Abstract

The following project is based on Machine Translation. We will test the accuracy of some translators, with regards to some langague idioms.The translations will be from English into Spanish. In this particular case, we have chosen 2 different translators: SYNTRAN and FREETRANSLATION. We can see that though Machine Translation System is useful in a first approach to an unknown language, a great improvement is needed; since the translations are made word by word, this is, literally. This fact, makes the understanding difficult for the receptor of the translations. A great effort of intelligibility is required to the user of the translator.

Introduction

AThis report, is based on the analysis of Machine Translation System. In this aim, we have chosen two different translators; SYSTRAN and FREETRANSLATION. 40 English idioms have been introduced in each translator, so that we could compare the results, and this way, the differences and similitude's of the two translators. In the case of FREETRANSLATION, we observe that the translations are quite literal. By the term literal we mean, that words are translated literally, though with quite sense with regards to sentence structure. Another thing to take into account is that, though the words in English do not vary in gender, the translator translate the proper gender in the articles, determiners and words. Still, the proper translation of words, as well as, the correct word order are not enough for the receptor to get the sense of the idiom. A great deal of knowledge of the language is required to the user, in order to acknowledge what the real translation, or the proper idiom would be in Spanish. Something to underline on this translator is the fact that some words are not translated, which makes the understanding a bit more difficult for the interpreter. Looking now at the SYSTRAN, we see that it is less accurate than the already mentioned FREETRANSLATION. This second translator those not always distinguish the correct gender. Furthermore, in some particular cases it does not even use the corresponding number in adjectives, articles or determiners. "Mucho un verdad ". This may lead the user to grammatical mistakes if he/she does not have a high level of the language into which it has been translated, in this case, Spanish. What is more, these grammatical mistakes might be fossilised by the person, this is, internalised. This system, does also apply the same kind of translation system, which as we already said in the previous case is the word by word translation. However, we must admit that the results obtained from FREETRANSLATIONS are easier to interpret. Though it does not offer the corresponding idiom in Spanish, the sense of it is easier to guess. "ser verde con la envidia" , in this case, it would not be the correct expression but the reader would be able of getting the essence of the sentence, whereas in the translation by SYSTRAN, "para ser verde con el asimiento oscuro del caballo de la envidia", the translation does not make any clear the meaning of the idiomatic expression. Apart from all these, we must also say that both systems have similar types of grammatical errors. The infinitival "to", is translated as a preposition, mostly "para", making this way, the verb tensed, instead of infinitival. But this occurs more often in SYSTRAN, which also tends to include extra words, making then translation a bit more complicated. All these characteristics, can clearly be seen in the table presented above. .

