Lexical Mediation of Reduplication Effects in Arabic Speakers: Implications for Associationist Accounts of Word Formation
Poster D2 in Poster Session D with Social Hour, Friday, October 7, 5:30 - 7:15 pm EDT, Millennium Hall
Enes Avcu1, Skyla Lynch1, Seppo Ahlfors2, David Gow1,2,3; 1Massachusetts General Hospital / Harvard Medical School, 2Athinoula A. Martinos Center for Biomedical Imaging, 3Salem State University
In Semitic languages like Arabic or Hebrew, words (e.g., janna ‘garden’) are created by a combination of a three-consonant root (e.g., jnn) and a wordform pattern (e.g., C1VC2C3V). These languages may also have a root that reduplicates, or repeats, the second consonant, (jnn) but not the first (jjn). Berent (2002), using a lexical decision task, showed that Hebrew speakers were faster at rejecting nonwords with unattested repetition patterns (e.g. [jjn]) than they were at rejecting nonwords with attested repetition patterns (e.g. [jnn]). These results were used as evidence for the symbolic account that attributes linguistic generalizations to the use of abstract rules. However, it is also possible that Hebrew speakers were only trying to recognize words in the task, and since no words exist that have a root [jj_], they recognize the item as not a word immediately when they hear the second consonant of the root. Such an interpretation could support associationist accounts that attribute linguistic generalizations to statistical properties like similarity to existing words in the lexicon. In this work, we asked what mechanism is responsible for assisting a speaker in knowing that [jnn] could be an acceptable root for a word, but [jjn] could not, even when the speaker is hearing items they have never heard before? We collected simultaneous magnetoencephalography (MEG) and electroencephalography (EEG) data from native speakers of Standard Arabic while they completed an Arabic lexical decision task. In each block, subjects heard a series of trisyllabic CVCVC nonsense words with a common syllable reduplication pattern (e.g., AAB as in jijin, ABB as jinin, and ABC as jinik), and they were tasked to press one of two buttons to signify if they thought the item was an Arabic word or nonword. There were 1160 trials divided into ten blocks. Reaction time and accuracy data were both collected during the experiment. We compared the neural responses in the three-syllable reduplication conditions (AAB, ABB, and ABC) using effective connectivity analyses of brain activity to view how different root representations are utilized. Behavioral results showed participants performed the lexical decision task with high accuracy and were faster at rejecting AAB nonwords than they were at rejecting ABB nonwords replicating Berent’s results. Neural results showed both attested and unattested repetition patterns produced stronger influences of brain regions implicated in wordform representation and phonetic/motor functions on acoustic-phonetic regions. These results suggest that speakers are making a comparison between an unattested root pattern and an existing root pattern in the lexicon, and judgments are shaped by how close the given root pattern is to an existing root. While these preliminary results do not resolve questions about the mechanisms responsible for assisting a speaker in knowing whether unattested root patterns could be an acceptable root or not, future progress will benefit from developing independent, empirically derived characterizations of the representations upon which any mechanism must depend.
Topic Areas: Phonology and Phonological Working Memory, Speech Perception