Sunday, February 13, 2011

Le problemas le plus grande que io ha habite con le traductor electronic de Google


(Languages of this post: Interlingua, English)


Generalmente, secundo mi experientia, le traductor electronic de Google, como il ha un base exclusivemente statistic, pote producer traductiones ubi le ordine de parolas es correcte, in le caso de pares de linguas con le quales io ha travaliate pro producer le textos trilingue publicate in “Intelingua multilingue”. In altere textos ubi il non ha un quantitate grande de textos que le systema de Google pote digerer, lo que es ver in le caso de linguas ultra le familia anglo-romanic, su traductiones es plus inexacte e confundente.

Il ha problemas special con traductiones del anglese al germano e al hollandese, le quales in lor propositiones subordinate ha sequentias de parolas que es multe differente de lor equivalentes in anglese.

E in le caso del anglese e le linguas romanic, sovente le traductiones de Google, ben que illos ha parolas con sequentias appropriate in le lingua al qual on traduce, sovente ha errores in le plurales, in le correspondentia de genero inter adjectivos e nomines, e de numero inter le subjectos e le verbos de un proposition, e il es necesse que un persona qui usa iste systema sape corriger los.

Sovente quando un expression in anglese non provide un traduction adequate, il es possibile trovar un altere simile expression que produce un traduction satisfactori. Como mi lingua native es le anglese, pro me il es possibile trovar un grande varietate de tal expressiones alternative.

Il anque me pare que sovente le traductor de Google functiona plus efficacemente si on traduce del anglese a un altere lingua in vice de un altere lingua al anglese o a un lingua que non es le anglese.

Il esserea possibile reducer iste problemas si on poterea incorporar un componente syntactic al algorithmo statistic de Google, que es le sol methodo usate per su systema de traduction.

Io es assatis confidente que altere systemas de traduction electronic in le futuro potera combinar iste duo typos de algorithmos pro meliorar considerabilemente le qualitate de lor traductiones.

---

The biggest problems that I have had with Google’s electronic translator

Generally, according to my experience, Google’s electronic translator, since it has an exclusively statistical basis, can produce translations where the word order is correct in the case of languages I have worked with to produce the trilingual texts published in “Interlingua multilingue.” In other texts where there is not such a large quantity of bilingual texts, which often is true in the case of languages outside of the Anglo-Romance family, its translations are more inexact and confusing.

There are special problems with translations from English to German and Dutch, which in their subordinate clauses have sequences of words that are very different from their equivalents in English.

And in the case of English and the Romance languages, often Google’s translations, though they have words with appropriate sequences in the language into which the translation is made, often have errors in plurals, in gender agreement between adjectives and nouns, and in number agreement between subjects and verbs within clauses, and it is necessary for someone using this system to know how to correct them.

Often when an expression in English does not provide an adequate translation, it is possible to find another similar expression that produces a satisfactory translation. Since my native language is English, for me it is possible to find a great variety of such alternative expressions.

It also seems to me that Google’s translator would work more efficiently if a person translates English into another language instead of another language into English or into a language other than English.

It would be possible to reduce these problems if a syntactic component could be incorporated into Google's statistical algorithm, which is the only method used by its translation system.

I am rather confident that other electronic translation systems in the future will be able to combine these two different kinds of algorithms to considerably improve the quality of their translations.

No comments: