General design information

SIMPLE LE4-8346

WP3.8

SIMPLE - GREEK LEXICON DOCUMENTATION

* * *

Document first version date

26/06/00

Document date

30/06/00

Document ID

WP3.8

Version

01

Doc. type

Document status

to be validated

Validation type

Comments

Name

Organisation

Purpose

From

Maria Gavrilidou

ILSP

documentation

Penny Labropoulou

ILSP

documentation

Elena Mantzari

ILSP

documentation

Sophia Roussou

ILSP

documentation

Danai Anagnostopoulou

ILSP

documentation

Elina Desipri

ILSP

documentation

To

TM

UB-FBG-TM

validation

General design information

Lexicon population

The Simple lexicon in its final form includes 10.000 semantic units from the categories of nouns, adjectives and verbs. These units correspond to a subset of the Parole lexicon, which currently includes morphological and syntactic information for 20.000 lemmas.

For the population of the Greek Simple lexicon, the following factors have been taken into consideration:

wide coverage of the template types,

selection of lemmas to be encoded only from those present in the Parole lexicon,

encoding of the most frequent lemmas,

encoding of the lemmas represented in the set of Basic Concepts,

encoding of the most important readings of the selected lemmas,

lexicon closure (i.e. all semantic units that are present in the Simple lexicon participating in a certain relation must also be fully encoded) within certain conditions.

The starting point for building the Simple lexicon is supplied by the set of Basic Concepts as provided by the Specifications Group. The Greek team has transferred the Basic Concepts into the Greek language. In this task, the aim was to select for each concept the prototypical lemma, taking into consideration two more points:

presence of the prototypical lemma in the Parole lexicon,

selection of a single word lemma, given that the Greek Parole lexicon does not contain compounds.

These two conditions could not always be satisfied. In this case, the relevant Basic Concepts have not been encoded in the Greek Simple lexicon.

Therefore, for each grammatical category, nouns and verbs, the initial set of 462 clustered Basic Concepts for Nouns and 199 for Verbs (English version) gave rise to 306 distinct lemmas of nouns and 150 distinct lemmas of verbs in Greek, which form the first source for the population of the Simple lexicon.

The Greek Simple lexicon finally includes 479 Semantic Units from this set of Basic Concepts; these, however, correspond to more Basic Concepts, since we have further merged some concepts, when their meanings were considered close enough in the Greek language to be coded as one Semantic Unit.

In addition, responding to the criterion of "wide coverage", a list of lemmas is selected for each template type. The selection of the lemmas is based on the frequency criterion, i.e. for each template type we select the most frequent lemmas from the Parole lexicon. It is important to stress that corpus frequency is derived from the Parole corpus, and that it refers to lemma frequency and not to sense frequency.

Finally, a third source for the lexicon population is the list of "dummy" semantic units that arises from the encoding of the first two sets of lemmas. Encoding the various readings of a lemma requires the encoding of its relations to other semantic units. Given the lexicon closure criterion, we have tried to fully encode all the target semantic units, provided they are already morphologically and syntactically encoded in Parole. However, this criterion was not entirely met, since the criterion of wide coverage of template types was considered more important and preference was given to encoding different readings rather than encoding "dummy" entries.

Coverage and completeness

Coverage and completeness of the lexicon refers to two aspects:

full encoding of the semantics of each lemma and, for each reading, full encoding of all the obligatory data for discrimination,

encoding of all the lemmas that are present in the lexicon, i.e. encoding of the target semantic units participating in a relation.

Our original aim was to fully encode the lemmas that have been selected in order to get the full representation of the semantics of each lemma, imposing only two conditions on this aim:

not all readings of the lemma are to be encoded : metaphorical readings are excluded; thus, the semantic units encoded for each lemma correspond roughly to the senses included also in the dictionaries of Modern Greek

each semantic unit should comprise all information that is deemed necessary for its distinction; our general principle consists in encoding only what is necessary and linguistically relevant for sense discrimination. At the same time, we avoid encoding "non-distinctive" and "encyclopaedic" semantic relations, especially if they point to dummy semantic units that cannot be encoded in the Simple lexicon.

However, in order to speed up the encoding process, this criterion is considered a secondary priority. This means that our attention was shifted to locating quickly lemmas having senses for a particular template type, and encoding them, instead of exhausting all the senses of a lemma. Still, the major senses of a given lemma have been encoded.

Use of background resources

No computational semantic lexica exist for Modern Greek; therefore, for the construction of the Simple lexicon in what concerns sense discrimination and gloss writing, two large and one medium size Greek dictionaries (in printed format) are being consulted.

Dictionary of Modern Greek language, G. Babiniotis, Centre Of Lexicology, Athens 1998

Dictionary of Common Modern Greek language, Institute of Modern Greek Studies, Aristotle University of Thessaloniki, Thessaloniki 1998

Modern Greek Dictionary, E. Kriaras, Ekdotiki Athinwn, Athens 1995

Current lexicon contents

Table of full and dummy SemU's

The Greek lexicon currently includes in total 10.009 semantic units and 1079 dummy units that have arisen from the semantic relations encoded for the full SemU's.

