Cross-li linguistic res esearch in into deriv ivational networks - - PowerPoint PPT Presentation

SLIDE 1

Cross-li linguistic res esearch in into deriv ivational networks

Lívia Körtvélyessy P.J.Šafárik University, Košice, Slovakia

SLIDE 2

Structure of the lecture

• brief introduction of the project
• brief introduction of the methodology
• linguistic aspects

SLIDE 3

Brief introduction of the project

• 2015
• L. Körtvélyessy, P. Štekauer, S. Valera, A. Bagasheva
• derivational networks
• 40 European languages
• 49 experts
• http://www.ugr.es/~svalera/Monika
• workshop
• volume: Mouton de Gruyter

SLIDE 4

Methodology

• selection of languages
• selection of words
• development of the networks
• tools for evaluation of the networks
• their typological comparison

SLIDE 5

Selection of languages

• European languages
• HSK. Word-Formation. An International Handbook of the Languages
• f Europe
• reliable sources (corpora, dictionaries)

SLIDE 6

Selection of languages

Indo-European (29) Slavic (9): Bulgarian, Croatian, Czech, Polish, Russian, Serbian, Slovak, Slovene, Ukrainian Germanic (8): Danish, Dutch, English, Frisian, German, Icelandic, Norwegian, Swedish Romance (7): Catalan, French, Galician, Italian, Portuguese, Romanian, Spanish Celtic (2): Irish, Welsh Baltic (2): Latvian, Lithuanian Greek Uralic (4) Estonian, Finnish, Hungarian, North Saami Altaic (2) Tatar, Turkish Nakh-Daghestanian (2) Chechen, Dargwa Kartvelian (1) Georgian Afro-Asiatic (1) Maltese Isolate (1) Basque

SLIDE 7

Selection of words

• 30 actual (existing) underived (simple) words in each language
• equivalent words in 40 languages
• Swadesh basic vocabulary list (200 words)
• 10 nouns, 10 verbs, 10 adjectives

Nouns bone eye tooth day dog louse fire stone water name Verbs cut dig pull throw give hold sew burn drink know Adjectives bad new black straight warm

• ld

long thin thick narrow

SLIDE 8

What are derivational networks?

• Dokulil, Horecký, Furdík – derivational nests

→ vod-ár → vod-ák → vodáč-ka → vodác-tvo → vodác-ky → vod-áreň → vodáren-stvo → vodáren-ský → po-vod-eň → povodň-ový → vod-ička → vodičk-ový → vod-ný → vodník → vodnícky → vod-natý → z-vodnat-ieť → zvodnat-enie → vod-ový → vodov-o → pod-vod-ný → za-vod-niť → zavodn-enie → zavodň-ovať → zavodňova-nie → zavodňova-cí → od-vod-niť → odvodn-enie → odvodň-ovať → odvodňova-nie

VODA ‘water’

derivational series/chain

• rders of derivation

→ vod-ár → vod-ák → vod-áreň → po-vod-eň → vod-ička → vod-ný → vod-natý → vod-ový → pod-vod-ný → za-vod-niť → od-vod-niť

P a r a d i g m a t i c c a p a c i t y

SLIDE 9

What are derivational networks?

ORDERS OF DERIVATION PARADIGMATIC CAPACITY network of derivatives

• derived from the same

word-formation base (simple underived word)

• each affix formally

representing specific semantic categories

SLIDE 10

Development of the networks

1st order, Bulgarian, PIE ‘to drink’

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák pijač piteen pivnica pijan pivo pitie pijavica dopie (si) izpie vpie (se) napie (se)

• pie (se)

zapie (se) razpie (se) propie (se) nadpie prepie

• tpie

popie pijne pie mu se

1st order, Bulgarian, PIE ‘to drink’

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák pijač piteen pivnica pijan pivo pitie pijavica dopie izpie vpie napie

• pie

zapie razpie propie nadpie prepie

• tpie

popie pijne pie mu

SLIDE 11

Development of the networks

Bulgarian, pie

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák pijač piteen pivnica pijan pivo pitie pijavica dopie (si) izpie vpie (se) napie (se)

• pie (se)

zapie (se) razpie (se) propie (se) nadpie prepie

• tpie

popie pijne pie mu se

Development of the networks

1st order , Bulgarian, PIE ‘to drink’

