Wednesday, January 15, 2020

Real Language Examples I: Comparison

I have over the years read a lot of descriptions about linguistic structures. Seldom do conlangers ever even approach the level of intricacies that natural languages do, and there are of course natural reasons for this – a population of a million, or even just a few dozen will be exposed to more different real situations than a single conlanger will, and thus need to communicate more things.

Over decades or centuries, this may lead to established patterns that slowly shift around.


Anyways, looking at a few of these at some level of detail - and also discuss mistaken "models" for how they work - may be of interest. I figure I'd start out with two very similar words - Swedish än and English than.

This is not meant as me taking sides (although I think the side I am on with regards to prescriptivism will be clear), it is me showing just how convoluted grammar can be.

1. Etymology
Both of these words, funnily enough, are closely related to temporal adverbs - than originates as a spelling variant of then, än can still be used to signify still, although some speakers may prefer ännu for that.

2. Are they prepositions?
Some speakers (especially in the case of Swedish) object to the idea that they are prepositions, citing a supposed predicate that should be possible to insert. Thus 'A is bigger than B' really is 'A is bigger than B is (big)'. No speakers, as far as I know, deny that this construction can be used, though, and one can also compare things with different adjectives: 'A is longer than B is tall'. Similar objections sometimes are voiced in English, and thus you may have heard 'it should be "taller than I"'. And of course, case should follow concord in that case: "it made me taller than him", in case it made both of us grow taller. In this model, they are exclusively conjunctions.*

However ...

3. Are they subjunctions?
In both English and Swedish, they behave syntactically in ways that don't really fit subjunctions, but does line up with prepositions - and the pro-conjunction gang generally do not object to these behaviors, and sometimes even demand them. This requires some introduction.

3.1 English 'whom', but also preposition stranding
Some prescriptivist English authorities who otherwise prefer the subjunction model, demand 'than whom'. This despite it breaking their subjunction model. Also, the syntax of 'than whom' is decidedly unsubjunctionlike! Consider if "I am bigger than you" is really short for "I am bigger than you are", then "Than whom are you older" should be long for "Than whom is are you older". This seems to be a badly formed sentence even with the nominative who: "Than who is are you older" is just as bad English as the parallel construction is bad Swedish.

It gets even weirder to pretend the subjunctive model has any relevance when you hit it with stranding: "who are you older than (*is/?he is)?". And relative clauses would have a relative pronoun referring to a noun outside of the scope of a subclause!

The man than who(m) I am taller -> the man than who is (tall) whom I am taller  ... but "who(m)" refers to "man", which is not even in the same scope - "who" should now be inside the scope of "than". This is like having something like 'the man said that who came here yesterday he is sick' where 'who' in 'who came here yesterday' refers to 'the man'.

There are also other transformations that usually can hit prepositions, but can't hit subclauses that than can take. (This also holds for Swedish.) Swedish even more agressively strands prepositions than English does, and 'than' definitely can be hit by preposition stranding for most speakers of Swedish. No subjunction stranding exists. Also, subclauses have more restrictions on them during clefting than do prepositions, and 'än' seems to be able to fill both of those roles for most speakers.

3.2 Swedish reflexive possessive pronouns

A relevant piece of evidence in the case of Swedish is its reflexive possessive pronouns. Unlike western Germanic languages, the north Germanic languages kept a distinct reflexive possessive pronoun. This is used (mostly) when a third person subject is the possessor of some other noun in the clause. I will use the invented pronoun sy and syne for these in examples:
Manneni kör sini bil
The mani is driving syi car
Manneni kör hansj bil
The mani is driving hisj car

Jag fann mitt paket och hani fann sitti
I found my package and hei found synei
So, this gets relevant due to a few reasons. All Swedish-speakers have these in their vocabulary, but in southern Sweden, due to the Danish influence/substrate/superstrate(!?) many speakers will use the regular third person pronouns anyway. Immigrants also tend to do so, or in the case of Slavic immigrants use them in all persons. So, correct use of these has become a shibboleth. Native speakers of northern varieties usually have no problems.

However, edge cases exist, and comparison is one of them. So, two observations: än, by one of the models introduces a subclause. For nearly all  speakers, sin cannot ever be the attribute of a subject.

However, speakers who long back to the day when everyone spoke proper Swedish and knew when to use the reflexives right tend to get infuriated whenever anyone says 'than his X' rather than 'than sy X'. Even when 'sy X' is the subject. And you ask them whether they can accept 'than sy X is' and they say no, and wonder why you even ask something silly like that**, and they often fail to grasp that they're being inconsistent.

So... the same person often will demand that when comparing subjects, subject forms be used, but when comparing with reflexive possessors involved, the only way of getting a permissible subject in there is strictly forbidden.

3.3 What is the expected verb phrase?

The idea that than/än always serve to introduce subclauses further runs into problems with things like this little 'story': Alice is short, but Bob is tall. Alice concocts a potion that makes her taller than Bob. Is Alice now supposed to say
"this potion made me taller than him"
or
"this potion made me taller than he"?
In the subclause model presented by Svenska Akademiens Grammatik, the actual subclause model copies the entire main clause into the subclause, substituting only whichever constituent(s?) is provided after 'än'. Thus, we are left with two optional interpretations:
'this potion made me taller than it made him'
or
'this potion made me taller than he made me'
In fact, Svenska Akademiens Grammatik only permits for using the nominative on the comparand after 'än' if the compared noun in the main clause is the subject. However, teachers who never learned how this is supposed to work think the implicit verb is 'är' or 'gör' (is or does), and so think "proper grammar" prescribes 'he' and thus 'taller than he (is)', which by the rules in SAG clearly is not the case.

