Slide 1
Slide 1 text
CLICS 2.0
Towards an Improved Handling of Cross-Linguistic Colexification
Patterns
Johann-Mattis List
Research Group “Computer-Assisted Language Comparison”
Department of Linguistic and Cultural Evolution
Max-Planck Institute for the Science of Human History
Jena, Germany
2017/07/03
very
long
title
P(A|B)=P(B|A)...
1 / 33
Slide 2
Slide 2 text
A long, long time ago...
A long, long time ago...
2 / 33
Slide 3
Slide 3 text
A long, long time ago... Predecessors
Predecessors: People and Ideas
3 / 33
Slide 4
Slide 4 text
A long, long time ago... Predecessors
Predecessors: People and Ideas
Haspelmath (2003): The geometry of grammatical meaning.
3 / 33
Slide 5
Slide 5 text
A long, long time ago... Predecessors
Predecessors: People and Ideas
Haspelmath (2003): The geometry of grammatical meaning.
François (2008): Semantic maps and the typology of colexification.
3 / 33
Slide 6
Slide 6 text
A long, long time ago... Predecessors
Predecessors: People and Ideas
Haspelmath (2003): The geometry of grammatical meaning.
François (2008): Semantic maps and the typology of colexification.
Cysouw (2010): Drawing networks from recurrent polysemies.
3 / 33
Slide 7
Slide 7 text
A long, long time ago... Predecessors
Predecessors: People and Ideas
Haspelmath (2003): The geometry of grammatical meaning.
François (2008): Semantic maps and the typology of colexification.
Cysouw (2010): Drawing networks from recurrent polysemies.
Steiner, Stadler, and Cysouw (2011): A pipeline for computational
historical linguistics.
3 / 33
Slide 8
Slide 8 text
A long, long time ago... Predecessors
Predecessors: People and Ideas
Haspelmath (2003): The geometry of grammatical meaning.
François (2008): Semantic maps and the typology of colexification.
Cysouw (2010): Drawing networks from recurrent polysemies.
Steiner, Stadler, and Cysouw (2011): A pipeline for computational
historical linguistics.
Urban (2011): Assymetries in overt marking and directionality in
semantic change.
3 / 33
Slide 9
Slide 9 text
A long, long time ago... Predecessors
Predecessors: Data
4 / 33
Slide 10
Slide 10 text
A long, long time ago... Predecessors
Predecessors: Data
Intercontinental Dictionary Series (IDS, Key and Comrie 2016)
offers 1310 concepts translated into about 360 languages, an
earlier version offered ca. 200 languages.
4 / 33
Slide 11
Slide 11 text
A long, long time ago... Predecessors
Predecessors: Data
Intercontinental Dictionary Series (IDS, Key and Comrie 2016)
offers 1310 concepts translated into about 360 languages, an
earlier version offered ca. 200 languages.
World Loanword Typology (WOLD, Haspelmath and Tadmor 2009)
offers 1430 concepts translated into 41 languages (some overlap
with IDS).
4 / 33
Slide 12
Slide 12 text
A long, long time ago... Predecessors
Predecessors: Techniques
Steiner, Stadler, and Cysouw (2011) present the idea to model
similarities between concepts by constructing a matrix from parts
of the IDS data that shows how often individual languages colexify
certain concepts.
5 / 33
Slide 13
Slide 13 text
A long, long time ago... Predecessors
Predecessors: Techniques
Steiner, Stadler, and Cysouw (2011) present the idea to model
similarities between concepts by constructing a matrix from parts
of the IDS data that shows how often individual languages colexify
certain concepts.
Cysouw (2010) shows how to use polysemy data to draw networks.
5 / 33
Slide 14
Slide 14 text
A long, long time ago... Initial Ideas
Initial Ideas
6 / 33
Slide 15
Slide 15 text
A long, long time ago... Initial Ideas
Initial Ideas
List, Terhalle, and Urban (2013) build on ideas of Cysouw (2010)
and Steiner, Stadler and Cysouw (2011) in using IDS data for
polysemy studies and in using network techniques to study the
data.
6 / 33
Slide 16
Slide 16 text
A long, long time ago... Initial Ideas
Initial Ideas
List, Terhalle, and Urban (2013) build on ideas of Cysouw (2010)
and Steiner, Stadler and Cysouw (2011) in using IDS data for
polysemy studies and in using network techniques to study the
data.
In contrast to earlier approaches, they use techniques for
community detection (Girvan and Newman 2002) to further analyse
the network, and to partition the concepts into communities which
seem to make intuitively sense, reminding of naturally derived
semantic fields.