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák pijač piteen pivnica pijan pivo pitie pijavica dopie izpie vpie napie

• pie

zapie razpie propie nadpie prepie

• tpie

popie pijne pie mu

State Manner Agent Relational Diminutive Action Inceptive pijanstvo pijano pijanica pijanski pijaničâk pijanstva vpijanči se

SLIDE 12

1st order, Bulgarian, PIE ‘to drink’

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák pijač piteen pivnica pijan pivo pitie pijavica dopie izpie vpie napie

• pie

zapie razpie propie nadpie prepie

• tpie

popie pijne pie mu

State Manner Agent Relational Diminutive Action Inceptive pijanstvo pijano pijanica pijanski pijaničâk pijanstva vpijanči se

SLIDE 13

1st order, Bulgarian, PIE ‘to drink’

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák pijač piteen pivnica pijan pivo pitie pijavica dopie izpie vpie napie

• pie

zapie razpie propie nadpie prepie

• tpie

popie pijne pie mu

State Manner Agent Relational Diminutive Action Inceptive pijanstvo pijano pijanica pijanski pijaničâk pijanstva vpijanči se Saturative Diminutive napijanstva (se) popijanstva

SLIDE 14

Development of the networks

1st order, Bulgarian, pie

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák pijač piteen pivnica pijan pivo pitie pijavica dopie izpie vpie napie

• pie

zapie razpie propie nadpie prepie

• tpie

popie pijne pie mu

Number of derivatives (paradigmatic capacity): 22

SLIDE 15

Development of the networks

1st order, , pie

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák pijač piteen pivnica pijan pivo pitie pijavica dopie izpie vpie napie

• pie

zapie razpie propie nadpie prepie

• tpie

popie pijne pie mu

1st order, , drekka Quality Process drekkanlegur drekking drekkandi

SLIDE 16

Evaluation of the networks – maximum derivational network

1st order, Bulgarian, pie

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák pijač piteen pivnica pijan pivo pitie pijavica dopie izpie vpie napie

• pie

zapie razpie propie nadpie prepie

• tpie

popie pijne pie mu

1st order, Bulgarian, pie

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák 1 pijač piteen 1 pivnica pijan 1 pivo pitie

3

pijavica dopie izpie vpie napie

• pie

zapie razpie propie nadpie prepie

• tpie

popie pijne pie mu

1st order, Bulgarian, pie

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák 1 pijač1 piteen 1 pivnica 1 pijan 1 pivo pitie

3 1 1

pijavica dopie izpie vpie napie

• pie

zapie razpie propie nadpie prepie

• tpie

popie pijne pie mu

1st order, Bulgarian, pie

Quality Agent Location Entitiy Finitive Directional Saturative Inceptive Augmentat Singulative Diminutive Desiderative pivák 1 pijač1 piteen 1 pivnica 1 pijan 1 pivo pitie

3 1 1

pijavica dopie izpie vpie

3 2

napie

• pie

zapie razpie propie nadpie prepie

• tpie

popie pijne pie mu

SLIDE 17

Evaluation of networks

Quality Entity Location InceptiveResultat Diminut Agent Saturativ DirectionPrivative cut 2 1 1 1 1 3 1 4 dig 1 1 1 1 2 5 1 pull 1 2 1 1 2 throw 1 1 give 1 1 8 hold 1 2 1 1 3 sew 1 1 3 2 1 3 1 1 burn 2 2 2 2 drink 3 3 1 3 2 1 2 1 know 1 1 MAX 3 3 1 3 3 2 1 3 8 4 MAX 3+ 3+ 1+ 3+ 3+ 2+ 1+ 3+ 8+ 4+…..

