work related material in around a
third of the control group students’ homes, while only a few non-Ethiopians and one
Ethiopian student also observe it in their homes. Note that reading of work-related
material was not mentioned by any of the students. Writing on the computer was not
mentioned by any of the students in the three groups. Note that a third of the students
in the control group report on reading done on the computer in their homes. Writing
110
of lists and notes is reported by a fifth of the non-Ethiopian and the control group
students. In comparison only a few Ethiopian students also report on such types of
writing in their homes.
In sum, it appears that in all the groups reading at home is mainly aimed at
enjoyment and is done for pleasure (reading of papers, T.V. subtitles and books). In
addition, only in the upper-class homes the common purposes of writing are studying
and communication (writing of study and work-related material and letters). These
writing activities are usually obligatory and required by others. The two research
groups seem to observe almost the same amount of reading and writing activities in
their homes. In comparison, the control group students observe more a greater variety
of literacy activities, and their family members are more engaged in doing them.
3.18.3 Sources of assistance
Sources of assistance refers to the more general context in which students
engage in writing activities. Specifically, students were asked to indicate what they do
and whom they turn to for help when they encounter difficulty with writing. Siblings
are mentioned as the people they turn to for help among around 15% of the research
population, both Ethiopian and non-Ethiopian, but by none of the control group.
Parents are listed by around one third of both “veteran” Israeli groups, non-Ethiopian
and controls alike, but by almost none of the Ethiopian students. Friends are listed by
a third of the Ethiopian and by only a fifth of the non-Ethiopian and control group
students. Teachers are most given as the people all the students turn to for help: two
thirds of Ethiopian students, 40% of the control group students, and a third of non-
Ethiopians. In other words, students in the research populations turn mostly to their
teachers for help with writing difficulties. Next the “veteran” Israeli groups most turn
to their parents, while the Ethiopian students ask their friends for help more than their
parents.
3.19 Summary of the Results In conclusion, the results presented here derive from four different sources:
a) students’ responses to the questionnaire,
b) interviewers’ reports,
c) language teachers’ evaluations,
111
d) parental reports on their own formal education.
Information is provided about the language experiences of students both inside and
outside of school, about their attitudes towards language in general and towards
writing in particular, and about home literacy.
The main results show that both research groups from low SES backgrounds
are quite similar, as far as language achievements, attitudes towards writing and even
literacy are concerned. These students are rated by their language teachers as
“average”. At home they mainly observe newspapers being read and letters being
written. They generally define writing as a notational system aimed at achieving
specific and practical instrumental goals and characterized mainly by its conventions
such as spelling. In addition, both research groups show the same preference towards
the kind of interviewer they prefer working with. Thus, it seems that the two research
groups appear to have very similar images of themselves as members of the school-
going population, as Israelis, and as people required to function in a world where
“linguistic literacy” is important for success.
In contrast, the control high SES group differs markedly from the two research
groups in all the above respects. They are rated by their language teachers as
“excellent”. At home their parents are more educated and they observe their family
members occupied with writing study and work-related material - some done on the
computer. They also tend to define writing as a special discourse style aimed at
developing one’s cognition and self expression and categorized by its content,
organization, language use and communicativeness. Thus, it seems that the control
students are more successful in school, and they observe more various and complex
literacy activities in their homes. Moreover, they define writing not only as a
notational system aimed at achieving practical goals, but rather like a more abstract
process, which can serve as a tool for self expression and cognitive enrichment. With
age, students from all backgrounds evidently come to share this last definition of
writing. Certain other developmental trends, which emerged, are that, with age,
students express more tolerance towards language diversity, and they have a higher
opinion of their own writing skills.
Across the age groups all the subjects from the different backgrounds express
the same opinions as to the easiest and the most difficult writing activities. All of
them also share higher self-evaluations of language achievements than the ones given
112
by their language teachers. Moreover, students and teachers alike evaluate that
expository writing is more difficult and more poorly performed than narrative writing.
Where reading at home is concerned, all the students report on newspapers as the
primary reading activity. All students also report that they primarily turn for help to
their teachers when having problems with their writing. According to the
interviewers’ reports the questionnaire was quite an easy task for all the students.
However, the Ethiopians needed rather more time to fill it out and the control group
did this far more quickly across the board. The subjects’ behavior during the task
reflects the normative modes of behavior and attitudes of their culture to this activity.
In sum, the literacy questionnaire provides us with the means for
characterizing what is involved in "becoming literate" in different socio-cultural
environments. Such a profile is indispensable especially for the recently immigrated
Ethiopian children, since there are no studies concerning home literacy of the
Ethiopian community.
113
Chapter 4: Discussion
The final chapter of this study is devoted to interpretation of its findings
complemented by comments on new directions for research and practice arising from
the results. The chapter is divided into two parts: a summary of findings from the
analysis of the subjects' texts and the literacy questionnaires (Section 4.1). This
section presents findings concerning linguistic and communicative behaviors that are
either shared or differ across populations ending with relevant comments on
methodological implications for further research and/or recommendations for
intervention by educators and teachers.
Section 4.2 elaborates on four key themes relating to issues of particular
relevance to the main goals of my study and suggests directions for further research in
relevant domains. The first of these themes - more is not necessarily better - relates to
the amount of verbal outputs as a characteristic typical of the language use of low
SES children. The second theme - early distinctiveness late command of genres -
seems to describe a shared developmental stage across populations in the process of
acquiring proficiency in writing. The third topic - research involvement of community
representatives - relates to a methodological issue, which I found to be of prime
importance in the process of analyzing texts written by members of different cultures.
Although this issue was discussed at some length in the section on methods (Section
2.4.5), I elaborate on it here in view of its relevance to both collection and
interpretation of data. The fourth theme - what makes writing communicative? -
reflects my conviction that effective interaction between writers and readers is a
critical feature of literate language use. The four themes are discussed below in light
of their relevance to this study, and in each case I conclude by considering
methodological implications for researchers and/or educators 11.
11 In order to maintain flow of discourse, this chapter avoids mention of relevant research literature except a few items not previously mentioned in this study. Elsewhere the reader is referred back to the sections where works were cited in earlier chapters.
114
4.1 Overview of the Research Findings
This study presents two sets of analyses. The first is a detailed linguistic analysis
of 192 written texts of Israeli low SES Ethiopian and non-Ethiopian schoolchildren
from the same three age groups, compared with their middle class peers of the same
groups and grade levels. The second set of analyses describes similarities and
differences in language proficiency, in literacy activities, and in attitudes at school
and at home, that are liable to affect the language use of children from different
backgrounds. This information on background factors was gathered by means of 114
literacy questionnaires12.
The two sets of findings together provide an integrated insight into the
development of children’s writing and serve to characterize what is involved in
“becoming literate” in different socio-cultural environments. The following
discussion thus aims: (1) to teasing apart the variables of age, genre, origin, and
population with evidence from different populations which differ from the original
study in which middle-class "mainstream" Hebrew native speakers were investigated;
(2) to establishing a profile of lexical, syntactic and thematic repertoires that which
characterize and differentiate the written language use of schoolchildren of different
ages across two discourse genres; (3) to identifying the complex interaction between
factors of cultural conventions, environmental literacy, attitudes towards writing, and
parental backgrounds elements affecting children's linguistic usage.
4.1.1 Patterns shared by the Ethiopian and non-Ethiopian subjects
Results from the two sets of analyses confirm the main research assumption to
the effect that immigrant and native-speaking children from similar low SES
backgrounds manifest similar writing behaviors across age and genre. Moreover,
analysis of their home and school literacy-related activities, their language
achievements, and attitudes towards writing also show no significant differences.
These findings do not support the popular view which equates Ethiopian origin with
12 The research population consisted of 96 subjects, each of whom produced 2 texts.
The questionnaire was administered to the 96 research subjects and additional 12 controls.
115
lower literacy interests and achievements compared to native speakers of Hebrew.
Moreover, the findings of this study indicate that some parental characteristics which
other studies view as affecting children's language proficiency (as detailed in Section
1.3) are less significant as has been claimed in relation to the writing proficiency of
low SES children from different cultural backgrounds. The parental characteristics at
issue here were measured by years of parental formal education (7 years difference
between non-Ethiopian and Ethiopian parents) and parental proficiency in Hebrew
(about half of the Ethiopian parents do not read and write Hebrew, in comparison to
95% proficiency in Hebrew in the non-Ethiopian group). The lack of obvious impact
of these two very different elements in home background leads me to believe that the
students’ own experience in literately-related activities at home and/or at school might
be more significant for the development of their text production abilities than the
formal education or writing proficiency of their parents. Since most complex literate
experiences occur in school, it is also probable that teaching is more influential than
home environment with regard to the development of writing in low SES children.
In order to tease apart these variables further research is required that would
examine attitudes to language and learning of low SES families from different cultural
backgrounds. Such research would need to consider parental perceptions of learning
in general and language achievements in particular together with investigation of
parental support and stress on intellectual development.
4.1.2 Patterns shared by low and middle-class populations
The findings of this study show no difference between children from different
SES backgrounds with regard to social-cognitive and higher-order information
processing and metacognitive skills of the kind that are required for global text
construction. Thus, the findings of this study are consistent with those of the original
study in which the youngest subjects, like those in the research study, proved well
able to write adequate narratives with well-constructed openings and episodes. In
addition, the subjects of this study like their "mainstream" peers manifested canonical
construction of narrative earlier than of expository texts - most likely for similar
reasons, since even preschoolers are exposed to and can produce narratives, whereas
expository texts are typical of school language and literacy. Moreover, for children in
all the populations the closing is the hardest and the last to be acquired. Production of
116
well-constructed narratives does not mean having the ability to organize the
information in these texts at a fully hierarchical level - and indeed only about one
third of the high-school children across the populations organized the information in
their narratives in a fully proficient hierarchical organized fashion. Taken together,
these results confirming findings from earlier research as described in Section 2.6.6 to
the effect that such high order information processing in narrative continues to
develop up to and beyond adolescence revealing greater and protracted construction
and comprehension of complex literacy-type texts (Kaplan, 2003).
Production of global expository structure reveals similar patterns across the
populations. It seems that all children acquire this ability only in the course of
schooling. The ability to engage in pre-planning of an entire expository text as
evidenced by appropriate topic introduction emerges generally between junior-high
and high-school age, since about half the students in the junior-high group of both
research and comparison populations still fail to write well-constructed openings in
their expository texts. In fact, even high-school children are not able to produce fully
organized expository texts, since they fail to provide a gradated flow of information
between general statements and specific comments, nor do they explicitly mark such
transitions by appropriate linguistic means. In addition, the ability to sum up ideas by
referring back to the main idea or generalization stated in the opening of the text is an
even late development one which only emerges in high-school. Data from the
comparison publications suggest that only adult writers are capable of organizing
information at such a level in the course of expository text construction. As noted in
Section 1.2. constructing a “good” expository text requires complex and highly
developed cognitive and linguistic abilities, such as the ability to coherently
formulate, organize and give verbal expression to ideas, the command of linguistic
means to express abstract content, metalinguistic awareness and self-monitoring of all
the above (Berman & Katzenberger, in press).
In terms of linguistic rather than discourse criteria, lexical density (measured
by number of content words per total number of words) and clause package length
(measured by number of clauses per clause package) were also found to be similar
across the populations. A prototypical clause package lies within the scope of two to
three clauses, and is insensitive to age and population. These two aspects of language
use might also represent universally shared abilities that are part of text construction. I
had thought middle-class children would score higher on these measures, since they
117
have more developed vocabularies (Section 2.6.2). The results indicate that these
abilities may be less connected to the size and scope of children’s lexicons, but rather
relate to general patterns of information organization. These shared patterns would
also explain the similar behavior of all the children with regard to levels of
information organization in both the narrative and expository texts. Memory
limitations might also play a role in this respect suggesting that extended research
across populations and in different languages is needed in order to confirm the
hypothesis that the ability to construct a well-formed piece of discourse involves
general cognitive processing in the organization of information that goes beyond
language-specific types of knowledge.
4.1.3 Patterns shared by low SES subjects from poor educational systems and
low SES native-speaking subjects in well-established schools (comparison to
Salmon, 2003)
No difference was found with regard to grammatical accuracy as measured by
number of grammatical errors per clause between low SES children from poor
educational systems and low SES children attending well-established schools (Section
3.5.1). Thus, these findings are not in line with my initial prediction, that low SES
children attending well-established schools would have less grammatical errors. The
heads of the language staff in schools attended by low SES children in the research
population reported that the language teachers tend not to correct the children’s
grammatical errors, so as not to impair their motivation to take part in the lessons of
their own free will. Besides, the teachers themselves were heard to produce
grammatical errors, mainly in verb agreement. Two of the heads of the language staffs
suggested that the teachers do this in order to adjust their way of expression to that of
their students in believing that they are developing positive feelings of students
towards teacher and creating a more congenial, less threatening atmosphere in their
lessons. In addition, the curriculum of grammar teaching in Israel, has paid little
attention to corrections of grammatical errors since the eighties, nor is there a specific
section on this topic in the matriculation examinations (Shalom, 1999). Thus, it seems
that low SES children produce more grammatical errors especially in verb agreement,
since they extract them from the language input they hear in their immediate
surroundings, both at home and in school. More research on the language use of
children from different SES environments with regard to specific types of
118
grammatical errors is needed in order to investigate the connection between language
input and grammatical error.
In terms of teaching strategies, teachers especially those of low SES children, I
feel, should be encouraged to correct grammatical errors, since low SES children are
exposed to grammatical inaccuracies not only at home but even in school. Teachers
need to provide a model to the correct use of language so as to promote their students’
awareness that complying with linguistic rules and conventions is a necessary step to
becoming proficient writers. The effect of schooling on grammatical error of low SES
children can be examined in schools where teachers emphasize the importance of
grammatical accuracy.
Another finding of this study is that low SES children who attend well-
established schools produce more words per clause than their peers from the same
SES background attending less achieving schools. These findings are consistent with
my prediction and may point to the significance of teaching in promoting writing
skills of low SES children. More research using a representative sample that includes
statistical analyses is needed in order to confirm such an assumption.
4.1.4 Effect of schooling in immigrant low SES subjects from poor educational
systems and immigrant subjects attending well-established schools (compared to
Rabukhin, 2003)
In an attempt to examine the effect of school and of literate background, a
recent study using the same methodology as mine, shows that low SES Russian
immigrant students, who have been raised in literate families and attend well-
established schools produce well-organized narratives and expository texts and use
rich literate language. In order to examine the effect of school on the language
achievements of the Ethiopian children, similar research is needed. Such a population
is harder to achieve since a large proportion of the Ethiopian parents do not read and
write either in Amharic nor in Hebrew, and their children attend low-achieving
schools (Habib, 2001).
119
4.1.5 Different language skills of the Ethiopian and non-Ethiopian subjects
The results of this study confirmed that the Ethiopian adolescents at junior-
high and high-school have more grammatical errors than their non-Ethiopian peers.
Although they are considered native speakers of Hebrew, since they have been using
the language for at least eight years, their deficit in exposure to literate language
seems to be the reason for their variability in performance concerning grammatical
accuracy. In contrast, 4th graders of Ethiopian origin have fewer grammatical errors
than their non-Ethiopian peers evidently because they were all born in Israel like their
non-Ethiopian classmates. It seems that when these two groups of children are all
Israeli-born and attend the same educational institutions, Ethiopian children are as
good as and even better in mastery of core grammar. In fact they might have even
transferred the demand for linguistic accuracy from Amharic to Hebrew, which is an
important norm in their culture.
Comparison of lexical density in the two research groups shows that the
Ethiopian children manifest identical means of lexical density in their narratives,
while the expository texts of the non-Ethiopian children show higher lexical density.
