Content uploaded by Anna Endresen
Author content
All content in this area was uploaded by Anna Endresen on Jun 09, 2021
Content may be subject to copyright.
Content uploaded by Anna Endresen
Author content
All content in this area was uploaded by Anna Endresen on Mar 26, 2021
Content may be subject to copyright.
John Benjamins Publishing Company
This is a contribution from Belgian Journal of Linguistics 34
© 2020. John Benjamins Publishing Company
This electronic le may not be altered in any way. The author(s) of this article is/are permitted to use
this PDF le to generate printed copies to be used by way of oprints, for their personal use only.
Permission is granted by the publishers to post this le on a closed server which is accessible only to
members (students and faculty) of the author's/s' institute. It is not permitted to post this PDF on the
internet, or to share it on sites such as Mendeley, ResearchGate, Academia.edu.
Please see our rights policy on https://benjamins.com/content/customers/rights
For any other use of this material prior written permission should be obtained from the publishers or
through the Copyright Clearance Center (for USA: www.copyright.com).
Please contact rights@benjamins.nl or consult our website: www.benjamins.com
How to build a constructicon in ve years
The Russian example
Laura A. Janda,
Anna Endresen,
Valentina Zhukova,
Daria Mordashova, and Ekaterina Rakhilina,
UiT The Arctic University of Norway |National Research University
Higher School of Economics |Institute of Linguistics of the Russian
Academy of Sciences |Lomonosov Moscow State University |
Vinogradov Institute for Russian language of the Russian Academy of
Sciences
We provide a practical step-by-step methodology of how to build a full-scale
constructicon resource for a natural language, sharing our experience from
the nearly completed project of the Russian Constructicon, an open-access
searchable database of over , Russian constructions (https://site.uit.no
/russian-constructicon/). The constructions are organized in families, clus-
ters, and networks based on their semantic and syntactic properties, illus-
trated with corpus examples, and tagged for the CEFR level of language
prociency. The resource is designed for both researchers and L learners of
Russian and oers the largest electronic database of constructions built for
any language. We explain what makes the Russian Constructicon dierent
from other constructicons, report on the major stages of our work, and
share the methods used to systematically expand the inventory of construc-
tions. Our objective is to encourage colleagues to build constructicon
resources for additional natural languages, thus taking Construction Gram-
mar to a new quantitative and qualitative level, facilitating cross-linguistic
comparison.
Keywords: constructicon, construction grammar, Russian, corpus
1. Why build a constructicon?
If you are a linguist working on individual constructions in a language X, you
might wonder why one should bother building a constructicon resource, and even
if you accept this challenge, you might wonder where to start, how to proceed,
and how to organize this endeavor.
https://doi.org/10.1075/bjl.00043.jan
Belgian Journal of Linguistics Volume 34 (2020), pp. 161–173. issn 0774-5141 |e‑issn 1569-9676
© John Benjamins Publishing Company
The primary objective of this article is to address linguists working in the
framework of Construction Grammar in order to inspire and motivate them to
build constructicon resources for their languages, by presenting the ideas and tools
we utilized in building a constructicon for Russian.
Constructions are the elements that structure languages (Fillmore, Kay, and
O’Connor ; Cro ; Goldberg ). In essence, each language is a struc-
tured inventory of constructions, and thus it is theoretically possible to model an
entire language as a constructicon. The term constructicon refers to both a struc-
tured inventory of grammatical constrtuctions and a description of this inventory.
Today, constructicon resources are under development for only a handful of lan-
guages, namely English, Swedish, German, Brazilian Portuguese, Japanese, and
Russian (Lyngfelt, Borin, et al. ).
The growth of this emergent sub-discipline of Construction Grammar, termed
‘constructicography’, promises crucial benets both for linguists and for language
learners. Our understanding of how networks of constructions work largely de-
pends on the amount of publicly available data on constructions. Moreover, thor-
oughly annotated and searchable databases of constructions can serve the needs of
Natural Language Processing (NLP). Recognizing semi-compositional construc-
tions in running text is crucial for machine translation, extraction of information
and other applications (Dunietz, Levin, and Petruck ). It is now high time to
build comparable constructicon resources for additional natural languages.
In what follows, we provide a practical guide for how to build a full-scale con-
structicon resource for a natural language, sharing our experience from the Russ-
ian Constructicon project (https://site.uit.no/russian-constructicon/). We report
on a group project carried out over a ve-year period (–) that succeeded
to collect, describe and illustrate an inventory of over multi-word construc-
tions of Contemporary Standard Russian (Janda et al. ; Endresen et al. ).