	Number of SemU's
Encoded SemU's	10009
Dummy SemU's	1079

Table of Syntactic Units encoded in Simple

The 10.009 full Semantic Units of the current Simple lexicon correspond to 5.748 morphological units, which have been encoded as 7.341 syntactic units in the Parole lexicon. Not all of these syntactic units have yet been encoded at the semantic level, as they may correspond to senses that have not yet been treated. Only 6.464 of them have been encoded at the semantic level.

	Number of SynU's
SynU's in Parole	7341
SynU's in Simple	6464

Table of SemU's per Grammatical Category

Grammatical category	Number of SemU's
Nouns	7001
Verbs	2000
Adjectives	1008
Total

Table of SemU's per Template type

Template	Number of SemU's
WVSFTemplateEntityPROT	29
WVSFTemplatePartPROT	107
WVSFTemplateBodypartPROT	174
WVSFTemplateGroupPROT	51
WVSFTemplateHumanGroupPROT	462
WVSFTemplateConcreteEntityPROT	30
WVSFTemplateLocationPROT	103
WVSFTemplate3DLocationPROT	28
WVSFTemplateGeopoliticalLocationPROT	185
WVSFTemplateAreaPROT	66
WVSFTemplateOpeningPROT	12
WVSFTemplateBuildingPROT	300
WVSFTemplateArtifactualareaPROT	51
WVSFTemplateMaterialPROT	10
WVSFTemplateArtifactPROT	299
WVSFTemplateArtifactualmaterialPROT	36
WVSFTemplateFurniturePROT	28
WVSFTemplateClothingPROT	105
WVSFTemplateContainerPROT	46
WVSFTemplateArtworkPROT	37
WVSFTemplateInstrumentPROT	76
WVSFTemplateMoneyPROT	34
WVSFTemplateVehiclePROT	81
WVSFTemplateSemioticartifactPROT	179
WVSFTemplateFoodPROT	12
WVSFTemplateArtifactFoodPROT	85
WVSFTemplateFlavouringPROT	17
WVSFTemplatePhysicalobjectPROT	28
WVSFTemplateOrganicobjectPROT	8
WVSFTemplateLivingentityPROT	4
WVSFTemplateAnimalPROT	9
WVSFTemplateEarth-AnimalPROT	127
WVSFTemplateAir-AnimalPROT	44
WVSFTemplateWater-AnimalPROT	31
WVSFTemplateHumanPROT	162
WVSFTemplatePeoplePROT	101
WVSFTemplateRolePROT	80
WVSFTemplateIdeoPROT	94
WVSFTemplateKinshipPROT	91
WVSFTemplateSocialstatusPROT	142
WVSFTemplateAgentoftemporaryactivityPROT	409
WVSFTemplateAgentofpersistentactivityPROT	396
WVSFTemplateProfessionPROT	609
WVSFTemplateVegetalentityPROT	8
WVSFTemplatePlantPROT	145
WVSFTemplateFlowerPROT	30
WVSFTemplateFruitPROT	59
WVSFTemplateAgentivePROT	12
WVSFTemplateCausePROT	9
WVSFTemplateConstitutivePROT	38
WVSFTemplateMicro-organismPROT	6
WVSFTemplateTelicPROT	2
WVSFTemplateSubstancePROT	52
WVSFTemplateNaturalsubstancePROT	94
WVSFTemplateSubstancefoodPROT	50
WVSFTemplateDrinkPROT	4
WVSFTemplateArtifactualdrinkPROT	47
WVSFTemplateAmountPROT	57
WVSFTemplatePropertyPROT	41
WVSFTemplateQualityPROT	138
WVSFTemplatePsychpropertyPROT	22
WVSFTemplatePhysicalpropertyPROT	34
WVSFTemplatePhysicalpowerPROT	6
WVSFTemplateColorPROT	28
WVSFTemplateShapePROT	20
WVSFTemplateSocialPropertyPROT	9
WVSFTemplateAbstractEntityPROT	21
WVSFTemplateDomainPROT	68
WVSFTemplateTimePROT	150
WVSFTemplateMoralstandardPROT	17
WVSFTemplateCognitivefactPROT	12
WVSFTemplateMovementofthoughtPROT	55
WVSFTemplateInstitutionPROT	207
WVSFTemplateConventionPROT	42
WVSFTemplateRepresentationPROT	26
WVSFTemplateLanguagePROT	104
WVSFTemplateSignPROT	50
WVSFTemplateInformationPROT	173
WVSFTemplateNumberPROT	11
WVSFTemplateUnitofmeasurementPROT	122
WVSFTemplateEventPROT	3
WVSFTemplatePhenomenonPROT	27
WVSFTemplateWeatherVerbPROT	11
WVSFTemplateDiseasePROT	25
WVSFTemplateStimulusPROT	9
WVSFTemplateAspectualPROT	15
WVSFTemplateCauseAspectualPROT	16
WVSFTemplateStatePROT	8
WVSFTemplateExistPROT	13
WVSFTemplateRelationalStatePROT	14
WVSFTemplateIdentificationalStatePROT	20
WVSFTemplateConstitutiveStatePROT	5
WVSFTemplateStativeLocationPROT	34
WVSFTemplateStativePossessionPROT	8
WVSFTemplateActPROT	4
WVSFTemplateNonRelationalActPROT	40
WVSFTemplateRelationalActPROT	70
WVSFTemplatecooperativeActivityPROT	86
WVSFTemplatePurposeActPROT	59
WVSFTemplateMovePROT	35
WVSFTemplateCauseMotionPROT	11
WVSFTemplateCauseActPROT	5
WVSFTemplateSpeechActPROT	23
WVSFTemplateCooperativeSpeechActPROT	23
WVSFTemplateReportingEventPROT	62
WVSFTemplateCommissiveSpeechActPROT	10
WVSFTemplateDirectiveSpeechActPROT	26
WVSFTemplateExpressiveSpeechActPROT	37
WVSFTemplateDeclarativeSpeechActPROT	12
WVSFTemplatePsychologicalEventPROT	14
WVSFTemplateCognitiveEventPROT	38
WVSFTemplateJudgementPROT	12
WVSFTemplateExperienceEventPROT	88
WVSFTemplateCauseExperienceEventPROT	54
WVSFTemplatePerceptionPROT	14
WVSFTemplateModalEventPROT	24
WVSFTemplateChangePROT	3
WVSFTemplateRelationalChangePROT	12
WVSFTemplateConstitutiveChangePROT	19
WVSFTemplateChangeofStatePROT	148
WVSFTemplateChangeofValuePROT	23
WVSFTemplateChangeofPossessionPROT	43
WVSFTemplateTransactionPROT	89
WVSFTemplateChangeofLocationPROT	121
WVSFTemplateNaturalTransitionPROT	36
WVSFTemplateAcquireKnowledgePROT	28
WVSFTemplateCauseChangePROT	4
WVSFTemplateCauseRelationalChangePROT	16
WVSFTemplateCauseConstitutiveChangePROT	37
WVSFTemplateCauseChangeofStatePROT	232
WVSFTemplateCauseChangeofValuePROT	41
WVSFTemplateCauseChangeLocationPROT	88
WVSFTemplateCauseNaturalTransitionPROT	36
WVSFTemplateCreationPROT	27
WVSFTemplatePhysicalCreationPROT	20
WVSFTemplateMentalCreationPROT	22
WVSFTemplateSymbolicCreationPROT	25
WVSFTemplateCopyCreationPROT	14
WVSFTemplateGiveKnowledgePROT	15
WVSFTemplateADJModal	16
WVSFTemplateADJEmotive	1
WVSFTemplateADJManner	18
WVSFTemplateADJObjectRelated	81
WVSFTemplateADJPhysicalProperty	281
WVSFTemplateADJPsychologicalProperty	276
WVSFTemplateADJSocialProperty	264
WVSFTemplateADJTemporalProperty	71

