Hacker News Reader

Comment by toasterlovin | Hacker News Reader

Heritability has a different meaning from how you are using it. A trait is heritable if a child displays the trait regardless of their environmental milieu. So, language is not heritable because the language you speak depends solely on the environment you grow up in; the child of English speakers will speak Spanish if they are raised in an environment where only Spanish is spoken. Height, on the other hand, is heritable, because, absent major malnutrition, the child of tall parents will be tall regardless of the environment they grow up in.

And regarding status and wealth: genetics probably play more of a factor in these than you think.

View on HN

Not sure I agree. Inheritance refers to flow of influence on properties from parent to child. I think anything beyond that high level definition requires additional qualifiers (e.g. genetic heritability). Inheritance is typically not deterministic, even in the case of genetic heritability.

For example, the statement that tall parents will always have tall children is incorrect, though there is a strong bias. I would be curious to see how the probabilistic linkage between parent and child height compares to that for language. My guess is that parent language has a stronger effect than 99.9% of SNPs identified in GWAS.

If you really think about the difference between the two inheritance scenarios you outlined, it's that one mechanism of influence occurs at a scale that readily admits direct observation and perturbation while the other doesn't.

I don't think we (humans) know anywhere near enough about how genes compare and interact with other factors to support an argument either for or against your final statement.

https://en.wikipedia.org/wiki/Heritability

Heritability has a very specific meaning. The first sentence from that article: "Heritability is a statistic used in the fields of breeding and genetics that estimates the degree of variation in a phenotypic trait in a population that is due to genetic variation between individuals in that population."

The key part in that sentence is "due to genetic variation between individuals in that population".

Yes, there is a very strong relationship between first spoken language of parent and first spoken language of child. But there is no genetic element to that relationship (Asian children adopted by American parents will speak English, not the language of their parents), thus first spoken language is not heritable.

> I don't think we (humans) know anywhere near enough about how genes compare and interact with other factors to support an argument either for or against your final statement.

There is a difference between knowing that an effect exists and knowing how an effect works. We know, via twin studies, that IQ and height are highly heritable (50-80% for IQ; 80% for height), even though we don't know all the genes responsible or the way in which they contribute. Similarly, wealth/financial status have been shown to be heritable. Siblings raised apart will have levels of wealth that are more similar than random strangers. Identical twins raised apart, even more so.

I find your confident view loaded.

How can you say what heritability means ontologically, when the only way of ascertaining it is through mathematical statistics that doesn't care about your distinctions between genetic or cultural origins of the effects. Unless very carefully controlled, which is apparently too much to ask form the crude tools used in these studies.

Specifically about height you are mistaken this is clear-cut. First there is influence if populace is not under closure. Immigrants and emigrants affect height. A study that controlled for this was taken on pretty stable populace of Norway. Since height is also perceived as attractive and matters in sexual selection it was no surprise that Norse rose 10 cm over a century, or 2.5 cm in a generation. For such an effect to work exclusively through selection though the average height of parents must have been several centimetre higher than the average height of populace! The magnitude of the effect is just too large to work solely through selection at that pace. Thusly it is to be concluded that height is not purely genetic but also an environmental adaptation, both captured in heritability.

Heritability is a term from genetics with a very specific meaning:

https://en.wikipedia.org/wiki/Heritability

A trait, such as height, can be heritable without being 100% heritable. On the other hand, you have traits, like the language you speak, which are not heritable at all.

Sorry, but language acquisition has been found heritable.

I am aware of the definition. I fear it is you who doesn't have the epistemological clarity. It is not words that define but procedures with which you arrive at your results. And they do not discern culture, environment, flesh unless you take an extra care.

How do you propose to measure heritability of the first spoken language? What would be the variable here? It is perhaps hard to define variance in the language you speak, no? Are you aware of any such study, proving as you say that language one happens to speak is not heritable at all (as opposed to obviously)?

To see about an obviously cultural trait you'd like the first spoken language to be, it would be perfectly possible to measure accent variance, and perform a GWAS or twin study on that to see whether this is heritable or not. I am not aware of such study.

The ability to produce and understand language, in general, is heritable. Speaking a specific language is not. Have you spent any time around the children of immigrants? They speak the language of their peers fluently, but often have a limited ability to speak and understand the language of their parents.

This is something different. As I already said acquisition of first, second &c. language, pace of this has been found heritable.

