The Spanish Morphology in Internet
Departamento de Informática
y Sistemas,
Universidad de Las Palmas
de Gran Canaria,
Campus Universitario de
Tafira,
35017 Las Palmas de Gran
Canaria, Spain
Abstract. This Web service tags morpholexically any Spanish word and it gets the corresponding forms starting from a canonical form and from the flexion asked for. In the verbs, it deals with the simple and compound conjugation, the enclitic pronouns, the flexion of the participle like verbal adjective and the diminutive of the gerund. With the nonverbal forms, this web service considers: gender and number, heteronomy for change of sex, superlative degree, adverbiation and the appreciative derivation. In the tag and in the generation the prefixation is taken into account. It allows the manipulation of morpholexical relationships. It offers a global vision of the behavior and productivity of the Spanish words in the principal processes of formation (sufixation, prefixation, parasinthesis, suppression, regression, zero‑modification, apocopation, metathesis and others which are unclassifiable and that generate alternative graphical forms). It includes the principal Spanish lexicographic repertoires. It considers 151103 canonical forms that produce more than 4900000 flexioned and derived forms and about 90000 morpholexical relationships are established.
1 Introduction
As result of the work done by the Group of Data Structures and Computacional Linguistic of the University of Las Palmas de Gran Canaria (http://www.gedlc.ulpgc.es), a Web application has been created able to interpret and to handle with versatility different relevant aspects of the Spanish morphology. The application represents more a form to show the potentiality of a Web System of management of the morphology of the Spanish language than a finalist tool. This system allows its integration in other useful tools for the natural language processing like orthographic correction, advanced search information, text analysers, disambiguousers, lexicographical station, parsers, information extraction, text automatic generation, syntactic correction and extraction of summaries, among other applications, all of them aimed at offering interactive services distributed through Internet.
The automatized treatment of the Spanish morphology arouses great interest because it constitutes the first touchstone on which to construct any natural language processor and starts the way towards future Web services more specialized in the handling, learning and control of that great human potential –the language.
The available system in the Web tags any Spanish word identifying its canonical form, grammar category and the flexion or derivation that produces it, and is able to produce the corresponding forms from a canonical form and from the flexion or derivation asked for; both the recognition and the generation operate on a same data structure, to cross it in opposite senses implies that the tool works in one or another modality.
In the verbs, it deals with the simple and compound conjugation, the enclitic pronouns, the flexion of the participle like verbal adjective (gender and number) and the diminutive of the gerund. With the nonverbal forms, it considers: the gender and the number in the nouns, adjectives, pronouns and articles; heteronomy by sex change in the nouns; the superlative degree in the adjectives and adverbs; the adverbiation and the superlative adverbiation in the adjectives; the appreciative derivation in the nouns, adjectives and adverbs; the graphical variants in all the grammar categories and the invariant forms such as preposition, conjunctions, exclamations, words of other languages and locutions or phrases. As much in the tag as in the generation the incorporation of prefixes is discretionarily considered.
In addition it allows the recognition, the generation and the manipulation of the morpholexical relations of any word, it includes the recovery of all its lexicogenetic information until arriving at a primitive one, the management and control of the affix in the treatment of its relations, as well as the regularity in the established relation. It provides a global vision of the behavior and productivity of the Spanish words in the main processes of formation (sufixation, prefixation, parasinthesis, suppression, regression, modification‑zero, apocopation, metathesis and nonclassifiable others that generate alternative graphical forms).
For the accomplishment of
this work, a corpus of Spanish words with the lexicon of different dictionaries
has been created: the Diccionario de
2 The Tagger
This service (http://www.gedlc.ulpgc.es/investigacion/scogeme02/lematiza.htm) allows the user to tag any Spanish word ―it identifies its canonical form, grammar category and the flexion, derivation and prefix that affects to it. The tag is the first advisable step for the users of the system: on the one hand, it makes the entry flexible when allowing to have access to the rest of the available services without previously having neither morphologic nor grammar knowledge and, on the other hand, it identifies without ambiguity the different morphologic interpretations of the inserted form. Thus, for example, the answer for the entry habiéndose sobreaterrado is:
Fig. 1. Example of tagger
Next to each canonical form two links are provided that allow to flexion it and to obtain their morpholexical relations. These links lead in an intelligent way to the suitable services: for the verbs, the conjugater, and for the nonverbal forms the possible flexions are supplied according to the grammar category of the canonical form associated ―the prefixation is also transferred. Additionally, in the verbs, semantic information useful in several contexts is shown and this can help to identify better the infinitive and, consequently, the entry.