Table of SemU's per Domain

Domain	Number of SemU's
TSVP_HEALTH_AND_MEDICINE_TS_domaine_D	12
TSVP_ADMINISTRATIVE_LAW_TS_domaine_D	12
TSVP_CONSTITUTIONAL_LAW_TS_domaine_D	6
TSVP_GENERAL_TS_domaine_D	10009
TSVP_LIFE_SCIENCES_TS_domaine_D	3
TSVP_MEDICINE_TS_domaine_D	81
TSVP_OBSTETRICS_TS_domaine_D	6
TSVP_ZOOLOGY_TS_domaine_D	101
TSVP_PUBLISHING_TS_domaine_D	37
TSVP_FILM_TS_domaine_D	44
TSVP_THEATER_TS_domaine_D	48
TSVP_DERMATOLOGY_TS_domaine_D	7
TSVP_LINGUISTICS_TS_domaine_D	157
TSVP_CHEMISTRY_TS_domaine_D	56
TSVP_BOTANY_TS_domaine_D	243
TSVP_ECONOMICS_TS_domaine_D	27
TSVP_ANATOMY_TS_domaine_D	149
TSVP_PSYCHOANALYSIS_TS_domaine_D	5
TSVP_BANKING_TS_domaine_D	10
TSVP_AUDIOVISUAL_TS_domaine_D	2
TSVP_SPORTS_AND_LEISURE_TS_domaine_D	22
TSVP_MILITARY_TS_domaine_D	41
TSVP_SPORT_TS_domaine_D	68
TSVP_WASHING_TS_domaine_D	1
TSVP_PSYCHOLOGY_TS_domaine_D	154
TSVP_MAMMALOGY_TS_domaine_D	103
TSVP_SURGERY_TS_domaine_D	2
TSVP_DENTISTRY_TS_domaine_D	4
TSVP_RELIGION_TS_domaine_D	27
TSVP_ASTRONOMY_TS_domaine_D	70
TSVP_SAILING_YACHTING_AND_BOATING_TS_domaine_D	7
TSVP_SEA_TRANSPORT_TS_domaine_D	36
TSVP_AIR_TRANSPORT_TS_domaine_D	22
TSVP_COSMETICS_TS_domaine_D	4
TSVP_ARMY_TS_domaine_D	34
TSVP_LAW_ENFORCEMENT_TS_domaine_D	12
TSVP_POLITICS_TS_domaine_D	95
TSVP_CLOTHING_INDUSTRY_TS_domaine_D	9
TSVP_PENAL_SYSTEM_TS_domaine_D	3
TSVP_CRIME_TS_domaine_D	16
TSVP_PHYSIOLOGY_TS_domaine_D	9
TSVP_BUSINESS_TS_domaine_D	133
TSVP_LIVESTOCK_FARMING_TS_domaine_D	4
TSVP_MONARCHY_TS_domaine_D	29
TSVP_EDUCATION_TS_domaine_D	96
TSVP_PRIMARY_AND_SECONDARY_EDUCATION_TS_domaine_D	17
TSVP_GARDENING_TS_domaine_D	3
TSVP_ICHTHYOLOGIE_TS_domaine_D	15
TSVP_LAW_TS_domaine_D	88
TSVP_COMMERCE_TS_domaine_D	77
TSVP_MANAGEMENT_TS_domaine_D	6
TSVP_ORNITHOLOGY_TS_domaine_D	43
TSVP_BASEBALL_TS_domaine_D	1
TSVP_HERPETOLOGY_TS_domaine_D	6
TSVP_HEALTH_TS_domaine_D	2
TSVP_FINANCE_TS_domaine_D	46
TSVP_MAIL_TS_domaine_D	9
TSVP_INSURANCE_TS_domaine_D	3
TSVP_DRINK_TS_domaine_D	32
TSVP_DRUGS_TS_domaine_D	3
TSVP_MATHEMATICS_TS_domaine_D	66
TSVP_GOVERNMENT-ADMINISTRATION_TS_domaine_D	4
TSVP_TRANSPORT_TS_domaine_D	96
TSVP_NAVY_TS_domaine_D	10
TSVP_TEXTILES_TS_domaine_D	20
TSVP_JEWELRY_TS_domaine_D	13
