Adjectives in Text Production

January 28, 2006, Antwerp
March 26, 2006, Home

April 9, 2006, Bangor
May 15, 2006, Home
Adjectives in later language development:
text-embedded analyses
Dorit Ravid and Ronit Levie
Tel Aviv University
Ms in progress
The category of adjectives provides an interesting window on processes of lexical
acquisition in later language acquisition, i.e., the school years. Adjectives constitute a less
primary content-word class in both typological and psycholinguistic perspectives, and
emerge later on in language acquisition than the primary content-word classes of nouns
and verbs. Their distribution in Hebrew texts produced by children, adolescents and
adults offers rich insights on the consolidation of a lexical category in the literate lexicon
and on its interface with semantic, pragmatic, syntactic and discursive factors.
1.0 INTRODUCTION
The three well-known content-word (or open class) categories in the world’s
languages are nouns, verbs, and adjectives. These categories differ in their degree of
universality and proto-typicality, as defined by a number of criteria. Semantically, a
lexical category is characterized by the concepts it refers to, and syntactically, by the
syntactic functions it fulfills. It has also been proposed that lexical categories have
discourse roles, and that the prototypical status of category members depends to what
extent they introduce participants or events into the discourse (Hopper & Thompson,
1984). Finally, a lexical class may also have language-specific morphological
characteristics. Typical members of the class of lexical categories fulfill these
requirements. Thus the two basic lexical categories that participate in “making up” a
language either onto- or phylo-genetically are nouns and verbs. Although these two
categories contain more and less typical members (e.g., concrete vs. abstract nouns,
dynamic vs. state verbs), they are both primary lexical classes in the sense of referring to
the basic lexical notions of objects and events, and implementing primary syntactic
functions such as arguments and predicates (Schachter, 1985). In both English and
Hebrew, nouns and verbs are rather easy to characterize in uniform grammatical and
structural terms such as morpho-syntactic behavior: for example, nouns take possession
markers and verbs decline in tenses.
1.1 Adjectives: a typological and psycholinguistic perspective
Adjectives constitute a third content-word class, which is less primary in a
number of senses. From a pragmatic point of view, Thompson (1988) shows that
adjectives have differential discourse functions in spontaneous conversation, together
with nouns and verbs. Semantically, adjectives denote attributes or properties of nouns,
that is, they serve in a secondary function to a primary class. Syntactically, adjectives
fulfill two functions, again, in relation to nouns: Predicative adjectives have the function
of predicate heads (e.g., Mary is smart); attributive adjectives have the function of NP
modifiers (e.g., the smart student). In both cases, the adjective denotes a property
attributed to a noun - either the subject of the sentence or the NP head (Ferris, 1993).
In classical linguistic terms, nouns are those terms that refer, describe or designate
objects in some way, whereas adjectives characterize them (Lyons, 1966). This is
reflected in the fact that in many languages, adjectives agree with the noun they modify
in number, gender, and in many others also in additional values such as definiteness or
case (e.g., French, Latin, Hebrew). In his survey of linguistic universals, Greenberg
(1966) notes that in all languages where the adjective follows the noun, it expresses all
the inflectional classes marked by the noun, even in cases where the noun itself may lack
overt expression of one or all of them. This implies that nouns have a fixed form
independent of any modifier they receive, whereas adjectives presuppose a noun and
adjust their form to correspond to its inflection (Markman, 1989).

The secondary nature of adjectives is expressed typologically in the fact that
while all languages have an adjective class, its size varies extremely and in many
languages it is not a canonical open-class category (Dixon & Aikhenvald, 2004). In some
languages this is a small and closed-class system expressing mainly dimensions, color,
age and value. Typological surveys show that adjectival meanings are often expressed by
(mainly abstract) nouns (e.g., in Hausa), and (again, mainly) relativized stative verbs
(e.g., Bemba and Mandarin Chinese) (Dixon, 1977; Schachter, 1985). Biblical Hebrew,
for example, lacked a well-defined adjective class and expressed adjectival meanings by
verb and noun categories (Gesenius, 1910). This interim status of adjectives is described
by Thompson (1988) from a pragmatic point of view to show that adjectives indeed share
discourse functions in spontaneous conversation with nouns and verbs.
In psycholinguistic terms, the representation of adjectives in the mental lexicon is
less richly structured and more arbitrary than that of nouns. Working within a
categorization framework, Markman (1989) presents evidence that people expect nouns
but not adjectives to refer to concepts that have considerably enduring and permanent
inferential depth, that provide fundamental, essential information about the object and its
identity, that are more readily placed in a taxonomy, and are difficult to combine with
richly structured categories (1989:116-135). Adjectives, in contrast, are less dense in
meaning and have a less correlated structure than nouns, and they are more prone to
adjusting not only their form but also their meaning according to the modified noun.
Compare, for example, good person – good knife; large house – large mouse; and even
more dramatically, criminal act vs. criminal lawyer (Bolinger, 1967). Comparing the two
lexical categories, Markman claims that frequently used nouns tend to convey richer,
stronger, more stereotyped information than do common, frequently used adjectives.
Adjectives point to arbitrary categories – where a single property might be the defining
characteristic implying a contrast between members of the same noun category and
specifying subdivisions within a richer category along many different dimensions. It
seems that adjectives presuppose nouns in some way, whereas nouns do not presuppose
adjectives. The interdependence of nouns and adjectives is demonstrated in Sedivy,
Tanenhaus, Chambers & Carlson (1999), who present eye-tracking experiments to show
that various types of adjectives modifying nouns are interpreted contrastively with the
help of contextual clues in cases of referential indeterminancy.
Another facet of the representation of adjectives vs. nouns is the different ways
the two lexical classes function in the way people organize and retrieve information in
memory. Markman (1989) presents evidence from studies of paired associate learning
and semantic memory in English that suggest that nouns may have some privileged status
in memory, allowing more accurate, quicker access to information, and being more
effective as memory cues than adjectives and verbs. For example, nouns are better
retrieval cues than adjectives, and when nouns precede adjectives, N-A pairs are learned
better despite the word order mismatch in English. This critical difference between the
lexical classes of nouns and adjectives emerges early on in development. Gelman &
Markman (1985) report an experimental study of noun and adjective interpretation in
young children (aged 2;6-3;6) who were asked to “find the ball” or to “find the red one”.
When asked to interpret adjectives, children tended to focus on a contrast between
members of the same object category, but nouns prompted children to select the more
distinctive exemplar of the category.

1.2 Adjective acquisition
The secondary nature of adjectives is demonstrated in a set of experiments
involving adult L2 learners of Russian (Polinsky, 2004) who were tested their
comprehension and translation of nouns, verbs, and adjectives. Polinsky found that
incomplete L2 learners do most poorly on adjectives, and explained her findings in the
claim that “it is possible that the knowledge of adjectives is more or less a luxury”, given
that adjectives are ‘rhetorical devices’ whose comprehension is not essential to the noun
phrase.
A recent spurt of experimental studies with young children and toddlers
demonstrate to what extent learning adjectives depends on the nature of the noun they
modify or predicate. Mintz & Gleitman’s (2002) study of novel adjective interpretation
by 2- and 3-year-old children provides evidence that they lean heavily on object construal
of the noun: Novel adjectives such as drin were reliably assigned to taxonomically
specific lexical nouns such as car or ball, but only at chance level to taxonomically
underspecified nouns – that is, generic thing and pronominal one. Mintz & Gleitman
claim that object construal, which is enabled by taxonomically specific nouns, is a
necessary step on the way to determining a novel adjective’s meaning. Other studies have
shown that learning adjective meaning is facilitated when basic-level nouns support
adjective mapping (Klibanoff & Waxman, 2000; Waxman & Klibanoff, 2000; Waxman &
Markow, 1998). A recent study (Mintz, 2005) also indicates that object familiarity
facilitates young children’s mapping of adjective meaning. And Manders & Hall (2002)
showed that word teachers, like word learners, are sensitive to the conditions under which
contrasts promote novel adjective learning. Likewise, Ninio (2004) shows that children
younger than 4;0 can use adjectives to make simple discriminations, but face difficulties
in comprehending N-A combinations even when both are basic, frequent lexical items.
Finally, children, like adults, assign contrastive functions to adjectives in the
interpretation of noun phrases (Prasada & Cummins, 2000).
But adjectives are interesting in and by themselves in the domain of language
acquisition. As a lexical class with universal and language-specific semantic, syntactic
and morphological characteristics, they can shed light on the nature of lexical
categorization and deserve to be investigated from a developmental perspective. Being
less robust and canonical than either nouns or verbs, their emergence and consolidation in
child language development should serve as a window on aspects of lexical acquisition
that might be obscured in the earlier and more rapid acquisition paths of the more
canonical lexical classes.
One obvious direction of investigation would be to examine the rate of acquisition
of adjectives in early childhood and their relative proportion among words in general and
content words in particular: Since adjectives represent what might be construed as ‘non-
essential’ information, their emergence should herald a larger, richer and more complex
lexicon with more subtle distinctions than the initial core noun-verb lexicon. Another
thrust of investigation would be to examine the development of the various semantic
classes expressed by adjectives, such as dimension, color, internal and external
properties. In languages with rich morphology, this analysis would be supplemented by
an examination of morphological patterns in adjective learning. Adjectives also constitute
a nicely delimited class for the study of syntactic acquisition: While nouns in noun
phrases can take all kinds of syntactic roles, and verbs occupy only predicative position,
adjectives fulfill two canonical syntactic sites – predicative and attributive, with
implications for syntactic complexity and for the information structure of discourse
(Englebretson, 1997). Finally, the fact that adjectives are descriptors and modifiers rather
than primary meaning-bearers reduces their capacity for being syntactically modified
within the adjectival phrase. This too points at a potentially interesting direction for
investigation.
Studies of adjective acquisition have focused to date on early language
development, mostly from distributional and semantic points of view. Diary studies and
surveys of natural language acquisition show that adjectives appear later in child speech
than do nouns and verbs (Casseli, Bates, Casadio & Fenson, 1995; Rice, 1990; Sommers,
Kozarevich & Michaels, 1994). They also constitute an extremely low-frequency class
when compared to other content words in children’s early lexicons in various languages
(Dromi, 1987; Marvin, Beukelman & Bilyeu, 1994; Sandhofer, Smith & Luo, 2000;
Valian, 1986). Nevertheless, after an early phase of acquiring predominantly common
nouns followed by verbs, children come to acquire adjectives as well (Barret, 1995).
Ninio (1988) claims that the emergence of abstract predicative categories such as verbs
and adjectives in child language follows the emergence of hierarchical syntax involving
the insight of creating higher-order complex units. Adjective comprehension is especially
difficult, since it has to be interpreted relative to the noun it attributes to generate a subset
of the category of objects sharing the property designated by the adjective (Ninio, 2004).
According to Berman (1988), adjectives enter the child’s repertoire relatively later than
do verbs and nouns since they share features with both, and are therefore less prototypical
than verbs and nouns. It thus takes time for children to integrate semantic, syntactic and
morphological cues to make the necessary distinctions between nouns and verbs, on the
one hand, and adjectives, on the other (Mintz, Newport & Bever, 2002).
A systematic and detailed study by Blackwell (2005) investigated the adjective
lexicons of Adam and Sarah from ages 2;3 to 5;0, and of their mothers (Brown, 1973).
Adjectives were coded by semantic field, and the study establishes an order of acquisition
by semantic properties. Bt age 2;4, most semantic classes were already represented in the
children’s productions. Color, dimension, physical property and value adjectives emerged
and increased in the third year of life. Human propensity, behavior, and various sub-
categories of physical property emerged between 3;0-4;0, and were followed by mental
state adjectives. Age of acquisition was found to be correlated with three properties of the
input: input frequency, syntactic diversity, and variety in noun-type co-occurrence. Older
studies of adjective acquisition beyond its early phases indicate that a complex of factors
continues to direct patterns of acquisition (Berman, 1994; Valian, 1986; Yuill, 1992). For
example, a universal hierarchy has been found in the acquisition of dimensional
adjectives from global size to specific dimensions, so that big / small precedes the
acquisition of high / low (Kallio, 1988; Richards, 1979). However the semantic
components making up words such as long and high differ in different languages and
may affect rate and order of acquisition (Harris, Morris & Terwogt, 1986). With growth,
adjectives express finer semantic distinctions and eventually metaphorical meanings
(Blodgett & Cooper, 1988).
A study of Hebrew adjective development (Ravid & Nir, 2000) examined the
same age span as Blackwell’s study. Fifty kibbutz children aged 2;0 to 6;0 were recorded
in peer interaction, and the adjectives they produced were counted and analyzed. The
number of adjectives produced increased with age from 0.1 per utterance in the youngest
group to 0.2 in the oldest. The earliest semantic classes to emerge were color, dimension
and physical property adjectives, as in Blackwell’s English study. Later emerging
adjectives denoted more abstract and internal states and attributes. As in Blackwell’s
study, by 3;0 most – but not all - adjective categories had at least one representative, with
semantic and morphological diversity increasing with age. From a syntactic perspective,
most adjectives in all transcripts were predicative, with 5.5 predicatives for every
attributive adjective in the 2-year-olds, but the proportion of attributive adjectives rises
steadily with age until there are 1.5 predicative for every attributive adjective in the 6-
year-olds. At the same time, adjective modification increased both quantitatively and
qualitatively. All three domains analyzed in the Hebrew study – semantics, morphology,
and syntax – were highly correlated in all age groups.
The results of the studies reviewed above in various languages indicate that the
category of adjectives emerges during the second year of life and is an important
yardstick for semantic and syntactic development in preschool. The literature review also
shows that children are aware of the additional informative value of adjectives in relation
to nouns from early on. However, the full array of adjective categories is far from present
at age 6;0, and its development is intertwined with the emergence and consolidation of
the ‘advanced’, high-register, literate lexicon and its cognitive correlates (Dockrell &
Messer, 2004; Ravid, 2004a; Ravid & Berman, submitted). In English, for example,
denominal adjectives of Latinate origin such as informative or industrial are not in
productive usage before highschool (Bar-Ilan & Berman, in press). Likewise, Hebrew
denominal and reduplicated adjectives are late-emerging morphological categories with a

