Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Awareness of linear and nonlinear morphology in Hebrew:A developmental study
Running head: Morphological awareness
Dorit RavidSchool of Education and the Department of Communications Disorders,
Tel Aviv University
Adina MalenkySchool of Education, Tel Aviv University
Address for correspondence:Dorit RavidSchool of Education, Tel Aviv UniversityTel Aviv 69978, ISRAELTelefax: 972 3 5360394email: [email protected]
Ravid, D. & D. Malenky. Awareness of linear and nonlinear morphology in Hebrew: A developmental study. First Language, 21, 25-56, 2001.
Abstract
The study explores the interface of language typology, universal predispositions,
language awareness, and school instruction through the examination of two morphological
domains in Hebrew: linear formation of stem-and-suffix words, and nonlinear Semitic
formation of root-and-pattern affixation. 100 children, adolescents and adults were
administered five tasks testing awareness of roots, morphological patterns, stems, and
suffixes in inflection and in derivation. Two major findings are reported and analyzed:
awareness of linear constructions emerges earlier than awareness of nonlinear forms; and
stems (roots and word stems) are easier to construe than affixes (morphological patterns and
suffixes). The paper discusses the interaction of language acquisition and use with linguistic
awareness, and the effect of tasks on different degrees of morphological awareness in .
2
1.0 Introduction
This paper investigates the development of morphological awareness in Hebrew from
preschool to adulthood in two morphological domains, aiming to shed light on a number of
developmental and linguistic issues, focusing on the role of typological constraints in
developing sensitivity to language-specific morphological structures.
1.1 Language use and language awareness
Learning a first language is a long and complex route that extends over close to two
decades (Nippold, 1998). There appear to be two modes in which language knowledge is
negotiated by children and adults. One is language use, which employs language as a means
towards carrying out cognitive and communicative goals, and is involved in holistic,
automatic and fast processing of speech and writing. Language use requires the synthesis of
separate systems of linguistic knowledge in order to map meaning onto structure in the
conventional, context-appropriate language-specific form - spoken, written, or thought
(Chafe, 1994).
Throughout the course of linguistic development, language users of different ages and
literacy backgrounds sometimes also consciously access, discuss and verbalize their language
knowledge. In such cases, children and adults evidence metalanguage or language
awareness, the ability to think about language as an object from without (Chaudron, 1983;
Gombert, 1992; Karmiloff-Smith 1986, 1992). This alternative mode treats language as a
formal problem space, focusing analytically on its components as a cognitive goal in its own
right. Metalanguage requires the ability to introspect on the linguistic components that blend
together naturally in language usage - phonemes, morphemes, words, syntactic structures,
and discourse types. Thus it involves an analytical perception of units of language, the ability
to represent on each unit separately, disassociating form from semantic content, and
conscious monitoring of one’s own linguistic knowledge (Bialystok, 1986; Valtin, 1984).
3
The emergence of a self-reflective, “aware” language mode that involves conscious
thought about linguistic units and processes has been noted at various stages of language
development and for various systems (Clark, 1978; Karmiloff-Smith, Grant, Sims, Jones &
Cuckle, 1996; Van Kleeck, 1982). Its role in first language learners is controversial: Marshall
& Morton (1978), for example, take a Piagetian view of language as a system with a
monitoring component necessary for error detection and for self-correction that will lead to
behavioral mastery. But instead of diminishing when the number of errors decreases, as
would be expected of such a system, metalinguistic capacities increase with age and the
growth of linguistic knowledge (Gombert, 1992); and since behavioral mastery of a language
is achieved quite early on, it seems that metalanguage has functions that constitute part of
natural language development (Karmiloff-Smith, 1986, 1992).
Young children display emergent metalinguistic awareness in natural interaction
through spontaneous self-repairs, “practice” sessions, questions and observations about
language (Clark, 1978). Children’s ability to perform structured linguistic tasks such as
inflectional changes in non-natural, experimental contexts implies a rudimentary
metalinguistic capacity (Ravid, 1995a). However, tasks requiring controlled, analytical,
explicit verbalization of linguistic processes and constructs are beyond the capacities of
young children, and may not be fully performed adequately before adolescence (Ashkenazi
& Ravid, 1998; Menn & Stoel-Gammon, 1995). Moreover, metalinguistic insights reflect
different perceptions of language at different ages (Nippold, Uhden & Schwarz, 1997;
Tunmer & Herriman, 1984; Van Kleeck, 1982).
Literacy and schooling have a central role in metalinguistic development. There is
evidence that specific aspects of language awareness, especially phonological and
morphological awareness, both promote and are promoted by learning to read and write
through the establishment of links between phonemes, syllables and morphemes and their
4
written representations (Bentin, 1992; Gillis & de Schutter, 1996; Fowler & Liberman, 1995;
Levin, Ravid & Rappaport, 1999, in press; Rubin, 1988). Abilities requiring more integrated
knowledge such as reading comprehension are also related to analytic metalinguistic skills
(Demont & Gombert, 1996; Yuill, 1998). Sensitivity to more specific language domains,
such as derivational morphology, has been shown to play a role in reading ability in higher
school grades and among college students (Henry, 1993; Mahony, 1994; Smith, 1998).
Metalinguistic development is thus clearly related to the acquisition of literacy and school-
based knowledge.
1.2 Morphological awareness
Morphology is closely related to other linguistic domains. Its phonological facet is
expressed in the structural components of words and the types of formal changes they
undergo in morphophonological operations. Its semantic facet is expressed in the meanings
of words and morphemes, and morphological relations play an important role in the
organization of the mental lexicon (de Jong, Schreuder & Baayen, 2000). Morphology is also
related to syntax through derivational processes expressed in the argument structure of the
sentence and through inflectional marking (Spencer, 1991). Testing morphological awareness
may thus provide a window on other structural and semantic aspects of language and their
interaction (Anglin, 1993).
A number of studies have shown that the ability to judge, segment and extract
morphological units emerges early on, but that explicit explanations using relevant
terminology occur later on. For example, kindergartners are already able to segment bi-
morphemic agentive nouns such as writer and to correct grammatical mistakes such as He
likes to watched movies. However, they find it hard to explain their morphological
corrections (Jones, 1991; Smith-Lock & Rubin, 1993). Ashkenazi & Ravid (1998) show that
5
explicitation in the explanation of morphological riddles and jokes is not fully achieved
before adolescence.
Studies indicate that morphological awareness contributes to success in the beginning
phases of literacy instruction since morphology links together phonological and semantic
facets of language (Brittain, 1970; Carlisle, 1995; Carlisle and Nomanbhoy, 1993; Levin et
al., in press). This is especially evident in studies comparing learning- and reading-disabled
children with normally achieving readers (Webster, 1994). Anglin (1993) provides evidence
that vocabulary growth in gradeschool years increases together with children’s ability to
perform morphological problem solving. In subsequent stages of becoming an efficient
reader, the most important morphological aptitude is the growing ability to segment, extract
and discuss stems and affixes from the multimorphemic vocabulary of the “literate” English
lexicon (Derwing & Baker, 1986; Freyd & Baron, 1982; Lewis & Windsor, 1996; Nagy &
Scott, 1990; Wysocki & Jenkins, 1987). Nippold (1998) shows that during the 9-through-19
age range English speakers’ understanding and use of various word classes increases together
with their ability to define words and their awareness of the lexical network which facilitates
the acquisition of new words.
1.3 Nonlinear and linear structures in Hebrew morphology
Studying the development of morphological awareness in Hebrew presents a unique
challenge of metalinguistic analysis. Hebrew morphology makes use of two major types of
word formation devices: root-and-pattern Semitic forms alongside with concatenated, linear
structures, and thus permits testing contrasts which are not found in non-Semitic languages
(Blau, 1971; Bolozky, 1997).
INSERT TABLE 1 ABOUT HERE
1.3.1 Nonlinear formation
6
The Hebrew lexicon is root-based. All verbs and most nouns and adjectives contain a
tri- or quadri-consonantal core, the Semitic root, which carries the main lexical substance of
the word (Ravid, 1990), shared by other words with the same core meaning (Table 1). This
structural core appears discontinuously in the word, since it is interdigitated by vowels
provided by the complementary structure of the pattern. The combination of root and pattern
into a word is termed nonlinear affixation (Berman, 1987a), and is illustrated in Table 1 in a
set of words related by the root k-l-t ‘take in, absorb’. A number of studies have looked at
readers’ perception of the Hebrew root. Ephratt (1997) tested 4th to 6th graders and adults on
root awareness by having them color letters within the word. She found that they consistently
colored root letters inside the words rather than using other strategies such as coloring initial
or final letter sequences. Berent & Shimron (1997) also found sensitivity to roots in two
experiments on adult Hebrew readers. No developmental study has so far explored speakers’
awareness of the Semitic root in a structured elicitation from kindergarten to adulthood.