TSVP_CHRISTIANITY_TS_domaine_D	23
TSVP_ARCHITECTURE_TS_domaine_D	75
TSVP_GRAPHIC_ARTS_TS_domaine_D	25
TSVP_HOTEL_BUSINESS_TS_domaine_D	8
TSVP_RESTAURATION_TS_domaine_D	2
TSVP_MUSIC_TS_domaine_D	145
TSVP_LEISURE_TS_domaine_D	12
TSVP_ARTS_TS_domaine_D	87
TSVP_MANUFACTURING_INDUSTRY_TS_domaine_D	20
TSVP_SHIP_BUILDING_TS_domaine_D	13
TSVP_ENTOMOLOGY_TS_domaine_D	25
TSVP_AEROSPACE_ENGINEERING_TS_domaine_D	9
TSVP_COMPUTING_TS_domaine_D	11
TSVP_BASKETBALL_TS_domaine_D	13
TSVP_AMERICAN_FOOTBALL_TS_domaine_D	1
TSVP_RADIO-TELEVISION_TS_domaine_D	25
TSVP_BREWING_TS_domaine_D	3
TSVP_DISTILLING_TS_domaine_D	2
TSVP_SOCCER_TS_domaine_D	6
TSVP_ELECTRICAL_ENGINEERING_TS_domaine_D	6
TSVP_ALCHEMY_TS_domaine_D	3
TSVP_AGRICULTURE_TS_domaine_D	5
TSVP_WOODWORKING_TS_domaine_D	4
TSVP_MINING-GENERAL_TS_domaine_D	1
TSVP_EARTH_SCIENCES_TS_domaine_D	9
TSVP_PSYCHIATRY_TS_domaine_D	8
TSVP_GAMES_TS_domaine_D	10
TSVP_CHESS_TS_domaine_D	5
TSVP_DIPLOMACY_TS_domaine_D	9
TSVP_AIRFORCE_TS_domaine_D	2
TSVP_ETHNOLOGY_TS_domaine_D	101
TSVP_PHONETICS_TS_domaine_D	1
TSVP_GEOLOGY_TS_domaine_D	21
TSVP_AUTOMOBILE_ENGINEERING_TS_domaine_D	5
TSVP_METEOROLOGY_TS_domaine_D	40
TSVP_ELECTRONIC_ENGINEERING_TS_domaine_D	4
TSVP_CONSTRUCTION_TS_domaine_D	331
TSVP_MECHANICAL_ENGINEERING_TS_domaine_D	15
TSVP_ROAD_TRANSPORT_TS_domaine_D	15
TSVP_CREATIVE_WRITING_TS_domaine_D	41
TSVP_PHYSICS_TS_domaine_D	26
TSVP_EMPLOYMENT_TS_domaine_D	1
TSVP_FORESTRY_TS_domaine_D	6
TSVP_MEDIA_TS_domaine_D	32
TSVP_HIGHER_EDUCATION_TS_domaine_D	23
TSVP_TELECOMMUNICATIONS_TS_domaine_D	14
TSVP_ACCOUNTING_TS_domaine_D	8
TSVP_PAPERMAKING_TS_domaine_D	3
TSVP_GEOPOLITICS_TS_domaine_D	1
TSVP_POLITICS_AND_GOVERNMENT_TS_domaine_D	96
TSVP_BAKERY_TS_domaine_D	5
TSVP_BUILDING_CRAFTS_TS_domaine_D	1
TSVP_PHOTOGRAPHY_TS_domaine_D	10
TSVP_FURNITURE_TS_domaine_D	8
TSVP_MARKETING_TS_domaine_D	2
TSVP_DANCE_TS_domaine_D	14
TSVP_BUS_TRANSPORT_TS_domaine_D	1
TSVP_CAR_TRANSPORT_TS_domaine_D	8
TSVP_TRUCKING_TS_domaine_D	6
TSVP_FURNISHING_TS_domaine_D	32
TSVP_ISLAM_TS_domaine_D	5
TSVP_GLASSMAKING_TS_domaine_D	1
TSVP_PRINTING_TS_domaine_D	4
TSVP_OPTICS_TS_domaine_D	6
TSVP_BALLET_TS_domaine_D	2
TSVP_CERAMICS_TS_domaine_D	2
