6

I am trying to analyze

Arthur doesn't discipline his children because he loves them

to show the structural ambiguity using phrase structure rules that precede X' rules, and that because is throwing me off.

hippietrail
  • 14,687
  • 7
  • 61
  • 146
Matt
  • 61
  • 1
  • 1
    Hello Matt and welcome! Can you elaborate on what have you got so far and why is the "because" throwing you off? (You can edit your own question by clicking edit, below your question text.) – Alenanno Mar 20 '14 at 09:19
  • Are you sure it's a structural ambiguity and not semantic or pragmatic? – jlawler Mar 20 '14 at 16:46
  • @jlawler, this isn't my area of expertise, but there definitely seems to be at least a structural ambiguity. Doesn't the meaning depend in part on whether the negation scopes over 'discipline his children because he loves them' or just over 'doesn't discipline his children'? In which case the semantics of 'because' might be a bit different in each case... – musicallinguist Mar 20 '14 at 21:06
  • Scope ambiguity is not necessarily a structural phenomenon. It has to do with, among other factors, stressed focus/bound constituents, movement rules, conversational implicatures, and presuppositions triggered by lexical items. Few if any of these have much to do with structure. Note that the ambiguity would not occur in speech, and that virtually all written sentences are multiply ambiguous because intonation and stress are not indicated. I mean, I can tell you how *I'd* analyze it, but it wouldn't make anybody's professor happy. – jlawler Mar 20 '14 at 21:18
  • I know nothing of the topic (sorry), and I do not know the definition of structural ambiguity. But the ambiguity is whether the negation is on the first proposition alone, or on the complete sentence, i.e. on the causality relation. I think "because"is not the only conjunction that creates this kind of ambiguity. But I did not find it easy to imagine another similar example (not as crisp as the previous one): Arthur doesn't discipline his children whenever they are disobedient. I thought that having more examples might help resolve the problem. – babou Mar 20 '14 at 21:37
  • Right, more data is always welcome. Does moving the because-clause change the ambiguity? "Structural" ambiguity usually refers to syntax, but scope is a logical phenomenon. – jlawler Mar 21 '14 at 00:36
  • 2
    @jlawler in my experience it does not make professors happy to make the ambiguity only in writing point during the lecture! –  Mar 21 '14 at 01:28
  • Another example (hoping it helps): I do not call to make you understand. All examples seem to have an undertone of causality (not always in the same direction). - - - - - -
      • cc @jlawler: What I mean about structural ambiguity is that I

    understand ambiguity from a formal language point of view, but syntax in NL theory may go beyond this, thus outside my own area of knowledge (cf. X' ).

     – babou Mar 21 '14 at 10:55
  • @jlawler point taken. I am no stranger to the issue of syntacticians making structural claims based on examples that don't take intonation into account; prosody is my area of expertise. But for me there is a possible intonation for this sentence (in which the speaker is not going out of her way to disambiguate the meaning) that still has the two possible scope readings. – musicallinguist Mar 21 '14 at 12:10
  • 1
    @jlovegren for the record, I wrote a term paper for a syntax class in which I debunked a previous claim in the literature about island effects in Japanese by taking all of the supposedly ungrammatical examples from the previous literature and presenting them orally to native speakers, who ended up judging them all to be perfectly grammatical. And my syntax professor was thrilled! – musicallinguist Mar 21 '14 at 12:15
  • @jlawler Is this question a real problem, or is it an exercise? It amused me, but it is not my area at all, and I cannot tell. Still I would think such issues have been pretty much explored. I wonder what characterizes these ambiguities (even if intonation can make a difference: the fact that intonation may be needed is significant by itself). – babou Mar 21 '14 at 22:44
  • Intonation is always needed, and if it's not present in the signal it will be inferred by the recipient. And it's a real and fairly famous example sentence, illustrating a real ambiguity. The question is the number of presuppositions one needs to make in order to describe the ambiguity. – jlawler Mar 21 '14 at 22:58

3 Answers3

3

I do not know that I have a proper answer, but I would like to summarize my findings. Recall that I am not a native speaker, and it may impact my perception of the language. Also, I am not taking in consideration other aspects that may help disambiguation, such as prosody.

