A few days ago, I saw an interesting claim in my Twitter timeline. A grammatical feature of language influences people’s suicidal behaviour. Here is a link to an account where you can find a link to the original paper. I find such a claim extraordinary and, to declare my hand, very suspicious. This post explains why.

The grammatical structure in question is future time reference (FTR). Basically, there are languages in which future is always explicitly encoded in language, and there are those in which it’s not and the present tense is used. The examples can be English in which in order to say that it will rain tomorrow, you will say:

• It WILL rain tomorrow.

And the verb ‘will’ is an explicit marker of reference to the future. In contrast, in German you will say something like

• Morgen regnet es.

in which future reference is achieved with the use of the present tense (you can find considerably more in the paper here). This means that English is a strong-FTR and German a weak-FTR language.

Of course, things as ever are more complex and my native Polish is a strange language in which you can say:

• Jutro będzie padać. (Tomorrow will be rain (verb)).

and

• Jutro popada (Tomorrow rains).

Although there is a slight difference in the meaning, the second sentence suggests that raining will stop, Polish has two ways of expressing the future. On the one hand, Polish uses a complex future tense (with ‘will be’ + infinitive or past tense) in the case of imperfective verbs (basically, those suggesting the action does not end) and, on the other hand, a simple present tense in the case of perfective verbs (the action ends) which has the function of referring to the future.

This duality of the future tense means that I have no idea how to classify Polish and some other Slavonic languages – they use both a strong and a weak FTR forms. I don’t think it’s a major problem for the idea, still a problem of sorts. Incidentally, the paper I review has Polish as strong-FTR language which is at the very least problematic. One also immediately wonders how reliable the remainder of the classification is.

Incidentally, if a claim is made about FTR having an effect on cognition, I am not certain why only FTR in main clauses is taken into account. If, for example, you take subordinate clauses, English can be a weak-FTR language too. For example,

• If you do this, I will be unhappy.
• I will be unhappy until you have done this.

where future is referenced via a present tense (simple and perfect, respectively).

Anyway, the author of the paper, Shuo Zhang, says that the mechanism is that

The potential mechanism is that, individuals will pay more attention to the part of reality that is emphasized by language. The lack of the distinction about time in a weak FTR language will shorten the distance between present and future and foster future-oriented behaviours of its speakers.

That, it seems to me, is counterintuitive. If you pay more attention to the part of reality emphasised by language, that would mean it is strong-FTR languages that should foster future-oriented behaviour. Am I missing something here? But, hey, let’s not quibble. I can just about see the idea that if future, sort of, happens now, it’s closer to us and that has some effect on how we prepare for it.

Let’s turn to suicide. Here is the central claim of the paper I am commenting on:

We find countries with primary languages having obligatory marker of future tense have lower suicide rates.

The explanation is:

There are two mechanisms to understand how future time reference influences suicide decisions. The first mechanism is time distance-based, in which individuals speaking languages encoding future event obligatorily in grammar are always reminded of the distinction between present and future. That is to say, at the time point of t, the pain felt by weak FTR language speakers is larger than strong FTR speakers, as the marker for future event zooms in the disutility that will occur in the future. The other mechanism is associated with the discount factor. The absence of marker for future events leads to a lower discount factor as the speakers of weak FTR language tend to equalize the sense between today and tomorrow.

The way I understand it is (and I might well be wrong, it took me some time to get it) that if language offers you a clear distinction between the present and the future, you might think that things will get better in the future. And I must admit that it does sound plausible. However, I find the claim about pain just plucked out of thin air, there is no evidence for this at all, but the present-future distinction seems legit.

So, why am I having problems with it? Well, there are a number of them. Before I get to the issues I really want to focus on, let me deal with a number of important issues that I don’t want to deal with.

