Is the difference in cognitive load among folks who speak different languages is consistent or even measurable? Does it bias a developer toward any specific pattern/mistake/misunderstanding/bug? Are there subtle mental-model-building advantages to decoupling software commands from their linguistic meaning through a language difference even if it's initially more difficult? Does the presence or quality of English instruction affect adoption of development beyond the first few weeks?
Could be cool to explore these things to see how they might inform pedagogy or even language design. Then again, my relatively naive suppositions might lend entirely too much weight to it.
I also regularly can’t recall a german word, when there is an english one that is more precise and vice versa.
I always assumed it was just a random keyword for a programming concept. Same for “thread” by the way (also took me a while to understand how to pronounce it as I only read it).
> depreciate. to reduce the purchasing value of (money)
> deprecate. Computers. to cease supporting or recommending the use of (older elements, features, or versions of software)
(Source: dictionary.com)
Secondly the keywords don't really have any strong prior meaning to overload, so in a way they're more "pure" cognitively and easier to grok. The word "for" in everyday English really has nothing to do with "iterative loop with variable definition", the word "use" has nothing to do with symbol imports, but that's all it means to a non-English coder.
Now libraries are another thing. Typically they start using higher level abstractions that start to map to everyday constructs (eg, in a drawing library you might find things like "stroke width" which are trivial to understand for a native speaker)
The two main points I’ve seen are:
- it’s a daily pain to handle two vocabulary sets and lead to undue errors and bugs. When you have a “FishingBoat” class which related to a “boat” table and all the documentation points to “Bateau de pêche” and “barque” it’s 2x harder to follow things. Then people will accidently translate “bateau” to “ship” and it slips through a few reviews of mostly independent features, and you’ll have both words in the code base before you realize.
- it becomes a swamp the day a non native speaker touches your code base. For instance you need to share part of it with a company next door that happens to be the other side of the border. Or you ask a Romanian dev team to port your code to another platform.
This leads to business domain objects to be translated in English by the devs and they will make different or incorrect translations.
Then, all the stakeholders and requirements will be in French. All developers need to constantly map the real world names to the English names used in the code base, but that nobody else uses.
This leads to bugs, inconsistencies and extra cognitive load. It comes back to how hard "naming things" is.
I would disagree with this. A lot of programming keywords have enough context to help new people. I.e. "for" or "forEach" maps almost perfectly to English. "For each friend you know, call them and ask what their birthday is and what kind of cake they like".
friends.forEach(friend => { friend.askBirthday(); friend.askFavoriteCake(); })
While loops are the same way. "While there's still boxes on the truck, you need to be moving those boxes".It requires some slightly awkward contortions, but you can phrase English so that it somewhat resembles code.
"Use" and "import" both also have relatively clear parallels in English. import meaning bringing something from a foreign space into your own, and "use" being equivalent to using a tool.
English is my only real language (other than horribly botched Spanish) and the similarities to English really helped me when learning Python. Both in understanding and just simply remembering the keywords.
method: "a particular form of procedure for accomplishing or approaching something, especially a systematic or established one."
class: "a set or category of things having some property or attribute in common and differentiated from others by kind, type, or quality."
I agree that virtual is a little shakier, but: "almost or nearly as described, but not completely or according to strict definition."
I think as a developer one quickly realizes the value of sticking with the defaults (in this case English). Like configuring your system in English to be able to google the error messages; or Excel, which changes key shortcuts based on your language (ctrl+G for Guardar, to Save; ctrl+S in English); checking documentation in English or buying books in English because they are 2-3 years earlier than the local ones.
Using an US keyboard configuration also makes it much easier to type certain common characters in programming, although that might be too much to ask for :)
Companies with a poor engineering culture often name things in their own language, which makes an awkward and confusing mix because everything else is in English.
I have noticed interesting regional differences too. In France it is hard to get people to use English, but otherwise the grammar is mostly fine. But in Asia it's easy to get people to use English, but it's not easy to get the grammar right. For example they have a hard time with plurals when naming variables.
Of course it's just my data point and a broad generalization, because there are also very skilled developers everywhere.
It initially supported both English and French keywords: https://github.com/nlehuen/babble/commit/eb090fd3e80238166ac... - that was fun but for some reason I found it cringy and removed it (as if the rest was not cringy either).
Completely off topic, but I also implemented Hodor - "Its main objective is to teach programming to Hodor.": https://github.com/nlehuen/babble/tree/hodor
You just associate the words with what they do, but it doesn't matter what it's written. You just associate in the same way the operators +, - work in math. It just happens.