Elsevier

Cognition

Volume 64, Issue 1, July 1997, Pages 39-72
Cognition

The representation of Hebrew words: Evidence from the obligatory contour principle

https://doi.org/10.1016/S0010-0277(97)00016-4Get rights and content

Abstract

The Hebrew root morpheme typically consists of three consonants. Hebrew allows a gemination of a root consonant, but constrains its location [McCarthy (1979). Formal problems in semitic phonology and morphology. Cambridge, MA; MIT Ph.D. dissertation. Distributed by Indiana University Linguistics Club. Garland Press, New York, 1985]. A gemination of a root-consonant is permitted at the end of the root (e.g., [mss]), but not at its beginning (e.g., [ssm]). Two experiments examined readers' sensitivity to the structure of the root morpheme by obtaining ratings for nonwords derived from nonroots. Root-initial gemination (e.g., [ssm]) was judged unacceptable compared to root-final gemination (e.g., [mss]) or no gemination controls (e.g., [psm]). The sensitivity to root structure emerged regardless of the position of the root in the word. These results have several implications. (1) Our findings demonstrate morphological decomposition. Hebrew speakers' ratings reflect a phonological constraint on the location of geminates. Being the domain of this constraint, the root morpheme must form a separate constituent in the representation of Hebrew words. (2) The rejection of root-initial gemination supports the psychological reality of the Obligatory Contour Principle, a pivotal constraint in autosegmental phonology. (3) A sensitivity to the location of geminates presupposes a distinction between the representation of geminate and nongeminate bigrams. Such a distinction, however, requires the implementation of a symbol. Our findings converge with numerous linguistic evidence in suggesting that the representation of constituency structure is necessary to account for linguistic generalizations.

Introduction

There is a considerable body of evidence demonstrating the sensitivity of speakers of a variety of languages to the morphological structure of their native language (e.g., Caramazza et al., 1988; Emmorey, 1995; Feldman, 1994; Fowler et al., 1985; Grainger et al., 1991; Gordon, 1989; Kim et al., 1991; Marcus et al., 1995; Marslen-Wilson et al., 1994). However, there is an ongoing debate regarding the representational properties that must be postulated in order to account for this sensitivity. At the heart of the debate is the question of whether an account for speakers' morphological knowledge requires an explicit decomposition of words onto their formal constituents (e.g., Fodor and Pylyshyn, 1988; Pinker, 1991; Pinker and Prince, 1988), or, rather could this knowledge be explained without recourse to explicit formal constraints on word structure, in fact, without postulating explicit lexical entries at all (Rumelhart and McClelland, 1986; Seidenberg, 1987; Seidenberg and McClelland, 1989). On this latter view, morphological knowledge could be fully explained in terms of the co-occurrence of phonological, orthographic, and semantic properties of morphological units without assuming any formal morphological distinctions per sé. Indeed, most of the existing evidence for morphological compositionality comes from languages whose morphology is concatenative. In those languages, morphologically complex words may be formed by affixation without interrupting the integrity of the base morpheme. Consequently, morphemes tend to correspond to discrete orthographic and phonological units that are often associated with some well defined semantic features. An empirical dissociation of the contribution of formal units, morphemes, from that of their nonmorphological correlates is not easily achieved (Stemberger, 1995).

In the present research, we examine evidence for morphological compositionality in Hebrew, a language whose morphology is nonconcatenative. In this language, distinct morphemes are often interwoven, temporally co-occurring units. The root morpheme is not a linearly discrete unit on either the orthographic or the phonological dimension. Morphological structure in Hebrew is thus fairly opaque. Yet, we demonstrate that Hebrew speakers are sensitive to a phonological constraint that specifically concerns the structure of the root morpheme. The domain to which this constraint applies, the set of all possible Hebrew roots, is very large, and cannot be defined by any orthographic, phonological or semantic features. Despite the fact that Hebrew roots do not share any distinctive nonmorphemic features, subjects nevertheless treat them as a single linguistic class that is subject to a common constraint. This finding, we believe, supports the view of the root morpheme as a discrete constituent in the representation of Hebrew words.