One explanation could be that in comparison to their non-Ethiopian peers, Ethiopian
children have less developed genre distinction due to less exposure in their homes to
written language. Another interpretation could be that due to their extensive
experience in listening to narratives in their homes they have developed better
abilities to produce narratives.
In particular terms, this suggests schools dealing with immigrant children from
this background need special attention to developing the students' command of non-
narrative texts that are so critical for academic success.
4.1.6 Different patterns in the three populations
The results of the present study point to SES as an important factor in
linguistic proficiency, since both research groups (immigrant and non-immigrant)
from low SES used fewer words per clause and fewer content words per clause across
genre (Section 3.2). The low SES groups of this study also displayed less proficient
usage in lexical selection, in discourse conventions and in register consistency
compared with the middle class children of the original study (for a discussion
concerning these results see section 4.2.1 below). Thus, the results of this study add
new insights on what characterizes the differences in language use between the two
120
populations. Interestingly enough, the results of this study indicate that SES is
sensitive to lexical richness but not to lexical density. Thus, the ability to use rich
language might be related to both formal learning and direct teaching either at home
or at schools, while the ability to use more content words might be related to general
cognitive development. A recent Israeli study also shows that socio-economic status is
the main predictor of language achievements in immigrant students as compared to
native speakers (Levin et al, 2003). They show that in Hebrew it takes 8-11 years for
immigrants to reach achievements similar to native speakers.
4.1.6.1 Thematic themes
The results of this study confirm the prediction that differences in thematic
content would be sensitive to population (Section 2.6.4). Children from different
backgrounds differ in what they choose to write about, and children from the same
background tend to rely on their mutual cultural repertoire in telling stories and
writing texts. My study is consistent with prior research on personal oral narratives
showing that it is a type of discourse used as a means of presenting one’s self-identity
to society (Section 3.4). Further, a typical tradition in story telling regards personal
experienced stories as a channel whereby narrators position themselves and are
positioned by others as the representatives of their cultural heritage, and they do this
via the specific thematic content they select. The findings of my study show that
around half the Ethiopian children wrote about affronts to their honor in comparison
to only 10% of the non-Ethiopian children. In this the non-Ethiopian low SES
subjects are far more similar to their middle-class "mainstream" peers. It seems as if
the "voice" of the Ethiopian culture is echoed in the children's narratives. This “voice”
acknowledges the Ethiopian children’s socialization and acquisition of one of their
culture’s most important norms - the honor code. Moreover, the children "tell" the
others that by Ethiopian standards, affronts to someone’s honor are considered as a
reprehensible type of behavior. This is in fact a message to the Israeli society sent by
members of a minority group. The Ethiopian children note what is culturally non-
canonical and "tell" the receiving society what ought to be done about it.
Another example of difference in thematic content is the refraining of
Ethiopian children from mentioning conflicts in their homes. It seems that they
expressed a cultural norm that prohibits sharing with outsiders information about the
121
private life of their family members. Such information is usually kept to oneself and is
shared only with friends who have proven their ability to keep secrets.
In addition, the texts of low SES children in general contain more content
words that express physical aggression than middle-class children (Section 3.4.2).
This is probably because children raised in high-risk neighborhoods either in Israel or
abroad are more exposed to situations of violence in their immediate surroundings.
Consequently, they tend to report more about violent behaviors by using expression of
aggression than upper SES children (Section 3.4.2).
In conclusion, it seems that children express their cultural norms in their
writing behavior. In order to build positive communication in a multicultural society
each of its members should be attentive to such cultural messages. Developing more
quantitative measures for measuring thematic content would enable researchers to
examine this issue by less subjective methods.
4.1.6.2 Avoidance of talk
The results of the pilot study showed that the Ethiopian children refrained from
talking in front of the researcher probably because they employed a cultural decree of
refraining from speaking to authoritative figures in general, and abstaining from
initiating any conversation with them in particular (Section 2.1.1). Consequently it
was decided to use only their written texts as adequate data. This avoidance of talk
was also confirmed by the teachers of the Ethiopian children, who reported these
children tend not to take part in conversations held in class, in which an adult was
presented. This behavior was explained by Mr.Yirga Erate, an Ethiopian educational
counselor, as complying with the Ethiopian way of learning at school - an Ethiopian
student is expected only to listen attentively to the teacher and not to hurry to flaunt
his knowledge.
The results of this study also show that Ethiopian students use less lexical
infelicities than their non-Ethiopian Israeli-born peers (Section 3.5.2). This avoidance
of using linguistic expressions, not fully understood, was also explained by Mr. Yirga
Erate as complying with another accepted practice - refraining from using unfamiliar
expressions in public, until one has mastered them in private. Thus, it is better for a
member of the Ethiopian community to keep silent than to make mistakes in front of
others.
122
All the above behaviors seem to reflect specific socio-cultural practices in
learning situations. The results of this study show that Ethiopian children need more
time to fill out questionnaires across age (Section 3.16.1). They need this time in
order to overcome the typical suspicion and caution characterizing Ethiopian
normative modes of behaviors. These results support Ben Ezer (1992) who claims that
the “internal time” of the members of the Ethiopian community is different to that of
others Israelis. They take more time to perform tasks during group dynamics and also
during psychological treatment. Their relaxed, unhurried pace characterizes their
working and learning. Thus, it seems that cultures also differ in their orientation and
perception of time, consequently more time on tests and during interactions with
figures of authority is recommended to members of a community who perceive time
in such a way. In my opinion the extra time that should be given to Ethiopian students
on tests would allow establishment of trust and complying with the politeness norm
that prevents them from getting straight to the point (Section 2.2.1).
Some studies on different cultures point to difference in communicational
behaviors at school. Phillips (1983) describes a specific minority behavior, which
contradicts the conventional practice of non-Indian teachers in the U.S. She shows
that these teachers of Indian children often misunderstand the Indian cultural aspects,
especially those that go beyond semantics. Cultural difference in communicational
behaviors, like non-verbal communication confuses teachers and students alike. On
the one hand, the teachers interpret and assess the children’s seemingly “inattentive”
behaviors as rude and as a sign for their inability to learn. On the other hand, the
children use them for expressing their respect for their teacher, and their
unwillingness to express competitive behavior towards their friends. These modes of
behavior reflect important values in the Indian culture. Moreover, the children
consider the teacher’s high voice and demand for direct eye contact as an invasion of
privacy, an act of defiance, or a demonstration of lack of trust (Smith, 1984; Phillips,
1983). Ben Ezer (1999) reports similar behaviors in the Ethiopian community.
Physical gestures of honor, like avoiding direct eye contact with elder members of the
family or figures of authority such as teachers, are preferred values in this community.
As in the Indian culture, direct gazing into the eyes is considered insolent and is
castigated. The teachers of the Ethiopian subjects in this study confirm that their
Ethiopian students manifest such behaviors in class.
123
4.1.6.3 Cultural conventions of structures
In some cases social-cultural orientations serve also as examples of structures
by which a particular culture expresses meaning using a specific type of discourse
component. The following example refers to the extensive reference of the Ethiopian
children to the video scenes, as part of the discussion on how the Ethiopian culture
expresses main ideas in moral stories by means of proverbs. In this example the socio-
cultural orientation is reflected in the relation between the type of a discourse
component and its function. In the Ethiopian culture, a proverb often begins a moral
story. It indicates the adequate moral judgement to a specific story telling situation
(Ben Ezer, 1992; Noy, 1988; Rozen, 1999). 13
As noted in the original study, the subjects were shown a short wordless video
clip depicting different conflict situations in a school setting. The video clip served as
a trigger for producing texts with a shared thematic content. The children were
explicitly instructed not to refer to the content of the video in their writing. In spite of
the prediction that the Ethiopian children would refer more to the video clip, it was
decided to keep the same methodology used in the original study in order to allow
comparison between the research and the original studies. The findings show that,
indeed, about 40% of the Ethiopian children in comparison to only 20% of the non-
Ethiopians referred to the conflict scenes in their expository texts. I think the
Ethiopian children regarded the video scenes as if they were proverbs, which precede
text construction. They paid more attention to them, in order to pin point the requested
moral. Thus, researchers should be aware that any text structure might also be cultural
specific with regard to where main ideas are expressed. Such information may be
relevant to data collection procedures.
Along similar lines, Ethiopian high-school children produce more clause
packages per narrative component especially in their settings and episodes. These
children belong to a culture where oral narratives are used as a central means of
socialization and education. Sometimes the same story is told to teach several
different morals. In order to clarify to the listener which is the “right” moral, the
13 In oral African societies proverbs are used as an indirect means of conveying
morals. For example, in western Kenya and northern Uganda the use of proverbs as an indirect style of moral indicating is more highly valued among speakers of Western Nilotic languages than the direct style typically adopted in Western cultures. (Personal communication with Prof. Edith Bavin, who examined motivation for language changes in Western Nilotic languages).
124
storyteller begins his or her story with a proverb and/or elaborates the context of the
story in the setting or the episodes by means of a lot of details. In these cultures clear
stated morals are not used, but rather are left for the listener to figure them out for
him/herself. The cultural norm of avoiding specifically articulated conclusions might
be the explanation why the Ethiopian subjects wrote less closings than their non-
Ethiopian peers. It seems the Ethiopian subjects acquired the conventional pattern of
story-telling in their community, while their non-Ethiopian peers are more used to the
western way of expressing morals in the final part of a discourse.
In conclusion, cultural background serves as a rich context for interpretation,
understanding and recall of specific elements found in texts. When analyzing different
types of discourse the researcher should be familiar with cultural variables, which
might affect text production.
4.1.7 Home literacy
With regard to home literacy, the responses to the literacy questionnaires in
this study show that upper-class children observe more various and complex literacy
activities in their homes. These activities include writing for studying or writing of
work-related materials - some done on the computer. Low SES children observe letter
writing mainly by family members, who do not use computers for writing (Section
3.18.2 ). In the research literature, evaluation of the child's literacy-related
environment is generally done by direct observations of the variety of stimuli in the
child's home that enable him/her to act in the realm of literacy (Bradley & Carldwell,
1979). An Israeli study used such measure for examining the literacy-related
environment of low SES preschoolers (Aram, 1998). In general, the researcher found
that in all low SES homes there were writing implements and in most of them (66%)
there were at least ten children's books.
A similar research using anthropological methods of direct observations has
not been done yet in the Ethiopian community. Such a research using naturalistic
observations of literate activities carried out in homes of these children is hard to
employ, since members of this community express caution and reluctance in
cooperating with researchers on practices carried out in their homes. Without such
research it would be difficult to scientifically confirm the assumption of a relation
between home literacy and language proficiency.
125
4.1.8 Attitudes towards writing
The results of this study show that the two groups also differ with regard to
attitudes towards writing (Section 3.14). The low SES children tend to define writing
as a notational system aimed at achieving practical instrumental goals. In comparison,
the upper-class children generally define writing as a mode to express one’s ideas and
feelings, and as a means to develop one’s thinking. These may be affected by the
students' own experience in writing. For example, Levi (1991) who examined the
writing of sixth graders which showed that half the children wrote texts out of school
mainly for practical instrumental purposes, mostly directed to their teachers or school
friends. Writing for self-expression or communicating with others was not common.
The tendency of low SES children to define writing primarily as a notational
system may also be related to differences in focus in the teaching of writing.
Examination of the amount, type, and uses of print activities in first grade classes in
different SES environments in the U.S showed substantial differences (Duke, 2001).
High SES students spent a larger percentage of print time with extended texts, and
had more opportunities to develop agency as print users, for example to write for
audiences beyond the teacher. In comparison, low SES students devoted a greater
proportion of time to text at the letter, word and phrasal/sentential levels. Anyon's
(1981) study showed that different school curricula and different content instructions
influenced the definition of knowledge in various socioeconomic school settings.
Working-class students defined knowledge with a concern for personal meaning using
the terms "practical", and "mechanical", whereas in "executive elite" schools students
defined knowledge in terms as "academic, intellectual, rigorous". I think that similar
research on the potential influence of differences in literacy instruction on the
conceptualization of writing could help to explain why low SES children acquire
different attitudes towards writing in comparison to high SES students. Particularly,
the question why only children of highly advantaged and literate backgrounds define
both knowledge and writing as connected to reason and to intellectual development?
The "instrumental" approach of low SES subjects appears, to my regret, to be in line
with the by-now discredited approach by Bernstein in the 1960 (Section 1.3), as well
as in line with the ideas of the Israeli scholar Frankenstein (1976) on lower-class
children being incompetent of abstract thinking.
126
The results of this study also show that only upper class children define writing
as a mode to express the writer’s thoughts, feelings and also the values of his culture.
I believe high SES subjects develop such attitudes because their parents put a great
emphasis on the importance of writing. Different parental mediation of writing in the
different populations was proven to affect children’s definition of writing. Aram
(1998) showed that one of the central contributors to the mother's mediation of
writing in low SES population in Israel is related to the way the mother presents
written language in instructional interaction with her preschool child, expressed in the
way the mother analyzes the word and explains the principles and rules of language to
the child. Few such explanations have been found among less literate mothers. Along
similar lines, Korat and Levin's (2001) study of the nature of mother-child
collaborative text writing, maternal pedagogical beliefs, and children's independent
text writing in two socio-economic (SES) groups showed that spelling issues occupied
a more dominant place in interaction among low SES than genre, monitoring of text,
and other discursive or linguistic domains. The claim made in this study that
difference between low SES and high SES populations lies in parental perceptions of
literacy is also based on the study of Fitzerald, Spiegel and Cunningham (1991) who
found that low SES less literate parents regard the development of literacy only as
acquisition of skills and rules of reading, while more literate high SES parents regard
it as a cultural process.
Additional research is needed to clarify the relation between attitudes to
writing and writing proficiency. Whatever the results of such studies, it is clear that
teachers in general and language educators in particular should expose their students
to a variety of writing activities and openly discuss the contribution of these activities,
especially to their learning, thinking, and preserving their cultural heritage. Such an
experience is critical since low SES children seem not to acquire familiarity with the
written language as a discourse style without external intervention. Familiarity with
writing and written language both as a discourse style and as a notational system is
crucial in providing young language users with an important extra-linguistic source of
language relevant knowledge: the encounter with a variety of genres (Gillis & Ravid,
in press). This encounter with language characteristics typical of various text types is
a necessary step in acquiring linguistic literacy (Ravid & Tolchinsky, 2002). Thus,
extended material-based experiences with a variety of texts is important. However,
these class activities alone will not be sufficient to address lower levels of literacy
127
achievement among low SES students. Such interventions should be followed by
discussions on the contribution of writing to achieving practical goals, and its
potential for self-expression and cognitive enrichment. In other words, in order to
educate students to higher levels of literacy, teachers should increase both the amount
and type of texts which are available, used and elicited in the classroom environment,
and so also enhance the opportunities of their students to develop control as readers
and writers. In addition, teachers should also help their students to develop attitudes
that regard writing as an important tool for reasoning.
4.2 The Four Themes The following is a discussion of four key themes which have emerged from
this study: (1) more is not necessarily better; (2) early distinctiveness, late command
of genres; (3) research involvement of community representatives; and (4) what
makes writing communicative?
4.2.1 More is not necessarily better
This section deals with a marked characteristic of the language use of low SES
children - the unexpected phenomenon of "verbosity", that is lots of verbal output.
The results of the present study show that mean number of words across genre was
significally higher in the low SES groups than in the middle class group. Moreover,
mean number of content words in the narratives was also significally higher in the low
SES groups (Section 3.3). I had expected opposite results, since upper-class children
have more developed lexicons and usually score higher on vocabulary tests than low
SES children (Section 2.1.1). One explanation could be the higher motivation of the
low SES groups to cooperate with the researcher and to do their best during the
process of text production. This collaboration arose out of deliberate attempts to foster
this motivation by culture-oriented adaptations in the methodology of data collection.
