The Polish Valence Dictionary (Walenty) is an electronic dictionary of subcategorisation frames for Polish verbs and quasi-verbal predicates. Some textual snapshots of the dictionary are made available at the bottom of this page, together with articles describing its format in detail. What follows is an increasingly obsolete overview, based on earlier versions of the dictionary.
The dictionary represents valence as a list of individual frames describing a particular verbal base with a particular aspect (perfective, imperfective, or bi-aspectual, listed as _). The actual argument structure is presented as a set of positions which must be filled by phrases of appropriate types and parameters. Individual positions may be marked for their status as a subject (subj) or a passivisable direct object (obj), and for their role in control relations with other positions in the argument structure (controller and controlee).
The resource has been produced as part of the CESAR project (Central and Southeast European Resources), as well as other projects carried out at ZIL IPI PAN, and is made available on META-SHARE.
See the bottom of this page for the latest versions of the dictionary and for papers describing the format of Walenty in more detail.
Format
The format of the dictionary (devised by the authors listed below) is based on the electronic version of Świdziński's dictionary, but includes a number of significant changes:
- Arguments and their parameters consistently use Latin- or English-based terms.
Sentential phrases (sentp) are split into three categories based on the number of parameters they have (cp for bare complementiser clauses, ncp for complementiser clauses with a correlative pronoun, prepncp for prepositional phrases involving a complementiser clause with a correlative pronoun).
Phrases representing arguments restricted in terms of semantic categories which may be expressed by a wider scope of syntactic constructions ("adverbial" phrases) are represented as xp and classified into specific subtypes.
Multi-word prepositions are represented as comprepnp.
Case requirements listed as the nominative and the accusative case in Świdziński's dictionary are represented as the structural case (str) in order to capture their alternation with other cases (as opposed to e.g. a lexical accusative present in prepositional phrases). In the subject position, the structural case may be nominative (in noun phrases) or a case represented as the accusative in the LFG grammar (in non-agreeing numeral phrases). In other (object) positions, the structural cases represent the accusative case or the genitive case (when the predicate is negated).
Subjects positions are listed as subj. This includes sentential subjects.
Passivisable (direct) objects are listed as obj. This includes non-accusative direct objects.
Control relations are represented, marking the controller and the controlee arguments. Control relations are involved e.g. in establishing the source of agreement for adjectival arguments.
- Coordination of different types of arguments is marked by listing arguments within a single syntactic position
- Information about implicit subjects and raised subjects is included
- Idiomatic arguments and structures are included where they cannot be captured by more general valence frames
Entry structure
The dictionary, in text format, consists of a list of valence frames. Every frame is associated with a lemma, in the following format:
base form: aspect: frames
The actual valence information is represented as a list of syntactic positions, expressed within curly braces and separated with plus signs. A position may include more than one type of argument if the arguments can be coordinated within the same position (arguments within a position are separated by semicolons).
Positions may bear special categories (e.g. subject, passivisable object), listed before the relevant position. The following categories are distinguished:
subj: grammatical subject (including non-nominal subjects)
implicit subjects are marked as subj{E}
subjects whose shape is transmitted from an embedded infinitival phrase are marked as subj,controller{E}
obj: passivisable object (regardless of case)
controller, controlee: control relations between arguments, playing a role in:
- agreement between adjectival phrases and controller arguments
agreement between prepositional phrases involving jak, jako, niż and controllier arguments
- establishing the controller subject of infinitival phrases
In the situation where different types of arguments may be coordinated within a single position, certain categories are only relevant for a subset of listed arguments:
possibility of passivisation (obj) applies to nominal and sentential arguments, but not infinitival ones (which may sometimes be coordinated with sentential arguments)
control (controlee) is not relevant for sentential arguments
Since the dictionary is syntactic in nature, only longest possible frames are listed - shorter frames are included within broader ones, regardless of differences in semantics, unless they differ in terms of control relations or the presence of the subject.