Before describing our evidence, a brief exposition of some of the central properties of Hebrew morphology is in order. Two of the properties of Hebrew morphology, its productivity and nonlinearity, are potentially important for the representation of morphological constituency. We then review existing empirical evidence for morphological decomposition in Hebrew. Finally, we describe the Obligatory Contour Principle as a means for investigating the structure of morphologically complex words in Hebrew.

Like other Semitic languages, Hebrew morphology is both highly productive and nonlinear. Hebrew words are formed by inserting a root morpheme, an abstract sequence of generally 3 consonants, in a word pattern containing vowels, and sometimes, additional consonants. The root itself is not an independent word, but it may be realized in several words that are morphologically (and often semantically) related. These words are generated by inserting the root into one of several verbal and nominal word patterns, called binyanim and mishkalim, respectively. For instance, the root [ktb] conveys the general meaning of writing. This root may form the verb katav, he wrote, by inserting it into a prosodic pattern that is characteristic of the third person, masculine, singular, past tense form in binyan qal (i.e., C1aC2aC3; see Table 1). Similarly, the noun mixtav, letter, is formed by inserting the root [ktb] in a miC1C2aC3 noun template. Each of these words, in turn, may further be subjected to a rich inflectional system. Specifically, the root [ktb] may be conjugated in four of the seven verb patterns, binyanim, and each such binyan could yield approximately 30 inflectional forms (Aronoff, 1994, p. 124). Thus, the conjugation of the verb [ktb] alone yields about 120 distinct, morphologically related, words1.

This brief example is sufficient to illustrate the productivity of Hebrew morphology. In view of this rich productivity, a representation of Hebrew words in reference to their common root would appear to save considerable rote learning and storage space. However, a second characteristic of Hebrew morphology, its nonlinearity, may prove an obstacle for morphological decomposition. In Hebrew, the root morpheme and the word pattern are not linearly discrete units. Instead, they are interwoven, temporally co-occurring entities. Thus, root consonants are often interrupted by a series of vowels, and sometimes, additional nonroot consonant, provided by the prosodic template. The structure of Hebrew words is well captured by nonlinear autosegmental theories of phonology. Autosegmental theories of phonology represent phonological constituents on distinct levels of representation, i.e., planes. These planes are interconnected by the skeleton, a sequence of timing units. In our example, the root [ktb] is represented on a single plane, whereas the vowels are represented on a separate plane. These planes are interconnected by the skeleton, which specifies the word patterns of katav and mixtau (see Fig. 1).

This representation neatly captures the fact that, despite their temporal discontinuity, root consonants and vowels each form distinct morphemic constituents. At the same time, however, it illustrates the potential problems posed for the processing of morphological structure in Hebrew. Because Hebrew morphemic constituents often do not correspond to linearly discrete units in either spoken or written language, the root morpheme cannot be easily parsed by simple cues of transitional probability, often available in concatenated morphologies (Seidenberg, 1987). The transparency of the root is further reduced by phonological processes that obscure the similarity between the root and its derivations. For instance, the words mixtav (letter), and katav (wrote), are both derived from the root [ktb], whose initial radical is realized as /x/ as a result of a spirantization rule (Glinert, 1989)2. Thus, not only does the root morpheme in Hebrew lack linear discreteness but it is also fairly opaque phonologically. How, then, do Hebrew readers represent morphologically complex words? In particular, does the representation of Hebrew words decompose the root and the word pattern?

Despite the opaqueness of morphological constituents in Hebrew, there is considerable empirical evidence suggesting morphological compositionality in Hebrew (for a review, see Bentin and Frost, 1992). Several recent studies specifically address the role of the root as a morphemic constituent among adult Hebrew speakers (Bentin and Feldman, 1990; Feldman and Bentin, 1994; Feldman et al., 1995; Frost et al., 1994). Underlying these works is a common rationale concerning the availability of the root to mental process. If the root consists of a distinct constituent in mental representations, then the root should be more readily available for processing compared to a sequence of consonants that is not a morphemic constituent. The processing availability of morphological constituents should further increase with the productivity of the root: Highly productive roots should be more easily decomposed from their word patterns compared to unproductive roots. In a segment shifting task, Feldman et al. (1995)instructed subjects to strip the word patterns from a Hebrew word and shift it to a novel root. Feldman et al. (1995)found that the productivity of the root facilitated performance, regardless of its orthographic transparency and surface frequency. This finding suggests that productivity increases the availability of both the root and the word pattern to mental processes.