TSVP_POTTERY_TS_domaine_D	4
TSVP_PETROLOGY_TS_domaine_D	1
TSVP_SCIENCES_TS_domaine_D	16
TSVP_MAGIC_AND_WITCHCRAFT_TS_domaine_D	6
TSVP_COKING_INDUSTRY_TS_domaine_D	1
TSVP_ROMAN_CATHOLICISM_TS_domaine_D	3
TSVP_OCEANOGRAPHY_TS_domaine_D	1
TSVP_JUDAISM_TS_domaine_D	2
TSVP_CYTOLOGY_TS_domaine_D	1
TSVP_ATHLETICS_TS_domaine_D	12
TSVP_SEA_FISHING_TS_domaine_D	4
TSVP_POLO_TS_domaine_D	1
TSVP_SCULPTURE_TS_domaine_D	19
TSVP_SMOKING_TS_domaine_D	6
TSVP_ENOLOGY_TS_domaine_D	10
TSVP_STATISTICS_TS_domaine_D	2
TSVP_HYDROGRAPHY_TS_domaine_D	2
TSVP_OPERA_TS_domaine_D	3
TSVP_UTILITIES_TS_domaine_D	3
TSVP_NEWSPAPER_PUBLISHING_TS_domaine_D	17
TSVP_PLUMBING_TS_domaine_D	3
TSVP_THEOLOGY_TS_domaine_D	10
TSVP_FASHION_TS_domaine_D	122
TSVP_HUNTING_AND_SHOOTING_TS_domaine_D	4
TSVP_GEOMETRY_TS_domaine_D	27
TSVP_ADVERTISING_TS_domaine_D	3
TSVP_PAINTMAKING_TS_domaine_D	9
TSVP_TILING_TS_domaine_D	1
TSVP_PHILOSOPHY_TS_domaine_D	44
TSVP_PROTESTANTISM_TS_domaine_D	3
TSVP_ELECTRICITY_TS_domaine_D	9
TSVP_TYPOGRAPHY_TS_domaine_D	1
TSVP_MARTIAL_ARTS_TS_domaine_D	1
TSVP_HISTORY_TS_domaine_D	9
TSVP_BEEKEEPING_TS_domaine_D	1
TSVP_FISHING_TS_domaine_D	4
TSVP_BACTERIOLOGY_TS_domaine_D	3
TSVP_OIL_INDUSTRY_TS_domaine_D	1
TSVP_RAIL_TRANSPORT_TS_domaine_D	8
TSVP_SOCIOLOGY_TS_domaine_D	113
TSVP_PHILATELY_TS_domaine_D	1
TSVP_ANTIQUITY_TS_domaine_D	1
TSVP_PHARMACY_TS_domaine_D	2
TSVP_TOBACCO_INDUSTRY_TS_domaine_D	2
TSVP_VIROLOGY_TS_domaine_D	1
TSVP_HEATING_TS_domaine_D	2
TSVP_POETICS_TS_domaine_D	2
TSVP_SEISMOLOGY_TS_domaine_D	6
TSVP_MINERALOGY_TS_domaine_D	6
TSVP_MILITARY_LAW_TS_domaine_D	1
TSVP_ARCHAEOLOGY_TS_domaine_D	9
TSVP_WOOL_INDUSTRY_TS_domaine_D	1
TSVP_KITCHEN_EQUIMENT_TS_domaine_D	1
TSVP_MYTHOLOGY_TS_domaine_D	2
TSVP_GEOGRAPHY_TS_domaine_D	221
TSVP_ASTROLOGY_TS_domaine_D	6
TSVP_BUDDHISM_TS_domaine_D	4
TSVP_GAS_TS_domaine_D	1
TSVP_SERVICE_INDUSTRY_TS_domaine_D	3
TSVP_SUBWAY_TRANSPORT_TS_domaine_D	1
TSVP_ACOUSTICS_TS_domaine_D	4
TSVP_SOCIAL_ACTION_TS_domaine_D	1
TSVP_CIVIL_LAW_TS_domaine_D	1
TSVP_CRIMINAL_LAW_TS_domaine_D	6
TSVP_VITICULTURE_TS_domaine_D	3
TSVP_HYDROLOGY_TS_domaine_D	4
TSVP_MYCOLOGY_TS_domaine_D	4
TSVP_INTERNATIONAL_LAW_TS_domaine_D	3
TSVP_FOOD_TS_domaine_D	217

