Construction of the Lexicon

The words in the Finnish lexicon are selected from three sources. First, all the simple nouns and all the proper nouns in the Frequency Dictionary of Finnish (Saukkonen, Haipus, Niemikorpi & Sulkala 1979) are included in the Finnish lexicon. The second source is the FINTAG-corpus compiled of texts representing different text types (for example academic prose, textbooks, newspapers, and magazines). The size of this corpus is about 1,3 million running words. The criteria for including a word in the lecixon was a frequency of no less than 4 for nouns and a frequency of no less than 2 for verbs and adjectives in the above mentioned corpus. The third source is the sample of the hs90-corpus containing about 1,0 million running words from the newspaper Helsingin Sanomat 1990.

2. Current Lexicon Contents

2.1 Morphological layer

2.1.1 Summary of the morphological information

Number of simple morphological units	25421
Number of compound morphological units
Number of affix morphological units	60
Number of agglutinated morphological units
Number of graphical morphological units	25421
Number of simple inflection modes
Number of simple compound inflection modes

Category	Subcategory	Number of Units
Noun	common	17785
Noun	proper	746
Verb	normal	2941
Verb	impersonal	88
Adjective	quali	2931
Adjective	pronominal	23
Adjective	noninflecting	11
Adjective	ordinal	10
Adverb		569
Ad-adjective		42
Adposition	postposition	151
Adposition	preposition	44
Pronoun	personal	6
Pronoun	demonstrative	6
Pronoun	reflexive	1
Pronoun	relative	2
Pronoun	interrogative	5
Pronoun	indefinite	17
Numeral	cardinal	24
Conjunction	coordinative	8
Conjunction	subordinative	11

2.1.2 General remarks

In the Finnish lexicon the lemmatised form for nominals is nominative singular, or, if the word does not have this particular form, the lemmatised form is one of the existing word forms (for example nominative plural for pluralia tantum). For verbs the lemmatised form is 1st infinitive, or when this does not exist, the lemmatised form is 3rd person singular in active indicative present tense. For adverbs and adpositions inflected in locative cases there is one Morphological Unit for every different word form. The criteria for splitting morphlogical units is considered to be in line with the GENELEX principles.

There are some non-standardized categories, features and feature values added in the morphological layer in order to make the model more suitable for the Finnish languge.

Among these are two minor grammatical categories postulated for the Finnish lexicon, namely ad-adjectives and post-adverbs. Ad-adjectives are adverbs used to modify adjectives and other adverbs. Ad-adjectives never modify a verb, which is considered an argument to separate them from other adverbs and to postulate a new grammatical category 'ad-adjective'. Post-adverbs are adverbs that require a nominal complement. In this regard they behave like adpositions, but the difference is that a post-adverb does not determine the case of its complement like an adposition does.

Adjectives with no inflection are classified as a separate grammatical subcategory, i.e. NONINFLECTING. They never display the predicative function, and their position in the syntagma is immediately before the headword and after other noun modifiers that agree with the headword in case and number. (Vilkuna 1996.) Also pronouns that function like adjectives, i.e. modify nouns and agree with their headword in the normal way, are considered to form a separate grammatical subcategory, PRONOMINAL.

The Finnish case system contains four grammatical cases (i.e. NOMINATIVE, PARTITIVE, GENITIVE, and ACCUSATIVE), six locative cases, which are structured according to two dimensions, i.e. location and direction (see the table below), two abstract locative cases (ESSIVE and TRANSLATIVE) and finally the so called 'marginal cases' (ABESSIVE, COMITATIVE and INSTRUCTIVE). (See for example Karlsson 1987 and Vilkuna 1996.)

The system of the Finnish local cases according to Karlsson (1987:99):

		LOCATION
		INSIDE	OUTSIDE
	STATIC	inessive	adessive
DIREC-	AWAY FROM	elative	ablative
TION	TOWARDS	illative	allative

An additional personal ending needed in the Finnish personal inflection is the fourth person indefinite, which is a special personal ending for the passive verb forms referring to personal but indefinite actor(s) of the process or action described by the verb.