To have heritability at all you have to have variance. People having hands and legs: heritability not defined. Has there been a language study around such immigrant children you mention to see if their language is heritable? Or do you rule out heritability here because <words>, which has nothing to do with how heritability is actually measured whenever it can be (there being underlying variance). Or have you made up that unconducted example like mine with accent? I would gladly accept my wrong if presented with actual conducted study you got that fallacious argument (language: 100% not heritable) from. Or was that just your personal understanding?

You don't need to design an experiment, because it is trivial to show that the specific language you speak is not heritable by looking at natural experiments.

For instance, the United States has experienced waves of immigration from various European countries in the last several hundred years. Specifically, there have been large waves of immigration from Germany, Poland, Italy, and Sweden. Yet there is no meaningful population of German, Polish, Italian, or Swedish speakers, fluent or otherwise, in the United State. Also, most Caribbean islands feature extremely high degrees of African ancestry, but the language of these nations is whatever the language of the colonizer was (so Haitians speak French, Dominicans speak Spanish, Jamaicans speak English, etc.). Native Americans speak English and, as a result, their native languages are in danger of becoming extinct. American blacks do not speak African languages, despite being descended from Africans.

This is in stark contrast to skin color, which is highly heritable. So people descended from European immigrants have the same light skin as their ancestors. And people descended from Africans have the same dark skin as their ancestors.

there is no meaningful population of German, Polish, Italian, or Swedish speakers, fluent or otherwise, in the United State.

I can't believe you wrote that. Have you been to Chicago? I encourage you to get out your bubble, while admitting failure to reason with you. Parting rhetorical question: what about Hispanics, Asians? https://en.wikipedia.org/wiki/Languages_of_the_United_States

This is of course totally apart from the asked question and the notion of heritability, which is dear sir a compound statistical and not a causal notion where you can handwave something away because obviously.

Okay, I'm going to try one more time, despite your belligerence.

From my link on heritability:

> Heritability is a statistic used in the fields of breeding and genetics that estimates the degree of variation in a phenotypic trait in a population that is due to genetic variation between individuals in that population.

As best as I can tell, you are pointing out that children often speak the same language as their parents, therefore the language spoken is heritable. But that is an exceedingly naive interpretation, because the result is confounded by the fact that children inhabit a very similar environment to their parents. Somehow you need to remove the environmental confound.

What I am saying is that there are natural experiments where this has happened, as I outlined above. Black Americans who are descended from slaves have very similar genetics to their African ancestors (despite some admixture with Europeans and Native Americans). Yet the prevalence of African languages among black Americans who are descended from slaves is 0. Therefore the heritability of African languages is 0, since African genetics has zero relationship to African languages once you control for environmental confounds.

Now lets look at the German language. According to this[0] Wikipedia article, there are 44 million American who trace their ancestry to German immigrants. Yet, according to the link you shared on languages in the United States, there are 0.91 million fluent German speakers, just 2% of those descended from Germans. In this case, again, it is clear that genetics play an exceedingly small role, if they play any at all, in the prevalence of spoken German. For Polish, there are 9.5 million Americans descended from Polish immigrants[1], but only 0.54 million fluent speakers of Polish. Again, that points to, at best, extremely low heritability of the Polish language.

[0]: https://en.wikipedia.org/wiki/German_Americans

[1]: https://en.wikipedia.org/wiki/Polish_Americans

Why cherry-pick? Where is the Spanish? Is this a heritability study by someone who just outright denied any foreign languages being spoken in the USA?

The natural experiment was conducted by Omniscient Statistician? This was published somewhere?

You do realise that heritability has no sense outside particular time, particular controls and particular populace? There is no number to heritability (in particular not 0) unless you precisely specify on what probe (and that number is only valid to that specific probe). These are not constants but relative variables being measured.

Let me illustrate that with Flynn's effect of IQ increasing over the last century. A today's IQ test administered on 1950's populace would average below 100. In populace today heritability of IQ test-taking was measured to be something. If one were to combine these two populaces, the distribution's variance partition would change (the distribution would become bimodal, with new environmental factor: whether someone is from today or the 50's). Suddenly heritability would be severely diluted by the new environmental factor. Does that make existence of heritability suspicious? No. Does that invalidate your estimates? Yes.

Reply on news.ycombinator.com