A closely related argument for compositionality is based on the contribution of root priming to word processing. If the root is a mental constituent, then increasing its availability by priming should contribute to word processing more than a non-constituent sequence of consonants. Supporting this view are the results of several priming studies demonstrating that the priming of the target by a prime sharing the target's root facilitates processing. This facilitation was obtained regardless of the semantic relatedness between the prime and the target, and even when the target and the prime were separated by 15 items (Bentin and Feldman, 1990). In a subsequent study, Feldman and Bentin (1994)observed morphological priming for derivationally and inflectionally related targets and primes, similar in its magnitude to identity priming. Supporting the automaticity of these priming effects, a facilitation by word prime sharing the target's root was obtained regardless of the type of foils (Feldman and Bentin, 1994) and even when the prime was heavily masked (Frost et al., 1994).

These studies suggest that the root is more readily available to mental processes than non-root controls. A permanent or temporary increase in the availability of the root (due to high productivity or priming, respectively) facilitates word recognition. This finding is readily explained by assuming morphological decomposition. However, other explanations for the increased availability of the root are possible as well. Hebrew roots are strongly associated with a core meaning that is often shared with the target word. Furthermore, because the Hebrew orthography is consonantal, the root morpheme is often represented by a linearly discrete orthographic unit. Thus, in order to attribute the increased availability of the root to its morphological status, it is necessary to rule out the possibility that its greater availability is in fact due to orthographic, phonological and semantic factors. Although existing studies have invested considerable effort in ruling out such nonmorphemic factors, their empirical control cannot be easily achieved. The elimination of semantic, orthographic and phonological explanations for morphological effects is typically inferred from the manipulation of these variables in separate studies, rather than from directly controlling all variables within a single design (Stolz and Feldman, 1995). Evidence for the separate morphemic status of the root could thus be strengthened by the convergence of findings from distinct theoretical perspectives and empirical manipulations.

In the present study, we provide converging evidence for the representation of the root as a separate morpheme. The rationale guiding our demonstration is different than that of existing research. Rather than probing for the effect of morphological constituency on the ease of mental processing, we assess Hebrew speakers' tacit knowledge regarding the structure of the root morpheme. We exploit the fact that Hebrew roots are subject to a phonological constraint that is specifically sensitive to location within the root morpheme. Importantly, this constraint cannot be explained by merely referring to the location of the segment in the word or syllable. We consider this formal constraint on the structure of the root morpheme as evidence for morphological decomposition. If Hebrew speakers' knowledge of their native language entails a constraint that specifically concerns the root, then, in order to account for their competence, it is necessary to postulate that the root is a separate morphemic constituent.

The constraint in question concerns the gemination of root consonants. Semitic languages allow a gemination of adjacent root consonants, but constrain its location: A gemination of the second and third root consonants (e.g., [smm]) is frequent whereas a gemination of the first and second root consonants (e.g., [ssm]) is rare (Greenberg, 1950). In his seminal work, McCarthy (1979)provided an elegant explanation for this asymmetry within the framework of autosegmental phonology. McCarthy (1979)departed from the proposal that Semitic languages represent the root consonants on a separate autosegmental plane. He further assumed that the root plane is constrained by the Obligatory Contour Principle (OCP), a principle initially proposed by Leben (1973)in studying tonal languages and subsequently documented in a wide variety of phenomena (e.g., Goldsmith, 1990; Kenstowicz, 1994; McCarthy, 1986; Yip, 1988; Yip, 1989). The OCP prohibits the occurrence of adjacent identical elements on the same plane. A gemination of a root consonant is thus banned from underlying representations. The ubiquitous, root-final gemination, (e.g., [smm]), as well as the rare, root-initial gemination (e.g., [ssm]) are considered surface manifestations of a common underlying biconsonantal representation (e.g., [sm]). To derive a word, the root must be associated with a skeletal template that is characteristic of a particular verb or noun pattern, binayan or mishkal. Fig. 2 illustrates this process for the verb samam. First, the underlying representation of the root, [sm], is associated with the consonant skeletal positions. This association is assumed to proceed from left to right3. Hence, the two initial consonant positions are filled, leaving the third one empty. This empty slot is later filled by spreading of an adjacent consonant, resulting in a surface geminate. Importantly, please note that only the second consonant, /m/, is free to spread. The initial consonant /s/ cannot spread because its adjacent slot is already filled. A gemination of the first and second root consonants is indeed rare. In contrast, because the slot adjacent to the second consonant, /m/, is empty, it is free to spread, resulting in the frequent gemination of the second and third radicals, [smm].

