8

A recent question posed by another user observed that the following sentence is ambiguous:

 (1) Arthur does not discipline his children because he loves them.

This sentence can mean either that Arthur refrains from punishing his children because he loves them or he actually does punish them, but he does so for some other reason than that he loves them (perhaps he just wants to be cruel). The two readings can be understood in terms of the scope of the negation. The first reading has the causal adjunct clause scoping over the negation (because [not]), and the scond reading has the negation scoping over the causal adjunct clause (not [because]).

My question concerns a closely related sentence that has the adjunct clause fronted:

 (2) Because he loves them, Arthur does not discipline his children.

For me, the ambiguity has disappeared. The only reading in (2) is the first reading where cause scopes over negation (because [not]). I have two precise questions in this regard:

  1. Do others agree? Does sentence (2) lack the ambiguity that is present in (1)?

  2. If the answer is yes, then why should this be the case? Why is the negation capable of scoping forwards over the cause in (1), but incapable of scoping backwards over it in (2)?

Concerning the second question, note that negation is easily capable of scoping backwards in other cases, e.g. All that glitters is not gold -- not scopes backwards over all in this case.

Based upon the many helpful answers and comments below, I would like to give the original question a new direction. The answer to the first part of the question is apparently a strong "yes, sentence (2) is NOT ambiguous". The answer to the second part of the question remains somewhat open, although the suggestion and comments below present avenues for exploration. One of the avenues concerns the role of intonation, and another suggests a strong role for pragmatics. While I cannot discount these possibilities outright, I currently have a hypothesis I am entertaining that has not yet been disproven. This hypothesis is expressed as follows:

 Hypothesis
 Negation can scope forwards over arguments and adjuncts, but it can scope backwards
 only over arguments (not over adjuncts).

The data that have appeared in the question, the answers, and the comments support this hypothesis. I have not yet encountered an example in my own explorations that contradict it. Here are the examples that appear on this page:

 (2) **Because** he loves them, Arthur does **not** punish his children.

 (3) At **every** party, Fred did **not** dance.

 (4) He **definitely** did **not** do it.

 (5) #**Because** he loves chicken, Arthur does **not** eat.

In each of these sentences, the first bold operator is (in) an adjunct, and in each case, there is no ambiguity. The negation cannot scope backwards over the adjunct. In cases where the negation does scope backwards, it does so over an argument:

 (6) **All** that glitters is **not** gold.  (*all that glitters* is the subject argument) 

 (7) He was helping **every** student at **no** time. (*every student* is the object argument)

These sentences both allow the reading in which the negation scopes backwards over the preceding operator.

The long and the short of all this is that the original question can now be redirected as follows:

  1. Is there any evidence suggesting that negation can ever scope backwards over an adjunct?

If the answer to this question is no, then I think it has become possible to produce a coherent and principled account of the scope of negation.