long developmental route across adolescence (Ravid, 2004b). A systematic study of how
the adjective category expands and diversifies during the school years to include all of the
subcategories found in adult language and to fulfill all syntactic positions and functions
would rectify this lacuna in our knowledge.
Moreover, in later language development, it seems desirable to investigate
adjectives in context rather than as isolated lexical items. The close, interdependent
relationship between nouns and adjectives suggests that syntactic and lexical knowledge
go hand-in-hand and may pace each other in development (Ravid & Cahana-Amitay,
2005). For example, the higher the occurrence of content words, especially nouns and
adjectives, the higher the complexity of the syntactic architecture that frames them
(Ravid, van Hell, Rosado, and Zamora, 2002). The syntactic distribution of adjectives as
either predicative or attributive has been shown to be determined by text genre (Chafe,
1982; Englebretson, 1997; Thompson, 1988). Beyond syntax, the occurrence of higher-
register adjectives has been shown to correlate with both genre and modality (Ravid,
2004; Ravid & Zilberbuch, 2003) and to interact with other constructions expressing
detached and abstract stance, such as complex syntax, passive voice, nominalizations,
impersonal pronouns, irrealis and non-finite constructions (Berman & Nir-Sagiv, 2003).
Against this background, the aim of the current study is to take up adjective
development in Hebrew from the point where previous studies have left off – the school
years. It examines changes in adjective distribution in oral and written narrative and
expository texts from lexical, semantic, morphological and syntactic perspectives.
1.3 Hebrew adjectives

Hebrew offers an interesting case of a language in which a morphological class
has evolved, and where child language acquisition mirrors this process. Biblical Hebrew,
despite being a morphologically rich language, did not have a morphological class of
adjectives (Gai, 1995; Gesenius, 1910). Primary adjectival notions such as tov ‘good’, ra
‘bad’ were mainly expressed by present-tense participial (benoni) verb forms which share
many features with nouns. At the same time it had a small class of nouns denoting ethnic
origin with the suffix -i, e.g., yevusi ‘belonging to the nation of Yevus’. These constitute
the roots of present-day adjectival derivation in Hebrew.
Modern Hebrew has four major structural classes of adjectives: One, an
essentially closed class of primary CVC adjectives originating in Biblical present-tense
participials (e.g., xam ‘hot’). These adjectives are morphologically simplex, since despite
their verbal origin, they are monomorphemic as well as monosyllabic, having lexicalized
into a single unit. They also designate basic semantic relations such as good, bad, hot and
cold (Ravid & Nir, 2000). As a result, they are very early acquisitions. A second class of
adjectives contains a variety of nonlinear root+pattern structures (e.g., mahir ‘fast’, root
m-h-r, agentive noun pattern CaCiC). Except for a class of color terms, which is
inherently adjectival (e.g., kaxol ‘blue’, sagol ‘purple’), almost all of these are
appropriated from either verbal or nominal patterns, For example, mahir ‘fast’, and axil
‘edible’ take the agentive noun pattern CaCiC (cf. pakid ‘clerk’); while mafxid ‘scary’
and mevushal ‘cooked’ use participial maCCiC and meCuCaC verbal patterns. The
semantic content of these verbal / nominal adjectives varies across structural categories,
and designates a range of general and specific properties, attributes, and states. Times of
acquisition also vary in accordance with the semantic content of the adjective class. For
example, color terms are acquired and conjugated correctly early despite their structural
complexity due to their frequency in CDS (Ravid, 1995). Resultative adjectives, in
contrast, emerge and consolidate between the ages of 4 and 6 (Berman, 1994; Ravid &
Yagev, 2003).
A third structural device is reduplication, which is used mostly for adjective
diminutivization, e.g., vradrad ‘light pink’ from varod ‘pink’. It is a late-emerging device
which is accessible mostly to mature users of Hebrew (Hora, Avivi-Ben Zvi, Levie &
Ravid, in press; Ravid & Geiger, submitted). The fourth and most productive class of
adjectives in Modern Hebrew is a late historical development deriving from those
Biblical ethnic nouns which evolved in Medieval Hebrew into a full-fledged class of
denominal adjectives, e.g., xashmal-i ‘electr-ic’, tsibur-i ‘publ-ic’. Structurally,
denominal adjectives are simpler than the root-and-pattern class, since they involve linear
formation of a nominal stem and the addition of the adjectival suffix –i. However, they
are typical of higher-register, written Hebrew, such as literary prose, journalistic and
expository texts, and their meaning is quite complex (Ravid, 2004). About one third of
the adjective types occurring in journalistic Hebrew are denominal (Ravid & Shlesinger,
1987). Apart from lexicalized forms such as xagigi `festive’ and the original Biblical
ethnic-attributive meaning (e.g., dati `religious’, rusi `Russian’), they are completely
absent from child-directed speech. Denominal i-suffixed adjectives are the last type of
adjectives to emerge in Hebrew child language around age 6, and they do not emerge in
text production before highschool (Levin, Ravid & Rapaport, 2001; Ravid & Zilberbuch,
2003a).
2.0 DATABASE, ANALYSES, AND PREDICTIONS