Types of arguments
Actual arguments listed in valence frames are categorised into following types, listed as argument(parameter1,parameter2,...):
np(case): nominal phrase
str: structural case (nominative if subject, accusative/genitive of negation otherwise)
pred: predicative case
part: partitive case (accusative or partitive genitive)
adjp(case): adjectival phrase
prepnp(preposition,case): prepositional-nominal phrase
comparative conjunctions jak, jako and niż are treated as prepositions governing the structural case
prepadjp(preposition,case): prepositional-adjectival phrase
postp: postprepositional case
comprepnp(complex preposition): complex (i.e. multi-word) preposition
cp(type): sentential (complementiser) phrase
ncp(case,type): sentential phrase with a correlative pronoun
prepncp(preposition,case,type): sentential phrase with a prepositional phrase correlative
- nonch: "nonchromatic" phrase (pronominal element not replaceable by a nominal phrase)
infp(aspect): infinitival phrase
perf: perfect aspect
imperf: imperfect aspect
_: any value of aspect
xp(category): "adverbial" phrases involving semantic requirements (expressible through adverbs, prepositional phrases, or sentential phrases)
locat: locative
abl: ablative
adl: adlative
perl: perlative
temp: temporal
dur: durative
mod: manner
advp(category): adverbial phrase expressible through adverbs only
pron: anaphoric adverbs tak and jak
misc: adverbs of degree and evaluation
locat, abl, adl, perl, temp, dur, mod: as in xp(category) above
lexnp(case,number,lemma,modification): idiomatic nominal arguments with lexical restrictions
natr: the structure of the argument cannot be expanded beyond the required form
atr: the structure of the argument can be freely expanded
ratr: the structure of the argument must be expanded (it occurs only with modifiers or complements)
batr: the argument must be modified by a bound possessive pronoun (własny/swój)
preplexnp(preposition,case,number,lemma,modification): idiomatic arguments embedded within prepositional phrase
fixed(string): fixed expressions
or: oratio recta, i.e. direct speech
refl: reflexive use marked through the word się
- E: implicit subject, transmitted subject when marked as controller
Sources
Świdziński, M. (1992). Realizacje zdaniowe podmiotu-mianownika, czyli o strukturalnych ograniczeniach selekcyjnych, in: A. Markowski (ed.), Opisać słowa, pp. 188–201, Dom Wydawniczy Elipsa, Warsaw.
Świdziński, M. (1994). Syntactic Dictionary of Polish Verbs, Uniwersytet Warszawski / Universiteit van Amsterdam.
Expanded electronic version of the Syntactic Dictionary of Polish Verbs, distributed with the Świgra parser (Słownik walencyjny analizatora), available at the Składnica page.
Authors
The format of the dictionary has been devised by:
- Filip Skwarski
Elżbieta Hajnicz <hajnicz AT SPAMFREE ipipan DOT waw DOT pl>
Agnieszka Patejuk <aep AT SPAMFREE ipipan DOT waw DOT pl>
Adam Przepiórkowski <adamp AT SPAMFREE ipipan DOT waw DOT pl>
- Marek Świdziński
Marcin Woliński <wolinski AT SPAMFREE ipipan DOT waw DOT pl>
Manual edition has been carried out by:
- Piotr Batko
- Joanna Filipczak
- Anna Grzeszak
- Marta Kalużna
- Marcin Opacki
- Paulina Rosalska
- Filip Skwarski
- Jakub Szymczak
- Maciej Zgondek
- Sebastian Żurowski
License
The data are available under a CC BY-SA license.
Available resources
The following text versions of the dictionary are available:
the January 2013 release of the Polish Valence Dictionary (30 January 2013)
the September 2013 release of the Polish Valence Dictionary (09 September 2013) – with a paper in the Polonica journal (in Polish) describing the formalism used in this version and providing some quantitative information (see below)
the March 2014 release of the Polish Valence Dictionary (04 March 2014) – with a paper (in English) in the proceedings of LREC 2014, describing the formalism used in this version and providing some quantitative information (see below)
the September 2014 release of the Polish Valence Dictionary (14 September 2014) – accompanied by a poster in Polish (to be) presented at the LXXII Annual Meeting of the Polish Linguistic Society in Cracow on 23rd September 2014
the May 2015 release of the Polish Valence Dictionary (27 May 2015)
the April 2016 release of the Polish Valence Dictionary (18 April 2016), available in plain text, PDF, and TEI format
The complete dictionary database, including corpus examples illustrating the use of individual frames, may also be viewed through the online application Slowal, which was used during the creation of the dictionary.
Errors
Please report any errors here.
Publications
