Translators’ Resource Dominance and the Success of Finding the Target Terms in Human Translation and Post-editing of Machine Translation

136 T ansla o s’ Resou ce Dominance and he Success o Finding he Ta ge Te ms in Human T ansla ion and Pos -edi ing o Machine T ansla ion Esz e Se mann Má a Lesznyák doi.o g/10.35321/ e m31-07 T ansla o s’ Resou ce Dominance and he Success o Finding he Ta ge Te ms in Human T ansla ion and Pos -edi ing o Machine T ansla ion Ve ėjų iš eklių domina imas i sėkminga e imo kalbos e minų paieška e čian žmogui a ba pos edaguojan mašininį e imą ESZTER SERMANN Depa men o I alian Language and Li e a u e, Uni e si y o Szeged, Hunga y ORCID id: h ps://o cid.o g/0000-0001-8753-5312 MÁRTA LESZNYÁK Ins i u e o English and Ame ican S udies, Uni e si y o Szeged, Hunga y ORCID id: h ps://o cid.o g/0000-0002-9354-0633 ABSTRACT This s udy p esen s he i s esul s o p ocess-o ien ed esea ch on he ypes o on- line ansla ion esou ces used by i s - and second-yea ansla ion ainees when ansla ing and pos -edi ing a legal ex om English in o Hunga ian. Based on he sc een eco dings o he s uden s’ wo k low, he possible ela ions be ween esou ce dominance ( e mino-lexicog aphic o ex -based), ime on ask and he success o inding he co ec a ge e ms we e analysed. Ou esul s indica e ha s uden s gene ally p e e e mino-lexicog aphic sou ces o ex -based sou ces. In e es ingly, in mos cases, he success o inding he co ec a ge e ms showed no signi ican co ela ions ei he wi h ime on ask o wi h esou ce dominance. The only excep- ion was he pos -edi o g oup, whe e he e was a signi ican co ela ion be ween he equency o using ex -based sou ces and he success o inding co ec e ms. In addi ion, e idence was ound ha pos -edi o s wo ked mo e e icien ly han om-sc a ch human ansla o s in e ms o ime and sea ch e o . The pape ends wi h possible explana ions o he indings and sugges ions o ansla o aining. KEYWORDS: ansla ion, pos -edi ing, ansla ion p ocess esea ch, ansla ion esou ces, adequacy o e minology. 137Te minologija | 2024 | 31 ANOTACIJA Šiame y ime pa eikiami pi mieji į e imo p ocesą o ien uo o y imo apie in- e ne inių e imo iš eklių ipus, ku iuos naudoja pi mo i an o ku so e ėjai p ak ikan ai, e sdami eisinį eks ą iš anglų kalbos į eng ų kalbą i jį pos - edaguodami, ezul a ai. Remian is s uden ų da bo eigos ek ano į ašais, bu o analizuojami galimi yšiai a p iš eklių ( e minog a inių i leksikog a inių (angl. e mino-lexicog aphic) a eks inių) domina imo, užduočiai ski o laiko i sėkmin- go eisingų e imo kalbos e minų adimo. Mūsų ezul a ai odo, kad s uden- ai dažniausiai eikia pi menybę e minog a iniams i leksikog a iniams, o ne eks iniams šal iniams. Įdomu ai, kad daugeliu a ejų eisingų e imo kalbos e minų sėkmingas su adimas nepa odė eikšmingos ko eliacijos nei su užduočiai ski u laiku, nei su iš eklių domina imu. Vienin elė išim is bu o pos edak o ių g upė – nus a y as eikšmingas yšys a p eks inių šal inių naudojimo dažnumo i sėkmingo eisingų e imo kalbos e minų adimo. Be o, nus a y a, kad pos - edak o iai di bo e ek y iau nei e ėjai p ak ikan ai iek laiko, iek i e imo kalbos e minų paieškos pas angų a ž ilgiu. S aipsnis baigiamas galimais y imo ezul a ų paaiškinimais i pasiūlymais dėl e ėjų mokymo. ESMINIAI ŽODŽIAI: e imas, pos edaga imas, e imo p oceso y imas, e imo iš ekliai, e minijos inkamumas. 1. INTRODUCTION T ansla ion ools ha e changed adically o e he las wo decades, wi h online digi al esou ces almos comple ely eplacing pape -based ools. In oday’s echnological en i onmen , many online digi al esou ces a e a ailable o ansla o s, bu choosing eliable and e ec i e ools equi es ca e and ou ine. Resea ch has con i med ha he use o online ansla ion esou ces cons i u es a signi ican p opo ion o he o e all ime spen on ansla ion (H elplund 2017); mo eo e , cons an echnological changes ha e led o an inc eased in e es in he ypes o ansla ion esou ces and hei use in he p ocess o human ansla ion and pos -edi ing o machine ansla ion (e.g. Gough 2019; H elplund 2016, 2017; P ie o Ramos 2021). In ecen yea s, pos -edi ing o machine ansla