The abstraction that this system offers on the morphologic irregularities of the Spanish language makes its use easier and, in no case, diminishes its precision. Thus, for example, for the entries with irregular flexions cuélguemelo, bendigo, óyelo, fue, habiéndole dicho, nazca, degüellan, muévete, riño, tengo and vi, the corresponding infinitives are provided: colgar, bendecir, oír, ir or ser, decir, nacer, degollar, mover, reñir, tener and ver; either to identify the forms ―irrespective of the existence of enclitic pronouns in the words―, or to have access later to the conjugation service Flexioner: verb, with which all previous morphologic knowledge of the language on the part of the user is eluded. And in the nonverbal examples, also with irregular flexion: paupérrimo, sapientísimo, bestezuelo, blanquecino, boyazo, ovecico, hijosdalgo, cualesquiera, robaliza, flámines, princesa, the service provides all the canonical forms from which these words can come from with the corresponding information of the flexion applied in each case: pobre, sabio, bestia, blanco, buey, huevo, hijodalgo, cualquiera, róbalo or robalo, flamen and príncipe. This allows to accede to the suitable service Flexioner.
Special mention deserves the incorporation of enclitic pronouns to the verbal forms. The system admits the appearance of up to three pronouns for a verbal form ―either simple or compound―, thus comerme, habérselo comido or comiéndosenosla lead us to the infinitive comer; and, of course, the service takes into account the necessary rules of union, not only as far as the incorporation of accents is concerned, but also, as far as the modificationns in original forms is concerned ―temeos is temed with the enclitic pronoun os and amémonos is amemos with the enclitic pronoun nos.
Another important flexibility is the recognition of the existence of prefixes, discretionarily incorporated in the form to tag. It avoids, therefore, knowledge in this area ―as to which prefixes exist in Spanish and which the rules of union and their irregularities are― for the use of the system if on starts from words such as: predigestión, reatacar, irrealización, coadmitido, etc.
3 The verbal flexioner.
This service (http://www.gedlc.ulpgc.es/investigacion/scogeme02/flexver.htm) allows the user to obtain, from a verb, the conjugation of a simple or composed tense, nonpersonal simple or compound forms, the flexion of the participle as a verbal adjective and the diminutive of the gerund. In addition, it is possible to select from three pull-down lists the valid combinations of one, two or three enclitic pronouns, so that the system incorporates them correctly to the required conjugated forms. Thus, for example, the past perfect of indicative of the infinitive temer with the enclitic pronoun le is:
Fig. 2. Example of verbal flexioner
Like it happens with the tagger, in the generation of flexioned forms, the irregularities that appear in the conjugation models are considered, in addition to all the changes in the graphical form as a result of the incorporation of the pronouns to the conjugated forms. Thus, for example, the indicative present of the infinitive oler with the enclitic pronouns se and lo is presented in Table 1.
Table 1. For input oler
|
The indicative present of oler with se and lo |
|
|
First person of the singular |
huéloselo |
|
Second person of the singular |
huéleselo |
|
Third person of the singular |
huéleselo |
|
First person of the plural |
olémoselo |
|
Second person of the plural |
oléiselo |
|
Third person of the plural |
huélenselo |
It is possible to add up to three prefixes discretionarily to the conjugated forms, selecting them of three pull‑down lists alphabetically organized. The system considers for each prefix the general rules of application in this word formation process and the specific ones. Thus, for example, if the prefix is gotten up re‑ to the first and third person of the singular of the indefinite past of the verb hacer ―hice and hizo― rehíce and rehízo are obtained; and the simple nonpersonal forms for emitir with the prefixes trans‑, re‑ and co‑ are: corretransmitir, corretransmitiendo and corretransmitido.
Since any Spanish word is admitted like input to the conjugater, all the requests are previously treated by the tagger in order to apply the flexions on the infinitives from which the inputs come. If the input is a infinitive is conjugated, if is a conjugated form its infinitivo is flexioned and if the input does not come from any verb is the result of its tag with intelligent links to the flexion pages. In all the cases it can be implied more than one canonical form. Thus, for example, if it is the present of indicative asked for aterrar, fue and óptimas, the different answers are presented in Table 2, Table 3 and Fig. 3.
