How Language Affects Savings Rate

[ooeei]

I watched this video awhile back, and didn't see a thread on it here except for one in the welcome section two years ago. It's an interesting analysis of how language, specifically the way language conveys the future, influences savings rate in a significant way.

Essentially, people whose languages make the future a discreet time separate from the present tend to save less than people whose languages don't differentiate between the future and present. They also tend to smoke less, use contraceptives more, and generally do things better for their future.

My girlfriend speaks Chinese and English, so this was especially interesting for her. His first point about naming of relatives is something I've directly learned being around her family, and is another interesting language component that is something that's just fundamentally different about the languages. It's easier to understand now how "____ doesn't really translate" can be true, language is far more complex than just ___ means ___.

https://www.youtube.com/watch?v=lw3YTbubyjI

[Optimiser]

This is something that I know very little about, but find very interesting. Language shapes how we experience the world, as someone who is only fluent in one language, I wonder if my human experience is somewhat limited.

[Financial.Velociraptor]

I remember reading a sci-fi novel in the early 90s I enjoyed immensely: The languages of Pao.  A planet's society was purposefully transformed over decades by introducing new languages to slowly change people's perceptions and values. 

[GuitarStv]

I remember reading a sci-fi novel in the early 90s I enjoyed immensely: The languages of Pao.  A planet's society was purposefully transformed over decades by introducing new languages to slowly change people's perceptions and values.

That was one of Orwell's central concepts in 1984, and why Newspeak was the official language of Ingsoc.  By purposely altering the way that language is spoken you can eventually control people's ability to even have certain thoughts.

[maizefolk]

So this seems like an extension of the Sapir-Worf model that the way we perceive the world is shaped by the language(s) we speak. Sapir-Worf isn't particularly popular in the linguistic community, but non-linguists like it a lot more.

Googling this turned up an awesome analysis where someone who was skeptical of the results tried a whole bunch of other possible language features using the same datasets as Chen, and found that the presence/absence of a future tense was a better predictor of savings rate than 143 out of 144 other characteristics of languages for which all the languages in the dataset had been tested.

http://www.replicatedtypo.com/whorfian-economics-reconsidered-residuals-and-causal-graphs/6011.html

[ooeei]

So this seems like an extension of the Sapir-Worf model that the way we perceive the world is shaped by the language(s) we speak. Sapir-Worf isn't particularly popular in the linguistic community, but non-linguists like it a lot more.

Googling this turned up an awesome analysis where someone who was skeptical of the results tried a whole bunch of other possible language features using the same datasets as Chen, and found that the presence/absence of a future tense was a better predictor of savings rate than 143 out of 144 other characteristics of languages for which all the languages in the dataset had been tested.

http://www.replicatedtypo.com/whorfian-economics-reconsidered-residuals-and-causal-graphs/6011.html

Yeah I found the number of factors he controlled for in the data to be pretty convincing. Based on the original thread in the welcome forum I think a lot of people didn't watch the video and assumed he just compared savings rates between countries with different languages.

That extra analysis is cool too, although I confess I don't really understand all of it. It does point out that this is all correlation, and proving causation would be a much more difficult thing (if possible at all) to do. Also of note is the only factor more strongly correlated is whether there is a separate way to say an event is completed vs it being ongoing. If there is a distinction between an event being ongoing vs completed, people save less. This doesn't seem totally unrelated to the future tense characteristic the study is based on, since they both have to do with perception of events over time. Perhaps cultures with defined "completed" events tend to focus more on short term goals in order to get that satisfaction from completing something?

[joonifloofeefloo]

...save less than people whose languages don't differentiate between the future and present. They also tend to smoke less, use contraceptives more, and generally do things better for their future

ooeei, was that a typo or do they in fact save less while doing all the other things well?

My own language/savings rates piece was around the words "expenses" and "expenditures". Using one vs the other changed everything. When I called an outlay an "expense", my body believed it was necessary, unavoidable. When I changed it to "expenditure", my body experienced a sense of agency and choice, and I easily stopped. So weird.

[maizefolk]

Yeah I found the number of factors he controlled for in the data to be pretty convincing. Based on the original thread in the welcome forum I think a lot of people didn't watch the video and assumed he just compared savings rates between countries with different languages.

A good random rule of thumb is that for links to text-based articles only 1/2 to 1/3 of commenters on any given website are going to have read the linked article. For videos it's probably more like 5-10%.

That extra analysis is cool too, although I confess I don't really understand all of it. It does point out that this is all correlation, and proving causation would be a much more difficult thing (if possible at all) to do.

Proving causality with observational (rather than experimental) studies is always an extremely hard and high bar to reach. But that fact that the correlation stands up to scrutiny, rather than looking like an artifact of p-value hacking plus an enjoyable story, is already quite cool. Whether it turns out causation going in one direction, or the other, or both savings rate and linguistic features are correlated with some third factor not in the dataset the ultimate explanation is going to be something fascinating.

Also of note is the only factor more strongly correlated is whether there is a separate way to say an event is completed vs it being ongoing. If there is a distinction between an event being ongoing vs completed, people save less. This doesn't seem totally unrelated to the future tense characteristic the study is based on, since they both have to do with perception of events over time. Perhaps cultures with defined "completed" events tend to focus more on short term goals in order to get that satisfaction from completing something?

A very Worfian interpretation of the results. ;-)

[ooeei]

...save less than people whose languages don't differentiate between the future and present. They also tend to smoke less, use contraceptives more, and generally do things better for their future

ooeei, was that a typo or do they in fact save less while doing all the other things well?

My own language/savings rates piece was around the words "expenses" and "expenditures". Using one vs the other changed everything. When I called an outlay an "expense", my body believed it was necessary, unavoidable. When I changed it to "expenditure", my body experienced a sense of agency and choice, and I easily stopped. So weird.

It was a typo. People who saved more smoked less, used contraceptives more, etc. My bad!

[shelivesthedream]

I was quite "meh" about this idea but I watched the video and WOW. I was extremely impressed with all the variables he controlled for. Causation or not, the correlation is amazing.