The database for this analysis consists of 252 texts produced by 63 Hebrew-
speaking children, adolescents and adults. The participants consisted of three groups of
4th graders (aged 9-10), 7th graders (aged 12-13), and 11th graders (aged 16-17), 16 in each
age group, compared with a group of 15 adult college graduates. All participants were
native speakers of Hebrew from middle-to-high SES background. They were asked to tell
and write a story about a personal fight or a quarrel, and to present and write an
expository text about school violence. Each participant thus produced four texts – two
spoken and two written, in two genres: two personal “fight” narratives, and two
expositions about school violence.
The choice of two modalities and two different text genres has important
implications for the study of content words in textual context. Written texts are
informatively and linguistically denser than spoken texts, with implications for the
proportions of content words and complex syntax (Chafe, 1994; Ravid & Berman, in
press; Strömqvist, Nordqvist, & Wengelin, 2004). Narratives focus on people, their
actions and motivations, and express the unfolding of events in a temporal framework
(Berman & Slobin, 1994). In contrast, expository texts, as a special type of non-narrative
genre (distinct from description, for example), focus on ideas and concepts, and express
the unfolding of claims and argumentation in a causal context (Britton, 1994; Mosenthal,
1985). Recent studies demonstrate the impact of genre and modality on the selection of
rhetorical / expressive devices and grammatical constructions in non-expert text
production (Berman, 2005; Berman & Nir, 2004; Ravid & Zilberbuch, 2003b). While
spoken texts, on the one hand, and narratives, on the other, are easier to produce, written
texts, and especially written expositories are more linguistically complex in the sense of
having greater lexical density and diversity and more complex syntactic architecture
(Ravid, 2004; Ravid & Zilberbuch, 2003a). These studies enable us to generate general
predictions for the current study, namely, that genre and modality should determine the
distribution of adjectives in texts, with more and more diverse categories of adjectives in
written than spoken texts, and likewise in expositories versus narrative texts.
2.1 Text analyses
In preparation for adjective analysis, we carried out a protocol for text size and
lexical text analysis (Ravid, 2004). The two basic units measuring text size are words and
clauses: Words are defined for Hebrew as graphemic sequences separated by spaces,
while for clauses we follow the definition of a clause as “a unified predication” in
Berman & Slobin (1994: 660-664). The following were identified and counted in each
text: All clauses, word tokens, content word tokens, and adjective types and tokens.
Based on previous analyses, we predicted a rise in all of these measures with age and
schooling, and in particular in written and expository texts.
2.1.1 Adjective categories
Two different categorizations were applied to the adjectives in the texts: Morpho-
semantic and syntactic.
(i) Morpho-semantic categorization. Hebrew Adjective tokens were classified
into five categories, based on the psycholinguistic and developmental literature, as
follows:
(1) Core adjectives. This category is based on the list of adjectives frequently
produced by children aged 2-5, documented and compiled by Nir (1997). It
contains a list of basic monosyllabic adjectives of the type described in 1.3 above
(e.g., kal ‘light’), and in addition, canonical color terms (e.g., yarok ‘green’),
basic size adjectives (katan ‘small’), basic evaluative adjectives (yafe ‘pretty’,
xazak ‘strong’) and some adjectives with modal functions (tsarix ‘necessary /
needful’).
(2) Resultative participial adjectives. This category of root-and-pattern adjectives
typically develops in Hebrew-speaking children between the ages of 4-6 (see 1.3
above). It contains passive adjectives, mostly with resultative meaning (e.g., tafus
‘occupied’, muxba ‘hidden’), based on three different participial present-tense
verb patterns: CaCuC, meCuCaC, and muCCaC.
(3) Adjectives with verbal / nominal patterns. This category contains adjectives which
share their forms with nouns and verbs, as explained above. For example, zahir
‘careful’, nominal pattern CaCiC, salxan ‘indulgent’, nominal pattern CaCCan, or
ne’eman ‘loyal’, verb pattern niCCaC.
(4) Reduplicated diminutive adjectives, e.g., shxarxar ‘darkish’ from shaxor ‘black’.
(5) Denominal i-suffixed adjectives. This class of literate, late-emerging adjectives is
extremely productive in Modern Hebrew. Examples include mosadi ‘institutional’
and ma’asi ‘practical’.
(ii) Syntactic categorization. Adjectives were classified by three syntactic criteria:
(1) Syntactic site: (i) Predicative, e.g., ha-délet hayta ne’ula ‘the door was locked’
(7th grader, written narrative); (ii) Attributive, e.g., xadar kosher zamin ‘(an)
accessible fitness room’ (adult, written expository); (iii) Adverbial function, e.g.
hem ne’henim lir’ot yeladim axerim sovlim ‘they enjoy watching other kids
suffer’; (iv) NP head (noun ellipsis), e.g., ba-dérex ha-zot eyn mafsidim ve-eyn
menatsxim ‘this road has no losers and no winners’ (7th grader, written
expository).
(2) Participation in compelx structure: (i) Adjective conjoining, e.g., metuxkamot
ve-adinot ‘sophisticated and subtle’; (ii) Adjective stacking, e.g., hashpa’a
xevratit gdola ‘great social influence’; (iii) Compound head, e.g., alimut tluyat
matsav ‘situation-dependent violence’.
(3) Internal AP modification: (i) basic modification, e.g., haxi gdola ‘most big =
biggest’, ma-ze alimim ‘so violent’; (ii) advanced modification, e.g., yaxasit
katan ‘relatively small’; (iii) multiple modifications, e.g., harbe yoter lo ne’imim
‘much more unpleasant’.
3.0 RESULTS
In order to create a framework for adjective analysis, we measured text size in
terms of words, clauses, and mean clause length (number of words divided by number of
clauses). Table 1 presents these measures by age, genre, and modality.
PLEASE INSERT TABLE 1 ABOUT HERE
Three-way ANOVAs (age x genre x modality) on the data in Table 1 showed the
following: Age group. Increase in numbers of words (F(3,59)=16.27, p<.001) and clauses
(F(3,59)=8.99, p<.001), and mean clause length (F(3,59)=11.29, p<.001). According to
post-hoc Bonferroni analyses, adults had the most words and clauses, and the two older
groups had the longest clauses. Narratives had more words (M=122.89) (F(1,59)=4.58,
p<.04) and also more clauses (M=27.75) (F(1,59)=17.74, p<.001) than expositories
(M=101.87 and M=19.45 respectively). Expositories had longer clauses (M=5.28) than
narratives (M=4.51), (F(1,59)=24.19, p<.001), but this difference begins only in 11th
grade (F(3,59)=3.58, p<.02), as shown in Figure 1.
PLEASE INSERT FIGURE 1 ABOUT HERE
Spoken texts had more words (M=123.26) (F(1,59)=10.6, p<.003) and also more
clauses (M=26.29) (F(1,59)=12.56, p<.002) than written texts (M=101.5 and M=20.91
respectively).
3.1 Lexical density
To neutralize different text lengths, we measured lexical density by number of
content words and adjectives per clause (Table 2). All analyses were performed at the
token (rather than type) level, since we were interested in actual usage (Bybee, in press;
Elman, in press).
Content words. A three-way ANOVA showed that the number of content words
per clause increases with age and schooling (F(3,59)=6.39, p<.001), between 7th grade
(age 12-13) and 11th grade (16-17). Expositories contained more content words (M=2.97)
than narratives (M=2.1), (F(1,59)=54.75, p<.001), but this difference emerges only in 7th
grade, increasing dramatically in the expositions of the three older age groups
(F(3,59)=3.76, p<.02) (Figure 2).
Written texts contained more content words (M=2.77) than spoken texts (M=2.3),
(F(1,59)=24.34, p<.001), but this difference stems only from the expository texts
(F(1,59)=17.74, p<.001), as shown in Figure 3.

Adjectives. A three-way ANOVA showed that the number of adjectives per clause
increases with age and schooling (F(3,59)=5.45, p<.003), with the largest number in the
adults, differing from the two youngest age groups. Expository texts contained more
adjectives (M=0.53) than narratives (M=0.21), (F(1,59)=99.03, p<.001). Written texts
contained more adjectives (M=0.44) than spoken texts (M=0.31), (F(1,59)=15.89,
p<.001). This difference emerges only in the written texts of the two older age groups
(F(3,59)=2.95, p<.05), as shown in Figure 4; and it stems from a difference between
written texts of the two different genres (F(1,59)=4.91, p<.04), as shown in Figure 5.
PLEASE INSERT FIGURES 4 AND 5 ABOUT HERE
3.2 Morpho-semantic analyses
Three analyses were performed on adjective morpho-semantic categories.
3.2.1 Mean score on the scale
The adjective scale provides one way of assessing morpho-semantic development.
Table 3 presents the mean scores on the morpho-semantic scale described in 2.1.1 above.
Note that we collapsed ranks 4 and 5 so that the scale runs only up to 4, since a negligible
number of diminutive adjectives occurred in the texts.
A three-way ANOVA showed that score on the scale increases with age and
schooling (F(3,59)=31.3, p<.001), and most of all in the adults. Expository texts had a
higher score (M=1.88) than narratives (M=1.17), (F(1,59)=25.27, p<.001). There were no
other effects or interactions.
3.2.2 Number of different adjective categories

The number of different adjective categories in a text is another measure of
adjective "richness". Table 4 shows the number of different adjective categories in the
analyzed texts.
A three-way ANOVA showed that the mean number of different adjective
categories increases with age and schooling (F(3,59)=23.77, p<.001), with two cut-off
points - after 4th grade, and after 11th grade. Expository texts had a higher number of
adjective categories (M=2.52) than narratives (M=2.24), (F(1,59)=6.02, p<.02). Also,
written texts had more adjective categories (M=2.65) than spoken texts (M=2.11),
(F(1,59)=20.64, p<.001). There were no interactions.
3.2.2 Size of different adjective categories
We performed two analyses to assess the size of each of the adjective categories
while neutralizing differences in text lengths. One examined the relative proportion of
adjectives in each category out of all adjectives, and another looked at the numbers of
adjectives in each category per clause. While both analyses reveal similar trends, together
they provide a fuller picture of the development of the adjective class.
(i) Number of category items per clause. Table 5 presents mean number of
adjectives per clause from each of the four adjective categories.
Three-way ANOVAs on each of the adjective categories in Table 5 showed the
following: Age group: decrease in Category 1 core adjective tokens (F(3,59)=3.43,
p<.03): 4th graders have the most core adjectives, and 11th graders the fewest; increase in
Category 2 resultative adjectives (F(3,59)=5.51, p<.003), with a cutoff point after 7th
grade; increase in Category 3 verbal / nominal adjectives (F(3,59)=18.61, p<.001), with
cutoff points after 7th and 11th grades; Category 4 shows no effect for age group. Genre:
no effect in Category 1 and 2; more Category 3 in expositories (M=0.13) than in
narratives (M=0.08), (F(1,59)=22.1, p<.001); and more Category 4 in expositories
(M=0.13) than in narratives (M=0.06), (F(1,59)=7.26, p<.01). Modality: Spoken texts
have more Category 1 adjectives (M=0.18) than written texts (M=0.09), (F(1,59)=23.37,
p<.001); and the number of Category 1 adjectives declines sharply after 4th grade in
written texts (F(3,59)=6.94, p<.001), as shown in Figure 6.
There is no effect for modality in Category 2 resultative adjectives. Written texts
have more Category 3 adjectives (M=0.16) than spoken texts (M=0.05), increasing
gradually in spoken texts and exponentially in written expositories (F(3,59)=2.95, p=.04).
Finally, denominal adjectives are more numerous in written (M=0.16) than in spoken
texts (M=0.03), with no interactions.
(ii) Proportion of adjective categories. Table 6 presents the percentage of
adjectives in each of the categories.
Three-way ANOVAs on the data in Table 6 showed the following: Age group:
Decrease in Category 1 adjectives (F(3,41)=19.55, p<.001): Cutoff points after 4th grade
and 7th grade; increase in Category 2 adjectives (F(3,41)=3.26, p<.04), cutoff point after
7th grade; increase in Category 3 adjectives (F(3,41)=4.83, p<.007), with cutoff points
after 4th and 11th grades; Category 4 shows no effect for age group. Genre: More Category
1 adjectives in narratives (M=48.87%) than in expositories (M=38.91%), (F(1,41)=7.78,
p<.009); more Category 3 adjectives in expositories (M=32.08%) than in narratives
(M=21.57%), (F(1,41)=4.15, p<.05); and no genre effects in Categories 2 and 4.
Modality: Spoken texts have more Category 1 and 2 adjectives (M=49.84%, M=16.75%)
than written texts (M=37.93%, M=7.12%), (F(1,41)=5.09, p<.04), (F(1,41)=9.96,
p<.004), respectively. Written texts have more Category 3 and 4 adjectives (M=32.08%,
M=22.87%) than spoken texts (M=21.57%, M=11.85%), (F(1,41)=8.02, p<.008),
(F(1,41)=7.54, p<.01), respectively. No interactions emerged in this analysis.
3.3 Syntactic analyses
We analyzed adjectives by (i) syntactic position and (ii) modification types.
3.3.1 Adjectives in different syntactic positions
Two analyses of different syntactic positions are presented below.
(i) Adjectives per clause by syntactic positions. Table 7 presents the mean number
of adjectives per clause by syntactic position, age, genre, and modality.
Three-way ANOVAs on the data in Table 7 showed the following: Age group: No effect
in predicative, adverbial, and NP head adjectives; increase in number of attributive
adjectives per clause (F(3,59)=6.09, p<.001), cutoff between oldest and youngest age
groups. Genre. No effect in predicative, adverbial, and NP head adjectives; expository
texts have more attributive adjectives (M=0.26) than narrative texts (M=0.15),
(F(1,59)=25.98, p<.001). Attributive adjectives level off in narratives from age 12, while
continuing to increase in expositories (F(3,59)=3.34, p<.03), as shown in Figure 8.