ion has become a common p ac ice o ansla o s, bu one o he majo sho comings o machine ansla ion engines is ha hey p oduce a high numbe o e minological e o s and inconsis encies, so pos -edi o s mus check he co ec ness and consis ency o a ge language equi alen s, which is also a ime-consum- ing p ocess equi ing a high deg ee o p ecision. The esea ch p esen ed he e ocused on de e mining wha esou ces ansla ion ainees p e e when ansla ing o pos -edi ing a legal ex , and 138 T ansla o s’ Resou ce Dominance and he Success o Finding he Ta ge Te ms in Human T ansla ion and Pos -edi ing o Machine T ansla ion Esz e Se mann Má a Lesznyák how esou ce use dominance is linked o pe o mance indices like ime on ask and co ec e minology use. Fi s , a b ie li e a u e e iew is p o ided ocusing on ansla ion p ocess esea ch, online ansla ion esou ces and ansla ion-o ien ed esea ch ac i i y. Nex , he Szeged T ansla ion Com- pe ence esea ch p ojec is desc ibed, and he me hodology o he p esen in es iga ion is de ailed. This is ollowed by he p esen a ion and discus- sion o he esul s and a conclusion. 2. LITERATURE REVIEW 2.1. T ansla ion P ocess Resea ch (TPR) T ansla ion p ocess esea ch began o p oli e a e in he 1980s (Ká oly 2022; Lesznyák 2024) when esea che s u ned hei a en ion o he cogni i e, psychological and beha iou al aspec s o ansla ion. Ea ly p ocess-o ien - ed esea ch ocused on ansla ion compe ence and he de elopmen o esea ch me hods (e.g. Wilss 1988; Ti kkonen-Condi , Jääskeläinen 2000). In he 21s cen u y, echnological ad ances ha e b ough adical changes no only o ansla ion p ac ice, bu also o ansla ion esea ch (see Klau- dy 2022 o mo e de ails), including new esea ch ools and me hods on, among o he hings, he psychological ac o s o ansla ion (Angelone 2010), ansla ion s a egies, ansla o beha iou (D ags ed 2005; Lesznyák 2008), pedagogical issues in ansla ion (Sh e e 2006; E icsson 2010), he p ocess o simul aneous ansla ion (Seebe 2015), and he physical en i- onmen o ansla o s (Eh ensbe ge -Dow, Hunzike Heeb 2016). In ansla ion p ocess esea ch, schola s use he ollowing me hods o obse e ansla o s: hink-aloud (TAP), keys oke logging, sc een eco d- ing, eye- acking, e ospec i e in e iews, and di ec obse a ion o he ansla ion p ocess (Risku 2019; Ká oly 2022; Lesznyák 2024), and hen y o iden i y he egula i ies ela ed o he ansla ion p ocess. TPR is usually no indi idual esea ch, as da a collec ion and analysis equi e collabo a i e wo k (Risku 2019), and quan i a i e and quali a i e me hods a e o en combined. 2.2. The use o ansla ion esou ces as an elemen o ansla ion compe ence Types and use o ansla ion esou ces, cogni i e e o , and pa e ns o in- e ac ion be ween p o essional ansla o s and online esou ces a e widely s udied opics (Gough 2018; Lesznyák 2008; H elplund 2016, 2017; P ie- o Ramos 2021 e c.). Di e en classi ica ions o ansla ion esou ces can 139Te minologija | 2024 | 31 be ound in he e minological and lexicog aphical li e a u e (Fó is 2019; Gaál 2016; Se mann 2021; Tamás 2014). The skills ela ed o he use o ansla ion esou ces a e included – ei- he explici ly o implici ly – in he di e en models o ansla ion com- pe ence. The PACTE Resea ch G oup (Hu ado Albi 2017) classi ies he use o esou ces as an ins umen al sub-compe ence, while in he la es e sion o he EMT model (2022), in o ma ion mining compe ence does no appea as a sepa a e elemen , bu he co ec use o sea ch engines and he use o co pus-based esou ces a e included in echnological com- pe ence. The T ansCe model (Budin e al. 2013) includes in o ma ion mining and echnological compe ence, while he eT ansFai model (2016) includes in o ma ion e ie al and e minology compe ence. In summa y, he e ec i e use o ansla ion esou ces plays a ole in some o m in each o he ansla ion compe ence models examined. 