Subjects’ text production was encouraged and positively promoted by detailed
explanation of the purpose of the study; promise of full confidentiality; and the
presence of respectful authority figures during data collection, prior to its inception.
Another explanation might be that low SES subjects have more developed abilities to
produce narratives as a means of communication, since oral narratives are more
prevalent in their communities as a vehicle for expressing idea and messages
128
(Section 2.2.1). An additional reason might be pragmatic failure, due to unintentional
lack of adherence to normative rules of writing behaviors. Blum-Kulka and Olshtain
(1986) found that deviation from native norms of utterance length might be a potential
cause for pragmatic failure. Their comparison of the number of words used by non-
native and native speakers showed that non-natives used significally more words per
utterance. They explained this phenomenon of “too many words” (determined by
normative native use) in terms of a pragmatic failure caused by different expectations
of the non-native speakers regarding the normative amount of talk. This pragmatic
failure characterizes the learners’ lack of confidence in their communicative
competence, and is caused by inadequate awareness of specific cultural norms of
pragmatic appropriateness. They conclude that “the less confident you are that you
can get the meaning across, the more words and contextualized information you use”
(Blum-Kulka & Olshtain, 1986, p. 177). Along similar lines, Scott (1988) claimed
that sentence length should not be considered as an indicator of syntactic
development, especially if conciseness is important. Thus, verbosity does not
necessarily mean proficiency, and longer texts are not always better. On the contrary,
overindulgence in words may create a lack of appropriateness and make the
interlocutor less patient. The language use of the proficient writer is content-directed
and implicit and avoids verbosity by not including unnecessary details. That is, it
follows the Gricean principle of being informative and not redundant (Grice, 1975).
Along similar lines, this study shows that native speaker-writers from low SES
background are probably unsure of their effectiveness to produce communicative
written discourses, since their written production still suffers from verbosity. It also
seems that low SES writers are unaware of their superfluous use of words, since the
ratio of number of words in their texts and those of the subjects of the original study
increases with age. Thus, verbosity is evident especially among high-school students.
On the one hand, these writers possess the necessary linguistic knowledge to put their
ideas into words. On the other hand, they seem to be even less confident than their
middle-class peers that they can get their meaning across. Their over-lengthy texts
produce as a result a sense of impatience in the native-speaker reader. This is a sign
that writer and reader do not share the same norms of quantity in writing, and cause
for miscommunication and breakdown in the interaction. In sum, the difference
between the two populations should not be attributed to the relative scopes of their
129
lexicons, but rather to their level of communicative competence and/or to their
awareness of regulative norms of literacy.
An interesting finding is that when text size is neutralized, the subjects of the
original study use more words per clause across age groups. These findings show that
verbosity, as an indication of pragmatic infelicity, should be measured by deviation
from norms of text length, rather than from utterance length. Analogously to what
Brown (1973) and his colleagues found for mean length of utterance as a diagnostic of
early preschool linguistic development, so mean length of clause in texts produced at
school age is a clear indicator of later language development (Chafe & Danielewicz,
1989; Halliday, 1988; Saltzman & Reilly, 1999). In this respect, the middle-class
children show higher achievement than their low SES peers.
Two features also point to communicative problems of low SES children. As
noted, it seems that children use “more words” when they have difficulties in getting
the meaning across. This inappropriateness appears not only at the level of isolated
words, but also how words are combined into set phrases or collocations. This study
indicates that collocational appropriateness can serve as a criterion for accuracy and
fine-tuning of lexical usage. Deviations from collocational appropriateness (both
single words and multilexemic expressions) were common in our database. These
lexical infelicities which strike the reader as deviating from accepted or conventional
native usage show a clear developmental pattern in this study being typical of high-
school students, and not found at all in the texts of fourth graders, and relatively rarely
in the junior-high school texts. Further, Ethiopian children produce fewer lexical
infelicities than their non-Ethiopian peers, and middle-class children do so less even
than their low SES peers.
One explanation for this phenomenon may be that the writers do not appear to
be aware of the fact that they are violating conventional usage, since there was no
evidence that they attempt to correct or change these discordant or inappropriate
expressions. That is, they seem to be lacking the requisite monitoring skills or
“executive control” abilities for proficient written expression of their ideas. On the
other hand, these writers do have access to a relatively sophisticated, high-level, and
extensive lexical repertoire. From this repertoire, they select (or even sometimes
invent) items that neighbor on (but do not precisely convey) the intended meaning.
This type of linguistic usage involves a special kind of creativity or lexical innovation.
It indicates that the writers are in fact in the process of developing their
130
"sprachgefühl", and that they are working on precision and elegance of vocabulary
selection. Thus, being literate means not only having access to a rich repertoire of
words, styles, and codes, but also the ability to control them (Bialystok, 1986).
Besides, this study revealed that lexical infelicities typically occur in situations where
writers were attempting to raise the linguistic register, which they use. “Register” is
used here in the sense of level of linguistic usage, ranging from everyday casual or
colloquial style to highly formal, carefully monitored forms of expression. These
speaker-writers seem able to use a variety of linguistic patterns suitable to different
contexts and social situations in their language (Biber, 1995; Ferguson, 1994). The
fact that these attempts are sometimes unsuccessful suggests that these students do in
fact have considerable awareness of written language "as a special discourse style"
(Ravid & Tolchinsky, 2002), since they produce more "discordancies" in their
expository writing, which are characterized by complex linguistic expressions along
with a variation of linguistic register. But evidently, low SES children have not yet
developed a fully mature communicative awareness, since the results of this study
show that they produce three times more instances of register mixing in their writing
compared with "mainstream" children (Section 3.5.3).
Low SES children in this study, as in other studies, lack command of
conventionally appropriate forms of expression as shown by their use of the lexicon
of the target language. In this, they reveal lack of an ability confined to highly literate
language users - the inability to monitor one’s linguistic output so as be fully in
keeping with a given communicative situation. Thus, the fourth graders do not
produce lexical infelicities since their lexicons are still underdeveloped. Ironically, it
keeps them closer to native norms of lexical usage. In comparison, the low SES high-
school children deliberately attempt to use “elegant” or formal language. This not yet
fully literate attempt leads them to violate appropriateness. In comparison, upper-class
children typically use simpler or more everyday forms of expression in such contexts,
and so avoid inappropriate usage. The occurrence of discordant lexical collocations
should thus not be attributed to lack of metalinguistic awareness per se. On the
contrary, production of lexical infelicities especially by high-school children points to
the consolidation of their reflective metalinguistic powers in language (Karmiloff-
Smith, 1986, 1992) during this age and their rhetorical expressiveness (Slobin, 1977).
The occurrence of discordant lexical collocations seems, rather to be connected to
lack of two other, related types of knowledge. The first is linguistic proficiency in
131
terms of mastery of a full repertoire of expressive and rhetorical options in the target
language. This lack of mastery may be the reason why the Ethiopian children produce
fewer lexical infelicities. From a cultural viewpoint, the Ethiopian children might
have transformed their L1 norm of vocabulary accuracy and specificity from Amharic
to Hebrew (Section 2.2.1). A second factor underlying lexical infelicity might be
literacy-engendered monitoring of one’s own linguistic output. This control over
linguistic variation, which is a major characteristic of linguistic literacy, seems not to
be fully developed in low SES high-school children in general.
In conclusion, conventional lexical selection shows fine-grained properties of
developing text production abilities, and so serves as a good diagnostic tool for
examining textual maturity. Along similar lines, teachers should look at the children’s
unconventional lexical selection not only as errors, but as expressions of the
children’s implicit knowledge about the lexicon, and as steps in the long way
adolescents still have to go in the development of writing skills. Developing a feel for
language and an ear for idiomatic appropriateness probably requires extensive
exposure to contextualized literate language usage at home and in the school. Where
this is lacking, as the case of the research populations of this study, opportunities need
to be sought and provided elsewhere - by encouraging genuine communicative
reading and writing activities as part of education in the language arts. It is not clear,
however, if rote-learning and rote-teaching of lists of set expressions is the way to
overcome these deficiencies. Like Schulte (1988) I believe that although choosing the
appropriate word is probably something a person either has or lacks, it is a technique
which can be taught to some extent. Children can elaborate their vocabulary usage by
continuous practice. Teachers should find indirect ways of encouraging their students
to develop their linguistic awareness. Exposing their children, especially those from
low SES backgrounds to different types of texts, ones which use rich lexicons and
different registers, may contribute to the development of their “sprachgefühl".
A more detailed examination of the reasons behind SES differences with
regard to lexical selection is needed. The attitudes of the different populations towards
the value of lexical accuracy should also be probed in order to find possible relations
between vocabulary selection and usage and proficient text production.
132
4.2.2 Early distinctness late command of genres
A common developmental pattern emerged across populations in the process
of acquiring proficiency in writing; early acquisition but late command of genre
distinctiveness. The findings of this study confirm that even low SES fourth-graders
manifest clear genre-like distinctions in constructing narratives compared with
expository texts. Fourth-graders across populations from all backgrounds construct
well-formed narratives, but have difficulty in constructing well-structured expository
texts (Section 3.6). The subjects of this research use more words per clause, more
clauses per clause package and more content words per clause in their expository texts
than in heir narratives (Sections 3.2.1; 3.2.2.1; 3.3). Content words per clause are
sensitive to age only in the expository texts. Syntactic-discursive and discursive links
in clause packages are also sensitive to genre, and are produced more in expository
texts than in narratives (Section 3.2.2.2). The increase in syntactic complexity and in
lexical richness points to a greater deployment of expressive options in expository
texts and shows that with schooling, children improve their ability to construct
expository texts more than they refine their ability to write narratives. As noted,
mastery of expository discourse structure is reached only in high-school, since this
top-down hierarchical organization is too difficult for the younger children. The
relative complexity of the expository text is also reflected in the evaluations of both
the teachers and the students, who claim that it is more difficult and that the perform
worse in expository than narrative writing (Sections 3.10.2 and 3.10.5).
Nonetheless, since the ability to write, particularly expository prose is acquired
in school and is essential for academic success, teachers especially of low SES
children, should put more emphasis on it. Research shows that direct teaching of
narrative structure facilitated the production of better narrative texts (Dickson,
Simmons & Kameenui, 1995). McGee and Richards (1985) showed that school
graders who were directly taught global text structure in general and expository global
structure in particular recalled a lot of information. I believe that explicit instruction
of how to construct well-structured expository texts, and how to coherently organize
the information they contain is critical right from the beginning of high-school, since
by this age but not before, children from different backgrounds have the required
linguistic and cognitive skills to cope with these demanding school-based literacy-
minded tasks. Teaching strategies in relation to text structure and information
133
organization may have an effect on students' sensitivity to genre appropriateness,
which is a major feature of "later language development”.
Further research is needed on the affect of culture and background not only on
narrative writing, but also on expository prose in order to understand better the
development of this kind of ability in different populations.
4.2.3 Research involvement of community representatives
This heading concerns a methodological recommendation. Members of the
Ethiopian community were directly and deliberately involved in this study. This
enabled a wider cultural perspective “from the inside” as to suitable data collection
procedures as well as interpretations of the data (Section 2.4.5). Ethiopian adults
participated as part of the research team during the pilot study, which included one-
on-one meetings with the researcher for data collection. The Ethiopian children
appeared to cooperate best with a familiar adult from their own community than with
"a stranger"- the non-Ethiopian researcher whom they met for the first time. The
Ethiopian adults interpreted the children's reluctance to produce oral discourses in a
face to face interaction with the researcher as an expression of a normative sign of
respect for a figure of authority in their community. As a result it was decided to
include only the written texts as part of the research (Section 2.1). The Ethiopian
adults also participated as interviewers during data collection by means of the literacy
questionnaire, since the pilot study showed that the Ethiopian children felt more at
ease with an Ethiopian adult. Other research done in the U.S showed that participation
of Afro-American role models in class helped young Afro-American students
(Ascher, 1991). This study shows that both the Ethiopian children and their non-
Ethiopian peers expressed the same patterns in choosing Ethiopian and non-Ethiopian
adults for the literacy questionnaire task. The students, especially the Ethiopians, were
already familiar with the interviewers, and probably did not consider the non-
Ethiopian ones as "strangers" since these adults had helped them during the texts'
production. In comparison, it seems that native-born Israeli students regard the
Ethiopian adults as equally professional “teachers”, and their choice of adults is
devoid of ethnic prejudices. This is an encouraging sign for the recent process,
encouraged by the Ministry of Education, of integrating teachers from Ethiopian
origin into the Israeli educational system.
134
The Ethiopian interviewers also helped in interpreting the behavior of the
Ethiopian children towards the interviewers when filling out the literacy
questionnaire. Their sensitivity to cultural norms of behavior helped us to understand
that there are probably different reasons why Ethiopian and non-Ethiopian children
need to "interview" the adult interviewer. It seems that children of different socio-
cultural backgrounds pose questions to the adult interviewer for different reasons.
Thus, children of high SES background asked questions about the nature of the task in
order to perform it better. In contrast, the Ethiopian children elicited information
about the interviewer's intentions, attitudes and knowledge. I asked one of the
Ethiopian adults, a specialist in educational counseling, to explain this phenomenon.
He began his explanation with an Amharic proverb: “He who rushes into speaking his
mind is like a tree whose fruit prematurely ripen, so they are only useful as bird food”.
He added that he thought he knew what was going on. He told us that in the Ethiopian
community, any conversation begins unconsciously with a kind of a “test”. Before
starting to speak to the point, a feeling of mutual trust must be established. This is
done by gathering information about each other’s intentions, attitudes and knowledge.
Members of the Ethiopian community usually speak their mind only to well-trusted
people who have no ill-intentions, reflected in the way the speaker expresses his or
her respect. A respectful speaker avoids showing anger by raising his/her voice or by
making unjust humiliating or discriminatory remarks towards the other speaker or
towards his people. Refraining from using verbal insults or openly discussing private
matters or the troubles of the other is also considered polite behavior. A good
candidate for talking is someone who underestimates his own understanding
compared with the other's. If the other speaker does not comply with these norms, he
will not pass the “test”, leading to a breach of trust, and hence termination of the
conversation. This kind of behavior is typical of the basic suspicion and caution
characterizing normative modes of behavior in the Ethiopian community (Section
2.2.1). Thus, the goal of the questions of the Ethiopian children was probably to
establish a necessary feeling of trust, which is a prerequisite for any conversation
about personal information in the Ethiopian community. Hence, before testing or
posing questions to Ethiopian subjects, it is recommended that a trustful atmosphere
be established by clarifying intentions, showing respect, and relatively some
knowledge concerning the Ethiopian culture.
135
The Ethiopian adults also directed our intention to cultural messages, like the
common theme in their narratives - affronts to their honor (Section 3.3). Thus, the
participation of the Ethiopian adults helped in reaching interpretations, which are
culturally sensitive and usually inaccessible to researchers outside the specific
community under study. Hencefore, I strongly recommend that any research on
minority groups should include members of the specific community as part of the
research team.
In conclusion, “the world is not a loose collection of unconnected entities: We
are all part of the suprasystem of human society, and we are interdependent with
members of other societal subsystems throughout that larger community” (Ellsworth,
2000, p. 235). Thus, researchers, as examiners of human behaviors, should become
responsive to the different cultural aspects of text production and adjust their
methodology and interpretations accordingly.
