C ea ing and Consul ing Local Sel -
Compiled Co po a o Teaching Fo eign
Languages wi h Tools A ailable in
CLARIN Filip Zeman
Au onomous Uni e si y o Mad id
[email protected]
Thesis supe iso : Olga Ba iuko a
CLARIAH–ES
Acknowledgemen s
This pos e is pa o he p ojec Co pus o composi ionali y and lexical
in o ma i eness: anno a ion, analysis and applica ions (INFOLEXIS), (PI
Olga Ba iuko a, e e ence PID2022-138135NB-100), inanced by
MCIN/AEI/10.13039/501100011033 and by he ESF+.
The pos e is pa o he g an PREP2022-000882, inanced by
MCIN/AEI/10.13039/501100011033 and ESF+.
This pos e was unded by Comunidad de Mad id unde he ag eemen
“The node CLARIAH-CM: Digi al Humani ies and Language Technologies”
(4180134).
P ojec o e iew
Sel -compiled co po a: guidelines and illus a ions
•PhD hesis cu en ly unde way: The Use o Sel -Compiled
Co po a o Teaching Spanish as a Fo eign Language: Selec ion
Requi emen s and Lexical Selec ion.
•Academic use o co po a: su ge o p oposals in 1990s, low
implemen a ion due o echnical cons ain s. Re i ed a en ion
since he 2010s.
•Aim: De elop a echnique o eaching Spanish ocabula y and
lexical selec ion equi emen s ha builds on he o iginal da a
d i en lea ning p oposals and makes use o sel -compiled
co po a as didac ic ools.
Guidelines o sel -compiled co po a use
Minimum size: 50,000 wo ds (100,000 wo ds
p e e ed o mo e accu a e Wo d Ske ch
esul s; compila ion ime ± 1 hou ).
Me hod o Compila ion: Di ec sea ch (mo e
p ecise, abili y o cleanse da a noise) s.
c awling ( as e , mo e noise in da a).
Compila ion me hodology: S uden
pa icipa ion encou aged, eache o e sigh
necessa y.
Minimum CFER le el: B1-B2 (some linguis ic
knowledge equi ed).
Sea ch (Sk. En. speci ic): “BASIC” p e e ed.
Ske ch Engine
+ mode n, la ge a ie y o ools
+ uses LogDice o e alua e lexical cohesion (unlike MI sco e does
no o e alua e a e combina ions, is co pus-size-independen )
– Paid so wa e, ee e sion NoSke chEngine (also a ailable
h ough CLARIN) o e s limi ed unc ionali y
Lancs Box
+ ee, ela i ely easy o use, au oma ic, a ie y o ools
(conco dances, N-g ams, colloca ion sea ch)
– O e s limi ed unc ionali y compa ed o Ske ch Engine, uses MI
o colloca ion s abili y indica ion
Tei ok
+ F ee
– Complex o use, XML/TEI knowledge needed o un
CLARIN ools o co pus compila ion and use Exempla y ac i i ies wi h Ske ch Engine ools
Conco dance
Tool: Shows keywo d in con ex o sea ches in a co pus.
Ac i i ies: lookup, e i y exis ence o wo ds, mul iwo d uni s.
Wo d Ske ch
Tool: Summa izes wo d’s syn ac ic and colloca ional beha iou .
Ac i i ies: c ea e syn ac ic mind maps, di e en ia e ixed s. non-
ixed combina ions, g amma ical uses
Wo d Ske ch Di e ence
Tool: Compa es Wo d Ske ch esul s o wo wo ds.
Ac i i ies: disce n cuasisynonyms based on colloca ional pa e ns
N-G ams
Tool:Lis s equen wo d sequences o chosen leng h.
Ac i i ies: 2+ wo d o mulaic sequence ex ac ion (combina ions
no de ec able wi h Wo d Ske ch), seman ic mind map c ea ion.
Fig. 1. Wo d Ske ch Di e ence esul s o p oblema (p oblem)
and as o no (diso de ),
Fig. 2. Wo d Ske ch esul s o salud
(heal h).
F om CLARIN
•Co pus que y ools
a ailable in CLARIN we e
used o design he
eaching echnique and
will be used as lea ning
ools o he expe imen al
pa o he hesis.
Fo CLARIN
•Guidelines on co pus
compila ion and didac ic
use wi h CLARIN-a ailable
so wa e, incl. exempla y
ac i i ies (LRC)
•Da a on he me hod’s
e ec i eness
•Upload o eady-made
co po a o CLARIN ( u u e)