2.3 T ansla ion-o ien ed esea ch ac i i y (TRA) We based ou s udy on H elplund’s esea ch abou ansla o s’ use o digi al esou ces du ing ansla ion (H elplund 2016, 2017) and Gough’s esea ch abou ansla ion-o ien ed esea ch ac i i y (Gough 2016, 2018, 2019, 2023). H elplund’s indings (2017) show ha digi al esou ce consul a ion cons i- u es a conside able amoun o he o e all ansla ion ask ime (20 pe cen ), and ansla o s mos ly used gene al language dic iona ies, bilingual dic io- na ies and In e ne sea ch engines o bo h li e a y and echnical ansla ion. Gough (2019) used a quasi-na u alis ic, obse a ional me hod ia sc een e- co dings o s udy p o essional ansla o s’ esea ch ac i i y. The au ho iden- i ied ou esou ce beha iou ypes: Dic iona y En husias , Pa allel Tex Fan, Mixed Use and MT Adop e . He esul s sugges ha domain expe ise o a lack o i plays a ole in he esou ce beha iou p o ile. In a p e ious s udy, we in es iga ed he ypes o online ansla ion e- sou ces used by i s - and second-yea s uden s en olled in a Mas e o T ansla ion and In e p e ing p og amme when ansla ing and pos -edi ing he same legal ex om English in o Hunga ian. Resul s show ha i s - yea s uden s we e mo e likely o use monolingual and bilingual dic io- na ies, while second-yea s uden s op ed o mo e co pus-based esou ces. The choice o he online ansla ion esou ces was no de e mined by he mode o ansla ion (HT o PE), bu by he ansla o ’s le el o expe ise (Se mann 2023). 140 T ansla o s’ Resou ce Dominance and he Success o Finding he Ta ge Te ms in Human T ansla ion and Pos -edi ing o Machine T ansla ion Esz e Se mann Má a Lesznyák In he p esen s udy, based on he sc een ideos o s uden s’ wo k low, we ied o ind co ela ions be ween ansla o ainees’ esou ce domi- nance, he ime on ask and he success o inding he co ec a ge e ms. Resou ce dominance was ope a ionalized as he numbe o he ypes o di e en esou ces used and he equency o consul ing hese sou ces, which led o wo ca ego ies: (a) e mino-lexicog aphic and (b) ex -based. 3. THE SZEGED TRANSLATION COMPETENCE RESEARCH PROJECT – AIMS AND OBJECTIVES The esea ch epo ed he e is pa o he T ansla ion and MT Pos -Edi - ing Compe ence Resea ch Cen e’s o e a ching esea ch p ojec (Uni e - si y o Szeged, Hunga y), which aims o in es iga e he ole ha elemen s o ansla ion compe ence (PACTE) play in human ansla ion and he pos -edi ing o machine ansla ion. In he p ojec , we ha e been wo king wi h he a iables o sou ce language skills (English), L1/ a ge language (Hunga ian) compe ence, p o essional backg ound knowledge ela ed o he con en o he sou ce ex (bilingualism, copy igh ), decla a i e knowl- edge o ansla ion, he sou ce language ex ype, ansla ion expe ience, wo k mode (HT o PE), and s uden s’ pe cep ions o he ad an ages and disad an ages o wo king wi h each me hod (Lesznyák, Bak i, Se mann 2023, 2024). Ha ing been in ol ed in p oduc -o ien ed esea ch, we a e now p esen ing he i s esul s o ou ansla ion p ocess esea ch. In he in es iga ion epo ed he e, ou esea ch ques ions we e he ol- lowing: 1. Wha a e he esou ce p e e ences o he di e en subg oups (1s and 2nd yea s uden s, HT and PE s uden s)? 2. How is s uden s’ esou ce dominance ela ed o he success o ind- ing he a ge equi alen s o he key e ms? 3. Is he e a co ela ion be ween esou ce dominance and ime on ask? 4. Is he e a co ela ion be ween he success o inding he co ec a ge e ms and ime on ask? 4. METHODOLOGY 4.1. Pa icipan s and ma e ial 14 i s -yea s uden s (a he beginning o hei s udies) and 12 second-yea mas e ’s s uden s o ansla ion (a he end o hei s udies) o med he sam- ple o he p esen s udy. I is impo an o s ess he e ha a he Uni e si y 141Te minologija | 2024 | 31 o Szeged, in Hunga y, ansla o aining akes place a he mas e ’s le el exclusi ely, wi hin he amewo k o a ou -semes e ansla o and in e - p e e aining p og am. The e o e, i s -yea mas e ’s s uden s could be iewed as highly skilled language lea ne s o use s wi h no p io ansla- ion expe ience. Addi ionally, i ’s wo h no ing ha he s uden s did no ecei e any speci ic aining in pos -edi ing a he ime he da a was col- lec ed, bu second-yea s uden s had al eady comple ed he cou ses ‘Basics o Law’ and ‘Legal T ansla ion’. The sample was di ided in o wo subg oups ( ansla o o pos -edi o ), and hey we e asked o ansla e o pos -edi a 350-wo d long legal ex (a pa o a copy igh ag eemen , see Appendix). The