We start with a brief overview of characteristics of the Russian Constructicon
resource (Section ), then outline the major stages of our work, focusing on meth-
ods for expanding and structuring the inventory of constructions (Sections and
). Section presents an illustration of our method. The article concludes with
recommendations based on our experience.
2. Features of the Russian Constructicon resource
The Russian Constructicon resource provides a large-scale model of the system of
Russian constructions for the benet of linguists, second language learners, and
NLP. The goal of modelling a language as a constructicon and the needs of users
have motivated the design of the project. The scope and organization of the pro-
ject are detailed in this section.
162 Laura A. Janda et al.
© 2020. John Benjamins Publishing Company
All rights reserved
2.1 The scope of the project
In the broadest sense, a construction is any recurrent form-meaning pairing in
a language, at any level of complexity, from morpheme through lexeme through
phrase to discourse structure (Goldberg , ). The constructicon of a language
is an open-class inventory that is potentially limitless. Therefore it would be unre-
alistic to expect to produce a comprehensive constructicon resource. Furthermore,
many items that a comprehensive constructicon should contain are already avail-
able in existing reference works, such as dictionaries (that contain lexeme-level
constructions), phraseological dictionaries (that contain idioms where all the slots
are xed), and grammars (that explain basic schematic types of sentences and use
of function words).
What remains are entrenched multi-word expressions that contain at least one
open (not xed) slot, and these are the strategic target of the Russian Constructicon
resource. More precisely we have collected partially schematic phrases that are
repeatedly used in Russian to convey meanings that range along a scale from fully
transparent (compositional) to opaque. A salient feature of such constructions is
the fact that their form, while motivated, is also to some extent arbitrary.
The following examples illustrate the type of constructions targeted in the
Russian Constructicon resource, namely constructions that are neither merely
schematic sentence types nor fully xed idioms. A typical construction in the re-
source includes a xed part, called the ‘anchor’ and one or more slots that can
be lled with a restricted set of lexemes. This type of construction is partially
schematic because part of it (the anchor) is xed, while the rest is variable. Par-
tially schematic constructions are likewise the focus of the Swedish constructicon
resource (Lynfgelt, Bäckström, et al. , ), and are referred to as ‘constructions
of microsyntax’ in the Russian linguistic literature. For example, in the construc-
tion net čtoby VP-Inf, Cl [instead of X-ing, Y] illustrated in (), the anchor is net
čtoby literally ‘no in-order’, and the open slots are the innitive verb (here lled by
podoždat’ ‘wait’) and the following clause.
() Net
no
čtoby
in.order
podoždat’,
wait-
on
he
uše-l-ø
leave--.
bez
without
nas!
we.
‘Instead of having waited for us, he just le!’
This example strongly illustrates non-compositionality since it is not possible to
predict the meaning of this construction based on its components. Linguists, learn-
ers, and NLP specialists face challenges in accounting for such constructions.
In addition to non-compositional constructions like the one above, high-
frequency compositional constructions are targeted in our project, such as (NP-
Dat) Cop možno VP-Inf [possible to X ] illustrated in (), where the adverb možno
‘possible’ is added to an innitive to mean ‘it is possible to X’.
How to build a constructicon in ve years 163
© 2020. John Benjamins Publishing Company
All rights reserved
() Do
to
Moskv-y
Moscow-
iz
from
London-a
London-
možno
possible
dolete-t’
fly-
za
behind
četyr-e
four-
čas-a.
hour-.
‘It is possible to y from London to Moscow in four hours.’
Even such a construction is somewhat arbitrary, since it would be theoretically
possible to use a dierent adverb or a dierent form of the verb (perhaps a gerund
or a deverbal noun), however in Russian the usual way to express this meaning is
with precisely this construction.
Further types of compositional but arbitrary constructions targeted in the
Russian Constructicon resource include constructions where the anchor is a verb
with a specic argument structure or where a derivational morpheme serves as part
of the anchor. For example, in NP-Nom načinat’ NP-Ins [X begin as Y ] illustrated
in (), the conventionalized choice of the instrumental case with the verb načinat’
‘begin’ indicates the status of the person as a salient and temporary property.
() On
he
načina-l-ø
begin--.
učitel-em.
teacher-.
‘He began his career as a teacher.’