Table 2. For input aterrar
|
Of the infinitive aterrar (transitive pronominal intransitive) |
|
|
First person of the singular |
atierro |
|
Second person of the singular |
atierras |
|
Third person of the singular |
atierra |
|
First person of the plural |
aterramos |
|
Second person of the plural |
aterráis |
|
Third person of the plural |
atierran |
|
Of the infinitive aterrar (transitive pronominal intransitive) |
|
|
First person of the singular |
aterro |
|
Second person of the singular |
aterras |
|
Third person of the singular |
aterra |
|
First person of the plural |
aterramos |
|
Second person of the plural |
aterráis |
|
Third person of the plural |
aterran |
Table 3. For input fue
|
Of the infinitive ser |
|
|
First person of the singular |
soy |
|
Second person of the singular |
eres |
|
Third person of the singular |
es |
|
First person of the plural |
somos |
|
Second person of the plural |
sois |
|
Third person of the plural |
son |
|
Of the infinitive ir |
|
|
First person of the singular |
voy |
|
Second person of the singular |
vas |
|
Third person of the singular |
va |
|
First person of the plural |
vamos |
|
Second person of the plural |
vais |
|
Third person of the plural |
Van |
Fig. 3. For input óptimas
4 The nominal flexioner
This service allows the user to obtain, from a noun, the changes of gender and number and the appreciative derivation for the selected gender and number. Thus, for example, the diminutives in masculine plural of pez are: pececitos, pececillos, pececicos, pececines and pecezuelos.
In the generation of flexionned forms, those irregularities that appear in the formation of feminine and plural are considered, apart from a great variety of irregularities that the appreciative suffixation in Spanish has. Someone outstanding examples are: the feminine of rey is reina, the feminines of actor are actora or actriz, the plural of régimen is regímenes, the plurals of maniquí are maniquíes or maniquís, velón exists as augmentative masculine of the feminine noun vela and the irregular appreciative forms of chico are chicorrotín, chiquitín, chicorrotico, chicorrotillo, chicorrotito, chiquirritín, chiquilín, chiquirritico, chiquirritillo and chiquirritito. If one asks for the plural feminine form of toro, the service indicates that it does not admit it and offers vacas like an option ―heteronomy.
Like in the verbal conjugator, it is also possible to add prefixes to the flexioned forms. Thus, for example, if the preffix anti‑ is joined to the form son, we obtain antisón; if we join nutación to the preffix circun‑ we read circumnutación; and if the prefix exo‑ is joined to the form ósmosis it generates exosmosis and exósmosis.
Since the flexioner admits in any Spanish word, the requests are previously examined by the tagger in order to apply the corresponding flexions on the canonical forms, just like the conjugater does. If the entry is a nominal canonical form, it is flexioned; if it is a flexioned nominal form, the canonical form from which it comes is flexioned; and if the entry does not come from any nominal form, the result of its tag is shown with intelligent links to the suitable flexion pages. Thus, for example, if the masculine plural of libro, barcucho and bebió are asked for, the different answers are: libros, barcos ―from the canonical form barco― and bebió is not recognized as a noun ―the system offers to conjugate the verb beber.
5 The adjectival flexioner
This service (http://www.gedlc.ulpgc.es/investigacion/scogeme02/flexadj.htm) is similar to the one for nouns, although it adds the specific flexions of the adjectives: the superlative degree, the adverbiation and the superlative adverbiation. In addition to the formation rules, the existing irregularities for these flexions are considered. Thus, for example, the superlative in masculine plural of blanco is blanquísimo, the adverbiation of pobre is pobremente and the adverbiation of the superlative is pobrísimamente and paupérrimamente.
The treatment of prefixes and the flexibility in the processing of the entries are completely analogous to the one offered by the Flexioner: noum service.
6 The flexioner of other forms
This service (http://www.gedlc.ulpgc.es/investigacion/scogeme02/flexotra.htm) allows the user to obtain, from pronouns, articles, adverbs, prepositions, conjunctions, exclamations, words and expressions from other languages, changes in gender and number and appreciative forms when possible. In addition to the formation rules, the existing irregularities for these flexions are considered. Thus, for example, the neutral singular of the personal pronoun él is ello, the feminine plural of the article el is las and the diminutive forms of the adverb cerca are cerquita and cerquininga.
The treatment of prefixes and the flexibility in the processing of the entry are completely analogous to those offered by the previous services, although in these forms the incorporation of prefixes is not admitted.
