Discontinuous morphology in time: Incremental masked priming in Arabic

Abstract

Semitic morphology is based on the combination of two abstract discontinuous morphemes, the word pattern and the root. The word pattern specifies the phonological structure and morpho-syntactic properties of the surface form, while the consonantal root conveys core semantic information. Both units play a crucial role in processing Semitic languages such as Arabic and Hebrew. Here we use incremental masked priming to probe the time-course of word pattern and root activation in reading Arabic deverbal nouns and verbs. The morphological (word pattern and root), orthographic, and semantic relationship between prime and targets is varied over four stimulus onset asynchronies (SOAs) (32, 48, 64, and 80 ms). Results show distinctive patterns of activation for the two morphemic entities. Word pattern effects are transient, and detectable only at SOAs 48 and 64 in deverbal nouns and SOA 48 in verbs. Root effects are strong at all SOAs. This may reflect differences in the timing with which word pattern and root information can be extracted from the orthographic input, as well as differences in the roles of these morphemes in building internal lexical representations. Both types of morphemic effect contrast strongly with the effects of orthographic and semantic primes, where reliable facilitation is only obtained at the longest SOA (80 ms). The general pattern of results is consistent with the view that morphological effects in Semitic languages represent distinct structural characteristics of the language.

Acknowledgments

We thank Mike Ford for his assistance in the preparation of the two experiments and Abdallah Megbli, headmaster of the High School in Tataouine, Tunisia for his generous help in providing testing facilities and access to participants for testing. We also thank Matt Davis, Dave Plaut, Kathy Rastle, and an anonymous reviewer for helpful comments on an earlier version of this work. This work was supported by the Medical Research Council (UK).

Notes

¹The letters are traditionally used as place holders indicating where the first, second and third letter of the root are to be inserted.

²This is true of masked priming in Arabic and Hebrew. However, cross-modal priming offers a different picture with a slight increase in the magnitude of priming among transparent pairs in Hebrew (Frost et al., 2000), but not in Arabic (Boudelaa & Marslen-Wilson, 2000).

³Conditions 1 and 2 were also matched in terms of their underlying phonological overlap (i.e., counting the short vowels not specified in the orthography), which averaged 2.3 and 2.1 phonemes in common, respectively. However, we assume that the critical factor here is orthographic overlap, since this is the overt form dimension directly shared by prime and target.

⁴As is generally the case in research with Semitic scripts, it is not possible to fully match orthographically without introducing morphological confounds. It is hard to find pairs of words that share three consonants that do not share a root as well, especially given the other constraints on the stimuli.

⁵ Asterisks indicate a pseudoword throughout.

⁶ A word pattern has a causative morpho-syntactic interpretation or meaning when the surface form where it appears is glossable as cause someone to do X, where X is the core semantic meaning associated with the root. For example, in the form , where the root is {skt} keeping quiet and the word pattern is , the latter is considered to have a causative meaning in the sense that is glossable as cause someone to keep quiet. Analogously, a word pattern is intensive when the surface form of which it is part has to be glossed as do X with intensity, and a reciprocal word pattern is one which entails a gloss like X is done mutually by two agents (Holes, 1995; Wright, 1995).

⁷Average underlying phonological overlap was 2.5 phonemes for Condition 1 and 2.0 phonemes for Condition 2.

⁸Because of the much more restricted range of word patterns in the verbal morphology (10 instead of hundreds), and because of the constraints imposed by the need to use long vowels where possible, it was not possible to fully match Condition 1 in number of overlapping letters without introducing potential morphological confounds.