An example of a derivational morpheme embedded in a construction is NP-Nom
pere-Verb vse NP-Acc.Pl[re-X all Ys] as in (), where the prex pere- species dis-
tributive semantics.
() Ja
I
pere-my-l-ø
-wash--.
vs-e
all-.
tarelk-i
dish-.
v
in
dom-e.
house-.
‘I washed all of the dishes in the house.’
While the instrumental case and the prex pere- are motivated from the perspective
of Russian grammar, their use in these constructions is also an arbitrary language-
specic fact that must be accounted for by linguists and mastered by learners.
In sum, the Russian Constructicon resource targets recurrent linguistic pat-
terns that ‘fall between the cracks’ of dictionaries and grammars, yet are essential
to full mastery of the language.
Some constructicons are connected to a FrameNet resource, based on Fill-
more’s work on frames. Accordingto Fillmore and Atkins (, ), a frame is a cog-
nitive structure, the knowledge of which is presupposed for the concepts encoded
by constructional constituents. Though Russian lacks a fully developed FrameNet
resource, there exists a FrameBank (https://github.com/olesar/framebank) that
focuses primarily on verbs and their argument structure. The data of FrameBank
and the Russian Constructicon partially overlap. In the future, we might add cross-
references to frames described in the Russian FrameBank where appropriate.
164 Laura A. Janda et al.
© 2020. John Benjamins Publishing Company
All rights reserved
2.2 The presentation of constructions
The presentation of constructions in the Russian Constructicon resource is tai-
lored to the needs of the projected users: linguists, second language learners, and
NLP researchers. To this end, we provide both detailed linguistic classication
and user-friendly guidance. Each construction is supplied with:
– a name, which is a schematic description of the construction; such as net čtoby
VP-Inf, Cl [instead of X-ing, Y]
– a brief illustration; such as ()
– a denition stated in non-technical language in Russian (with translations
into English and Norwegian); in this case: “The construction indicates that
the speaker expresses dissatisfaction with the fact that the interlocutor has not
taken a given action or is undertaking or has undertaken a dierent action.”
– a CEFR language prociency level (from A to C) to help learners target appro-
priate constructions; in this case C
– a series of semantic and syntactic tags
– a list of common llers for the open slot(s)
– a usage label specifying the type of speech (Neutral, Colloquial, Formal, Ob-
solete)
– a structure in terms of Universal Dependencies (https://universaldependencies
.org)
– three to ve corpus examples from the Russian National Corpus (www.rus
corpora.ru)
In addition, both the denition and the corpus examples are tagged for semantic
roles (Agent, Experiencer, etc.). All of the information about each construction is
searchable. For example, linguists can search for semantic and syntactic parame-
ters, learners can search for constructions at a given prociency level, and both
types of users can enter strings (for example, of anchor words) to search for specic
constructions. The Universal Dependency structure, the glossing system, and the
lists of common llers of the slots serve the purposes of Natural Language Process-
ing, facilitating automatic recognition of constructions in authentic Russian texts.
The system of semantic tags is based on terminology from typological literature
(cf. the “universal grammatical set of meanings”, Plungian , ). Taken together,
these features make the Russian Constructicon a multi-functional resource, de-
signed for language pedagogy, language research, and language technology. Among
other constructicon projects, only the Swedish Constructicon (Lyngfelt,
Bäckström, et al. , , ) pursues pedagogical goals and has been created not
only for linguists but also for learners of Swedish.
How to build a constructicon in ve years 165
© 2020. John Benjamins Publishing Company
All rights reserved
3. Reaching and exceeding a critical mass of constructions
Linguistically, we can classify constructions according to their semantics and their
formal structures. However, the classication becomes reliable only aer a repre-
sentative sample has been obtained. A critical mass of constructions is needed in
order to establish their classication, which is uncertain prior to that point. In other
words, we had to repeatedly cycle through the tasks of collecting and classifying
constructions in order to arrive at a stable system which could then be exploited for
further expansion of the constructicon with only minor adjustments. Our process
proceeded in three stages, visualized in Figure as the Initial inventory, Corpus-
based expansion, and System-based expansion. Numbers inside the bars reect the
quantity of constructions added in each stage, and dates indicate the approximate
timing of the stages.