1. The paper looks at countries as if they were all ethnically and linguistically homogeneous. This is far from being the case and it is well-established, for example, that migrants are at higher risk of suicide. Unfortunately, the paper does not deal with it. No thought is given to how many people do or do not speak the language, let alone to bi- or multilingualism, which even in Europe is not all that rare.
2. The paper does not address sometimes drastically varying methods of national collection of data. There are countries where suicide is still not a separate category for classifying deaths and is reported as ‘other’.
3. There are also countries where suicide is still illegal and one can assume that statistics at least can be lowered. This is the case of Cyprus, for example, a country where suicide is illegal and (therefore?) a country which has the lowest suicide rate in Europe.
4. The paper does not differentiate between suicide attempts and suicide, focusing only on suicide rates.

In my view, all these issues make the results considerably weaker. But let me deal with the main problem as I see it. The problem is that there is a very significant intra-language variation in suicides committed. Well, very significant variation in suicides in individual countries, neither Zhang nor I have data for speakers of languages.

The first difference is in gender. Considerably more males commit suicide than females. The highest ratios are in Eastern Europe. For example, male-female ratio in Slovakia is 8:1, in Poland is over 7:1, in Lithuania just under 7:1. Shuo Zhang does see the issue and says:

It suggests that females are more sensitive to the implications and molds that are conveyed by linguistic structure.

Bloody hell, that’s quite a statement. I mean, really? All that cultural, political, social, historical dynamics in which gender is entangled are dismissed by a simple sentence that females are more sensitive to linguistic structures? I must admit that the first time I read it, I immediately thought that of course men cannot be more sensitive to anything. We stand firm and will not be influenced by any stupid grammatical structures.

There is another problem with the statement. In my view, it would stand to reason that if women were more sensitive to linguistic structures, they should commit more suicides than men in weak-FTR countries. Let’s unpick it. In English, which is a strong-FTR language, i.e. leading people to commit fewer suicides, women, who respond better to that structure, should commit fewer suicides than men, and indeed, this is the case. But, surely, the reverse should be the case when weak-FTR countries are concerned. If the language leads people to commit more suicides, like German, surely, women should commit more suicides! Shouldn’t they? After all, they are more sensitive to the structure!

Unfortunately, either my logic is flawed or reality doesn’t want to oblige. In Germany, it’s still 3:1 for men (as, roughly, in Sweden). In Brazil, it’s 4:1 (here, I started wondering why Portugal is a strong-FTR language and Brazil is a weak one). You can find WHO’s data on suicide rates and ratios here.

I must admit that the language FTR and suicide theory started looking very very shaky for me after this. As does the claim of females’ sensitivity to linguistic structures.

The second problem I have is that the effects of future time reference seem to fluctuate with age. If you take the UK’s statistics (you can find them here), you will see significant changes in suicide rates across the lifespan. How do you explain the initial rise in suicide rates? Surely, as the linguistic structures get embedded, shouldn’t the rates at least flatten out? Then they drop, expectedly, they do take their time, don’t they? But how do they rise again? Is the sensitivity to linguist structures wearing out? Note, incidentally, that the drops and rises in suicide rates are both for males and females. Surely, if women are more sensitive to linguistic structures, this should not be the case. Should it?

And so, alas, I will not be buying the idea that one linguistic structure is responsible for suicide rates. A closer examination of it makes it fall apart, in my view. In addition, I am baffled by the author’s claims to causality, there is no evidence for it at all. And the article that started it all (here is the link again) is very cautious about claiming causality. Yet, the author of the paper I am reviewing is not so cautious at all. Language structures affect the rates of suicides. That’s a claim I really find too much.

But I want to finish with another extraordinary claim. Both Shuo Zhang and the author, Donald Lien, commenting on the article says (the link is at the beginning of this post) are also suggesting that languages and their structures should be changed. Lien writes:

Although it is difficult and costly to modify people’s language, it is worth noting that suicide is likely to be a rational behavior and affected by language FTR through the association between the way language encodes time and the forward planning behaviors of its speakers.

Are you for bloody real? Are you really saying that governments, academies of science, or some other agencies should now start changing grammatical structures of English, Polish or Dutch, because one person (or even a hundred) says that suicide rates are influenced by those linguistic structures? All in an effort to get people to think right! Gosh, let’s not limit ourselves, let’s make sure that the bad thinkers regret their thoughtcrimes!

I think I’d better pause here before I go too far. Thank goodness, I was still able to think all those things I haven’t written about.