Ser wrote: ↑29 May 2020 21:43
I find the construction easy to understand when I think of it as "Of the man, there exists a house of his", thereby retaining -(s)i as a modifier of "house"(!), meaning 'The man has a house'. If your struggle is that "Of the man... of his" seems horribly redundant to you, you may recall many similar constructions in polypersonal languages, and I should emphasize you do not need to have full-blown polypersonal agreement to have
some of it (e.g. Spanish has it for indirect objects but not direct objects, especially Latin American Spanish).
This makes a bit clearer to me my own earlier point, thank you. Basically, what I was trying to say is that we can distinguish two constructions, using approximate English translations to convey the gist:
"There is a house at the man"
"Regarding the man, there is a house"
In the first case there, we can say that the preposition 'at' (or 'with', or 'to', or case in some languages) semantically, in a certain language, specifically includes the concept of possession. Hence, no further marking is necessary (at least perhaps except in the rare cases where a plausible ambiguity arises between the possessive and physical meanings - but then again, even English 'have' is routinely ambiguous in this way).
In the second case, "regarding" does not semantically specifically include possession. "Regarding" only creates a semantic topic [not, of course, necessarily a syntactic topic!], so the possibility of ambiguity is great. Therefore there's a much greater need (not absolute, clearly, since some language do just use this sort of bare topic+existential for possession, as I understand it) to clarify possession in some other way.
I don't know Hungarian or Turkish. But the genitive case in many language has exactly this sort of vague 'regarding' function - a 'genitive of relation', we can call it. So I'm wondering whether it's a coincidence that the two languages given as examples of requiring the overt possessive marker use the genitive, while the two languages given as examples of not requiring or even forbidding the overt possessive marker use the dative and/or a preposition.
I wonder whether Turkish and/or Hungarian may have other examples of a broad relational genitive?
[and now I'm wondering how this sort of clause interacts with Hungarian's focus marking!]
-------------
Another way of thinking about the difference, if the above turns out not to be valid, is in terms of the syntax of possessive marking more generally.
Clearly, there is a continuum here. Some languages, like English, have virtually no necessary possessive marking. For instance, in English you can say "My spanner and wrench are over there. Could you pass me the spanner?" - whereas I believe some languages would require (or at least prefer) that to be "my spanner" in the second sentence.
But at the other end of the scale, in some languages at least some highly-possessed words can NEVER appear without a possessive marker attached.
So it seems reasonable to imagine that some languages may have a much stronger preference for overt possessive marking than others, and one would indeed reasonably suspect that this would be more likely in a language in which possessive marking had become morphological (since morphologically-encoded distinctions are typically (though not always) more obligatory than periphrastically-encoded ones). And again, perhaps it's not a coincidence that Turkish and Hungarian are both examples of languages with morphological possession.
So then I'd wonder: do Turkish and/or Hungarian show a wider pattern of requiring overt possessive marking where Latin and Irish (and English) do not?
And since the alienability of possession is lexical in the languages with the least alienability, I'd also wonder whether that was a factor here. In phrases like "I have a persecutor" or "the shop has an owner" or "I have a phone technician", where the 'possessed' noun is of relatively high independence (compared to houses, feet, children, etc), is the overt possessive affix still present? Or does an entirely different structure have to be used?
-----
Finally, I think I might be seeing the relevance of definiteness, even if I don't understand the point Omzinesy was making about it.
"My" is prototypically definite (which is why, for example, English doesn't let you say "the my cat"). ['prototypically' - as Ser has pointed out, in English it's not always truly definite]
A true existential clause, on the other hand, always has an indefinite subject. [in English, you can say "there is a house", but not "there is the house" - well, you can, but only as an ostensive, not as an existential]. [yes, you can say "the house exists", but that's not an existential in the conceptual sense - in calculus terms, it's predicating a (controversial) property, rather than positing a term]
Therefore you can't normally say something like "there is my house by me". Instead, for the existential, you need an indefinite possessive, equivalent to English "of mine". And indeed, Irish does apparently require "an X of mine" in semantically indefinite situations where English cheats and uses "my X". So this may be why they do not say "my house is by me".
So now I wonder: are the Turkish and Hungarian possessives really best translated as definite "my", or are they more like indefinite "of mine"?
If the latter, it may be that this is inherently less semantically precise and more versatile - "there is a house of mine" might introduce many different lines of thought, not just possession, whereas "my house exists" is rather more to the point! - and this may be why they require the (what looks like) double marking of the overt nominal possessor.