Despite the pivotal role of the OCP in explaining a wide variety of linguistic phenomena, the psychological reality of this principle has not been corroborated yet in experimental settings. Thus, although the OCP is observationally adequate in accounting for the statistical distribution of different root types, it is unclear whether this principle actually reflects the linguistic competence of modern speakers of Semitic. Specifically, in the case of modern Hebrew, the psychological reality of the OCP is questioned by the existence of several violations of this principle. Even-Shoshan's New Hebrew dictionary (Even-Shoshan, 1993) lists 12 roots manifesting root-initial gemination, 3 of them are frequently used in modern Hebrew. Furthermore, a sensitivity to co-occurrence of root consonants presupposes the representation of the root as a separate morpheme: Note that the asymmetry predicted by the OCP constrains the adjacency of elements as a function of their morphological structure, not merely their temporal or spatial adjacency. As previously noted, however, the nonlinearity of Hebrew morphology may obscure morphological constituency. Thus, the opaque nature of the root morpheme and the presence of counter-examples may prevent modern Hebrew speakers from internalizing this phonological constraint on root structure.

The goal of the present research is to examine the sensitivity of Hebrew speakers to the constraint on root structure predicted by the Obligatory Contour Principle. We view the psychological reality of the OCP not only as an intriguing research question in its own right, but further, as a means for uncovering the morphological structure of Hebrew words. If it can be shown that Hebrew speakers possess a knowledge that constrains the position of geminates relative to the root, then it follows that the root is a separate constituent in the representation of Hebrew words. Specifically, a constraint commonly affecting all Hebrew roots would suggest that they all form a single linguistic class. Note, however, that there are no semantic, orthographic, or phonological characteristics that define Hebrew roots as a class. The feature that unites all Hebrew roots must then be formal in nature, i.e., their status as a distinct morphological unit. The sensitivity of Hebrew speakers to a restriction whose domain is specifically the root morpheme would thus provide strong evidence for morphological decomposition.

Our method of choice in examining readers' sensitivity to the OCP is a rating task. The use of this technique in assessing tacit linguistic knowledge is sometimes criticized on the grounds that the rating decision entails a conscious, problem solving component, its outcomes may reflect meta-linguistic knowledge, and its measurement scale is coarse. These arguments were countered by Prasada and Pinker (1993), who noted that although the rating decision is a conscious act, the computations leading to the decision are not. Indeed, Kim et al. (1991)demonstrated that the ratings assigned to past tense inflections reflect sensitivity to grammatical categories of which subjects are utterly unaware. We return to the demonstration of the tacit nature of the OCP in the discussion of Experiment 1. Like Prasada and Pinker (1993), we further believe that in exploring a previously uninvestigated phenomenon such as the OCP, establishing what subjects consider to be “an acceptable form” is logically prior to investigating the effect of well formedness on performance in speeded response tasks. Rating appears to provide a direct and simple reflection of subjects' notion of well formedness. Conversely, the influence of task-specific strategies in speeded response tasks is well documented (e.g., Balota and Chumbley, 1984; Stone and Van Orden, 1993), and their sensitivity does not necessarily exceed the rating technique due to fluctuations in attentional and sensorimotor factors.

To assess subjects' sensitivity to the OCP, we presented them with a series of nonwords created from nonroots (combinations of three consonants that do not correspond to any existing Hebrew root) and asked them to determine the extent to which they sound like possible Hebrew words. The structure of our materials is illustrated in Table 2. The critical items violated the OCP due to the gemination of root-initial consonants4. If subjects are sensitive to the OCP constraint, then nonwords created from such roots should receive low acceptability ratings. To secure the attribution of the low acceptability of these items to a constraint on root structure, rather than a general unacceptability of gemination or a particular ill-formedness of the non-geminating consonants, each of the roots with initial gemination was matched to two control roots.

The first control was designed to examine whether the low acceptability of [ssm]-type roots is indeed due to the location of the gemination. Although Hebrew contains numerous roots manifesting final-gemination, it is possible that subjects have a general bias against gemination that is unrelated to its location. To examine this possibility, each of the initial-gemination roots was matched against a control root geminating the second and third consonants (e.g. [mss]5). This control root had exactly the same phonemes as the critical root and altered only the location of the gemination. A second alternative explanation to the ill-formedness of the [ssm]-type roots may be unrelated to gemination per sé. On this view, this group of roots exhibits some systematic ill-formedness in the sequence of the second and third consonants, and it is this ill-formedness, rather than the initial-gemination, that accounts for their low acceptability. Although the likelihood of creating such a systematic bias in a large group of items is rather low, we nevertheless decided to examine this possibility by designing a second control for the critical root. This control maintained the sequence of the second and third consonants of the initial-gemination root, but differed with respect to the first consonant, which was not a geminate of the second (e.g., [psm]). Together, the triplet of the initial-gemination (e.g., [ssm]), the final-gemination (e.g. [mss]) and the no-gemination roots (e.g. [psm]) permits determination of whether Hebrew speakers have internalized a constraint on the location of gemination.

Although a low acceptability of [ssm]-type roots would suggest that root-initial gemination is sufficient for low acceptability, it still leaves open the critical question of whether root-initial gemination is necessary for low ratings. Is it the location of the gemination within the root that accounts for its unacceptability, or could it be explained instead by a general prohibition of geminating the word's initial consonants? To demonstrate subjects' sensitivity to the OCP, and hence, to abstract root structure, it is necessary to show that it is the location of the gemination in the root, rather than in the word, that accounts for the unacceptability of the critical items. Although there is no reason to believe that a word-initial gemination is ill formed, it is nevertheless possible that subjects' sensitivity to the OCP may be determined by the transparency of the root which, in turn, may be affected by its location in the word. The transparency of the gemination may further depend on the particular phonotactics of the word pattern.

To examine the effect of the transparency of the root on subjects' ratings, each root triplet was conjugated in one of three classes of word patterns6. In the first class, the roots were presented with no additional prefixes or suffixes (e.g., SaSaM7). The location of the gemination in this class was therefore highly transparent. However, because, in this class, root-initial gemination is also word-initial, it is impossible to determine from this class alone whether the unacceptability of root-initial geminates reflects the location of gemination in the word or in the root. The distinction between word and root location was achieved in the second and third classes. Both classes presented the root sandwiched in between a prefix and a suffix. A sensitivity to the OCP in such circumstances requires a representation of the root. The second and third classes nevertheless differed in the transparency of the gemination due to their particular phonotactic properties. In the second class, the first and second root consonants were adjacent in the word's surface form (e.g. maS-Si-Mim). The failure to separate the geminates by a full vowel appears to increase the ill-formedness of the gemination. In contrast, geminates in the third class were separated by at least a full vowel and sandwiched between affixes (e.g. hiS-ta-SaM-ti). Furthermore, a metathesis rule affecting the formation of some of the word patterns in the third class also disrupted the succession of root consonants (Glinert, 1989). In our example, hiStaSaMti, the root-initial geminates are separated by the affix /t/. Such items permit examination of subjects' sensitivity to the OCP under extreme conditions of opacity.

In summary, the following studies assess subjects' sensitivity to root structure by obtaining ratings for words derived from a set of three matched roots (Please note that subjects were presented only with the derived verbs and nouns, and never with the roots themselves). These roots were each conjugated in three different classes of word patterns which differed in the transparency of the root. Two questions are at the center of this investigation: (1) Do subjects consider derivations of roots with initial-gemination as unacceptable? (2) Does the sensitivity to root structure require a transparent morphological structure? A demonstration of subjects' sensitivity to root structure would support the psychological reality of OCP and the morphological decomposition of Hebrew words.
EXPERIMENT 1

Experiment 1 was designed to examine subjects' sensitivity to the OCP and determine the effect of root transparency on its magnitude. Subjects were presented with triplets of nonwords and asked to determine the extent to which each member of the triplet sounds like a possible Hebrew word. The members of the triplets were all conjugated in one of the three word classes described above. They shared precisely the same derivational and inflectional patterns, and differed only with respect to their root structure. Specifically, these roots exhibited either an initial-gemination, final-gemination or no-gemination. If Hebrew speakers represent the root separately, and if roots are further subject to the OCP constraint, then root-initial gemination is expected to result in low acceptability ratings regardless of the word pattern in question. Thus, we predict that root-initial gemination should receive lower acceptability ratings compared to both root-final gemination and no-gemination. However, subjects' sensitivity to the location of gemination may depend on the transparency of the root and the idiosyncratic phonotactics of the word pattern. Thus, the acceptability of root-initial gemination may be modulated by word class.