Modality: Written texts have more predicative adjectives (M=0.23) than spoken
texts (M=0.07), (F(1,59)=63.49, p<.001), and also more attributive adjectives (M=0.29)
than spoken texts (M=0.13), (F(1,59)=38.37, p<.001). There are no modality effects in
adverbial and NP head adjectives.
(ii) Proportions of adjectives by syntactic position. Table 8 presents the
percentages of adjectives out of all adjectives by site, age, genre, and modality.
Three-way ANOVAs on the data in Table 8 showed the following: Age group: Decrease
in predicative adjectives (F(3,41)=2.91, p<.05), but no differences detected among age
groups; increase in attributive adjectives (F(3,41)=2.95, p<.05), but no differences
detected among age groups. No effects in adverbial and NP head adjectives. Genre. More
predicative adjectives in narratives (M=48.5%) than in expositories (M=39.99%),
F(1,41)=5.55, p<.03). More attributive adjectives in expositories (M=54.32%) than in
narratives (M=47.25%), (F(1,41)=3.97, p=.05). No genre differences in the other two
syntactic positions. Modality: No difference in predicative, attributive, and adverbial
adjectives. More NP head adjectives were found in spoken (M=5.57%) than in written
texts (M=2.03%), F(1,41)=6.74, p<.02). This difference comes mainly from the teenage
groups F(3,41)=5.16, p<.005), as shown in Figure 9.
3.3.2 Participation in complex structures
We now proceeded to analyze adjectives in complex adjectival structures
including adjective conjoining, stacking, and in the function of compound heads. Since
complex structures involve more than one adjective, only the first analysis was possible.
Adjectives in complex structures per clause. Table 8 presents the mean number of
adjectives per clause by complex structure type, age, genre, and modality.
Three-way ANOVAs on the data in Table 8 showed the following: Age group: No
effect in conjoined and stacked adjectives; increase in number of adjectives as compound
heads (F(3,59)=3.25, p<.03), with the significant difference between the youngest 4th
graders and the adults. Genre. Expository texts have more conjoined adjectives
(M=0.019) than narrative texts (M=0.008), (F(1,59)=7.98, p<.007). No other effects or
interactions emerged in the other complex categories. Modality: Written texts have more
conjoined adjectives (M=0.02) than spoken texts (M=0.007), (F(1,59)=9.14, p<.005). No
other effects or interactions emerged in the other complex categories.
3.3.3 Internal AP modification
Our last analysis focused on three categories of internal AP modifications - basic,
advanced, and multiple modifications. Since this analysis always involves a single
adjective, we were able to examine both percentages and numbers per clause.
(i) Adjectives with internal modification per clause. Table 9 presents the mean
number of adjectives per clause by internal AP modification category, age, genre, and
modality.
Three-way ANOVAs showed as follows: Age group: No effect in any of the
adjective modification categories. Genre. No effect in any of the adjective modification
categories. Modality: Written texts have more adjectives with advanced modification
(M=0.04) than spoken texts (M=0.002), (F(1,59)=30.2, p<.001), and more adjectives with
multiple modifications (M=0.09) than spoken texts (M=0.001), (F(1,59)=68.85, p<.001).
(ii) Proportion of adjectives with internal modification. Table 10 presents the
mean number of adjectives by three internal AP modification categories, age, genre, and
modality.
Three-way ANOVAs showed as follows: Age group: No effect in any of the
adjective modification categories. Genre. Spoken texts have more adjectives with basic
modification (M=17.89%) than written texts (M=12.47%), (F(1,41)=7.64, p<.01). No
effects in the other modification categories. Modality: Spoken texts have more adjectives
with basic modification (M=19.22%) than written texts (M=11.13%), (F(1,41)=8.81,
p<.006). Written texts have more adjectives with advanced modification (M=5.76%) than
spoken texts (M=0.77%), (F(1,41)=16.72, p<.001), and more adjectives with multiple
modifications (M=16.89%) than spoken texts (M=0.47%), (F(1,41)=68.4, p<.001).
4.0 DISCUSSION: ADJECTIVES IN THE DEVELOPKMENT OF TEXT
PRODUCTION
Taken together, our findings show that as predicted, the adjective class grows
larger, richer and more diverse with age and schooling in lexical, morpho-semantic and
syntactic terms. Moreover, and again as predicted, adjectives also configure according to
text types and modalities in ways that provide independent support for text type
classification based on different analyses (Berman, Nir-Sagiv & Bar-Ilan, in press; Ravid,
2004, in press). These two clear findings converge to provide a picture of the evolution of
the category of adjectives as a yardstick in later language development, and specifically
in the development of text production.
First, consider the findings regarding the size and complexity of the texts within
which the adjectives were analyzed. Text size as measured by words and clauses, which
increases mainly in later adolescence, gives some indication of the increase in amount of
content and hence informativeness with age and schooling. However what we can learn
from text size is confounded by ancillary and less informative material such as
repetitions, reformulations, false starts, and discourse markers, which render narratives
longer than expositories and spoken texts longer than written texts (Ravid, 2004; Ravid &
Berman, 2006; Ravid & Zilberbuch, 2003).
A clearer and more consistent picture emerges from the analysis of mean clause
length, a measure that neutralizes different texts lengths. It shows that adolescents and
adults produce longer clauses only in expository texts. Longer clauses consist of more
words per clause, pointing to expositories as the locus of lexical and syntactic
complexity: Lexical complexity is indicated since vocabulary denoting complex
temporal, logical and causal configurations is multi-lexemic, e.g., Hebrew ela im ken
‘unless’ which would be counted in our study as three words. Syntactic complexity is
indicated since a clause length results from a larger number of optional phrases, in most
cases intra-sentence modifiers such as adjectival and adverbial phrases, and / or longer
and more internally complex phrases such as heavy NPs. Mean clause length is thus
associated with lexical and syntactic growth, which are in their turn necessary for the
expression and organization of information in text development. This picture gains
support and focus from the analysis of mean number of content words per clause, which
examines only those open-class words that contribute to text lexicality. Our findings also
reveal that content words become more numerous in written expositories from 7th grade
upwards. Taken together, these analyses point to the very late consolidation (around 16-
18 years of age) of a rich lexical and syntactic text architecture, mostly concentrated in
written texts of the expository genre (Nippold, 1998; Ravid & Zilberbuch, 2003;
Tolchinsky, 2004). The abstract and academic nature of these texts entails densely
informative linguistic constructions with complex, hierarchical and diverse syntactic
architecture, but also lexical density and diversity.
4.1 Adjective density
Against this background, we now turn to the distribution and properties of
adjectives in the texts. First, consider lexical density from the narrower point of view of
number of adjectives per clause, which follows closely the same trends indicated in the
general text-size analyses described above: increase with age and schooling, more so in
written than in spoken texts, more in expository than narrative texts, and specifically in
the written expositories of 11th graders and adults. As the primary function of adjectives is
noun modification, expressing specific nominal distinctions and sub-categories, increase
in the number of adjective tokens reflects richer and more informative text content. For
example, an 11th grade girl writes in a narrative about her former friend (adjectives
italicized): "There was a scene of exaggerated jealousy which resulted in a verbal
argument, a scathing argument, filled with reciprocal accusations and yelling". Moreover,
rich adjectival texture is always grounded in complex nominal and syntactic structure,
since adjectives participate in the construction of noun phrases (Ravid, in press). Finally,
the picture provided by adjective tokens is closely mirrored by type distribution, since
expository texts in almost all age groups contain more adjective types than do narrative
texts: 4th grade expositories contain 61 adjective types, compared with 36 in narratives;
7th grade expositories contain 73 adjective types, compared with 78 in narratives; 11th
grade expositories have 84 adjective types, while narratives have 92; and adult
expositories contain 222 adjective types, while narratives contain 178. It thus seems that
the category of adjectives grows and diversifies at the same time that more adjective
tokens increase in actual usage.
Although it may initially seem counter-intuitive that expository texts are richer in
adjectives than narratives, this finding is not surprising at all and in fact is in line with our
predictions. One reason is that expositories are more nominally dense, and are therefore
more likely to be richly modified by adjectives than narratives, which are less nominal
(Ravid, in press). Another reason is the abstract and academic nature of expositions,
which requires higher-register, foreign and lexically specific adjectives such as
mashma'utit 'significant,Fm', nefoca 'wide-spread', or stati 'static.
Note, moreover, that this developmental pattern in adjective density is mirrored
by the growing complexity and abstractness of the noun category, and in fact is co-
dependent on it. Ravid (in press) shows how nominal density, which underlies much of
the referential and syntactic architecture of texts (Biber, 1995; Halliday, 1989; Ravid,
2004), increases dramatically at the same time that adjectives increase in number -
between high school and adulthood, especially in written expositories. The current
analysis will demonstrate how adjectives contribute to nominal complexity by
participating in NP structure.
4.2 Adjective diversity