Community
  • 1
Tim Osborne
  • 5,747
  • 1
  • 17
  • 40
  • 2
    In a word, constituents. The VP discipline his children because he loves them in (1) is no longer a constituent in (2), and thus it is not a possible focus of the not; this was one of the two interpretations, so removing it makes it unambiguous. – jlawler Sep 02 '14 at 19:05
  • @jlawler, thanks for the explanation. I must ponder that possibility. At present I'm skepticle that that will work in other cases. – Tim Osborne Sep 03 '14 at 00:29
  • 1
    There's quite a literature on this, as I recall, starting around 1970. Lakoff and McCawley both contributed, and possibly Ross as well. – jlawler Sep 03 '14 at 00:49
  • @Jlalwer, Can you provide a more exact citation or two. I'm really interested in this. Note, however, that your explanation fails in the following case: "...but discipline his children Arthur does not because he loves them." For me, the negation can now again scope over the cause despite the fact that "discpline because he loves them" is not a constituent. The explanation lies elsewhere. Linear order is somehow playing a role. – Tim Osborne Sep 03 '14 at 01:23
  • Of course. I assume it's a variant on the "cannot both precede and command" constraint of anaphor coreference. – jlawler Sep 03 '14 at 02:03
  • 1
    Maybe the disambiguation effect has to do with intonation and information structure. My intuition is that the reading of (1) where negation scopes over the causal adjunct requires something like a fall-rise intonation at the end of the sentence, which might designate both a rejection of something that has been implied by the preceding discourse and an anticipation of an assertion of the true cause. Maybe fronting the causal adjunct doesn't support this kind of discourse function. If this is the case, then the question is why negating the cause requires this kind of discourse function. – Shai Cohen Sep 03 '14 at 06:42
  • @Shi Cohen, while I agree that intonation plays a role in disambiguating scope ambiguities, I think the explanation for the phenomenon at hand is situated more concretely in the syntax. The negation seems to be incapable of scoping backwards over an adjunct, whereas at can easily scope forwards over an adjunct. But why should this be the case? Here's another example: "Fred did not dance at every party" vs. "At every party Fred did not dance". The first example is ambiguous (not [every] vs. every [not]), whereas I think the second sentence lacks this ambiguity. – Tim Osborne Sep 03 '14 at 08:39
  • "The negation seems to be incapable of scoping backwards over an adjunct, whereas it can easily scope forwards over an adjunct". If I knew of tests for what you call "adjunct", I might be able to help. But I think it's straightforward perception. Negation is not anaphora -- it's an operator, like quantification and modality, and it applies to what follows it because it turns on a perceptual switch that can't apply to material perceived before it was activated. Except for material in the speech buffer, which is usually the last big unparsed constituent or two. That's your "backwards" scope. – jlawler Sep 03 '14 at 14:25
  • @Jlawler, my understanding of adjuncts is standard. Most temporal, locational, causal, concessive, consecutive, modal, final, etc. expressions are adjuncts. They are adverbial expressions. Many PPs are adjuncts as well as many subordinate clauses introduced by a subordinator (=subordinate conjunction). In the example in the question, the adjunct is the causal clause "because he loves them". In the second example in my comment just above, the adjunct is "at every party". – Tim Osborne Sep 03 '14 at 22:32
  • @Jawler, my main observation is that there seems to be argument vs. adjunct distinction that helps define the scope of negation. A negation can scope backwards over an argument, but it cannot scope backwards over an adjunct, whereas it is capable of scoping forwards over both. I suspect that what you are calling a "buffer" is in fact more accurately characterized in terms of the argument vs. adjunct distinction. Consider this example: "He definitely did not do it". The negation is incapable of scoping backwards over the adjunct "definitely" in this case; the sentence is unambiguous. – Tim Osborne Sep 03 '14 at 22:39
  • @Jawler, if Ross or McCawley or anyone else has already pursued the sort of account of the scope of negation that is evident in my comments, I really want to know about it, and will be very thankful to anyone who can point me in that direction. Thanks. – Tim Osborne Sep 03 '14 at 22:45
  • Start with Horn, then. Or anything else in the bibliography in my encyclopedia article on negation. McCawley 1998 has a whole chapter (#17) on the syntax of negation. – jlawler Sep 03 '14 at 23:27
  • @TimOsborne Looking at the scope of negation in Japanese (different word order than English, i.e. what is "backwards"?) might yield some insight. Your general hypothesis (i.e. "backwards" scoping over ARGs, but not over ADJs) strikes me as right, and I can give you a "standard" (within one theoretical framework) account of why that is.

    Aoun & Li wrote a monograph (1993?) called The Syntax of Scope that would surely be informative, and Anna Szabolcsi (1999) has a paper (http://bit.ly/1rQqwui) that covers this in detail (covering semantics too), with a great bibliography.

     – Fred Sep 06 '14 at 22:07
  • @Fred, this is really helpful! Thank you so much. – Tim Osborne Sep 07 '14 at 00:00
  • In Japanese, it seems to me that negation can very well "scope backwards" in terms of linear order (using the terminology of the question) over an adjunct. For instance, the direct translation of (2) retains both meanings to me (though the relevant one would be singled out by intonation). You might want to check with a native speaker. – Olivier Sep 08 '14 at 08:42

2 Answers2

3

The citation is probably

Ross, John (1984) "Inner Islands". In Proceedings of 10th Berkeley Linguistics Society.

It is likely hard to find (although I'd bet jlawler has a copy available). It deals (inter alia) with the interaction of negation, adjuncts, and questions in scope/"extraction" phenomena. It's a fun paper.

One "standard" explanation is in terms of constituency, as mentioned, more specifically in terms of a hierarchically organized phrase structure tree and making use of the notion of c-command.

Alternatively, Kuno & Kuroda (I think?) had a paper circa 1996 or 1997 in Linguistic Inquiry describing/explaining inner (aka "weak") islands in terms of semantic and pragmatic properties. Also a good paper.

