11 th WoSSP, June 5-6 2014 Autonomous University of Barcelona 1 How to say ‘no’ to a ‘no’: polarity reversal in English and Russian Lena Borise, Harvard University [email protected]1. 1. 1. 1.1. 1. 1. 1. Some background: Negative polar questions Some background: Negative polar questions Some background: Negative polar questions Some background: Negative polar questions There are two ways of responding to a standard yes/no question (YNQ): with a ‘yes’ or a ‘no’, which correspond straightforwardly to the two alternatives, affirmative and negative. Answers to negative polar questions like (1), however, lack this clean-cut correlation, often being a puzzle for linguists and speakers alike: (1) A: Is John not coming to the party? B: No. (= No, he is not. OR = No, it is not the case that he is not coming= he is coming. ) Yes. (= Yes, he is. OR = Yes, it is the case that he is not coming OR *Yes) Cross-linguistic variation with respect to the meaning of the alternative replies is wide, and the ways in which languages cope with this communicative problem seem to differ equally widely. However, I will show in this talk that the polarity-reversal strategies used in languages otherwise significantly different, namely English and Russian, are syntactically exceedingly similar. I will concentrate on bare Yes and No replies, but full versions will become relevant later on. 1. 1. 1. 1.2. 2. 2. 2. Polarity and Polarity and Polarity and Polarity and polarity heads. polarity heads. polarity heads. polarity heads. • Most proposals on the structure of negative YNQs recognise that there are several sites within a clause where negation can be merged/interpreted. • According to Laka (1990), there are two Σ (polarity) heads in a clause, a high one in the CP-domain and a low one within the TP, c-commanded by T. • In a similar vein, Ladd (1981) tentatively proposes to differentiate between inner and outer negation in negative polar questions and Kramer & Rawlins (K&R) elaborate on this intuition in their latest work (2010, 2011, 2012), using Laka’s idea of the Σ-head: (2) (from K&R, 2012:36) • Holmberg (2013) distinguishes three heights of negation - high, low and middle - for English, and a subset of these - high and middle - for Swedish. High negation in his proposal is interpreted within
12
Embed
How to say ‘no’ to a ‘no’: polarity reversal in English ... · How to say ‘no’ to a ‘no’: polarity reversal in English and Russian Lena Borise, Harvard University
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
11th WoSSP, June 5-6 2014 Autonomous University of Barcelona
1
How to say ‘no’ to a ‘no’: polarity reversal in English and Russian
• Some languages have developed a special polarity-reversing particle: French si, German doch, Polish
owszem, Swedish jo, etc.:
(7) A: - Jean il (ne) parle pas français? French
John he NEG speak NEG French
Doesn’t John speak French?
B: - Non. ( = he doesn’t)
- *Oui.
- Si. ( = he does)
(8) A: - Kom Johan inte i tid? Swedish
came Johan not on time
Didn’t Johan come on time?
B: - Nej. ( = he didn’t come on time)
-*Ja.
- Jo. ( = he did come on time)
• Structurally, the polarity-reversing particle behaves like the two other answer particles: it is an
operator merged in SpecFocP that seeks to bind a Pol variable lower in the IP, but its polarity feature
is specified as [REV], that is, ‘reverse polarity to Aff’.
• What if a language doesn’t have a polarity-reversing particle though?
11th WoSSP, June 5-6 2014 Autonomous University of Barcelona
4
• This is where the different structural positions of negation become relevant. As the next section will
show, in languages that do not have a special polarity-reversing particle, the polarity-reversing
strategy depends on the ‘height’ of negation used.
2.2. 2.2. 2.2. 2.2. High negation: no special strategyHigh negation: no special strategyHigh negation: no special strategyHigh negation: no special strategy
• High negation, in Holmberg’s analysis, is merged within the CP; it is roughly equivalent to outer
negation in Ladd’s and K&R’s terminology.
• -n’t attached to the auxiliary verb as opposed to the main verb is (in most cases - see below)
indicative of high negation in English.
