Dr Dagmar Divjak

Research interests

My main research interests are in understanding how our cognitive capacities give rise to the patterns and structures we see in language and in charting what language has to offer the learner in his/her quest for the meaning of words and constructions. Because of my background in usage-based cognitive linguistics, frequency in all its guises plays a central role in my work.

My fascination with the grammar-lexis interface first led me to explore these questions for the category of Russian and Polish complex events, e.g. I´ve always wanted to be a linguist. Finite verbs that participate in complex events are quite peripheral in the lexicon: less than 1% of all verbs combine with an infinitive, and many of these verbs are extremely rare in corpora. Nevertheless, meta-linguistic acceptability ratings reveal that speakers´ judgments are consistent even for the rarest verbs yet none of the currently available corpus-derived frequency measures predicts this correctly: less than 30% of the ratings can be accounted for in a regression model that takes into account frequency measures as well as rater generosity. I am currently exploring how speakers succeed in correctly accepting some and rejecting other equally rare constructions by comparing strategies they may entertain when judging sentences off-line. In a follow-on study I take a closer role at the role frequency plays in the mature language system by comparing the predictive power of a range of standardly used measures of frequency for a variety of off-line and on-line behavioural tasks.

Another way of looking at these questions can be found in my work on near-synonyms expressing abstract concepts, such as intend, try and contrive. Given that we lack sensory-motor experience for these concepts and that the differences between near-synonyms such as plan and intend or try, attempt and endeavor are hardly observable, how do we find out about them? How far can we get by tracking frequency distributions in input? Surprisingly far: on the basis of information contained within clause boundaries a statistical model predicts the choice between near synonyms as well as the average non-English US college applicant. Ongoing research looks at these results from both a prototype and exemplar-based perspective, viewing them as opposite ends along a continuum that might be activated at different stages in the acquisition process dependent on the amount of input received. In a next step, I plan to measure the power of distributional learning for concept formation by studying blind speakers’ understanding of verbs of seeing from a corpus-based and experimental perspective.

A third study on the cognitive motivations for a seemingly aberrant linguistic pattern zooms in on the interaction between aspect and modality in the Slavic languages. Slavic languages reverse the allegedly universal relation between aspect and modality by reserving imperfective aspect to express deontic modality, and perfective aspect to render dynamic modality. Statistically modelling data from a comparative Slavic corpus revealed that aspectual choice in modal constructions is not defined by modality type, but is the result of an analogical mapping between the conceptualizations underlying the domains of aspect and modality.
This poses an interesting question for acquisition research: given that children are not deemed capable of entertaining deontic concepts at least until the age of 5, what does an imperfective in a modal context mean to younger children? Does aspect function as scaffolding for the acquisition of modality? A corpus-based study contrasting Dutch and Russian children is underway that will shed light on the contribution made by input frequency and the role played by moral maturation in the acquisition of modality.

I have also been awarded a small research grant to collect longitudinal data on early trilingual first language acquisition of Dutch, English and Swedish. Recording of the first child (12-30 months) is now complete. I plan to conduct a corpus-based pilot study on 1) the effect different word stress patterns in closely related languages may have on initial word segmentation and 2) the communicative value prosody may have for infants acquiring three closely related languages. Please contact me if you would be interested in working with me on one of these two projects.