An explanation for the counterfactual/past-tense conflation in conditional sentences

Paper Title : "A statistical explanation for the counter-factual/past-tense conflation in conditional sentences"

..

7 June 2018

..

I hit upon why past tense morphology often (always ?) is associated with counter-factuality in conditional sentences about two weeks ago. Now I am feeling very pleased with myself. Probably I will never achieve anything quite like this for the rest of my life (I am 58 by the way). I look at the totality of human knowledge as a big pile. Every day people are adding to this pile, building on what others have layed down before. I see the information I will set out below, as my little contribution to this growing pile of knowledge.

..

Now it seems no professional linguist has discovered what the past tense counterfactual was all about. Two weeks ago I thought I must have merely rediscovered something that was already common knowledge ... but apparently not.

But why me ? There must have been thousands and thousands of manhours spent pondering this very question. Why did nobody else hit upon the solution. Perhaps it is because I think very visually. As you will see below, the explanation is very visual.

..

Metaphor doesn't come in to it. The only mechanism that is necessary is the usual mechanism of grammatical change. Namely ... that people learning a language give an affix/particle a meaning appropriate to the environments in which they find it. Nothing could be simpler.

..

..... One example of a particle's meaning getting refashioned

..

Looking up the meaning of a word on an online device or even in a bookform dictionary is a very modern habit. But I guess children have always ask the meaning of a word occasionally ...

Q. "Daddy, what does dour mean ?" A. "Well Johnny, its means something like unhappy."

Usually they are given a very rough and ready equivalent. But it is enough ... and after Johnny hears "dour" used 7 or 8 times in conversation, he has a pretty good idea when it is appropriate for he himself to use the word.

However most words are learnt by children without having to ask explicitly. Also I have never heard a child ask about the meaning of a particle (those short common words that have a grammatical meaning).

Q. *"Daddy, what does "if" mean ?.

Q. *"Daddy, what does "since" mean ?.

[A star, as above, before a sentence means "does not occur"]

Because of this method of children (and other language learners) picking up the meaning of a word from from the environment/situation it is usually found in. Well it can facilitate the spread of meaning a word or particle might have, into other areas. As an example of this, lets look at the English word "since". Nowadays "since" can be said to have two meanings, “a time span from an event in the past up until the present time” and "because".

But if we visited "since" in an earlier era, we would find it only had one meaning, ... “a time span from an event in the past up until the present time”.

At that time we would hear such expressions as ...

A) I haven’t eaten since breakfast.

B) Our local football team hasn’t been doing so good since Peter McCallum broke his leg.

Now the speaker of English only thought "since" in the above two examples meant “a time span from an event in the past up until the present time”. However a language learner, hearing "since" in a sentence such as (B) might think it meant "because". And this is exactly what happened. And a generation or two later, we would hear expressions like ...

C) Since you’re so clever, you work it out yourself.

Perhaps nowadays in speech, 75%* of the time you here "since" it means “a time span from an event in the past up until the present time” and 50% of the time it means "because. This makes 125% of course. Perhaps a quarter of the time it can mean either.

If one meaning of a word significantly outweighed its secondary meaning(in terms of frequency) it denoted the first meaning and connoted the second meaning (“denoted” and “connoted” are cognates by the way). However in the case of "since" this is probably not appropriate as both meanings occur equally frequently ... more or less.

When a grammatical word (particle) changes its meaning, often the original meaning is lost. Or alternately the two different usages can quite happily live on in parallel indefinitely. Presumably in the latter case, the ambiguity does not cause that much confusion.

More can be said about particles/affixes changing their meaning, but I think the above encapsulates the basics quite well.

..

* Of course this is a rough ballpark figure.

..

..... Terms used here

..

I call the clause containing "if" the antecedent, and the other clause the consequent.

These names are derived from Latin, meaning "what goes before" and "what follows".

But notice in English, "what goes before" can follow, and "what follows" can go before.

English indicates which is the antecedent and which is the consequent by having "if" to the left of the antecedent. Mandarin differentiates between the two by having "rúguǒ" to the left of the antecedent and "jiù" to the left of the consequent. (The English "then" can be thought of as equivalent to "jiù". However "then" is not mandatory, or even common, to the left of the English consequent.)