(9) - Isn’t John coming, (too)? (genuine question, positive bias)
- Yes. ( = John is coming)
- No. ( = John is not coming)
• Unlike English, Russian doesn’t have high negation. A possible factor contributing to the lack of high
negation in Russian is the fact that Russian doesn’t have auxiliary verbs, so there is no reliable way to
tell whether the negation is in the CP- or IP-domain - unlike in English, where both auxiliary verbs
and main verbs can bear negation (Isn’t John coming? vs. Is John not coming?) With modals and
passives, both of which make use of auxiliary verbs, Russian employs middle negation.
• In English high negation the negative head of a high negation YNQ is interpreted in its derived
position in the CP. In this way, the Pol head within the IP of a YNQ (recall that Pol is the highest
projection within the IP) doesn’t have a negative value, because the negation is interpreted higher in
the CP.
(10)
• Therefore, when the IP of the YNQ is copied into a YNA, there is no feature clash and the answer
particle operator can bind the unvalued Pol in the IP, just like in a non-negative YNQ (cf. (2)):
11th WoSSP, June 5-6 2014 Autonomous University of Barcelona
5
(11)
⇒ In other words, there is no need for a special polaritythere is no need for a special polaritythere is no need for a special polaritythere is no need for a special polarity----reversing strategy. reversing strategy. reversing strategy. reversing strategy. In this sense, high
negation is not interesting for our purposes, since YNQs with high negation behave essentially
as non-negative YNQs.
2.3. Middle and low negation2.3. Middle and low negation2.3. Middle and low negation2.3. Middle and low negation
2.2.2.2.3.1. English 3.1. English 3.1. English 3.1. English
• Middle and low negation in English share the same polarity reversal strategy, so I will discuss them
together. Let us start with middle negation.
• Middle negationMiddle negationMiddle negationMiddle negation is interpreted inside the IP, but with sentential scope.
• Unavailability of bare yes as a felicitous answer to a negative YNQ seems to be the main diagnostic
for middle negation in English. Variants with both not and -n’t attached to the auxiliary verb are
possible, though there is some inter-speaker variation in acceptability of the variants:
(12) - Isn’t John coming (either)?
(negative bias; unacceptable for some speakers, see below)
Or
- Is John not coming?
- No. ( = John is not coming)
- *Yes. (infelicitous/indeterminate)
- Yes, he ishe ishe ishe is. (= John is coming; emphatic)
“Some speakers of English find [the first question in (12)] sharply ungrammatical. There may be a
partial correlation with American vs. British English, with British speakers more often accepting the
construction as perfectly well formed. But there are American English speakers who accept it …
and some very preliminary investigation of mine indicates that not all British speakers do.”
(Holmberg, 2013:38)
• In middle negation YNQs, the negative head is interpreted within the IP, thus making the IP of the
YNQ bear Pol valued [Neg]:
11th WoSSP, June 5-6 2014 Autonomous University of Barcelona
6
(13)
• A bare yes is then an infelicitous answer, because of the feature clash between the [Aff] value of yes
and the [Neg] value of the IP inherited from the question (as discussed in section 1, see (6)).
• To reverse the polarity, it is necessary for the Pol of the IP in the YNA to be valued [Aff].
• We can make sure it is by copying (and therefore eliding) just the vP/VP from the YNQ to YNA;
because the negation is merged in the IP, the vP/VP alone won’t bear the negative value.
• This prediction is confirmed by the fact that the IP he is cannot be elided from Yes, he is in (12). This
is because ellipsis is possible under polar identity of the IPs of the YNQ and YNA; if they are not
identical, the IP in YNA needs to be spelled out in full:
(14)
(From Holmberg 2013:37)
⇒ Therefore, the polarityTherefore, the polarityTherefore, the polarityTherefore, the polarity----reversing strategy for reversing strategy for reversing strategy for reversing strategy for middle negation is to merge the affirmative particle middle negation is to merge the affirmative particle middle negation is to merge the affirmative particle middle negation is to merge the affirmative particle
yesyesyesyes to a [Aff] polarity PolPto a [Aff] polarity PolPto a [Aff] polarity PolPto a [Aff] polarity PolP,,,, and copyand copyand copyand copy from the YNQfrom the YNQfrom the YNQfrom the YNQ (and then elide) just (and then elide) just (and then elide) just (and then elide) just the the the the vp/vp/vp/vp/VP.VP.VP.VP.