While all the words in Table 1 share the consonantal skeleton k-l-t, they differ in the
other component of their structure, the pattern. A pattern is a phonological template
associating a stress pattern with a set of vowels, sometimes with additional affixes (e.g., the
prefix hi- in the verbal pattern hiCCiC, the suffix -an in the nominal agentive pattern
CoCCan). Each pattern is mapped onto a set of root consonants in a different way. For
example, in miklat ‘shelter’ the first two root consonants are preceded by prefix mi- and
followed by the infixed vowel a; whereas in kélet ‘input’ root consonants are separated by
the vowel set é-e with a penultimate stress pattern. The combination of root and pattern is
thus nonlinear and gives the surface form of a Semitic word (McCarthy, 1982). Each of the
components of this template occurs at a different representational tier or plane (McCarthy,
1981), and only their combination results in a possible word. Neither root nor pattern is
pronounceable, nor do they have lexical status as words. This makes them less accessible to
7
speakers than linear segments, which are pronounceable, with stems almost always extant
words.
The syntax and semantics of patterns is comparable to that of linear derivational
suffixes (Lieber, 1981). They have classificatory functions indicating syntactico-semantic
verbal and nominal classes. The seven verbal patterns are termed binyanim (literally,
buildings), and they indicate transitivity values (Berman, 1993). For example, kalat
‘absorbed’ is a transitive verb, while hiklit ‘recorded’ is causative and huklat ‘was recorded’
is its passive counterpart. Nominal patterns, a few dozen in number, indicate ontological
categories (Clark, 1993) such as agent, instrument, place, abstract nominal. For example,
CoCCan is an agent pattern, while maCCeC indicates instruments.
There is evidence that the extraction of patterns may be more difficult and require
even more explicit awareness than roots: Wile the former carry the main lexical substance of
the word and are represented by salient consonants, patterns classify words and are mostly
vocalic. Word-internal vowels, which constitute the major part of the pattern, have been
shown to be less important to Hebrew speakers than consonants (Ravid, 1995a) and may thus
be less linguistically salient. Moreover, being mostly vocalic, patterns are under-represented
in the spelling system (Shimron, 1993). For example, pattern miCCaC is represented only by
the letter M in the written word MQLT miklat ‘shelter’. Hebrew speakers, including teachers,
often claim that they do not “hear vowels” when asked to segment words. Frost, Forster &
Deutsch (1997) studied the lexical representation of Hebrew words by testing adults on
reading non-vowelled Hebrew words. They found that previous exposure to the root letters
(but not to the morphological pattern) facilitated lexical access and naming of targets that
were derivations of the root, that is, evidence for the primacy of root representations over
pattern representations in gaining access to words. Frost et al. conclude that primary
8
morphological analysis in Hebrew entails the extraction of the root and not of the word
pattern. To date, no study has examined roots versus patterns in development.
1.3.2 Linear formation
In addition to root-and-pattern structures, the Hebrew lexicon contains linearly
concatenated stem-and-suffix forms, e.g., mal’ax-i ‘angel-ic’. Table 2 illustrates linear
structures in inflection and derivation.
INSERT TABLE 2 ABOUT HERE
The major difference between linear and nonlinear operations is formal. Linear
structure consists of a pronounceable stem containing vowels, which is in most cases a word
in its own right (e.g., mada ‘science’, kos ‘glass’). Unlike the Semitic root, which is
interdigitated by the pattern, the stem is followed (in some cases, preceded) by a separate
suffix (e.g., agentive –an, diminutive –it). The result is a linear structure, e.g., mada-an
‘scientist’, kos-it ‘wine glass’. Although stems often undergo morphophonological changes
(e.g., layla / leyli ‘night / nocturnal’, cf. English five / fifth), the concatenated stem-and-suffix
form always marks the boundaries of discernibly distinct entities. In contrast, the result of
root and pattern affixation is a fused form where the two components are interdigitated. The
components of linear formation are thus more analytic than those of nonlinear forms: Unlike
the abstract, discontinuous, consonantal Semitic root, linear morphemes are identifiable,
distinct, pronounceable units.
The linear / nonlinear contrast is relevant to the inflectional / derivational dichotomy
in Hebrew. Linear structures occur in both domains, as shown in Table 2, although their
historical origins are different: inflectional suffixes are typically Semitic and practically
unchanged since Biblical Hebrew, while derivational suffixes are more recent and have
increased in number and productivity as a result of language contact with Arabic and with
9
European languages (Berman, 1987a; Gesenius, 1910; Holes, 1995; Nir, 1993). Nonlinear
formation is mostly restricted to derivation, except for tense shifting in verbs (Schwarzwald,
1996). This again leads to the expectation that linear structures should be more accessible to
analysis than nonlinear ones, as inflection is more transparent, productive, predictable and
regular than derivation.
Inflection is widespread in Hebrew and mostly obligatory. Nouns are inflected for
number and gender (e.g., sus / susa / susim / susot ‘horse / mare / horses / mares’), and
adjectives agree with them (e.g., sus gadol ‘horse big = big horse’ / susa gdola ‘mare big =
big mare’). Verbs inflect for tense, and agree with their subjects in number, gender, and (in
past and future tenses), also in person (e.g., hiskamta / hiskamt ‘you, Sg agreed / you, Fm,Sg
agreed’). Prepositions inflect for number, gender and person (e.g., mimex ‘from-
you,Fm,Sg’).
Two knowledge sources are relevant to the development of meta-morphological
perception: natural language acquisition and school instruction.
1.4 Hebrew morphology in acquisition
Grammatical morphological alternations - first number / gender distinctions, then
verb tense and person - initially appear in the speech of normally developing Israeli children
before the end of the second year of life (Dromi, 1987; Kaplan, 1983; Levy, 1980; Ravid,
1997). By the end of their third year, children make productive use of all obligatory
inflectional marking on nouns, verbs and adjectives, though mastery of stem changes and
idiosyncratic forms takes until age 7 at least (Berman, 1981a,b, 1983; Ravid, 1995a,b).
Optional inflectional marking of genitive nouns first appears productively in the early years
of gradeschool (Berman, 1985; Levin et al., in press), and opaque systems such as numerals
are not marked correctly before highschool, and even that takes school intervention (Ravid,
1995c).
10
Hebrew-speaking children display knowledge of word-formation early on, though full
mastery takes until adolescence since the derivational system is less regular and predictable
than the inflectional one, and not as productive (Berman, 1995). The literature describes
differential paths of acquisition for the three major lexical classes - nouns, verbs and
adjectives - and their structural components. Verbs. This is the most structured system in
Hebrew, with all verbs consisting of a combination of root and one of the seven verbal
patterns, termed binyanim. Most early verbs take the semantically and syntactically neutral
verbal pattern P1 (Qal), and by age three basic transitivity relations such as causativity and
incohativity are expressed in the rest of the non-passive binyanim. Though children do not
find novel verb formation, especially denominal verbs, to be easy, by age 4 they are able to
coin semantically appropriate novel verbs from other verbs in a form consistent with the
structural stipulations of their grammar, and showing knowledge of Semitic root-formation in
forming verbs with irregular roots (Berman, 1999). By kindergarten, children are familiar
with a large number of root-related word families, and make productive and mostly correct
use of salient transitivity relations (Berman, 1993). However, more marked and less
transparent operations such as morphological passive formation are still not completely
mastered by 9 years of age.
