PanLex documentation includes expository reference documents and comments within program code files.
Documentation on the PanLex project is located or indexed at this website.
PanLex code consists principally of programs written in Perl, SQL, PL/Pgsql, PL/Perl, and Bash, and configuration files for various programs by others. This code is generally equipped with liberal comments translating the functionality of each line into ordinary English.
The PanLex documentation would ideally be accessible to interested readers in all languages, particularly given the panlingual-communication mission of the project. One can imagine editorially translating the documentation into multiple languages. Systems supporting such site translation include:
Alternatively, one could convert the documentation into lemmatic form for PanLex-based automatic translation. The PanLem interface implements that strategy by making all of its control labels and messages lemmatic. All of them are expressions in an artificial language variety “PanLem”, whose translatability has been seeded with translations in the PanLex database into at least 6 other language varieties.
Users of the rest of the website, aside from the PanLem interface, can request no-cost automatic translation from various services, including Google Translate, Microsoft Translator, and SYSTRANet. Such automatic translation suffers from mediocre quality, but allows the PanLex personnel to change the website content at any time with immediate reflection in subsequent translations at no cost to either PanLex or website visitors.
In principle, PanLex personnel, instead of investing effort into custom translations of the website, can offer strategically selected corrections to the automatic services so that visitors will get better translations of the site’s pages in the future. Google Translate invites such corrections (with no promise on whether and when they will have an effect), and Microsoft Translator offers collaborative translation editing.
Another strategy for multilingual accessibility would be to reconfigure the documentation as an implementation of a content-management system that provides standardized and human-translated interface elements with content negotiation. Examples of such systems are:
We have begun to investigate how to let visitors know what automatic translation options exist and to choose among them. Although some users who want web pages translated have equipped their preferred browsers with personalized one-click translation controls, the PanLex project has a mission-based reason to show its visitors what the existing automatic translation services can do, preferably head-to-head.
By trial and error, we have found that it is not feasible to use, simultaneously, the embedded-code fragments offered by three different automatic translation sites. Embedding all three on the same page bloats the page visually (with various-shape, various-color widgets), makes the page load painfully slowly (each code fragment imports content from a remote server), and also produces bizarre results if a user requests translation from one service and then asks another service to translate the first translation (the services use framesets for output). We have not thought of a method to implement an inconspicuous and efficient multilingual, multi-service translation feature. Therefore, we have provisionally left the site’s pages without any translation functionality.
