A study of relations between associative structure and morphological structure of Hungarian words
D´aniel Cz´egel
Budapest University of Technology and Economics
czegel d@yahoo.com
Zsolt Lengyel
Pannon University, Veszpr´em
Csaba Pl´eh
Central European University, Budapest vispleh@ceu.hu
The paper mainly aims to reanalyze data with the presently available corpus linguistics tools from a relatively large scale paper-and-pencil based Hungarian verbal association dictionary with regard to two aspects. i)The mental lexicon issue. How are associative overlaps representing structural relations in the mental lexicon? ii)The systemic variability of the associative fields mo- bilized by the stimulus words: how variable the responses are, and how these associative entropies are related to morphological entropies of the same words.
1 Methods and materials
For the associative corpora, two dictionaries of Lengyel [3] were used. They are based on the responses of2000students between 10−14and 18−24 to about 200stimulus words. Digitized responses from this dictionary were related to the frequency distribution of 800 million web-based Hungarian words from the MOKK corpus [2].
2 Results
2.1 Associative overlaps and lexical fields Based on the associative overlap measure intro- duced by Deese [1], a multidimensional scaling method was used to obtain associative fields de- picting the pairwise associative distance of stim- ulus words in a two-dimensional figure. The re- sults indicate that young adults have a more dense structure, their associative clusters are more tight compared to those of children of age 10-14, as il- lustrated in figure (1) and (2) and shown quantita- tively in figure (3) and (4).
window
give
bed
shape to stand
dream
to sleep
mother father
woman
table baby
doll
problem friend
enter
impression speak
illness do
group work
matter dear
expensive
sweet
health simple
hungry life go
away
to live
man
to remember
to eat
interesting
forest
result
power strong
to understand
year village
white head
milk
black answer
light male
man
young
boy river
earth run
think
child
stomach
quick
fruit war to
hear
to besilent
to listen to sound
voice
loud
wrath
house mountain place
cold
moon long
time
weather truth
promise
to write
school acquaintance
known
familiar
drink mark
ticket good law
come hammer
stove soldier
dear
kind
blue have to
must
need
bread hard
picture to ask
earnlook forsearch
bitter
hand little
wish book send
footleg
soft
lamp daughter
slow
see
girl sit down
high
tall
hungarian excuse
catch sight of
go
deep say
cinema job
big
great grandmother
day
sun square
difficult
heavy
people name to
look at
tranquil
ocean page
side clock
hour
watch
lesson
russian
lion
country
doctor
physician old
joy
paper money
red dot
point
score
morning order part
bad
short
yellow corner
sour help
walk
smooth
salt
dark free
allowed
outdoors
stem chair
eye beautiful
love
like
colour word
room
saturday thirsty
carpet find
teach
learn hold
keep
full space
square
turn
production do
put
clean thief
law can
know
new newspaper
road
way
street sit
holiday
butter
town purchase
guest
world
flower
water music
green
ADJ ADJMODN ADJN ADVN INF MOD MODV N NV V
Figure 1: Associative field of children (age 10-14)
window
give
bed figure
to stand
dream
to sleep
mother father
woman
table baby doll
trouble friend
enter
impression speak
illness do
group
work matter
dear
sweet health
simple
hungry life
go away to live
man
to remember
to eat
interesting
wood result
power strong to
understand
year
village
white head
milk
black answer
light man young boy
river
earth run
think
child
stomach quick
fruit
war to hear
to listen to to be silent
voice sound
loud wrath
house mountain
place
cold moon long
time weather
truth promise
to
write school
acquaintance
drink
ticket good
law come
hammer
stove
soldier dear
blue to be
wanting
hard bread
picture
to ask
look for
bitter hand
little
wish book send
foot
soft lamp daughter
slow see girl
sit down
high hungarian
excuse catch sight of
go
deep say
cinema
work
big
grandmother sun day square
heavy
people
name
to look at
thanquil
ocean side
clock hour lesson
russian
lion country
doctor
old joy
paper
money
red full
stop
morning order
part
bad short
yellow
corner
sour help
walk smooth
salt dark free allowed outdoors
stem chair
eye beautiful
like
colour word
room
saturday
thirsty carpet find
teach learn
hold
full square
space production
do
clean thief
law
know new
news newspaper
way street
sit
feast
butter city
buy guest
world
flower
water music
green
ADJ ADJ MOD N ADJ N Adv N INF MOD N N V V
Figure 2: Associative field of young adults (age 18-24)
Copyrightc by the paper’s authors. Copying permitted for private and academic purposes.