Nouns are less restricted morphologically than verbs in Hebrew and offer a challenge
of a wide range of structures from non-derived forms, through zero-conversion, to
morphologically complex blends, compounds, linear and nonlinear forms. Children readily
coin novel nouns as early as 3 years and earlier, but they often violate structural and semantic
constraints in doing so (Berman, 1999). Clark & Berman (1984) found that in younger
children the preferred structural option for agent and instrument nouns was the -an suffix,
and they also used zero-conversion from present participles (e.g., xotex ‘cuts’ for ‘cutter’),
whereas older children employed as wider variety of structures. Unlike English-speaking
11
children, the compound option was not a preferred option by Hebrew speakers in the Clark &
Berman study. Free innovation of compounds starts only at age four, and kindergartners still
do not have full command of stem changes in compounds (Berman, 1987b). Sensitivity to the
classic Semitic word-structure, however, is apparent in Hebrew early on. Children as young
as 3 were able to interpret novel nouns, indicating their ability to extract the root from the
given test item (Clark & Berman, 1984). In both a structured elicitation test of deverbal noun
coinage, and in a corpus of nearly one thousand unconventional lexical usages recorded from
children aged 2 to 8 years, Berman (1999) found that the majority of novel noun coinages
took the form of some possible root plus noun-pattern rather than linear stem-and-suffix
forms. In contrast, in a structured task eliciting novel nouns from verbs and nouns in
normally developing 8-10 and 6-8 year olds, as compared to SLI children aged 8-10,
nonlinear root-and-pattern responses were found to increase with age and to be significantly
more prevalent in the older controls (Ravid, Avivi Ben-Zvi & Levie, 1999). Not all types of
nouns are acquired productively early on. While children in all of the previously described
studies find the semantics of agent and instrument easy to express early on, coining action
and state nominals and other abstract nouns starts later, around age 5, and takes until
highschool to consolidate as a productive system. Children initially attach the abstract suffix
-ut to a variety of stems, then move on from the least marked, transitively neutral action
nominal pattern CiCuC to the appropriate action nominal pattern for each binyan and finally,
by adolescence, learn the complex lexical network of deverbal nouns (Berman, 1997; Ravid
& Avidor, 1998).
Adjectives. Morphologically complex adjectives in Hebrew take two main structures:
Nonlinear form, based on present-participle verb forms, e.g., matsxik ‘funny’ (P5), and linear
form, consisting of a nominal stem with the suffix -i, e.g., xashmal-i ‘electr-ic’. In a study of
spontaneous speech in preschoolers, Ravid & Nir (2000) found that nonlinear affixation
12
precedes linear suffixation in adjectives, which occurred only in the oldest group of
kindergartners. This finding is supported by other studies on the morphological development
of adjectives in Hebrew (e.g., Berman, 1994, 1997; Levin et al., in press; Ravid, Levie &
Avivi Ben-Zvi, in press).
A major source of information about morphological structure and semantics in
schoolage children is formal and informal school instruction. Recent studies indicate that
learning about written Hebrew both requires and fosters morphological knowledge, since
Hebrew spelling consistently represents both roots and affixes. Israeli gradeschoolers make
use of these morphological consistencies from early on in their spelling strategies (Gillis &
Ravid, 2000). Exposure to highly synthetic written Hebrew texts encourages young readers to
employ syntactic and morphological cues in analyzing long strings of letters designating
morphemes, e.g., the written string WBMHBRTH pronounced u-ve-maxbart-a ‘and-in-
notebook-hers’. Moreover, formal language instruction in Israel focuses on Hebrew structure,
and especially on its morphology and phonology, from the youngest grades.
1.5 Study hypotheses
It is clear that both linear and nonlinear formation are essential components of
Hebrew morphology, and that they both occur in natural language acquisition in different
morphological classes. Focusing on the development of meta-linguistic awareness, for which
salience and analyticity of structures are crucial (Gombert, 1992), we make two predictions:
(1) Linear precedence. Awareness of linear morphemes will precede that of nonlinear
morphemes, since linear morphemes are easier to perceive and isolate than nonlinear
morphemes which are fused together. (2) Lexical precedence. Awareness of lexical
morphemes - roots and stems, which carry the main lexical substance of the word, will
precede awareness of affixal patterns and linear suffixes, which have categorical functions in
both inflection and derivation.
13
2.0 Method
This study examines the development of morphological awareness in Hebrew
speakers, assessed by an experimental design testing two main types of processes: nonlinear
word-formation by the Semitic combination of consonantal root and affixal pattern in the
derivational domain; and linear concatenation of stem and suffix in the inflectional domain.
2.1 Participants
100 children, adolescents and adults, 20 in each of the five age groups and divided
equally by gender participated in this study: Kindergarten (mean age: 5;5; range: 5;1-5;11).
This was the youngest group that piloting showed could handle the study tasks. By this age,
children are familiar with most of the morphological constructions in Hebrew, though
mastery of exceptions and of literate forms is far in the future. Kindergartners do not receive
formal language instruction, but they are exposed to stories and poems and encouraged to
experiment with writing. 3rd grade (mean age: 8;7; range: 8;2-9) represents middle
childhood. By this age, all children without special needs have reached behavioral mastery in
reading and are well on the way to grasping spelling regularities. They are familiar with
Biblical Hebrew and in many schools have started learning English as a second language. 6th
grade (mean age: 11;7; range: 11;3-13) marks the end of gradeschool. 6th graders, on the
verge of adolescence, have virtually no spelling errors and are familiar with a range of
textual genres. They have been learning English and sometimes another foreign language for
a number of years. 9th grade (mean age:14;5; range:14;1-14;11) represents highschool, a
time when formal, analytic language instruction is at its most demanding, alongside with
reading and writing texts of all types. The oldest group is that of adults, students or college
graduates in subjects other than Hebrew language, literature or linguistics, (mean age: 27;
14
range: 20-39). All participants were native speakers of Hebrew from a middle to high socio-
economic background.
2.2 Procedure
Participants were tested individually and orally at school, in a quiet room. Adults
were tested in their homes. Kindergartners were tested in two separate meetings to reduce test
pressure. Responses were recorded and transcribed by the tester.
2.3 Materials
There were five study tasks, each containing 6 test items, in two sections: I Nonlinear
tasks: 1. Nonce words, 2. Noun patterns, 3. Analogies; II Linear tasks: 4. Backformation,
5. Grammatical judgments. Each task was preceded by 2 training items. Order of the five
tasks, of task sections, and of items within each task was randomized. The full tasks appear
in the Appendix.
2.3.1 Nonlinear formation tasks
Since root-and-pattern word formation constitutes the major lexical device in
Hebrew, we examined participants’ awareness of these two components in three lexical
categories (nouns, adjectives and verbs), by using three different tasks which provide
different ways of looking at how Hebrew speakers construe roots and patterns.
1. Nonce word task. The participant was presented with a set of 6 sentences
containing a nonce word (2 nouns, 2 verbs, and 2 adjectives) constructed of an active root
(occurring in several Hebrew words) and a pattern, again common in Hebrew words. For
example The boy *hidrig `(possible interpretation) graded’ the blocks (nonce word *hidrig
composed of root d-r-g ‘grade’ and causative verb pattern hiCCiC; cf. the extant word dereg
‘ranked’). S/he was then presented with 2 extant words, related to the nonce word
15
morphologically through the root (e.g., madrega `step’ with the same root d-r-g) or through
the pattern (hisbir `explained’ with the same pattern hiCCiC) and asked if the nonce word
was connected to the extant word. Responses were scored as correct if the participant replied
“yes”, identifying a relationship between the two words. The requirement for root and pattern
awareness in this task is thus basic, on the one hand, since it is enough to establish a
connection between two words by saying “yes” to score correct. On the other hand, the fact
that the stimulus word is an unfamiliar nonce word means participants have to focus on root
and pattern alone with no clue provided by lexical meaning of the whole word. Each root and
pattern response received a scoring of 1, so that the maximal points for each response type
was 6.
2. Noun patterns task. This task tested awareness of root and pattern in nouns by
open-ended questions. The participant was asked, for example, why a librarian (Hebrew
safran, root s-p-r ‘tell’, agent noun pattern CaCCan) is called that way. The task contained 6
nouns, each representing an ontological category: Agent, instrument, collective noun, disease
name and abstract nominal (see Appendix). A correct response contained the same root as in
the stimulus word, and a designation of its category (person, disease, etc.). An example of a
fully correct response would be: It’s a man (indicating awareness of the agent noun CaCCan)
working in the library (Hebrew sifriya, containing the root s-p-r `tell’). This task does not
require the analysis of the components of nonce words, but it does direct participants’
attention to the internal structure of a known word, seeking a same-root word to serve as a
focal point in a proposition that would explain the stimulus word. Note that a correct root
score required the actual production of a word sharing the same root, but a correct pattern
score only required a generalization about the pattern meaning – e.g., person, instrument,
disease, etc. Each root and pattern response received a scoring of 1, so that the maximal
points for each response type was 6.
16
3. Analogies task. This task tested awareness of root and pattern indirectly by
eliciting analogies to 6 words - 3 nouns, 2 verbs, one adjective. For example, the participant
was presented with the action nominal ktiva ‘writing’ (root k-t-v ‘write’, pattern CCiCa) and
asked to provide similar words. The training part directed the participants to both root and
pattern responses (see Appendix). A correct response was related to the stimulus word in
either root or pattern. Root responses could be, for example, katav `wrote’ or mixtav `letter’,
and pattern responses could be glisha `surfing’ or sgira `closing’. In this case, both response
types required an analysis of the stimulus word into its components, in addition to actually
seeking words containing one of these components. The maximal score for each response
type was 6.
