The Arabic project will be lead by Dr. Samantha Wray (Dartmouth College), with significant input from Suhail Matar (NYU). Arabic is defined by Ethnologue as a ‘macro- language’ with dozens of distinct, but largely mutually intelligible sister languages. For the purposes of this project, we focus on Modern Standard Arabic (MSA), the variety used most consistently in written texts. Arabic words are formed through the interleaving of two discontinuous bound morphemes: the root, which encodes the broad, core semantic meaning, and the pattern or template which encodes syntactic and semantic information such as part of speech, number, person and case. For example, the root /k-t-b/ ‘writing’ may be combined with several different patterns, including ma–a- ‘place’ for ‘office’, -a:-i- male singular nominative person for ‘writer’, and -a-a-a male singular verbal perfective for ‘he wrote’. Critically, the Arabic root has no grammatical category and many roots have a very broad range of meanings depending on their pattern (eg. X & Y share the root CCC). Because of this typologically distinct root-and-pattern system, Arabic has been well studied by psycholinguistic work addressing the ability of the speaker to extract and access morphemes in on-line spoken and written processing (including Boudelaa & Marslen-Wilson 2000, 2001, 2004a, 2004b, 2005, 2011; Prunet et al. 2000, Idrissi et al. 2008, among others). In neurolinguistics, the distinctness of these morphemes in processing has also been demonstrated (EEG: Boudelaa et al. 2010, MEG: Gwilliams & Marantz 2015). However, research beyond basic demonstration of the salience of the root, and occasionally pattern, morpheme is minimal, and no research has focused on the neural signatures of word-internal syntactic and semantic processing. Kastner (2017) argues on the basis of Hebrew that Semitic templates are grammatical affixes, with the same syntactic and semantic distribution and restrictions as concatenative affixes, but this proposal has not been experimentally tested.
We have initially identified one affix which places semantic restrictions on the kind of verb it can attach to. In this case, the ‘affix’ is a discontinuous string of phonemes making up the bound morpheme known as the pattern. The root /k-t-b/ ‘writing’ changes its part of speech, in addition to other information such as tense, aspect, number, and gender, when combined with different patterns. Because roots are acategorical, the category violation items will be built with roots that are highly noun dominant in their use (as was done for the English experiment). However, not every root+pattern combination is possible. For example, the pattern /it-a-:-a-a/ ‘reflexive/passive/effective’ can only be able to be interleaved with roots which have a causative/intensitive verb form. Thus, the root /H-s-n/ ‘good’ can be interleaved with the /-a-:a-a/ causative pattern to make /Has:ana/ ‘he improved (it)’, and with the /it-a-:-a-a/ reflexive to make /itHas:sana/ ‘he got better’. Contrast this with the root /sh-r-k/ ‘sharing’ which cannot interleave with reflexive/passive/effective or causative/intensive patterns, resulting in the ungrammatical */shar:aka/ *‘he force-shared’ and */itshar:aka/ *‘he shared himself’.