And again, this ties in to my earlier suggestion about alienability and obligatory marking. Because if "-1.POSS" is seen not as "my" (i.e. indicating a definite entity) but as only "of mine" (i.e. indicating membership of a class or sort), then semantically it's much more plausible that it would be obligatorily marker - who a thing belongs to is an external relation, but what sort of thing a thing is (including being the 'belonging to me' sort) is an internal characteristic, and the latter is more easily regarded as inalienable.
-------------
----------------------------
--------
I think a large part of why people hate syntactic analysis is that syntacticians have generally done a bad job at choosing their jargon (Chomsky did a huge disservice to linguistics when he came up with the likes of "theta-role" and "c-command", IMO), and arguably also at explaining their concepts and why they're useful at modelling real languages. People have a hard time learning to think in terms of trees, so it needs a good justification. It was also my own experience at university that I was given nowhere near enough tree examples from English and other languages when learning about syntax, and this lack of examples made learning everything so much harder.
I'm obviously not a linguist, but FWIW my objection to "syntactic analysis" in the academic sense is twofold. On the one hand, I've never seen it having any practical application. Grammatical descriptions of individual languages, and even comparison between languages, seems to have no need of it. Indeed, quite the opposite. I sigh when I realise that a paper is going to be 'theoretical', because it means that its wordcount is going to be devoted to whether this this should be analysed as part of this, or that, and whether X is really a Y or a Z, and how many phrases can fit on the head of this or that pin... while I'm much more interested in a paper where the wordcount is devoted to describing what does or does not actually happen in the language, in a factual sense. I am interested in syntax, and in syntactic analysis in the genuine sense, of actual languages, but Syntactic Analysis in the Neoplatonist sense just seems to be a constant unwelcome distraction from the real business of linguistics.
On the other hand, what a lot of linguists say about this sort of business is deeply confused and naive. They talk about one model being better than another, they talk about what they do as a 'science' - when really, they are engaged in metaphysics. Certainly at least it must be seen as a philosophical (or literary) project rather than a scientific one - they are using a specialist, literary language to describe the facts, they are not using the facts to test and falsify definite predictions. The word 'analysis' is kind of a giveaway! As, indeed, is the proliferation of 'models' despite very little disagreement on the actual facts - because very few of the models can actually be disproven!
And of course, I'm a philosophy graduate, not a scientist, so I have no inherent problem with all this. My objection is that 'linguists' are doing philosophy, believing themselves under no obligation to actually study prior philosophers in order to do so, refusing to admit they're doing philosophy, and generally doing philosophy rather badly. Regarding which I guess I have two specific objections: first, that they're engaged in speculative metaphysics - debating 'rationally' the nature of these invisible, undisprovable, causally superfluous things, 'grammars' and 'models' - which I have little time for; and second, that they're doing it for no apparent reason. When mediaeval philosophers speculated about angels, it was at least an important question for them (even if they had no way to answer it). When modern philosophers speculate about possible worlds and personal identity, it's just as vacuous, in my opinion, but at least I recognise that the questions are on important issues. But when linguistic philosophers speculate about parameters and heads, I don't know why anyone (even someone who wants to understand languages) should care - it's like debating smeeps and flurks.
Also, I'd argue Construction Grammar is not necessarily "the most realistic grammar theory", and that here you may be expressing some confusion between being metalinguistically aware and the grammatical structure speakers carry in their heads, which is clearly there to some extent from their ability to make grammatical judgements.
How is that clear?
I'm afraid I see this an example of naive philosophy. Specifically, I think that when Chomsky was borrowing the concept of a universal grammar from Wittgenstein, he took the word 'grammar' too literally, as meaning what a linguist means by it, as though we all had a big book of rules 'carried in our heads'. But this is actually not what Wittgenstein meant at all.
People make judgements, certainly. But it's naive, IMO, to think that judgements are performed by comparing the world to a pre-existing (in some strange mental space!) model, or following some objective set of rules by which we can spit out an 'affirmative!' or 'that is not logical, captain!'.
Wittgenstein points out that when we follow the rules of the game, it is always possible that we may encounter a situation for which the rules no longer provide a ruling - where either they underdetermine (there are no rules for this situation) or overdetermine (the rules conflict). In this cases, we must make, as it were, a pure judgement of our own. And this situations occur all the time with linguistic games - giving rise to, depending on their tricksiness, either lingering philosophical controversies, or cheap jokes and riddles to amuse children (or anywhere in between).