..

..... Preamble

..

What is this paper about ?

Imagine you heard a sentence starting "if you left tomorrow ..." and a sentence starting "if you leave tomrrow ...". Which form suggests the least chance of the verb (to leave) actually being performed ?

After a moments consideration, it should be obvious that the first form suggests the least chance of the action actually happening ? We can say that the first form has more counter-factuality.

Looking at "if you left tomorrow ..." we see that the verb is marked past tense, yet we can see the action is for tomorrow, the future.

This has always been a bit of a mystery. What exactly is going on here ?

..

Well I will throw some light on this later. But for now lets talk a bit about "if" and "-t" as in ... "if you left tomorrow ..."

"if" marks the antecedent of a conditional sentence. It tells you we are talking about a contingency. As far as I know all languages have a means of expressing contingencies. Having a means to express contingencies is just too important to do without. If somehow “if” were to be banished overnight from the English language I can see something being quickly roped in to express this role … perhaps “ink” <= “in case” or something similar.

And of course "-t" expresses past tense. WALS reckons that about 60% of the world languages have a past tense ... http://wals.info/feature/66A#6/-7.856/290.625 ... [Feature 66a]

Below I will argue that if a language possesses these two components, then it is highly likely that past tense in a conditional sentence will acquire a certain shade of counter-factuality. I fact if a language has these components but lacks this shade of counter-factuality ... well THEN you have an enigma that should be investigated.

Now with so many of the world languages possessing this shade of counter-factuality, a counter-factuality forged by exactly the same forces, it would be useful to give this counterfactuality a name.

I would suggest "the ngali* counter-factual" ... pronounced [ŋgali] ..

..

In the above diagram you can see that the consequent always follows on from the antecedent. That is shown by a thick black line. But how much does the speaker of the conditional sentence insist it is real. Well that can vary a lot. (For the moment lets restrict our self to telic events that will happen in the future ... we need to simplfy to see the wood from the trees.)

The speaker can have 100% confidence that the event will take place (A). In English the speaker would use "when" instead of "if" in this situation. Or the speaker can be 100% confident that the event will NOT take place in this Universe (C). He is talking about an alternative Universe. However since this alternative Universe has the same physical laws as this Universe, since the people behave similar to people in this Universe, there is wisdom to be learnt by hearing conditional sentences set in this alternative Universe.

But I would say overall the chances of a telic event happening in the future is about 50% (B) ... the future is always uncertain ... things turn up thwart the antecedent.

Lets say that the speaker of "if you leave tomorrow" gives the "leaving" about a 50% probability.

..

.* Other names I have considered are "the accidental counterfactual" and the "half-assed counterfactual". Some might ask why we need to give it a special name. If this counterfactual exists in every language with a past-tense, why not just call it the "past-tense-counterfactual". I think "past-tense-counterfactual" is not specific enough : it doesn't distinguish between "past tense" meaning and "past tense" morphology. But I am not sure. Any thoughts ?

..

..... The thought experiment

..

Imagine a pre-industrial society. The fastest way of getting about is by walking : the fastest way of sharing information is by word of mouth.

[The above is important for my argument : the below is just adding colour]

Lets make this society an isolated village … comprising of about 200 adults ... perhaps in Southern Sudan ... about 20,000 years ago. It is a hot climate so people spend a lot of their life out of doors. Lets imagine these people as 200 dots on a piece of paper. You can see these dots mingling/moving about ... a bit like seeing brownian motion in a petri dish.

..

..

These people have a language. Lets call it the thought experiment language (TEL). TEL is remarkably like English. It has a past tense (PST), non past tense (NPST) distinction. PST is represented by the affix “-ed”, NPST is unmarked.

No perfective/imperfective distinction or perfect. Lets keep it simple.