51 Maximum Derivational Network – 1st order

SLIDE 18

Evaluation of networks

Order of derivation verbs Sum total of maximum derivatives in individual semantic categories 1st 51 2nd 61 3rd 22 4th 2 5th 1 Σ 137 Maximum Derivational Network

SLIDE 19

Evaluation of networks – calculation of the saturation value

• MDN – potential
• actualization of the potential – SATURATION VALUE (SV)
• structural richness
• evaluative morphology, word-formation
• proportion between the number of actual derivatives in a

particular derivational network and the maximum derivational network

• pie (‘drink’), 1st order of derivation: 22 derivatives; MDN=51

SV=

𝑬 𝑵𝑬𝑶 x100

SV = 22:51x100= 43.14%

• rder

1st 2nd 3rd 4th 5th TOTAL SV 43.14 45.9 18.18 39.42 SV = 54:137x100= 39.42%

SLIDE 20

Evaluation of networks – scope of data

• 1200 networks
• number of derivatives – maximum derivational network
• number of orders of derivation
• saturation value
• each word – each order of derivation + total SV
• each word class – each order of derivation + total SV
• semantic categories:
• order of derivation
• combinations
• blocking effect
• multiple occurrences

SLIDE 21

Number of derivatives

Order of derivation 1st 2nd 3rd 4th 5th TOTAL Number of derivatives 22 28 4 54 Bulgarian, pie ‘to drink’

• total number of words

derived from 10 verbs in Bulgarian: 328 (average 32.8) Hungarian, iszik ‘to drink’ Order of derivation 1st 2nd 3rd 4th 5th TOTAL Number of derivatives 11 9 20

• total number of words

derived from 10 verbs in Hungarian: 351 (average 35.1)

SLIDE 22

Number of derivatives

• the average number of derivatives in Bulgarian
• the average number of derivatives in Hungarian
• average number of derivatives by order of derivation for all languages

nouns verbs adjectives 16.5 32.8 19.5 nouns verbs adjectives 32.2 35.1 27.6 1st order 2nd order 3rd order 4th order 5th order Nouns 7.71 6.26 2.21 0.43 0.06 Verbs 10.82 12.84 6.13 1.46 0.18 Adjectives 8.17 6.86 3.56 0.86 0.09

SLIDE 23

Num umber of

• f de

deriv ivativ ives - con

• nclu

lusio ions

• There are considerable differences among languages in their

paradigmatic capacity.

• The most prolific base is verb.
• Verbs have clearly the highest MDN value in every order of

derivation.

SLIDE 24

Number of orders of derivation

Verbs cut 3 dig 3 pull 3 throw 2 give 5 hold 3 sew 3 burn 4 drink 3 know 3 average 3.2

Bulgarian

Verbs Adjectives Nouns cut 3narrow 3bone 3 dig 3old 5eye 4 pull 3straight 2tooth 3 throw 2new 3day 3 give 5long 4dog 2 hold 3warm 4louse 2 sew 3thick 3fire 2 burn 4bad 4stone 2 drink 3thin 4water 4 know 3black 1name 4 average 3.2average 3.3average 2.9

SLIDE 25

Number of orders of derivation

Verbs Bulgarian Hungarian cut 3 3 dig 3 3 pull 3 2 throw 2 3 give 5 5 hold 3 4 sew 3 3 burn 4 4 drink 3 2 know 3 4 average 3.2 3.3 Verbs Bulgarian Hungarian Adjectives Bulgarian Hungarian Nouns Bulgarian Hungarian cut 3 3 narrow 3 3 bone 3 4 dig 3 3

• ld

5 3 eye 4 5 pull 3 2 straight 2 3 tooth 3 3 throw 2 3 new 3 4 day 3 5 give 5 5 long 4 4 dog 2 3 hold 3 4 warm 4 4 louse 2 4 sew 3 3 thick 3 3 fire 2 4 burn 4 4 bad 4 2 stone 2 4 drink 3 2 thin 4 3 water 4 4 know 3 4 black 1 4 name 4 4 average 3.2 3.3 average 3.3 3.3 average 2.9 4

SLIDE 26

Number of orders of derivation

Order of derivation 1st 2nd 3rd 4th 5th nouns 40 38 34 20 9 verbs 40 40 36 24 13 adjectives 40 31 34 25 10

Number of languages per order of derivation

SLIDE 27

Number of orders of derivation - con

• nlu

lusions

• The maximum number of orders of derivation is 5.
• There are six languages reaching five orders of derivation in all three

word-classes, none of them belonging to the Romance or Germanic genus.

• The average number of affixation steps is very similar for verb-based

and adjective-based derivation. It is lower for nouns.

• The richness of derivational networks is sensitive to the order of

derivation.