In marking possession in Finnish the word signifying what is possessed also takes an ending. The possessive suffixes in the 3rd person singular and in the 3rd person plural are the same, and the (additional) value SGPL3 for the morphological feature POSSESSOR is thus meant to indicate the fact that a word form containing the possessive suffix in question is ambiguous between the two possible interpretations. (See e.g. Karlsson 1996.)

The morphological features TENSE and MOOD are classified as one distributional class in Finnish morphology, i.e. all the sequences standing for some possible combination of tense and mood are always indicated by only one surface morph. Although the functional endings of the infinitives and participles are in complementary distribution with the morphemes indicating tense and mood, they are considered to form a separate morphological feature (called NONFINITE here). Non-finite forms can take a case ending and a possessive suffix, and participles are also inflected for number. (Karlsson 1987.)

2.1.3 Inflection modes

2.1.3.1 General principles

In the Finnish lexicon the information needed to generate the different word forms (i.e. about 6000 word forms for a noun, and about 12000 word forms for a verb) is included in the different kinds of stems a word can have and in the different kinds of attributes that are linked to the stems. A noun can have either 5 or 6 stems, an adjective 7 or 8 stems, and a verb either 6, 7 or 8 stems.

In the same way every single variant of an inflectional ending with the corresponding attributes is listed in the lexicon (for example the illative ending has no less than 42 different allomorphs).

The idea is basically that an ending can be matched with a word stem if, on the one hand, one can find exactly the same attributes in the element Radg describing the behavior of a stem and in the Radg describing the behavior of an ending and if, on the other hand, the set of attributes have exactly the same values. In addition some attributes are needed in linking different suffixes.

Below are the lists containing the attributes needed to describe the Finnish inflectional morphology in the LE-PAROLE lexicon model. All the possible combinations of the word stems for the main parts of speech are also stated in the tables below. All frequencies of the combinations are given in the table at 2.1.3.6 (Frequencies of the different combinations of stems). One example of the inflection of the nominals is also given in the appendix.

The most frequently used inflected forms of pronouns are listed in the lexicon according to the common LE-PAROLE lexicon model for encoding morphological information.

The nouns, verbs, and adjectives in the lexicon are also marked with a code indicating the inflectional category the word belongs to. The categories used are those defined in "The Basic Dictionary of Finnish" ("Suomen kielen perussanakirja"). We have used the codes referring to the different inflectional categories as id-numbers, which identify the different MFGs in the SGML-encoded lexicon. In some instances there may be the symbol "DEFAULTMF" instead of the specific code for an inflectional category.

2.1.3.2 Attributes and Combination of stems for nouns and adjectives

Attributes needed for adding suffixes to nominal stems

ATTRIBUTES	VALUES
stemtype
BASE	base form
SSG	strong vowel stem
WSG	weak vowel stem
SPL	strong plural stem
WPL	weak plural stem
CON	consonant stem
dstem
PVE	positive stem
CVE	comparative stem
SVE	superlative stem
back	does the stem contain back or front vowels
has2v	is there a long vowel in the end of the vowel stem
endingv	is there a vowel in the end of the stem in question
basei	is there the vowel 'i' in the end of the base form
syll1p3	is the word 2-syllabic or not
syll2p3	is the word monosyllabic or not
contexte_var	this attribute is used mainly for restricting the application of certain rules
vq	vowel quality
stemc	does the word have a consonant stem
ending2v	is there two different vowels in the end of the vowel stem

Attributes needed for linking different suffixes

px	This attribute tells if a case ending must/must not/can be followed by one of the possessive suffixes.
px3	This attribute tells if a certain type of case endings and a certain type of possessive suffixes match.
vqending	Vowel quality linked to the suffixes.