• Low negationLow negationLow negationLow negation, , , , in turn, is interpreted within the vP/VP.
• Availability of bare yes as a felicitous answer to a negative YNQ is indicative of low negation used in
the YNQ:
(15) - Is John not coming?
- Yes. ( = John is not coming)
- No. ( = John is not coming)
-Yes, he ishe ishe ishe is. ( = John is coming; emphatic)
• Because the negative head is merged within the vP/VP, it turns out to be too deeply embedded to
clash with the Pol head at the top of the IP.
• In the YNA, the answer particle operator assigns an [Aff] or [Neg] value to the Pol in the IP, but that
doesn’t affect the contents of the vP/VP. Therefore, regardless of whether yes or no is used, the
negation of the YNQ is confirmed in the YNA - so called negative neutralisation in K&R’s terms.
11th WoSSP, June 5-6 2014 Autonomous University of Barcelona
7
(16)
(17)
• In order to get rid of the negative neutralisation effect, an non-negative IP needs to be merged into
the YNA.
⇒ In low negation, tIn low negation, tIn low negation, tIn low negation, the polarityhe polarityhe polarityhe polarity----reversing strategy isreversing strategy isreversing strategy isreversing strategy is thereforethereforethereforetherefore the same as inthe same as inthe same as inthe same as in middle negation middle negation middle negation middle negation ----
spelling out the IP with [Aff]spelling out the IP with [Aff]spelling out the IP with [Aff]spelling out the IP with [Aff]----valued Pol (within the IP).valued Pol (within the IP).valued Pol (within the IP).valued Pol (within the IP).
- Pridёt/Pridёt, pridёtPridёt/Pridёt, pridёtPridёt/Pridёt, pridёtPridёt/Pridёt, pridёt. ( = John is coming)
Will come/ will come, will come.
⇒ The polarityThe polarityThe polarityThe polarity----reversing strategreversing strategreversing strategreversing strategy for middle negation, as in (18y for middle negation, as in (18y for middle negation, as in (18y for middle negation, as in (18), is uttering or reduplicating ), is uttering or reduplicating ), is uttering or reduplicating ), is uttering or reduplicating (to (to (to (to
make the polarity contrast more prominent) the bare verbmake the polarity contrast more prominent) the bare verbmake the polarity contrast more prominent) the bare verbmake the polarity contrast more prominent) the bare verb.
11th WoSSP, June 5-6 2014 Autonomous University of Barcelona
8
• Recall that the characteristic features of low negationlow negationlow negationlow negation are the availability of yes as a felicitous answer,
as well as the negative neutralisation effect.
• It turns out that there is some dialectal (idiolectal?) variation with respect to the availability of low
negation in Russian (recall the inter-speaker variation with respect to middle negation in English
Holmberg reports):
(19) - Džon ne pridёt? (negative bias, falling intonation)
John NEG will-come
Is John not coming?
- Net. ( = John is not coming)
- %Da. (= John is not coming)
- Pridёt/ PridёtPridёt/ PridёtPridёt/ PridёtPridёt/ Pridёt, pridёtpridёtpridёtpridёt. ( = John is coming; emphatic)
Will come.
• Some speakers find da ‘yes’ infelicitous as a reply to the question in (19), which leads to the
conclusion that those speakers’ Russian lacks low negation and thus has middle negation only. This
pattern requires further investigation, but it is not very relevant for the present purposes, since the
polarity reversing strategy for middle and low negation is the same.