sou ce language was English, he a ge language was Hunga ian and English was he B o C language o he ansla o ainees, and Hunga ian was hei A language. The MT ou pu o he pos -edi o s was p oduced by eT ansla ion, he ma- chine ansla ion ool o he EU. 4.2. The da a collec ion p ocedu e Da a collec ion ook place in Au umn 2022 and in Sp ing 2022. All he ansla ions we e p epa ed in a class oom se ing, and he e was a ime limi o 120 minu es o he ask, bu ha limi was ne e ac ually eached. S uden s had in e ne access and we e allowed o use wha e e sou ces and web pages hey wan ed o use. Ne e heless, s uden s in he HT condi ion we e ins uc ed o e ain om MT. S uden s we e asked o p oduce a a ge ex o publishable quali y (i.e., ull pos -edi ing o he MT ou pu ), wi h- ou using TM so wa e. S uden s wo ked in T anslog, as da a was collec ed on he ansla ion/pos -edi ing p ocess, oo. In addi ion, OBS S udio was used o eco d he compu e sc een while s uden s we e wo king on he ansla ion/pos -edi ing ask. To es ablish ime on ask, ideo eco dings we e used. The leng h o he ansla ion/pos -edi ing p ocess was coun ed om he momen he ‘s a logging’ bu on was pushed in T anslog un il he ‘s op logging’ bu on was pushed. Because o a echnical b eakdown, some da a was los , as a esul , we could use 22 eco dings in o al. 4.3. Me hods o da a analysis Based on he sc een ideos o s uden s’ wo k low, measu es on esou ce use we e c ea ed. Fi s , each esou ce s uden s consul ed was ca ego ized ei he as e mino-lexicog aphic (LEX) o as ex -based (TXT). Wi hin 142 T ansla o s’ Resou ce Dominance and he Success o Finding he Ta ge Te ms in Human T ansla ion and Pos -edi ing o Machine T ansla ion Esz e Se mann Má a Lesznyák each g oup wo ca ego ies we e c ea ed: (1) he numbe o he speci ic ypes o esou ces used (e.g. 3 dic iona ies) and (2) he o al equency o consul ing hese sou ces (e.g. 11 sea ches /in he 3 dic iona ies). This way ou a iables we e c ea ed ha we e used in he quan i a i e analysis: LEX numbe , LEX equency, TXT numbe , and TXT equency. A e wa ds, 10 key e ms we e selec ed om he sou ce ex (see Ap- pendix), and hei equi alen s we e e alua ed in he a ge ex s, which we e ansla ed o pos -edi ed by he s uden s. Se en o he e ms belong o he legal domain and h ee o hem o he academic domain. The co - ec Hunga ian equi alen s o he legal e ms we e de e mined by an ex- pe in copy igh law. Fo he s a is ical analysis, SPSS . 26 was used. The analysis ocused on de e mining he esou ce dominance o he subg oups wi h he help o pai ed samples - es s, compa ing he esou ce dominance o di e en subg oups wi h independen samples - es s, and we also ied o ind co - ela ions be ween esou ce dominance, success in inding app op ia e e - minology and ime on ask. Because o space es ic ions, we will e ain om sha ing la ge amoun s o da a wi h non-signi ican esul s and will ocus on signi ican indings. 5. RESULTS 5.1. Di e ences in esou ce dominance wi hin he subsamples Ou i s esea ch ques ion aimed a inding ou wha ype o esou ces he di e en subg oups p e e ed. Fi s , pai ed samples - es s we e ca ied ou o compa e he p e e ences wi hin he i s -yea and second-yea samples. On he one hand, we compa ed how many di e en ypes o esou ces hey used, and on he o he hand, he equency o consul ing hese esou ces was con as ed. Wi hin he i s -yea sample (n = 12), i was ound ha s uden s used signi ican ly mo e e mino-lexicog aphic (M = 2.42, SD = 1.68) han ex -based esou ces (M = 1.33, SD = 1.23, (11) = 3.463, p = .005). In line wi h his, hey also did signi ican ly mo e sea ches in e mino-lexi- cog aphic (M = 18.00, SD = 13.86) han is ex -based sou ces (M = 5.17, SD = 5.80, (11)= 3.736, p = .001). In he second-yea sample (n = 10), no signi ican di e ences we e ound in he ypes o sou ces used by he s uden s ( e mino-lexicog aphic M = 1.90, SD = 0.74, ex -based M = 1.40, SD = 1.27, (9) = 1.246, p = .244). 