I have read this question. My interest is not in the h-index, but in a colleague listing a paper in two different versions and expecting credit (and teaching release) for both. (They got both versions published in peer-reviewed journals, but one is a straight translation of the other --- I checked.) True, it takes time to do a translation, but the intellectual effort that goes into translating one's own paper is rather less than producing the original research. I think the colleague is behaving selfishly and in an intellectually dishonest way. Before I make a big deal out of it, however, I want to make sure I talk to the right person(s). Is this behavior as big a deal as I think it is?
Regardless of what one might call this behavior -- self-plagiarism, double-dipping, etc. -- your colleague's behavior is an example of gaming the system. It is a common behavior observed by dishonest and insecure academics in response to various metrics that are introduced to measure their productivity and performance. Some examples related to the one under discussion are:
Listing a paper legitimately published in two versions (one as a conference proceedings and one in a journal) as two separate listings in one's publication list and without pointing out the connection. See this recent question.
Splitting a paper into several smaller papers to artifically inflate one's number of publications while possibly decreasing the work's overall scientific usefulness and impact.
Strategically "saving up" work done in one merit review cycle to write it up in the next cycle due to a feeling that one has done "enough" work for the current cycle and feeling insecure about one's ability to continue to generate meaningful output in the future.
Choosing the journal one is submitting to based on Impact Factor, h-index or other arbitrary and mostly meaningless indices.
Not all of these practices are equally unethical or dishonest, but they share the common feature that the practitioner is letting their actions be influenced by factors that are tangential to, and in some cases opposed to, the academic's main goal of advancing scientific knowledge. They can also have the effect of distorting institutional decisions about who is performing good work, who deserves to be promoted or offered jobs, etc., and to the extent that that is the case, in my opinion they cross a clear boundary between ethical and unethical behavior.
In the case under discussion, based on your description there is no doubt in my mind that your colleague is behaving dishonestly. It is fine for him to publish his paper in two languages if he feels that that adds some value, but he must clearly indicate that the two publications are duplicate versions of the same work and not try to claim credit for both.
Let's split this into two different aspects: honest representation and bean-counting.
Your colleague has control of whether the publication is represented honestly. Any listing of the two works should clearly declare their relationship to one another, e.g.:
"A translation of this work into LANGUAGE is available as TITLE "
"This work is a translation of the original document TITLE"
Not making such a declaration is dishonest, essentially a more difficult to detect form of self-plagiarism, and should be called out as such.
Bean-counting to determine whether a translation counts as a separate publication, on the other hand, is the problem of the various organizations that want to compute metrics. As noted in various other questions on this site, these metrics are highly variable in how different organizations compute them, quite noisy, and should be regarded are a rough estimate at best. Thus, I would not recommend worrying about that aspect of it.
I assume that in the later translation paper, the article states something like "this is a translation of X", since that is standard practice enforced by editors. For the purpose of institutional evaluation (tenure, promotion, raises), the question is what rules have been established regarding summary documents like the CV. Standards for what goes on a CV and how it is presented are constantly in flux, generally driven by institutional requirements. 30 years ago (at least in my field), research output was reported in two groups, "publications" and "presentations", but since then, universities are increasingly requiring more nuanced subdivisions into book chapters, working papers, conference proceedings, peer-reviewed journal articles, translations, and so on, at least for promotion-type purposes. While my university required faculty to submit a current CV annually, it only required the minutely-subdivided CV for promotion purposes. Department chairs vary substantially in their documentation requirements for the annual raise-review, sometimes not even asking for the titles of papers published, just the number, or sometimes including a paragraph of general discussion regarding research activities.
I disagree with Jakebeal's assessment that it is dishonest to not label a translation in listings like the CV. Dishonesty means that there is an unquestionable and known rule that something must be done, yet it wasn't done without openly declaring resistance to the rule. There is no such rule. It would not be irrational to create such a rule, as universities have done regarding the division of the CV into more labeled categories, but it would also not be irrational to create a rule that requires each article to contain contextual information such as "This study repeats the procedures of papers A, B, C using a new dataset" or "This study repackages the ideas of paper Z more efficiently". When a scholar keeps saying the same thing over and over, we don't claim that he is dishonest, we claim that he is boring or something along those lines. Accusations of "dishonesty" should be held in check, until it is clear that there is an attempt at deception, rather than a disagreement about what standard one should follow.