3.4 Swedish reflexive Verbs

Some verbs in Swedish are innately reflexive, or require reflexive marking when English would not: "I am washing up" would come out as 'I wash myself'. NB: in Swedish, reflexives do not require the suffix själv (cognate of self), but can take it. Reflexive pronouns are not formed using genitives, but accusatives, so essentially "me(self)", not myself.

So, which one are we to pick:
I wash me more often than he?
I wash me more often than him?
Both should, according to SAG, lead to weird meanings:
I wash me more often than he (washes me)
I wash me more often than (I wash) him

When asking a group of grammar nazis***, ** only a few out of about thirty responses even spotted the problem. Most called for 'he', rather than 'him', due to 'I wash me more often than he does'. This doesn't even, imho, really justify or specify anything. Than he does what? Wash me?

The standard reference work for Swedish grammar states about elliptical clauses with 'än' that they need to copy the entire main clause except the one constituent that follows 'än', be that the verb, subject, object, some adverbial or some prepositional argument. Thus ... Svenska Akademiens Grammatik demands the interpretation I gave above. With regards to reflexives, it does not state (in that chapter) whether copying the main clause also adjusts reflexives, but other chapters that deal with coordination and with reflexives imply that one cannot assume reflexives to remain reflexives over coordination except in the case of the explicitly reflexive 'sig' on both arguments, i.e. when there's only third persons involved.

In the group I asked, no one came up with any other solution than using the full verb phrase, or solutions that their own rules preclude. A few "liberals" that - much like me - accept än as a preposition also accepted 'than me' as the trivial solution, and that is a solution I can accept.

Now, I did provide my own conservative solution, that was accepted by most:
än han sig
than he himself

 I realize this also does violate some of the nitty-gritty of the Svenska Akademiens Grammatik's description of how subjunction-like elliptical än works. However, I am not entirely sure this is a subjunction!

I imagine this could be considered a rare example of a preposition that takes both a subject and an object, rather than a subjunction with ellipsis!

The fact that no one else came up with this idea seems to suggest to me that the subjunction-with-ellipse model is not genuinely present in people's mental grammar, and if it were, they'd faster have realized the problem with the reflexive verbs.

3.5 Impossible Verbs
In some constructions, there are no reasonable subclause to posit after than/än:
"Fewer than two people know this"
"No one other than you knew of it"
The main clause's verb phrase is 'know this'. What is the supposed subclause 'than' would introduce? 'Know this'?!?
*fewer than three people know this know this.
*No one other than you knew of it knew of it
*no one else than I/me was there
'Do'? 'Are?'
*fewer than three people are know this.
*no one other than you are knew of it
*no one else than I was (there) was there
I am aware some English speakers might prefer 'but' for some of these, but even there the question about potential subclause remains, as some speakers would prefer 'but I/he' over "but me/him". In Swedish, 'än' is probably predominant here, as 'utom' (but) requires some rephrasing, and even then doesn't really permit any actual subclause in these cases.
Superficially, 'do' might look okay, but if we switch to a different verb phrase, e.g. 'are running', we immediately find out what the issue is. The 'other than'-example is also immediately exposed due to a tense mismatch:
*fewer [than three people do] know this
*fewer than three people do are running
*no one other than you did knew of it
*no one else [than I did] was there

Swedish provides similar examples with 'more than' (fler/mer än), fewer than (färre än/mindre än), 'other than' (annan/annat/andra än)

Weirdly, even though I find no way of turning these nouns into subjects of VPs, I prefer the nominative here when using pronouns, as do most conservative speakers of Swedish.

4 Conclusion

I am not a big fan of prescriptivism****. However, in this case they've created some interesting issues!
  • They have provided inconsistent rulesets that are impossible for speakers to navigate. The only way to win is not to play.
  • They exist at tension with the usage in large parts of the speaker community.
  • Some prescriptivist-bent members of the speaker community have not properly understood the rules crafted by the authorities in the prescriptivist camp, and thus use home-crafted, different versions that may be superficially similar. These think they adhere to the strict rules, but fail to do so and create even more confusion.
  • 'Than'/'än' themselves by nature exist in a weird tension between the two word classes among almost all members of the speaker community.
  • The tension between different speakers' different mental models, the inconsistent ruleset and the strong beliefs about how it should be creates a fascinating grammatical situation, where also beliefs about the justifications for different case forms or different constructions are not consistently interpreted.
I would be very happy to see even a single conlang contain a single type of construction or a single word with a similar depth of complexity to it.



* Swedish grammar traditionally cuts conjunctions in two: conjunctions and subjunctions, where subjunctions subordinate one of the sides, i.e. almost always particles that introduce subclauses.


** I've done my research on this in a Swedish "grammar police group" on facebook.

*** The Swedish term is less offensive.

**** Although I generally am mostly in favour of a descriptive approach to language, but also of maintaining a literary standard (that does not force itself into people's daily conversations or light writing and light reading too hard), this might seem as though I am criticizing the conservative prescriptive language authorities very strongly - often, their advice is inconsistent, makes unjustified assumptions, and at least in bygone days even was phrased in a very unjustifiably elitist way (if someone is as inconsistent as prescriptivists often are, they do not deserve the right to lambast others for inconsistencies or failures to spot patterns or whatever).

No comments:

Post a Comment