|
{ |
|
"paper_id": "W89-0112", |
|
"header": { |
|
"generated_with": "S2ORC 1.0.0", |
|
"date_generated": "2023-01-19T03:45:19.926262Z" |
|
}, |
|
"title": "Msiskinoversaettelse fra esperanto til islandsk", |
|
"authors": [ |
|
{ |
|
"first": "Stefan", |
|
"middle": [], |
|
"last": "Briem", |
|
"suffix": "", |
|
"affiliation": {}, |
|
"email": "" |
|
} |
|
], |
|
"year": "", |
|
"venue": null, |
|
"identifiers": {}, |
|
"abstract": "Machine Translation from Esperanto to Icelandic It has become evident that Esperanto, due to its logicality and regu larity, is well suited as an intermediate language in m2tchine translations between ethnic languages. The project presented here is intended to be a contribution to a later inclusion of Icelandic in a larger multilingual system for machine translation, where Esperiinto is used as an intermediate Icinguage. In fsict such a multilingual machine translation system, DLT (Distributed Language Translation), is being developed at BSO/Research in Utrecht (Netherlands). M y project started in the year 1981. It was a spare time 2ictivity exclusively on a private basis until two years ago when it received public support in Iceland. This is a machine translation system applicable as an instrument for studying various ideas relating to machine translation from Esperanto to Iceleindic. An effective machine translation system must be capable of coping with unexpected words, i.a. words created by spontaneous need in the daily use o f the language. Those new creations follow rules and conventions that can be different from one language to another. This theme, that involves automatic word creation, is discussed with respect to Esperanto and Icelandic. The derivation o f words from words o f different word classes by changing the endings o f the words, is a common feature of Esperanto and Icelandic. In cases where this happens in a similar and systematic way in both languages, automatic word creation is feasible. A noticeable charttcteristic o f Esperanto is the extensive use of affixes, both prefixes and suffixes, in order to modify the meaning of words and to create new words. In Icel2indic affixes are used less frequently. This means that when translating between Esperanto and Icelandic sometimes suffixed words in Esperanto correspond to different word constructions in Icelandic.", |
|
"pdf_parse": { |
|
"paper_id": "W89-0112", |
|
"_pdf_hash": "", |
|
"abstract": [ |
|
{ |
|
"text": "Machine Translation from Esperanto to Icelandic It has become evident that Esperanto, due to its logicality and regu larity, is well suited as an intermediate language in m2tchine translations between ethnic languages. The project presented here is intended to be a contribution to a later inclusion of Icelandic in a larger multilingual system for machine translation, where Esperiinto is used as an intermediate Icinguage. In fsict such a multilingual machine translation system, DLT (Distributed Language Translation), is being developed at BSO/Research in Utrecht (Netherlands). M y project started in the year 1981. It was a spare time 2ictivity exclusively on a private basis until two years ago when it received public support in Iceland. This is a machine translation system applicable as an instrument for studying various ideas relating to machine translation from Esperanto to Iceleindic. An effective machine translation system must be capable of coping with unexpected words, i.a. words created by