A study of relations between associative structure and morphological structure of Hungarian words

(1)

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 introduced by Deese [1], a multidimensional scaling method was used to obtain associative fields de- picting the pairwise associative distance of stimulus words in a two-dimensional figure. The results 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

(2)

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 overlaps (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 overlaps (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 morphological and associative structure of words. As shown in figure (5), there is an interesting difference of the relations between associative entropy (defined as H_A = −^Pip_ilog₂p_i, where p_i is the relative frequency of the ith associated word) and corpus morphological entropy (defined asH_M =

−^Piqilog₂qi, whereqi is the relative frequency of form i in the MOKK corpus) between nouns and verbs. In nouns, the more varied the morphology 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 frequency: 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

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 morphological 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 ææ

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

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] Péter Halácsy, András Kornai, László Németh, András Rung, István Szakadát, and Viktor Trón. Creating open language resources for hungarian. InLREC, 2004.

[3] Zsolt Lengyel. Magyar asszociációs normák enciklopédiája I-II. TINTA könyvkiadó, Bu- dapest, 2008-2010.

118

(3)

[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