Here is the original example, and two others I found, with different conjunctions:

Arthur doesn't discipline his children because he loves them.

Arthur doesn't discipline his children whenever they are disobedient

I do not call to make you understand.

What is common to all three examples is an undertone of causality in the conjunction and the negative form. The issue is whether the negation applies to the whole sentence, thus dominating (?) the conjunction, or only to the first proposition, which changes the meaning.

Regarding syntax, and I am considering only formal languages, I am not sure I like to define it with a Context-Free Grammar.

Here is a suggestion (approximately):

S -> NP VP
S -> NS NCoord S

The NS would be defined so that it derives only on negative sentences, and NCoord would derive only on conjunctions that "can be negated" and thus cause the observed ambiguity.

However, I find this a bit awkward, and I wonder whether it would not be more natural with a Tree-Adjoining Grammar.

Then, the question remains of when such an ambiguous analysis is acceptable. What are the conjunctions that allow the observed ambiguity. We have so far examples involving exclusively conjunctions with a causal undertone, though the direction of the causality may vary.

I found another example, without causality, which may shed some light on this. Unfortunately, it is based a conjunctive form that is understandable, but rarely used (according to my search engine).

He does not speak and thinks like an adult

He does not speak and think like an adult

In the first case the subject is mute, while in the second case he is immature.

Unfortunately, people will usually say for the second case:

He does not speak or think like an adult

But let's ignore this, since the rare form is understandable.

Then, if you take the plural for the same examples, you get a single form:

They do not speak and think like adults

which is then ambiguous since there is no verbal inflection to tell the difference.

The interesting part, in this, is that the conjunction no longer has a causality undertone.

What is left in common between the conjunctions in the examples is that they are related to a logical operator, for which negation also has meaning.

Could it be that the distinctive character for relevant conjunctions is that they have an associated logical semantics, which naturally has a negative interpretation.

But I am not a native speaker, and you may be able to come up with better examples.

Community
  • 1
babou
  • 1,477
  • 11
  • 18
1

So if you are starting from your rewrite rules, you could either start from S -> NP VP or S -> S Coord S, then you would get two different bracketings of your phrase markers and still manage to insert all of the same words in the same order once you get to lexical insertion.

  • All examples I know have an undertone of causality. It may be because I am trying to mimic the first one. Or is it the reason why they work? The negation seems also to play an important rôle in the ambiguity, which is not accounted for in your suggested syntax. The syntactic (structural?) ambiguity you suggest could apply as well to any coordinated structure, but there is no semantic ambiguity in most cases, I think: "He does not eat when she comes.", or even "Arthur disciplines his children because he loves them.". – babou Mar 21 '14 at 11:01
  • The reason why they all "have an undertone of causality" is because CAUSE is the main predicate of both readings. Because, after all, means 'be (the) cause (of)', which is just a nominalization of cause (vt), which takes two complements, one subject, one object. The ambiguity comes about from different placements of NOT on this tree. – jlawler Mar 21 '14 at 15:50
1

I may be missing something but isn't this more a question of negation scope. In which case a simple case of bracketing would explain the difference:

[[NOT discipline] [because he loves them]]

[NOT [discipline because he loves them]]

You'd have to be more explicit about the X' problem.

Community
  • 1
Dominik Lukes
  • 10,588
  • 27
  • 49
  • I think that's more or less right. My preferred analysis would be that negation takes rigid scope and the ambiguity arises depending on where the because-clause attaches - Either inside or outside of the scope of negation. – P Elliott Aug 20 '14 at 18:37
  • But wouldn't a fixed scope obscure the difference in presupposition? In the first case the children are not disciplined and in the second they are. – Dominik Lukes Aug 21 '14 at 05:17
  • Well your analysis and mine both make the same predictions as far as i can see - only in your analsysis the because clause has fixed scope, whereas in mine negation has fixed scope. As long as the X because Y is in the scope of negation, the sentence is compatible with the truth of X. – P Elliott Aug 22 '14 at 14:03