6 / 33
Slide 17
Slide 17 text
A long, long time ago... Further Ideas
Further Ideas
7 / 33
Slide 18
Slide 18 text
A long, long time ago... Further Ideas
Further Ideas
Mayer, List, Terhalle, and Urban (2014) present an interactive way
to visualize cross-linguistic colexification data.
7 / 33
Slide 19
Slide 19 text
A long, long time ago... Further Ideas
Further Ideas
Mayer, List, Terhalle, and Urban (2014) present an interactive way
to visualize cross-linguistic colexification data.
List, Mayer, Terhalle, and Urban (2014) publish the database and
the web-application online, under the name CLICS (Database of
Cross-Linguistic Colexifications).
7 / 33
Slide 20
Slide 20 text
A long, long time ago... Further Ideas
Further Ideas
Mayer, List, Terhalle, and Urban (2014) present an interactive way
to visualize cross-linguistic colexification data.
List, Mayer, Terhalle, and Urban (2014) publish the database and
the web-application online, under the name CLICS (Database of
Cross-Linguistic Colexifications).
In contrast to earlier attempts, they increased the data by merging
IDS, WOLD, and additional datasets which they collected
themselves, thus containing 220 languages in total.
7 / 33
Slide 21
Slide 21 text
A long, long time ago... Further Ideas
Further Ideas
Mayer, List, Terhalle, and Urban (2014) present an interactive way
to visualize cross-linguistic colexification data.
List, Mayer, Terhalle, and Urban (2014) publish the database and
the web-application online, under the name CLICS (Database of
Cross-Linguistic Colexifications).
In contrast to earlier attempts, they increased the data by merging
IDS, WOLD, and additional datasets which they collected
themselves, thus containing 220 languages in total.
They also improved the community detection procedure by using
Infomap (Rosvall and Bergstrom 2008), an advanced algorithm
based on random walks in complex networks.
7 / 33
Slide 22
Slide 22 text
CLICS 1.0
CLICS 1.0
8 / 33
Slide 23
Slide 23 text
CLICS 1.0 Data
Data
9 / 33
Slide 24
Slide 24 text
CLICS 1.0 Data
Data
IDS (Key and Comrie 2007 version), of 233 language varieties,
178 included in CLICS.
9 / 33
Slide 25
Slide 25 text
CLICS 1.0 Data
Data
IDS (Key and Comrie 2007 version), of 233 language varieties,
178 included in CLICS.
WOLD (Haspelmath and Tadmor 2009), of 41 languages in
WOLD, 33 are included in CLICS.
9 / 33
Slide 26
Slide 26 text
CLICS 1.0 Data
Data
IDS (Key and Comrie 2007 version), of 233 language varieties,
178 included in CLICS.
WOLD (Haspelmath and Tadmor 2009), of 41 languages in
WOLD, 33 are included in CLICS.
Logos Dictionary (Logos Group), of dictionaries for more than 60
different languages, 4 languages were manually extracted and
included in CLICS.
9 / 33
Slide 27
Slide 27 text
CLICS 1.0 Data
Data
IDS (Key and Comrie 2007 version), of 233 language varieties,
178 included in CLICS.
WOLD (Haspelmath and Tadmor 2009), of 41 languages in
WOLD, 33 are included in CLICS.
Logos Dictionary (Logos Group), of dictionaries for more than 60
different languages, 4 languages were manually extracted and
included in CLICS.
Språkbanken project (University of Gothenburg) offers 8 word lists
for SEA languages, 6 were included in CLICS.
9 / 33
Slide 28
Slide 28 text
CLICS 1.0 Methods
Methods
Problems
10 / 33
Slide 29
Slide 29 text
CLICS 1.0 Methods
Methods
Problems
(A) Data cannot be displayed fully, complexity needs to be
reduced.
(B) Data is noisy and needs to be corrected.
10 / 33
Slide 30
Slide 30 text
CLICS 1.0 Methods
Methods
Problems
(A) Data cannot be displayed fully, complexity needs to be
reduced.
(B) Data is noisy and needs to be corrected.
Solutions
10 / 33
Slide 31
Slide 31 text
CLICS 1.0 Methods
Methods
Problems
(A) Data cannot be displayed fully, complexity needs to be
reduced.
(B) Data is noisy and needs to be corrected.
Solutions
(A) Show communities instead of showing all the data, offer a
subgraph-view that cuts out the nearest neighbors of one concept
to compensate for data loss in the community view.