spontaneous need in the daily use o f the language. Those new creations follow rules and conventions that can be different from one language to another. This theme, that involves automatic word creation, is discussed with respect to Esperanto and Icelandic. The derivation o f words from words o f different word classes by changing the endings o f the words, is a common feature of Esperanto and Icelandic. In cases where this happens in a similar and systematic way in both languages, automatic word creation is feasible. A noticeable charttcteristic o f Esperanto is the extensive use of affixes, both prefixes and suffixes, in order to modify the meaning of words and to create new words. In Icel2indic affixes are used less frequently. This means that when translating between Esperanto and Icelandic sometimes suffixed words in Esperanto correspond to different word constructions in Icelandic.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Abstract", |
|
"sec_num": null |
|
} |
|
], |
|
"body_text": [ |
|
{ |
|
"text": "I d e s id ste o t t e \u00e5 r h a r je g b e sk ae ftig e t m ig m ed m a sk in ov ersae ttelse fra esp e ra n to til isla n d s k , g a n sk e v ist ikke er v e le g n e t til d a ta lin g v is tis k b ru g . D e t er n e to p d en n e id \u00e9, s o m je g har slu ttet m ig til, d a je g h a r v a lg t a t b e sk ae ftig e m ig m e d m a sk in ov ersae ttelse fr a esp era n to til isla n d sk . G ru n d e n til a t je g h id til h ar b eg rae n set m ig til ov e rsae tte lse i d en n e re tn in g , m en ikke fr a isla n d sk til e s p e ra n to , er sim p elth en d en a t p a rsn in g a f e s p e r a n to er m eg e t n e m m e re e n d p a rsn in g a f a n d re sp r o g ", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Indledning", |
|
"sec_num": "1" |
|
}, |
|
{ |
|
"text": "J e g v il nu v ise e t p a r e k sem p ler p \u00e5 a u to m a tisk ord d a n n e lse v e d a fled n in g fra a n d r e ord k la sser.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": ".1 A fledning fra en 2inden ordklasse", |
|
"sec_num": "4" |
|
}, |
|
{ |
|
"text": "I e s p e r a n to er d e n n e ty p e a f o rd d a n n e ls e m e g e t a ", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": ".1 A fledning fra en 2inden ordklasse", |
|
"sec_num": "4" |
|
}, |
|
{ |
|
"text": "E n a n den ty p e a f o r d d a n n e ls e r p \u00e5 e s p e r a n to g \u00e5 r u ", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "E speranto-ord m ed affikser", |
|
"sec_num": "4.2" |
|
} |
|
], |
|
"back_matter": [ |
|
{ |
|
"text": "Reykjavik.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "annex", |
|
"sec_num": null |
|
}, |
|
{ |
|
"text": "Proceedings of NODALIDA 1989", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "145", |
|
"sec_num": null |
|
} |
|
], |
|
"bib_entries": {}, |
|
"ref_entries": { |
|
"FIGREF0": { |
|
"type_str": "figure", |
|
"uris": null, |
|
"text": "u a fb ru d t. D e f\u00f8 r s te 6 \u00e5 r var d et u d elu k ken d e 138 Proceedings of NODALIDA 1989, pages 138-145 Stefan Briem: Maskinoversaettelse fra esperanto til islandsk 139 & itid sb esk ae ftigelse h elt p \u00e5 e g n e v eg n e. M e n i d en s id ste tid h a r m it p r o je k t m \u00f8 d t sta d ig v o k se n d e in teresse o g h ar f\u00e5 et o ffe n tlig s t\u00f8 t te . J eg v il b e g y n d e m e d a t p la ce re m it p r o je k t i s t\u00f8 r r e sa m m en h ae n g . H vis m an b e tr a g te r m a sk in o v ersae