5 ms·

Not a linguist, but I took some classes ;) I think the Sapir-Whorf hypothesis is a pretty imperfect fit for the question. Even the weaker form doesn't really g

Not a linguist, but I took some classes ;) I think the Sapir-Whorf hypothesis is a pretty imperfect fit for the question. Even the weaker form doesn't really gel well. The question seems more do certain modes of thought make you tend to think of certain languages, and not so much "do certain languages impact behavior". Basically, its generally formulated in the opposite manner, and like many things, cause and effect can not usually be reversed and still have the reasoning hold ;) Also, I grew up effectively trilingual as well. I simply associate the different languages with different areas of thought. Math/science is English, food/home-y things is Chinese (and er, well, Taiwanese+Japanese also food/home), and I really don't think in terms of imagery. So I think you're definitely in anecdote territory. (BTW: I actually count in Chinese, because its a tad more uniform and the words are shorter, but operators I use English.) I wonder if language learning methods will vary from person to person, because most of your advice totally disagrees with my experiences learning Spanish later in life. I learnt through rote memorization, but followed up with a conversation partner, so, to continue the lookup table analogy, I simply used the lookup table until it became so commonly used that it stayed in cache and not memory. Then over time, segments were pulled out and their structure transformed (length of cache value increased, keys switched to a general emotion), until more and more of it was "fluid". (The other possibility, which I'm very inclined to believe, is that describing your own learning process is a crap shoot. You find analogies you're pleased with, and selection bias takes over.)

whacked_new 18y ago
I take a medium-strong form of S-W. It makes sense in the connectionist model (which I also take a medium-strong interpretation). The strongest form says that concepts in one language are completely alien to another, such that if you don't know the concepts in one language, you cannot perceive them. This is obviously false. However, as related concepts are evoked by spread-activation, it is completely plausible that one language allows more efficient retrieval of certain concepts, such that very strong biases result. The famous example of the many Inuit words for snow comes to mind. It is true that a non-Inuit speaker is able to discern between different kinds of snow, but the language guides the mind in terms of where to look for details. Without this guide, it is easy to get lost and confused. Here's the familiar example for hackers: code in C/Basic/Pascal for a while, and switch to Python/Ruby (or vice versa). After you get used to the target language, and look back at the code right after the switch, you'll notice that a lot of programming constructs are out-of-place and "counterintuitive." This is, so to speak, Sapir-Whorf "at work." This is also why when LISPers say that you should learn lisp, because it will make you a better programmer, they are correct. :)
- bd 18y ago
  Wasn't Inuit snow words story just an urban legend? http://en.wikipedia.org/wiki/Eskimo_words_for_snow http://en.wikipedia.org/wiki/Eskimo_words_for_snow
  - whacked_new 18y ago
    My fault for not confirming the veracity of that story in particular, but I was using it as an illustration of accessibility to concepts in different language, independent of its truth. I just supposed that this story is widely known, but I could have (perhaps should have) easily used another example. There are several words in Chinese that describe taste without direct English translations that I know of (it is harder to sustain these ideas nowadays, given the quickness of word-borrowing like "umami"). One of them is "fish stink," which really isn't fish stink. A strong S-W will say the concept is inaccessible to the non-Chinese speaker, and, of course, that isn't true. In fact you don't even need to cross language boundaries to see the same effect: layman versus technical terminology reveals analogous differences in "perception." Me not knowing the name of industry colors does not mean I cannot perceive these colors (let's assume a single-sex population), but I will gladly group together large groups of colors into single words.
- pfedor 18y ago
  The comparison with programming languages is extremely misleading, because, unlike natural languages, programming languages do differ in the expressive power. An algorithm coded in Haskell or Lisp may well be 5x less code than an equivalent thing in C or Java. On the other hand, when a text is translated from one natural language to another the result is typically of the same length and typically there's even an almost one-to-one mapping between individual sentences. To anyone interested in the relations between the language and thinking, I recommend an excellent book "The Stuff of Thought" by Steven Pinker. One chapter is devoted to the Sapir-Whorf hypothesis, more specifically, to debunking it.
  - litewulf 18y ago
    This is probably more a function of the fact that languages change over time, so if a common task is hard to talk about, new forms will appear that make it easier. Programming languages see very little change (and even rarer are ambiguity or "non backwards compatible" changes introduced.) Stuff like this is why I became interested in PL theory in the first place actually.

whacked_new 18y ago

I take a medium-strong form of S-W. It makes sense in the connectionist model (which I also take a medium-strong interpretation). The strongest form says that concepts in one language are completely alien to another, such that if you don't know the concepts in one language, you cannot perceive them. This is obviously false. However, as related concepts are evoked by spread-activation, it is completely plausible that one language allows more efficient retrieval of certain concepts, such that very strong biases result. The famous example of the many Inuit words for snow comes to mind. It is true that a non-Inuit speaker is able to discern between different kinds of snow, but the language guides the mind in terms of where to look for details. Without this guide, it is easy to get lost and confused. Here's the familiar example for hackers: code in C/Basic/Pascal for a while, and switch to Python/Ruby (or vice versa). After you get used to the target language, and look back at the code right after the switch, you'll notice that a lot of programming constructs are out-of-place and "counterintuitive." This is, so to speak, Sapir-Whorf "at work." This is also why when LISPers say that you should learn lisp, because it will make you a better programmer, they are correct. :)

bd 18y ago
Wasn't Inuit snow words story just an urban legend? http://en.wikipedia.org/wiki/Eskimo_words_for_snow http://en.wikipedia.org/wiki/Eskimo_words_for_snow
- whacked_new 18y ago
  My fault for not confirming the veracity of that story in particular, but I was using it as an illustration of accessibility to concepts in different language, independent of its truth. I just supposed that this story is widely known, but I could have (perhaps should have) easily used another example. There are several words in Chinese that describe taste without direct English translations that I know of (it is harder to sustain these ideas nowadays, given the quickness of word-borrowing like "umami"). One of them is "fish stink," which really isn't fish stink. A strong S-W will say the concept is inaccessible to the non-Chinese speaker, and, of course, that isn't true. In fact you don't even need to cross language boundaries to see the same effect: layman versus technical terminology reveals analogous differences in "perception." Me not knowing the name of industry colors does not mean I cannot perceive these colors (let's assume a single-sex population), but I will gladly group together large groups of colors into single words.
pfedor 18y ago
The comparison with programming languages is extremely misleading, because, unlike natural languages, programming languages do differ in the expressive power. An algorithm coded in Haskell or Lisp may well be 5x less code than an equivalent thing in C or Java. On the other hand, when a text is translated from one natural language to another the result is typically of the same length and typically there's even an almost one-to-one mapping between individual sentences. To anyone interested in the relations between the language and thinking, I recommend an excellent book "The Stuff of Thought" by Steven Pinker. One chapter is devoted to the Sapir-Whorf hypothesis, more specifically, to debunking it.
- litewulf 18y ago
  This is probably more a function of the fact that languages change over time, so if a common task is hard to talk about, new forms will appear that make it easier. Programming languages see very little change (and even rarer are ambiguity or "non backwards compatible" changes introduced.) Stuff like this is why I became interested in PL theory in the first place actually.