(B) Filter by language families and weight the concept links by
frequency of occurrence, following Dellert’s (2014) suggestion.
This will cut most of the links resulting from homophony and leaves
the links which are due to polysemy.
10 / 33
Slide 32
Slide 32 text
CLICS 1.0 Interface
Interface
11 / 33
Slide 33
Slide 33 text
CLICS 1.0 Interface
Interface
Interface is written in JavaScript for the visualizations and PhP for
querying the data.
11 / 33
Slide 34
Slide 34 text
CLICS 1.0 Interface
Interface
Interface is written in JavaScript for the visualizations and PhP for
querying the data.
The interactive component of the network browser was specifically
designed for CLICS and builds on the D3 framework by Bostock et
al. (2011).
11 / 33
Slide 35
Slide 35 text
CLICS 1.0 Interface
Interface
Interface is written in JavaScript for the visualizations and PhP for
querying the data.
The interactive component of the network browser was specifically
designed for CLICS and builds on the D3 framework by Bostock et
al. (2011).
The underlying network with the inferred communities is offered for
download from the website, and the whole code which was used to
create the website is available for download at
http://github.com/clics/clics.
11 / 33
Slide 36
Slide 36 text
CLICS 1.0 Interface
DEMO
12 / 33
Slide 37
Slide 37 text
CLICS 2.0
CLICS 2.0
13 / 33
Slide 38
Slide 38 text
CLICS 2.0 Motivation
Motivation
14 / 33
Slide 39
Slide 39 text
CLICS 2.0 Motivation
Motivation
Problems in CLICS 1.0
difficult to curate (error-correction, linking data, adding data)
14 / 33
Slide 40
Slide 40 text
CLICS 2.0 Motivation
Motivation
Problems in CLICS 1.0
difficult to curate (error-correction, linking data, adding data)
difficult to collaborate (the CLICS team is separated and everybody
is extremely busy with stuff other than CLICS
14 / 33
Slide 41
Slide 41 text
CLICS 2.0 Motivation
Motivation
Problems in CLICS 1.0
difficult to curate (error-correction, linking data, adding data)
difficult to collaborate (the CLICS team is separated and everybody
is extremely busy with stuff other than CLICS
difficult to communicate (not all users understand how we arrived at
the data, and often think that it is us who messed datasets up, etc.,
although we only take the data to produce something new out of it)
14 / 33
Slide 42
Slide 42 text
CLICS 2.0 Motivation
Motivation
Problems in CLICS 1.0
difficult to curate (error-correction, linking data, adding data)
difficult to collaborate (the CLICS team is separated and everybody
is extremely busy with stuff other than CLICS
difficult to communicate (not all users understand how we arrived at
the data, and often think that it is us who messed datasets up, etc.,
although we only take the data to produce something new out of it)
difficult to expand (new datasets cannot be added without having a
true guiding principle)
14 / 33
Slide 43
Slide 43 text
CLICS 2.0 Motivation
Motivation
Problems in CLICS 1.0
difficult to curate (error-correction, linking data, adding data)
difficult to collaborate (the CLICS team is separated and everybody
is extremely busy with stuff other than CLICS
difficult to communicate (not all users understand how we arrived at
the data, and often think that it is us who messed datasets up, etc.,
although we only take the data to produce something new out of it)
difficult to expand (new datasets cannot be added without having a
true guiding principle)
difficult to catch up (we know much, much better now, how to curate
datasets, but we did not know this when preparing CLICS initially)