2.3.2 Linear formation tasks
We examined awareness of linear components in the backformation and in the
grammaticality judgment task, focusing on awareness of the semantics of linear inflectional
affixes.
4. Backformation task. Backformation is a reverse process of deriving a simplex
stem from a complex form by removing a suffix, which occurs naturally in both child
language and language change (Ravid, 1995b). Here, we made use of this morphological
device to see if participants were aware of the relationship between complex inflectional
forms and their stems by asking them to find the word hidden within the stimulus word. This
task contained 6 inflected words with a linear structure, e.g., adumim `red, Pl’ consisting of
singular adom `red’ and the plural suffix -im. Each correct response was given a score of 1,
so that the maximal score was 6.
5. Grammatical judgment task. Participants were presented with 6 sentences with
ungrammatical inflected words and 2 grammatical sentences. They were asked to judge
which sentences were “funny” (for the children) or incorrect (the adults). They were also
17
asked to correct the incorrect items and to explain what was wrong with them. Examples: ani
halaxta `I went, 2nd person’ to the store (cf. correct ani halaxti ‘I went,1st’ or ata halaxta
‘you went,2nd’. Participants’ responses on the grammaticality judgment task were scored 1
point for each identified and corrected structure, a maximal score of 6.
3.0 Results
INSERT TABLE 3 AND FIGURE 1 ABOUT HERE
Looking at the two major morphological devices in Hebrew, we hypothesized linear
precedence: Awareness of linear formation would precede nonlinear structure. In order to
test this hypothesis, we combined the scores of the three nonlinear tasks (nonce words, noun
patterns, and analogies tasks) and the two linear tasks (backformation and grammaticality),
and compared their mean percentages (Table 3). A two-way analysis of morphological task
type (2 - linear / nonlinear) by age (5) with repeated measures showed an effect of age
(F(4,95)=97.34, p<.001), of morphological task (F(1,95)=292.97, p<.001), and an interaction
of age by morphological task type (F(4,95)=6.63, p<.001), depicted in Figure 1. Simple
effects analyses confirmed an effect of age (p’s<.001 for both tasks). The post-hoc Scheffé
procedure revealed that on the linear tasks, kindergartners differed from all other groups
(p< .001); while on the nonlinear tasks, each of the younger groups of kindergartners, 3rd and
6th graders differed from all other groups (p’s<.03 or better).
We now turn to results on the nonlinear task types.
INSERT TABLE 4 ABOUT HERE
3.1 Nonlinear formation
The three nonlinear tasks examined speakers’ awareness of the two major lexical
components of Hebrew: roots and patterns. The three tasks described above tapped
participants’ awareness of root and pattern in different ways, each requiring different degrees
of awareness. The mean percentages of the three tasks are presented in Table 4. A two-way
18
ANOVA (age (5) X nonlinear task type (3)) with repeated measures found an effect for age
(F(4,95)=56.34, p<.001) and for task type (F(2,190)=194.56, p<.001: Analogies (M=84.00)
> Noun patterns (M=74.83) > Nonce Words (M=47.00) The post-hoc Scheffé procedure
revealed that each of the younger groups of kindergartners, 3rd and 6th graders differed from
all other groups (p’s<.03 or better). There was no interaction.
We hypothesized that lexical morphemes would precede categorial elements -
patterns and suffixes - in morphological awareness. A specific form of this prediction within
the nonlinear tasks is root precedence: Roots will precede patterns. In order to test this
prediction, we compared the root and pattern scores on each of the three nonlinear formation
tasks (Table 5).
INSERT TABLE 5 AND FIGURE 2 ABOUT HERE
3.1.1 The nonce words task
Participants were asked a direct question about the stimulus nonce word and an extant
word sharing a root or a pattern with it (e.g., hidrig `graded’ and madrega `step’): “Are these
words connected”? Table 5 presents the mean percentages of the responses indicating root
and pattern connections between words, showing that the root precedence hypothesis was
confirmed. A two-way ANOVA (age (5) X response type (2)) with repeated measures shows
a main effect for age (F(4,95)=29.80, p<.001), for response type (F(4,95)=324.12, p<.001),
and an interaction of age and response type was found (F(4,95)=6.48, p<.001), depicted in
Figure 2. Simple effects analyses confirmed an effect of age (p’s< .001 for both response
types). The post-hoc Scheffé procedure revealed that on root responses, kindergartners
differed from all other groups, and 6th graders differed from the adults (p’s<.005 or better);
while on pattern responses, kindergartners differed only from the two oldest groups, 3rd
graders differed from both oldest groups, while 6 graders differed only from 9th graders
(p’s<.004 or better).
19
Root awareness starts as early as kindergarten and increases steadily, while pattern
awareness stays low, peaking in 9th graders who are studying formal Hebrew grammar, and
declining again in adults.
3.1.2 The noun patterns task
This task tested root and pattern awareness in a more roundabout way. Participants
were asked to explain the names of task nouns, e.g., “Why is a librarian (Hebrew safran)
called that way”?
Table 5 gives the mean percentages of responses indicating awareness of root and
pattern responses. A two-way ANOVA (age (5) x response type (2) with repeated measures
reveals an effect for age (F(4,95)=21.23, p<.001); for response type (F(1,95)=309.80,
p<.001); and an interaction of age and response type (F(4,95)=6.07, p<.001), depicted in
Figure 3. Simple effects analyses confirmed an effect of age (p’s< .001 for both response
types). The post-hoc Scheffé procedure revealed that on root responses, kindergartners
differed from all other groups (p<.001); while on pattern responses, kindergartners differed
only from the three oldest groups, and 3rd graders differed from both oldest groups (p’s<.03
or better).
Again, root responses behave differently from pattern responses: Kindergartners
already give 80% root responses, and from grade 3 onwards, this response type reaches
ceiling. Pattern responses start low, increase slowly with age, and do not exceed 80% in any
of the groups.
INSERT FIGURE 3 ABOUT HERE
3.1.3 The analogies task
Finally, we tested pattern and root awareness by asking participants to produce words
in analogy to stimulus items, Table 5 shows the root precedence hypothesis confirmed once
again. A two-way ANOVA (age (5) X response type (2)) with repeated measures reveals an
20
effect for age (F(4,95)=18.27, p<.01); for response type (F(1,95)=48.68, p<.01); and an
interaction of age and response type (F(4,95)=13.75, p<.01), depicted in Figure 4.
Simple effects analyses confirmed an effect of age only on the pattern response type
(p< .001). The post-hoc Scheffé procedure further revealed that on pattern responses,
kindergartners and 3rd graders differed from the three oldest groups (p’s<.001, except for the
difference between 3rd and 6th graders, which is marginal at .052).
All age groups were able to give root-based responses, but only 25% of the
kindergartners’ responses included a pattern-related word. By grade 9, all participants were
able to do so, yielding the highest overall pattern score of all three nonlinear tasks.
INSERT FIGURE 4 ABOUT HERE
3.2 Linear formation
We have already seen that awareness of linear forms both precedes and is easier than
nonlinear formation (Table 3). The separate mean scores of the two linear tasks are given in
Table 6. The two linear tasks probed different aspects of awareness of linear structures,
focusing on the stem in the backformation task and on the suffix in the grammaticality
judgment task. To test the prediction that stem awareness precedes suffix awareness, we
performed a 2-way ANOVA (age (5) x linear task type (2 - backformation and
grammaticality judgment) with repeated measures, which found an effect for age
(F(4,95)=57.12, p<.001), for linear task type (F(1,95)=52.67, p<.001), and an interaction
between age and task type F(4,95)=22.73, p<.001), depicted in Figure 5. Simple effects
analyses confirmed an effect of age (p’s< .001 for both linear task types). The post-hoc
Scheffé procedure revealed that on backformation, kindergartners differed from all other
groups (p’s<.009 or better); while on grammaticality judgments, kindergartners differed from
all other groups (with p’s<.001). This youngest group starts at around 85% on the
backformation task and at about half that score on the grammaticality task.
