Report B by Ager Gondra, Unai Diego de Somonte, Andrew San Juan, Stéphane Cos.



The main aim of this report is to study the efficiency of translators. For that purpose we have chosen two popular translators: Systran and Google. The report is divided in two parts. First we are going to work with some coloquial expressions that could lead to confusion, due to the various meanings the same word may have. And the second part includes a series or popular sayings and proverbs.




This report is going to deal with the issue of Machine Translation. The purpose of this tool is to be able to write asking for information in your mother language, and get a response with the information requested in the same language you asked for it, although that information might be in another different language.

Nevertheless, this task may be more complicated than what it might seem at first sight. This is because every language has its own twists and turns that make one single word have not only more than one meaning, but also certain connotations.

As said before in the abstract, to accomplish our task we haver chosen two different translators, Systran and Google (references below), as well as divided the report into two different parts. The first one include a series of expressions used in every day life, and the second one includes a number of English popular sayings and proberbs.

After we have used both translators, we will procede to compare the results obtained with the real meaning these expressions have. Both parts of the report are displayed in two charts. And after each chart there will be a small analysis comenting on the outcome.



Every Day Expressions

Sintagma en inglés

Traducción de Google

Traducción en castellano de Systransoft

Traducción real

Football pitch

Pitch del balompié

Campo de fútbol

Campo de fútbol

Oh, my God

Oh mi Dios

Oh mi Dios

Oh, Dios mío

I´m so happy

Soy tan feliz

¡Soy tan feliz!

¡Qué contento estoy!

In about three hours

Sobre cerca de tres horas

Sobre cerca de tres horas

En unas tres horas

About 7 or 8

Cerca de 7 o 8

Cerca de 7 o 8

Unos 7 u 8

A brick-wall

Un ladrillo-pared

Un ladrillo-pared

Un muro de ladrillos

Whatever it may be/seem

Lo que puede ser

Lo que puede ser

Lo que puede ser

Bull shit

Mierda del toro

Mierda del toro


Bull´s eye

Ojo del toro

ojo del toro

Diana, blanco

Straight ahead

Todo derecho

Todo derecho

Directo al grano

Straight forward

Derecho remita

Derecho remita

Todo recto

Right in the middle

Derecho en el centro

Derecho en el centro

Justo en el blanco, medio

In this list we are able to see lots of similarities between Google and Systransoft translators. Both of them fail in mistakes such as:



Sayings and Proberbs

Refran original

Traducción con Systran

Traduccion con Google

Traduccion correcta

And the pot called the kettle black

Y el pote llamado el negro de la caldera

Y el pote llamado el negro de la caldera.

Y la sartén le dijo al cazo: "apártate que me tiznas"

Barking dog seldom bites

El raspar persigue raramente la mordedura

Mordeduras del perro el raspar raramente.

Perro ladrador poco mordedor

Once in a year never hurts anyone

Una vez en un año nunca lastima cualquier persona

Una vez en un año nunca lastima cualquier persona.

Una vez al año no hace daño

Let the bygones be bygones

Deje los bygones ser bygones

Deje los bygones ser bygones.

El pasado, pasado está

Many a true word is spoken in jest


Mucho una palabra verdadera se habla en broma

Mucho una palabra verdadera se habla en broma.

Entre broma y broma la verdad asoma

Brain is better than brawn

El cerebro es mejor que el queso de cerdo

El cerebro es mejor que el queso de cerdo

Más vale maña que fuerza

As you sow, so shall you reap

Pues usted puerca, así que usted cosechará

Pues usted puerca, así que usted cosechará

Quien siembra vientos recoge tempestades

You can't make a silk purse out of a sow's ear

Usted no puede hacer un monedero de seda fuera del oído de una puerca

Usted no puede hacer un monedero de seda fuera del oído de una puerca

Aunque la mona se vista de seda, mona se queda

He who laughs last laughs longest

Él que ríe por último ríe lo más de largo posible

Él que ríe por último ríe lo más de largo posible.

Quien ríe el último ríe mejor

Better safe than sorry

una caja fuerte mejor que apesadumbrada

Una caja fuerte mejor que apesadumbrada.

Es mejor prevenir que curar

As we have seen, both Google and Systran shares similar mechanics in their translation methods and the same errors issued when translating colloquial expressions appears with sayings:




    After having tested two diferent translators, we have come to the conlusion that automatic translators need great improval. The more common errors made by automatic translators are the gramatical ones. This is mainly due to the incapacity to overcome the diferences between English and Spanish gramatical structure.

    The following problems are the more common ones:

1- Identification between gramatical terms is not perfect

2- Words that the system doesn't understand are not translated.

3- Some ortographic rules are not considered.

    Finally, last but not the least, for an efficient machine translation, errors of a too categorical translation may be eliminated by the human presence; after all, they are only machines, and should be completed with the help of a human mind until they are developped to a higher level.