While the general number of adjectives in the texts is a measure of lexical density,
it also implies lexical diversity, i.e., the more adjectives, the more diverse categories.
Adjective diversity and 'richness' were measured by examining the morpho-semantic
adjective categories in the texts from three different perspectives: morpho-semantic
ranking, number of different adjective categories, and their size. Taken together, these
analyses provide a well-rounded picture of adjective development in the school years and
their patterning in different text types.
One perspective on adjective diversity relates to their morpho-semantic ranking.
Recall that the morpho-semantic categories were ranked on the scale according to
developmental criteria. We found that score on the scale indeed increases with age and
schooling, especially between adolescence and adulthood. In fact, this textual measure
provides independent support for the gradual developmental increase in adjective 'quality'
– from core adjectives such as adom 'red' and xadash 'new' in 4th grade to resultative
me'orav 'mixed' in 7th grade, nominal-pattern ragish 'sensitive' in 11th grade, and finally
diminutive and denominal shavririt 'fragile, Fm' (from shavrir 'fragment') in the adult
group. In line with the findings about adjective density, we also found that the adjective
score is higher in expository than in narrative texts. This difference is reflected, for
example, in the fact that 4th grade expository texts contain 'advanced' adjectives such as
Category 3 no'az 'reckless' and Category 4 pir'i 'wild' – a far cry from the uniform
Category 1 items which constitute the adjective lexicon in the 4th grade narratives.
A second perspective on adjective distribution and richness is the number of
different adjective categories present in the texts of each age group. This is a measure of
lexical diversity, which does not suffer from the methodological problems pointed out in
Malvern & Richards (1997) regarding token and type counts, since it refers to categories
rather than to lexical items. Here we found a steady increment in number of adjective
categories from one or two categories in 4th grade to well over thee in the adults. This
finding indicates the consolidation of a rich and diverse lexicon which relies on all
structural options in Hebrew and expresses a range of semantic notions. This analysis
again revealed more adjective diversity in expositories than in narratives in all age
groups, except for the adults. For example, the compiled list of adjective types in 4th
grade expositories already displays the full variety of adjective categories from basic yafe
'nice' and kashe 'hard' through resultative mexumamim 'heated, Pl' and mesuyémet 'certain,
Fm', verbal / nominal patterns meca'er 'distressing' and mevix 'embarrassing', to
denominal xinuxi 'educational' and tarbuti 'cultural'. In contrast, the list of adjective types
in 4th grade narratives contains mostly core adjectives.
However, unlike the morpho-semantic scale, adjective diversity as measured by
number of categories also distinguished written from spoken texts. Spoken texts,
constrained by the narrow window of on-line processing in real time and relying heavily
on shared knowledge and cooperation with addressee, do not permit much organization
and planning. As a result they contain a large amount of ancillary non-informative
material (Ravid & Berman, 2006) and are less lexical and informative. Written texts,
which encourage planning, revision, review and rewriting, allow the retrieval of literate
lexical items and morpho-syntactic structures without the pressures of on-line processing,
and are consequently more informative. Adjectives, as a secondary, modifying category,
not only reflect this general distinction, they are extremely susceptible to it. Compare, for
example, the two expository texts produced by the same adult in speech and writing: The
spoken text contains 81 words in 17 clauses, with only 2 adjectives – ra 'bad' (Category
1), and racuy 'desirable' (Category 2). In contrast, the written version, 107 words in 20
clauses, contains 10 different adjectives – one Category 1 adjective (tov 'good'), three
Category 2 resultative adjectives (e.g., mexo'éret 'ugly,Fm'), three Category 3 adjectives
(e.g., mazika 'harmful'), and three denominal adjectives (e.g., sportivi 'sportive'). As we
show below, this lexical diversity accompanies a more variegated syntactic structure in
which these adjectives are embedded.
A third and final point of view on adjective diversity examines the distribution of
adjective categories in the different age groups and text types. The double analysis we
conducted permits their examination both as relative proportions and as items per clause,
unrelated to other adjective categories. Taken together, these analyses highlight the
developmental change in the internal construction of the Hebrew adjective category.
Core adjectives. All analyses point to core adjectives (tov 'good', xazak strong',
yafa 'pretty,Fm') as the most prevalent class in all age groups. This is a morphologically
simplex category, the earliest to emerge in acquisition, and part of the core lexicon of
Hebrew (Ravid & Nir, 2000). They are most numerous in the youngest group, and to a
lesser extent in the 7th graders. But in all groups, those text types that are less lexical and
informative and which less readily rely on rich nominal structure – narratives and spoken
texts – contain more core adjectives than written expositories.
Resultative adjectives. This is the first morphologically complex class of
adjectives to emerge in pre-school language (Berman, 1994; Yagev, 2000) to describe
resultative states of objects (e.g., šavur 'broken' from šavar 'broke'). However, the three
resultative patterns are also occupied by other adjectives, not strictly passive or
resultative, such as everyday modal adjectives (e.g., asur 'forbidden'), on the one hand;
and a host of lexically specific, high-register adjectives with no active counterparts, such
as mesugar 'withdrawn' – on the other. As a result, this category is pervasive in both
genres, and though it clearly increases with age and schooling according to both analyses,
resultative adjectives are more numerous in spoken texts. To illustrate this issue, consider
one adult written expository which contains four different resultative adjectives
(underlined) of all types and registers: alimut hi tofa'a mexo'éret u-mazika 'violence is an
ugly and harmful phenomenon'; beyt ha-séfer alul lihyot gehinom avur yeladim
mesuyamim 'school may be hell for certain kids'; and …ma še-mitraxeš be-nafšam ha-
mesuxséxet 'what goes on in their troubled mind'.
Verbal / nominal adjectives. This category contains an array of adjectives based
on verb and noun patterns, and its development has not been previously investigated
systematically from a psycholinguistic perspective. Both analyses conducted in the
current study indicate that verbal/nominal adjectives constitute a late-emerging category
which continues to increase throughout the school years, with cutoff points in each age
group. These adjectives emerge gradually in spoken texts and increase dramatically in
written texts, more in expositories than in narratives. The most frequent nominal-based
adjective is CaCiC (sharing nominal pattern with nouns such as pakid 'clerk'). This
pattern shows up in diverse adjectives covering broad range of meanings and appearing
in a variety of contexts, starting with basic modal carix 'have to' in younger age groups. A
survey of nominal-based adjectives (underlined) in the texts of the older age groups
yields a rich variety of CaCiC–based forms. For example, ha-šulxan alav katavti lo haya
yatsiv 'the-desk on which I wrote was not steady'; yexasim kalilim beyn xaverot 'light
relationships between girlfriends' (written 11th grade narratives); šerut ya'il, levavi ve-
zariz 'efficient, cordial and speedy service' (written adult narrative); alimut… hi davar
šaxíax 'violence… is a frequent thing' (written 11th grade expository); ha-alimut eyna
stam mila tmima 'violence is not just an innocent word'; xadar kóšer zamin 'an available
gym' (written adult expositories). Alongside with it we find other nominal patterns such
as CaCaC (shared with agentive nouns such as nagar 'carpenter'), e.g., alimut be-beyt-
séfer kayémet 'violence is existent in schools'; and CaCCan (shared with agentive nouns
such as safran 'librarian'), e.g., horim savlanim…klapey ha-yeladim 'parents who are
patient towards children' (written adult expositories).
In younger age groups, frequent and transparently structured verb-based
adjectives occur, as in the written 4th grade expository excerpt with hif'il and pi'el-derived
adjectives alimut ze davar mag'il…ze gam ktsat metsa'er 'violence is something
disgusting… it is also a bit saddening'. Other examples of adjectives based on verbal
patterns include, for example, participle Qal forms such as CoCeC, e.g., he'evarnu méser
šone 'we conveyed a different message' (written 11th grade narrative). In older age groups
we find more opaque nif'al-based forms such aszo axen truma nixbada 'this is indeed a
considerable contribution', he wanted axrayut ve-samxut noséfet 'he wanted additional
responsibility and authority' (written adult narratives), and alimut be-beyt-séfer ze davar
nadoš 'violence in schools is a banal thing' (written 11th grade expository). Taken together,
these examples of the verbal/nominal adjective acquisition indicate to what extent the
adjective lexicon diversifies and becomes enriched with linguistic and cognitive
development coupled with schooling.

Denominal adjectives. The most 'advanced' adjective category in Hebrew is
denominal –i suffixed adjectives, the only genuine adjectival class which does not share
form with verbal or nominal patterns, and which emerges last in acquisition (Ravid,
2004). As the smallest class, it does not show age effects; however it is the hallmark of
written, and especially expository, texts in all age groups. In younger age groups, most
denominal adjectives are frequently-occurring and formulaic, as in the following excerpt
from a 4th grade expository beyt-séfer im alimut ze lo beyt-séfer normali 'a school with
violence is not a normal school'; or alimut kolšehi – im milulit ve-im fisit 'some violence –
whether verbal or physical' (written 7th grade expository). But starting in pre-adolescence,
written texts – and especially expositories - already contain some genuinely derived
denominal adjectives such as yeladim še-yeš lahem hašpa'a xevratit 'kids who have social
influence' (written 7th grade expository), or zo hitnahagut lo humanit 'this is non-humane
behavior' (written 11th grade expository). The most sophisticated denominal adjectives are
found in the adult group, e.g., hi grama le-mifga betixuti 'she caused a safety-wise
obstruction', or netuley xešbonot katnuniyim 'devoid of petty ???' (written adult
narratives).‫להכניס דוגמאות מאקספוזיטורי‬
‫להתפתחות תחבירית‬
Adjectives participate in this rich nominal structure by modifying nouns in noun
phrases, as in ma'aréxet xevratit mele'a 'full social network' (spoken adult expository).
Moreover, NPs with abstract nominal heads tend to attract more morphologically and
semantically complex adjectives (Ravid & Cahana-Amitay, 2005), e.g., kol ha-ká'as ha-
‫‪mitstaber sheli klapéyha 'all of my accumulating anger against her' (written 11th grade‬‬
‫‪narrative).‬‬
‫מה אני חושב על אלימות בבית ספר? אני חושבת שזה דבר רע אין אה קשה לראות את ה‪ ...‬כאילו‬
‫אין פה דיון בכלל‪ .‬כמובן שזה דבר אמממ דבר לא רצוי זה מאמלל אנשים אני מניח יש אנשים שסוחבים את‬
‫זה אתם כל החיים‪ ,‬כל מינידברים שעשו להם בבית ספר‪ .‬אממ זהו כל מה שאני יכול להגיד אם את לא רוצה‬
‫שאני יכנס לנסיון אה‪ ..‬זה מה שיש לי להגיד‪.‬‬
‫‪Shlomi, Humanities, male‬‬
‫‪1.‬‬ ‫‪eh: ma ani xoshev al alimut be-beyt^sefer‬‬

‫‪2.‬‬ ‫‪ani xoshev‬‬
‫‪3.‬‬ ‫‪she ze davar ra‬‬
‫‪4.‬‬ ‫]‪eyn+ eh kashe lirot et ha,+ [trs: unfinished‬‬
‫‪5.‬‬ ‫‪keilu eyn po diyun bixlal,‬‬
‫‪6.‬‬ ‫‪kamuvan she ze: davar+ em: davar loh racuy‬‬
‫‪7.‬‬ ‫‪ze meamlel anashim‬‬
‫‪8.‬‬ ‫‪ani meniax‬‬
‫‪9.‬‬ ‫‪yesh anashim‬‬
‫‪10.‬‬ ‫‪she soxvim et ze itam kol ha-xayim, kol miney dvarim‬‬
‫‪11.‬‬ ‫‪she asu lahem be-beyt^sefer #‬‬
‫‪12.‬‬ ‫‪em, zehu ze kol ma‬‬
‫‪13.‬‬ ‫‪she ani yaxol lehagid‬‬
‫‪14.‬‬ ‫‪im at loh roca‬‬
‫‪15.‬‬ ‫]‪she ani yikanes le-nisayon eh # [trs: unfinished‬‬
‫‪16.‬‬ ‫‪ze ma‬‬
‫‪17.‬‬ ‫‪she yesh li lehagid‬‬
‫מה יש לומר? אלימות היא תופעה מכוערת ומזיקה‪ ,‬בלי קשר‬
‫היכן _ היא מתרחשת‪ .‬אני מניח שאלימות בבתי ספר היא‬
‫בעייתית עוד יותר כיוון שהיא משאירה את רישומה בנפשותיהם‬
‫‪.‬של ילדים שאלו הן השנים המעצבות שלהם‬
‫בית הספר יכול להיות גיהנום עבור ילדים מסויימים‪ ,‬מבלי‬
‫שלמורים שלהם יש רמז על מה שמתרחש בנפשם‬
‫המסוכסכת‪ .‬יש ילדים שמפחדים להגיע לביה"ס כיוון שיש‬
‫‪.‬מי שמתנכל להם‬
‫נראה כי יש להסביר למורים שעליהם למנוע ולהפריע לכל‬
‫אפילו‬
‫‪.‬אקט של אלימות ‪ ,‬יהא זה אקט ספורטיבי או תוך כדי משחק‬
‫זה יכול להיות עוד צעד בדרך לניטרול המצ'ואיזם של‬
‫‪.‬החברה הישראלית‬
‫ילדים שגדלו בסביבה לא אלימה הם אזרחים טובים יותר‬
‫‪.‬בבגרותם‬
‫ניתוחים תומכים אחד בשני‬
‫המריבה‪ ,‬שהייתי שותפה בה‪ ,‬היתה חוויה לא קלה‪.‬‬