ttelse p a rv is m ellem m a n g e s p r o g s \u00e5 k u n n e d en tae nk es a t fo r e g \u00e5 d ire k te m ellem h v ert s p ro g p a r. E n amden m u lig h ed er a t b ru g e et fae lles m e lle m s p ro g , h v o r f.ek s. e s p e r a n to tje n er so m m e llem sp rog . V a lg et a f e s p e ra n to s o m m e lle m s p ro g er ik ke tilfae ld ig t, m en b a seres p \u00e5 , at e sp e ra n to er e t sae rd eles lo g isk s p r o g m e d reg e lm ae ssig g ra m m a tik , h v o r fo r d et 139 Proceedings of NODALIDA 1989 140 Computational Linguistics -Reykjavik 1989", |
|
"num": null |
|
}, |
|
"FIGREF1": { |
|
"type_str": "figure", |
|
"uris": null, |
|
"text": ", o g d er er j o alligevel p r o b le m e r n o k a t g lae d e sig o v er. M e d h en b lik p \u00e5 d en n e g ru n d lae g g e n d e id \u00e9 g \u00e5 r m it p r o je k t i ta k t m ed det k e n d te D L T -p r o je k t fo r m u ltis p ro g lig m a sk in ov ersae ttelse, h v o r m an o g s \u00e5 har v a lg t e s p e r a n to s o m m e lle m sp ro g . I tid e n s l\u00f8 b h a r je g u d v ik le t et o v ersae ttelsesp rog ra m sa m m en m ed en o r d sa m lin g p \u00e5 g o d t 6 .0 0 0 o r d . P r o g r a m m e t er i sta n d til a t g e n g iv e esp era n to -tek st p \u00e5 isla n d sk , s\u00e5 le d es a t d e isla n dsk e o r d h ar d en \u00f8n sk ed e b \u00f8 jn in g s fo r m o g sae tn in g ern es stru k tu r er ae n d re t i re tn in g a f k orrek t islan dsk sy n ta k s. J eg h a r fu n d e t d e t n y ttig t a t h a v e et o v e rsae tte lsesp rog ra m , s o m fu n gerer. S elv o m p r o g r a m m e t m e g e t lid t b e r \u00f8 r e r sem a n tik s\u00e5 kan d et b ru g e s til at stu d ere fo rsk e llig e p r o b le m e r v e d m a sk in ov ersae ttelse, isae r sy n ta k tisk e o g m orfo lo g isk e p r o b le m e r . M a n kan e k sp erim en tere m e d id \u00e9er o g m u lig e l\u00f8 sn in g e r til disse p r o b le m e r v e d a t ae n d re lid t p \u00e5 p ro g ra m o g ord sa m lin g o g , h vad d e r kan vae re m e g e t o p m u n tr e n d e , h u rtig t se et resu lta t p \u00e5 d a ta sk ae rm eller p \u00e5 p a p ir.", |
|
"num": null |
|
}, |
|
"FIGREF2": { |
|
"type_str": "figure", |
|
"uris": null, |
|
"text": "r d a n fu n g e re r o v e r sae tte lse sp ro g ra m m e t? v e rs ae tte ls e o r d fo r o r d 3. S tru k tu rae n d rin g e r P r o g r a m m e t a r b e jd e r m e d \u00e9n sae tn in g a d g a n g e n o g g \u00e5 r fra d e t en kle til d e t m e re k o m p lic e r e d e . I n d d a ta e r a lts \u00e5 en sae tn in g p \u00e5 e sp e ra n to . P ro g ra m m e ts f\u00f8 r s te o p e r a tio n e r a t d e le sae tn in g en o p i en k elte o r d . D e n a n d en o p e r a tio n er at s l\u00e5 o p i o r d r e g is tr e s o m g iv e r fo r h v ert en k elt e s p e r a n to -te k s to r d d et tilsvaren de 140 Proceedings of NODALIDA 1989 Stefan Briem: Maskinoversaettelse fra esperanto til islandsk 141 islan dske o r d sa m m e n m e d m o rfo lo g is k e o p ly s n in g e r. D e n tr e d je o p e r a tio n b e s t\u00e5 r a f stru k tu rae n d rin g er i h e n h o ld til forsk ellen p \u00e5 e s p e r a n to o g isla n d sk . T il slu t afleverer p ro g ra m m e t s o m u d d a t a en isla n d sk s ae tn in g m e d n ae ste n isla n d sk syn tak s o g fo r h \u00e5 b e n tlig n o g e n lu n d e sa m m e m en in g s o m d en o p r in d e lig e sae tn in g p \u00e5 e sp e ra n to har. D e t, s o m je g isae r v il d is k u te r e v e d d e n n e le jlig h e d , er h v ilk e u d v e je m a n har n \u00e5r et e s p e r a n to -o r d ikke fin d es d ire k te i o r d s a m lin g e n . L \u00f8 s n in g e n a f d e tte p r o b le m in d e b ae re r a u to m a tis k ord d a n n e lse . 