But once you admit the possibility of having to make what I've called pure judgements - ones that do not just abide by the rules - you eventually realise that you're having to do them all the time. This is due to what's known as Wittgenstein's "rule-following considerations". To simplify greatly: to follow the rule in a situation, you must know what the rule tells you to do in this situation. But of course, the rule does not list every possible situation explicitly. You derive the specific situational guidance from the abstract, general rule. But how do you do that? You cannot, of course, simply pluck something out of your imagination - if you interpreted a rule however you liked, on the spur of the moment, you wouldn't really be 'following' the rule. Your interpretation must be derived in a particular way from the rule. It must, in other words, follow the rules for deriving an interpretation from that rule. But what interpretation would follow the interpretation rules? You must deduce that from the interpreting-the-interpretation-rules rules. And so on, in an infinite regress. You end up with Lewis Carroll's paradox.
What we must therefore conclude is that people do not in fact deduce their actions and judgements from a pre-existing, objective table of instructions: they cannot do.
Yes, it's very common to hear from language learners that they never understood a thing about English grammar in school (whatever little they were taught) until they studied the grammar of another language. But it is also the case that when people hear non-native speakers, they can perfectly make grammatical judgements, correct them, and sometimes even figure out patterns and structures on their own
But this gives the game away. Yes, they 'figure out' the patterns, on hearing them. This is what we do with rules - we do not follow them, we use them to describe our actions. A rule is, in other words, a post hoc narrative for understanding ourselves. A rule is an identity - the identity of a form of life.
Of course, given that coherent forms of life - coherent practices, like German, or chess, or haiku, or wine-tasting - do exist, it's reasonable to ask how this is possible, if they are not in fact the result of obedient following of objective rules. If a wine-taster does not algorithmically produce their description of a wine from an internal rulebook, how do they do it? And why do different wine-tasters often agree? Equivalently, how do German speakers know when in which situations a particular case should be used, and why do they often use the same case as one another?
Analogy.
Wittgenstein gives the famous example of word definitions. For many words, it is difficult to give a single, unobjectionable definition. He gives the example of 'game'. It's easy to think of characteristics of games, but equally easy, he says, to think of counterexamples in each case. He cannot think of any single "rule" that "defines" whether something is a game or isn't. And yet he can want to say that something is a game. He can recognise that it is very like other games, alike enough that, at least for some purposes, it is reasonable that it should share the name. But it need not be like ALL other games, in all ways. He coined the term "family resemblance" for this. We can say that John looks much like his brother Bob, and that Bob looks much like his sister Alice, and that Alice looks much like her mother Charlotte. The family shares a certain resemblance. We may in general feel that this family look alike, in a way that they do not much look alike the members of the family of John's neighbours, who look quite different. And yet, we may not be able to produce a single definitive characteristic that all members of the family share. Indeed, some members of the family, like John and Charlotte, may not really look much alike at all, in pairwise comparison. But because they both, in different ways, look like other family members, they form, as it were, a cluster of appearances; John and Charlotte are united as 'looking like Andersons' not because they, in isolation, look like each other, but because they each resemble other Andersons - it's only the existence of the other, median Andersons that results in John and Charlotte being bundled into the same group. We are, as it were, extrapolating from our other experiences of the Anderson family, by analogy. You see John, you think of other Andersons, and you think "he looks like a member of the family!" And yet there is no single "rule" defining the appearance of all the family's members.
Likewise, when I judge whether someone has used a word or a tense correctly, I'm not comparing the usage to a gigantic tome of abstract rules I carry around in my head. I'm comparing it to other actual usages that I have encountered, and considering whether those other usages were 'correct' (socially approved') or 'incorrect' (socially disapproved). From a series of such comparisons, I reason by analogy as to whether this new usage will similarly be correct or incorrect. My judgement is therefore not so much rule-
following (moral) as rule-
predicting (scientific). When I say that that's not how the word is used, I'm issuing a prediction (I may also, of course, be attempting to bias the experiment for personal reasons, if I happen to like or dislike a certain usage for whatever idiosyncratic reason of my own). If the comparisons are too strained - the analogy too weak - or if different comparisons give different predictions, then I'm left not "knowing" (i.e. being reluctant to be called to the stand to testify) whether the usage is correct or not. Perhaps I say I just don't know; perhaps I say I can't quite put my finger on what the rule is - but that doesn't mean there's a rule pre-existing in my head but somehow escaping my conscious attention, it's just a way of expressing that I can't give a confident ruling. [our judgements do not follow rules, but issue them]
[language is like that game where you compete to give the most (or least) common answer to a question: your answer both predicts the correct answer, and helps to define it. But that doesn't mean the correct answer is carried in your head before you hear the question!]
Very often what happens, of course, is that we give one ruling at one time, when a certain analogy is in our minds; but later, when other possible comparisons are called to mind, other analogies becomes more persuasive, and we rescind our ruling, or even reverse it.