143Te minologija | 2024 | 31 In o he wo ds, he e mino-lexicog aphic and ex -based sou ces we e used o app oxima ely he same ex en , and he mino di e ence in mean alues may be due o measu emen e o . Howe e , analysing he equency o sea ches has shown ha second-yea s uden s, jus like hei i s -yea pee s, we e engaged in a signi ican ly highe numbe o e mino-lexico- g aphic sea ches (M = 17.2, SD = 11.10) han in ex -based sea ches (M = 5.4, SD = 5.83, (9) = 3.190, p = .011). As can be seen, he s anda d de i- a ion o ex -based sea ches was highe han he mean, which is a signal o high a ia ion in and abno mal dis ibu ion o he da a. Consequen ly, medians we e also compa ed o he equency o sea ches. The pai ed sam- ples Wilcoxon- es showed ha he medians we e signi ican ly di e en , oo, wi h he e ec size being la ge ( e mino-lexicog aphic MDN = 14.50, ex -based MDN = 3.50, Z (9) = -2.497, p = .013, = 0.79). In he nex s ep, he sample was di ided in o human ansla o s and pos -edi o s, and esou ce p e e ences we e analysed wi h pai ed samples - es s once again. In he human ansla o g oup (n = 10) no signi ican di e ence was ound be ween he numbe o he wo ypes o esou ces used ( e mino-lexicog aphic M = 2.70, SD = 1.16, ex -based M = 1.90, SD = 0.99, (9) = 1.809, p = .104). Ne e heless, igu es showed ha s uden s did signi ican ly mo e sea ches in e mino-lexicog aphic (M= 27.8, SD = 9.36) han in ex -based sou ces (M = 8.5, SD = 5.87, (9) = 4.636, p= .001). Resul s o he - es s indica ed ha he pos -edi o s used signi ican ly mo e e mino-lexicog aphic (M = 1.75, SD = 1.36) han ex -based esou c- es (M = 0.92, SD = 1.24, (11) = 2.802, p = .017). As he s anda d de ia ion o ex -based sou ces was la ge han he mean indica ing abno mal dis i- bu ion, he pai ed samples Wilcoxon- es was used o compa e he medians. The es esul s showed ha he e was a signi ican di e ence be ween he medians, al hough he p obabili y o measu emen e o was sligh ly highe ( e mino-lexicog aphic MDN = 2, ex -based MDN = 0, Z(11) = -2.226, p = .026, = 0.643), bu he e ec size was s ill mode a e o s ong. The pai ed samples - es showed ha pos -edi o s ca ied ou signi i- can ly mo e sea ches in e mino-lexicog aphic (M = 9.17, SD = 7.03) han in ex -based sou ces (M = 2.58, SD = 3.99, (11) = 4.026, p = .011). Be- cause o abno mal dis ibu ion, he pai ed samples Wilcoxon- es was ap- plied again, o check whe he medians we e signi ican ly di e en . Resul s o he es s we e signi ican again, sugges ing ha he di e ence be ween he median numbe o sea ches in e mino-lexicog aphic sou ces (MDN = 8) 144 T ansla o s’ Resou ce Dominance and he Success o Finding he Ta ge Te ms in Human T ansla ion and Pos -edi ing o Machine T ansla ion Esz e Se mann Má a Lesznyák and in ex -based sou ces (MDN = 0) is signi ican ly di e en (Z(11) = - 2.805, p = .005, = 0.810) wi h he e ec size being la ge. 5.2. Resou ce dominance di e ences be ween he subsamples A e cha ac e izing he indi idual subsamples, we compa ed he g oups, ha is, i s -yea and second-yea s uden s, on he one hand, and human ansla o s and pos -edi o s, on he o he hand, wi h each o he . No signi - ican di e ences we e ound be ween i s -yea and second-yea s uden s’ esea ch ac i i ies on ei he o he measu es. Howe e , when human ansla o s and pos -edi o s we e compa ed, all he di e ences in he equency o sea ches we e signi ican and he di - e ences conce ning he ypes o sou ces used we e ma ginally signi ican (see able 1). Table 1. Resou ce dominance di e ences be ween human ansla o s and pos -edi o s. Resul s o he independen samples - es s HUMAN TRANSLATORS (N = 10) POST-EDITORS (N = 12) MSD MSD T-TEST P Numbe o e mino- lexicog aphic esou ces 2.70 3.08 1.16 1.36 1.74 .096 Numbe o ex -based esou ces 1.90 0.99 0.92 1.24 2.02 .057 To al numbe o esou ces used 4.60 1.65 2.75 2.38 2.08 .051 F equency o sea ches (LEX)* 27.80 9.60 9.17 7.03 5.33 .000 F equency o sea ches (TXT)* 8.50 5.87 2.50 3.99 2.81 .011 F equency o sea ches ( o al)* 36.30 8.41 11.75 9.93 6.18 .000 No e: * = signi ican di e ences 5.3. Numbe o co ec equi alen s o e ms ound To answe esea ch ques ions 2 and 4, he success o inding he co ec a - ge equi alen s had o be de e mined. Mean alues (HT = 1.40, PE = 1.67, 1s yea = 1.42, 2nd yea = 1.70) show ha s uden s we e no pa icula ly 