21
INSERT TABLE 6 AND FIGURE 5 ABOUT HERE
4.0 Discussion
This study examined the acquisition of morphological awareness in Hebrew, a
morphologically rich language, in 100 native speakers of Hebrew aged five years to
adulthood. Focus was on awareness of two major morphological operations in Hebrew:
nonlinear combination of root and pattern in derivational constructions, tested by three
different tasks; and linear juxtaposition of stem and suffix in inflectional constructions, tested
by two different tasks. These two word-formation devices yield almost all structures in
Hebrew, apart from compounding (Berman, 1987a; Ravid, 1990). The unique characteristics
of Hebrew morphology permitted the testing of two major hypotheses: One, that the
transparency and distinct boundaries of linear, concatenated structure would make it easier to
construe metalinguistically by Hebrew speakers than the interdigitated affixation of root and
pattern; and two, that within each construction type, stems, carrying lexical content, would be
easier to process than affixes which convey categorial value. For nonlinear structures, this
means that roots should be perceived earlier than patterns, and for linear structure, this means
that stems should be perceived earlier than suffixes. Both hypotheses were confirmed.
4.1 Analyzing linear vs. nonlinear morphology
On the tasks described in the current study, Hebrew speakers found it easier to
formally detect the structure of linear structures than that of nonlinear structures. The
combined score on awareness of nonlinear morphology starts below chance level at
kindergarten, proceeds gradually to 6th grade, and the major rise takes place in the
highschoolers (85%), who do not differ from the adults. The combined score on the linear
morphology tasks, in contrast, starts around 2/3 in kindergarten, rises steeply, and by third
grade reaches over 90% (Figure 1).
22
The precedence of linear over nonlinear structures in the language awareness tasks
seems to be in discord with the centrality of root-and-pattern word formation in natural
acquisition: Israeli children manipulate nonlinear structures in all lexical classes by
preschool, as evidenced by the growth and diversification of their vocabulary as well as by
spontaneous coinages.
A number of reasons conspire to help participants on linear tasks here. First, linear
items in this study were all inflectional, while all nonlinear items were derivational.
Inflection, being obligatory, productive, regular, semantically predictable and generally
applicable, is more accessible than derivation, which is semi-productive, nonobligatory,
irregular and non-predictable. In analyzing an inflected word into its components, the
removal of one morphological segment entails the other, even in words with changed stems.
For example, when removing the feminine suffix -a from shvura ‘broken,Fm’, the remaining
part (shavur) will always be ‘broken -Fm’, by default ‘broken,Masc’, and vice versa. In
contrast, removing the pattern or the root from a derivational form does not yield a
predictable form. This problem is not isomorphic with the type of structure that carries
derivational semantics: It is rather difficult to analyze shnaton ‘cohort’ into shana ‘year’ and
suffix -on, and sfaton ‘lipstick’ into safa ‘lip’ and suffix -on.
Moreover, metalanguage crucially differs from implicit language usage by the need
for analytic attention to distinct linguistic units. While holistic language use can draw on
long-term memory of similar forms sharing structural and semantic elements, our tasks
required specific, explicit identification and verbalization of morphological components. This
is easier to do when the boundaries between morphemes are clearly drawn, as in linear
formation, even with stem changes (e.g., adom / adum-a ‘red, Masc / Fm’), then when
morphemes are interdigitated or even fused together, as in mavreg ‘screwdriver’ consisting
23
of root b-r-g and noun pattern maCCeC. Linear stem and suffix components are thus more
transparently perceptible in the word than nonlinear roots and patterns.
In addition to the more transparent nature of linear structure, it is easier to analyze
than nonlinear formation since its components are more salient and less abstract: stems in
Hebrew (unlike English) are almost always words, while roots are never words. In carrying
out a task of isolating a morphological component, focusing on extant, pronounceable,
concrete, “real” elements such as words is easier than detecting an string of unpronounceable
consonants which does not constitute a lexical entry. Furthermore, the meaning of roots is
different from that of words (lexical entries) in being non-specified or, in a sense, potential:
root h-l-x has the generalized meaning of ‘walk’, but takes on various shades of meaning in
hélex ‘wanderer’, tahaluxa ‘parade’, mahalax ‘move’, hilux ‘gait’, tahalix ‘process’ and
halaxa ‘religious law’.
4.2 Isolating stems and affixes
Figures 2-5 indicate that as predicted, stems are more accessible than affixes within
words. Figures 2-4 show that for each nonlinear task, root scores are high even in
preschoolers, while pattern scores are lower, especially in kindergarten and 3rd grade. Figure
5 shows that the grammaticality judgment task, which required the isolation and explanation
of inflectional suffixes, was more difficult for preschoolers than the backformation task,
which demanded that they isolate the base word.
Taken together, these results show that stems, the basic lexical building blocks of
words, are more accessible to metalinguistic analysis than affixes, which are categorizing
elements, especially in the younger age groups. This has already been shown for English by
Anglin (1993) in discussing morphological problem solving strategies among gradeschool
children presented with inflected or derived words and asked to provide their meanings. Most
subjects manifested a “part to whole” pattern in which they discussed the stem embedded
24
within the target word - either a root word or an embedded derived word. Only the oldest
children were occasionally able to relate to the affixes as well as to the stems (1993:116). In
the current study, however, the unique characteristics of Hebrew morphology enable us to
perceive further differences between stems and affixes in inflected linear constructions in
contrast to derived nonlinear forms.
The difference between the accessibility of stems and affixes in linear constructions is
valid only in the youngest age group, the kindergartners. Beyond that age group, stems and
affixes were equally detectable by our subjects. However, root perception emerges very early
on, mitigated by the difficulty of the task, while pattern perception continues to challenge
older age groups. On the easier analogies and noun patterns tasks, root responses were very
high even in kindergarten, compared with much lower pattern responses, which showed
gradual rise until they converged in highschool. On the more difficult nonce words task,
pattern responses were almost nonexistent until the end of gradeschool. Thus, despite the
discontinuous, consonantal-only, nature of the Semitic root, it is easily identified as a
morphological unit early on. The complementary structure, the pattern, is a very late
development in metalinguistic abilities.
This dichotomy reflects the differential status of roots and patterns, consonants and
vowels, in Hebrew. Pattern vowels are less salient than root consonants in Hebrew, and carry
less semantic information (Ravid, 1995a). Many words in Hebrew have double and triple
vowelling options in spoken Hebrew. For example, the word for cut is xatax, but has two
other colloquial variants - xétex and xétax. The vowel pattern changes, but the root
consonants stay the same. Given the lexical vs. categorial function of roots and patterns, as
well as the under-representation of vowels in the Hebrew writing system, it is clear why the
representation of roots as a lexical prime is so stable even in preschoolers, whereas that of
patterns is so precarious even in adults. An exceptional response by one preschooler
25
demonstrates this issue. Asked if *le-hadrig ‘to-grade’ and le-hasbir ‘to-explain’ were
related, he said yes, and explained that ‘if you whisper both of these words and listen to them
from a far, they sound the same”. And he demonstrated this by whispering the two words,
eliminating the sound of the consonants to emphasize the similarity in the vowel patterning
of the two words
4.2.1 Task-related differences in nonlinear formation
While all three nonlinear tasks show the same patterns of early root and late pattern
perception, they differ in degree of difficulty due to the degree of analyticity and abstractness
required in carrying them out, and to whether the task was one of comprehension or
production.
The hardest task was the nonce words, where participants were asked to state whether
a nonce word was related in either root or pattern to extant words. This was essentially a
comprehension or judgment task, but it proved to be the hardest. Counter to what might be
expected, inquiring about a possible relationship between words did not result in a flood of
“yes” responses, and in fact on patterns there were more negative than positive responses
across the board. This is because this judgment test did not provide participants with a
familiar lexical anchor, while requiring them to analyze the nonce word into its components,
holding both nonce and target words in short-term phonological memory in order to detect a
discontinuous phonological sequence in both of them. This task thus required a high degree
of morphological analyticity in order to abstract interdigitated morphemes away from the
word.
The analogies task was the easiest of all three tasks. This was a production test,
requiring the retrieval of any two isolated words, one sharing a root and another sharing a
pattern with the stimulus word. The examples preceding the test itself required participants to
focus directly on the two discontinuous elements and retrieve words containing them with no
26
context provided. Unlike the nonce words task, target words were provided by the participant
rather than by the investigator, requiring active lexical manipulation of the “morphological
family” (de Jong, Schreuder & Baayen, 2000); but there was no need to abstract away the
morphological components of the word, rendering the task less abstract and analytic. In the
younger age groups, participants were mostly able to activate root-related words, and it was
only from 6th grade onwards that pattern-related words were also retrieved.
Another easy task was the noun patterns, where participants were asked to explain
why a certain referent was called that way. This task in fact required the production of a
proposition with two elements – a word containing the target root and a pronoun or a full NP
paraphrasing the categorical function of the pattern, e.g., because he (= agent noun) works in
a library (= target root); or because they cure (= target root) people there (= place noun).