4.2.4 What makes writing communicative?
This section deals with the main goal of this study: what it means to "become
literate" in a given socio-cultural background? Ravid and Tolchinsky (2002) define
"becoming literate" as having control over linguistic variation, possessing
metalanguage awareness, and familiarity with writing as well as written language both
as a discourse style and a notational system. This study regards a piece of discourse as
a piece of life, and supports the above definition of linguistic literacy by additional
data. The results show that understanding of the intended meaning of a written text is
better achieved by avoiding inaccuracies and ambiguity on the part of the writer, and
misunderstanding or uneasiness on the part of the reader. Both interlocutors should
ideally share similar attitudes towards written communication in general and actual
usage in particular. In every written discourse writers choose to present themselves to
the readers in their special way in accordance with the norms and values of their own
culture. The communication between the two depends on the writer's ability to
properly exploit a variety of lexical, syntactic, and rhetorical means. Text efficiency
and effectiveness also depends on the writer's ability to mark his/her text as a piece of
discourse that suits the writer's purpose and is easy to process. This is achieved by
using clearly identified and appropriate linguistic devices, strategies for marking of
information organization and global text structure. Proficient readers use these
136
markings for understanding the text, and try to reveal to whom the writer directed his
or her writing. They also look for the writer’s specific message or purpose, by being
sensitive to the cultural uniqueness of the writer and the special way of his/her culture
to put ideas into words. When these conditions are met, the two literate interlocutors
have a better chance to achieve understanding of the intended meaning.
Underlying this enterprise is the belief that in order to become fully-fledged
members of a literate society, children need to learn how to integrate writing in their
everyday life using their sensitivity to differences across text types. This is expressed
in their selected linguistic forms and words against the common ground of their own
culture.
137
References
Aarssen, J. (1996). Relating events in two languages. Tilburg: Tilburg University
Press.
Aisenman, R. (Ed.). (1999). Developing literacy across genres, modalities and
languages. Working Papers. (Volume I). Tel-Aviv University: International
Literacy Project.
Aisenman, R., & Berman, R. (2000). Rethinking lexical analysis. In M. Aparrici,
N. Argerich, J. Perera, E. Rosado, & L. Tolchinsky (Eds.), Developing literacy
across genres, modalities, and languages. Working Papers. (Volume III,
pp. 187-197). University of Barcelona. International Literacy Project.
Akinçi, M.A. (1999). Development of narrative competence of Turkish-French
bilingual children, aged 5-10 years, in France. Unpublished doctoral dissertation.
Lumiere-Université. Lyon 2. [In French]
Akinçi, M.A., Jisa, H., & Kern, S. (2001). Influence of L1 Turkish on L2 French
narratives. In L. Verhoeven & S. Srtömqvist (Eds.), Narrative development in a
multilingual context (pp. 189-208). Amsterdam/Philadelphia: John Benjamins.
Alexander, K.L., & Entwisle, D.R. (1988). Achievements in the first 2 years of school:
Patters and process. Monographs of the Society for Research in Child
Development, 52. (2 serial, No .218).
Allen, L., Cipielewski, J., & Stanovich, K.E. (1992). Multiple indications of
children's reading habits and attitudes: Construct validity and cognitive correlates.
Journal of Educational Psychology, 84, 489-503.
Anderson, S. (1985). Inflectional morphology. In T. Shopen.(Ed.), Language,
typology, and syntactic description, 2, (pp.150-201). Cambridge University Press.
Anglin, J.M. (1993). Vocabulary development and the growth of morphological
knowledge. In J.M. Anglin (Ed.), Vocabulary development: A morphological
analysis. Monographs of the Society for Research in Child Development, 58.
(10 serial, No. 118-152).
Anyon, J. (1981). Social class and school knowledge. Curriculum Inquiry, 11, 3-42.
138
Aparici, M., Tolchinsky, L., & Rosado, A. (2000). On defining Longer-Units in
narrative and expository Spanish texts. In M. Aparici, N. Argerich, J. Perera, E.
Rosado, & L Tolchinsky (Eds.), Developing literacy in different contexts and
different languages. Working Papers. (Volume III, pp. 95-122). University of
Barcelona: The International Literacy Project.
Aram, D. (1998). How mothers assist their children in coping with writing
assignments: Analysis of relations literacy among kindergartens in a development
town. Unpublished doctoral dissertation. Tel-Aviv University. [In Hebrew]
Ascher, C. (1991). School programs for African American students. New York:
ERIC Clearinghouse on Education. Columbia University. Teachers College.
Au, K.H. (1998). Social constructivism and the school literacy learning of students of
diverse backgrounds. Journal of Literacy Research, 30, 297-319.
Bahns, J., & Eldaw, M. (1993). Should we teach EFL students collocations?
System, 21, 101-114.
Bakhtin, M. M. (1986). Speech genre and other late essays. Austin: University of
Texas.
Bamberg, M. (1997). A constructivist approach to narrative development.
In M. Bamberg (Ed.), Narrative development: Six approaches (pp. 89-132)
Mahwah, NJ: Lawrence Erlbaum.
Bavin, E.L. (in press). Focussing on ‘where’: An analysis of Warlpiri frog stories. In
S. Stromqvist, L. Verhoeven (Eds.), Relating events in narratives: Typological and
contextual perspectives. Mahwah, N.J: Lawrence: Erlbaum Associates.
Beals, D., & Tabors, P. (1993). Arboretum, bureaucratic, and carbohydrates:
Preschoolers' exposure to rare vocabulary at home. Paper presented at the
Biennial Meeting of the Society for Research in Child Development,
New Orleans.
Ben Ezer, G. (1992). As a light in a clay pot: Migration and absorption of Ethiopian
Jews in Israel. Ruben Mass. [In Hebrew]
Ben Ezer, G. (1999). Cross-cultural misunderstandings: The case of the Ethiopian
immigrants in Israeli society. International Journal of Adolescent Medicine and
Health, 11 (1-2), 21-37.
Ben Ezer, G. (2002a). The mutual creative space as a tool for multicultural work.
In L. Kacen & R. Levi-Wiesel. Group work in multicultural society (pp. 161-149).
Tel-Aviv. Tchericover. [In Hebrew].
139
Ben Ezer, G. (2002b). The Ethiopian Jewish exodus: Narratives of the migration
journey to Israel 1977-1985. London: Routledge.
Ben-Rafael, E. (1994). Language, identity and social division. Oxford: Clarendon
Press.
Ben-Rafael, E. (2002). Multiculturalism and multilingualism in Israel. In S. Israel
(Ed.), Talking Hebrew 18, (pp. 67-84). Tel-Aviv University [In Hebrew].
Ben-Rafael, E., Olshtain, E., & Geist, E. (1994). Aspects of identity and language
acquisition among immigrants from the Commonwealth of Independent States.
The NCJW Research Institute for Innovation in Education. School of Education.
The Hebrew University of Jerusalem [In Hebrew].
Berman, R.A. (1988). Word-class distinctions in developing grammars. In Y. Levy,
I.M. Schlesinger & M.D.S. Braine (Eds.), Categories and processes in language
acquisition (pp. 45-72). Hillsdale, NJ: Erlbaum.
Berman, R.A. (1996a). Developing literacy in different contexts and in different
languages. Research proposal submitted to the Spencer Foundation, Chicago,
Illinois.
Berman, R.A. (1996b). Form and function in developing narrative abilities: The case
of ‘and’. In D.I. Slobin, J. Gerhardt, A. Kyratzis & G. Jiansheng (Eds.), Social
interaction, social context, and language. Essays in honor of Susan Ervin-Tripp
(pp. 243-268). Mahwah, NJ: Lawrence Erlbaum.
Berman, R.A. (1997a). Developing form/function relations in narrative texts.
Lenguas Modernas, 24, 45-61.
Berman, R.A. (1997b). Theory and research in the acquisition of Hebrew as a first
language. In Y. Shimron (Ed.), Studies in the psychology of language (pp. 37-69).
Jerusalem: Magnes. [In Hebrew].
Berman, R.A. (1997c). Narrative theory and narrative development: The Labovian
impact. [Special issue]. Journal of Narrative and Life History, 7, 235- 244.
Berman, R.A. (1998). Typological perspectives on connectivity. In N. Dittmar &
Z. Penner (Eds.), Issues in the theory of language acquisition (pp. 203-224).
Bern: Peter Lang.
Berman, R.A. (1999). Bilingual proficiency/proficient bilingualism: Insights from
Hebrew-English narratives. In G. Extra & L. Verhoeven (Eds.), Bilingualism and
Migration (pp. 187-208). Berlin: Mouton.
140
Berman, R.A. (1999a). Form/function relations in narratives. In R.A. Aisenman
(Ed.), Developing literacy across genres, modalities, and languages. Working
papers. (Volume I, pp. 80-93). Tel-Aviv University : International Literacy
Project.
Berman, R.A. (2000). Thematic perspectives on how children talk about interpersonal
conflict. In M. Aparici, N. Argerich, J. Perera, E. Rosado, & L Tolchinsky (Eds.),
Developing literacy in different contexts and different languages. Working
Papers. (Volume III, pp. 65-76). University of Barcelona: International Literacy
Project.
Berman, R.A. (2001a, May). (Why) are expository texts more difficult than
narratives? Talk given at Laboratoire Cognition and Development. Université
René Descartes Paris V.
Berman, R.A.(2001b, December). New perspectives in lexical categorization. Plenary
lecture, Israel Association of Dyslexia. Bar Ilan University.
Berman, R.A. (2002). Crosslinguistic comparison in later language development.
In S. Strömqvist (Ed.), The diversity of languages and language learning.
(pp. 25-44). Lund: Center for Languages and Literature.
Berman, R.A. (in press). Genre and modality in developing discourse abilities.
In C.L. Moder & A. Martinovic-Ziv (Eds.), Discourse across languages and
cultures:Typological studies in language series. Amsterdam: John Benjamins.
Berman, R.A. (Ed.). (2003). Developing discourse stance in different text types and
Languages [Special Issue]. Journal of Pragmatics, Discourse stance:
Crosslinguistic and developmental perspectives.
Berman, R.A., & Nir, B. (submitted). Linguistic indicators of inter-genre
differentiation in later language development. Journal of Child Language.
Berman, R.A., & Katzenberger, I. (submitted). Form and function in introducing
narrative and expository texts: A developmental perspective. Discourse Process.
Berman, R. A., Ragnarsdóttir, H. & Strömqvist, S. (2002). Discourse stance. Written
Language & Literacy, 5 (2). 255-290.
Berman, R.A., & Ravid, D. (1999). The oral/literate continuum: Developmental
perspectives. Final Report to Israel Science Foundation, Jerusalem.
141
Berman, R.A., & Slobin, D.I. (Eds.). (1994). Relating events in narratives: A
cross-linguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum.
Berman, R.A., & Verhoeven, S. (2002). Crosslinguistic perspectives of the
development of text-production abilities in speech and writing. Written Languages
and Literacy, 5 (1), 1-43.
Bernstein, B. (1960). Language and social class. British Journal of Sociology, 11,
261-276.
Bialystok, E. (1986). Factors in the growth of linguistic awareness. Child
Development, 57, 498-510.
Bialystok, E. (1993). Metalinguistic awareness: The development of children’s ideas
about language. In C. Pratt & A.F. Carton (Eds.), Systems of representation in
children: development and use (pp. 211-233). Chichester: John Wiley and Sons.
Biber, D. (1995). Dimensions of register variation: A crosslinguistic comparison.
Cambridge: Cambridge University Press.
Biskup, D. (1992). L1 influence on learners’ renderings of English collocations:
A Polish/German empirical study. In P.J.L. Arnaud & H. Bejoint (Eds.),
Vocabulary and Applied Linguistics (pp. 85-93). London: Macmillan Academic
and Professional Ltd.
Blum-Kulka, S. (1993). “You gotta know how to tell a story”: Telling tales and teller
in American and Israeli narrative events at dinner. Language in Society, 22 (3),
361-402.
Blum-Kulka, S., & Olshtain, E. (1984). Requests and apologies: A cross-cultural
study of speech act realization patterns. (CCSARP). Applied Linguistics, 5, 196-
213.
Blum-Kulka, S., & Olshtain, E. (1986). Too many words: Length of utterance and
pragmatic failure. Studies in Second Language Acquisition, 8 (2), 165-179.
Bos, P.(1997). Development of bilingualisim. Tilburg:Tilburg University Press.
Bosher, S., & Rowekamp J. (1992). Language proficiency and academic success:
The refugee/immigrant in higher Education. (ERIC Document Reproduction
Service No. ED 353 914)
Bowey, J.A. (1995). Socio-economic status differences in pre-school, phonological
sensitivity and first grade learning achievements. Journal of Educational
Psychology, 87, 476-487.
142
Boyd, S., & Nauclér, K. (2001). Socio-cultural aspects of bilingual narrative
development in Sweden. In L. Verhoeven & S. Strömqvist (Eds.), Narrative
development in a multilingual context (pp. 129-151). Amsterdam/Philadelphia:
John Benjamins.
Bradley, R. H., & Carldwell, B. M. (1979). HOME observation for measurement of
the environment: A revision of the preschool scale. American Journal of Mental
Deficiency, 84 (3), 235-244.
Britton, J. (1994). Understanding expository text: Building mental structures to
indicate insights. In M.A. Gernsbacher (Ed.), Handbook of psycholinguistics
(pp. 641-674). NY: Academic Press. Chicago: University of Chicago Press.
Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard
University Press.
Bruner, J. (1986). Actual minds, possible words. Cambridge, MA: Harvard University
Press.
Bruner, J. (1990). Acts of meaning. Cambridge, MA: Harvard University Press.
Cahana-Amitay, D. (2003). Where syntax meets discourse: Clause packaging in
narrative texts. Unpublished report.
Cahana-Amitay, D., & Berman, R.A. (1999). Initial ideas on L(onger)- Units.
In R.A. Aisenman (Ed.), Developing literacy in different contexts and different
languages. Working Papers. (Volume I, pp. 114-127). Tel Aviv University:
International Literacy Project.
Cahana-Amitay, D., & Sandbank, A. (2000). Clause packaging in written expository
texts. In M. Aparici, N. Argerich, J. Perera, E. Rosado, & L. Tolchinsky (Eds.),
Developing literacy in different contexts and different languages. Working
Papers. (Volume III pp.129-148). University of Barcelona: International Literacy
Project.
Chafe, W. L., & Danielewicz, J. (1987). Properties of spoken and written language. In
R. Horowitz and S. J. Samuels (Eds.), Comparing oral and written language
(pp. 35-112). New York: Academic Press.
Chall, J., & Snow, C. (1988). School influences on the reading development of
low-income children. Educational Letter. Harvard Graduate School of Education:
Harvard University Press
Collier, V.P. (1987). Age and rate of acquisition of a second language for academic
purposes. TESOL Quarterly, 21, 617-641.
143
Collier, V.P., & Thomas, W.P. (1988). Acquisition of cognitive-academic second
language proficiency: A six-year study. Paper presented at the American
Education Research Association conference. New Orleans.
Crystal, D. (1985). Dictionary of Linguistics and Phonetics. Oxford: Blackwell.
Cummins, J. (1981a). The role of primary language development in promoting
educational success for language minority students. Schooling and language
minority students: A theoretical framework. Los Angeles: California State
University Evaluation, Dissemination, and Assessment Center.
Cummins, J. (1981b). Age on arrival and immigrant second language learning in
Canada: A reassessment. Applied Linguistics, 2 (2), 131-149.
Daiute, C., & Griffin, T.M. (1993). The social construction of written narratives.
In C. Daiute (Ed.), The development of literacy through social interaction
(pp. 97-120). SF: Jossey–Bass.
Davis, L. (1978). Register and the teaching of language to the disadvantaged.
