Annette Hohenberger & Joerg Keller
University of Frankfurt/Main
hohenberger@lingua.uni-frankfurt.de
The comparison of signed and spoken language production yields a unique opportunity to assess modality effects in processing. Although the PF-interfaces are clearly different in both modalities (aural-oral vs. visual- gestural), under the null-hypothesis, processing is amodal in nature. There is only one structural format for the representation of language.
From a typological perspective, languages serialize their units to different degrees. Basically, all sign languages display a very high degree of simultaneity in phonology, morphology, and syntax whereas spoken languages display a broader range of variation. With regard to morphology, for example, we distinguish concatenative languages such as Turkish and English and non-concatenative languages such as Hebrew and other semitic languages. Sign languages pattern with the latter.
We will contrast data from German Sign Language (Deutsche Gebaerdensprache DGS) and spoken German, more precisely, repairs of slips of the hand and tongue. Until now, there has been no research on repairs of slips in sign languages. Our knowledge of monitoring relies exclusively on the investigation of spoken language production errors.
Given that representations are grammatically structured and that processing accesses representations, processing is structure-sensitive, too. This is what is found in spoken language error repairs: The major cut-off point in repairs of slips is after a possible syllable.
Typically, signs are monosyllabic. The articulation of a syllable/sign in sign language, however, takes twice as long as the articulation of a syllable in spoken language. This asymmetry offers a fantastic opportunity to investigate the nature of monitoring: is it anchored structurally (with respect to representations) or temporally (in terms of internal time units irrespective of content)? If the monitor is structure-sensitive, we expect the major cut-off point to be located after the syllable/sign. This, however, is not borne out by the data. The monitor typically interrupts erroneous signs within a syllable.
We conclude that monitoring, in general, is not structure-sensitive. Rather, monitoring disregards the content of what is processed. In sign language, the mean physical duration of truncated slips corresponds to the physical duration of spoken syllables. This means that processing is indeed amodal: Sign languages are processed and monitored on a par with spoken languages.
We will present data in support of this conclusion and discuss further consequences.