how much does first langauge matter when perceiving (and reacting to) other languages?
If you asked me that in fourth grade, I would’ve said yes. My reasoning would have been simple: you often translate your thoughts into a certain language, after thinking your thoughts in your first language (first language here is whatever language your mind performs cognitive tasks in, apart from visuals - but usually those abstract visuals are also a product of your environment and the language you think in so it gets circular). This happens without your knowledge if you’re relatively used to speaking that language (usually in multilingual communities). If you’re new to this, you might want to think of how you would approach writing something in a language you’re just learning. You explicitly say something in the language you’re accustomed to thinking in, in your head, as you write down or say whatever you have to in this other, new language.
This would lead to inconsequential translational errors, including but not limited to the omission of articles, SOV misplacement, gender misappropriation (basic syntactic errors), and metaphorical transliteration. But does it affect pragmatic perception in any way?
Obviously it’s established science that the language you think in/ grow up with affects the way you view the world, but I posit (as many probably would have already) that being proficient in a language doesn’t necessitate actual proper understanding of what was set to be conveyed (John Searle’s Chinese Room problem all over again). You could take this a couple steps ahead and say no one would ever precisely know what someone means to say simply by what they’re saying since everyone has different cognitive directions and experiences, but let’s stay at the ghost language barrier hill just for a bit.
my experience: semantics are conveyed fairly well, but the pragmatics are a whole other story
I’m really old now and I don’t think of it as something as trivial as I deemed it in fourth grade. When you’re not as well-read in this other language, or as exposed to it, when you hear or read something in a second language, and translate it back to one of the default languages you think in, you are attaching the connotations of the words and sentence structure from this language instead of the source language. That way, intentions can be misplaced. One can seem evil, clownish, or angry even if they didn’t mean to. This goes both ways. If the source’s speech is in a language they don’t think in, they’re likely to get misinterpreted even if they meant no harm.
This stuff has the power to really stir things up a lot in a troubled household.
an example:
(this will get rather specific because specifics are quite important to gauge the connotations attached)
SCENE: Imagine you are watching TV and your playful adult family member who you are close to (relationships are important when trying to dissect what language/sentence structure/vocabulary is acceptable and appropriate) tries to keep blocking the TV in attempts to tease you and you’re done humoring them. Also assume this person knows English but thinks in Telugu, an Indian language.
English: Please don’t trouble me :/
Telugu adult hears: దయచేసి నన్ను ఇబ్బంది పెట్టకు
Dissecting the connotations attached: దయచేసి is “please” when used with an instruction or request, and is often a formal word not used in close relationships. It’s assigned a greater weight in communication. When you use this word, you are, in a sense, alienating the other person and rendering your closeness void by doing so - which is seen as an insult (probably because historically Indian families lacked boundaries). This is still fine, but then there’s ఇబ్బంది for “trouble”. Now this is a word that can be used two ways: first when you’re a bit annoyed or uncomfortable because of an itch or a weird sitting position OR– note that since I said uncomfortable– could have “harassment” connotations. Essentially it’s a euphemism for that kind of stuff in polite society when discussing affairs with people you aren’t close to. They’re making you feel REALLY uncomfortable. Since you’ve already alienated them by saying దయచేసి for “please”, and that set the tone for the rest of the sentence, you have likely insulted them.
A lot of that explanation would be easy to discard as overthinking but I believe it’s important to recognize and acknowledge the cognitive processes that dictate how you react to something, so we can all understand each other better.
etymological influence too?
‘దయచేసి’ is, literally, “to do దయ”, or dayā. This word, dayā, is directly borrowed from Sanskrit, and the Sanskrit word दया means sympathy or mercy. Dayā is associated with charity.
On the other hand, ‘please’ comes from the Latin word placere which means to please or make happy. You’d also find it used in sentences where one expresses appreciation or fondness for something, like ‘hoc gemma mihi placet’, which means ‘I like this jewel’ or literally ’this jewel pleases me’.
Now it’s getting a bit easier to see where these connotations may have arisen from.
I also find it interesting to note that the origins of dayā and please root from the same language family, but these words are part of two separate language families (Germanic & Dravidian). So my regards to the Indo-Aryans and William the Conquerer?
machines in the picture
Imagine a t-SNE or UMAP representation of a smallish multilingual language model. You would expect to find most translations of words in a similar area. When they’re not aligned on any dimension, I would assume my model just sucks. I know data miners would try to find other examples and try to ascribe meaning to them, and one of the dimensions surely would be attributed to etymology (or is it? do we know? I’m not looking up any research tonight). But it would make a bit more sense why
- the words are far apart in the embedding tensor: they are used differently in local texts
- the model isn’t using the word appropriately.