Language identification is a term used to describe the process by which one recognizes the language of written or verbal works. Though it can be quite easy to differentiate certain categories of languages, such as Romance and Germanic languages, it is sometimes more difficult to tell specific similar languages apart. Language identification is important for the purpose of categorizing various works of literature and for computerized translation. Many languages have characteristic words or letters that can allow one to recognize the language without understanding it. Many computational approaches, mostly based in statistics, also exist for the purpose of determining the language of a given text or spoken work.
Many people, even those without a great deal of formal education, are generally capable of some limited level of language identification. An individual asked whether a given language is German or Chinese, for instance, will generally be able to tell based either on the sound of the words or on the appearance of the written language. Different languages are commonly used in movies and books that reach wide audiences, so even those who seldom travel and never study foreign languages are generally capable of rudimentary language identification.
In libraries and online databases, it is sometimes necessary to categorize texts according to the languages in which they are written. In some cases, particularly when a digital copy of a work does not exist, language identification must be performed without the help of computational methods. Difficulties arise in situations involving highly similar languages, such as Portuguese and Spanish or Swedish and Norwegian, as a cursory glance at the text may not necessarily be sufficient to differentiate such similar languages. Upon narrowing the list of possible languages down to only a few, though, one can generally consult a chart of words and characters that are characteristic of only one language.
Manual language identification is not generally necessary for texts that have been digitized, as there are many different computational methods of language identification. Texts are generally statistically analyzed and compared to reference texts, though other methods of computational language identification do exist. Such computational methods can be used for sorting purposes. They are also particularly useful in computational translation programs, as it is necessary to identify a language before properly translating it to another language. Some computational language translation or recognition tools are able to adapt as more information is given — one or two words may lead the program to the conclusion that a text is in one language while a full paragraph could reveal that it is, in fact, in a different but similar language.
