Modal Concepts and Compositionality: New Directions in Experimental Semantics

Modality is found in all human languages. It poses a special kind of interest because it concerns meanings where we reason about the `non-actual' or the hypothetical. This is something unique to human communication systems within the animal kingdom. Words where modal meanings are traditionally involved include modal auxiliaries such as `can' and `must' . However, in formal semantic treatments of natural languages, modal analyses have been proposed for more and more kinds of language phenomena, in lexical and aspectual meanings as well. We are working with the hypothesis that some of these so called modal meanings are not as cognitively complex as others. In particular some types of `modality' are actually cognitively basic and involve a kind of essential understanding of concepts, while other kinds of modality are more internally complex and require more complex reasoning. We are testing our hypothesis by constructing and devising experimental methodologies to probe the processing time course of different species of `modal´ constructions, across languages with different morphological make up. We specifically compare English and Russian in this area, which are different in the ways in which they encode modal and aspectual properties. We expect that our results will have an important impact on mathematical treatments of modality in linguistic theory. The project is also pioneering in the sense that very few experiments so far have been devised that actually test the predictions of formal semantic theory and the stages of meaning composition. The results from our eye tracking studies have shown us so far that the Russian aspectual system is processed rapidly and effortlessly in adults and acquired early by children. One exciting early result of our work is the discovery that eye tracking can reveal quite fine details of temporal processing--- we have detected reliable differences in incremental processing of forms depending on whether prefixes or suffixes are involved in Russian. On the English side, while judgements on the English progressive closely match Russian speakers´ responses to the imperfective, the English past tense does not behave like a perfective in either processing or behavioural response. But even for the progressive, the eye tracking data from English shows a slower and delayed response, indicating a construction level rather than lexical response. The next stage of the project will compare the processing trajectory for modal auxiliaries to these aspectual and lexical effects. The major exciting development in this year of the project has been the discovery that the eyes really are windows to the soul, or at least to the soul of meaning composition! We have a robust methodology which we can now turn on a variety of other constructions in English and Russian. During the most recent reporting period, our experimentation plans have been affected by Covid-19. However, we have innovated this year in pioneering online web based technologies for conducting eyetracking experiments via participants' own webcams in their own homes. We have successfully replicated our infra red experiments on English and Russian in the new web-based technology (in experiments that are the first of their kind) and are now ready to roll out that methodology on a variety of experimental conditions in the future.

We have developed a robust methodology which is sensitive to the dynamic unfolding of meaning composition, and which we have now pioneered in a online web based format, which preserves all of the subtle effects found in the in person experiments. This means that we now have a paradigm that is fully and easily portable to a variety of other languages without requiring in person testing. The presentation of our work at major psycholinguistic and semantic conferences has already led to other researchers requesting access to our materials for adaptation to other language pairs in the domain of tense and aspect. Vos has become an expert in online methodologies, will complete her PhD within the year, and has already been hired at a major Dutch university helping researchers in online experimentation. Minor has become and expert in experimental visual world paradigm and statistical analysis. He has now been hired as an experimental linguist in a førsteamenuensis position at UiT. This has amply fulfilled the training goals and expansion of the skill set for the post doctoral researcher on the project.

Modal meanings are found in all human languages. They pose a special kind of interest because they concern meanings where we reason about `non-actual', potential states of affairs. Typological work tells us that verbal modal marking systematically and robustly occurs further away from the verbal root than causational or aspectual morphology, an interesting fact in and of itself. Despite this strict morphological separation and ordering found typologically, much recent work on formal semantics employs modal analyses of verbal and aspectual notions. If modal notions pervade verbal semantics in this way, we lose an explanation of the ordering from general cognitive factors. Further, we reject the idea that there is an innate pre-formed universal template for verbal expressions in natural language. We explore instead the hypothesis that there are (at least) two types of intensionality involved. One, is lexical and conceptual (we call this `internal' intensionality) and inheres in the way roots label natural kinds. The other is inferential and compositional and can be encoded by explicit semi-functional morphology (we call this `external' intensionality). The latter notion is strictly ordered with respect to the root, not the former. We test our hypothesis through a variety of offline experiments involving prediction and classification using the visual world paradigm and and eye tracking as a basic methodology. We zero in on cases where formal semantic analyses mismatch with typological predictions in the domain of modality. Our test languages are English and Russian, since there has been much formal semantic work on both, and because both possess a wealth of aspectual and modal interactions, with crucially different morphological exponence. Since work in experimental semantics is relatively new, we expect the project to contribute new insights and methodologies for testing aspects of verbal composition in a psychologically grounded way.