Table stating SemU's per Semantic Class

Semantic class	Number of SemU's
TSVP_ABSTRACT_TS_classificateur_de_nom_C	429
TSVP_ACTIVITY_TS_classificateur_de_nom_C	39
TSVP_ADJ_COULEUR_TS_classificateur_d_adjectif_C	58
TSVP_ADJ_GEO_TS_classificateur_d_adjectif_C	95
TSVP_ADJ_PERIOD_TS_classificateur_d_adjectif_C	71
TSVP_AFFECTION_TS_classificateur_de_nom_C	7
TSVP_AGENCY_TS_classificateur_de_nom_C	212
TSVP_AMOUNT_TS_classificateur_de_nom_C	57
TSVP_AMPHIBIAN_TS_classificateur_de_nom_C	3
TSVP_ANIMAL_TS_classificateur_de_nom_C	21
TSVP_APPARATUS_TS_classificateur_de_nom_C	3
TSVP_ARTIFACT_TS_classificateur_de_nom_C	634
TSVP_ATTRIBUTE_TS_classificateur_de_nom_C	309
TSVP_BIO_TS_classificateur_de_nom_C	251
TSVP_BIRD_TS_classificateur_de_nom_C	37
TSVP_BODY_PART_TS_classificateur_de_nom_C	174
TSVP_BODY_TS_classificateur_de_verbe_C	22
TSVP_BUILDING_TS_classificateur_de_nom_C	302
TSVP_CHANGE_TS_classificateur_de_verbe_C	638
TSVP_COGNITION_VB_TS_classificateur_de_verbe_C	109
TSVP_COGNITIVE_FACT_TS_classificateur_de_nom_C	13
TSVP_COLOR_TS_classificateur_de_nom_C	28
TSVP_COMMUNICATION_TS_classificateur_de_verbe_C	193
TSVP_COMPETITION_TS_classificateur_de_verbe_C	1
TSVP_CONCRETE_TS_classificateur_de_nom_C	29
TSVP_CONTACT_TS_classificateur_de_verbe_C	27
TSVP_CONTAINER_TS_classificateur_de_nom_C	46
TSVP_CREATION_TS_classificateur_de_verbe_C	108
TSVP_CURRENCY_TS_classificateur_de_nom_C	34
TSVP_DAY_TS_classificateur_de_nom_C	22
TSVP_EMOTION_VB_TS_classificateur_de_verbe_C	149
TSVP_ENTITY_TS_classificateur_de_nom_C	60
TSVP_ETHNOS_TS_classificateur_de_nom_C	101
TSVP_FISH_TS_classificateur_de_nom_C	14
TSVP_FLOWER_TS_classificateur_de_nom_C	32
TSVP_FORM_TS_classificateur_de_nom_C	20
TSVP_FRUIT_TS_classificateur_de_nom_C	62
TSVP_FUNCTIONAL_SPACE_TS_classificateur_de_nom_C	1
TSVP_FURNITURE_TS_classificateur_de_nom_C	33
TSVP_GARMENT_TS_classificateur_de_nom_C	108
TSVP_GEOGRAPHY_TS_classificateur_de_nom_C	185
TSVP_HUMAN_TS_classificateur_de_nom_C	1112
TSVP_IDEO_TS_classificateur_de_nom_C	94
TSVP_ILLNESS_TS_classificateur_de_nom_C	21
TSVP_INANIMATE_TS_classificateur_de_nom_C	13
TSVP_INSECT_TS_classificateur_de_nom_C	21
TSVP_INSTRUMENT_TS_classificateur_de_nom_C	58
TSVP_LETTER_TS_classificateur_de_nom_C	23
TSVP_LIVING_BEING_TS_classificateur_de_nom_C	10
TSVP_LOCATION_TS_classificateur_de_nom_C	260
TSVP_MAMMAL_TS_classificateur_de_nom_C	103
TSVP_MATTER_TS_classificateur_de_nom_C	92
TSVP_MEASURE_UNIT_TS_classificateur_de_nom_C	133
TSVP_MEASURING_INSTRUMENT_TS_classificateur_de_nom_C	3
TSVP_MICROORGANISM_TS_classificateur_de_nom_C	6
TSVP_MOLLUSC_TS_classificateur_de_nom_C	10
TSVP_MONTH_TS_classificateur_de_nom_C	25
TSVP_MOTION_TS_classificateur_de_verbe	255
TSVP_MUSHROOM_TS_classificateur_de_nom_C	4
TSVP_MUSICAL_INSTRUMENT_TS_classificateur_de_nom_C	39
TSVP_NOTION_TS_classificateur_de_nom_C	111
TSVP_OBJECT_TS_classificateur_de_nom_C	48
TSVP_OCCUPATION_AGENT_TS_classificateur_de_nom_C	1149
TSVP_OCCUPATION_TS_classificateur_de_nom_C	71
TSVP_ORGANISM_TS_classificateur_de_nom_C	1
TSVP_PERCEPTION_TS_classificateur_de_verbe_C	22
TSVP_PERIOD_TS_classificateur_de_nom_C	149
TSVP_PHENOMENON_TS_classificateur_de_nom_C	34
TSVP_PLANT_TS_classificateur_de_nom_C	99
TSVP_POSSESSION_TS_classificateur_de_verbe_C	132
TSVP_PROCESS_TS_classificateur_de_nom_C	1
TSVP_PSYCHOLOGICAL_FEATURE_TS_classificateur_de_nom_C	22
TSVP_REPTILE_TS_classificateur_de_nom_C	6
TSVP_SHRUB_TS_classificateur_de_nom_C	9
TSVP_STATE_TS_classificateur_de_nom_C	13
TSVP_STATIVE_TS_classificateur_de_verbe_C	113
TSVP_SUBSTANCE_TS_classificateur_de_nom_C	247
TSVP_SYSTEM_OF_THOUGHT_TS_classificateur_de_nom_C	55
TSVP_TIME_PERIOD_TS_classificateur_de_nom_C	144
TSVP_TREE_TS_classificateur_de_nom_C	38
TSVP_VEHICLE_TS_classificateur_de_nom_C	86
TSVP_WEATHER_VB_TS_classificateur_de_verbe_C	11

