From my perspective, such information is essential for attracting new contributors, as well as giving prospective users an immediate idea of how much work it might entail to integrate the application into their system.
- A general aspect is the libraries used when running the application.
Some installations are restricted to a few selected toolkits, like GTK+ but not QT, or vice versa.
For an administrator who maintains a system and regularly updates its components over a long period of time, this may solely be a practical and not a religious question.
- Another aspect is the libraries used and prerequisites necessary in order to compile the application.
I.e. for users of a source-based Linux distribution it makes a big difference whether an application is written in C, or in Objective-C, because their compiler needs to support the language in the first place.
Other languages may make it necessary to install a huge stack of libraries.
The question then is, again, how much work you're willing to accept in order to compile this application.
- A different aspect is the intention to attract contributors.
Most developers have a preference for a small number of languages, or may simply lack experience in other ones. In order to allow a greater number of people to contribute to an application, some projects even split their sources into two different languages (like Wesnoth, Vega Strike, Naev, only to name a few). One of them for the core application (like C or C++), the other one for easy modification (like Python or Lua). Here's a link to a chapter of "The Architecture of Open Source Applications" that describes how and why this was done in Wesnoth.
- Finally, there's obviously a lot of bias for and prejudice against some languages.
I'll just say that I've seen horribly inefficient software written in about any language. If you ask me, for efficiency, the code quality of the application is much more important than the language it is written in.