Assume here that “-ed” has only past tense meaning in all environments … maybe only grammaticized a few days ago (I know, unlikely, but please bear with me). There is no future tense … well they have a word meaning “intend” but it hasn’t been grammaticized yet. For human subjects with volition "intend" usually translates the English future tense. For non-sentient subjects, such adverbs as "soon", "tomorrow", "next year" suffice to show the future.

..

Building the scene

..

Old Umara is feeling poorly. She loves blood-pudding. Her grandson Puntu is to undergo the initiation into manhood rite quite soon. Only adults from Puntu’s family and the village shaman may attend. The rite is held in the shaman's hut. To encourage Umara out of her sick bed, Puntu's mother and elder sisters dangle the prospect of blood-pudding. One of the talking points of this community (the word on the street) is …

(1) “If Umara attends* Puntu’s event, they intend to serve blood-pudding.”

Now (1) is a valid statement right up to the time of the rite.

The event is to happen at midnight on the first day of the new moon. All the community knows this.

After the event (1) is no longer valid, but (2) is valid ... (2) = “If Umara attended Puntu’s event …” ... (We are only considering the antecedent here)

Now after the rite, news of it will spread … people meeting and chatting like they do.

Now lets go back to our piece of paper with these dots. Imagine if you will a cone under the piece of paper. The point of the cone meets the piece of paper at one point. This point on the paper is the position of the shaman's hut. Sheet of paper is 2-D, cone is 3-D, so we have introduced another dimension. This is the dimension of time. The point where the sheet meets the cone is the place and time of the event (the rite). The cone can be thought of as a “cone of knowledge about the event”. Of course the cone as an idealized shape, the actual shape of the “knowledge volume” will be quite irregular as it depend upon people going about their usual business and chatting together.

..

..

Anyway … the point I am trying to make is that in a short period of time, everyone in the village will know of the event, they will know if Umara managed to go to the rite. If she, in fact, attended the rite, (2) is obsolete. You would only hear (2) if Umara had been too sick to attend. Assuming that there was a 50% chance that Umara made it to the rite. To account for all eventualities we construct two universes*** ... one in which she attended the rite, and one in which she didn't.

..

..

The chart below I only produced because colouring-in 3D is too difficult for me. It is a 2D slice though the 3D volumes above.

..

..

It is worth emphasizing again ... IN UNIVERSE 1, (2) WILL NEVER BE SAID AFTER THE PERIOD OF IGNORANCE IS PASSED ... I guess this is the kernel of my whole proposition. ..

If you do the maths**** (that is compare the yellow volume to the light blue volume) you will find that (2) is pronounced in situations of ignorance 28 % of the time and in counterfactual situations 72 % of the time.

Hence it is inevitable in this community that a clause containing “if” + “-ed” quickly gets associated with counter factuality.

Now in TEL (as in English), “if” + “-ed” is nearly completely associated with counterfacuality. Hence the PST/NPST distinction has been lost. This distinction might be missed. Maybe there will be a future developement***** in the language to re-instate this distinction.

Remember at the beginning of this section I said "-ed" meant past tense in all environments. However after about 80 years (all the population has changed) this will no longer be true. In one environment (the antecedent clause) "-ed" will mean, more or less, counterfactuality

..

.* Notice that in TEL as in English. “if” plus a verb in indicative mood produce a verb with future** meaning. This isn’t surprising as the main point of conditional sentences is to evaluate contingencies ... to make plans for the future.

.** Just to complicate things a little. We can say that there are two types of verb. Telic verbs and Stative verbs. Telic verbs are verbs where an outside observer would see something happening. “drink” is such a verb. Stative verbs are verbs where an outside observer would not see anything happening. “believe” is such a verb. “if” plus a telic verb in indicative mood produce a verb with future meaning. “if” plus a stative verb in indicative mood produce a verb with future meaning, however this future meaning stretches down to the present (time of speaking). It is this tense that is the most pertinant hence it is said … “if” plus a stative verb in indicative mood produce a verb with present tense meaning.