14 / 33
Slide 44
Slide 44 text
CLICS 2.0 Ideas
Ideas
15 / 33
Slide 45
Slide 45 text
CLICS 2.0 Ideas
Ideas
use the state of the art of available data
15 / 33
Slide 46
Slide 46 text
CLICS 2.0 Ideas
Ideas
use the state of the art of available data
separate data from display (CLICS 2.0 does not host data, but
simply uses it)
15 / 33
Slide 47
Slide 47 text
CLICS 2.0 Ideas
Ideas
use the state of the art of available data
separate data from display (CLICS 2.0 does not host data, but
simply uses it)
assemble data with help of the Concepticon (List, Forkel, and
Cysouw 2016)
15 / 33
Slide 48
Slide 48 text
CLICS 2.0 Ideas
Ideas
use the state of the art of available data
separate data from display (CLICS 2.0 does not host data, but
simply uses it)
assemble data with help of the Concepticon (List, Forkel, and
Cysouw 2016)
assemble information on languages exclusively from Glottolog
(Hammarström et al. 2017)
15 / 33
Slide 49
Slide 49 text
CLICS 2.0 Ideas
Ideas
use the state of the art of available data
separate data from display (CLICS 2.0 does not host data, but
simply uses it)
assemble data with help of the Concepticon (List, Forkel, and
Cysouw 2016)
assemble information on languages exclusively from Glottolog
(Hammarström et al. 2017)
curate the code and the polysemy data with help of a transparent
API
15 / 33
Slide 50
Slide 50 text
CLICS 2.0 Ideas
Ideas
use the state of the art of available data
separate data from display (CLICS 2.0 does not host data, but
simply uses it)
assemble data with help of the Concepticon (List, Forkel, and
Cysouw 2016)
assemble information on languages exclusively from Glottolog
(Hammarström et al. 2017)
curate the code and the polysemy data with help of a transparent
API
regularly release the data in release circles of about 1 per year
(following the practice of Glottolog and other CLLD projects)
15 / 33
Slide 51
Slide 51 text
CLICS 2.0 Ideas
Ideas
use the state of the art of available data
separate data from display (CLICS 2.0 does not host data, but
simply uses it)
assemble data with help of the Concepticon (List, Forkel, and
Cysouw 2016)
assemble information on languages exclusively from Glottolog
(Hammarström et al. 2017)
curate the code and the polysemy data with help of a transparent
API
regularly release the data in release circles of about 1 per year
(following the practice of Glottolog and other CLLD projects)
normalize the data which is analysed by CLICS
15 / 33
Slide 52
Slide 52 text
CLICS 2.0 Excursus
Excursus: Concepticon
Concept List # Items Concept Label Concept ID
Allen (2007) 500 animal oil; 动物油(脂肪) GREASE (CONCEPTICON-ID: 3232)
Gregersen (1976) 217 fat-grease*fat-grease GREASE (CONCEPTICON-ID: 3232)
Heggarty (2005) 150 fat (grease); grasa GREASE (CONCEPTICON-ID: 3232)
Swadesh (1955) 100 fat (grease) GREASE (CONCEPTICON-ID: 3232)
Alpher and Nash (1999) 151 fat, grease GREASE (CONCEPTICON-ID: 3232)
Hale (1961) 100 fat, grease GREASE (CONCEPTICON-ID: 3232)
OGrady and Klokeid (1969) 100 fat, grease GREASE (CONCEPTICON-ID: 3232)
Blust (2008) 210 fat/grease GREASE (CONCEPTICON-ID: 3232)
Matisoff (1978) 200 fat/grease GREASE (CONCEPTICON-ID: 3232)
Samarin (1969) 218 fat/grease GREASE (CONCEPTICON-ID: 3232)
Dunn et al. (2012) 207 fat GREASE (CONCEPTICON-ID: 3232)
Swadesh (1950) 215 fat GREASE (CONCEPTICON-ID: 3232)
Zgraggen (1980) 380 fat GREASE (CONCEPTICON-ID: 3232)
Jachontov (1991) 100 fat n. GREASE (CONCEPTICON-ID: 3232)
Wiktionary (2003) 207 fat (noun) GREASE (CONCEPTICON-ID: 3232)
Starostin (1991) 110 fat n.; жир GREASE (CONCEPTICON-ID: 3232)