‫עד עתה נמנעתי מלדבר עליה‪ ,‬ושיתפתי בה רק אנשים‬
‫ספורים ביותר‪ .‬ואולם זו היתה חוויה כה חזקה ‪,‬‬
‫מבגרת ומחשלת ואף עצובה‪ ,‬שאיני יכולה לכתוב על‬
‫אף מריבה אחרת‪.‬‬
‫‪ XX‬בהיותי בתיכון התמזל מזלי לפתור לפני מבחן‬
‫מסויים שאלות רלבנטיות‪ .‬וכיוון שחשדתי ש"פגעתי בול"‬
‫שיתפתי מספר מועט של בני כיתה ב"מימצאי"‪ .‬מכיוון‬
‫)אכן "עליתי" על השאלות הנכונות‬
‫בבית(‬ ‫שהעניין יצא מכלל שליטה ‪XXXXXXXXXXX‬‬
‫החלטתי כי זה יהיה לא הגון מצידי )ואף טיפשי‪ ,‬כי‬
‫"טובי לב" נתנו את התשובות לכל דיכפין( לא לספר‬
‫למורה על מה שאירע‪ .‬ואכן‪ ,‬החלטתי לספר על מה שהיה‪,‬‬
‫למרות שכאמור אני הייתי זו שבאמת ישבה ופתרה את‬
‫הבחינה בבית‪ ,‬כדין‪.‬‬
‫מה שאירע לאחר מכן‪ ,‬היה מאוד לא נעים בלשון המעטה‪.‬‬
‫פרט לחרם שהוחרמתי ע"י בני כיתתי‪ ,‬שלא עצרו לתת‬
‫לעצמם דין וחשבון על כך שבעצם היתה פה רמאות‪,‬‬
‫למדתי הרבה על אנשים‪ ,‬יושר‪ ,‬צביעות ועוד‪ .‬אבל‬
‫גם למדתי‪ ,‬שחברה כמו החברה שעמדה לצידי והגנה‬
‫עלי בחירוף נפש )ומדובר בבחורה מאוד עדינה(‪,‬‬
‫היא פשוט נכס‪ .‬חברת אמת‪ ,‬שלעולם‪ ,‬אבל לעולם‪,‬‬
‫לא אשכח לה את מה שעשתה למעני‪ .‬ואני פשוט אוהבת אותה‬
‫‪The rich syntactic structure that is typical of adult texts thus consolidates very late‬‬
‫‪in terms of language acquisition, around 16-18 years of age, and is highly dependent on‬‬
the combined effect of literacy and socio-cognitive factors (Nippold, 1998; Ravid &
Zilberbuch, 2003; Tolchinsky, 2004).
general issues, score higher on the scale than personal-experience narratives,
which focus on people participating in events. In all four age groups (except for
narratives in the adolescent groups), the Noun Score has the following ascending order:
Spoken narratives  Written narratives  Spoken expositories  Written expositories.
Written expositories have the highest Noun score of all four text types. The scores within
each age group illustrate this robust finding, rising not only with age and schooling but
also within each group across text types (Table 2). This foreshadows one of the main
findings of this paper – the fact that counter to popular thought, gradeschoolers employ
genuinely individuated, concrete nouns (1-2 on the scale) only sparsely, even in their
spoken personal narratives; instead, they recruit general, categorical nouns with
appropriate modulations for genre and modality in their text production. True, this finding
is limited to the types of discourse investigated in this study and does not cover spoken
conversation (which even so contains few lexical nouns, see Chafe, 1994). But the fact
that each of the participants produced four different texts in two modalities and two
genres does provide us with sufficient data on language variation to support my
conclusions.
Discussion – all of the adjective checklists in psychology /psychiatry
‫לל לללל‬ ,‫לללל ללללל‬
Nouns derived from adjs
‫ לללללל ללללל ללללל‬SLI
‫לללללל ללללל לללל לל ללל‬
experts
References
Blackwell, A.A. (2005). Acquiring the English adjective lexicon: relationships with input
properties and semantic typology. Journal of Child Language. 32 (3). 535-562.
Bar-Ilan, L. & R. A. Berman (in press).. Developing register differentiation: The Latinate-
Germanic divide in English. Linguistics.
Berman, R.A. (1994). Formal, lexical, and semantic factors in the acquisition of Hebrew
resultative particles. In S. Gahl, A. Dolbey, & C. Johnson (eds.) BLS 20, 82-92.
Berman, R.A. (2005). Editor: Journal of Pragmatics, 37,2. Special Issue on Developing
discourse stance across adolescence.
Berman, R.A. & B. Nir-Sagiv. 2004. Linguistic indicators of inter-genre differentiation
in later language development. Journal of Child Language, 31, 339-380.
Berman, R.A., B. Nir-Sagiv, & L. Bar-Ilan. In press. Vocabulary development across
adolescence: Text-based analyses. In I. Kupferberg & A. Stavans, eds.
Language education in Israel: Papers in honor of Elite Olshtain. Jerusalem:
Magnes Press.
Blodgett, E.G., & E.B. Cooper. (1988). Talking about it and doing it: metalinguistic
capacity and prosodic control in three to seven year olds. Journal of Fluency
Disorders 13, 283-290.
Bybee, J. In press. From usage to grammar: The mind’s response to repetition. Language.
Chafe, Wallace L. 1982. Integration and involvement in speaking, writing, and oral
literature. in Deborah Tannen, Ed. Spoken and written language: exploring orality
and literacy. Norwood NJ: Ablex, 35-54.
Chafe, W. L. (1994). Discourse, consciousness, and time: The flow of language in
speech and writing. Chicago: Chicago University Press.

Dixon, R. M. W. & Aikhenvald, A.Y. (eds.) 2004. Adjective classes: A cross-linguistic
typology. NY: Oxford University Press.
Elman, J.L. In press. An alternative view of the mental lexicon. Trends in Cognitive
Science.
Englebretson, R. 1997. Genre and grammar: predicative and attributive adjectives in
spoken English. Berkeley Linguistics Society, 23, 411-421.
Ferris, C. (1993). The meaning of syntax: A study in the adjectives of English. Longman:
London.
Hall, D.G. & Waxman, S.R. & Hurwitz, W.M. 1993. How 2-year-old and 4-year-old
children interpret adjectives and count nouns. Child Development, 64, 1651-1664.
Harris, P.L., J.E. Morris & M.M. Terwogt. (1986). The early acquisition of spatial
adjectives: a cross-linguistic study. Journal of Child Language 13, 335-352.
Kallio, K.D. (1988). Developmental differences in the comprehension of simple and
compound comparatives relations. Child Development 59, 397-410.
Klibanoff, R.S. & Waxman, S.R. 2000. Basic level objects categories support the
acquisition of novel adjectives: Evidence from preschool-aged children. Child
Development, 71, 649-659.
Manders, K. & Hall, D.G. 2002. Comparison, basic-level categories, and the
teaching of adjectives. Journal of Child Language, 29, 923-937.
Mintz, T.H. 2005. Linguistic and conceptual influences on adjective acquisition in 24-
and 36-month-old. Developmental Psychology, 41, 17-29.
Mintz, T.H., & Gleitman, L.R. 2002. Adjectives really do modify nouns: The incremental
and restricted nature of early adjective acquisition. Cognition, 84, 267-293.

Mintz, T.H., Newport, E.L., & Bever, T.G. 2002. The distributional structure of
grammatical categories in speech to young children. Cognitive Science, 26, 393-
424.
Nadig, A., Sedivy, J., Joshi, A., & Bortfeld, H. 2002. The development of discourse
constraints on the interpretation of adjectives. In Beachley, B., Brown, A., &
Conlin, F. BUCLD 27, pp. 568-579. Boston: Cascadila Press.
Ninio, A. 2004. Young children's difficulty with adjectives modifying nouns. Journal of
Child Language, 31, 255-285.
Nir, M. 1997. Adjective development in Hebrew-speaking children: a naturalistic study of
children aged 2-5. MA thesis, the Department of Communications Disorders, Tel
Aviv University, Israel.
Prasada, S. & Cummins, M. 2001. Structural Constraints on the Interpretation of Novel
Count Nouns. In Do, A. H.-J., Domínguez, L. & Johansen, A. (eds) BUCLD 25,
pp. 623-632. Boston: Cascadila Press.
Polinsky, M. 2004. Word class distinctions in an incomplete grammar. In D. Ravid & H.
Bat Zeev Shyldkrot (eds.). Perspectives on language and language development.
Dordrecht: Kluwer, 419-434.
Ravid, D. & M. Nir. 2000. On the development of the category of adjective in Hebrew. 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.
Richards, M.M. (1979). Adjective ordering in the language of young children: an
experimental investigation. Journal of Child Language 6, 253-277.

Sandhofer, C.M., Smithm, L.B., & Luo, J. 2000. Counting nouns and verbs in the input:
differential frequencies, different kinds of learning? Journal of Child Language,
27, 561-685.
Sedivy, J.C., Tanenhaus, M.K., Chambers, C.G., & Carlson, G.N. 1999. Achieving
incremental semantic interpretation through contextual representation. Cognition,
71,109–147.
Shlesinger, Y. & D. Ravid. 2003. Distribution of adjectives in expert and non-expert
texts: A linguistic, cognitive and developmental perspective. Paper presented at
the AAAL meeting, Arlington University, Virginia, USA, March 22-25.
Strömqvist, S., Nordqvist , Å., & Wengelin, Å. (2004). Writing the frog story:
developmental and cross-modal perspectives. In S. Strömqvist & L. Verhoeven
(Eds.), Relating events in narrative: Typological and contextual perspectives
(pp.359-394). Mahwah, NJ: Lawrence Erlbaum.
Thompson, Sandra A. 1988. A discourse approach to the category adjective.
In John A. Hawkins, ed. Explaining language universals. 167-210. Oxford:
Blackwell.
Valian, V. (1986). Syntactic categories in the speech of young children. Developmental
Psychology 22, 562-579.
Waxman, S.R. & Markow, D.B. 1998. Object properties and object kind: Twenty one
month old infants’ extension of novel adjectives. Child Development, 69, 1313-
1329.
Yuill, N. 1992. Children’s production and comprehension of trait terms. British Journal
of Developmental Psychology 10, 131-142.

Figure 1. Interaction of age group and genre in mean clause length
Figure 2. Interaction of age group and genre in content words per clause
Figure 3. Interaction of modality and genre in content words per clause
Figure 4. Interaction of age group and modality in adjectives per clause
Figure 5. Interaction of modality and genre in adjectives per clause
Figure 6. Interaction of age group and modality in Category 1 core adjectives per clause
Figure 7. Interaction of age group, genre and modality in Category 3 verbal / nominal
adjectives per clause

Figure 8. Interaction of age group, genre and modality in Category 3 verbal / nominal
adjectives per clause

Figure 4. Interaction of age group and modality in percentage of adjectives as NP heads
Age group / Genres # Words # Clauses Mean clause length
Spoken Written Spoken Written Spoken Written
9-10 Narratives 96.56 74.19 22.44 18.44 4.47 4.25
(51.36) (35.59) (12.47) (10.3) (0.83) (0.75)
Expositories 59.94 48.31 14.56 10.25 4.15 4.68
(25.8) (28.03) (6.14) (5.27) (0.79) (1.29)
12-13 Narratives 118.87 108.00 27.5 24.88 4.37 4.45
(76.26) (52.69) (18.02) (13.19) (0.62) (0.71)
Expositories 99.75 81.5 21.81 16.88 4.81 5.04
(54.02) (25.9) (13.6) (6.25) (0.60) (1.00)
16-17 Narratives 108.62 104.19 25.00 22.69 4.49 4.55
(100.85) (68.48) (24.48) (14.85) (0.61) (0.65)
Expositories 84.56 78.31 16.13 14.94 5.77 5.64
(57.19) (48.93) (11.74) (10.25) (2.16) (1.64)
Adults Narratives 218.53 154.13 48.33 32.73 4.64 4.87
(105.00) (107.62) (25.76) (25.11) (0.63) (0.85)
Expositories 199.27 163.33 34.53 26.47 5.67 6.47
(115.66) (79.93) (17.36) (12.33) (0.91) (1.91)
Table 1. Means and standard deviations of number of words, number of clauses, and
mean clause length (words per clause)
Age group / Genres Content words per clause Adjectives per clause
Spoken Written Spoken Written
9-10 Narratives 2.07 2.07 0.11 0.12
(0.47) (0.54) (0.12) (0.16)
Expositories 2.01 2.76 0.41 0.49
(0.48) (0.9) (0.24) (0.25)
12-13 Narratives 1.91 2.01 0.17 0.16
(0.36) (0.39) (0.11) (0.14)
Expositories 2.37 3.05 0.45 0.54
(0.55) (0.84) (0.34) (0.24)
16-17 Narratives 2.01 2.19 0.22 0.24
(0.36) (0.47) (0.16) (0.2)
Expositories 2.78 3.55 0.44 0.7
(1.26) (1.5) (0.27) (0.61)
Adults Narratives 2.22 2.26 0.23 0.45
(0.57) (0.58) (0.19) (0.25)
Expositories 3.02 4.26 0.45 0.79
(0.87) (1.48) (0.19) (0.31)
Table 2. Mean number of content word and adjective tokens per clause, by age group,
genre and modality
Age group / Genres Spoken Written
9-10 Narratives 0.54 0.39 (0.36)