Figure 1. Stages of the Russian Constructicon project
The Initial inventory of constructions was amassed manually in Stage from
a variety of sources including textbooks for learners of Russian (especially Janda
and Clancy ) and scholarly literature on Russian constructions (especially
Rakhilina ), as well as a crowd-sourced Google spreadsheet. At this stage we
decided what kinds of constructions to focus on in our project (see Section .),
established most of the conventions that would be used in the presentation of con-
structions (see Section .) and began to explore the semantic and syntactic sys-
tem of the constructicon (see Section ). This stage involved continuous revisions
in our procedure as we grappled with the dimensions of the project.
The Corpus-based expansion in Stage continued the manual heterogeneous
collection of constructions, at this stage culled from running texts of various kinds,
particularly those that contain dialogues and spoken discourse, as well as an auto-
matically extracted list of highly frequent collocations attested in the Russian
National Corpus. In this stage, we added constructions to the Initial inventory.
In addition to adding constructions, we continued the work on classication of
semantic and syntactic types, using the new constructions to verify and rene the
166 Laura A. Janda et al.
© 2020. John Benjamins Publishing Company
All rights reserved
classication. Once we had reached a critical mass of over one thousand construc-
tions, the classication became stable and robust enough to facilitate the identica-
tion of ‘families’ of constructions (see Section ). In other words, on the basis of our
semantic and syntactic tags we were able to discover groups of constructions that
were internally relatively homogeneous.
Families of constructions served as the basis for the more rapid and extensive
System-based expansion of the constructicon in Stage , which more than doubled
the size of the inventory to over , items. We examined semantic families of
constructions found in the database and searched for their synonyms, antonyms,
and related constructions containing the same or similar anchor words in order
to ll gaps in each family. Thus the classication system facilitated addition of
constructions in a signicantly more ecient manner. This stage yielded not only
quantitative but also qualitative change in the constructicon: semantic classica-
tion of constructions turned what initially was a list of unrelated items into a struc-
tured system of constructions.
4. Identifying families: Theoretical motivation and methodology
4.1 Theoretical motivation
One of the tenets of Construction Grammar is the idea that constructions are
related to each other. Following the example of Goldberg () and her analysis
of the English Subject Auxiliary Inversion family of constructions, we have devel-
oped the means to transform the inventory of constructions into a structured sys-
tem. One of the crucial challenges of a constructicon resource is to reveal and
represent this system, that is, the complex relationships (both hierarchical and lat-
eral) between constructions. One strategy is to focus on the relationship of parent
vs daughter constructions, i.e. a more abstract schema vs its specic instantiation.
In addition, we identify meaningful groupings: families that form clusters, and
ultimately networks.
We dene a family of constructions as a relatively homogeneous group of about
two to nine constructions that exhibit family resemblance in that they share some
semantic, syntactic (function in a clause and structure of the xed part), and struc-
tural properties (e.g. reduplication, negation, inversion, etc.). Family resemblance
means that the constructions in a family share various subsets of these properties.
The families within a cluster in turn share properties in a prototypical vs. periph-
eral distribution. We have elaborated a multi-level set of semantic and syntactic
tags that facilitate identication of families and clusters.
How to build a constructicon in ve years 167
© 2020. John Benjamins Publishing Company
All rights reserved
4.2 Methodology
Annotation was undertaken by a panel of three native speakers who worked to
achieve consensus on the tagging of each construction. A number of semantic and
syntactic tags were assigned to each construction by the panel. The annotation was
continuously rened and cross-checked by the entire panel, minimizing subjec-
tivity and guaranteeing consistency. In this process we took into account existing
scholarship relevant to semantic and syntactic classication, from both Russian
and typological scholarly traditions (Plungian ).
In all we employ general semantic tags, many of which have subtags, yield-
ing an overall inventory of subtags. Over of the constructions bear more
than one semantic tag. Figure displays the distribution of the most frequent gen-
eral semantic tags. Figure displays the distribution of constructions across eleven
syntactic tags.
Figure 2. Distribution of constructions across twenty most frequent general semantic
tags
We investigated the intersection of semantic and syntactic classications to identify
meaningful groupings of constructions. Among the constructions that received
each general semantic tag, we examined syntactic patterns in order to nd more
homogeneous groups of constructions. Thus, we arrived at smaller groups of –
constructions that shared more or less the same syntactic structure and more nar-
rowly specied semantics. These smallest groups we call families. We furthermore
168 Laura A. Janda et al.
© 2020. John Benjamins Publishing Company
All rights reserved
Figure 3. Distribution of constructions across general syntactic tags
examined how families are related to each other within clusters and how clusters
comprise networks. As a rule, our general semantic tags correspond to networks,
and the subtypes correspond to clusters. An illustration of this approach is pre-
sented in Section .