.*** To account for this you could imagine that the universe split at the time of the rite. Resulting in Universe 1 where Umara attended the rite and Universe 2 where Umara failed to attend the rite. But don't worry your head about this, I am not advovating the "multiverse theory". An alternative way is to look at thing is that there is only one Universe, and some times future contingencies discussed happen and sometimes they don't. I am too lazy to dream up another contingency so I am going with the "multiverse" view. But it doesn't matter one way or the other ... this paper is about linguistics and not theoretical cosmology.

.****

.***** I am of course thinking of English using “if” + “pluperfect” to indicate “counterfactual past” … I am not so sure how to explain of this developement. If our TE language had a past perfect would the third conditional be formed at the same time as the second conditional or would it be a subsequent developement.

..

..... Assumptions I have made

..

4 constants that I picked

(a) Radius of the community is 4 km

(b) Speed of propogation of information is 1 km/hour

(c) Time that disinterest sets in is 16 hours

(d) A considered future conditional situation will actually materialize half the time.

..

3 things that I assumed

(a) The community area has uniform population density

(b) People are equally talkative 24 hours a day

(c) Disinterest is sudden instead of gradual

..

It doesn't really matter, but say "rate of utterance of (2)" = 0.2 per kmkmhour (i.e. utterred 0.2 times in a square km over a one hour period)

..

Now these are just assumptions I have built my model on. If you change any of these you will get a change in the amount of counter-factuality implied by "if" + "past tense". However it will not change the basic counter-factuality distribution ... i.e. something like this ...

... the large volume in "unknown" (the yellow) stops the "if" + "past tense" ever being accepted as 100% counterfactual.

..

Now of course I have used only one example of a telic verb using a third person (3S) subject in the antecedent and 3P subject in the consequent. I guess if first person or second person were in used in any of these positions the amount of "unknown" would be reduced. But I will say it again, if a language learner averages out the counter-factuality in all the thousands and thousands of different past tense conditional sentences they hear as they are learning the language, the counter-factuality they will come to associate with "if" + "past tense" will be around about the range I have given in the above diagram (45% - 90 %).

..

So a language learner brought up in Puntu's community ... if he wanted to say "If you leave tommorrow ..."/"if you left tomorrow ..." ... well low levels of counterfactuality he would of course say "If you leave tomorrow ...". As the counter-factuality of the situation increases, he has the choice of saying "If you left tomorrow ... ". In the interests of being precise he would probably switch over to saying "if you left tomorrow". He would use this right up to 100% counterfactuality. However if he wanted to impress on his listener that the contingency was 100% counterfactuality, he would have to add the tag " ... but you won't leave tomorrow", if he wanted to impress on his listener that the contingency was 95% counterfactuality, he would have to add " ... but you probably won't leave tomorrow",

..

..... How counterfactual

..

Now primary function of the English word "if" is to indicate contingecy. It's job is not to show counterfactuality, so it ranges over the entire counterfactuality range ... well nearly.

Below I have drawn 4 sketches graduated 0 => 100. This represents degree of counterfactuality. At zero you have a event/state which is dead certain, at 100 you have something which will definitely not happen.

"if" [bluish] is represented in the top sketch. You notice that it spans nearly the entire counterfactuality range. Only when something is held to be totally real or realizable does it disappear. In the extreme left of this continuum "if" would be replaced by "when" in English.

This division is not really necessary. The German particle "wenn" [pinkish] is shown in the second top sketch. As you see it covers the entire continuum.

The third sketch represents "if" + "past tense". Notice that it does not go all the way to the right. The explanation for this is the considerable amount of "unknown" that contributes to its meaning (remember the volume coloured yellow in U1 and U2 (U = Universe).

..

..

The bottom sketch represents the Slovenian particle "da". Slovenian also has the particle "bi" for expressing contingency. I presume 'bi" would have a distribution similar to "if" or "wenn" if it was sketched, but with it stopping about where '"da'" starts.

Because of the distribution of "if" + "past tense" (its RHS cut-off point in fact), to make sure a proposition is understood 100% as counterfactual you have to add a "tag" to the to an "if" + "past tense" conditional sentence. For example, to "if you left tomorrow, you would arrive on Tuesday" you would have to add the tag " ... but you can't leave tomorrow".