151Te minologija | 2024 | 31 pa icipa ed in he da a collec ion. Fu he mo e, only one ex , a legal one was used in he s udy and pa icipan s came om one ins i u ion. These ac o s limi he gene alisabili y o he esul s, al hough signi ican ind- ings may indica e alid endencies o se ings simila o ha o he s udy. The lack o e bal da a om pa icipan s p e en ed he in e p e a ion o ce ain indings, al hough i is clea ha one mus se limi s o he com- plexi y o he esea ch design, oo. The limi a ions o his s udy indica e he di ec ions o u u e esea ch. The s udy could be epea ed wi h a la ge sample in ol ing p o essionals and s uden s om o he uni e si ies o coun ies. A di e en ex ype could be used and o he ypes o p ocess da a could be collec ed. Mo eo e , he da a se o he p esen s udy has emained pa ly unex- plo ed. Time spen on esea ch has no been measu ed o ela ed o indices o esea ch beha iou o pe o mance. The use o ansla ion esou ces in he di e en s ages o HT and PE (d a ing, e ision) could be s udied, oo, jus like he numbe o e minology changes pos -edi o s make and hei ela ion o esea ch s a egies. Despi e i s limi a ions, he s udy ce ainly adds o ou unde s anding o ansla ion ainees’ esou ce use. Ne e heless, mo e wo k needs o be done in his ield o disco e wha esea ch s a egies and beha iou s a e needed in he digi al e a whe e pos -edi ing will likely domina e ansla o s’ wo k. REFERENCES Angelone E ik 2010: Unce ain y, Unce ain y Managemen and Me acogni i e P oblem Sol ing in he T ansla ion Task. – “T ansla ion and Cogni ion”, ed. Sh e e G. M., Angelone E., Ams e dam: Ame ican T ansla o s Associa ion Schola ly Monog aph Se ies XV, 17–40. Budin Ge ha d, K ajcso Zi a, Lommel A le 2013: The T ansCe P ojec : Ensu ing Tha T ansna ional T ansla o Ce i ica ion Mee s S akeholde Needs. – The In e na ional Jou nal o T ansla ion & In e p e ingResea ch 5(1), 143–155. DOI: 10.12807/ i.105201.2013.a08. D ags ed Ba ba a 2005: Segmen ing in T ansla ion: Di e ences ac oss le els o expe ise and di icul y. – Ta ge 17(1), 49–70. DOI:10.1075/ a ge .17.1.04d a. E icsson K. Ande s 2010: Expe ise in In e p e ing: An expe -pe o mance pe spec i e. –“T ansla ion and Cogni ion”, ed. Sh e e G. M., Angelone E., Ams e dam: Ame ican T ansla o s Associa ion Schola ly Monog aph Se ies XV, 231–262. EMT 2022: Upda ed e sion o he EMT compe ence model amewo k. A ailable a : h ps://commission. eu opa.eu/news/upda ed- e sion-em -compe ence- amewo k-now-a ailable-2022-10-21_en. eT ansFai 2018: Compe ence Ca d o Specialised T ansla o s. A ailable a : h ps://e ans ai .inyk.bme.hu/. Fó is Ágo a 2019: Új endenciák és módsze ek a o dí ási olyama ban: a e minológia és a dokumen áció sze epe. – “Diszciplínák alálkozása a o dí ásban”, ed. Szo ák Sz., Budapes : OFFI, 211–219. Gaál Pé e 2010: A hagyományos szó á ak izsgála i szempon jainak alkalmazha ósága az online szó á ak ese ében. – “Az alkalmazo nyel észe ma: inno áció, echnológia, adíció. XX. Magya Alkalmazo Nyel észe i Kong esszus”, ed. Boda I. K., Mónós K., Budapes /Deb ecen: MANYE–Deb eceni Egye em, 432–437. 152 T ansla o s’ Resou ce Dominance and he Success o Finding he Ta ge Te ms in Human T ansla ion and Pos -edi ing o Machine T ansla ion Esz e Se mann Má a Lesznyák Gough Joanna 2016: The pa e ns o in e ac ion be ween p o essional ansla o s and online esou ces. PhD Thesis, Uni e si y o Su ey. A ailable a : h ps://open esea ch.su ey.ac.uk/esplo o/ou pu s/99514578402346. Gough Joanna 2018: In es iga ing he use o esou ces in he ansla ion p ocess. – “T ends in e- ools and esou ces o ansla o s and in e p e e s”, ed. Pas o G. C., Du án-Muñoz I., Leiden: B ill, 9–36. Gough Joanna 2019: De eloping ansla ion-o ien ed esea ch compe ence: wha can we lea n om p o essional ansla o s? – The In e p e e and T ansla o T aine 13:3, 342–359. DOI: 10.1080/1750399X.2019.1656404. Gough Joanna 2023: Indi idual Va ia ions in In o ma ion Beha iou o P o essional T ansla o s: Towa ds a Classi ica ion o T ansla ion-o ien ed Resea ch S yles. T ansla ion S udies. Taylo and F ancis Online. A ailable a : h ps://www. and online.com/doi/ ull/10.1080/14781700.2023.2231933. Eh ensbe ge -Dow Mau een, Hunzike Heeb And ea 2016: In es iga ing he E gonomics o a Technologized T ansla ion Wo kplace. – “Reembedding T ansla ion P ocess Resea ch”, ed. R. Muñoz Ma ín, Ams e dam/Philadelphia: John Benjamins, 69–88. Hu ado Albi Ampa o (ed.) 2017: Resea ching T ansla ion Compe ence by PACTE G oup, Ams e dam/ Philadelphia: John Benjamins. DOI: 10.1075/b l.127. H elplund K is ian Tangsgaa d 2016: Cogni i e e iciency in ansla ion. – “Reembedding T ansla ion P ocess Resea ch”, ed. R. Muñoz Ma ín, Ams e dam/Philadelphia: John Benjamins, 149–170. H elplund K is ian Tangsgaa d 2017: T ansla o s’ Use o Digi al Resou ces du ing T ansla ion. – He mes 56, 71–87. Ká oly K isz ina 2022: A nyel i köz e í és empi ikus ku a ásának módsze ei. – “Be eze és a o dí ás és a olmácsolás ku a ásmódsze anába I. Ál alános ész”, ed. K. Klaudy, E. Robin, O. Seidl-Péch, ELTE FTT– MANYE Fo dí ás udományi Szakosz ály, 93–108. Klaudy Kinga 2022: A o dí ás udomány dilemmái a 21. század elején. – “Be eze és a o dí ás és a olmácsolás ku a ásmódsze anába I. Ál alános ész”, ed. K. Klaudy, E. Robin, O. Seidl-Péch, ELTE FTT– MANYE Fo dí ás udományi Szakosz ály, 13–25. DOI: h ps://doi.o g/10.21862/ku modsze an1/2. Lesznyák Má a 2008: A szó á használa jellegze ességeinek ál ozása a o dí ói apasz ala ügg ényében. – “Az I. Alkalmazo Nyel észe i Dok o andusz Kon e encián elhangzo előadások kö e e”, ed. Vá adi T., Budapes : MTA Nyel udományi In éze , 67–79. Lesznyák Má a 2024: Folyama o ien ál ku a ások a o dí ás udományban. – “Be eze és a o dí ás és a olmácsolás ku a ásmódsze anába II. Speciális ész”, ed. K. Klaudy, E. Robin, O. Seidl-Péch, ELTE FTT– MANYE Fo dí ás udományi Szakosz ály, 71–94. Lesznyák Má a, Bak i Má ia, Se mann Esz e 2023: T ansla ion and pos -edi ing pe o mance o ansla ion s uden s –a c oss-sec ional analysis. – L10N Jou nal 2(2), 7–23. Lesznyák Má a, Bak i Má ia, Se mann Esz e 2024: Reading akes i all? – he ole o language compe ence and subjec knowledge in legal ansla ion. – The In e p e e and T ansla o T aine , 1–20. A ailable a : h ps://doi.o g/10.1080/1750399X.2024.2344177. P ie o Ramos Fe nando 2021: T ansla ing legal e minology and ph aseology: be ween in e - sys emic incong ui y and mul ilingual ha moniza ion. – Pe spec i es 29:2, 175–183. DOI: 10.1080/0907676X.2021.1849940. Risku Hanna 2019: P ocess esea ch. – “Rou ledge Encyclopedia o T ansla ion S udies. 3 d Edi ion”, ed. M. Bake , G. Saldanha, London: Rou ledge, 437–441. Seebe Kilian G. 2015: Cogni i e Load in Simul aneous In e p e ing: Measu es and me hods. – “In e disciplina i y in T ansla ion and In e p e ing P ocess Resea ch”, ed. M. Eh ensbe ge -Dow, S. Göp e ich and S. O’B ien, Ams e dam/Philadelphia: John Benjamins, 18–33. Se mann Esz e 2021: Te minológiai szab ányosí ás és nyel i köz e í és, Szeged: Juhász Gyula Felsőok a ási Kiadó. Se mann Esz e 2023: Online o dí ói segédeszközök használa a humán o dí ás és a gépi o dí ás u ósze kesz ése so án. – “XXIX. Magya Alkalmazo Nyel észe i Kong esszus” (accep ed o publica ion). Sh e e G ego y 2006: The Delibe a e P ac ice: T ansla ion and expe ise. – Jou nal o T ansla ion S udies 9(1), 27–42. Ti kkonen-Condi Sonja, Jääskeläinen Rii a 2000: Tapping and mapping he p ocesses o ansla ion and in e p e ing, Ams e dam/Philadelphia: John Benjamins. A ailable a : h ps://doi.o g/10.1075/b l.37. Tamás Dó a Má ia 2014: Gazdasági szakszö egek o dí ásának e minológiai ké dései ől, Budapes : ELTE. Wilss Wol am 1988: Kogni ion und Übe se zen: Zu Theo ie und P axis de menschlichen und de maschinellen Übe se zung, Tübingen: Niemeye . 