Producing a spontaneous utterance which provides a context for real words functioning in
their natural syntactic sites is holistic and natural, and does not require as much analysis as
identifying and relating word components in the nonce words task. Moreover, this task was
less abstract since did not require specific attention to each of the nonlinear morphemes, but
rather elicited them indirectly by asking to motivate the word.
27
References
Anglin, J.M. (1993). Vocabulary development: a morphological analysis.
Monographs of the Society for Research in Child Development 58, 10.
Ashkenazi, O. & D. Ravid. (1998). Children’s understanding of linguistic humor: an
aspect of metalinguistic awareness. Current Psychology of Cognition 17, 367- 387.
Bentin, S. (1992). Phonological awareness, reading, and reading acquisition: a survey
and appraisal of current knowledge. In Katz, L. & R. Frost (Eds.) Orthography, phonology,
morphology and meaning. Amsterdam: Elsevier Science Publishers.
Berent, I. & J. Shimron. (1997). The representation of Hebrew words: evidence from
the contour principle. Cognition 64, 39-72.
Berman, R.A. (1981a). Regularity vs. anomaly: The acquisition of inflectional
morphology. Journal of Child Language 8, 265-282.
Berman, R.A. (1981b). Children’s regularizations of plural forms in the Hebrew noun
system. Papers and Reports on Child Language Development 20, 34-44.
Berman, R.A. (1983). Establishing a schema: Children’s construal of verb-tense
marking. Language Sciences 5, 61-68.
Berman, R.A. (1985). Acquisition of Hebrew. In D.I. Slobin (ed.) The crosslinguistic
study of language acquisition, Vol I. Hillsdale, NJ: Erlbaum, 255-371.
Berman, R.A. (1987a). Productivity in the lexicon: new-word formation in Modern
Hebrew. Folia Linguistica 21, 225-254.
Berman, R.A. (1987b). A developmental route: leaning about the form and use of
complex nominals in Hebrew. Linguistics 25, 1057-1085.
Berman, R.A. (1993). Developmental perspectives on transitivity: A confluence of
cues. In Y. Levy (ed.) Other children, other languages: Issues in the theory of acquisition.
Hillsdale, NJ: Erlbaum, 189-241.
28
Berman, R.A. (1994). Formal, lexical, and semantic factors in the acquisition of
Hebrew resultative participles. In S. Gahl, A. Dolbey, & C. Johnson (eds.) Berkeley
Linguistic Society, 20, 82-92.
Berman, R.A. (1995). Word-formation as evidence. In D. McLaughlin & S. McEwen
(eds.) Proceedings of the 19th Annual Boston University Conference on Language
Development, Vol I. Somerville, Mass: Cascadilla Press, 82-95.
Berman, R.A. (1997). Preschool knowledge of language: what five year olds know
about language structure and language use. In C. Pontecorvo (ed.) Writing development: an
interdisciplinary view. Amsterdam: John Benjamins, 61-76.
Berman, R.A. (1999). Children’s innovative verbs versus nouns: Structured
elicitations and spontaneous coinages. In L. Menn & N. Bernstein-Ratner, eds. Methods in
studying language production. Mahhwah, NJ: Erlbaum, 69-93
Bialystok, E. (1986). Factors in the growth of linguistic awareness. Child
Development 57, 498-510.
Blau, Y. (1971). Biblical Hebrew phonology and morphology. Tel Aviv: Hakibbutz
Hameuchad. [in Hebrew]
Bolozky, S. (1997). Israeli Hebrew phonology. In A.S. Kaye (ed.) Phonologies of
Asia and Africa. New York: Eisenbrauns, 287-311.
Brittain, M.M. (1970). Inflectional performance and early reading achievement.
Reading Research Quarterly 6, 34-48.
Carlisle, J. (1995). Morphological awareness and early reading achievement. In L.B.
Feldman (Ed.) Morphological aspects and early reading achievement. Hillsdale, NJ:
Erlbaum. Carlisle, J. & Nomanbhoy, D. (1993). Phonological and morphological
awareness in first graders. Applied Psycholinguistics 14, 177-195.
29
Chaudron, C. (1983). Research on metalinguistic judgments: a review of theory,
methods, and results. Language learning 33, 343-377.
Chafe, W. (1994). Discourse, consciousness and Time. Chicago: University of
Chicago Press.
Clark, E.V. (1978). Awareness of language: some evidence from what children say
and do. In A. Sinclair, R. Jarvella, & W. Levelt (eds.) The child’s conception of language.
New York: Springer, 17-43.
Clark, E.V. (1993). The lexicon in acquisition. Cambridge: Cambridge University
Press.
Clark, E.V. & R.A. Berman. (1984). Structure and use in the acquisition of word-
formation. Language 60, 542-590.
Demont, E. & Gombert, J-E. (1996). Phonological awareness as a predictor of
recoding skills and syntactic awareness as a predictor of comprehension skills. British
Journal of Educational Psychology 66, 315-332.
Derwing, B. & Baker, W. (1986). Assessing morphological development. In P.
Fletcher & M. Garman (Eds.) Language acquisition (2nd edition). Cambridge: Cambridge
University Press.
Dromi, E. (1987). Early lexical development. Cambridge: Cambridge University
Press.
Ephratt, M. (1997). The psycholinguistic status of the root in Modern Hebrew. Folia
Linguistica 31, 77-103.
Fowler, A.E. & Liberman, I.Y. (1995). The role of phonology and orthography in
morphological awareness. In L.B. Feldman (Ed.) Morphological Aspects of Language
Processing. Hillsdale, NJ: Erlbaum, 157-188.
30
Freyd, P. & Baron, J. (1982). Individual differences in the acquisition of derivational
morphology. Journal of Verbal Learning and Verbal Behavior 21, 282-295.
Gesenius. (1910). Gesenius' Hebrew Grammar, edited by E. Kautzsch, revised by
A.E. Cowley. Oxford: Clarendon Press.
Gillis, S. & D. Ravid. (2000). Effects of phonology and morphology in children’s
orthographic systems: a cross-linguistic study of Hebrew and Dutch. In E. Clark (ed.) The
Proceedings of the 30th Annual Child Language Research Forum. Stanford: Center for the
Study of language and Information, 203-210.
Gillis, S. & G. de Schutter. (1996). Intuitive syllabification: universals and language
specific constraints. Journal of Child Language 23, 487-514.
Gombert, J.E. (1992). Metalinguistic development. Translated by T. Pownall. New
York: Harvester Wheatsheaf.
Henry, M.K. (1993). Morphological structure: Latin and Greek roots and affixes as
upper grade code strategies. Reading and Writing 5, 227-241.
Holes, C. (1995). Modern Arabic: structures, functions, and varieties. London:
Longman.
Jones, K.N. (1991). Development of morphophonemic segments in children’s mental
representations of words. Applied Psycholinguistics 12, 217-239.
de Jong, N.H., R. Schreuder, & R.H. Baayen. (2000). The morphological family size
effect and morphology. Language and Cognitive Processes 15, 329-366.
Karmiloff-Smith, A. (1986). From meta-processes to conscious access: evidence from
metalinguistic and repair data. Cognition 23, 95-147.
Karmiloff-Smith, A. (1992). Beyond modularity: a developmental perspective of
cognitive science. Cambridge, MA: MIT Press.
31
Karmiloff-Smith, A., Grant, J., Sims, K., Jones, M.C., & Cuckle, P. (1996).
Rethinking metalinguistic awareness: representing and accessing knowledge about what
counts as a word. Cognition 58, 197-219.
Kaplan, D. (1983). Order of acquisition of morpho-syntactic elements in Hebrew
speaking nursery schoolers. Master’s thesis, Tel Aviv University. [in Hebrew]
Levin I., Ravid, D., & Rappaport, S. (1999). Developing morphological awareness
and learning to write: A two-way street. In T. Nunes (Ed.) Learning to read: An Integrated
view from research and practice. Amsterdam: Kluwer, 77-104. 1999
Levin I., Ravid, D., & Rappaport, S. (In press). Morphology and spelling among
Hebrew-speaking children: From kindergarten to first grade. Journal of Child Language.
Levy, Y. (1980). The acquisition of gender. Unpublished doctoral dissertation,
Hebrew University in Jerusalem. [in Hebrew]
Lewis, D. & Windsor, J. (1996). Children’s analysis of derivational suffix meanings.
Journal of Speech and Hearing Research 39, 209-216.
Lieber, R. (1981). Morphological conversion within a restricted theory of the lexicon.