SLIDE 28

Saturation value

• potential (MDN value) is actualized by the individual sample words
• actualization of the potential capacity of derivation
• individual words and compare the richness of derivation by orders of

derivation;

• compare the saturation values of all sample words of a specific word-

class

• compare nouns, verbs and adjectives as a whole
• to evaluate the data from the perspective of groups of languages
• bone

SLIDE 29

Language 1st 2nd 3rd 4th 5th Total Bulgarian 36.67 26.09 26.67

• 28

Croatian 22.22 23.68 6.9

• 19.54

Czech 32.26 24.56 30.3 16.67 50 27.91 Polish 33.33 35.71 12.5

• 30.3

Russian 29.41 38.89 23.08

• 30.61

Serbian 31.37 33.9 7.41

• 26.21

Slovak 36.17 48.89 40.74 40

• 41.6

Slovene 22.5 24.32 23.53 100

• 25

Ukrainian 16 23.81 38.89 33.33

• 25.35

Danish 33.33 100

• 36.36

Dutch 81.82 60

• 65.22

English 30 45.45

• 35.48

Frisian 22.22 50

• 27.27

German 45 55.56 33.33

• 45.45

Icelandic 8.33

• 6.67

Norwegian 27.78 14.29

• 24

Swedish 35.71 29.41 16.67

• 28.95

Latvian 30.43 37.5

• 30.95

Lithuanian 41.38 39.13 25

• 39.29

Language 1st 2nd 3rd 4th 5th Total Catalan 26.83 8.11 25

• 18.18

French 38.1 66.67 100

• 60

Galician 22.22 25.71 66.67

• 25

Italian 25 20

• 18.07

Portuguese 36.11 26.92

• 31.25

Romanian 34.15 32.61 23.81

• 30.09

Spanish 19.57 6.06

• 13.1

Irish 17.65 26.32 33.33

• 21.43

Welsh 17.14 6.67

• 13.73

Greek 50 69.57 47.06

• 56.45

Estonian 22.58 24

• 19.4

Finnish 24.24 35.29 19.05

• 24.74
• N. Saami

34.78 23.68 40 25

• 29.9

Hungarian 25.93 25.58 12.5 20

• 20.16

Tatar 38.89

• 16.28

Turkish 25 20 12.5

• 16.44

Basque 27.78

• 17.86

Dargwa 25

• 25

Chechen 16.67

• 16.67

Georgian 39.29 9.09 3.7

• 15.31

Maltese 37.5 25

• 30.77

SLIDE 30

Saturation value - bone

• SAT. value

Languages HIGH Dutch, French, Greek, German, Slovak MEDIUM Lithuanian, Danish, English, Portuguese, Latvian, Maltese, Russian, Polish, Romanian, North Saami, Swedish, Bulgarian, Czech, Frisian, Serbian, Ukrainian, Slovene, Galician, Dargwa, Finnish, Norwegian, Irish, Hungarian LOW Croatian, Estonian, Catalan, Italian, Basque, Chechen, Turkish, Tatar, Georgian, Welsh, Spanish, Icelandic

SLIDE 31

Saturation value - bone

high medium low

SLIDE 32

Saturation value - word-classes - nouns

NOUNS LANGUAGES HIGH 30-40% German, Dutch, Dargwa, Greek, North Saami MEDIUM 20- 29.99% Finnish, English, Czech, Turkish, Lithuanian, Swedish, Chechen, Slovak, Frisian, Estonian, Italian, Hungarian, Norwegian, Romanian, Portuguese, Serbian, Georgian, Latvian, Spanish, Croatian, Catalan, Bulgarian, Russian, Ukrainian, Galician, Slovene, Irish LOW <20% Basque, Tatar, Polish, Danish, Welsh, Icelandic, French, Maltese

high medium low

SLIDE 33

Saturation value - word-classes - verbs

VERBS LANGUAGES HIGH 30-46% Greek, North Saami, Turkish, Georgian, German, Maltese, Dargwa Dutch, Lithuanian