Iyonim Bahinux, 14, 113-138. [In Hebrew]
Day, T.M., & Tappan, M.B. (1996). The narrative approach to moral development:
From the epistemic subject to dialogical selves. Human Development, 39, 67-82.
De Fina, A. (2000). Orientation in immigrant narratives: the role of ethnicity in the
identification of characters. Discourse Studies. 2 (2), 131-157.
Denzine, N.K., & Lincoln, Y.S. (1994). Handbook of qualitative research.
London and New Delhi: Sage.
Derive, J. (1994). La pratique lettre dans les societes de culture orale, un exemple
negro-african. In U. Frith, G. Ludi, M. Egli, & C-A. Zuber (Eds.). Texts of
literacy. (Vol.III). ESF Network on Written Language and Literacy.
Dickinson, D. K., & Tabors, P.O. (1991). Early literacy: linkages between home,
school and literacy achievements at age five. Journal of Research in Childhood
Education, 6, 30-46.
Dickinson, D. K., & Tabors, P.O. (Eds.). (2001). Building literacy with language.
Baltimore: Brooks.
Dickson, S.V., Simmons, D.C., & Kameenui, E.J. (1995). Text organization:
Curricular and instructional implications for diverse learners. National Center to
Improve the Tools of Educators. Technical Report, No. 18. Eugene, OR: NCITE.
DiLeonardi, J. W. (1993). Families in poverty and chronic neglect of children.
Families in Society, 74 (9), 557-562.
144
Donitsa-Schmidt, S. (1999). Language maintenance or shift: Determinants of
language choice among Soviet immigrants in Israel. Unpublished doctoral
dissertation. University of Toronto.
Dromi, E. (2002). Stages in the development of Hebrew as a native language.
In P. S. Klein, & D. Givon (Eds.), Language learning and literacy in early
Childhood (pp. 9-41). Tel- Aviv. Ramot. Tel-Aviv University. [In Hebrew].
Dubow, E.F., & Ippolito, M.F. (1994). Effects of poverty and quality of the home
environment on changes in the academic and behavioral adjustment of elementary
school-age children. Journal of Clinical Child Psychology, 23, 401-412.
Duke, N. (2001). Print experiences and environments offered to children in very
low- and very high- SES first grade classrooms. Teacher Education Research, 37
(2), 441-478.
Ellsworth, J. B. (2000). Surviving change: A survey of educational change models.
N.Y: Eric. Clearinghouse on Information and Technology.
Eshet, T. (1980). The connection between visual perception and decoding of the
orthographic symbols in three groups of children: Disadvantaged, learning
disabled and mild retarded. M.A. Dissertation. Tel-Aviv School of Education,
Tel-Aviv University. [In Hebrew].
Erlich, S. (2001). Development of narrative and expository writing among junior-high
students. M.A. Dissertation. Tel-Aviv School of Education, Tel-Aviv University.
[In Hebrew].
Erlich, H., Kaplan, S., & Salmon, H. (Eds.). (2003) Ethiopian Christianity Islam and
Judaism. The Open University. [In Hebrew].
Extra, G., & Verhoeven, L. (1993). Immigrant languages in Europe. Clevedon, UK:
Multilingual Matters.
Farghal, M., & Obiedat, H. (1995). Collocations: A neglected variable in EFL.IRAL,
33, 315-331.
Ferguson, C.A. (1994). Dialect register and genre: Working assumption about
conventionalization. In D. Biber & E. Finegan (Ed.), Sociolinguistic perspectives
on register (pp. 15-30). New York: Oxford University Press.
Ferreiro, E., & Teberosky, A. (1982). Literacy before schooling. Exeter, NH:
Heinemann.
Finnegan, R. (1970). Oral literature in Africa. Oxford: Clarendon Press.
Firth, J.R. (1957). Papers in linguistics. Oxford: Oxford University Press.
145
Fitzgerald, J., Spiegel, D. L., & Cuningham, J. W. (1991). The relationship between
parental literacy level and perception of emergent literacy. Journal of Reading
Behavior, 23 (2), 191-214.
Foley, W.A., & Van Valin, R.D. (1984). Functional syntax and universal grammar.
Cambridge: Cambridge University Press.
Fox, B. (1987). Rhetorical structure analysis. In B. Fox. (Ed.), Discourse structure
and anaphora: Written and conversational English (pp. 77-92). Cambridge:
Cambridge University Press.
Frankenstein, C. (1976). Teaching as a social challenge. Jerusalem: School of
Education of the Hebrew University and the Ministry of Education and Culture. [In
Hebrew].
Freedman, A. (1987). Development in story writing. Applied Psycholinguistics, 8,
153-180.
Gayraud, F. (1999). The development of the differentiation between oral and written
language through the lexicon. Unpublished doctoral dissertation. Lumiere-Lyon 2
University. [In French]
Gayraud, F., Jisa, H., & Viguié, A. (1999). The development of syntactic packaging in
French children’s written and spoken texts. In R. A. Aisenman (Ed.), Working
papers in developing literacy across genres, modalities, and languages.
(Volume 1 pp. 169-181). Tel-Aviv University: International Literacy Project.
Gillis, S., & Ravid, D. (In press). Language acquisition. In J. Verschueren, J.O.
stman, J. Blommaert, & C. E. Bulcaen (Eds.). Handbook of Pragmaticsײ
(pp. 1995-2003). Amsterdam: Benjamins.
Giora, R. (1990). On the so-called evaluative material in informative text. Text, 10,
299-320.
Golan–Kuk, P., Horowitz, L., & Shefatia, L. (1987). The adaptation of immigrant
students from Ethiopia to the school system. The NCJW Research Institute for
Innovation in Education. School of Education. The Hebrew University of
Jerusalem. The Szold Institute. [In Hebrew]
Gregory, E. (1999). Myths of literacy: Childhood memories of reading in London’s
East End. Writing language and literacy, 2 (1), 89-111.
Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax
and semantics. Vol. 3: Speech acts (pp. 41-58). New York: Academic press.
146
Habib, J. (2001). Integration of Ethiopian immigrants into Israeli society: Challenges,
programs and policies, and future directions. The JDC-Brookdale Institute.
Haller, R. (1988). On the feelings for language and its epistemic value. In C. Nyiri
and B. Smith (Eds.), Practical Knowledge: Outlines of a Theory of Traditions
and Skills (pp. 22-135). London: Croom Helm.
Halliday, M.A.K. (1989). Spoken and written language. Oxford: Oxford University
Press.
Haiman, J., & Thomson, S.A. (Eds.). (1998). Clause combining in grammar and
discourse. Amsterdam: Benjamins.
Hanushek, E.A. (1990). The impact of differential expenditures on school
performance. Issue Analysis. 1-22.
Harshav, B. (1993). Language in time of revolution. Berkeley: University of
California Press.
Hart, B., & Risley, T.R. ( 1995). Meaningful differences in the everyday experiences
of young American children. Baltimore: Paul H. Brooks.
Hartmann, R.R.K., & Stork, F.C. (1972). Dictionary of language and linguistics.
Applied Science Publishers.
Hatav, G. (1985). Criteria for identifying the foreground. Theoretical
Linguistics, 12, 265-273.
Hazan-Rokem, G. (1993). Man is a bridge to man. The proverbs of Georgian Jews in
Israel. Jerusalem: Ben-Zvi Institute. [In Hebrew].
Heath, S.B. (1983). Way with words. Cambridge: Cambridge University Press.
Hickmann, M. (1995). Discourse organization and the development of reference to
person, space and time. In P. Fletcher & B. MacWhinney (Eds.), The handbook of
child language (pp. 118-194). Oxford: Blackwell.
Holden, G.W. (1997). Parents and dynamics of child reading. Westview Press.
Harper Collins.
Hunt, K. W. (1965). Grammatical structures written at three grade levels (Research
Report No. 3). Champaign, IL: National Council of Teachers of English.
Invernizzi, M. A., & Abouzeiad, M.P. (1995). One story map does not fit all: A cross
cultural analysis of children’s written story retellings. Journal of Narrative and
Life History, 5 (1), 1-9.
147
Johannsson, V. (1999). Word frequencies in speech and writing: a study of expository
discourse. In R.A. Aisenman (Ed.), Developing literacy across genres, modalities,
and languages. Working papers. (Volume 1 pp. 182-198). Tel-Aviv University:
International Literacy Project.
Kais. Y. (1979). Researches on the language of the disadvantaged. Iyonim Bahinox, 4,
109-128. [In Hebrew].
Kaplan, D. (1983). Order of acquisition of morpho-syntactic categories among
Hebrew-speaking children, aged 1:9-3:6. M.A. Thesis. Tel-Aviv University. [In
Hebrew].
Kaplan, D. (2003, June).Linguistic characteristics of difficult and easy texts for
comprehension. A lecture given at Script, Zichron Yaakov.
Karmiloff-Smith, A. (1986). From meta-process to conscious access: Evidence from
metalinguistic and repair data. Cognition, 23, 95-147.
Karmiloff-Smith, A. (1992). Beyond modularity. Cambridge: MIT Press.
Katzenberger, I. (1994). Cognitive, linguistic and developmental factors in the
narration of picture series. Unpublished doctoral dissertation. Tel-Aviv
University. [In Hebrew].
Katzenberger, I. (submitted a). Expository text organization. Written Language
& Literacy.
Katzenberger, I. (submitted b) Clause packaging: The building blocks of global text
architecture. Pragmatics.
Katzenberger, I., & Chana-Amitay, D. (2002). Segmentation marking in text
production. Linguistics, 40, 1161-1184.
Kemp, J. (1984). Native language knowledge as a predicator of success in learning a
foreign language with special reference to a disadvantaged population.
M.A. Thesis. Tel-Aviv University.
Kess, J. F. (1993). Psycholinguistics: psychology, linguistics, and the study of
natural language. Philadelphia: John Benjamins.
Korat, O., & Levin, I. (2001). Maternal beliefs, mother-child interaction, and child's
literacy: Comparison of independent and collaborative text writing between two
social groups. Applied Developmental Psychology, 22, 397-420.
Kress, G. (1993). Genre as a social process. In B. Cope & M. Kalantzis (Eds.), The
Powers of Literacy: A Genre Approach to Teaching Writing (pp. 22-37). London:
Falmer Press.
148
Kriz, S., Reilly, J.S., & Saltzman, D. (2000). Longer-Units in English. In M. Aparici,
N. Argerich, J. Perera, E. Rosado, & L Tolchinsky (Eds.), Developing literacy in
different contexts and different languages. Working Papers. (Volume III,
pp. 77-93). University of Barcelona: International Literacy Project.
Kuntay, A., & Nakkamura, K. (1993). Evaluative strategies in monological Japanese
and Turkish narratives. Paper given at 6th International Congress for the Study of
Child Language. Trieste, Italy.
Labov, W. (1972). Language in the inner city. Philadelphia: University of
Pennsylvania Press.
Labov, W. (1998). Some further steps in narrative analysis. Journal of Narrative and
Life History, 7 (1-4), 395-415.
Lahire, B. (1995). Families’ charts. Hautes Etudes. Gallimard, Le Seuil. [In French]
Landau B., & Gleitman, L. (1985). Language and experience: Evidence from the
blind child. Cambridge, MA: Harvard University Press.
Laparra, M. (1990). A question about immigrant children bilingualism. Migrants-
Formation, 83, 40-53. [In French]
Laufer, B., & Yano, Y. (2001). Reading and understanding unfamiliar words:
Do L2 learners understand how much they don’t understand? Reading Foreign
Language, 13, 549-566.
Leseman, P., & De Jong, P. (2001). How important is home literacy for acquiring
literacy in school? In L. Verhoeven & C.E. Snow (Eds.), Literacy and motivation
(pp. 71-94). Mahwah, NJ: Lawrence Erlbaum.
Levi, A. (1991). The writing of 6th graders outside of school - Reading and writing in
school and in daily life. Maagaley Kria. 20, 77-88. [In Hebrew].
Levin, I., Share, D.L., & Shatil, E. (1996). A qualitative-quantitative study of
preschool writing: Its development and contribution to school literacy. In M. Levi
& S. Ransdell (Eds.), The science of writing (pp. 271-293). N.J: Erlbaum.
Levin, D. (1965). Wax and gold, tradition and innovation in Ethiopian culture.
University of Chicago Press.
Levin, I., Ravid, D., & Rappaport, S. (1999). Developing morphological
awareness and learning to write: A two way street. In T. Nunes. (Ed.), Literacy in
theory and practice (pp. 77-104). Amsterdam: Kluwer.
149
Levin, I., Ravid, D., & Rappaport, S. (2001). Morphology and spelling among
Hebrew-speaking children: From kindergarten to first grade. Journal of Child
language, 28, 741-769.
Levin,T., Shohami, E., & Spolsky, D. (2003). Academic achievements of immigrant
students: Findings and recommendations for decision makers. Ministry of
Education. Department of the chief scientist. [in Hebrew].
Lev-Weizel, R. (2001). Poor children-perception of socio-economic status and future
professional career. Mifgash Leavoda Hinuhit – Socialit, 14,101-120. [In Hebrew].
Lifshitz, C., Noam, G., & Habib, J. (1998). The absorption of Ethiopian immigrants’
youth: A multi-dimensional perspective. The Ministry of Education and Culture,
The Ministry of Absorption, and SDC Israel. [In Hebrew]
Lyons, J. (1977). Syntagmatic lexical relations. Semantics, 1, 261-266.
McGee, M., & Richards, J. D. (1985). Teaching expository text structure to
elementary students. The Reading Teacher, April, 739-748.
MacWhinney, B.(1995). The CHILDES Project. Hillsdale, NJ: Erlbaum.
MacWhinney, B. (2000) The CHILDES project: tools for analyzing talk (3rd ed.).
Lawrence Erlbaum.
Maratsos, M. (1988). The acquisition of formal word classes. In: Y. Levy.,
I.M. Schlesinger & M.D.S. Braine (Eds.), Categories and processes in
language acquisition. Hillsdale, NJ: Erlbaum.
Matthiessen, C., & Thompson, S.A. (1988). The structure of discourse and
‘subordination’. In J. Haiman & S.A. Thompson (Eds.), Clause combining in
grammar and discourse (pp. 275-329). Amsterdam: John Benjamins.
McCarty, T.L. (1989). School as community: The Rough Rock demonstration.
Harvard Educational Review, 59, 484-503.
Mero, I. (2003). Mirkam- A project for language skills enrichment in multi-cultural
learning environments: An Evaluation report for the year 2002. Center of
Educational Technology. [In Hebrew].
Michaels, S., & Collins, I. (1984). Oral discourse styles: Classroom interaction and
the acquisition of literacy. In D. Tannen (Ed.), Spoken and written language (pp.
219-244). Norwood, NJ: Ablex.
Miles, B.M., & Huberman, A.M. (1994). Qualitative data analysis. (2nd ed.).
London: Saga.
150
Miller, P.J., Fung H., & Mintz, J. (1996). Self construction through narrative
practices: A Chinese and American comparison of early socialization. Ethos, 24,
237-280.
Minami, M., & McCabe, A. (1991). Haiku as a discourse regulation device: A stanza
Analysis of Japanese children’s personal narratives. Language in Society, 20. 577-
599.
Minami, M., & McCabe, A. (1995). Rice balls and bear hunts: Japanese and north
American family narrative patterns. Journal of Child Language, 22, 423-445.
Ministry of Education, Culture and Sport. (1996). Policy for language in education
in Israel. (Revised ed.). Jerusalem: The Circular of the Director-General.
[In Hebrew].
Ministry of Education and Culture. (1999). Literacy skills in the first grade:
Expected results. Ministry of Education and Culture. [In Hebrew].