2.1.3.3 Nouns

Stemtypes and their attributes

stemtype=Base	back
	endingv
	has2v
	basei
	vq
	syll2p3
stemtype=SSg	back
	has2v
	stemc
	vq
	syll2p3
stemtype=WSg	back
	syll1p3
stemtype=SPl	back
	endingv
	ending2v
	has2v
	basei
	syll1p3
stemtype=WPl	back
	endingv
	has2v
	syll1p3
stemtype=Con	back

Different combinations of noun stems

Nouns with 5 stems

Stemtype	Examples
Base	kauppa (shop)	kivi (stone)
SSg	kauppa/na	kive/nä
WSg	kaupa/n	kive/n
SPl	kauppo/ina	kiv/inä
WPl	kaupo/issa	kiv/issä

Nouns with 6 stems

Stemtype	Examples
Base	vene (boat)	ajatus (thought)	käsi (hand)
SSg	venee/nä	ajatukse/na	käte/nä
WSg	venee/n	ajatukse/n	käde/n
SPl	vene/inä	ajatuks/ina	käs/inä
WPl	vene/issä	ajatuks/issa	käs/issä
Con	venet/tä	ajatus/ta	kät/tä

2.1.3.4 Adjectives

Stemtypes and their attributes

dstem=PVE	back
stemtype=BASE	endingv
	has2v
	basei
	vq
	syll2p3
dstem=PVE	back
stemtype=SSG	has2v
	stemc
	vq
	syll2p3
dstem=PVE	back
stemtype=WSG	syll1p3
dstem=PVE	back
stemtype=SPL	endingv
	ending2v
	has2v
	basei
	syll1p3
dstem=PVE	back
stemtype=WPL	endingv
	has2v
	syll1p3
dstem=PVE	back
stemtype=CON
dstem=CVE	back
dstem=SVE	back

Different combinations of adjective stems

Adjectives with 7 stems

Dstem	Stemtype	Examples
PVE	BASE	korkea (high)	vapaa (free)
PVE	SSG	korkea/na	vapaa/na
PVE	WSG	korkea/n	vapaa/n
PVE	SPL	korke/ina	vapa/ina
PVE	WPL	korke/issa	vapa/issa
CVE		korkea/mpi	vapaa/mpi
SVE		korke/in	vapa/in

Adjectives with 8 stems

Dstem	Stemtype	Examples
PVE	BASE	kirkas (bright)	lämmin (warm)
PVE	SSG	kirkkaa/na	lämpimä/nä
PVE	WSG	kirkkaa/n	lämpimä/n
PVE	SPL	kirkka/ina	lämpim/inä
PVE	WPL	kirkka/issa	lämpim/issä
PVE	CON	kirkas/ta	lämmin/tä
CVE		kirkkaa/mpi	lämpimä/mpi
SVE		kirkka/in	lämpim/in

2.1.3.5 Verbs

Attributes

Attributes needed for adding suffixes to verb stems

vstem	SSGVST	strong vowel stem
	WSGVST	weak vowel stem
	SSGPAST	strong vowel stem: past tense
	WSGPAST	weak vowel stem: past tense
	SSGCOND	conditional stem
	CONVST	consonant stem
	PASS	passive stem
	CONPOTN	potential stem
	SSGINF	infinitive stem
back	does the stem contain back or front vowels
endingv	is there a vowel in the end of the stem in question
has2v	is there a long vowel in the end of the vowel stem
stemc	does the verb have a consonant stem
stempotn	does the verb have a potential stem
steminf	does the verb have an infinitive stem
vq	vowel quality
cq	consonant quality

Attributes needed for linking different suffixes

gradation	weak/strong/strongpast
vqending	Vowel quality linked to the suffixes.

Stemtypes and their attributes

vstem=SSGVST	back
	endingv
	has2v
	stemc
	stempotn
	steminf
	vq
vstem=WSGVST	back
vstem=SSGPAST	back
vstem=WSGPAST	back
vstem=SSGCOND	back
vstem=SSGINF	back
vstem=PASS	back
	endingv
	has2v
	cq
vstem=CONVST	back
	cq
vstem=CONPOTN	back
	cq

Different combinations of verb stems

Verbs with 6 stems

Vstem	Examples
SSGVST	muista/vat (remember)	juo/vat (drink)
WSGVST	muista/n	juo/n
SSGPAST	muist/ivat	jo/ivat
WSGPAST	muist/in	jo/in
SSGCOND	muista/isin	jo/isin
PASS	muiste	juo

Verbs with 7 stems

Vstem	Example
SSGVST	luke/vat (read)
WSGVST	lue/n
SSGPAST	luk/ivat
WSGPAST	lu/in
SSGCOND	luk/isin
PASS	lue
SSGINF	luki