5. Turning a list into a structured inventory
We illustrate the method outlined in Section with the network of Prohibitive con-
structions diagrammed in Figure , consisting of two clusters and a total of eleven
families.
Figure 4. Network of prohibitive constructions
How to build a constructicon in ve years 169
© 2020. John Benjamins Publishing Company
All rights reserved
In Figure , boxes represent families indexed as cluster:family, followed by a brief
description and illustrative example. Thick boxes indicate prototypes. Lines with
arrows indicate semantic transitions. Lines without arrows indicate syntactic/for-
mal similarities. Dotted lines and arrows indicate weaker relationships. Thick ar-
rows indicate overlap with other networks of constructions.
Whereas constructions in Cluster ask a hearer to refrain from doing some-
thing, constructions in Cluster express ‘continuative prohibition’, asking a hearer
to stop doing something. All constructions in Cluster contain overt markers of
negation; such markers are absent from Cluster . Cluster is centered around its
prototype, family :, containing negated imperative constructions. Lines represent
the relationships that hold among families and are tagged for semantic transitions
and shared formal properties. A semantic transition to generalized prohibitions
connects : to :, with transitions to the remaining families in Cluster labeled
in Figure . Prohibitions in : can be either generalized or individual, indicated
by a dotted arrow, and : shares the syntactic form of predicative with :. Three
families in Cluster (:, :, and :) share constructions across other networks
(Request, Intensity, and Warning), indicated by the thick arrows.
Cluster is connected to Cluster through three pairs of families. In each pair,
the semantic transition is from standard prohibition in Cluster to continuative
prohibition in Cluster . In both clusters, families to the le represent general-
ization and attenuation, as opposed to more combative prohibitions on the right.
In addition, the two prototypical families (: and :) share the syntactic form of
imperative (also shared by :) and families : and : share the form of predica-
tive. The po- prex is a necessary feature of : and :, and optionally found in
:, where there is also some use of imperative forms.
The Prohibitive network demonstrates the complex of semantic and formal
properties that structure the constructicon.
6. Conclusion
We hope that this article will encourage the building of constructicons for a wider
variety of languages to serve both language learners and linguists. While the Russ-
ian Constructicon represents just one possible model, we can share lessons that
from our experience can be valuable to other similar projects. This is not a project
for an individual; it is essential to build a team of researchers because a construc-
ticon requires a variety of skills and a long-term commitment. As with any col-
laborative project, funding is essential. We found that it was possible to ‘package’
funding for the Russian Constructicon under the umbrella of grant projects pri-
marily aimed at language pedagogy and international cooperation. A strategic
170 Laura A. Janda et al.
© 2020. John Benjamins Publishing Company
All rights reserved
focus on constructions that are otherwise underrepresented in pedagogical and
reference works helps to keep the project manageable and also makes it easier
to ‘sell’ in grant proposals. A further ‘selling point’ is a user-friendly design that
addresses the needs of multiple audiences: the Russian Constructicon is a resource
both for learners and for linguists. In terms of presentation, we started by ‘piggy-
backing’ on an existing architecture (the Swedish Constructicon), making it possi-
ble to work through the rst two stages of our project without having to start from
scratch with the design of an interface. We are grateful for the big advantage this
gave us, which ultimately made it possible to envision something that would bet-
ter represent the Russian Constructicon. Once we began to uncover the relation-
ships among constructions (illustrated in Section ), we had something that was no
longer an inventory, but a system, and we needed a new interface that could do jus-
tice to that structure. We look forward to further expanding and rening the Russ-
ian Constructicon in its new design and welcome comments and critique.
Funding
We acknowledge funding from the Norwegian Agency for International Cooperation and Qual-
ity Enhancement in Higher Education: grants NCM-RU-/ and CPRU-/.
References
Cro, William. . Radical Construction Grammar. Syntactic Theory in Typological
Perspective. Oxford: Oxford University Press.
https://doi.org/10.1093/acprof:oso/9780198299554.001.0001
Dunietz, Jesse, Lori Levin, and Miriam R. L. Petruck. . “Construction Detection in a
Conventional NLP Pipeline.” In The papers from the 2017 AAAI Spring Symposium on
Computational Construction Grammar and Natural Language Understanding. Technical
Report SS-17-02, –.