In M. Moortgat, H. van der Hulst, and T. Hoekstra (eds.) The scope of lexical rules.
Dordrecht: Foris.
Mahony, D.L. (1994). Using sensitivity to word structure to explain variance in high
school and college reading ability. Reading and Writing 6, 19-44.
McCarthy, J. (1981). A prosodic theory of non-concatenative morphology. Linguistic
Inquiry 12, 373-418.
McCarthy, J. (1982). Prosodic templates, morphemic templates, and morphemic tiers.
In H. van der Hulst & Smith, N. (Eds.) The structure of phonological representation, Part I.
Dordrecht: Foris.
32
Marshall, J.C. & Morton, J. (1978). On the mechanics of EMMA. In A. Sinclair, R.J.
Jarvella & W.J.M. Levelt (Eds.) The child’s conception of language. Berlin: Springer-
Verlag.
Menn, L. & Stoel-Gammon, C. (1995). Phonological development. In P. Fletcher &
B. MacWhinney (Eds.) The handbook of child language. Oxford: Blackwell, 335-359.
Nagy, W.E. & Scott, J.A. (1990). Word schemas: expectations about the form and
meaning of new words. Cognition and instruction 7, 105-127.
Nir, R. (1993). Lexical devices in Modern Hebrew. Tel Aviv: The Open University of
Israel. [in Hebrew]
Nippold, M.A. (1998). Later Language development: The school-age and adolescent
years. 2nd Edition. Austin, Texas: PRO-ED.
Nippold, M.A, Uhden, L.D. & Schwarz, I.E. (1997). Proverb explanation through the
lifespan: A study of adolescence and adults. Journal of Speech, Language, and Hearing
Research 40, 245-253.
Ravid, D. (1990). Internal structure constraints on new-word formation devices in
Modern Hebrew. Folia Linguistica 24, 289-346.
Ravid, D. (1995a). Language change in child and adult Hebrew: A psycholinguistic
perspective. New York: Oxford University Press.
Ravid, D. (1995b). The acquisition of morphological junctions in Modern Hebrew:
the interface of rule and rote. In H. Pishwa & K. Maroldt (Eds.) The development of
morphological systematicity: A Cross-linguistic perspective. Tubingen: Gunter Narr, 55-77.
Ravid, D. (1995c). Neutralization of gender distinctions in Modern Hebrew numerals.
Language Variation and Change 7, 79-100.
Ravid, D. (1996). Accessing the mental lexicon: Evidence from incompatibility
between representation of spoken and written morphology. Linguistics 34, 1219-1246.
33
Ravid, D. (1997). Morphological development a duo: pre- and proto-morphology in
the language of Hebrew-speaking twins. Papers and Studies in Contrastive Linguistics 33,
79-102.
Ravid, D. & A. Avidor (1998). Acquisition of derived nominals in Hebrew:
developmental and linguistic principles. Journal of Child Language 25, 229-266.
Ravid, D., G. Avivi Ben- Zvi & R. Levie. (1999). Derivational morphology in SLI
children: structure and semantics of Hebrew nouns. In M. Perkins & S. Howard (Eds.) New
directions in language development and disorders. New York: Plenum, 39-49.
Ravid, D., R. Levie & G. Avivi-Ben Zvi. (In press) Hebrew adjectives in language-
impaired and typically developing gradeschoolers. In L. Verhoeven (ed.).
Ravid, D. & Nir, M. (2000). On the development of the category of Adjective in
Hebrew. In In M. Beers, B. van den Bogaerde, G. Bol, J. de Jong, & C. Rooijmans (eds.)
From sound to sentence: Studies on first language acquisition. Groningen: Center for
Language and Cognition, 113-124.
Rubin, H. 1988. Morphological knowledge and early writing ability. Language and
Speech 31, 337-355.
Schwarzwald (Rodrigue), O. (1996), Syllable structure, alternations and
verb complexity: The Modern Hebrew verb patterns reexamined. Israel Oriental Studies 16,
99-112.
Shimron, Joseph. (1993). The role of vowels in reading: A review of studies of
English and Hebrew. Psychological Bulletin 114, 52-67.
Smith, M.L. (1998). Sense and sensitivity: An investigation into fifth grade children’s
knowledge of English derivational morphology and its relationship to vocabulary and reading
ability. Unpublished doctoral dissertation, Graduate School of Education, Harvard
University.
34
Smith-Lock, K. & Rubin, H. (1993). Phonological and morphological analysis skills
in young children. Journal of Child Language 20, 437-454.
Spencer, A. (1991). Morphological theory. Oxford, UK: Blackwell.
Tunmer, W.E., &. Herriman, M.L. (1984). The development of metalinguistic
awareness: a conceptual overview. In W.E Tunmer, C. Pratt & M.L. Herriman (Eds.)
Metalinguistic awareness in children. Berlin: Springer-Verlag.
Webster, P.E. (1994). Linguistic factors in reading disability: a model for assessing
children who are without overt language impairment. Child Language Teaching and Therapy
10, 259-281.
Wysocki, K. & Jenkins, J.R. (1987). Deriving word meanings through morphological
generalization. Reading Research Quarterly 22, 66-81.
Valtin, R. (1984). The development of metalinguistic abilities in children learning to
read and write. In J. Downing & R. Valtin (Eds.) Language awareness and learning to read.
New York: Springer-Verlag, 207-226.
Van Kleeck, A. (1982). The emergence of linguistic awareness: a cognitive
framework. Merrill-Palmer Quarterly 28, 237-265.
Yuill, N. (1998). Reading and riddling: the role of riddle appreciation in
understanding and improving poor text comprehension in children. Current Psychology of
Cognition 17, 313-342.
35
Word Gloss Pattern Pattern meaning
kalat absorbed CaCaC Basic verb, P1
niklat was absorbed niCCaC Passive verb, P2
hiklit recorded hiCCiC Causative verb, P5
huklat was recorded huCCaC Passive verb, P6
Word Gloss Pattern Pattern meaning
klita absorption CCiCa Action nominal
haklata recording haCCaCa Action nominal
kélet1 input CéCeC Abstract noun
taklit record taCCiC Derived nominal
maklet receiver maCCeC Instrument noun
miklat shelter miCCaC Place noun
koltan receptor CoCCan Agent noun
kalit absorbable CaCiC -Able adjective
Table 1. Words related by root k-l-t ‘take in, absorb’
36
Derived Word Gloss Stem and
Derivational Suffix
Stem and Suffix meaning
mada’an scientist mada-an science-agent
pa’oton nursery school pa’ot-on toddler-collective
kosit wine glass kos-it glass-diminutive
ma’afiya bakery ma’afe-iya baked product-place
lamdanut scholarship lamdan-ut scholar-abstract
Inflected Word Gloss Stem and
Inflectional Suffix
Stem and Suffix meaning
nafalti I fell nafal-ti fell-1st,Sg
hidpisu they printed hidpis-u printed-Pl
kosot glasses kos-ot glass-Pl,Fm
susa mare sus-a horse-Fm
armonénu our palace armon-énu palace-1st,Pl
Table 2. Linear structures in Hebrew
1 Stress is not marked unless penultimate.
37
Test groups Study tasks
Nonlinear Tasks Linear Tasks
1. K Mean
SD
45.32
7.0
64.67
16.63
2. G3 Mean
SD
60.60
13.43
91.49
5.60
3. G6 Mean
SD
70.83
10.31
93.98
4.32
4. G9 Mean
SD
84.84
7.0
97.97
2.36
5. Adults Mean
SD
81.46
8.73
97.52
2.75
Table 3. Combined mean percentages and standard deviations of participants’ scores on the
three nonlinear vs. the two linear tasks, by age group.