Table

Source SYSTRAN Target 2 Dictionaryla casa de un hombre es su
A man's house is his castle la casa de un hombre es su castillo un hombre la casa de ese es su castillo Mientras en mi casa estoy, rey soy
In many words, a lie or two may scape en muchas palabras, una mentira o dos puede scape en muchas palabras, una mentira o dos pueden escapar A mucho decir, mucho mentir
Where there's a sweet, there's always a bitter donde hay un dulce, hay siempre un amargo donde hay un dulce, hay siempre un amargo No hay miel sin hiel
God tempers the wind to the shorn lamb el Dios templa el viento al cordero cortado el Dios templa el viento al cordero cortado Dios aprieta, pero no ahorca
Many a true is spoken in jest mucho un verdad se habla en broma muchas una palabra verdadera se habla en la broma Entre broma y broma, la verdad se asoma
Once in a while never hurt anyone de vez en cuando nunca lastimado cualquier persona de vez en cuando nunca lastimó nadie Una vez al año, no hace daño
Poverty is the mother of all arts la pobreza es la madre de todos los artes la pobreza es la madre de toda artes Más discurre un hambriento que cien letrados
If you scratch my back, I'll scratch yours si ustez rasguña mi parte posteriora, rasguñare el tuyo si usted rasguña mi espalda que rasguñaré suyo Favor, con favor se paga
There's many a good tune play on an old fiddle hay mucho un buen juego de la consonancia en viejo violin hay muchos un aire bueno jugado en un campo viejo Gallina vieja hace buen caldo
Of evil grain, no good seed can come de grano malvado ninguna buena semilla puede venir de grano malo, ninguna semilla buena puede venir De mala mata, nunca buena zarza
That's the thanks you get ésa es las gracias que usted consigue eso es la gracias que usted obtiene Cría cuervos, y te sacaran los ojos
Words hurt more than swords las palabras lastimaron más que las espadas palabras lastiman más que espadas Palabras de boca, piedra de honda
To wise up a ascendente sabio a sabio arriba caer en el chiste
to buy a pig in a poke para comprarme un cerdo en un empuje compar un puerco en un pinchazo comprar una casa sin verla
as if he owned the place como si él poseyera el lugar como si el poseyera el lugar como Pedro por su casa
on the rocks en las rocas en las piedras en mala situacion
to be over the moon para estar sobre la compañia de la luna estar sobre la luna estar en el septimo cielo
two's company, three's a crowd dos, una muchedumbre de tres dos son las compañías, tres son una multitud dos son pareja, tres son multitud
according to the book segun el libro según el libro como Dios manda
dressed to kill vestido para matar vistió para matar de punta en blanco
for love and money para el amor y el dinero para el amor o el dinero a cualquier precio
how thing tum out como tum de la cosa fuera cómo cosas resultan segun salga el dado
right off (the bat) la derecha apagado el derecho lejos de buenas a primeras
to hand something to somebody on a plate dar a algo alguien en una placa para entregar algo a alguien en un plato ponerselo a alguien en bandeja de plata
he's not my cup of tea lo no es mi taza de te uno es mi copa de te no es santo de mi devocion
to be as cool as a cucumber estar tan fresco como un pepino ser refresca como un pepino estar mas fresco que una lechuga
till the cows come home hasta quevienen las vacas a casa hasta que las vacas viene hogar cuando las ranas crien pelo
to be green with envy para ser verde con envidia ser verde con la envidia morirsede envidia
dark horse caballo oscuro eso es un caballo oscuro incognita
hold your horses sostenga sus caballos el sostuvo sus caballos para el carro
to give somebody the red carpet treatment para dar a alguien el treatment rojo de la alfombra dar alguien el tratamiento rojo de alfombra tratar a alguien a cuerpo de rey
once in a blue moon una vez en una luna azul una vez en una luna azul de pascuas a ramos
to be the black sheep of the family para ser las ovejas negras de la famila ser la oveja negra de la familia ser la oveja negra de la familia
to turn as red as a beetroot para dar vuelta tan roja como una remolacha ir tan rojo como un beetroot ponerse como un tomate
you can shout till blue in the face usted puede gritar hasta que usted es azul en la cara usted puede gritar hasta usted es azul la cara puedes gritar hasta hartarte
for donkey's years por los años del burro para el asno años de s durante muchos años
to be straight from the horses mouth para ser recto de la boca de los caballos tiene su sea directamente de la boca de los caballos ser the buena tinta
that's ancient history esa es historia antigua eso es un cuento antiguo esa es cosa vieja
to swear black and blue that para jura negro y el azul eso jurar eso negro y azul jurar por todo lo mas santo que
to be as white as a sheep para ser tan blanco como una oveja ser tan blanco como una oveja estar palido como la muerte

Conclusion

In general , we see that both translators are quite similar. Similar in the sense that both use the same kind of translation system, word by word They also make similar grammatical errors, and non of them gives us the right form of the introduced idiom, into the required language. This is anyhow, quite an expected result. Since a very difficult problem to solve is the fact that different languages use different structures and sometimes there are terms for which no appropriate translation exists. In this point is where the human translator must act creatively, because computers, machine translators do not have that capacity of interpretation. From a more general approach, we see that translation is one of the solutions that we can find to overcome barriers of communication. This point drives us into the word of the machine translation, a ranch of which, we have analysed in this project. This kind of translator, include any computer based process that permit the transformation of a text into any other language. It quickens the work of the human translators, making work easier and much more comfortable in a way. Still, human contribution is always needed, due to unsolved problems, as shown in our project. .