7 The morpholexical relations
The primary target of this service (http://www.gedlc.ulpgc.es/investigacion/scogeme02/relmorfo.htm) consists of obtaining a set of morpholexical relations between useful Spanish words for applications of processing of the natural language. In a synchronous study, and with the glance put in the automatic treatment of the morphology with computer science means, the formal or theoretical aspects do not have to agree with the strictly linguistic ones. There exist Spanish words that maintain a strong semantic and functional relation ―the same that appears in the derivative or prefix level―, and that cannot take derivation or prefixation, although yes a formal relation through other stages in the evolution of the languages exists, that is why it is considered necessary to include them ―agua with acuoso, vejiga with vesical, conejo with cunicular. This concept must be restricted to avoid to arrive at the concept of related notion ―blanco with albura, sólido with endurecer, niño with pueril―, that is why a criterion of historic‑etimologic confluence is applied. It is obvious that, for the speaker, and therefore for computer science acuario, portuario and campanario must be places equally related to agua, puerto and campana. So, it is necessary to pass to another level beyond morphology level, in order to resolve linguistic barriers that they would prevent to treat relations beyond the derivation or the prefixation; the concept of morpholexical relation is extended in this way.
Fig. 4. Example for morpholexical relations
Derivative forms by grammar category ―nouns, adjectives, verbs, adverbs and other categories― can be obtained from a canonical form. Thus, for example, the nouns derived from peso are pesero, peseta and pesillo. Also it is possible to ask for the primitive form and the words that have undergone some process of formation from the entry ―suffixal, preffixal, parasinthetic and less frequent others like suppression, regression, zero‑modification, apocope, metathesis, aphaeresis and some nonclassifiable ones that generate alternative spellings. Thus, for example, the primitive form of cartero is carta and the formed ones from tubo are: tuba, tubería, tubular, tubuloso, multitubo, entubar and intubar ―they are showed according to the applied process of formation. All the requests allow filtering the words by its process of formation ―regular or irregular.
There is an option that facilitates to obtain the set of words morpholexically closer to the given one ―its primitive, its derivatives and the derivatives from its primitive. Another option serves to obtain the complete family of words with any morpholexical relation. These two possibilities allow in addition their graphical presentation, Fig. 4.
Due to any Spanish word is admited as input by the morpholexical relations service, the request is previously treated by the tagger in order to apply the adequate relations on the canonical forms. If the entry comes from more than one canonical form with relations, the requested relations for each canonical form are obtained.
References
1. Alsina, R.: Todos
los Verbos Castellanos Conjugados. 17th edn. Teide, Barcelona (1990)
2. Alvar Ezquerra,
M.: La formación de palabras en español. 5th edn. Arco/Libros, Madrid (2002)
3. Alvar Ezquerra,
M.: Nuevo diccionario de voces de uso actual. Arco/Libros, Madrid (2003)
4. Carreras, F.:
Sistema Computacional de Gestión Morfológica del Español (SCOGEME). Ph. Degree
Thesis directed by O. Santana and J. Pérez, Universidad de Las Palmas de Gran
Ca-naria (2002)
5. Casares, J.:
Diccionario Ideológico de
6. Corripio Pérez,
F.: Diccionario Práctico. Incorrecciones: Dudas y Norma Gramatical. Larousse
Planeta, Barcelona (1995)
7. Diccionario de
8. Diccionario de Uso
del Español Actual. Clave. Electronic edition, SM, Madrid (1997)
9. Diccionario de Uso
del Español de María Moliner. 2nd edn. Electronic edn., Gredos, Madrid (2001)
10. Diccionario
General de
11. Gran Diccionario
de
12. Gran Diccionario
de Sinónimos y Antónimos. 4th edn. Espasa Calpe, Madrid (1991)
13. Gómez Torrego,
L.: Manual de Español Correcto, 10th edn. Arco/Libros, Madrid, (2000)
14. Real Academia
Española: Esbozo de una nueva gramática de la lengua española. 1st edn. Espa-sa
Calpe, Madrid (1989)
15. Santana, O.,
Carreras, F., Hernández, Z., Pérez, J., Rodríguez, G.: Manual de la conjugación
del español. 12790 verbos conjugados. Arco/Libros, Madrid (2002)
16. Santana, O.,
Pérez, J., Carreras, F., Duque, J., Hernández, Z., Rodríguez, G.: FLANOM:
Flexionador y lematizador automático de formas nominales. Lingüística Española
Actual, Vol. XXI-2, Arco/Libros, (1999) 253–297
17. Santana, O.,
Pérez, J., Hernández, Z., Carreras, F., Rodríguez, G.: FLAVER: Flexionador y
lematizador automático de formas verbales. Lingüística Española Actual, Vol.
XIX-2, Arco/Libros, Madrid (1997) 229–282
18. Santana, O.;
Pérez, J.; Losada, L.: Generación automática de respuestas en análisis morfológico.
Estudios de lingüística, Universidad de Alicante, Vol. 14, (2000) 245–257
19. Seco, M.:
Diccionario de dudas y dificultades de la lengua española. 9th edn. Espasa
Calpe, Madrid (1991)