TeilDautrey et al. (2008) 430 fat, oil GREASE (CONCEPTICON-ID: 3232)
Swadesh (1952) 200 fat (organic substance) GREASE (CONCEPTICON-ID: 3232)
Shiro (1973) 200 grease (fat) GREASE (CONCEPTICON-ID: 3232)
Samarin (1969) 100 grease; graisse; Fett; grasa GREASE (CONCEPTICON-ID: 3232)
Wang (2006) 200 pig oil; 猪油 GREASE (CONCEPTICON-ID: 3232)
Haspelmath and Tadmor (2009) 1460 the grease or fat GREASE (CONCEPTICON-ID: 3232)
16 / 33
Slide 53
Slide 53 text
CLICS 2.0 Excursus
Excursus: Concepticon
Concept List # Items Concept Label Concept ID
Allen (2007) 500 animal oil; 动物油(脂肪) GREASE (CONCEPTICON-ID: 3232)
Gregersen (1976) 217 fat-grease*fat-grease GREASE (CONCEPTICON-ID: 3232)
Heggarty (2005) 150 fat (grease); grasa GREASE (CONCEPTICON-ID: 3232)
Swadesh (1955) 100 fat (grease) GREASE (CONCEPTICON-ID: 3232)
Alpher and Nash (1999) 151 fat, grease GREASE (CONCEPTICON-ID: 3232)
Hale (1961) 100 fat, grease GREASE (CONCEPTICON-ID: 3232)
OGrady and Klokeid (1969) 100 fat, grease GREASE (CONCEPTICON-ID: 3232)
Blust (2008) 210 fat/grease GREASE (CONCEPTICON-ID: 3232)
Matisoff (1978) 200 fat/grease GREASE (CONCEPTICON-ID: 3232)
Samarin (1969) 218 fat/grease GREASE (CONCEPTICON-ID: 3232)
Dunn et al. (2012) 207 fat GREASE (CONCEPTICON-ID: 3232)
Swadesh (1950) 215 fat GREASE (CONCEPTICON-ID: 3232)
Zgraggen (1980) 380 fat GREASE (CONCEPTICON-ID: 3232)
Jachontov (1991) 100 fat n. GREASE (CONCEPTICON-ID: 3232)
Wiktionary (2003) 207 fat (noun) GREASE (CONCEPTICON-ID: 3232)
Starostin (1991) 110 fat n.; жир GREASE (CONCEPTICON-ID: 3232)
TeilDautrey et al. (2008) 430 fat, oil GREASE (CONCEPTICON-ID: 3232)
Swadesh (1952) 200 fat (organic substance) GREASE (CONCEPTICON-ID: 3232)
Shiro (1973) 200 grease (fat) GREASE (CONCEPTICON-ID: 3232)
Samarin (1969) 100 grease; graisse; Fett; grasa GREASE (CONCEPTICON-ID: 3232)
Wang (2006) 200 pig oil; 猪油 GREASE (CONCEPTICON-ID: 3232)
Haspelmath and Tadmor (2009) 1460 the grease or fat GREASE (CONCEPTICON-ID: 3232)
16 / 33
Slide 54
Slide 54 text
CLICS 2.0 Excursus
Excursus: Concepticon
Concept List # Items Concept Label Concept ID
Allen (2007) 500 animal oil; 动物油(脂肪) GREASE (CONCEPTICON-ID:323)
Gregersen (1976) 217 fat-grease*fat-grease GREASE (CONCEPTICON-ID:323)
Heggarty (2005) 150 fat (grease); grasa GREASE (CONCEPTICON-ID:323)
Swadesh (1955) 100 fat (grease) GREASE (CONCEPTICON-ID:323)
Alpher and Nash (1999) 151 fat, grease GREASE (CONCEPTICON-ID:323)
Hale (1961) 100 fat, grease GREASE (CONCEPTICON-ID:323)
OGrady and Klokeid (1969) 100 fat, grease GREASE (CONCEPTICON-ID:323)
Blust (2008) 210 fat/grease GREASE (CONCEPTICON-ID:323)
Matisoff (1978) 200 fat/grease GREASE (CONCEPTICON-ID:323)
Samarin (1969) 218 fat/grease GREASE (CONCEPTICON-ID:323)
Dunn et al. (2012) 207 fat GREASE (CONCEPTICON-ID:323)
Swadesh (1950) 215 fat GREASE (CONCEPTICON-ID:323)
Zgraggen (1980) 380 fat GREASE (CONCEPTICON-ID:323)
Jachontov (1991) 100 fat n. GREASE (CONCEPTICON-ID:323)
Wiktionary (2003) 207 fat (noun) GREASE (CONCEPTICON-ID:323)
Starostin (1991) 110 fat n.; жир GREASE (CONCEPTICON-ID:323)
TeilDautrey et al. (2008) 430 fat, oil GREASE (CONCEPTICON-ID:323)
Swadesh (1952) 200 fat (organic substance) GREASE (CONCEPTICON-ID:323)
Shiro (1973) 200 grease (fat) GREASE (CONCEPTICON-ID:323)
Samarin (1969) 100 grease; graisse; Fett; grasa GREASE (CONCEPTICON-ID:323)
Wang (2006) 200 pig oil; 猪油 GREASE (CONCEPTICON-ID:323)
Haspelmath and Tadmor (2009) 1460 the grease or fat GREASE (CONCEPTICON-ID:323)
16 / 33
Slide 55
Slide 55 text
CLICS 2.0 Excursus
Excursus: Concepticon