Ministry of Education. (2001). Meizav.http://www.education.gov.il/haaracha/
download/etyopim-m.pdf. [In Hebrew].
Ministry of Education. (2002). Immigrant students in educational institutions.
[In Hebrew]. The information Unit in Electronic Mail. [In Hebrew].
Minkovich, A., Davis, I., & Bashi, I. (1977). Assessment of elementary school
educational achievements in Israel. Jerusalem: School of Education, The Hebrew
University. [In Hebrew].
Molokandov, N. (1999). Attitudes, motivation, self identity and language choice
among Ethiopian teenagers. Unpublished seminar paper, presented at The School
of Education, Tel-Aviv University. [In Hebrew]
National Center for Educational Statistics. (1998). National assessment of educational
progress (NAEP). http://nces.edgov/nationreportcard/site/home.asp
National Center for Educational Statistics. (1999). http://nces.edgov/
Nelson, K. (1986). Event knowledge. Hillsdale, NJ: Lawrence Erlbaum
Nelson, N.W. (1988). Reading and writing. In M.A. Nippold (Ed.), Later language
development: Ages 9 through 19. Austin, TX: pro–Ed
Nezer, V., & Polani, H. (1988). A legend of immigration: The Ethiopian Jews and
their linguistic absorption. Department of Adult Education, The Ministry of
Culture and Education. [In Hebrew]
151
Nicolopoulou, A. (1996). Narrative development in social context. In D.I. Sslobin, J.
Gerhardt, A. kyratzis Kyratzis & . G. Jiansheng. (Eds.). Social interaction, social
context and language. Mahwah. NJ: Lawrence Erlbaum.
Nicolopoulou, A. (1997). Children and narrative: Toward an interpretive approach.
In M. Bamberg (Ed.), Narrative development: Six approaches (pp. 179-215).
Mahwah, NJ: Lawrence Erlbaum.
Nippold, M.A. (Ed.). (1988). Later language development: Ages 9 through 19.
Austin, TX: pro–Ed.
Nippold, M. (1998). Later language development: ages 9 through 19.
Boston: Little, Brown.
Nir-Sagiv, B. (2002). Spelling mistakes among children from different populations - A
developmental perspective. Unpublished seminar paper, presented at The
Linguistic Department, Tel-Aviv University. [In Hebrew]
Noy, D.(1988). Animal stories of Beta Israel. In V. Nezer & H. Polani (Eds.),
A legend of an immigration (pp. 49-66). The Ministry of Culture and Education.
Olson, D.R. (1994). The world on paper. Cambridge: Cambridge University Press.
Ong, W.J. (1992). Writing is a technology that restructures thought. In P. Downing,
S.D. Lima, & M. Noonan (Eds.), The linguistic of literacy (pp. 293-319).
Amsterdam: John Benjamins.
Paltridge, B. (1997). Genre, frames and writing in research settings.
Amsterdam: Benjamins.
Payne, A.C., Whitehurst, G.J., & Angell, A.L. (1994). The role of literacy
environment in the language development of children from low-income families.
Early Childhood Research Quarterly, 9, 427-440.
Peled, N. (1996). On the shift from speech to writing: Children’s textual production
in different genres. In N. Peled (Ed.), From speech to writing (pp. 281-307).
Jerusalem: David Yellin State Teachers’ College. [In Hebrew]
Peterson, C., & McCabe, A. (1983). Developmental psycholinguistics: Three ways of
looking at a child’s narrative. New York: Plenum.
Pfaff, C. (2001). The development of co-constructed narratives. In L. Verhoeven &
S. Strömqvist (Eds.), Narrative development in a multilingual context
(pp. 53-188). Amsterdam: Philadelphia: John Benjamins.
Pflaum, S.W. (1986). The development of language and literacy. Columbus, OH:
Merill Publishing Company.
152
Phillips, S. U. (1983). The Invisible Culture: Communication in Classroom and
Community on the Warm Springs Indian Reservation. New York: Longman.
Plomin, R., DeFries, J.C., & Fulker, D.W. (1988). Nature and nurture during infancy
and early childhood. Cambridge: Cambridge University Press.
Polinsky, M. (1997). American Russian: Language loss meets language acquisition.
In W. Browne (Ed.), Formal approaches to Slavic linguistics (pp. 370-407).
Ann Arbor: Michigan Slavic Publications.
Pontecorvo, C., & Orsolini, M. (1996). Writing and written language in children’s
development. In C. Pontecorvo, M. Orsolini, M. Burge, & L. Renisk (Eds.),
Children’s early text construction (pp. 3-23). Mahwah, NJ: Lawrence Erlbaum.
Portes, A., & Rumbaut, R.G. (2001). Legacies: The Stories of the Immigrant Second
Generation. University of California Press.
Purcell-Gates, V. (1988). Lexical and syntactic knowledge of written narratives.
Research on Teaching of English, 22, 128-158.
Rabukhin, L. (2003). Written discourse in bilingual students: A developmental study.
M.A.Thesis. Tel-Aviv University [In Hebrew].
Ragnarsdóttir, H., Cahana-Amitay, D., van Hell, J., Rosado, E., & Viguié, A. ( 2002).
Verbal structure and content in narrative and expository texts in five languages.
Written Language and Literacy, 5(1), 95-125.
Ravid, D., (2002). Language and literacy in the transition from kindergarten to first
grade. In P.S. Klein & D. Givon. (Eds.), Language learning and literacy in early
childhood (pp. 43-70). Tel-Aviv. Ramot. Tel- Aviv University. [In Hebrew].
Ravid, D., (2003). Emergence of linguistic complexity in written expository texts:
Evidence from later language acquisition. In D. Ravid & H. Bat-Zeev
Shyldkrot (Eds.), Perspectives on language and language development. Dordrecht:
Kluwer.
Ravid, D., & Avidor, A. (1998). Acquisition of derived nominals in Hebrew:
Developmental and linguistic principles. Journal of Child Language, 25, 229-266.
Ravid, D., & Cahana-Amitay, D. (in press). Verbal and nominal expressions in
narrating conflict situations in Hebrew. Journal of Pragmatics.
Ravid, D. & Tolchinsky, L. (2002). Developing linguistic literacy: A comprehensive
Model. Journal of Child Language, 29, 453-457.
153
Reilly, J.S., Baruch, E., Jisa, H. & Berman, R.A. (2002). Propositional attitudes in
written and spoken language. Written Language and Literacy, 5 (2), 183-218.
Romaine, S. (1995). Bilingualism (2nd ed.). Oxford: Blackwell.
Rozen, H. (1999). The meaning of proverbs as a way of understanding the Ethiopian
Jews’ culture. In R. Izak (Ed.), The proverb as a bridge to communication
(pp. 12-13). The Ministry of Education, Culture and Sport. [In Hebrew].
Salmon, E. (2003). Lexical development in textual contexts across the school years:
the effect of the school system. M.A. Thesis. Tel-Aviv University. [In Hebrew].
Saltzman, D. & Reilly, J. (1999). English linguistic development in school-age
children and adolescents. In R.A. Aisenman. (Ed.), Developing literacy across
genres, modalities, and languages. Working papers. (Volume 1,. pp.224-236).
Tel-Aviv University: International Literacy Project.
Scarborough, H.S., & Dobrich, W. (1994). On the efficiency of reading to
preschoolers. Developmental Review, 14, 245-302.
Schleifer, M., & Katzenberger, I. (2001, July). Global structure of texts written by
children from different SES. A lecture given at Script, Tel Aviv.
Schulte, J. (1988). Remarks on sprachgefühl. In C. Nyiri & B. Smith (Eds.), Practical
Knowledge: Outlines of a Theory of Traditions and Skills (pp. 136-36). London:
Croom Helm.
Schwarzwald, O. (1978). The influence of demographic variables on linguistic
performance. Quantitative Linguistics, 1, 173-197.
Schwarzwald, O. (1981). Grammar and reality in the Hebrew verb. Ramat- Gan:
Bar Ilan University Press. [In Hebrew].
Scott, C. (1988). Spoken and written syntax. In M. A. Nippold (Ed.), Later language
development: Ages nine through nineteen (pp. 49-95). Austin, TX:PRO-ED.
Scott, C. (1995). Syntax for school-age children: A discourse perspective. In M. Fey,
J. Windsor, & S. Warren (Eds.), Language intervention: Preschool through the
elementary years (pp. 107-141). Baltimore, MD: Paul H. Brookes.
Scott, C., Windsor, J. (2000). General language performance measures in spoken and
written narrative and expository discourse of school-age children with language
learning disabilities. Journal of Speech, Language, and Hearing Research, 43,
324-339.
154
Shalom, T. (1999). Changes in language teaching. In R. Globman & I. Iram (Eds.),
The development of teaching in the educational institutions in Israel
(pp. 419-438). Tel-Aviv: Ramot. [In Hebrew].
Shen, Y. (1985). The structure action in the short narrative text. Unpublished
Doctoral dissertation. Tel-Aviv University. [In Hebrew]
Sherman, A., & The Children’s Defense Fund (1994). Wasting America’s future:
The children’s defense fund report on the costs of child poverty. Boston: Beacon
Press.
Short, D.J., & Spanos, G. (1989). Teaching mathematics to limited English proficient
students. ERIC Digest. Washington, DC: ERIC/CLL.ED 317086.
Shtal, A. (1977). Language and cognition of Israeli students. Ozar Hamore.
[In Hebrew]
Shuman, A. (1986). Storytelling rights: The uses of oral and written texts by urban
adolescents. Cambridge: Cambridge University Press.
Shuster, Y. (1997). Elementary education for grown-up immigrants: The
Hagoshrim Project. School of Education, The Hebrew University of Jerusalem.
[In Hebrew]
Skutnabb-Kangas, T., & Toukomaa, P. (1976). Teaching migrant children’s mother
tongue and learning the language of the host country in the context of the socio-
cultural situation of the migrant family. Helskinki: Finnish National Commission
for UNESCO.
Slobin, D. (1977). Language change in childhood and in history. In J. MacNamara
(Ed.), Language learning and thought (pp. 185-214). New York:Academic Press.
Slobin, D. (1996). From “thought and language” to “thinking for speaking”
In J.J.Gumperz & S.C. Levinson, (Eds.), Rethinking linguistic relativity
(pp. 70-96) Cambridge: Cambridge University Press.
Snow, C., Barnes, W., Chandler, J., Goodman, I., & Hemphill. L. (1991). Unfulfilled
expectations: Home and school influences on literacy. Boston: Harvard University
Press.
Smilansky, S., & Shefatia, L. (1979). Closing the gap in learning achievements by
Teaching reading in pre-schools. Hed Hagan, 3, 127-146. [In Hebrew].
Smith, H.A. (1984). Non-verbal behavior aspects of teaching in non-verbal behavior.
Lewiston, NY: C.J. Hogrefe.
155
Sperry, I. L., & Sperry, D.E. (1995). Young children’s presentations of self in
conversational narration. New Directions to Child Development, 69, 47-60.
Spolsky, B. (1996). Hebrew and Israeli identity. In Y. Suleiman (Ed.), Language and
identity in the Middle East and North Africa (pp. 181-189). Richmon: Curzon
Press.
Spolsky, B., & Shohamy, E. (1999). Language in Israeli society and education.
International Journal of the Sociology of Language, 137, 93-114.
Storch, S. A., & Whitehurst, G.J. (2001). The role of family and home in the
developmental course of literacy in children from low-income backgrounds. In
W. Damon (Series Ed.), P. R. Britto & J. Brooks- Gunn (Vol. Eds.). New
Direction for Child Development and Adolescent Development, 92, 53-71.
Strömqvist, S. (1996). Discourse flow and linguistic information structuring:
Explorations in speech and writing. Gothenburg Papers in Theoretical
Linguistics, 72.
Strömqvist, S., Johansoon, V., Kriz, S., Ragnarsdóttir, H., Aizenman, R.,
& Ravid, D. (2001). Towards a cross-linguistic comparison of lexical quanta
in speech and in writing. Written Language and Literacy, 5 (1), 46-67.
Suárez-Orozco, M.M. (1989). Central American refugees and U.S. high schools:
A psychosocial study of motivation and achievement. Stanford, CA: Stanford
University Press.
Suárez-Orozco, M.M., & Suárez-Orozco, C. (1996). Transformations: Immigration
family life and achievement motivation among Latino adolescents. Stanford
University Press.
Svirski, S., & Svirski, B. (2002). The Ethiopian Jews in Israel: Housing, employment
and education. Israel Association for Ethiopian Jews. Shivyon, 11, 1-39. [In
Hebrew].
Tatar, M., Kfir, D., Sever, R., Adler, C., & Regev, H. (1994). Integration of
Immigrant students into the Israeli elementary and secondary schools.
The NCJW Research Institute for Innovation in Education, School of Education,
The Hebrew University of Jerusalem. [In Hebrew].
Teliya, V., Bragina, N., Oparina, E., & Sandomirskaya, I. (1998). Phraseology as a
language of culture: Its role in the representation of a collective mentality.
In A.P. Cowie (Ed.), Phraseology: Theory, analysis and applications
(pp. 161-186). New York: Oxford University Press.
156
Thomas, W.P., & Collier, V. (1997). School effectiveness for language minority
students. Washington, DC. National Clearinghouse for Bilingual Education.
Tolchinsky, L., Johansson, V., & Zamora, A. (2002). Text openings and closings:
Textual autonomy and differentiation. Written Language and Literacy,
5 (1 ) . 2 1 9 - 2 5 4 .
Tough, J. (1982). Language, poverty and disadvantage in school. In L. Feagans &
D.C. Farron (Eds.), The language of children reared in poverty (pp. 3-17).
NY: Academic Press.
Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson
(Eds.), Organization of memory (pp. 382-403). NY: Academic Press.
Ure, J.N. (1971). Lexical density and register variation. In G.E. Perren &
J.L.M. Trim (Eds.), Application of linguistics: Selected papers of the 2nd
International Congress of Applied Linguistics (pp. 443-452). Cambridge
University Press.
Van Dijk, T.A. (1980). Macro structures: an interdisciplinary study of global
structures in discourse, interaction and cognition. Hillsdale, NJ: Erlbaum.
Verhoeven, L. (1991). Predicting minority children’s bilingual proficiency: Child,
family and institutional factors. Language Learning, 41, 205-233.
Verhoeven, L. (1994). Transfer in bilingual development: the linguistic
interdependency hypothesis revisited. Language Learning, 44, 381-415.
Verhoeven, L. (In press). Bilingualisim and narrative construction. In S. Strömqvist
& L.Verhoeven (Eds.), Relating events in narrative: Typological and
contextual perspectives. Mahwah, NJ: Lawrence Erlbaum.
Verhoeven, L., Aparici, M., Cahana-Amitay, D., Van Hell, J., Kriz, S.D., &
Viguié-Simon, A. (2002). Clause packaging: A cross-linguistic developmental
analysis. Written language and literacy, 5 (1). 135-161.
Vernez, G., & Abrahamse, A. (1996). How immigrants fare in U.S. education.
Santa Monica: RAND
Walker, D., Greewood, C., Hart, B., & Carta, J. (1994). Prediction of schools’
outcomes based on early language production and socioeconomic factors.
Child Development, 65, 606-621.
Walton, M.D., & Brewer, C.C. (2002). The role of personal narrative in bringing
children into the moral discourse of their culture. Narrative Inquiry, 11, 2, 307-334.
157
Weil, S. (1988). The religious beliefs and practices of Ethiopian Jews in Israel.
Jerusalem: NCJW Research Institute of Innovation in Education, School of
Education, Hebrew University of Jerusalem. [In Hebrew].