(0.52)
Expositories 1.06 1.01
(0.58) (0.75)
12-13 Narratives 1.05 0.89 (0.79)
(1.04)
Expositories 1.56 1.7 (0.89)
(0.84)
16-17 Narratives 0.86 0.94 (0.89)
(0.91)
(1.02) (0.66)
Adults Narratives 2.05 2.63 (1.93)
(1.15)
Expositories 2.99 3.93 (2.0)
(1.58)
Table 3. Mean score on morpho-semantic adjective scale, by age group, genre and
modality (maximum score: 4).
Age group / Genres Spoken Written
9-10 Narratives 1.13 1.19 (1.05)

(0.81)
(0.77) (1.15)
12-13 Narratives 1.75 2.13 (1.59)
(1.18)
(0.82) (0.81)
16-17 Narratives 1.75 2.19 (1.17)
(1.06)
(1.03) (0.77)
Adults Narratives 3.00 3.73 (0.46)
(0.85)
(0.72) (0.74)
Table 4. Mean number of adjective categories per text, by age group, genre and modality
Age group / Genres Category 1 Category 2 Category 3 Category 4
Core adjectives Resultative Verbal / nominal Denominals
participials patterns
Spoken Written Spoken Written Spoken Written Spoken Written
9-10 Narratives 0.09 0.07 0.02 0.01 0.01 0.03 0.001 0.01
(0.11) (0.14) (0.04) (0.02) (0.02) (0.06) (0.005) (0.03)
Expositories 0.28 0.29 0.02 0.02 0.07 0.05 0.05 0.13
(0.19) (0.22) (0.04) (0.04) (0.07) (0.07) (0.08) (0.17)
12-13 Narratives 0.1 0.08 0.02 0.02 0.02 0.03 0.02 0.04
(0.08) (0.07) (0.04) (0.03) (0.04) (0.03) (0.05) (0.06)
Expositories 0.18 0.21 0.03 0.02 0.1 0.15 0.14 0.17
(0.12) (0.14) (0.07) (0.03) (0.11) (0.12) (0.23) (0.23)
16-17 Narratives 0.08 0.07 0.05 0.06 0.05 0.07 0.04 0.04
(0.12) (0.08) (0.11) (0.06) (0.08) (0.1) (0.04) (0.08)
Expositories 0.14 0.16 0.03 0.05 0.13 0.24 0.14 0.25
(0.16) (0.15) (0.04) (0.09) (0.13) (0.21) (0.25) (0.49)
Adults Narratives 0.1 0.17 0.04 0.06 0.06 0.12 0.03 0.09
(0.07) (0.13) (0.04) (0.04) (0.04) (0.1) (0.04) (0.08)
Expositories 0.13 0.08 0.04 0.06 0.17 0.37 0.1 0.28
(0.09) (0.08) (0.04) (0.06) (0.11) (0.12) (0.12) (0.29)
Table 5. Mean tokens per clause and standard deviations of adjectives in different morpho-semantic categories, by age group, genre
and modality
Age group / Genres Category 1 Category 2 Category 3 Category 4
Core adjectives Resultative Verbal / nominal Denominals
participials patterns
9-10 Narratives 79.44 58.13 9.72 11.25 8.33 28.13 2.5 2.5
(20.3) (33.16) (18.25) (16.2) (17.82) (36.44) (7.07) (7.07)
Expositories 63.61 51.7 6.25 5.12 19.72 18.04 10.42 25.15
(21.09) (30.35) (12.4) (7.92) (19.05) (16.41) (17.68) (34.3)
12-13 Narratives 59.29 50.01 10.37 10.06 19.71 16.35 10.63 23.57
(33.61) (32.28) (18.77) (11.47) (30.6) (12.77) (18.44) (24.61)
Expositories 50.11 35.99 7.22 5.91 22.53 38.62 20.16 19.48
(54.02) (23.98) (13.8) (14.94) (14.23) (24.1) (26.77) (24.68)
16-17 Narratives 32.74 39.85 31.57 26.06 18.6 24.85 17.09 9.24
(40.53) (34.16) (39.67) (23.13) (32.5) (23.63) (25.79) (14.01)
Expositories 32.49 27.68 4.07 10.88 32.67 36.09 30.77 25.36
(24.15) (20.77) (9.97) (15.12) (26.59) (23.98) (31.12) (29.09)
(20.46) (18.37) (23.34) (11.28) (16.39) (14.81) (17.92) (11.01)
Expositories 30.52 11.38 10.14 7.34 38.95 50.03 20.39 31.25
(17.13) (10.82) (9.7) (7.12) (16.19) (13.75) (20.07) (18.05)
Table 6. Mean percentages and standard deviations of the different adjective categories, out of all adjective tokens, by age group,
genre and modality
Age group / Genres Predicative Attributive Adverbial NP head
9-10 Narratives 0.06 0.03 0.04 0.07 0 0.01 0.16 0.18
(0.07) (0.04) (0.05) (0.14) (0.01) (0.58) (0.63)
Expositories 0.28 0.28 0.13 0.21 0 0 0 0
(0.21) (0.17) (0.07) (0.19)
12-13 Narratives 0.09 0.04 0.08 0.12 0 0.002 0.002 0.01
(0.06) (0.05) (0.08) (0.12) (0.01) (0.01) (0.02)
Expositories 0.16 0.16 0.28 0.34 0 0.003 0.01 0.39
(0.1) (0.1) (0.29) (0.23) (0.01) (0.28) (0.84)
16-17 Narratives 0.1 0.06 0.1 0.16 0.01 0 0.02 0.02
(0.09) (0.06) (0.11) (0.17) (0.02) (0.05) (0.03)
Expositories 0.21 0.25 0.21 0.4 0.01 0.01 0.01 0.03
(0.17) (0.2) (0.25) (0.46) (0.03) (0.03) (0.03) (0.13)
(0.05) (0.06) (0.07) (0.2) (0.02) (0.02) (0.01) (0.03)
Expositories 0.19 0.29 0.24 0.49 0.01 0 0.01 0.01
(0.14) (0.35) (0.16) (0.25) (0.02) (0.02) (0.02
Table 7. Mean adjectives per clause and standard deviations of adjectives in the four syntactic positions, by age group, genre and
modality
Age group / Genres Predicative Attributive Adverbial NP head
9-10 Narratives 42.22 51.25 54.65 39.34 0 3.13 3.13 6.25
(30.03) (30.09) (32.18) (32.34) (8.84) (8.84) (17.68)
Expositories 63.61 57.74 36.39 42.26 0 0 0 0
(24.71) (32.82) (21.71) (32.82)
12-13 Narratives 52.71 27.86 45.83 65.16 0 1.3 1.47 5.68
(32.6) (33.68) (32.69) (32.29) (4.31) (3.53) (10.25)
Expositories 35.59 40.26 62.76 48.17 0 1.3 1.65 10.27
(26.57) (28.72) (25.29) (25.89) (4.31) (5.48) (6.67)
16-17 Narratives 53.13 33.94 28.13 57.27 4.55 0 14.2 8.79
(27.55) (28.59) (24.36) (28.32) (15.08) (21.75) (17.15)
Expositories 52.67 50.51 47.33 47.97 0 1.52 0 0
(27.83) (32.47) (27.83) (32.83) (5.03)
(21.01) (13.2) (22.3) (16.05) (9.9) (4.37) (3.7) (7.17)
Expositories 41.13 32.79 54.51 65.48 1.81 0 2.55 1.74
(27.16) (23.73) (26.88) (24.08) (4.17) (6.83) (2.72)
Table 8. Mean percentages and standard deviations of adjectives in the four syntactic positions, by age group, genre and modality
Age group / Genres Conjoining Stacking Compound head
9-10 Narratives 0 0 0 0 0 0
Expositories 0 0 0 0.004 0 0
(0.01)
12-13 Narratives 0.005 0.002 0 0 0 0
(0.01) (0.007)
Expositories 0.02 0.02 0.01 0.005 0.003 0.002
(0.07) (0.03) (0.03) (0.03) (0.08) (0.008)
16-17 Narratives 0.008 0.02 0 0.007 0.003 0
(0.02) (0.06) (0.02) (0.01)
Expositories 0.008 0.01 0 0.02 0 0
(0.02) (0.03) (0.06)
Adults Narratives 0.004 0.02 0 0.006 0.004 0.008
(0.009) (0.02) (0.02) (0.01) (0.02)
Expositories 0.01 0.05 0.007 0.01 0.01 0.004
(0.03) (0.05) (0.02) (0.03) (0.04) (0.01)
Table 9. Mean tokens per clause and standard deviations of adjectives in three different
complex structures, by age group, genre and modality
Age group / Genres Basic Advanced Multiple
9-10 Narratives 0.03 0.03 0 0 0.004 0
(0.04) (0.07) (0.06)
Expositories 0.04 0.04 0.03 0.03 0.07 0.07
(0.08) (0.07) (0.06) (0.08) (0.11) (0.1)
12-13 Narratives 0.04 0.03 0.002 0 0.003 0
(0.04) (0.04) (0.09) (0.01)
Expositories 0.04 0.06 0.01 0.04 0.06 0.11
(0.07) (0.09) (0.04) (0.08) (0.07) (0.12)
16-17 Narratives 0.06 0.02 0 0.003 0.001 0
(0.06) (0.03) (0.01) (0.003)
Expositories 0.04 0.06 0.02 0.04 0.06 0.09
(0.08) (0.1) (0.04) (0.06) (0.1) (0.11)
Adults Narratives 0.05 0.03 0.002 0.009 0 0
(0.06) (0.05) (0.004) (0.02)
Expositories 0.06 0.05 0.05 0.07 0.12 0.11
(0.04) (0.05) (0.06) (0.06) (0.09) (0.09)
Table 10. Mean tokens per clause and standard deviations of adjectives by AP
modification type, age group, genre and modality
Age group / Genres Basic Advanced Multiple
9-10 Narratives 26.6 27.5 0 0 2.78 0
(27.42) (21.04) (7.86)
Expositories 14.38 15.86 0 0 23.06 17.95
(19.98) (20.61) (24.85) (19.59)
12-13 Narratives 26.1 20.18 8.68 2.08 0.46 0
(18.95) (28.68) (17.28) (5.89) (1.51)
Expositories 8.92 9.27 1.82 3.55 11.95 12.83
(8.72) (11.9) (6.03) (6.66) (10.85) (13.04)
16-17 Narratives 22.51 3.64 0 1.51 0.51 0
(18.11) (6.9) (5.03) (1.68)
Expositories 11.23 9.27 3.28 6.6 14.51 15.87
(15.08) (12.06) (7.94) (10.34) (17.92) (17.17)
Adults Narratives 18.89 8.36 0.95 1.87 0 0
(17.5) (15.48) (2.54) (3.58)
Expositories 14.46 5.64 10.32 8.56 24.78 14.2
(9.15) (5.58) (12.36) (6.4) (19.25) (8.41)
Table 10. Mean percentages and standard deviations of adjectives by AP modification

