It is more than twenty years ago now that the notion of irrealis was introduced in the typological
literature. Carried over from the literature on native American languages, it was intended to
establish modality as a cross-linguistically valid category with a unitary conceptual content.
According to a formulation going back to Nancy Mithun (who, by the way, does not subscribe to it),
the realis “portrays situations as actualized, as having occurred or actually occurring”, while the
irrealis “portrays situations as purely within the realm of thought”. The notion of irrealis has been
criticized by Joan Bybee as an outdated Jakobsonian Gesamtbedeutung. In no two languages is the scope
of irrealis uses exactly the same; in every language it is just the sum of a number of individual
grammaticalization processes, and the highly general ‘irrealis’ meaning is epiphenomenal. While Bybee
is basically right, her view is too extreme. Not every irrealis use has its individual grammaticalization
history, and the wide functional scope of irrealis forms entails generalizations: e.g., the Baltic
(Lithuanian and Latvian) irrealis is based on the IE supine, which points to a functional shift from
purpose of motion to purpose in general, then to ‘purposive’ complement clauses (with desiderative verbs,
deontic modals) etc. It is therefore legitimate to ask what the nature of the generalizations is. In my
view there are two important generalizations: counterfactivity or emphatic non-factivity (the function
traditionally associated with irrealis) and unanchoring. Unanchoring is the inherent lack of situational
and temporal setting in non-propositional complements, e.g., with desiderative verbs (as in French il
faut que je vous dise ‘I have to tell you’), but also the extraction of an event from its situational
and temporal setting for the purpose of evaluation (as in French il est dommage que vous nous quittiez
‘it’s a pity you should be leaving us’).
The middle voice was, at one time, extremely unpopular in the linguistic literature, perhaps partly
as a result of people struggling to understand the rather unhelpful explanations offered for this
notion in Greek school grammars. Rehabilitated by Suzan Kemmer (1993), it has now become more broadly
accepted in typology, as seen in recent work by Guglielmo Inglese. In Generative Grammar, the notion
of middle has been firmly established for some time now, but in a different and much narrower sense
(apparently going back to John Lyons), with reference to what is also known (with a term introduced by
Faltz 1975) as the facilitative middle. One easily imagines the misunderstandings to which this divergent
terminological usage could have led if the disconnect between research traditions had not been as
absolute as it actually is. The work done on the middle (in the narrow sense) in the formal tradition
is both interesting and important (especially the discussions on the syntactic vs presyntactic, that
is, lexical character of the middle), but the view (prevailing in this school of thought) of the middle
as inherently generic is mistaken, resulting from ignorance of the Baltic and Slavonic facts. Below I
also list a few other middle-voice constructions on which I have worked in addition to the facilitative
[with Marta Grzybowska and Agnieszka Rembiałkowska] Middle voice reflexives and argument structure in Baltic.
In: Axel Holvoet & Nicole Nau, eds., Voice and Argument Structure in Baltic, Amsterdam-Philadelphia: John Benjamins, 2015, 181–208.
[with Anna Daugavet] Antipassive reflexive constructions in Latvian: A corpus-based analysis.
In: Studies in the Voice Domain in Baltic and Its Neighbours = Baltic Linguistics 11, 2020, 241–290.
[with Anna Daugavet] The facilitative middle in Baltic and North Slavonic: An overview of its variation.
In: Studies in the Voice Domain in Baltic and Its Neighbours = Baltic Linguistics 11, 2020, 291–342.
Finnish has a so-called agentive construction distinct from the passive. In the standard language
it is based on the so-called agentive participle, as in auto on pojan tekemä ‘the car is of the boy’s
making’; in the dialects it may be based on the passive participle. In a 2001 article I point out that
Latvian also has an agentive construction distinct from the passive (which is basically agentless
in this language); a similar construction in Lithuanian has developed into an agented passive. The
areal links are obvious and interesting in themselves, but there is a wider aspect to it. In their
2007 article on passives in the world’s languages, Keenan and Dryer mention ‘incorporation’ as one
of the strategies for expressing agents, citing not only English worm-eaten but also productive
syntactic patterns like Quechua Kuru miku-sqa-mi manzana-ø ka-rqa-n (bug eat-PTCP-COMMENT
apple-SUBJ be-PST-3) ‘The apple was bug-eaten’. This suggests the agentive construction might be a
minor cross-linguistic construction type alongside the passive and distinct from it.
Impersonals and Passives in Baltic and Finnic. In: Ö. Dahl & M. Koptjevskaja-Tamm, eds.,
The Circum-Baltic Languages: Typology and Contact. Vol. 2, Grammar and Typology,
Amsterdam: John Benjamins, 2001, 363–389. https://doi.org/10.1075/slcs.55.04hol
[with Anna Daugavet, Birutė Spraunienė and Asta Laugalienė], The agentive construction in Baltic and Fennic.
In: Minor Grams in Baltic, Slavonic and Fennic = Baltic Linguistics 10, 2019, 195–236.
