As we have shown in the first part, automatic text verification systems aim to become useful resources. However, these applications are by definition tools that help in writing, and they should never replace the human proofreader, especially if the goal is publishing. Until now, there were a lot of questions that technology could not face.
Where should we focus our attention?
We cannot trust technology when text revision involves a comprehensive and careful reading in order to find ambiguous sentences or inconsistencies from the author (e.g. changing in a story the name of the same character), or decide whether a footnote would be necessary, etc.
Apart from this, we must give attention to another type of revision. It is called conceptual or technical revision, and it consists in examining the text to see if it conforms to the terminological conventions which are typical of the related subject. In fact, this task should not be assigned to a specialist in spelling and style, but rather to a specialist in the given subject (a physician for a handbook of medicine, an engineer for a technical text, etc.).
Despite these facts, we must note that language technologies specialists have begun to handle information on a semantic basis. Examples of this are the recognition of anaphoras and coreferences. We believe that, in the near future, there will be major advances in the detection of certain lexical ambiguities or misuses.
Why should publishing professionals make use of automatic proofreading?
We assume that revising a text is a time-consuming task. Thus, we believe that publishing professionals can go a step further, and not just confine themselves to the process of looking for information in dictionaries, grammars, and other reference books. The new automatic proofreading systems are certainly helpful:
- You can save time on tedious tasks that the proofreader can perform easily.
- You can focus your efforts on activities that involve human processing.
- You can improve the quality of the final revision.
- You will have more time left to meet the tight deadlines imposed by the publisher.
In conclusion, you can be more productive, increase your profits and, at the same time, maintain the quality of your work.
[English version of ¿Qué aporta la corrección automática al profesional de la edición? (parte 2)]
A human proofreader is a professional in charge of revising materials written by an author. He tries to ensure that the readers receive the message clearly and free from errors.
The editing process is commonly comprised of several different levels of textual revision: spelling and typographical checking, style checking, conceptual revision, and revision of translated texts, were that the case. All of the publishing houses are aware of this process, but only a few put it into practice. In reality, it is not common for a publishing house to properly assign each revision type to specialized proofreaders. Usually, the proofreader of a given text gets far too much work, as he carries out all the revision work that three or four specialists should have done. He stands as a mediatory demiurge who links ideas to something legible. How much are they paid for this? 0,72 € per 1000 matrixes (or characters with spaces) for proofreading on screen, and around 0,50 € for second galleys (proofreading on paper). In conclusion, they are working for five or six euros per hour in the most profitable cases.
Thus, these edition demiurges may want to explore ways of increasing productivity and, at the same time, protecting the quality of their work.
How can the automatic text verification technology contribute to the proofreading process?
Granted, philologists and some other language professionals are very reluctant to anything related to “automatic proofreading”, however, we want to make clear that prejudging a last-generation software tool is somewhat unfair. Language lovers might congratulate themselves on the new Natural Language Processing technologies that make it possible to automatically proofreading a text. These automatic proofreaders are able to check, with a high degree of linguistic precision and recall, many items regarding spelling and typography (according to the application’s degree of processing). Equally, they can make a text conform to the spelling and grammar rules. On the other hand, the majority of these applications do not rewrite the text automatically, but rather they give the user a choice among the different proposals that the application makes.
What issues can be addressed by automatic proofreading?
- Spelling and typographic checking. An optimum level of orthographic recall can be reached if the system has a good lexical base. This avoids false warnings on existing words (even if they are not frequent), and also permits to check the spelling of national and foreign proper nouns (e.g. toponyms, persons’ names, institutions, brand names, etc.). In addition, many tools comprise personal dictionaries where new words are added, hence the lexical base is expanded. On the other hand, these new applications are becoming context-sensitive so that homophones and diacritic errors can be found. Finally, there are more issues concerning spelling and typography that a proofreading application considers too: it can now advise on the use of italics (e.g. foreign words), verify the opening and closing of pairs of signs, warn of wrong sequences of punctuation marks, verify the correct use of upper and lower case letters, check the spacing (double spaces, required spaces or joins between typographic signs and words), etc.
- Grammar checking. Last-generation proofreading applications have the potential to disambiguate different senses. It allows for finding many agreement errors at different sentence levels, and other syntactic violations such as mismatched verb tenses, or errors in prepositional government.
- Style checking. These applications are able to make suggestions about spelling variations that are much preferred, lexical misuses or very colloquial registers. They can also provide alternatives to foreign words, and warn of phenomena that can make reading confusing (abusive use of prepositions, word repetitions, too long sentences, redundancies, unwanted technical words, etc.).
- Revision of translations. These applications are able to find loan translations between the source and the target language. They can also warn of false friends or wrong transliterations.
What is left for a human proofreader?
[English version of ¿Qué aporta la corrección automática al profesional de la edición? (parte 1)]