type, age group, genre and modality
Robert Englebretson (reng@ruf.rice.edu) directed me tot he following three
references:
- Chafe, Wallace L. 1982. Integration and involvement in speaking, writing,
and oral literature. in Deborah Tannen, Ed. Spoken and written language:
exploring orality and literacy. 35-54. Norwood NJ: Ablex Publishing Corp.
- Thompson, Sandra A. 1988. A discourse approach to the category adjective.
In John A. Hawkins, ed. Explaining language universals. 167-210. Oxford:
Blackwell.
- Englebretson, Robert. 1997. Genre and grammar: predicative and attributive
adjectives in spoken English. BLS 23:411-421.
Also, he provided the following brief, but helpful, summary/discussion,
which I therefore quote in toto:
"Chafe's and Thompson's findings appear to be contradictory (Chafe reports a
nearly 2:1 ratio of attributive over predicative in a small corpus of
conversational English, while Thompson reports a nearly 2:1 ratio of
predicative over attributive in another conversational corpus). In my 1997
paper I argue that these findings are not contradictory, but in fact reflect
characteristics of the particular genre of conversational language each
researcher is looking at. The distribution of predicative and attributive
adjectives is closely tied to genre factors, and reflects the different
discourse functions for which they are used. I found roughly even
distribution in a larger corpus of conversational English (a subset of the
Santa Barbara Corpus of Spoken American English), but when viewed in terms
of individual speech events, I found the skewed distribution noted by both
Chafe and Thompson. In a nutshell, the determinants of which is more
frequent has to do with genre, but ultimately with the amount of "new"
versus "shared" information among the speech participants in the
interaction."
Finally, Hans Lindquist (Hans.Lindquist@hum.vxu.se) pointed me to the

following source:
Lindquist, Hans, 2000. Livelier or more lively? Syntactic and contextual
factors influencing the comparison of disyllabic adjectives. John M. Kirk
(ed.) Corpora galore. Analyses and techniques in describing English.
Amsterdam: Rodopi.
Anderson, S.R. (1985)
Inflectional morphology. In: T. Shopen (ed.) Language typology and syntactic
description, Vol III Grammatical categories and the lexicon. Cambridge:
Cambridge University Press, 150-201.
Barrett, M. (1995)
Early lexical development. In: P. Fletcher & B. MacWhinney (eds.) The
handbook of child language. Oxford: Blackwell, 362-392.
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. Hillsdale,
NJ: Erlbaum.
Berman, R.A. (1994)
Formal, lexical, and semantic factors in the acquisition of Hebrew resultative
particles. In S. Gahl, A. Dolbey, & C. Johnson (eds.) BLS 20, 82-92.
Berman, R.A. (to appear)
From known to new: how children coin nouns compared with verbs in Hebrew. In
L. Menn & N. Bernstein-Ratner (eds.) Jean Berko Gleason Festschrift.
Blodgett, E.G., & E.B. Cooper. (1988)
Talking about it and doing it: metalinguistic capacity and prosodic control in three
to seven year olds. Journal of Fluency Disorders 13, 283-290.
Casseli, M.C., E. Bates, P. Casadio & J. Fenson. (1995)
A cross-linguistic study of early lexical development. Cognitive Development 10,
159-199.
Chafe, Wallace L. 1982. Integration and involvement in speaking, writing,
and oral literature. in Deborah Tannen, Ed. Spoken and written language:
exploring orality and literacy. 35-54. Norwood NJ: Ablex Publishing Corp.
Clark, E.V. & R.A. Berman. (1984).
Structure and use in the acquisition of word formation. Language 60, 542-590.
Dixon, R.M.W. (1977)
A grammar of Yidin. Cambridge: Cambridge University Press.
Dromi, E. (1987)
Early lexical development. Cambridge: Cambridge University Press.
Ferris, C. (1993).
The meaning of syntax: A study in the adjectives of English. Longman: London.
Gai, A. (1995).
The category “adjective” in Semitic languages. Journal of Semitic Studies 1, 1-9.
Gesenius, (1910)
Gesenius' Hebrew Grammar, edited by E. Kautzsch, revised by A.E. Cowley.
Oxford: Clarendon Press.
Harris, P.L., J.E. Morris & M.M. Terwogt. (1986).
The early acquisition of spatial adjectives: a cross-linguistic study. Journal of
Child Language 13, 335-352.
Kallio, K.D. (1988)
Developmental differences in the comprehension of simple and compound
comparatives relations. Child Development 59, 397-410.
Langacker, R.W. (1991)
Concept, image and symbol: the cognitive basis of grammar. Berlin: Mouton de
Gruyter.
Marvin, C.A., D.R. Beukelman & D. Bilyeu. (1994) Vocabulary-use patterns in preschool
children: effects of context and time sampling. Augmentative and Alternative
Communication 10, 224-237.
Ninio, A. (1988)
On formal grammatical categories in early child language. n: Y. Levy, I.M.
Schlesinger & M.D.S. Braine (eds.) Categories and processes in language
acquisition. Hillsdale, NJ: Erlbaum.
Ravid, D. (1997)
Between Syntax and the Lexicon: the parallel between N-N compounds and N-A
strings in acquisition. In A. Sorace, C. Heycock & R. Shillcock (eds.)
Proceedings of the GALA ‘97 conference on language acquisition. Edinburgh:
University of Edinburgh, 138-141.
Rice, M.L. (1990)
Preschoolers’ QUIL: Quick incidental learning of words. In: G. Contini-Ramsden
& C. Snow (eds.) Children’s language, Vol 7. Hillsdale, NJ: Erlbaum.
Richards, B.J. & D.D. Malvern. (1997)
Quantifying lexical diversity in the study of language development. Reading: the
University of Reading.
Richards, M.M. (1979).
Adjective ordering in the language of young children: an experimental
investigation. Journal of Child Language 6, 253-277.
Schachter, P. (1985)
Parts-of-speech systems. In: T. Shopen (ed.) Language typology and syntactic
description, Vol I Clause structure. Cambridge: Cambridge University Press, 3-
61.
Sommers, R.K., M. Kozarevich, & C. Michaels. (1994)
Word skills of children normal and impaired in communication skills and
measures of language and speech development. Journal of Communications
Disorders 27, 223-240.
Thompson, S.A. (1988)
A discourse approach to the cross-linguistic category ‘adjective’. In Universals.
Oxford: Oxford University Press.
Valian, V. (1986)
Syntactic categories in the speech of young children. Developmental Psychology
22, 562-579.
Yuill, N. (1992)
Children’s production and comprehension of trait terms. British Journal of
Developmental Psychology 10, 131-142.
Instructions to Contributors: « click to expand/collapse
Language Learning and Development publishes refereed and invited articles reporting on
research. Submitted manuscripts should be double-spaced-including title page, text,
tables, figures, references, notes, and appendixes-and must adhere to the guidelines of the
Publication Manual of the American Psychological Association (5th ed.). The first page
should include the title, name(s), and affiliation(s) of author(s) and full contact addresses
for correspondence (including e-mail). The second page should include an abstract of not
more that 150 words. The text of the manuscript should begin on the third page, without
repeating the title. Use either American or British spelling consistently within an article.
Manuscripts should normally be no more than 25 to 30 double-spaced pages (including
references, notes, and tables). Minimize the number of notes. Tables and figures should
be placed after the references, each on a separate page with an indication as to where in
the text where they occur.
Submissions should be emailed to lld@uchicago.edu. Include in the email the title of the
paper, the authors, and the abstract. The preferred form of submission of the manuscript
is a PDF file including all tables and figures, with any special fonts (e.g., IPA) embedded.
It is also possible to submit the manuscript and tables as a Word .doc or .rtf file, with
figures in a separate .jpg or .gif file. After acceptance, a final version of the article will be
required on diskette/CD as well as in hard copy.
LL&D

Adjectives in Text Production

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Adjectives in Text Production

Uploaded by

Copyright:

Available Formats

January 28, 2006, Antwerp

March 26, 2006, Home

primary content-word class in both typological and psycholinguistic perspectives, and

universality and proto-typicality, as defined by a number of criteria. Semantically, a

1984). Finally, a lexical class may also have language-specific morphological

characteristics. Typical members of the class of lexical categories fulfill these

markers and verbs decline in tenses.

1.1 Adjectives: a typological and psycholinguistic perspective

Adjectives constitute a third content-word class, which is less primary in a

adjectives have differential discourse functions in spontaneous conversation, together

adjust their form to correspond to its inflection (Markman, 1989).

discourse functions in spontaneous conversation with nouns and verbs.

In psycholinguistic terms, the representation of adjectives in the mental lexicon is

richly structured categories (1989:116-135). Adjectives, in contrast, are less dense in

specifying subdivisions within a richer category along many different dimensions. It

adjectives. The interdependence of nouns and adjectives is demonstrated in Sedivy,

help of contextual clues in cases of referential indeterminancy.

Markman (1985) report an experimental study of noun and adjective interpretation in

When asked to interpret adjectives, children tended to focus on a contrast between

distinctive exemplar of the category.

The secondary nature of adjectives is demonstrated in a set of experiments

A recent spurt of experimental studies with young children and toddlers

claim that object construal, which is enabled by taxonomically specific nouns, is a

Finally, children, like adults, assign contrastive functions to adjectives in the

interpretation of noun phrases (Prasada & Cummins, 2000).

But adjectives are interesting in and by themselves in the domain of language

acquisition. As a lexical class with universal and language-specific semantic, syntactic

categorization and deserve to be investigated from a developmental perspective. Being

child language development should serve as a window on aspects of lexical acquisition

canonical lexical classes.

One obvious direction of investigation would be to examine the rate of acquisition

thrust of investigation would be to examine the development of the various semantic

classes expressed by adjectives, such as dimension, color, internal and external

properties. In languages with rich morphology, this analysis would be supplemented by

an examination of morphological patterns in adjective learning. Adjectives also constitute

Studies of adjective acquisition have focused to date on early language

Valian, 1986). Nevertheless, after an early phase of acquiring predominantly common

the insight of creating higher-order complex units. Adjective comprehension is especially

A systematic and detailed study by Blackwell (2005) investigated the adjective

example, a universal hierarchy has been found in the acquisition of dimensional

adjectives express finer semantic distinctions and eventually metaphorical meanings

(Blodgett & Cooper, 1988).

and physical property adjectives, as in Blackwell’s English study. Later emerging

and syntax – were highly correlated in all age groups.

denominal adjectives of Latinate origin such as informative or industrial are not in

denominal and reduplicated adjectives are late-emerging morphological categories with a

would rectify this lacuna in our knowledge.

Moreover, in later language development, it seems desirable to investigate

expository texts from lexical, semantic, morphological and syntactic perspectives.

1.3 Hebrew adjectives

the roots of present-day adjectival derivation in Hebrew.

Modern Hebrew has four major structural classes of adjectives: One, an

essentially closed class of primary CVC adjectives originating in Biblical present-tense

complexity due to their frequency in CDS (Ravid, 1995). Resultative adjectives, in

A third structural device is reduplication, which is used mostly for adjective

adjectives in Modern Hebrew is a late historical development deriving from those

denominal adjectives, e.g., xashmal-i ‘electr-ic’, tsibur-i ‘publ-ic’. Structurally,

2.0 DATABASE, ANALYSES, AND PREDICTIONS

expositions about school violence.

rhetorical / expressive devices and grammatical constructions in non-expert text

2.1 Text analyses

while for clauses we follow the definition of a clause as “a unified predication” in

schooling, and in particular in written and expository texts.

2.1.1 Adjective categories