Wigfield, A., & Asher, S.R. (1984). Social and motivational influences on reading.
In P.D. Pearson, R. Barr, M.L. Kamil, & P. Mosental. (Eds.), Handbook of
research on reading. (Vol.1, pp. 423-452). N.Y: Longman.
White, K. (1982). The relation between socioeconomic status and academic
achievements. Psychological Bulletin, 91, 461-481.
Whitehurst, G. J. (1997). Language processes in context: Language learning in
children reared in poverty. In L.B. Adamson & M.A. Romski (Eds.), Research on
communication and language disorders: Contribution to theories of language
development (pp. 233-266). Baltimore: Brooks.
Zehavi, A. (1992). The linguistic absorption of adolescents. In G. Ben Ezer. As a light
in a clay pot: Migration and absorption of Ethiopian Jews in Israel (pp. 156-159)
Ruben Mass. [In Hebrew].
158
Appendix I: Literacy Questionnaire Dear Students, We are trying to learn from you how you use your language at home and in school. We hope you’ll help us and answer all the questions. You can ask the interviewer to read and to explain the questions to you. Don’t feel ashamed to ask for help. We won’t pay attention to spelling mistakes. If you have to add comments, please write them in the appropriate place. If you do not know the answers to certain questions please leave the question blank.
1. Name: ____________________ 2. My mother tongue is: ____________________ 3. The language I speak with my parents: ____________________ 4. I have spoken Hebrew for the past eight years – please circle the correct
answer: a. Yes b. No 5. I was born in: ____________________ 6. My father was born in: ____________________ 7. My mother was born in: ____________________ 8. Write in the following table the names of languages you use. How would you
describe the way you speak, read and write in these languages: 5. excellent 4. very good 3. good 2. not so good 1. not at all
Name of language
I speak I read I write
Hebrew Other languages:
9.a. What languages do your parents speak? ____________________ 9.b. Who do they speak these languages to? ____________________ 9.c. In which languages do they read? ____________________ 9.d. In which languages do they write? ____________________ 10. People read all sorts of things: books, newspapers, stories, learning materials, T.V. subtitles, etc. Give some examples of things you have read at home in the last month: ______________________________________________________________________________________________________________________________________________________________________________________________________ 11. Give some examples of things your family has read at home during the last week: ______________________________________________________________________________________________________________________________________________________________________________________________________
159
12. Describe what happens when you bring home things you have written in class. Mark the appropriate column with an X:
Never Once in a while
Often Always
I tell my parents about what I wrote
I show my parents what I wrote
I read to my parents what I wrote
I save what I wrote, but don’t show it to anyone
My parents talk to me about what I have written
My parents show other people what I have written
Other:
13. Give examples for what your family has written at home during the last week. In which languages did they write? ______________________________________________________________________________________________________________________________________________________________________________________________________ 14. What do you like to write? Mark an X in the appropriate column:
Like very much
Like Don’t really like
Don’t like at all
Copying off the blackboard Writing answers to questions Writing a story Writing a summary Writing an essay Filling out a questionnaire Writing letters Writing poems Writing a personal diary Writing notes Other:
160
15. To what extent do you control each of the following writing activities? Mark an X in the appropriate column:
Do very well Do well Find it difficult
Find it very
difficult Copying off the blackboard Writing answers to questions Writing a story Writing a summary Writing an essay Filling out a questionnaire Writing letters Writing poems Writing a personal diary Writing notes Other: 16. What is the easiest writing activity for you, among the above writing activities? Explain: ______________________________________________________________________________________________________________________________________________________________________________________________________ 17: What is the most difficult writing activity for you, among the above writing activities? Explain: ____________________________________________________________________________________________________________________________________ 18. Do you think it is important to learn how to write? a. Yes b. No Explain: ____________________________________________________________________________________________________________________________________ 19. Give an example of something you have written in the past two weeks. What did you write? What did you do with what you wrote? ______________________________________________________________________________________________________________________________________________________________________________________________________ 20. When you find writing difficult, what do you do? Who do you turn to? ____________________________________________________________________________________________________________________________________ 21. Think about a person who writes well. What is good in this person’s writing? What is special in this person’s writing? ______________________________________________________________________________________________________________________________________________________________________________________________________
161
22. Below are some words describing feelings concerning the language your parents speak at home. Mark the words which describe your feelings towards the language your parents speak at home: (a) love (b) respect (c) proud of (d) don’t care (e) look down on (f) hate
(g) other: ____ Explain why you chose these words: ______________________________________________________________________________________________________________________________________________________________________________________________________ 23. Below are statements regarding language. Mark with an X the appropriate column:
Agree completely
Agree Disagree Disagree completely
In schools everybody should talk only Hebrew
It’s important that the new immigrants continue talking in their mother tongue
In a public place it is advisable that new immigrants don’t talk in their native language
You don’t have to know Hebrew in order to get along in Israel
If you want to feel Israeli you have to know Hebrew
It’s important to speak the language at home in order to stay in contact with the family
24. As you have noticed, the questionnaire is about writing, language and the attitude towards
them. If you have anything else to add, please comment here:
____________________________________________________________________________
____________________________________________________________________________
______________________________________________
Before returning the questionnaire, please read your answers again and check if you want to add or change something.
Thank you for your cooperation
162
Appendix II A narrative text produced by an Ethiopian high school girl according to the CHILDES convention of transcription Begin
@Participants: SBJ Subject @ID: th03fn1
@File name: th03fn1.cha
@Age of SBJ: 18;6.0
@Code of SBJ: h-03
@Name of SBJ: Bracha
@Sex of SBJ: female
@Origin of SBJ: Ethiopian
@Text Type: narrative
@Order: 1
@Number of Clauses: 76
@Number of LUnits: 1
*SBJ: alimut beyn ba'al le^ isha. [+ tit[ @bg: OPEN:SET
@bg: F-LU1/NS0//MC<S-rc
*SBJ: hayu shney xaverim.
%syn: MC
*SBJ: she^ ahavu ze et ze.
%syn: S-rc
@eg: F-LU1/NS0//MC<S-rc
@bg: LU2/NS0//MC;CDS-MC
*SBJ: ha^ xaver haya ben shloshim [@30.[
%syn: MC
*SBJ: ve^ ha^ baxura hayta bat esrim~ve~axat [@21.[
%syn: CDS-MC
@eg: LU2/NS0//MC;CDS-MC
@bg: LU3/NS0//MCD
*SBJ: ha^ baxura hayta baxura adisha.
%syn: MCD
@eg: LU3/NS0//MCD
@bg: LU4/SCN//MCD
*SBJ: ve^ hi gadla be^ bayit masorti. %syn: MCD
@eg: LU4/SCN//MCD
@bg: LU5/SCN//MC;CDS-MC
*SBJ: ve^ hayta la mishpaxa gdola shesh [@6] banot shiv'a [@7] banim.
%syn: MC
*SBJ: ve^ hi hayta axrona.
%syn: CDS-MC
@eg: LU5/SCN//MC;CDS-MC
@bg: LU6/NS0//MC<S-adv;CDS-MC;MCJ
*SBJ: imah niftera.
%syn: MC
163
*SBJ: she^ [*] hi hayta bat xamesh~esre [@15.[
%syn: S-adv
*SBJ: ve^ aviha haya ish kashuax meod.
%syn: CDS-MC
*SBJ: <hu loh haya ish ixpati mi^ yeladav <.[*]
%syn: MCJ
@eg: LU6/NS0//MC<S-adv;CDS-MC;MCJ
@bg: LU7/SCN//MC;CSSE-MC<S-adv<S2_S-ds-adv>S2_S-ds
*SBJ: ve^ ha^ baxura gadla be^ meshex xamesh [@5] shanim kashe.[*]
%syn: MC
*SBJ: ve^ savla meod.
%syn: CSSE-MC
*SBJ: she^ [*] hayta tamid xoshevet./"+
%syn: S-adv
*SBJ: +" im imi hayta xaya.
%syn: S-ds-adv
*SBJ: +" ani loh hayiti sovelet kax.
%syn: S-ds
@eg: LU7/SCN//MC;CSSE-MC<S-adv<S2_S-ds-adv>S2_S-ds
@eg: OPEN:SET
@bg: BOD:N
@bg: LU8/SC//S-adv>MC
*SBJ: ve^ az she^ [*] hi higi'a le^ gil esrim~ve~axat [@21.[
%syn: S-adv
*SBJ: hi hitxila lacet imm ha^ xaver shela.
%syn: MC
@eg: LU8/SC//S-adv>MC
@bg: LU9/SCN//MCD
*SBJ: ve^ hu meod [*] baxur akshan, bagdan, kanay bimyuxad ve^ kamcan.
%syn: MCD
@eg: LU9/SCN//MCD
@bg: LU10/NSC//MC;CSSE-MC
*SBJ: be^ meshex ha^ xaverut hu haya make ota be^ xagora, be^ matate, stirot lexi [stirot+lexi.[ %syn: MC
*SBJ: ve^ haya zorek ota mixuc le^ beyto aruma.
%syn: CSSE-MC
@eg: LU10/NSC//MC;CSSE-MC
@bg: LU11/SC//MC<S-rc><S-ds
*SBJ: ve^ kol ha^ xaverim shela ve^ ha^ mishpaxa [&she^ hayu makirim oto] amru la.
%syn: MC
*SBJ: <she^ hayu makirim oto> [CE./"+ [
%syn: S-rc
*SBJ: +" loh keday lax lehitxaten imm ha^ baxur ha^ ze.
%syn: S-ds
@eg: LU11/SC//MC<S-rc><S-ds
@bg: LU12/SC//MC<S-ds;S-dsj;S-ds-cds<S2_S-ds-pcomp;S-ds-css
*SBJ: ve^ hi amra./"+
%syn: MC
164
*SBJ: +" loh nora.
%syn: S-ds
*SBJ: +" anu stam ravnu al mashu katan.
%syn: S-dsj *SBJ: +" ve^ hu hivtiax li. %syn: S-ds-cds
*SBJ: +" she^ loh ya'ase zot shuv.
%syn: S-ds-pcomp
*SBJ: +" ve^ hu baxur tov.
%syn: S-ds-css
@eg: LU12/SC//MC<S-ds;S-dsj;S-ds-cds<S2_S-ds-pcomp;S-ds-css
@bg: LU13/NSC//MC;CDS-MC<S-nf *SBJ: axrey shnatayim hem hitxatnu.
%syn: MC
*SBJ: ve^ kol ha^ mishpaxa ve^ ha^ xaverim bau.
%syn: CDS-MC
*SBJ: rak bishvil lesameax ota.
%syn: S-nf @eg: LU13/NSC//MC;CDS-MC<S-nf @bg: LU14/SC//MC;CDS-MC
*SBJ: ve^ ha^ xatuna hayta yafa.
%syn: MC
*SBJ: ve^ kulam nehenu mimena.
%syn: CDS-MC
@eg: LU14/SC//MC;CDS-MC
@bg: LU15/NSC//MCD
*SBJ: axrey ha^ xatuna hi nixnesa le^ herayon.
%syn: MCD
@eg: LU15/NSC//MCD
@bg: LU16/SCN//MC<S-rc
*SBJ: ve^ hu haya shoter. %syn: MC
*SBJ: she^ hu [*] haya mefaked be^ taxant ha^ mishtara
]taxant+ha^mishtara.[
%syn: S-rc
@eg: LU16/SCN//MC<S-rc
@bg: LU17/NS0//MC;CSS-MC
*SBJ: hu acmo haya metapel be^ inyaney ha^ alimut [inyaney+ha^alimut.[ %syn: MC
*SBJ: ve^ hu mishtamesh ba^ bayit alimut [*] imm ishto.
%syn: CSS-MC
@eg: LU17/NS0//MC;CSS-MC
@bg: LU18/SCN//MC<S-adv
*SBJ: ve^ hu haya make.
%syn: MC
*SBJ: kol od she^ hayta be^ herayon.
%syn: S-adv
@eg: LU18/SCN//MC<S-adv
@bg: LU19/NS0//MC;CDS-MC
*SBJ: hi yalda et ha^ tinok.
165
%syn: MC
*SBJ: ve^ ha^ tinok haya beseder. %syn: CDS-MC
@eg: LU19/NS0//MC;CDS-MC
@bg: LU20/NS0//MC<S-pcomp;CDS-MC
*SBJ: ha^ rofim xasheshu.
%syn: MC
*SBJ: she^ yiye ba^ tinok eyzeshu pega.
%syn: S-pcomp
*SBJ: aval lemazalam ha^ tinok haya beseder. %syn: CDS-MC
@eg: LU20/NS0//MC<S-pcomp;CDS-MC
@bg: LU21/SC//S-adv>MC;CSS-MC<S-rc
*SBJ: axrey she^ hi xazra mi^ beyt+ha^xolim.
%syn: S-adv
*SBJ: hu natash ota.
%syn: MC
*SBJ: ve^ hu yaca imm baxura axeret. %syn: CSS-MC
*SBJ: she^ gam hi hayta be^ gila.
%syn: S-rc
@eg: LU21/SC//S-adv>MC;CSS-MC<S-rc
@bg: LU22/NS0//MC;CSS-MC
*SBJ: hu hizniax et ishto.
%syn: MC
*SBJ: ve^ hu loh haya noten kesef. %syn: CSS-MC
@eg: LU22/NS0//MC;CSS-MC
@bg: LU23/SCN//MC;CSS-MC<S-pcomp[S-adv>MC[
*SBJ: ve^ hu himshix lehakot ota.
%syn: MC
*SBJ: ve^ hu haya meayem aleha.
%syn: CSS-MC
*SBJ: she^ im tesaper le^ mishu.
%syn: S-pcomp+adv
*SBJ: hu yaharog ota.
%syn: MC
@eg: LU23/SCN//MC;CSS-MC<S-pcomp[S-adv>MC[
@bg: LU24/SCN//MC<S-pcomp
*SBJ: ve^ hi biksah mimeno.
%syn: MC
*SBJ: she^ yiten la get. %syn: S-pcomp
@eg: LU24/SCN//MC<S-pcomp
@bg: LU25/SC//MC<S-pcomp
*SBJ: ve^ hu amar la.
%syn: MC
*SBJ: she^ loh yiten la get. %syn: S-pcomp
@eg: LU25/SC//MC<S-pcomp
166
@bg: LU26/SCN//MC;MCJ<S-nf *SBJ: ve^ hi savla yom va^ layla.
%syn: MC
*SBJ: loh yad'a.
%syn: MCJ
*SBJ: ma la'asot. %syn: S-nf @eg: LU26/SCN//MC;MCJ<S-nf @bg: LU27/SC//MC<S-pcomp
*SBJ: ve^ levasof hi gilta la^ mishtara.
%syn: MC
*SBJ: she^ ba'ala hu make ota.
%syn: S-pcomp
@eg: LU27/SC//MC<S-pcomp
@bg: LU28/SC//MC;CSS-MC
*SBJ: ve^ hem tafsu et ba'ala.
%syn: MC
*SBJ: ve^ hem hixnisu oto le^ beyt+sohar. %syn: CSS-MC
@eg: LU28/SC//MC;CSS-MC
@eg: BOD:N
@bg: CLO:RES
@bg: LU29/SC//MCD
*SBJ: ve^ axshav hi xaya be^ rosh shketa [*] ve^ be^ shalva.
%syn: MCD
@eg: LU29/SC//MCD
@bg: LU30/NS0//MCD
*SBJ: sof ha^ sipur [sof+ha^sipur] le^ sikum. [+ tit[ %syn: MCD
@eg: LU30/NS0//MCD
@bg: LU31/SC//S-adv>MC;CSSE-MC
*SBJ: im loh mesaprim et ha^ beaya.