Also because of the distribution of "if" + "past tense" (its range quite far towards the LHS), it is possible to cancel* "if" + "past tense" conditional sentence. For example "If you had enough money, you could visit Australia", could be cancelled by adding "... well you have enough money, you can visit Australia". This sort of cancellation can not be used with "da" [the LHS of "da" would have to spread a lot further to the RHS to make it cancelable].

Various languages have such strong counterfactuals as Slovenian. For example mandarin has "yaobushi" as oppose to the normal contingence marker ruguo. Usually these strong counterfactuals have other functions in their language. I suspect the strong counterfactual meaning is the derived meaning (as "because" is the derived meaning of "since"). I would expect these strong counterfactuals to be more prevalent in languages with no past tense (maybe these WALS people could map "strong counterfactual" against "no past tense")

The strong counterfactuals in the various languages that possess them, may all have come about by different processes and hence all have slightly different ranges (an interesting topic for further research), however all the "if" + "past tense" languages should have a very similar counterfactuality range. [Note to self : is this right ? is there more complications?]

..

.*But this cancelation seems a bit awkward to me, is this because of a "logical" problem, or is it because the left hand edge of "if" + "past tense" is so far from zero : near the limit of "cancelability". I don't know.

..

... Making it real

..

bla bla bla

The Slovenian counterfactual "da" can never be cancelled ... too big a jump.

..

... Making it unreal

..

bla bla bla

A tag would be unnecessary for the Slovenian counterfactual "da". In fact it would be ungrammatical.

..

..... Other considerations

..

If a language has an indicative/subjunctive distinction as well as a past/non-past distinction (i.e. a four-way split), inevitably the "subjunctive + past" verb form is associated with counter-factuality. Well this is quite easy to explain .... subjunctives are used in the conditional environment. My explanation covers the subjunctive as well.

There also must be an explanation of why, if a language has a imperfective/perfective distinction as well as a past/non-past distinction (i.e. a four-way split), it is inevitable that the "imperfective + past" verb form is associated with counter-factuality. I don't know about this one.

..

..... Post Script

..

There is something a bit strange in the "if" + "past tense" counterfactual in that two elements combine to give one meaning. And these two elements are not adjacent. Maybe some people subscribe to a linguistic theory where this is not possible. If you do follow such a theory ... well ... I suggest you amend your theory :-)

I hope I have explained why (by way of typical human interaction) "past tense morphology" (by itself straight forward and unambiguous) plus "if" (by itself a pretty straight forward marker of conditional sentence) have come together. And together have a meaning "around 70% counter-factual".

I deliberately set my thought experiment in a pre-industrial age. Whether things like mobile phones and people zooming about in planes and automobiles would affect the process described is an interesting question.

..

..... One more thing

..

OK ... this has nothing to do with counterfactuality. In my life I have had one other original thought about linguistics. It is a very minor thought ... more like half a thought or a third of a thought. Far to insignificant to publish by itself, so I will append it here. I hope nobody minds.

..

How “have” picked up an additional use.

In Western Europe over a thousand years ago, you would have a priest saying his prayers with his Rosary Beads. If interupted, he might say “I have 15 beads prayed (prayed = past participle). A farmer at that time might say “I have 5 fields ploughed (ploughed = past participle)[and two field not yet ploughed … for example]. “have” which originally meant only possession came to take on the connotation (alongwith the PP) of “the action is done and dusted”, “it has been done at least once” and “the state resulting from this action still exists at the present time”.

Interestingly this extention of the use of “have” happened quite suddenly. There was a re-analysis of the sentence. The adjective (past participle) was suddenly bracketed with “have” and considered the core of the predicate. I think oof this re-analysis a bit like these line drawings … at first glance you see an old woman looking down … on second glanse you see a young woman looking away into the distance.

This particular reanalysis happened all over Western Europe at the same time. Now at this time the Catholic Church was the only institution common to all Western Europe. Also monks and priest tied the whole area together by their frequent pilgrimages. Also at that time rosary beads were very popular … the midieval equivalent of the mobile phone? Could a cultural artefact been the point of origin for a major areal linguistic innovation?

..