The chapter on weak islands in Partee et al (1990?) is also really good.

jlawler
  • 10,042
  • 1
  • 31
  • 55
Fred
  • 501
  • 2
  • 6
  • 1
    Thanks for the sources. I will indeed be taking a look at them. You seem to have a good overview of the area. Have you studied the scope of negation in particular? Are you available for further correspondence about the issue? – Tim Osborne Sep 03 '14 at 23:51
  • 2
    "Inner Islands" is available on Haj Ross's page, which I maintain. – jlawler Sep 04 '14 at 02:38
  • 1
    @Jlawler, thanks much. That's a fascinating little paper. My first reading of the paper suggests that there is some evidence there for my hypothesis. Ross seems to conclude that the argument vs. adjunct distinction is central to the phenomenon, although the terminology he uses it a bit different. This is promising. Many thanks! – Tim Osborne Sep 04 '14 at 05:06
  • This barely answers the question. Please summarise the paper and apply it to the sentences in the question above. – curiousdannii Sep 04 '14 at 08:08
  • 1
    @curiousdannii, the paper provides evidence for the assumption that the argument vs. adjunct distinction is part of the answer to the question. As far as summarizing the paper, I suggest you read it yourself. – Tim Osborne Sep 04 '14 at 11:34
  • @TimOsborne I was interested in (and did a teeny bit of work in) this area a long time ago. My syntax background is in late GB/early-to-mid-Minimalism (say, up to about 2005), so that definitely colours my approach to this problem. That being said, I think there are people who've done interesting work on this stuff within that subgenre of syntax. – Fred Sep 06 '14 at 21:45
0

The 'constituency' answer by @jlawler certainly explains the question but perhaps this is a good opportunity to question the whole constituency / negation scope approach to the problem. This may sound funny coming from someone who suggested it in the first place as an answer to the original question. But to explain the ambiguity of Arthur does not discipline his children because he loves them it seemed to be the most expedient approach.

However, just because bracketing the constituents seems to be a good heuristic, it does not necessarily mean that there is such a thing as a scope of negation. I would suggest that a construction / frame-semantic based approach to negation is far more fruitful and would avoid many of the issues Ross had to grapple with in his paper on "Inner Islands". Here are some off the cuff thoughts on this (if anyone has any references to construction work on negation - I'm sure there's plenty, I'd be happy to see).

First, negation with NOT is a construction with a form that looks something like this: NOT _____ with the meaning of 'NOT whatever'. Therefore, the example Because he loves them, Arthur does not discipline his children does not carry the meaning where the Arthur actually does discipline his children. That would have to be expressed as It is not because he loves them, that Arthur disciplines his children.

We need to return to pragmatics (or frame semantics - as I would call it) to explain the ambiguity. Compare the original sentence with:

Arthur doesn't eat chicken because he loves it.

Not only does that perfectly formed sentence not produce two meanings. It struggles to produce one. This is because in the frame of not eating chicken, the love of all chickens and not chicken as a dish is what matters. So we can have:

Arthur does not eat chicken because he loves chickens / animals.

or

Arthur does not eat his pet chicken because he loves it.

Of course, none of these produce a dual meaning because 'eating' something because of loving it, does not make sense in the blend of the frames of loving a living thing and eating a living thing.

Constituency has nothing to do here. Thus:

It is because he loves it, that Arthur doesn't eat his pet chicken.

is a lawful rephrasing of the previous sentence but

*It is because he loves it, that Arthur eats his pet chicken.

Although possibly, if we made the frame more plausible by placing it in time:

It is because he loved it, that Arthur finally ate is ailing pet chicken.

Similar procedure would help deal with many Ross's examples. For instance, he rules out

*How long didn't that concert last?

But when placed within a pragmatic context, it can be framed as in:

What did you say? How long didn't that concert last?

Or the same works for his two examples:

What did no imitation pearls touch? *What did no imitation pearls cost?

These two can be brought together with:

Did I hear you right? What did no imitation pearls cost?

Similarly to (but without the syntactic changes of):

What did you say did no imitation pearls cost?

What is the difference between touch and cost (if we indeed buy into that particular intuition)? Touching is an intrinsic property of all objects, whereas cost needs to brought into the frame with further elaboration.

Thus Ross's classification of adverbials can be rephrased as semantic frame specificity. They differ in how much work has to be done in order for the meaning to be brought into the frame and thus made available for negation.

Of course, there is also the construction _ NOT (intonation/irony) that produces sentences like.

I really love this itchy bright blue sweater my mother made me. NOT!

(PS: I understand, that this is very sketchy and welcome any comments of people who have thought harder about this than over a short breakfast.)