38
Test groups Study tasks
Nonce words Noun patterns Analogies
1. K Mean
SD
25.42
6.88
50.56
15.80
60.00
20.52
2. G3 Mean
SD
39.58
10.86
69.72
15.13
72.50
25.52
3. G6 Mean
SD
42.5
13.29
80.00
7.52
90.42
20.52
4. G9 Mean
SD
65.63
15.99
88.89
10.66
100.00
00
5. Adults Mean
SD
61.88
18.25
85.00
9.72
97.50
11.18
Table 4. Mean percentages and standard deviations of participants’ scores on the three
nonlinear study tasks, by age group
39
Age groups Nonce words Noun patterns Analogies
Root Pattern Root Pattern Root Pattern
Kindergarten Mean
SD
42.08
11.93
8.75
11.30
80.00
15.86
40.83
20.57
95.00
22.36
25.00
44.43
Grade 3 Mean
SD
72.50
20.06
6.67
10.33
96.67
11.60
55.00
19.57
100.00 45.00
51.04
Grade 6 Mean
SD
66.25
16.33
18.75
17.90
98.33
5.13
63.33
11.60
100.00 80.00
41.04
Grade 9 Mean
SD
80.83
19.52
50.42
26.14
98.33
5.13
79.17
16.11
100.00 100.00
Adults Mean
SD
88.33
16.75
35.42
27.95
97.50
6.11
74.17
17.50
95.00
22.36
100.00
Table 5. Mean percentages and standard deviations of participants’ root and pattern scores
on the three nonlinear study tasks, by age group
40
.Age groups Linear Tasks
Backformation
Mean SD
Grammaticality
Mean SD
Kindergarten 85.28 18.76 44.06 27.01
Grade 3 96.11 4.07 86.88 10.32
Grade 6 98.89 2.28 89.06 8.33
Grade 9 97.50 3.81 98.44 3.99
Adults 97.22 3.72 97.81 3.66
Table 6: Mean percentages and standard deviations of participants’ scores on the linear
tasks: Backformation and Grammaticality Judgments
41
Appendix: The five morphological awareness tasks
I Nonlinear tasks
1. Nonce words
Training: We have made up words that do not exist, but they are connected to other words
that you surely know. For example, listen to this sentence: We need to *shelf (Hebrew *le-
madef) the room. The word *le-madef is connected to a shelf (Hebrew madaf), right?
Because to*-shelf means to put up a shelf. [root relationship] And the word *le-madef is also
connected to the word le-lamed ‘to-teach’, because *le-madef is to do something, like le-
lamed [pattern relationship]
We have made up another word that does not exist, but it is connected to other words that
you surely know. Please listen to this sentence: This man is a *thrower (Hebrew *zarkan).
The word *zarkan is connected to throwing (Hebrew li-zrok), right? Because a* thrower
throws things [root relationship]. And the word *zarkan is also connected to the word baxyan
‘crybaby’ because like baxyan, *zarkan is someone who does something all the time [pattern
relationship].
Task: Now we’ll ask you about some other words that we have made up that are connected to
other words, and you’ll help us to find the connection. If participant answers ‘yes’, s/he is
asked: what is the connection?
Test Items
1. The family went *le-hitpaknek ‘to have a picnic’ in the wood (cf. conventional la-asot
piknik ‘to-make a picnic’).
Is the word *le-hitpaknek connected to...
(i) piknik ‘picnic’ (root relationship); (ii) le-hitxaten ‘to get-married’ (pattern relationship)
2. The child tried *le-hadrig ‘to grade’ the blocks (cf. conventional le-dareg).
Is the word le-hadrig connected to...
42
(i) madrega ‘step’ (root relationship); (ii) le-hasbir ‘to explain’ (pattern relationship)
3. The *mikpal ‘fold’ of the sleeve has unfolded (cf. conventional kיfel).
Is the word *mikpal connected to...
(i) hitkapel ‘folded’ (root relationship); (ii) mitbax ‘kitchen’ (pattern relationship).
4. The doctor examined the child with the *maksheva ‘listener (= stethoscope)’
Is the word *maksheva connected to...
(i) hikshiv ‘listened’ (root relationship); (ii) matslema ‘camera’ (pattern relationship)
5. This wall is *metsuba ‘colored’ white (cf. conventional tsavúa)
Is the word *metsuba connected to...
(i) tséva ‘color’ (root relationship); (ii) meshulash ‘triangle’ (pattern relationship)
5. This man is very hexleti ‘decisive’ and he always knows what is wants to do2.
Is the word hexleti connected to...
(i) hexlit ‘decided’ (root relationship); (ii) xagigi ‘festive’ (suffix relationship).
2. Noun patterns
Training: 1. Isn’t it right that a diving board (Hebrew makpetsa) is called makpetsa because
it is a thing (Hebrew maxshir ‘instrument’) that you jump (Hebrew kofets) from? [instrument
pattern]
2. And isn’t it right that a thief (Hebrew ganav) is called ganav because it’s a person that
steals (Hebrew gonev) things? [agent pattern]
Task: Now, you tell me please why ...
Test items
1. Why is safran ‘librarian’ called safran? [Agent noun]
2. Why is mavreg ‘screwdriver’ called mavreg? [Instrument noun]
3. Why is mirpa’a ‘clinic’ called mirpa’a [Place noun]
2 hexleti ‘decisive’ is an extant word, but -i suffixed adjectives are a later acquisition in Hebrew (Ravid & Nir, 2000; Levin et al, in press).
43
4. Why is tikun ‘fixing’ called tikun [Action nominal]
5. Why is tsahévet ‘jaundice’ called tsahévet [Disease-name noun]
6. Why is ףrex ‘length’ called ףrex [Abstract noun]
3. Analogies
Training: 1. The word sidewalk (Hebrew midraxa) resembles the word they step (Hebrew
dorxim) [root relationship], and also resembles the word fountain (Hebrew mizraka) [pattern
relationship], right?
2. And the word broke (Hebrew shavar) resembles the word broken (Hebrew shavur) [root
relationship], and also resembles the word wrote (Hebrew katav) [pattern relationship], right?
Task: Now tell me please words that resemble...
Test items
1. shakran ‘liar’;
2. mesaxek ‘playing’
3. ktiva ‘writing,N’
4. mutslax ‘good, successful’
5. hitnagev ‘toweled oneself’
6. dayar ‘tenant’.
II Linear tasks
4. Backformation
Training:
1. There’s a word that’s hiding inside the word tigress (Hebrew nemera). It’s the word tiger
(Hebrew namer), right? [gender inflection]
2. And there’s a word hiding inside the word they coughed (Hebrew hishta’alu). What is it?
It’s the word he coughed (Hebrew hishta’el), right? [number inflection]
44
Task: Now here are some other words with words hiding within them. Please tell me which
word is hiding within every word I’ll tell you, ok?
Test items - Inflected words:
1. shvura ‘broken,Fm’ [Masc shavur]
2. dapim ‘pages’ [Sg daf]
3. zamיret ‘singer, Fm’ [Masc zamar]
4. tslalim ‘shadows’ [Sg tsel]
5. mapiyot ‘napkins, Fm’ [Sg mapit]
6. adumim ‘red,Pl’ [Sg adom]
5. Grammaticality judgment
Training: There’s a doll here that sometimes talks funny, she can’t speak very well. Please
help us teach the doll how to speak well / correctly.
The doll says: 1. I wore a shirt (Fm) blue (Hebrew kaxol Masc). She talked funny, didn’t
she? We don’t say shirt kaxol, we say shirt kxula, because shirt is a girl (feminine gender),
right?
The doll says: 2. My phone is ringing now. She talked well, correctly, didn’t she?
Task: Now please listen to the doll and try to find out when she talks funny and makes
mistakes. Did she speak correctly or did she talk funny? Can you correct her so she speaks
correctly? Why did you have to correct the doll, what did she say that was funny?
Test items
1. I bought three doll. [noun number mismatch]
2. I have a small, Fm dog,Masc. [noun / adjective gender mismatch]
3. The drawer in my desk closed. [correct sentence]
4. I have a wound (Hebrew pיtsa) and another wound here. I have two ptsa’ot. (incorrect
form; cf. correct: ptsa’im)
45
5. The boy has a black,Fm TV remote control, Masc. [noun / adjective gender mismatch]
6. This girl drew,Pl a drawing for me. [noun / verb number mismatch]
7. My mummy bought me a teddy bear tomorrow [verb tense / temporal adverb mismatch]
8. Next week we’ll go to the beach. [correct sentence]
9. We will speak to you yesterday. [verb tense / temporal adverb mismatch]
46
0
20
40
60
80
100
K G3 G6 G9 Adults
NonlinearTasksLinear Tasks
Figure 1: Interaction of age group by morphological task type: Nonlinear and linear tasks
47
0102030405060708090
K G3 G6 G9 Adults
Root Pattern
Figure 2: The Nonce words task: Interaction of age group by root and pattern response type
48
0102030405060708090
100
K G3 G6 G9 Ads.
RootPattern
Figure 3: The Noun patterns task: Interaction of age group by root and pattern response type
49
0
20
40
60
80
100
K G3 G6 G9 Ads.
Root Patterns
Figure 4: The Analogies task: Interaction of age group by root and pattern responses
50
0
20
40
60
80
100
K G3 G6 G9 Ads
Backformation Grammaticality
Figure 5: Linear tasks: Interaction of age group by task type - Backformation and
Grammaticality
51
Notes
52