Endresen, Anna, Valentina Zhukova, Daria Mordashova, Ekaterina Rakhilina, and
Olga Lyashevskaya. . “Russkij konstruktikon: Novyj lingvističeskij resurs, ego
ustrojstvo i specika [The Russian Constructicon: A new linguistic resource, its design
and key characteristics].” In Computational Linguistics and Intellectual Technologies.
Papers from the Annual International Conference “Dialogue-2020”, –. Published
on-line.
Fillmore, Charles J., and Beryl T. Atkins. . “Toward a Frame-Based Lexicon: The
Semantics of RISK and Its Neighbors.” In Frames, Fields, and Contrast: New Essays in
Semantics and Lexical Organization, ed. by Adrienne Lehrer, and Eva Kittay, –.
Hillsdale, NJ: Lawrence Erlbaum.
How to build a constructicon in ve years 171
© 2020. John Benjamins Publishing Company
All rights reserved
Fillmore, Charles J., Paul Kay, and Mary C. O’Connor. . “Regularity and Idiomaticity in
Grammatical Constructions: The Case of Let Alone.” Language (): –.
https://doi.org/10.2307/414531
Goldberg, Adele. . Constructions at Work. The Nature of Generalization in Language.
Oxford: Oxford University Press.
Janda, Laura A., and Steven J. Clancy. . The Case Book for Russian. Bloomington, IN:
Slavica Publishers.
Janda, Laura A., Olga Lyashevskaya, Tore Nesset, Ekaterina Rakhilina, Francis M. Tyers. .
“A Constructicon for Russian: Filling in the Gaps.” In Constructicography: Constructicon
Development Across Languages, ed. by Benjamin Lyngfelt, Lars Borin, Kyoko Ohara, and
Tiago T. Torrent, –. Amsterdam: John Benjamins. https://doi.org/10.1075/cal.22.06jan
Lyngfelt, Benjamin, Linnéa Bäckström, Lars Borin, Anna Ehrlemark, and Rudolf Rydstedt.
. “Constructicography at Work: Theory Meets Practice in the Swedish
Constructicon.” In Constructicography: Constructicon Development Across Languages, ed.
by Benjamin Lyngfelt, Lars Borin, Kyoko Ohara, and Tiago T. Torrent, –.
Amsterdam: John Benjamins. https://doi.org/10.1075/cal.22.03lyn
Lyngfelt, Benjamin, Lars Borin, Kyoko Ohara, and Tiago T. Torrent (eds.). .
Constructicography: Constructicon Development Across Languages. Amsterdam: John
Benjamins. https://doi.org/10.1075/cal.22
Plungian, Vladimir A. . Vvedenie v grammatičeskuju semantiku: Grammatičeskie značenija
i grammatičeskie sistemy jazykov mira [An introduction to grammatical semantics:
Grammatical meanings and grammatical systems in the languages of the world]. Moscow:
Russian State University for the Humanities Press.
Rakhilina, Ekaterina V. (ed.) . Lingvistika konstrukcij [Linguistics of constructions].
Moscow: Izdatel’stvo Azbukovnik.
Abbreviations
nd person
accusative
dative
future
genitive
imperative
innitive
instrumental
imperfective
locative
NP noun phrase
masculine
plural
PP personal pronoun
past
singular
VP verb phrase
() optional element
172 Laura A. Janda et al.
© 2020. John Benjamins Publishing Company
All rights reserved
Authors’ addresses
Laura A. Janda
Department of Language and Culture
UiT The Arctic University of Norway
Hansine Hansens veg
N- Tromsø
Norway
laura.janda@uit.no
Anna Endresen
Department of Language and Culture
UiT The Arctic University of Norway
Hansine Hansens veg
N- Tromsø
Norway
anna.endresen@uit.no
Valentina Zhukova
School of Linguistics
National Research University Higher School
of Economics
Myasnitskaya Street
Moscow
Russia
valentina.zh@gmail.com
Daria Mordashova
Minority Language Research and
Preservation Lab
Institute of Linguistics, Russian Academy of
Sciences
bld. Bolshoy Kislovsky Lane
Moscow
Russia
mordashova.d@yandex.ru
Ekaterina Rakhilina
School of Linguistics
National Research University Higher School
of Economics
Myasnitskaya Street
Moscow
Russia
rakhilina@gmail.com
How to build a constructicon in ve years 173
© 2020. John Benjamins Publishing Company
All rights reserved
![Julia Prentice [Grosse]](https://i1.rgstatic.net/ii/profile.image/278460258897921-1443401507158_Q64/Julia-Prentice-grosse.jpg)