Verbs with 8 stems

Vstem	Example
SSGVST	valitse/vat (choose)
WSGVST	valitse/n
SSGPAST	valits/ivat
WSGPAST	valits/in
SSGCOND	valits/isin
CONVST	valit/koon
PASS	valit
CONPOTN	valin

2.1.3.6 Frequencies of the different combinations of stems

Number of	Category
stems	Noun (common)	Adjective (qualificative)	Verb (normal)
5	10305	-	-
6	7480	-	2204
7	-	92	187
8	-	4003	946

2.2 Syntactic layer

General

In the Finnish lexicon the main syntactic information encoded for verbs contains firstly the possible type of the object a verb can take, if any, and secondly, for intransitive verbs, the property of a verb to select a specific case and its possibility to take a clausal subject. We also have a category for intransitive verbs which can occur in existential constructions, i.e. for verbs which can take a subject in the partitive. In addition, there are some constructions for different mental and communicative verbs, which typically take a complement indicating RECIPIENT.

In addition to intransitive and transitive verbs, there is a third category for verbs in the lexicon, a category for impersonal verbs. By `impersonal' we refer to verbs, or uses of certain verbs, which lack a person contrast.

The syntactic encoding of nouns concerns the possible post-nominal complements a noun can take, i.e. different phrasal complements inflected in some local case, as well as infinitive constructions and clauses governed by a noun.

Adjectives in the lexicon are categorized according to three types of complements, which are phrasal complements inflected a) in the genitive and b) in some local cases. Thirdly there is a subgroup of adjectives which can be modified by an infinitival complement.

Prepositions and postpositions are categorized according to the form of the complement they occur with, i.e. a phrasal complement either in the genitive or in the partitive, or, in some rare cases, in some local case.

The so called ad-adjectives form a functionally motivated subcategory differentiated from adverbs. Adverbs in themselves in the lexicon are marked with information concerning e.g. polarity, the possible context they can occur in.

Syntactic functions

Syntactic functions used in the syntactic frames for verbs:

HEAD

ADVERBIAL for

NPs indicating MANNER or INSTRUMENT

CLAUSCOMP for

infinitive complements (i.e. 3rd infinitive in illative, elative or abessive, 1st infinitive, a participle or a permissive construction)

OBJECT realized as

an NP, a that-clause, a Wh-clause, a participal construction, an infinitive construction or a so called permissive construction

OBLIQUE for

phrasal complements in some local case or in essive or translative (the latter two cases are restricted here to predicative functions)

SUBJECT realized as

an NP in nominative, partitive or genitive or a clausal subject (a that-clause, a wh-clause or 1st infinitive construction)

Syntactic functions used in the syntactic frames for nouns:

HEAD

NCOMP for

phrasal complements inflected in some local case or in partitive

NCLAUSCOMP for

complements realized either as infinitive constructions or that- or Wh-clauses

Syntactic functions used in the syntactic frames for adjectives:

HEAD

SUBJECT

ACOMP for

phrasal complements inflected in some local case and for clausal complements realized as a third infinitive

SUBJPRED for

adjectives in predicative position with a clausal subject

NLEFTATTRIBUTIVE for

adjectives with no inflection. Noninflecting adjectives never display the predicative function, and their position in the syntagma is immediately before the headword and after other noun modifiers that agree with the headword in case and number.

Syntactic functions used in describing the syntactic behavior of prepositions, postpositions, ad-adjectives and adverbs:

HEAD

NPREPCOMP for

NPs depending on a preposition

NPOSTCOMP for

NPs depending on a postposition

AMODIFIER for

an ad-adjective modifying an adjective

ADVMODIFIER for

an ad-adjective modifying an adverb

ADVERBIAL for

an adverb as a sentence modifier

Our first approximation was to consider all the different forms of the phrasal post-nominal complements or modifiers (i.e. NPs inflected in different local cases) as different Syntactic units as well as the different kinds of clausal complements. The solution adopted for encoding adjectives was, however, to have one syntactic unit with different descriptions, so the encoding of nouns have been modified accordingly.