MEDIUM

20-29,99% Latvian, Croatian, Catalan, Hungarian, Slovak, English, Finnish, French, Ukrainian, Serbian, Portuguese, Estonian, Swedish, Slovene, Italian, Czech, Romanian, Tatar, Bulgarian, Polish, Icelandic, Frisian, Galician, Spanish, Basque LOW <20% Chechen, Irish, Norwegian, Danish, Russian, Welsh

high medium low

SLIDE 34

Saturation value - word-classes - adjectives

high medium low

ADJECTIVES LANGUAGES HIGH 30-60% DARGWA, French, Lithuanian, GREEK, Turkish, Maltese, NORTH SAAMI, Estonian, Hungarian, DUTCH, Latvian, Spanish MEDIUM 19-29.99% Swedish, Finnish, Irish, English, Romanian, Portuguese, Italian, Galician, Slovak, Georgian, Ukrainian, Czech, Polish, German, Tatar, Croatian, Icelandic, Frisian, Welsh, Bulgarian, Slovene, Norwegian, Catalan, Basque, Serbian LOW <19% Chechen, Russian, DANISH

SLIDE 35

Satu turatio ion val alue - con

• nclu

lusions

• The richness of derivational networks is:
• sensitive to the word-class of the basic word.
• sensitive to the order of derivation (tendency for SV to fall gradually with the rising
• rder of derivation in all three word-classes)
• There is a tendency for languages to actualize 20-29% of the derivational

potential of a word-class.

• A medium saturation value (20-30%) can be considered the most typical SV
• f all word-classes and the first three orders of derivation.
• There is a core group of languages that keep high saturation values across all

three word-classes.

• There is no geographically homogeneous territory on which the languages of

topmost saturation values are spoken.

• There is a tendency for languages at geographically peripheral areas to

feature low saturation values.

SLIDE 36

Semantic categories and orders of derivation

Nouns:

• 1st order: Diminutive (25), Quality (21), Privative (14), Relational (11),

Action (10)

• 2nd order: Action (15) and Stative (9)

Polish pies-ek dog-DIM ’doggy’ Spanish

• jo-ar

eye-ACTION ’to eye’ Icelandic nafn-laus name-QUALITY ’nameless’ Turkish kemik-siz bone-PRIVATIVE ’boneless’ Welsh enw-ol name-RELATIONAL ’ pertaining to a name; nominal’ Croatian zub-ar-iti tooth-AGENT-ACTION ‘to be a dentist’

SLIDE 37

Semantic categories – blocking effect

Nouns

• Diminutive displays in noun-based derivations in the first order (11

languages)

• no clear tendencies in 5 languages
• 35 languages: tendencies are language specific
• semantic categories with blocking effects in the 1st order: Agent,

Augmentative, Collective, Diminutive, Entity, Female, Location, Manner, Patient, Pejorative, Possessive, Process, Quality, Relational, Resultative, Similative, Stative, and Temporal.

SLIDE 38

Semantic categories – blocking effect

Adjectives

• Manner and Stative have blocking effects in 9 languages each.

Verbs

• the greatest diversity of distribution of semantic categories with

blocking effects

SLIDE 39

Semantic categories – combinations

• no detectable tendencies among genera and families
• noun-based derivation: Quality and Stative, Causative and Action

Italian

• ss-ifica-zione

bone-CAUSATIVE-ACTION ‘ossification’

SLIDE 40

Semantic categories – multiple occurrences

• multiple occurrences of a semantic category in successive orders of

derivation

• Quality (11):
• Action (10)
• Diminutive (8)

Slovak kôst-oč-ka bone-DIMINUTIVE-DIMINUTIVE ‘very little stone’ Serbian камен-ит-ост stone-QUALITY-QUALITY ‘stonyness’ Icelandic af-hend-ing

ACTION-throw-ACTION

‘deliverance’

SLIDE 41

Sem emantic ic categ egories – con

• nclu

lusions

• the most clearly correlated with the 1st order of derivation:
• Nominal bases: Diminutive, Quality, Privative, Relational and Action
• Verbal bases: Action, Agent, Resultative and Ability
• Adjectival bases: Manner and Stative
• reoccurrence: Quality, Action, Diminutive
• no recurrent patterns
• combinality of semantic categories
• blocking – exception Diminutive

SLIDE 42

CONCLUSIONS

• a new method
• emphasis on semantics
• 8 parameters:
• number of derivatives
• orders of derivation
• saturation value
• semantic categories:
• order of derivation, combinations, blocking effect, multiple occurrences
• partial results – to be confirmed/elaborated/refined
• different samples of basic words
• different languages outside Europe
• a viable way of exploring derivational networks from cross-linguistic

perspective

SLIDE 43

THANK YOU!