Well, actually we do. They are an important part of the English language. But more often than not do they get used incorrectly in writing and, what’s worse, documented incorrectly in terminology entries.
I have been asked at least a few times by content publishers whether they can use gerunds or whether a gerund would present a problem for translators. It doesn’t present a problem for translators, since translators do not work word for word or term for term (see this earlier posting). They must understand the meaning of the semantic unit in the source text and then render the same meaning in the target language, no matter the part of speech they choose.
It is a different issue with machine translation. There is quite a bit of research in this area of natural language processing. Gerunds, for example, don’t exist in the German language (see Interaction between syntax and semantics: The case of gerund translation). But more importantly, gerunds can express multiple meanings and function as verbs or nouns (see this article by Rafael Guzmán). Therefore, human translators have to make choices. They are capable of that. Machines are not. If you are writing for machine translation and your style guide tells you to avoid gerunds, you should comply.
Because gerunds express multiple meanings, they are also interesting for those of us with a terminologist function. I believe they are the single biggest source of mistakes I have seen in my 14 years as corporate terminologist. Here are a few examples.
| Example 1: | Example 2: |
 |
 |
In Example 1, it is clear that logging refers to a process. The first instance could be part of the name of a functionality, which, as the first instance in Example 2 shows, can be activated. In the second instance (“unlike logging”) is not quite clear what is meant. I have seen logging used as a synonym to the noun log, i.e. the result of logging. But here, it probably refers to the process or the functionality.
It matters what the term refers to; it matters to the consumer of the text, the translator, who is really the most critical reader, and it matters when the concepts are entered in the terminology database. It would probably be clearest if the following terms were documented:
- logging = The process of recording actions that take place on a computer, network, or system. (Microsoft Language Portal)
- logging; log = A record of transactions or events that take place within an IT managed environment. (Microsoft Language Portal)
- Process Monitoring logging = The functionality that allows users to …(BIK based on context)
- log = To record transactions or events that take place on a computer, network or system. (BIK based on Microsoft Language Portal).
Another example of an –ing form that has caused confusion in the past is the term backflushing. A colleague insisted that it be documented as a verb. To backflush, the backflushing method or a backflush are curious terms, no doubt (for an explanation see Inventoryos.com). But we still must list them in canonical form and with the appropriate definition. Why? Well, for one thing, anything less than precise causes more harm than good even in a monolingual environment. But what is a translator or target terminologist to do with an entry where the term indicates that it is an adjective, the definition, starts with “A method that…”, and the Part of Speech says Verb? Hopefully, they complain, but if they don’t and simply make a decision, it’ll lead to errors. Human translators might just be confused, but the MT engine won’t recognize the mistake.
So, the answer to the question: “Can I use gerunds?” is, yes, you can. But be sure you know exactly what the gerund stands for. The process or the result? If it is used as a verb, document it in its canonical form. Otherwise, there is trouble.
Very interesting topic and useful links, which stress the importance of having an excellent awareness of grammatical structures, both at source and target levels. In particular, Guzman’s article highlights an additional problem that can affect not only MT but also human translators – in English the gerund is identical in form to the present participle, so without context it might be difficult for a translator to decide if a word ending in -ing is a verb describing an action taking place in that specific moment (present participle) or the name of a process/function/action (gerund).
The one-word string Indexing, for example, could be a message from an indexing service (present participle, informing the user that the indexing process is in progress), the title of a dialog box (gerund with nominal function, the name of a feature that allows the user to set the relevant indexing options), or the title of a Help topic (gerund with a verbal function, telling users how to perform indexing). Failure to understand the correct grammatical function might result in translation errors. You can find an interesting example here (in Italian): the title Closing sessions, a gerund, was interpreted as a present participle and translated as if it meant “the sessions are [currently] being closed”.
FYI, my book, The Global English Style Guide, contains a chapter entitled “Clarifying -ING Words” (which include gerunds, of course). I think it offers a more-complete discussion of the contexts in which -ING words are and are not problematic for translation than you will find anywhere else. And it includes lots of examples.
This was a difficult chapter to write because I found that there is no general agreement among grammarians or linguists about how these words should be classified. I had to develop my own way of thinking about and explaining them.
John, I am glad you mention your book. I am a big fan and can highly recommend it, although I do not own a copy (yet). In researching this particular post, I thought of it. So, thank you very much for the comment.
I should add, however, that machine-translation systems often do not translate gerunds correctly even in the contexts in which the gerunds are unambiguous. That has been my experience with out-of-the-box Systran, at least. I think the inability to handle unambiguous gerunds is a major failing of some rule-based MT systems. The presence of a gerund can cause an MT system’s analysis of a phrase or clause to break down completely! Perhaps Systran and other systems could be customized to deal with gerunds appropriately–especially if the content is XML, where the tagging could perhaps be used to influence the system’s choice of target structures. In my small pilot project, my source was a Word document, and I did not have the opportunity to explore the possibility of customizing Systran, other than adding terms to its dictionaries.