Though the basic function of imperatives is to encode directive speech acts, it is well known that
they have non-directive uses as well: rhetorical imperatives (Go and catch a falling star) point out
the impossibility of an event, in some languages imperatives develop into a kind of narrative form
(Ukrainian A tut vin meni raptom i skažy... ‘then suddenly he said to me…’) etc. Until now there has
been no systematic research into the mechanisms leading from directive to non-directive meanings in
imperatives. It is sometimes suggested that the non-directive functions arise as a result of the basic
directive meaning fading away, being backgrounded etc. and other, less central aspects of imperatival
semantics thereby coming to the fore. In a series of articles I argue that it is precisely the
directive meanings themselves that develop into non-directive ones through semantic and pragmatics
mechanisms well attested in other domains of grammar. I discuss hypothetical, rhetorical, mirative,
historical, concessive, reportive and other imperatives. Different person and number forms of
imperatives have different non-directive extensions, and I have therefore availed myself of the
implicational model of the hierarchical structure of imperatival paradigms proposed in van der Auwera,
Goussev and Dobrushina’s 2004 article “A semantic map for imperatives-hortatives” in order to build an
enhanced semantic map also including the non-directive functions.
[with Anna Daugavet and Liina Lindström], Insubordinated concessive imperatives: An areal constructional idiom type.
In: Minor Grams in Baltic, Slavonic and Fennic = Baltic Linguistics 10, 2019, 307–354.
Towards an enhanced semantic map for imperatives. To appear in STUF Language Typology and Universals.
Sperber and Wilson (1986) introduce the notion of interpretive use, pointing out that linguistic utterances
may refer not only to states of affairs but also to other linguistic utterances in virtue of bearing a
resemblance to them. Within interpretive use they distinguish quotation and echoic interpretations. In my
2018 article I argue for distinguishing between the marking of quotative and echoic use, though this is not
generally practised in the literature. The question is partly terminological, and if people have no problem
with using the term ‘quotative’ where there is no actual speaker and no actual utterance to be quoted, just
a situation in which somebody could have said something, I have no problem with it either. In the domain of
grammatical marking, however, I prefer to say that one grammatical form echoically reproduces another form
rather than to say it quotes it. I have availed myself of the notion of echoic used to explain the East
Slavonic ‘necessitive imperative’ (Ukrainian Usi vidpočyvajut’, a ja pracjuj ‘Everybody is having a rest
while I have to work’), as well as a number of other grammatical forms and constructions. For this, see the
2011 article co-authored with Jelena Konickaja as well as my work on non-directive imperatives.
Epistemic modality, evidentiality, quotativity and echoic use. In Zlatka Guentchéva, ed.,
Epistemic Modalities and Evidentiality in Cross-Linguistic Perspective. Berlin: De Gruyter Mouton, 2018, 242–258.
Talmy’s typology of event integration (originally called ‘event conflation’) has spawned an impressive
body of research, but not all of its aspects have received the same amount of attention. Alternations
involving a spatial motion path (English the bottle floated into the cave vs Spanish la botella entrò
en la cueva flotando) are most prominent, with paths to result (English blew the candle out vs Spanish
apagué la vela soplándola) as a distant second. Hardly any thought has been given to constructions where
one incremental path is grafted onto another, as in the case of what Talmy calls ‘temporal contouring’
(German Ich habe den Brief fertiggeschrieben vs Spanish terminé de escribir la carta). In a recent article
I discuss a Lithuanian verb-framed construction for the achievement of an excessive value of some parameter
of an incremental process, as in pa-aukštino langus iš-kirsti (PFX-high-CAUS-PST.3 window.ACC.PL out.cut.INF)
‘cut out the windows too high’. Here as well we have an incremental path (viz. a path towards a normal value
of some parameter of a process) grafted onto another. Such constructions are not unimportant as discussions
have hitherto centered exclusively on the basic motion-path and path-to-result patterns, which has led some
researchers (like Beth Levin and Malka Rappaport) to describe the Talmyan typology as epiphenomenal, being
derived from the manner-result complementarity in lexicalization. I think, however, that the basic
motion-path and path-to-result patterns are only part of the story.
In several articles I have tried to make the point that while the notion of non-canonical subjects and
objects makes sense, that of ‘non-canonical marking of subjects and objects’ (the title of a 2001 volume
edited by Aikhenvald, Dixon and Onishi) basically does not. The idea of a non-canonically marked subject
might readily come to the mind when looking at Icelandic, which seems to have noun phrases displaying
(almost) every feature of subjecthood except nominative marking. But not all languages are like Icelandic;
when we look at Baltic or Slavonic, the idea seems less attractive. Here we find, e.g., verbs like ‘please’
taking datival experiencer arguments that display a set of typical subject properties in virtue of being
discourse-prominent (they are animate, default topics, occur clause-initially, may have certain control
properties etc.), but one needs only topicalize a nominative-marked stimulus argument to restore it to
canonical subjecthood. Subjecthood is a gradable property and nominative marking is itself a subject
property rather than just a means of encoding ‘underlying’ subjecthood. The very notion of underlying
subjecthood is mistaken, as subjecthood is a cluster of properties of different levels. In the Baltic
and Slavonic languages, which mostly have rich case systems allowing for relatively consistent marking
of semantic roles, canonical subjecthood is less generalized than in English or French.