In Vito Pirrelli, Claudia Marzi, Marcello Ferro (eds.):Word Structure and Word Usage. Proceedings of the NetWordS Final Conference, Pisa, March 30-April 1, 2015, published athttp://ceur-ws.org
117
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 10
100 1000
associative overlap
frequency
Figure 3: Histogram of pairwise associative over- laps (age 10-14)
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 10
100 1000 10 000
associative overlap
frequency
Figure 4: Histogram of pairwise associative over- laps (age 18-24)
2.2 Relationship between associative entropy, morphological entropy and frequency We followed the methods introduced by Osgood [5] for analyzing the variability of morphologi- cal and associative structure of words. As shown in figure (5), there is an interesting difference of the relations between associative entropy (defined as HA = −Pipilog2pi, where pi is the rel- ative frequency of the ith associated word) and corpus morphological entropy (defined asHM =
−Piqilog2qi, whereqi is the relative frequency of form i in the MOKK corpus) between nouns and verbs. In nouns, the more varied the morphol- ogy of the noun is in the corpus, the more variable the associative field is (r = 0.202, 0.175in the two ages). That can be interpreted as implying that the more varied the suffixation of a noun is, the more variable associative relations it enters with other words. In verbs, however, if the verb has a more varied morphology, it has less associations (r=−0.194, both groups). As figure (6) shows, a similar relationship has been obtained between as-
sociative entropy and the logarithm of corpus fre- quency: In the case of nouns the correlation is pos- itive (r = 0.134and0.367in the two ages) while the correlation is negative (r =−0.281,−0.222) for verbs. This peculiar relation would be further studied with considering morphological entropy in light of the argument frames of the verbs on the one hand, and the role of syntagmatic associations in the associative fields of verbs on the other [4].
æ æ
æ
æ æ æ
æ
æ æ æ
æ æ æ
æ
æ æ
æ æ æ æ
æ æ æ
æ
æ æ æ
æ æ æ
æ
æ æ
æ æ
æ
æ
æ æ
æ æ æ
æ æ æ
æ æ
æ
æ æ
æ æ
æ
æ æ
æ
æ æ
æ æ æ æ
æ
æ æ
æ
æ æ
æ æ
æ
æ æ
æ
æ
æ æ
æ
æ
æ
æ æ
æ æ
æ æ æ
æ
æ
æ æ
æ
æ
æ æ
æ
æ
æ æ
æ æ æ æ
æ æ
æ æ
æ
æ æ
æ
æ æ
æ æ
æ
æ æ
æ æ
æ
æ æ
æ
æ æ
æ æ
æ æ æ
æ
æ
æ æ æ
æ æ
æ
æ
æ æ
æ æ
æ æ æ
æ
æ æ
æ æ
æ æ
æ æ
æ
æ æ
æ
æ
æ æ
æ æ æ
æ
æ
æ
æ æ
æ
æ æ æ
æ
æ æ
æ æ
æ
æ æ æ
æ
æ æ
æ
æ æ
æ
3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
0 1 2 3 4 5 6 7
associative entropy
morphologicalentropy
æ ADJ 8 ADJMODN æ ADJN 8 AdvN æ INF
8 MOD
æ N
8 NV
æ V
Figure 5: Relation between associative and mor- phological entropy (age 18-24). The blue regres- sion line corresponds to nouns, while the red line corresponds to verbs.
æ æ
æ
æ æ æ
æ æ
ææ
æ æ
æ æ
æ
æ æ
æ æ
æ æ
æ
æ
æ
æ
æ æ
æ æ æ
æ
æ
æ æ
æ
æ
æ æ æ
æ æ æ
æ æ
æ æ æ
æ æ
æ
æ æ ææ
æ æ
æ æ
æ æ
æ
æ æ
æ æ
æ æ
æ æ
æ
æ
æ æ
æ æ
æ æ
æ
æ
æ æ
æ
æ æ
æ æ æ
ææ
æ æ
æ
æ
æ æ
æ
æ
æ æ
æ æ
æ æ
æ æ
æ æ
æ
æ æ
æ æ
æ æ
æ
æ
æ æ
æ
æ æ
æ
æ æ
æ æ
æ æ
æ æ æ
æ æ
æ æ
æ æ æ
æ
æ æ
æ æ
æ æ
æ æ æ
æ
æ æ
æ æ
æ æ
æ
æ æ æ
æ
æ æ
æ
æ æ
æ æ
æ æ æ
æ
æ
æ æ
æ
æ æ
æ
æ æ
æ æ æ
æ æ
æ æ
æ
æ
æ
3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
1000 104 105 106
associative entropy
frequency
ADJ ADJMODN ADJN AdvN INF MOD N NV V
Figure 6: Relation between associative entropy and corpus frequency (age 18-24). The blue re- gression line corresponds to nouns, while the red line corresponds to verbs.
References
[1] James Deese. The structure of associations in language and thought. Johns Hopkins Univer- sity Press, 1966.
[2] P´eter Hal´acsy, Andr´as Kornai, L´aszl´o N´emeth, Andr´as Rung, Istv´an Szakad´at, and Viktor Tr´on. Creating open language resources for hungarian. InLREC, 2004.
[3] Zsolt Lengyel. Magyar asszoci´aci´os norm´ak enciklop´edi´aja I-II. TINTA k¨onyvkiad´o, Bu- dapest, 2008-2010.
118
[4] Katherine Nelson. The syntagmatic- paradigmatic shift revisited: a review of research and theory. Psychological Bulletin, 84(1):93, 1977.
[5] Charles Egerton Osgood. Cross-cultural uni- versals of affective meaning. University of Illi- nois Press, 1975.
119