Section snippets

Subjects

The rating questionnaire was administered as part of a course at the School of Education of Haifa University. 18 native Hebrew speakers served as subjects. They were all students in the School of Education of Haifa University and received no compensation for their participation.

Materials

The items submitted to subjects' rating were all nonwords. Although we refer to these items as `words', it should be realized that neither the roots nor the conjugated items corresponded to existing Hebrew words. These

Results

Subjects' ratings were submitted to ANOVA's (3 root type × 3 word class) by subjects and items. The ANOVA indicated a significant main effect of root type (F1(2, 34)=111.22, p=0.0001; F2(2, 46)=174.82, p=0.0001) and a significant interaction of root type × word class (F1(4, 68)=23.16, p=0.0001; F2(4, 92)=10.17, p=0.0001). The main effect of word class was not significant (F1(2, 34)=1.43, p=0.25, n.s.; F2(2, 46)=1.47, p=0.24, n.s.). The effect of root type was further investigated using two

Discussion

The results of Experiment 1 demonstrate that subjects' ratings of nonwords generated from nonexisting Hebrew roots are sensitive to the structure of the root. Words formed by the conjugation of roots with initial-gemination (e.g. [ssm]) were unacceptable. Their unacceptability cannot be attributed to a general bias against gemination, since the ratings of root-initial gemination were significantly lower compared to final-gemination roots (e.g., [mss]). Similarly, the unacceptability of words

Method

The same set of 216 non-words employed in Experiment 1 was used in the present study. These words were formed by the conjugation of the 24 root triplets used in Experiment 1 within each of the 3 word classes described previously. The words were presented in a randomized list. The instructions for the subject were similar to those described in Experiment 1 with the following exceptions. First, subjects were asked to rate each word according to how it sounded individually, rather than in

Results

Subjects' ratings were submitted to analyses of variance by subjects and items (3 root type×3 word class). The main effects of root type (F1 (2, 28)=63.34, p=0.0001; F2(2, 46)=75.03, p=0.0001), word class (F1(2, 28)=16.50, p=0.0001; F2(2, 46)=39.44, p=0.0001) and the interaction of root type×word class (F1(4, 56)=27.31, p=0.0001; F2(4, 92)=41.69, p=0.0001) were all significant.

Subjects' mean ratings as a function of root type and word class are presented in Fig. 5. Subjects' sensitivity to the

Discussion

The results of Experiment 2 replicate the central findings of Experiment 1. Despite the fact that the rating procedure did not require special attention to the root, subjects nevertheless manifested a strong sensitivity to its structure. Words derived from roots with initial-gemination were unacceptable compared to roots with final-gemination or no-gemination. The sensitivity to root structure was not contingent on the transparency of morphological structure. In fact, it is the first class, in

General discussion

Hebrew speakers consider words derived from roots with initial-gemination as unacceptable. This finding clearly does not stem from a general bias against gemination

Acknowledgements

This research was supported by a National Research Service Award F32 DC 00186-01 to Iris Berent. Additional support was provided by National Institute of Health FIRST Award CM 5 R29 NS 26247-05 to Guy C. Van Orden. We are grateful to Guy Van Orden, Greg Stone and Steven Pinker for discussions of this work. We also thank Inon Berent and Mara Georgi for their technical assistance.

References (48)

  • Aronoff, M. (1994). Morphology by itself. MIT Press, Cambridge,...
  • D. Balota et al.

    Are lexical decisions a good measure of lexical access? The role of word frequency in the neglected decision stage

    Journal of Experimental Psychology: Human Perception and Performance

    (1984)
  • Bentin, S., Frost, R. (1992). Morphological factors in visual word recognition in Hebrew. In: Feldman, L.B. (Ed.),...
  • S. Bentin et al.