3 Ordsamlingens inddeling P \u00e5 g ru n d a f e sp e ra n to s s im p le g r a m m a tik afd ae k k es et te k s to rd s o rd k la s s e o g b \u00f8 jn in g s fo r m p \u00e5 g ru n d la g a f o r d e ts e n d else alen e, u a fh ae n g ig t a f d e a n d r e o r d i sa m m e sae tn in g . J eg h a r d e r fo r fu n d e t d e t n y ttig t a t d e le o rd sa m lin g e n o p i fe m o rd re g istre s o m vist her. o n o m in e r , p r ae p o s itio n e r . k o n ju n k tio n e r, a d v e rb ie r. n u m era lier o .fl.) R e g is te r n u m m er 5 in d e h o ld e r en lu k k et klasse a f ca . 2 0 0 o r d . D e a n d re re g istre in d e h o ld e r \u00e5 b n e ord k la sser. H v is et e s p e r a n to -o r d ik k e fin d es i re g ister 5 s\u00e5 tilh \u00f8 re r d e t m ed sik kerh ed \u00e9n b e s te m t a f d e fire a n d re reg istres o rd k la sse r. S\u00e5& em t o r d e t e r til s te d e i ord sa m lin g e n , fin d e r p r o g r a m m e t h u rtig t fre m til d e t. M en h va d h vis o r d e t allig ev el ikke fin d e s ? O g d et v il m a n j o ikk e k u n n e u n d g \u00e5 i tilfae ld e a f n y d a n n e lser. 4 Automatisk orddannelse M a n ku n n e se lv f\u00f8 lg e lig sim p e lth e n g iv e o p o g la d e v ae r e m ed a t o v e r s ae tte o r d e t. I tilfae ld e a f e s p e ra n to k u n n e m an a llig ev el h o ld e red e p \u00e5 sy n ta k sen , id e t o r d e ts ord k la sse o g b \u00f8 jn in g s fo r m e r k en dt fr a en d elsen , selv o m sta m m e n m \u00e5 sk e er u k en d t. M e n d e r er o g s \u00e5 d en m u lig h ed a t o r d e ts s ta m m e eller b e s ta n d d e le fa k tisk er til ste d e selv o m o r d e t ikke fin d es d ire k te i reg istre t fo r d e n r ig tig e o rd k la sse. M a n kan d ele d e u k e n d te e s p e r a n to -o r d i fe m tilfae ld e: 1. S a m m en sa tte o r d 2. O r d a fle d te & a a n d re ord k la sse r isse tilfae ld e h a r je g isae r ek sp e rim e n te re t m ed 2 til 4 o g i d en an led n in g in k lu d e re t n o g e n ord d a n n e lse sre g le r i m it o v e rsae ttelse sp ro g ra m . L a d o s f\u00f8 r s t b e tr a g te d e a n d re tilfae ld e . S a m m e n s a tte ord er m in d re a lm in d e lig e i e sp e ra n to e n d i islan dsk. I d e fleste tilfae ld e kan m a n n o k a u to m a tis k o v e r s ae tte et sa m m en sa t o r d d ire k te fr a esp er a n to til isla n d sk , d .v .s . m a n fin d er fre m til d e en k elte s ta m m e rs oversae ttelse r o g d^lnner a f d e m et sa m m e n sa t o r d p \u00e5 islan dsk . H v a d 5. tilfae ld e a n g \u00e5 r, an dre ord , s\u00e5 d rejer d et sig o m o r d s o m ikke kan o v e rs ae tte s i d e o m g iv e ls e r. M a n kan sim p e lth e n g en g iv e o r d e t u d en o v ersae ttelse i d e n isla n d sk e tek st. M a n kan j o allig ev el b e n y tte sig a f k en d sk a b e t til o rd ets o rd k la s s e o g b \u00f8 jn in g s fo r m til a t b e s te m m e den islan dske sae tn in gs stru k tu r.", |
|
"num": null |
|
}, |
|
"FIGREF3": { |
|
"type_str": "figure", |
|
"uris": null, |
|