153Te minologija | 2024 | 31 APPENDIX 1. The sou ce ex used o he s udy: LICENCE TO PUBLISH Manusc ip #: [name o jou nal] (“The Jou nal”) Ti le o he con ibu ion: (“The Con ibu ion”) Au ho (s): (“ he Au ho s”) To: Na u e Publishing G oup (“NPG”), a di ision o Macmillan Publishe s L d 1. In conside a ion o NPG ag eeing o publish he Con ibu ion he Au ho s g an o NPG o he ull e m o copy igh in he Con ibu ion and any ex ensions he e o, subjec o clause 2 below, he exclusi e licence (a) o publish, ep oduce, dis ibu e, display and s o e he Con ibu ion in all o ms, o ma s and media whe he now known o he ea e de eloped (including wi hou limi a ion in p in , digi al and elec onic o m) h oughou he wo ld, (b) o ansla e he Con ibu i- on in o o he languages, c ea e adap a ions, summa ies o ex ac s o he Con i- bu ion o o he de i a i e wo ks based on he Con ibu ion and exe cise all o he igh s ou lined in (a) abo e in such ansla ions, adap a ions, summa ies, ex ac s and de i a i e wo ks and (c) o license o he s o do any o all o he abo e. 2. Owne ship o copy igh emains wi h he Au ho s and p o ided ha , when ep oducing he Con ibu ion o ex ac s om i , he Au ho s acknowledge i s and e e ence publica ion in he Jou nal, he Au ho s e ain he ollowing nonexclusi e igh s: a) To ep oduce he Con ibu ion in whole o in pa in any p in ed olume (book o hesis) o which hey a e he au ho (s). b) They and any academic ins i u ion whe e hey wo k a he ime may ep o- duce he Con ibu ion o cou se eaching. c) To pos a copy o he Con ibu ion as accep ed o publica ion a e pee e iew (in Wo d o Tex o ma ) on he Au ho ’s own websi e o ins i u i- onal eposi o y six mon hs a e publica ion o he p in ed edi ion o he Jou nal, p o ided ha hey also gi e a hype link om he Con ibu ion o he Jou nals web si e d) To euse igu es o ables c ea ed by hem and con ained in he Con ibu- ion o o he wo ks c ea ed by hem. This Ag eemen shall be go e ned by and cons ued in acco dance wi h he laws o England wi hou ega d o he p inciples o con lic s o law. The pa ies he e o submi o he non-exclusi e ju isdic ion o he English cou s. APPENDIX 2. The 10 a ge e ms selec ed o he analysis: licence o publish, con ibu ion, ull e m o copy igh , including wi hou limi- a ion, o g an exclusi e licence, o ep oduce, de i a i e wo k, hesis, academic ins i u ion, pee e iew 154 T ansla o s’ Resou ce Dominance and he Success o Finding he Ta ge Te ms in Human T ansla ion and Pos -edi ing o Machine T ansla ion Esz e Se mann Má a Lesznyák VERTĖJŲ IŠTEKLIŲ DOMINAVIMAS IR SĖKMINGA VERTIMO KALBOS TERMINŲ PAIEŠKA VERČIANT ŽMOGUI ARBA POSTREDAGUOJANT MAŠININĮ VERTIMĄ San auka Ve imo p iemonių naudojimas iš esmės pasikei ė pe pas a uosius du dešim mečius – in e ne iniai skai meniniai iš ekliai be eik isiškai pakei ė popie ines p iemones. Šiuo- laikinėje echnologinėje aplinkoje e ėjai gali naudo is daugybe in e ne inių skai - meninių iš eklių, ačiau no in pasi ink i pa ikimas i eiksmingas p iemones, eikia k uopš umo i disciplinos. Šiame y ime nag inėjamas e ėjų p ak ikan ų iš eklių domina imas ą pa į eisinį eks ą e čian žmogui i pos edaguojan i analizuojami galimi yšiai a p iš eklių domina imo, užduo ies a likimo laiko i sėkmingai as ų eisingų e imo kalbos e minų. Čia pa eik as y imas y a Segedo e imo kompe encijos y imų g upės bend ojo y imo p ojek o, ku iuo siekiama iš i i e imo kompe encijos elemen ų (PACTE) aidmenį žmogaus e ime i mašininio e imą pos edaga ime, dalis. Kadangi da- ly aujame į p oduk ą o ien uo uose y imuose, daba p is a ome pi muosius e imo p oceso y imo ezul a us. Mūsų ezul a ai odo, kad e minog a iniai i leksikog a iniai šal iniai dominuoja su e imu susijusiuose s uden ų y imuose. Be o, s uden ams ganė inai nesisekė as i eisingų a i ikmenų e minams, jie ado idu iniškai ieną a du iš dešim ies. Be o, nei laikas užduočiai, nei iš eklių domina imas nepa odė jokių eikšmingų ko eliacijų su inkamų e imo kalbos e minų adimu. Ve ėjų i pos edak o ių palyginimas pa o- dė, kad pos edak o iai di bo g eičiau i a liko mažiau paieškų nei e ėjai. Nepaisan o, jų as ų e minų eisingumas bu o panašus į e ėjų. Aki aizdu, kad y imas u i am ik ų ūkumų, iš ku ių ienas y a maža im is, ačiau, nepaisan jų, iš ados ik ai p isideda p ie mūsų sup a imo apie e ėjų p ak i- kan ų iš eklių naudojimą. Gau a 2024-05-03 Esz e Se mann Depa men o I alian Language and Li e a u e, Uni e si y o Szeged, Hunga y 6724 Szeged, Pulz u ca 1. E-mail esz e .se [email protected] Má a Lesznyák Ins i u e o English and Ame ican S udies, Uni e si y o Szeged, Hunga y 6722 Szeged, Egye em u ca 2. E mail [email protected]