Framesets are not used in the Finnish lexicon (they are not obligatory information).

Number of Syntactic descriptions	124
Number of framesets

Category	Subcategory	Number of Units
Noun	common	17785
Noun	proper	746
Verb	normal	2941
Verb	impersonal	88
Adjective	qualificative	2931
Adjective	noninflecting	11
Adjective	pronominal	23
Adjective	ordinal	10
Adverb		569
Ad-adjective		42
Adpositon	postpositon	151
Adpositon	prepositon	44
Numeral	cardinal	24

3. Bibliography

Karlsson, Fred. 1987. FINNISH GRAMMAR. Second edition. WSOY: Juva.

Saukkonen, Pauli, Marjatta Haipus, Antero Niemikorpi & Helena Sulkala. 1979. SUOMEN KIELEN TAAJUUSSANASTO. WSOY: Porvoo.

Vilkuna, Maria. 1996. SUOMEN LAUSEOPIN PERUSTEET. Edita: Helsinki.

Dictionaries

SUOMEN KIELEN PERUSSANAKIRJA I-III. ("The Basic Dictionary of Finnish".) Kotimaisten kielten tutkimuskeskuksen julkaisuja 55. Valtion painatuskeskus: Helsinki.

Corpora

FINTAG. The Department of General Linguistics, University of Helsinki.

hs90. The Department of General Linguistics, University of Helsinki.

Appendix - An Example

Below are the relevant parts of the entries needed in building one of the approximately 6000 different forms of the Finnish noun aihe, 'subject'. The word form in question is aiheisiinsakin, 'even to his/her/their subjects':

aihe - i - sii - nsa - kin

SPL PL ILL SGPL3 CLITICPARTICLE

The order of the endings is indicated by means of the attributes linked to the element Um_Aff; e.g. 'mustbeattachedto' or 'mustbefollowedby'.

<Um_S

id="N60"

appellation="aihe"

catgram="NOUN"

sscatgram="COMMON"

autonomie="YES"

usyn_l="Usyn85"

<Umg

mf="DEFAULTMF">

<Lib>aihe</Lib>

<Radg

nieme="4"

back="YES"

stemtype="SPL"

endingv="YESEV"

ending2v="NOEN"

has2v="YESVV"

basei="NOBI"

syll1p3="NOSY1"

<Lib>aihe</Lib></Radg>

</Umg>

</Um_S>

<Um_Aff

id="NUM2"

typaff="SUFFIX"

mustbeattachedto="STEMORNONFINITE"

mustbefollowedby="CASE"

<Umg

mf="MFGEMPTY">

<Radg

nieme="5"

stemtype="SPL"

endingv="YESEV">

<Lib>i</Lib></Radg>

<MorphFeature

featurename="NUMBER"

featurevalue="PLURAL">

</MorphFeature></Umg>

</Um_Aff>

<Um_Aff

id="CASE4"

typaff="SUFFIX"

mustbeattachedto="NUMBERORNONFINITE"

<Umg

mf="MFGEMPTY">

<Radg

nieme="41"

stemtype="SPL"

endingv="YESEV"

has2v="YESVV"

px="YESPX"

px3="NOP">

<Lib>sii</Lib></Radg>

<MorphFeature

featurename="CASE"

featurevalue="ILLATIVE">

</MorphFeature></Umg>

</Um_Aff>

<Um_Aff

id="POSS5"

typaff="SUFFIX"

mustbeattachedto="CASE"

<Umg

mf="MFGEMPTY">

<Radg

nieme="1"

back="YES">

<Lib>nsa</Lib></Radg>

<MorphFeature

featurename="POSSESSOR"

featurevalue="SGPL3">

</MorphFeature></Umg>

</Um_Aff>

<Um_Aff

id="CPL2"

typaff="SUFFIX"

maybeattachedto="CASEORPOSSORPERSONORIMPERATIVE"

<Umg

mf="MFGEMPTY">

<Radg

nieme="1">

<Lib>kin</Lib></Radg>

<MorphFeature

featurename="CLITICPARTICLE"

featurevalue="YESCPL">

</MorphFeature></Umg>

</Um_Aff>