Concept List # Items Concept Label Concept ID
Allen (2007) 500 animal oil; 动物油(脂肪) GREASE (CONCEPTICON-ID:323)
Gregersen (1976) 217 fat-grease*fat-grease GREASE (CONCEPTICON-ID:323)
Heggarty (2005) 150 fat (grease); grasa GREASE (CONCEPTICON-ID:323)
Swadesh (1955) 100 fat (grease) GREASE (CONCEPTICON-ID:323)
Alpher and Nash (1999) 151 fat, grease GREASE (CONCEPTICON-ID:323)
Hale (1961) 100 fat, grease GREASE (CONCEPTICON-ID:323)
OGrady and Klokeid (1969) 100 fat, grease GREASE (CONCEPTICON-ID:323)
Blust (2008) 210 fat/grease GREASE (CONCEPTICON-ID:323)
Matisoff (1978) 200 fat/grease GREASE (CONCEPTICON-ID:323)
Samarin (1969) 218 fat/grease GREASE (CONCEPTICON-ID:323)
Dunn et al. (2012) 207 fat GREASE (CONCEPTICON-ID:323)
Swadesh (1950) 215 fat GREASE (CONCEPTICON-ID:323)
Zgraggen (1980) 380 fat GREASE (CONCEPTICON-ID:323)
Jachontov (1991) 100 fat n. GREASE (CONCEPTICON-ID:323)
Wiktionary (2003) 207 fat (noun) GREASE (CONCEPTICON-ID:323)
Starostin (1991) 110 fat n.; жир GREASE (CONCEPTICON-ID:323)
TeilDautrey et al. (2008) 430 fat, oil GREASE (CONCEPTICON-ID:323)
Swadesh (1952) 200 fat (organic substance) GREASE (CONCEPTICON-ID:323)
Shiro (1973) 200 grease (fat) GREASE (CONCEPTICON-ID:323)
Samarin (1969) 100 grease; graisse; Fett; grasa GREASE (CONCEPTICON-ID:323)
Wang (2006) 200 pig oil; 猪油 GREASE (CONCEPTICON-ID:323)
Haspelmath and Tadmor (2009) 1460 the grease or fat GREASE (CONCEPTICON-ID:323)
16 / 33
Slide 56
Slide 56 text
CLICS 2.0 Excursus
Excursus: Concepticon
Concepticon (List et al. 2016)
link concept labels in published concept lists
(questionnaires) to concept sets
link concept sets to meta-data
define relations between concept sets
never link one concept in a given list to more than one
concept set (guarantees consistency)
provide an API to check the consistency of the data
and to query the data
provide a web-interface to browse through the data
17 / 33
Slide 57
Slide 57 text
CLICS 2.0 Excursus
Concepticon
STONE
EGG
FOOT
THE STONE
THE EGG
THE LEG
STONE
(FRUIT)
EGG
(CHICKEN)
FOOT/LEG
STONE
EGG
LEG
FOOT
http://concepticon.clld.org
18 / 33
Slide 58
Slide 58 text
CLICS 2.0 Excursus
Concepticon
CONCEPT
SET
CONCEPT
CONCEPT
LIST
CONCEPT
LABEL
COMPILER
SOURCE
NOTE
CONCEPT
LABEL
CONCEPT
LABEL
CONCEPT
LABEL
CONCEPT
SET
CONCEPT
SET
18 / 33
Slide 59
Slide 59 text
CLICS 2.0 Excursus
http://concepticon.clld.org
19 / 33
Slide 60
Slide 60 text
CLICS 2.0 Excursus
Excursus: Data
DATASET EDITORS LANGUAGES CONCEPTS
IDS Key and Comrie (2016) 367 1310
WOLD Haspelmath and Tadmor (2008) 41 1430
BaiDial* Allen (2007) 8 500
HuberReed Huber and Reed (1992) 71 374
Kraft1981 Kraft (1981) 68 434
BantuBVD* Teil-Dautrey (2008) 10 430
Tryon1983* Tryon (1983) 111 324
Madang* Zgraggen (1980) 100 380
Cihui* Beijing Daxue (1964) 17 905
TBL* Huang (1992) 50 1800
NorthEuraLex Dellert and Jäger (2017) 106 1000
Datasets with an asterisk are currently in preparation and will be most likely
released already within this year.
20 / 33
Slide 61
Slide 61 text
CLICS 2.0 Excursus
Excursus: Data
21 / 33
Slide 62
Slide 62 text
CLICS 2.0 Excursus
Excursus: Data
By linking these datasets to the Concepticon (which we have
already done with most of them), we can easily combine the data
into a bigger dataset that we use as our basic data for CLICS 2.0.
21 / 33
Slide 63
Slide 63 text
CLICS 2.0 Excursus
Excursus: Data
By linking these datasets to the Concepticon (which we have
already done with most of them), we can easily combine the data
into a bigger dataset that we use as our basic data for CLICS 2.0.
