postdoctoral researcher at Universität Tübingen
University of Tübingen
Institute of Linguistics
72074 Tübingen, Germany
igor.yanovich at uni-tuebingen dot de
Curriculum Vitae (pdf)
My specialization is in formal semantics. The empirical domain I am now primarily working on is modality. I have proposed new analyses of epistemic modality and weak necessity, and argued for recognizing new types of modals: symbouletic modals of suggestion, and a "collapse variable-force" modal in Old English. Currently I am particularly interested in how modal meanings change over time; how multiply ambiguous modal words function in everyday speech; and how the modal systems of languages in contact situations influence each other. Over the years, I have also worked on the expressive power of backwards-looking operators like "now", de re attitudes, gender presuppositions of anaphoric pronouns, and indefinites.
I am a part of an emerging interest group for historical semantics. The group is now working on establishing a yearly series of workshopы on formal-semantic approaches to semantic change. I took part in organizing the second workshop of the to-be series as a special session at Sinn und Bedeutung 2014. The puzzles in historical semantics are relatively easy to find, but the formal methods needed to solve those puzzles are only starting to be developed. In my own historical-semantic work, I trace semantic and syntactic changes at the level of decades, paying attention to the overall distribution of meanings across each synchronic span. That provides me with more fine-grained data than is usually used, and thus allows for discovering otherwise indiscernible patterns, such as a variable-force modal in Old English, or a newly emerged symbouletic modal in Modern Russian.
I am an associate member of the EVOLAEMP project in Tübingen, which applies bioinformatic and artificial intelligence techniques to the study of family relations between languages. In this line of work, namely linguistic computational phylogenetics, I recently received results which correct widely publicized claims of strong support of the Anatolian homeland for Indo-European from phylogenetic methods. I found out that when we amend some interpretation problems in the earlier methods, and add more non-controversial prior information using radiocarbon dating, the very same phylogenetic methods show strong support for the Pontic steppe homeland of Indo-European. I also developed a technique for estimating the robustness of cognacy-based phylogenetic analyses to errors stemming from the possible insufficiency of our dictionaries. In the longer run, I believe that phylogenetic methods may be adapted for studying the evolution of syntactic and semantic features, and not just relationships between languages.
Sometimes I also do phonology, which resulted in some new results in mathematical Optimality Theory, and finding out from Ukrainian and Russian data, with Donca Steriade, that Base Priority effects also work within inflectional paradigms.
Selected papers are described below, grouped by topic. The CV contains the full list.
Currently (winter semester 2014/15) I teach Corpus Methods in Theoretical Linguistics at the Tübingen University. I will be happy to share materials by email request. Older lecture handouts may be found here.
Both in 16th century and in late 19th century Ukrainian, maty 'have' allows three types of modal/temporal uses: necessity, futurate, and possibility. The modal is genuinely ambiguous between those three, unlike true variable-force modals of the Pacific Northwest, and has no parallels described in the semantic literature. I am now investigating the finer details of the modal's semantics in Old and Pre-Modern Ukrainian. I am also looking into maty's place in the general Slavic landscape of 'have'-based modals: most Slavic languages had such at least at some stage, but few, with the possible exception of Polish, developed anything close to the possibility reading of Ukrainian maty.
In English, there is seemingly evidence for distinguishing "strong necessity" modals must and have to from "weak necessity" modals ought and should: the two groups of modals behave differently in several types of tests. I show that despite that, there is actually no such category as strength of necessity, and explain the observations appealing to basic properties of modals such as their modal flavor. My argument is based on three types of data. (i) In US court decisions, performative should carries the same strong force as must. (ii) In Russian and Ukrainian, the "weak necessity tests" that neatly divide English modals into two groups, fail to produce complementary distributions. (iii) In a pilot investigation of a parallel Bible corpus with English, German, French, Ukrainian, Russian, Czech and Bulgarian texts, English modals must vs. should and ought fail to correspond to categories of other languages.
I describe a previously unnoticed phenomenon of Russian syntax: the ability of certain predicative adjectives with infinitival complements to appear in a pre-copula position. The resulting structure resembles the so-called "long head movement" structures of the West and South Slavic. A crucial feature of the Russian pattern is that different predicative adjectives appear in the Adj-Aux-Inf order with different frequency: "more grammatical" adjectives almost always appear with the Adj-Aux inverted order, while other adjectives vary. For example, dolzhna 'must.ADJ' appears as Adj-Aux >1000 times in the 230M-words written portion of the Russian National Corpus, and only 1 time as Aux-Adj; the order of distinction is similar for the spoken subcorpus of 11M words. But "less grammatical" gotova 'ready' appears as Adj-Aux in 9.8% of written, and 57.9% of spoken instances. This distribution thus reflects the existence of a proto-category in Russian: adjectives like dolzhna do not currently form a syntactic category as such, as they still permit the dispreferred order in judgements, but there is apparently potential for the development of a new true syntactic category. I am now looking into the diachrony of this construction.
In this paper, I argue for distinguishing a novel modal flavor of symbouletic modality (< συμβoυλευω ‘advise’) — the modality of advice and suggestion. As symbouletic modality is often expressed by modals that also have deontic or teleological readings, it is not trivial to es- tablish its special character. Fortunately, Russian features modal stoit that is specialized for advice/suggestion uses. Using stoit as a convenient testing case, I propose a formal analysis of symbouletics within the framework for performative verbs and imperatives by Condoravdi and Lauer (2012), analyzing symbouletics as a kind of performatives.
In this short reply to Braun (2013), I provide an argument against his invariantist theory of might. Invariantism proposed by Braun aims to maintain full identity of semantic content between all uses of might. I invoke well-known facts regarding diachronic change in meanings of modals to argue that the invariantist view commits us to implausible duplication of familiar processes of lexical semantic change on the level of “lexical pragmatics”, with no obvious payoff.
This is an investigation into the expressive power over models that backwards-looking operators such as "now", "then" and "actually" add to the basic modal language. Contrary to popular belief in linguistics and philosophy, such operators do not push expressive power up to the level of explicit quantification over times and worlds.
(2011) How much expressive power is needed for natural language temporal indexicality? in Proceedings of WoLLIC 2011, LNAI 6642, Springer, pp. 293-309. doi:10.1007/978-3-642-20920-8_27. [preprint]
Superceded by "Expressive power of “now” and “then” operators".
(2010) Evaluation tree languages, abstract published in The Bulletin of Symbolic Logic, 17:2 (2011), p. 323. [presentation]
A stab at an analysis of context-sensitive modal operators, grown from earlier work on indexical presuppositions. Certain themes have been picked up in later work on backwards-looking operators, with much better technical precision.
Talk at the 2010 Logic Colloquium in Paris.
(2011) The problem of counterfactual de re attitudes in Proceedings of SALT 21, pp. 56–75.
I argue that counterfactual de re presents no analytical challenge (contra Ninan 2008) once it is analyzed as parasitic on beliefs. As an extra, a simple compositional way to derive de re readings is defined (see Sec. 2.3), not using covert concept generators, the movement of the res, or other syntactic entities not motivated by syntactic phenomena.
(2012) Indexical presuppositions of pronominal gender features [current version (July 2012)]
The paper shows that gender features on anaphoric pronouns trigger not regular presuppositions, but a peculiar set of restrictions on the context. Moreover, those restrictions only arise when pronouns refer to humans (which is harder to show in English, but is easy to observe in, e.g., Russian). I argue that the "presuppositions" of gender features are thus not a part of the compositional semantics, but instead the effects of a pragmatic rule of use for pronouns referring to humans.
(2012) What can Russian gender tell about the semantics of phi-features? (presented at FASL 21, Indiana University, May 2012) [presentation]
A conference-format presentation of the ideas here, with focus on data from Russian.
(2009) On the nature and formal analysis of indexical presuppositions, in Proceedings of LENLS 2009, LNAI 6284, Springer, pp. 272-291.
Superseded by this. The paper describes the peculiar projection pattern associated with gender features on English anaphoric pronouns, discusses the initial observations by Cooper (1983) capturing a part of that pattern, but forgotten in the later literature, and explains why conventional LF-semantics devices such as copying the antecedent's features or introducing special binding-theoretic rules for pronominal world variables cannot derive the observed facts.
The paper discusses the constraints on intermediate scope readings of some and a certain indefinites, and proposes a presuppositional analysis of those determiners.
The paper was presented at Sinn und Bedeutung 2007 in Oslo.
A short note discussing the semantics of Russian indefinites based on the adjectival root kakoj 'which'. Such indefinites are argued to be ambiguous between two readings, so that, e.g., kakoj-to is sometimes roughly synonymous with English 'some', and other times, with English 'of some kind'.
The paper proposes a choice-functional analysis of Russian indefinite series formed with -to and -nibud, and argues for a Hamblin analysis of bare indefinite roots in Russian.
Supersedes the previous two works, developing and sharpening the idea that the cycle has access to a pool of forms generated earlier rather than a single base form. In the new theoretical setup, the directionality of the cycle is preserved, but its reliance on unique representations at each stage is rejected.
Standard cyclic effects in derivational morphology involve inheritance of properties of the base by the derived form. Using data from Ukrainian and Russian, we argue that the base word provides not just a single form, but rather all the forms in its paradigm for the derived item to choose from.
Stress patterns within nominal inflectional paradigms in East Slavic are notoriously complex, and in the case of Ukrainian and Belarussian, still understudied in the theoretical literature. We analyze stress in Ukrainian nominal paradigms, and argue that everything falls into place once we accept that the singular subparadigm serves as a collective base for the plural subparadigm. Thus, effects of Base Priority may be observed not only in derivational morphology, but also within inflectional paradigms.
Many OT tableaux convey the same information about rankings, and operations preserving that information have been studied in Hayes (1997), Brasoveanu and Prince (2005), Prince (2006), a.o. The paper defines a functionally complete set of elementary information-preserving transformations, and provides a computable test for OT tableau equivalence via normal form techniques. One corollary of the results is that Brasoveanu and Prince's Skeletal Basis tableaux are unique not just for a single tableau, but for whole equivalence classes of comparative tableaux as well.
The paper develops the theory of sets of rankings compatible with a particular tableau, and provides methods for computing such sets for arbitrary tableaux, allowing for lossless and monotonic incremental OT learning.
Superseded by the two papers above.
What is good in the dissertation would not have existed without my MIT teachers and advisors, who helped me become the researcher I am today. The direct influence of my dissertation advisors Kai von Fintel, Irene Heim and Sabine Iatridou on the text would be evident for the reader. But beyond the dissertation advising, without their help and support over all my five years at MIT, I would not have grown into a linguist I am now. It is very much thanks to Kai, Irene and Sabine that I learned to be guided by empirical data, be ready to see beyond my current theoretical convictions, and also to be cautious when drawing theoretical inferences from observed data. I am also very grateful to my other MIT teachers, especially to Adam Albright, who served as my registration advisor for all my time at MIT; Donca Steriade, who infected me with excitement about phonology; and Martin Hackl, who taught me a lot about what a teacher needs to know.