    The contribution of morphological and semantic relatedness in visual word recognition: Evidence from Hebrew

    Quarterly Journal of Experimental Psychology

    (1990)
  • Berent, I., Everett, D. and Shimron, J. (1997). Do phonological representations specify formal variables? Evidence from...
  • I. Berent et al.

    A rose is a REEZ: The two cycles model of phonology assembly in reading English

    Psychological Review

    (1995)
  • Berent, I., Van Orden, G. (1996). Can we unmask the phonemic masking effect? The problem of methodological divergence...
  • Emmorey, K. (1995). Processing the dynamic visual-spatial morphology of signed languages. In: Feldman, L.B. (Ed.),...
  • Even-Shoshan, A. (1993). Ha'milon ha'xadash [The new dictionary]. Kiryat-Sefer,...
  • L.B. Feldman et al.

    Morphological analysis of disrupted morphemes: Evidence from Hebrew

    Quarterly Journal of Experimental Psychology

    (1994)
  • L.B. Feldman et al.

    Decomposing words into their constituent morphemes: Evidence from English and Hebrew

    Journal of Experimental Psychology: Learning, Memory and Cognition

    (1995)
  • C. Fowler et al.

    Relations among regular and irregular morphologically related words in the lexicon as revealed by repetition priming

    Memory and Cognition

    (1985)
  • Frost, R., Forster, K., Deutsch, A. (1994). Morphological and orthographic analysis in word recognition contrasted....
  • Glinert, L. (1989). The grammar of modern Hebrew. Cambridge University Press,...
  • Cited by (91)

    • The double identity of doubling: Evidence for the phonology–morphology split

      2017, Cognition
      Citation Excerpt :

      In line with this analysis, the OCP and contiguity have been each widely documented in the formal analysis of many languages (e.g., Inkelas & Zoll, 2005; Kager, 1999; Leben, 1973; McCarthy, 1979; McCarthy & Prince, 1995b; Suzuki, 1998; Walter, 2007).3 There is also a large experimental literature demonstrating that speakers productively extend identity restrictions to novel forms (e.g., Berent & Shimron, 1997; Berent et al., 2001; Berkley, 1994; Boll-Avetisyan & Kager, 2014; Buckley, 1997; Frisch & Zawaydeh, 2001; Kawahara, Ono, & Sudo, 2006). However, no previous experimental research has examined whether people productively obey contiguity.

    • The role of the morpho-phonological word-pattern unit in single-word production in Hebrew

      2016, Journal of Memory and Language
      Citation Excerpt :

      The morphological processes of de-composition, and specifically root extraction, were also seen in the ‘segment shifting task,’ where the ease with which participants were able to create a nonword derived from a three consonantal pseudo-root + a word-pattern of a source word depended on the prominence and productivity of the source word’s root morpheme (Feldman, Frost, & Pnini, 1995). Additional evidence that the root is a structural unit of words’ mental representation in Hebrew is derived from a set of studies which demonstrated that Hebrew readers are sensitive to the root morpheme’s phonological structure (Berent, Everett, & Shimron, 2001; Berent & Shimron, 1997; Berent, Shimron, & Vaaknin, 2001). In particular, these studies showed that subjects’ response to nonword was affected by whether the nonword obeyed or violated a specific phonological constraint that avoids duplication of the first two consonants of the Hebrew root.

    • Is speech processing influenced by abstract or detailed phonotactic representations? The case of the Obligatory Contour Principle

      2016, Lingua
      Citation Excerpt :

      When comparing the task used in our study to previous studies, differences in processing demands become evident. Most studies that suggest a psychological reality of abstract phonological knowledge used wellformedness judgments (Berent and Shimron, 1997; Coetzee, 2009) and lexical decision tasks (e.g., Berent et al., 2001b; Shatzman and Kager, 2007). Of course it has often been shown that decisions in these tasks are also affected by lower-level probabilistic information and lexical neighborhood density (e.g., Bailey and Hahn, 2001; Vitevitch and Luce, 1998, 1999).

    • Amodal phonology

      2021, Journal of Linguistics
    View all citing articles on Scopus
    View full text