Given problems with concept overlap in the datasets, we can make
different selections for the users, including datasets with many
concepts but not so many languages and datasets with many
languages but less concepts.
21 / 33
Slide 64
Slide 64 text
CLICS 2.0 Excursus
Excursus: Data
Subset Datasets Concepts Languages
High-Low >= 2 >= 1000 >= 300
Mid-Mid >= 5 >= 500 >= 600
Low-High >= 10 >= 250 >= 1000
22 / 33
Slide 65
Slide 65 text
CLICS 2.0 Excursus
Excursus: Data
Subset Datasets Concepts Languages
High-Low >= 2 >= 1000 >= 300
Mid-Mid >= 5 >= 500 >= 600
Low-High >= 10 >= 250 >= 1000
.
.
Effectively this means, that with CLICS 2.0, we can immediately offer
different views on the data, which allow scholars to investigate very
broad patterns of semantic associations, as well as fine-grained
patterns with a lower attestation.
22 / 33
Slide 66
Slide 66 text
CLICS 2.0 Excursus
Excursus: Software API
23 / 33
Slide 67
Slide 67 text
CLICS 2.0 Excursus
Excursus: Software API
With the Python API that we are currently preparing for CLICS 2.0,
users will be able to use their own data to run their own network
analyses, since all data is shipped with CLICS, users can also use
the data we selected for CLICS 2.0.
23 / 33
Slide 68
Slide 68 text
CLICS 2.0 Excursus
Excursus: Software API
With the Python API that we are currently preparing for CLICS 2.0,
users will be able to use their own data to run their own network
analyses, since all data is shipped with CLICS, users can also use
the data we selected for CLICS 2.0.
We will try to offer cookbooks accompanying the software API, to
help users to use it efficiently.
23 / 33
Slide 69
Slide 69 text
CLICS 2.0 Excursus
Excursus: Software API
With the Python API that we are currently preparing for CLICS 2.0,
users will be able to use their own data to run their own network
analyses, since all data is shipped with CLICS, users can also use
the data we selected for CLICS 2.0.
We will try to offer cookbooks accompanying the software API, to
help users to use it efficiently.
By shifting to the CLLD framework, scholars can also create their
own CLICS websites, since the source code for the creation of
interactive networks will be transparently shipped with the data.
23 / 33
Slide 70
Slide 70 text
CLICS 2.0 Excursus
Excursus: Software API
With the Python API that we are currently preparing for CLICS 2.0,
users will be able to use their own data to run their own network
analyses, since all data is shipped with CLICS, users can also use
the data we selected for CLICS 2.0.
We will try to offer cookbooks accompanying the software API, to
help users to use it efficiently.
By shifting to the CLLD framework, scholars can also create their
own CLICS websites, since the source code for the creation of
interactive networks will be transparently shipped with the data.
Spring schools and further events carried out at the MPI-SHH as
part of my ERC project on Computer-Assisted Language
Comparison will cover – among others – introductory tutorials to all
the software APIs that are shipped with the different tools and
datasets developed at our department.
23 / 33
Slide 71
Slide 71 text
CLICS 2.0 Features
Features
24 / 33
Slide 72
Slide 72 text
CLICS 2.0 Features
Features
drastic increase in data
24 / 33
Slide 73
Slide 73 text
CLICS 2.0 Features
Features
drastic increase in data
drastic increase in transparency
24 / 33
Slide 74
Slide 74 text
CLICS 2.0 Features
Features
drastic increase in data
drastic increase in transparency
drastic increase in replicability
24 / 33
Slide 75
Slide 75 text
CLICS 2.0 Features
Features
drastic increase in data
drastic increase in transparency
drastic increase in replicability
regular floating releases which feature new data
24 / 33
Slide 76
Slide 76 text
CLICS 2.0 Features
Features
drastic increase in data
drastic increase in transparency
drastic increase in replicability
regular floating releases which feature new data
strict and clear-cut collaboration guidelines
24 / 33
Slide 77
Slide 77 text
CLICS 2.0 Features
Features
drastic increase in data
drastic increase in transparency
drastic increase in replicability
regular floating releases which feature new data
strict and clear-cut collaboration guidelines
new methods (see demo on next slide)
24 / 33
Slide 78
Slide 78 text
CLICS 2.0 Features
Features
drastic increase in data
drastic increase in transparency
drastic increase in replicability
regular floating releases which feature new data
strict and clear-cut collaboration guidelines
new methods (see demo on next slide)
rigid policy towards open data (since we heavily profit from all of
our colleagues who publish their data!)