⇒ In Russian, like in English, tIn Russian, like in English, tIn Russian, like in English, tIn Russian, like in English, the polarityhe polarityhe polarityhe polarity----rrrreversing strategy is the same for middle and eversing strategy is the same for middle and eversing strategy is the same for middle and eversing strategy is the same for middle and low negationlow negationlow negationlow negation....
3. Middle and low negation reversal: 3. Middle and low negation reversal: 3. Middle and low negation reversal: 3. Middle and low negation reversal: ((((aaaan attempn attempn attempn attempt at) at at) at at) at at) a unified accountunified accountunified accountunified account
• It seems non-accidental that in Russian, like in English, the same polarity reversing strategy is
employed both in middle and low negation. The properties and the nature of this phenomenon
deserve further investigation.
• You could have noticed that in Russian, unlike in English, neither the answer particle nor the
arguments of the verb surface in the polarity-reversing YNA. In fact, constituents other than the verb
seem to be prohibited from appearing in a polarity reversing YNA (more on this below).
• The structure of a polarity-reversing reply in Russian, then, involves extracting the main verb and
merging it into the SpecFocP of a polarity-reversing YNA. The rest of the tree is then deleted (in the
spirit of Holmberg 2001, 2013):
(20)
11th WoSSP, June 5-6 2014 Autonomous University of Barcelona
9
• Stranded verbs as replies to YNQs are attested cross-linguistically and are usually analysed as
movement of the verb to the Foc projection with subsequent ellipsis of the rest of the clause.
Holmberg (2001) proposes such an analysis for Finnish, Morris-Jones (1999) for Welsh and Martins
(1994, 2006) for Romance varieties. Cf. also Merchant (2004) that thoroughly justifies the same
strategy - fronting and subsequent ellipsis - to derive fragment answers more generally.
• This analysis can be extended to Russian, especially since in Russian verbs can be fronted in other
contexts, when no ellipsis is involved:
o Thetic structures:
(21) Goreli fonari.
burned lanterns
There were lanterns burning. (Bailyn, 2012: 255)
o Corrective replies:
(22) A: Džon priedet na mashine?
John wil come by car
Will John come by car?
B: Priletit on, (a ne priedet).
Will fly in he but not will come
He will fly in (and not come by car).
• As the reviewer points out, a potential problem with this analysis is the fact that the overt correlate
of ellipsis in (20) is predicted to be grammatical, since ellipsis is usually considered to be an optional
phenomenon, but in fact the unelided version is nothing other than word-salad:
(23) *Pridёt Džon ne.
However, this fact - unavailability of the unelided full-version of the reply - is a part of a more general
problem with fragment answers, especially those involving vP/VP (as opposed to TP) fronting, as
identified, for example, by Merchant (2004):
(24) A: What is John doing?
B: John is washing the car.
Washing the car.
*Washing the car, John is.
According to Merchant (2004:697), the third reply in (24)
“…is distinctly odd in standard American English. For better or for worse, the reasons for this oddity (as
opposed to, say, their status in some British English varieties, where such VP-predicate fronting is fully
acceptable) are poorly understood at the moment. One safe conclusion that can be drawn … is that the
constraints that give rise to the oddity are constraints which can be (perhaps trivially) satisfied by non-
pronunciation, similar to a wide range of amelioration effects induced by ellipsis.”
• In both languages, therefore, in cases when the IP of the YNQ bears a [Neg] polarity value, which
needs to be reversed in the YNA, the IP of the YNQ cannot be copied into the YNA and then elided
11th WoSSP, June 5-6 2014 Autonomous University of Barcelona
10
under identity; instead, the IP of the YNA needs to be created afresh, with reversed polarity, and
needs to be overt.
• The key difference between the two languages lies in the fact that in an English polarity-reversing
YNA the answer particle and the subject are there (Yes, he is.), while in Russian it is the bare verb
(Pridёt.)
• This distinction is due to a combination of syntactic and non-syntactic factors.
• The relevant syntactic factor is EPP strength. English has strong EPP, therefore, subjects always need
to be overt - subjects of YNAs are no exception.