"text": "lm in d e lig , m eg e t m e re en d i isla n d sk o g d^lnsk f.ek s. E k s e m p e l 1 . a d v e r b iu m d a n n e t a f a d je k tiv a d je k tiv : fe lic a h a m in g ju sa m u r ly k k elig a d v e r b iu m ; fe lic e h a m in g ju s a m le g a ly k k e lig t N \u00e5 r p r o g r a m m e t skal o v e r s ae tte a d v e rb ie t fe l ic e til isla n dsk, s\u00e5 v il d e tte ord ik k e fin d e s i sa m lin g en a f a d v e rb ie r. P r o g r a m m e t v il d a s \u00f8 g e i sa m lin gen a f a d je k tiv e r fo r a d je k tiv e t fe lic a , s o m j o fin d es d er. R e s u lta te t b liv er d e t islan dske a d v e r b iu m h a m in gju sa m lega a fle d t a f d e t isla n dsk e a d je k tiv h am in gju sam u r. D e t a n d e t e k s e m p e l d re je r sig o m et a d je k tiv , eller sn arere o m et p a rticip iu m , s o m b e h a n d le s s o m et a d je k tiv . E k s e m p e l 2 . a d je k tiv (p a r tic ip iu m ) d a n n e t a f v e rb u e t ku ra n ta fin d es ik k e i sa m lin g en a f a d je k tiv e r. D a o r d e t en d e r p \u00e5 anta er d e t m u lig v is p rae sen s p a r tic ip iu m a f et v e r b u m . P r o g r a m m e t v il d e r fo r s \u00f8 g e i sa m lin g e n a f v e r b e r fo r v e r b e t k u ri. R e s u lta te t er p o s it iv t o g p r o g ra m m e t d an n er a u to m a tis k o r d e t hlaupandi v e d a fled n in g a f d e t isleindske v e rb u m hlaupa. D e t o e k se m p ler kan o g s \u00e5 k o m b in eres o g b liv e til en d o b b e lta fie d n in g . 142 Proceedings of NODALIDA 1989 Stefan Briem: Maskinoversaettelse fra esperanto til islandsk 143 E k s e m p e l 2 + 1 . a d v e rb iu m d a n n e t a f v e rb u m a d v e rb iu m : k u ra n te h la u p a n d i l\u00f8 b e n d e A d v e r b ie t k u ra n te ov e rsae tte s v e d a t p ro g r a m m e t f\u00f8 r s t s \u00f8 g e r fo rg ae v e s efte r a d jek tiv et ku ranta o g d e re fte r fin d e r fre m til v e r b e t kuri.", |
|
"num": null |
|
}, |
|
"FIGREF4": { |
|
"type_str": "figure", |
|
"uris": null, |
|
"text": "d p \u00e5 u d stra k t b r u g a f af& kser, b \u00e5 d e p rae fik ser o g su ffik ser. N \u00e5 r e s p e r a n to -o r d m e d affik ser skal k u n n e ov e rsae tte s til isla n d sk , kan d e m est a lm in d e lig e m ed fo r d e l in k lu d eres i o rd sa m lin g ern e. F or d e a n d re a ffik sed e o r d s v e d k o m m e n d e kan m a n p r \u00f8 v e a t v ae lg e en a f d e ov e rsae tte lse sm e to d e r s o m a n ty d es n e d e n for. e s p e r a n t o i s l a n d s f\u00f8 lg e n d e o r d fo rk la re n d e k o n stru k tio n B \u00e5 d e i tilfae ld e a f et p rae fik s o g et su ffik s k o m m e r alle d isse m u lig h e d e r i b e t r a g t n in g. M a n kan v ise ek sem p ler p \u00e5 alle m u lig h e d e rn e , m \u00e5 sk e m e d \u00e9n u n d ta g e lse. O m lid t v il v i se n o g le ek sem p ler. M en f\u00f8 rs t skal v i se p \u00e5 en sae rp rae g et b r u g a f p rae fik se r p \u00e5 e s p e ra n to . D e t d reje r sig o m k on stru k tio n stv illin g e r, h v o r en p r ae p o s itio n kan p la ce re s so m p rae fik s p \u00e5 et v e rb u m m e d in gen eller m in im a l ae n d rin g a f m en in g . 4 .2 . 1 S t r u k t u r ae n d r i n g p \u00e5 e s p e r a n t o e s p e ra n to : M i p a ro la s p r i la lib ro . isla n dsk: E g ta la u m b ok in a . d a n sk : J eg ta ler o m b o g e n . D en sa m m e m en in g kan y tre s v e d sae tn in ge n : e sp e ra n to : M i p r ip a r o la s la lib r o n . 143 Proceedings of NODALIDA 1989", |
|
"num": null |
|
} |
|
} |
|
} |
|
} |