Semantic encoding

The process of semantic encoding has been implemented in two main phases :

The semantic typing of SynU’s included in the Parole lexicon. During this phase our aim was to identify the entries to be encoded under the ontology of the concrete and abstract entities (events not included). We proceeded in two ways :

Template assignment to the 3.000 most frequent SynU’s.

Additional template assignment for some entries from the full list of the Parole nouns sorted by frequency in order to ensure that all templates will be covered to a satisfactory degree; for this reason, a minimum of 40 entries per template type has been conventionally set. However, this minimum has not been attained for certain templates, e.g. micro-organisms, given the restricted number of entries these templates include in general.

Further template assignment based on morphological and syntactic criteria, irrespective of the frequency criterion, in order to ensure quick identification of candidate semantic units. This mode of work was mainly explored for the template assignment of events and adjectives.

Actual semantic encoding of the identified SynU’s on the basis of groups of template types of the same "family" (e.g. all entries belonging to the template Human and its subtemplates). This approach ensures consistency and systematicity during codification and facilitates checking over the information of templates.

Criteria for Syntax-Semantics linking

The Syntax-Semantics linking is represented at the CorrespSynUSemU object, which is embedded in the SynU. The Correspondence object included in the CorrespSynUSemU determines the type of linking between syntactic positions and semantic arguments.

SynU- SemU relations

Three linking relations between SynU’s and SemU’s, depending on the number of SynU's and SemU's linked for each MuS, have been observed up to now.

One to one : when one SynU is linked to one SemU. That means that the SynU has only one meaning.

In the simplest case, the SynU is assigned one syntactic description and therefore one CorrespSynUSemU object. We distinguish between two cases :

Non-predicative SemU’s, i.e. entries that have no complements at the syntactic level and no arguments at the semantic level : these are not linked to any predicative structure and they have no Correspondence in the CorrespSynUSemU.