%syn: S-adv *SBJ: ha^ alimut timashex.
%syn: MC
*SBJ: ve^ loh tipasek.
%syn: CSSE-MC
@eg: LU31/SC//S-adv>MC;CSSE-MC
@bg: LU32/NS0//MCD
*SBJ: baxurot xayavot liyot ba'al [*] bitxon [ba'al+bitxon] gavoha.
%syn: MCD
@eg: LU32/NS0//MCD
@bg: L-LU33/NS0//MC<S-rc
*SBJ: isha [&ha^ sotemet et paneha] ba^ sof dofeket et acma.
%syn: MC
*SBJ: <ha^ sotemet et paneha [*]> [CE.[
%syn: S-rc
@eg: L-LU33/NS0//MC<S-rc
@eg: CLO:RES
@End
]ה[אל תניח לי להפוך אדיש: רק זאת אבקש מאלוהים"
"לקמילתם של אלה שנאלצו לעזוב ולחיות תרבות אחרת
ליאון גייקו
תקציר
הכוחות המשפיעים על ההתנהגות האוריינית של מהגרים החיים בקרב קבוצות אתניות אחרות
והתעצמות תופעת ההגירה הדמוגראפייםונה נושא חשוב למחקר עקב השינויים הפכו לאחר
" להיות אורייני"מהי המשמעות של : השאלה המרכזית של מחקר זה היא. בעולם המודרני
תרבותיות על התפתחות -מחקר זה מתרכז בבחינת ההשפעות החברתיות, לפיכך? בסביבות שונות
תרבותי שלהם שונים - פה המדוברת בבתיהם והרקע החברתיכישורי שיח כתוב בקרב ילדים שהש
המטרה הכוללת של מחקר זה היא להבין . מאלה של החברה הקולטת ושל המערכת הבית ספרית
חברתיות שונות כותבים טקסטים -ומסביבות תרבותיות, כיצד תלמידים בגילים שונים
. מונוליתיים
בגיל בית , תי קבוצות של תלמידים דוברי עבריתהמחקר מנתח טקסטים כתובים שהופקו על ידי ש
הקבוצה הראשונה מורכבת מילדי מהגרים בני העדה . אקונומי נמוך- ממיצב סוציו, הספר
שתי הקבוצות . ואילו הקבוצה השנייה מורכבת מילדי ותיקים שאינם ממוצא אתיופי, האתיופית
בחטיבת ' כיתה ז, הספר היסודיבבית' כתה ד: חולקו לשלוש רמות גיל המקבילות לשלוש כיתות
; הנבדקים התבקשו לכתוב שני טקסטים בשתי סוגות. א בבית הספר התיכון"הביניים וכיתה י
שני הטקסטים עוסקים בבעיות בין . אישי והשני טקסט אקספוזיטורי-האחד טקסט נרטיבי
עמדות , תהנבדקים התבקשו למלא בכתב שאלון שבדק התנסויות אורייניו, בנוסף לכך. אנשים
הוא על התפתחות האוריינות ועל , אם כך, הדגש במחקר זה. הבית- ואוריינות כלפי שפה וכתיבה
באמצעות בחינת יכולות של תלמידים , תרבותי-תוך התייחסות לרקע החברתי, השימוש הלשוני
מחקר זה השווה את , לשם כך. בגיל בית הספר להפיק סוגים שונים של טקסטים כתובים
בחינת משתנים אלה מתייחסת לשלוש . גיל וסוגה, מוצא, תרבותי- רקע כלכלי: הבאיםהמשתנים
:אינטראקציות
אינטראקציה בין הצורות הלשוניות לבין פונקציות השיח ) 1 (
של טקסט לבין שימוש הולם בצורות לשוניות ) מטה-מעלה(אינטראקציה בין ארגון כללי ) 2(
). מעלה-מטה(ומית של בניית טקסט ובאמצעים רטוריים ברמה מק
.הבית לבין יכולות הפקת טקסטים-אינטראקציה בין אוריינות ) 3 (
רקע של הגירה או , המחקר הנוכחי אינו תומך בגישות הרווחות המשוות בהכרח מצב כלכלי קשה
היא ההנחה. השתייכות לקבוצת מיעוט עם הישגים נמוכים בשפה ועם עניין בכתיבה ובקריאה
אקונומיים שונים עשויים להיות - שדווקא ההבדלים בשימוש הלשוני של ילדים מרקעים סוציו
בחינת . קשורים להתנהגויות אורייניות ספציפיות ולאו דווקא לכשירות לשונית כללית
. האוכלוסיות השונות מבוססת על ההנחה שהן יפגינו כישורי אוריינות אחידים ושווים ברמתם
יותר יפגינו אוריינייםשכותבים , מכאן. ה יהיה הן בתוך הקבוצות והן מעבר להןהגיוון בהפק
רגישות גבוהה יותר כלפי הבדלים בין סוגי הטקסטים השונים באמצעות בחירה של צורות
יסתמכו על ,מעבר לאוכלוסיות, הנבדקים, מחד גיסא. לשוניות המתאימות לכל סוג טקסט
קוגניטיביות לבניית -ס משותף של התנסויות חברתיותדפוסים התפתחותיים המבטאים בסי
או היכולת , לדוגמה אסטרטגיות לבניית ארגון כולל של טקסט ושל המידע בתוכו, טקסטים
אוכלוסיות אלה עשויות להציג רמות שונות של , מאידך גיסא. להבחין בין סוגי טקסטים שונים
ובמיוחד בתוכן , ים ולקסיקלייםתחבירי- שימוש בצורות לשוניות ברמה של מבנים מורפו
האוכלוסיות , לכן. מחקר זה תומך ברעיון שכל כותב משקף את התרבות בה הוא חי. התמאטי
בערכים התרבותיים ובעמדות המתארים את השימוש , יובחנו בנורמות הלשוניות המקובלות
.האורייני המיוחד להן
, גדת מגוון של היבטים לשונייםדיסציפלינארית המא-מחקר זה מיישם גישה מחקרית מולטי
המחקר הנוכחי הסתייע הן באמצעי מדידה כמותיים , משום כך. תרבותיים וחינוכיים- חברתיים
, והן באמצעי מדידה איכותיים השכיחים במדעי החברה, המאפיינים את המחקר החינוכי
השתמש בנוסף מחקר זה . ובניתוחים תרבותיים בשילוב ניתוחים מובנים של צורות לשוניות
. מארז הפסוקיות ועד לטקסט השלם, הפסוקית, מרמת המילה: שכבתית של ניתוח-בגישה רב
- מטה"גישה מעין זו מתאימה למחקר האורייני של הטקסט הכתוב מכיוון שהיא משלבת ניתוח
.של המבנה הכולל של הטקסט" מטה-מעלה"לפריטים לקסיקליים עם ניתוח " מעלה
192האחת מכילה ניתוח מפורט של . תייחסות לשתי סדרות של ממצאים תוצאות המחקר מ
אקונומי נמוך בני - טקסטים שנכתבו על ידי שתי קבוצות גיל של תלמידים ישראליים ממיצב סוציו
טקסטים כתובים אלה הושוו לסדרה של . וכאלה שאינם ממוצא אתיופי, העדה האתיופית
ספר - של תלמידי בית , ייכים לקבוצת ביקורת חיצונית השה סוגהמאותו , על אותו נושאטקסטים
הטקסטים . בני אותן קבוצות גיל הלומדים באותן שכבות לימוד, ישראליים מן המעמד הבינוני
& Berman 1996; Berman)לשוני רחב היקף בתנאים דומים -הופקו במסגרת מחקר רב
Verhoeven, 2002) .לשוני- למחקר הרבנית מחקר דומהמחקר זה אימץ תפיסה כללית זהה ותכ .
בדרך זו הושגה אפשרות להשוואה מרבית בין כשירויות שפה דומות ושונות של אוכלוסיות
הסדרה השנייה של הממצאים מתארת התנהגויות אורייניות ועמדות כלפי שפה . מרקעים שונים
קע המידע על משתני הר. וכתיבה העשויות להשפיע על השימוש הלשוני של ילדים מרקעים שונים
הושוו לאלה של והדמוגראפייםאקונומיים -המשתנים הסוציו. נאסף באמצעות שאלון אוריינות
מאותן קבוצות גיל ומאותן שכבות , גבוה-מהמעמד הבינוני, תלמידים בעלי הישגים גבוהים בשפה
.לימוד
- בסביבות של רקע סוציו" להיות אורייני"ניתוח הנתונים מאפשר להגדיר מהי המשמעות של
. ומחדד את ההבנה לגבי התפתחות הכתיבה של ילדים באמצעות יצירת פרופיל, אקונומי נמוך
חשוב במיוחד לגבי אוכלוסיית הילדים בני העדה האתיופית שעלתה לאחרונה , מעין זה, פרופיל
מאחר שעד היום לא קיימים מחקרים מובנים היטב על בסיס בלשני שבדקו את , לישראל
מחקר זה הנו מחקר התפתחותי ראשוני מסוגו . אלה מביעים עצמם בכתבשבהן ילדים, הדרכים
המחקר מספק . הבית של תלמידים ממוצא אתיופי-בישראל על רכישת הכתיבה ועל אוריינות
לחוקרים בתחום הבלשנות החברתית ובתחום החינוך סדרה ראשונה נגישה וממוחשבת של
ספר ממוצא אתיופי - ידי תלמידי בית טקסטים אותנטיים שהופקו בנסיבות מבוקרות על
י "הנתונים הממוחשבים על הטקסטים הכתובים שהופקו ע. ומקבוצות מעוטות יכולת אחרות
התלמידים בני העדה האתיופית מרחיב מאגר קיים אך לא ממוחשב של נתונים על שפתם של
).2003, סלמון; 2000, ארליך(תלמידים ישראלים ממיצב נמוך
אקונומיות נמוכות - מראות שתלמידים עולים ותלמידים ילידים מסביבות סוציותוצאות המחקר
של , הבית-ניתוחים של אוריינות , יתרה מכך. מפגינים יכולות כתיבה דומות מעבר לגיל ולסוגה
התנהגויות אורייניות בית ספריות ושל עמדות כלפי כתיבה אף הם מראים על העדר הבדלים
ההשוואה עם תלמידים מהמעמד הבינוני מראה שהמבנה הנרטיבי . להמובהקים בין אוכלוסיות א
הדחיסות הלשונית ואורך מארז הפסוקיות , ארגון המידע, המבנה האקספוזיטורי הכולל, הקנוני
נראה שיכולות אלה . אקונומיים שונים-גם אם הם מרקעים סוציו, דומים אצל כל הילדים
.את הבסיס להפקת טקסטיםמשקפות דפוסי התפתחות משותפים היוצרים
תלמידים מהמעמד הבינוני , בהשוואה שנערכה במחקר גילינו שכאשר מנטרלים את אורך הטקסט
תופעה המצביעה על מורכבות תחבירית גדולה יותר -מפיקים יותר מלים לפסוקית מעבר לגיל
וכים יותר אקונומי נמוך מפיקים טקסטים אר- תלמידים ממיצב סוציו, לעומתם). 2003 ,רביד(
עקב חוסר ביטחונם ביכולתם להעביר פרגמאטימכשל , כנראה, המכילים פלט לשוני רב הנובע
, שהפקת מלל רב אינה מהווה סימן לכשירות הלשונית של תלמידים ממיצב נמוך, מכאן. משמעות
נמצא , בנוסף לכך. אלא דווקא מצביעה על חוסר האימון שלהם ביכולותיהם התקשורתיות
מכיוון שתלמידים ממיצב נמוך לא רק שהפיקו פחות מלים , רגיש לעושר לקסיקלישמיצב
אלא גם הפגינו שימוש מיומן פחות בבחירה , לפסוקית ופחות מילות תוכן לפסוקית מעבר לסוגה
. במוסכמות שיח ובשמירה על משלב מתאים בהשוואה לחבריהם מהמיצב הבינוני, לקסיקלית
אקונומיים נמוכים לגבוהים - את ההבדלים בין מיצבים סוציולא נראה שיש לשייך , לפיכך
אלא דווקא לרמה של היכולות התקשורתיות , המנטאלייםלהיקפים השונים של הלקסיקונים
.ולמודעות לנורמות מקובלות של כתיבה
ושידע תרבותי , תוצאות המחקר מראות שתרבות היא גורם חשוב התומך בהתפתחות הלשונית
הנמצאים בסוגי הטקסטים השונים , שיר לפרשנות ולהבנה של אלמנטיםמשמש כהקשר ע
-שתי האוכלוסיות מהרקע הסוציו, כל שלוש האוכלוסיות. י מבוגרים וילדים כאחד"המופקים ע
נראה . התמאטימובחנות ביניהן בהשוואה לתוכן , אקונומי הנמוך והן אוכלוסיית המעמד הבינוני
כמחצית , לדוגמה. מות התרבותיות של קהילותיהם בכתיבתםשתלמידים בוחרים להביע את הנור
נושא - " פגיעה בכבודם"מהתלמידים בני העדה האתיופית בחרו לכתוב בנרטיבים שלהם על
.המשקף נורמה תרבותית חשובה בקרב העדה האתיופית
הבית ולעמדות כלפי כתיבה נראה שתלמידים ממעמד גבוה חשופים יותר-בהשוואה לאוריינות
ומגדירים כתיבה כתהליך מופשט המכוון להבעה עצמית , בבתיהם לפעילויות אורייניות מורכבות
אקונומי נמוך חשופים בבתיהם בעיקר -תלמידים ממיצב סוציו, לעומתם. ולהעשרה קוגניטיבית
.ומגדירים כתיבה כמערכת רישום המכוונת להשגת מטרות ברות מימוש, לכתיבת מכתבים
. תרבותי לבין התפתחות האוריינות- היחסים בין מרכיבים של רקע חברתימחקר זה בודק את
במיוחד של טקסטים , המחקר מראה שבכל סביבה ניסיון רחב בהפקת טקסטים מסוגים שונים
, מכאן. הינו הכרחי להתפתחות האוריינות הלשונית של התלמידים בבית הספר, אקספוזיטוריים
ורה כיצד לבנות טקסט אקספוזיטורי מאורגן כהלכה ההשתמעות למורי השפה היא שהוראה בר
. י בני הנעורים"בתחילת התיכון עשויה לעודד טיפוח מיומנויות מעין אלה הנרכשות רק ע
בחירות , לכן. תלמידים בעלי רקע אורייני נמוך יותר מתקשים יותר בפיתוח כשירות תקשורתית
.טקסטמשלביות ולקסיקליות עשויות לשמש כלי אבחון לבשלות ה
יש צורך במחקרים נוספים על השימוש הלשוני של אוכלוסיות שונות על מנת להבין מה משמעות
, וכיצד הטקסטים הכתובים משקפים את הנורמות הייחודיות, בסביבות שונות" להיות אורייני"
. תרבותיות שונות- את העמדות ואת ההתנסויות האורייניות של אנשים מקבוצות חברתיות
אביב- אוניברסיטת תל
ש לסטר וסאלי אנטין"הפקולטה למדעי הרוח ע
ש שירלי ולסלי פורטר"בית הספר למדעי התרבות ע
בקרב ילדים ובני נוער התפתחות יכולות הפקת שיח כתוב
-ילידי ישראל ומהגרים מאתיופיה
תרבותיים-היבטים לשוניים והיבטים חברתיים
" דוקטור לפילוסופיה"חיבור לשם קבלת התואר
מאת
מיכל שלייפר
אביב-הוגש לסנאט של אוניברסיטת תל
2003אוגוסט ג"אב תשס
עבודה זו נעשתה בהדרכת
ר גדי בן עזר"רות ברמן וד' פרופ