• Russian, on the other hand, has weak EPP (Bailyn, 2012) - consequently, the presence of a subject is
not obligatory.
• Though the conditions for subject-drop in Russian are still not well-understood (cf. McShane, 2009),
polarity-reversing YNAs seem to provide a context where subjects can - in fact, must - be dropped.
• Furthermore, recall that in Russian polarity-reversing YNAs it is not just subjects that are dropped - in
fact, the presence of any constituent apart from the verb makes the YNA strongly degraded:
(25) A: - Ty ne vzjala den’gi?
You NEG took money
Didn’t you take the money?
B: - (*Ja) vzjala (*den’gi).
I took money
(*I) took (*the money).
• This looks like an instance of a more general interdependency between subject-drop and object-drop
in ordinary YNAs in Russian: either both arguments are overt or both are silent (cf. McShane (2009)
for Russian, Huang (1984) for the same phenomenon in Chinese, and Liu (to appear) for a recent
analysis of the Chinese facts):
(26) A: Bill videl Džona?
Bill see-PAST John
Did Bill see John?
B: a. Da, on videl ego.
Yes he saw him
Yes, he saw him.
b. *Da, Ø videl ego.
*Yes, Ø saw him.
c. *Da, on videl Ø.
*Yes, he saw Ø.
d. Da, Ø videl Ø.
Yes, Ø saw Ø.
11th WoSSP, June 5-6 2014 Autonomous University of Barcelona
11
• This is due to a non-syntactic factor: Non-verbal constituents in Russian YNAs are interpreted as
contrastive, which is in conflict with the polarity answer - cf. (27) as potential replies to the question
in (25) above:
(27) - *Da, den'gi vzjala
(implies a contrast between money and something else, so not a felicitous reply to a
YNQ).
- *Da, vzjala den'gi
(implies a contrast between the action of money-taking and some other unelated
action).
• The general conclusion therefore is that the strategies of polarity reversal in YNAs in English and
Russian are more similar underlyingly than what the surface facts suggest: the general polarity-
reversing strategy is to front the unnegated version of the predicate, and then elide the rest of the
clause.
CCCConclusiononclusiononclusiononclusion and directions for further researchand directions for further researchand directions for further researchand directions for further research:
I have discussed in this talk the strategies of polarity reversal in replies to negative YNQs in English and
Russian, which turn out to be exceedingly similar, despite the fact that the negation mechanisms in
English and Russian are different. Section 1 introduced the basic facts about negative YNQs, while
section 2 turned to the analysis of language-specific polarity-reversal strategies. I adopted Holmberg’s
(2013) proposal about the different structural positions negation can occupy and showed that Russian,
unlike English, has middle and low negation only, with middle negation being the only option available
for some speakers.
Further, I showed that the strategies of polarity reversal in YNAs in English and Russian are more similar
underlyingly than what the surface facts suggest. This similarity might be part of a general trend -
regardless of the surface dissimilarity of language-specific negation patterns and availability of structural
positions in which negation can be merged, the polarity-reversing strategies do not vary equally widely
- instead, the general polarity-reversing strategy is to front the unnegated version of the predicate, and
then elide the rest of the clause. More general language-specific properties, like the EPP strength, and
pragmatic factors, can come into play too, obscuring the underlying mechanism of polarity reversal. The
unelided version of the reply is often infelicitous, which has been observed in the literature too. The
reasons for this are poorly understood at the moment.
Turning to directions for further research, prosody of a YNQ is a factor extremely relevant to the
phenomena observed, yet rarely brought up. A while ago, Ladd (1981:170) noted that the intonational
differences are not superficial phonetic phenomena, but factors important for distinguishing basic types
of questions. As research on the syntax-phonology interface receives more attention, it becomes
obvious that the prosody of polar questions is not just a superficial PF phenomenon, since the bias of
the question reflected in its prosody affects its interpretation and is one of the factors crucial for
identifying the height of negation.
11th WoSSP, June 5-6 2014 Autonomous University of Barcelona