Obliqueness, quasi-subjects and transitivity in Baltic and Slavonic. In:
Ilja A. Seržant & Leonid Kulikov, eds. The Diachronic Typology of Non-Canonical Subjects
(SLCS 140), Amsterdam & Philadelphia: John Benjamins, 2013, 257–282.
Among Indo-European languages, Germanic, Baltic and Slavonic are known as having developed definite and
indefinite adjectives. While in Germanic ‘strong’ and ‘weak’ forms are now mostly governed by determiners
(or their absence), and Slavonic has largely generalized the definite forms, in Baltic the distinction has
remained fully alive. The distribution of definite adjectives in contemporary Lithuanian has recently been
studied in detail in Ringailė Trakymaitė’s PhD thesis (2021). An interesting aspect of adjectival
definiteness marking is the occurrence of definiteness mismatches, that is, the use of definite adjectives
in indefinite noun phrases. They are associated mainly with the extension of generic definiteness to
specific noun phrases. Interesting developments can be observed especially in Latvian, which has a parallel
to the Scandinavian constructions known as ‘absolute positives’. Definite forms are used to refer to a
category created ad hoc (a device comparable to the ‘recognitional’ use of demonstratives). In Latvian it
is observed with negation, as in neesmu nekāds lielais eksperts ‘I am not a big (DEF) expert’ (or ‘I am not
one of these big experts’).
[with Birutė Spraunienė] Towards a semantic map for definite adjectives in Baltic,Baltic Linguistics 3, 2012, 65–99.
[with Birutė Spraunienė] Ad hoc taxonomies: A Baltic parallel to the Scandinavian absolute positives.
In Ērika Sausverde & Ieva Steponavičiūtė, eds., Fun and Puzzles in Modern Scandinavian Studies. Vilnius:
Vilnius University, 2014, 47–62.
My work on aspect is concerned mainly with what Bybee and Dahl have called ‘bounder-based aspect systems’,
that is, aspect systems where perfectivity is introduced by bounders—prefixes with an originally spatial
meaning. They have recently been studied in an areal-typological context by Peter Arkadiev. Languages with
bounder-based aspect include Baltic and Slavonic, Hungarian and Georgian. I have dealt with both Slavonic
and Baltic aspect. In a 1991 article, I argue for the relevance of what is now sometimes called ‘pofectives’
(perfectives expressing limited duration) for the grammaticalization of bounder-based aspect. In other work I
have taken issue with recent claims that there is a difference of principle between Slavonic and Baltic
verbal aspect, the former being grammatical while the latter is lexical. In a recent article I draw a parallel
between certain uses of ‘type-focusing’ imperfectives and irrealis forms in evaluating contexts, where I argue
both may have an ‘unanchoring’ function—for this notion cf. the section on irrealis semantics.
Semantic variables and the meaning of Polish verbal aspect. In: Barentsen, A.A., Groen, B.M.
& Sprenger, R., eds., Studies in West Slavic and Baltic Linguistics. Amsterdam-Atlanta: Rodopi, 1991,
The Baltic view on verbal aspect (to appear in the proceedings of the 13th International
Conress of Baltic Studies, Riga 2021)
In 2000, Casper de Groot introduced the notion of absentive construction, a cross-linguistic construction
type denoting the absence of a person from the deictic centre for the purpose of some specific activity.
The notion is now well established, but somehow authors dealing with it have failed to raise the question
why many language have only past-tense absentives, or past-tense absentives with rare and marginal
extensions into the present. In an article I co-authored in 2019 the notion of ambidirectional, a
construction denoting two-way motion with a specific purpose in the past, is proposed. Its extension into
the present leads to the rise of an absentive (though this need not be the only source for absentives).
It is noteworthy that the East Slavonic languages, which have so-called ‘indeterminate’ motion verbs used
(among other things) for two-way motion in the past, have not developed absentives.
[with Vaiva Žeimantiene], Absentives or ambidirectionals? Motion-cum-purpose constructions with
‘be’ and the infinitive in Baltic and elsewhere. In: Minor Grams in Baltic, Slavonic and Fennic
= Baltic Linguistics 10, 2019, 155–193.
I have dealt with a number of other topics, basically with reference to the Baltic (and often also
Slavonic) languages. These topics include: impersonals and passives, proximative constructions, secondary
predicates, extended uses of morphological causatives, semantic properties of complementizers, pain-verb
constructions, implicatives, predicative possession, external possessor constructions etc.