24 / 33
Slide 79
Slide 79 text
CLICS 2.0 Features
Features: Coverage
25 / 33
Slide 80
Slide 80 text
CLICS 2.0 Features
New Methods
26 / 33
Slide 81
Slide 81 text
CLICS 2.0 Features
New Methods
Following Urban (2011) we are currently testing an automatized
variant of partial colexifications which can help us to direct our
networks and shed light on compositional aspect of semantic
associations.
26 / 33
Slide 82
Slide 82 text
CLICS 2.0 Features
New Methods
Following Urban (2011) we are currently testing an automatized
variant of partial colexifications which can help us to direct our
networks and shed light on compositional aspect of semantic
associations.
By improving our insights into graph theory and available
algorithms, we can now enhance the analysis of the networks.
Articulation points, for example, show key players in a network
which connect between different communities.
26 / 33
Slide 83
Slide 83 text
CLICS 2.0 Features
New Methods
WASP
BEEHIVE
WINE
ALCOHOL (FERMENTED DRINK)
BEER
DRINK
MEAD
BEVERAGE
HONEY
BEESWAX
SUGAR
FRAGRANT
STINKING
BEE
SWEET
SMELL (STINK)
FEEL
SUGAR CANE
SNIFF
SMELL (PERCEIVE)
27 / 33
Slide 84
Slide 84 text
CLICS 2.0 Features
New Methods
CORNER SHORE
COAST
FRINGE
LAST (FINAL)
END (OF TIME)
FOR A LONG TIME
FAR
LENGTH
DEEP
LONG
BOUNDARY
SIDE
BESIDE
END (OF SPACE)
HIGH
UP
TOP
HEAVEN
TALL
ABOVE
SKY
NEAR
EDGE
BORDER
28 / 33
Slide 85
Slide 85 text
CLICS 2.0 Features
CLICS 2.0 DEMO
29 / 33
Slide 86
Slide 86 text
CLICS 2.0 Schedule
Schedule
30 / 33
Slide 87
Slide 87 text
CLICS 2.0 Schedule
Schedule
We are working hard on assembling more data and building up the
new API as well as the web-interface, but we are currently not
many who work on CLICS or in its periphery.
30 / 33
Slide 88
Slide 88 text
CLICS 2.0 Schedule
Schedule
We are working hard on assembling more data and building up the
new API as well as the web-interface, but we are currently not
many who work on CLICS or in its periphery.
We hope that we can publish CLICS 2.0 very late this year, and in
a worst case, in early 2018.
30 / 33
Slide 89
Slide 89 text
CLICS 2.0 Schedule
Schedule
We are working hard on assembling more data and building up the
new API as well as the web-interface, but we are currently not
many who work on CLICS or in its periphery.
We hope that we can publish CLICS 2.0 very late this year, and in
a worst case, in early 2018.
But we would argue that it is better to publish the next version a bit
later rather than publishing a version that we will need to update
immediately after we first published it.
30 / 33
Slide 90
Slide 90 text
CLICS 2.0 Schedule
Schedule
We are working hard on assembling more data and building up the
new API as well as the web-interface, but we are currently not
many who work on CLICS or in its periphery.
We hope that we can publish CLICS 2.0 very late this year, and in
a worst case, in early 2018.
But we would argue that it is better to publish the next version a bit
later rather than publishing a version that we will need to update
immediately after we first published it.
If we can learn one thing from CLICS 1.0, it is that we need to keep
the code and the data in a state that we can easily curate them.
We hope we will achieve this with CLICS 2.0.
30 / 33
Slide 91
Slide 91 text
Outlook
Outlook
31 / 33
Slide 92
Slide 92 text
It is still a rather long way from CLICS 1.0 to CLICS 2.0.
32 / 33
Slide 93
Slide 93 text
It is still a rather long way from CLICS 1.0 to CLICS 2.0.
But we hope that we are on the right track by now, and
that won’t disappoint those who came to like the
Cross-Linguistic Colexification Database.
32 / 33
Slide 94
Slide 94 text
It is still a rather long way from CLICS 1.0 to CLICS 2.0.
But we hope that we are on the right track by now, and
that won’t disappoint those who came to like the
Cross-Linguistic Colexification Database.
CLICS 2.0 won’t be perfect, but it will be entertaining
and hopefully very interesting for our colleagues
working on historical linguistics and lexical typology.
32 / 33
Slide 95
Slide 95 text
Thanks for your attention!
33 / 33