<SynU

id="trapezi"

description="Nnull">

<CorrespSynUSemU

targetsemu="SEMUtrapezi"></SynU>

Predicative SemU’s, i.e. entries that have at least one filled position at syntax projected onto the semantics. These are linked to the relevant predicate, according to the guidelines, and bear information on the Correspondence object; the appropriate value of the Correspondence relies on the number of complements at the syntactic level which are present at the semantic level.

<SynU

id="mytera"

description="Ncomplgenopt">

<CorrespSynUSemU

targetsemu="SEMUmytera"

correspondence="ISOmonovalent"></SynU>

In more complex cases, the SynU is assigned two syntactic descriptions linked through a frameset; for both these descriptions, the meaning does not change, and the SynU is linked to one SemU. Reciprocal alternation is such a case. For these entries, the syntactic unit is linked to the same predicate, and two correspondences are used in order to link the two syntactic positions to the appropriate arguments :

<SynU

id="sunoreuo"

description="VNomnpOblppme"

framesetl="fsunoreuo">

<CorrespSynUSemU

targetsemu="SEMUsunoreuo"

correspondence= "ISObivalent">

<CorrespSynUSemU

targetsemu="SEMUsunoreuo"

correspondence= "P0toArg0P0toArg1"

description= "VNomnpplu"></SynU>

One to many : when one SynU is linked to more than one SemU’s.

In the simple case, one SynU has more than one different meanings. This case applies to different readings of either truly polysemous or systematically polysemous lemmas.

<SynU

id="vivlio"

description="Nnull">

<CorrespSynUSemU

targetsemu="SEMUvivlio1">

<CorrespSynUSemU

targetsemu="SEMUvivlio2">  </SynU>

In more complex cases, the SynU is assigned two syntactic descriptions which are linked through a frameset but have a different meaning; therefore, the SynU is linked to two different SemU’s. Such an example is provided by the causative/inchoative alternation.

In these cases, the SynU is linked to two CorrespSynUSemU objects, with two SemUs and two correspondences to two different predicates; that is, we have decided to link each semantic unit to a different predicate :

<SynU

id="kleino"

description="VNomnpAccnpobl"

frameset="fanavo">

<CorrespSynUSemU

targetsemu="SEMUkleino1"

correspondence="ISObivalent">

<CorrespSynUSemU

targetsemu="SEMUkleino2"

correspondence="ISOmonovalent"

description="VactNomnp">

Many to one : when more than one SynU’s of the same MuS is linked to one SemU, i.e. different syntactic descriptions have the same meaning.

Up to now, we have met such examples from SynU's that have been split due to syntax constraints (e.g. presence of one complement affecting the way another complement is realised); for instance, deverbal nouns such as κατοχή [possession], are encoded as two SynU's : one with description "Nsubjapoobjgen" - with optional subject PP [apo] and an obligatory object NP [genitive]- and a second one with "Nnull". We have opted for this solution as the subject position is realised only when there is an object; therefore, a second description is required, where the object is not realised as well. We have not used the frameset mechanism to link such cases, since we have decided to use framesets only for cases referred to in the bibliography as "alternations".

<SynU

id="katohy1"

description="Nsubjapoobjgen">

<CorrespSynUSemU

targetsemu="SEMUkatohy"

correspondence= "ISObivalent"></SynU>

<SynU

id="katohy2" <!—possession2-->

description="Nnull">

<CorrespSynUSemU

targetsemu="SEMUkatohy"></SynU>

Types of Correspondence

We have used three types of Correspondence depending on the nature of mapping between syntactic positions and arguments :

ISO(morphic) when there is a "direct" mapping between positions and arguments.

CROSSED when there is a "crossed" mapping between positions and arguments.

ALT(ernation) when there is a mapping of one position to many arguments; this is the case of certain types of alternation (e.g. reciprocal alternation, coordinated objects).

We have decided not to use AUG(mented) correspondence, i.e. when the semantic representation includes "shadow" arguments.

Subtypes of Correspondence used include:

Correspondence	Comment	Example
ISOmonovalent	Isomorphic mapping for unary predicates	μητέρα [mother]
ISObivalent	Isomorphic mapping for bivalent predicates	απουσιάζω [be absent]
ISOtrivalent	Isomorphic mapping for trivalent predicates	κρατάω/ώ [keep]
ISOtetravalent	Isomorphic mapping for tetravalent predicates	μεταβιβάζω [transfer]

Document first version date	26/06/00
Document date	30/06/00
Document ID	WP3.8
Version	01
Doc. type
Document status	to be validated
Validation type
Comments

	Name	Organisation	Purpose

From	Maria Gavrilidou	ILSP	documentation
	Penny Labropoulou	ILSP	documentation
	Elena Mantzari	ILSP	documentation
	Sophia Roussou	ILSP	documentation
	Danai Anagnostopoulou	ILSP	documentation
	Elina Desipri	ILSP	documentation


To	TM	UB-FBG-TM	validation