Dominik Lukes
  • 10,588
  • 27
  • 49
  • I agree that frame semantics (or relevance theory) come into play here. But I don't see why Arthur doesn't eat chicken because he loves it is not ambiguous. In the case of Arthur doesn't eat chicken because he loves it, (but because he's forced to) Arthur does indeed eat chicken, but not out of love. Is this reading unavailable? – Thomas Gross Sep 04 '14 at 09:46
  • Thank you for the interesting and insightful answer. I can agree with what you write in two areas: 1) pragmatic considerations help to choose one of the meanings over the other(s) in cases of scope abmiguity, and 2) some of the acceptability judgments Ross produces are open to debate. My current hypothesis about what is going on is, however, much simpler than your answer suggests. I have yet to encounter examples in which a negation can scope backwards over an adjunct. When it scopes backwards, it's always over an argument. In other words, there may be a much simpler answer to all this. – Tim Osborne Sep 04 '14 at 11:32
  • @ThomasGross You are absolutely right. The sentence does have both readings. This only underscores the importance of pragmatics (framing) in the way we parse syntax. Because I activated a particular framing, I simply could not process the sentence fully.

    It also shows a weakness in this "linguistics by made up example", it doesn't really give us insight into underlying structure but rather into a process of linguists activating different frames in their minds and seeing what pops out. That's not to say there's nothing to learned from brilliant minds like Ross's.

     – Dominik Lukes Sep 04 '14 at 13:59
  • @TimOsborne I agree the argument/adjunct distinction is helpful in identifying whether negation can scope backwards. The ironic NOT which scopes exclusively backwards was already mentioned - but it is outside the clausal structure, so easy to rule out, even if it is not too dissimilar from (7) - as far as I can tell. It is a more elegant guide than a construction-based solution to someone who struggles with an intuition about these things. But still, I wonder if treating both as separate constructions (thus avoiding the concept of scope altogether) describes what productive speakers do better. – Dominik Lukes Sep 04 '14 at 14:37
  • 1
    Yes, "scope" is a metaphor that's used for negation and quantifiers (though negatives have a "focus" element, distinguished by stress, while quantifiers are said to "bind" elements, also distinguished by stress). My own favorite metaphor is a negative word or construction "triggering" a "negative field" with some of the properties of an electromagnetic (e.g.) field -- varying field strengths, diminution with distance from source, paranegation (aka "secondary triggering" or "parasitic triggers"), etc. This field, in fact. – jlawler Sep 04 '14 at 14:54
  • @jlawler I like the 'negative field trigger' metaphor. But I wonder if it's negation that's doing all of the business there. We also need something to trigger the 'text world'/'mental space' in which the negation makes sense in the first place. I think that way we can also account for the variation in the accessibility of certain readings across individual speakers. – Dominik Lukes Sep 04 '14 at 15:06
  • Possibly. Certainly shared context is key. As for mental spaces, here's a paper by me and Karen van Hoek that we call "Negs In Space!". – jlawler Sep 04 '14 at 15:32
  • @Dominik Lukes, the ironic "not" is indeed a strange beast. My interpretation of it is that it is a separate predicate that takes the entire preceding sentence as its argument. It is equivalent to "It is NOT the case that". The precedening sentence is therefore its argument -- thus no exception to my hypothesis. – Tim Osborne Sep 05 '14 at 00:36
  • @Jlawler, you have obviously worked on negation quite a bit. Are you aware of any cases where a negation clearly scopes backwards over an adjunct? – Tim Osborne Sep 05 '14 at 00:40
  • @Dominik Lukes, the same question for you: Are you aware of any cases where a negation clearly scopes backwards over an adjunct? – Tim Osborne Sep 05 '14 at 00:40
  • I gather you mean by "adjunct" what I would probly call "oblique" -- not an argument. As to "scoping backward", that's your term. – jlawler Sep 05 '14 at 00:49
  • @Jlawler, Do you not understand my question? Let me try to rephrase it in your terminology: Are you aware of any cases where the negation negates a non-argument that precedes it? – Tim Osborne Sep 05 '14 at 00:52
  • Of any flautist who ever played in America, he was the most famous. The superlative most famous at the end triggers the any and ever in the preceding phrase, which is transparently preposed. I have no idea why this is good and the other is bad, however. – jlawler Sep 05 '14 at 00:56
  • @jlawler, Interesting example. I'm not sure it qualifies as a counterexample, however, since it has to do with polarity sensitivity rather than scope ambiguities. Polarity sensitivity and scope ambiguities have distinct distributional patterns, e.g. "Every student did not pass" (negation negates "every") vs. "*Any student did not pass" (negation does not license "any"). – Tim Osborne Sep 05 '14 at 01:26
  • Whatever. I'm not exampling or counterexampling anything. I still don't know what you're after, and I'm not much interested in new syntactic analyses. – jlawler Sep 05 '14 at 04:09
  • 1
    @Sad, such a great mind, but closed. – Tim Osborne Sep 06 '14 at 23:54