a transferred negation ... - the grammar exchange

28
A Transferred Negation Interpretation from an Embedded to a Matrix Clause Sangsoo Park (Pusan University of Foreign Studies) Park, Sangsoo. 2003. A Transferred Negation Interpretation from an Embedded to a Matrix Clause. English Language and Linguistics 16, 271-294. In a negative complex sentence which is headed by the verb denoting the features of internalized epistemic modality and grammaticalized subjunctive mood, not- displacement is possible if and only if the P-feature Neg in the probe of a matrix and the Neg-feature in the goal of an embedded clause can be matched and identified by the operation Agree. This means that transferred negation interpretation in a negative complex sentence is guaranteed by the verb which has internalized epistemic modality and grammaticalized subjunctive mood feature. The P-feature Neg in the probe renders the negative feature in the goal active, which is able to implement the operation Agree and the internal Merge of Neg-operator triggering not-displacement. Therefore, it is reasonable to assume that the derivational process of the internal Merge of Neg-operator in an infinitival complement or a finite complement clause starts from the NegP of the embedded clause via CP and vP successively to the matrix NegP by phase to phase. Key words: negative complex sentence, not-displacement, P-feature Neg, phase, internalized epistemic modality feature, grammaticalized subjunctive mood feature 1. Introduction It seems that some negative complex sentences containing a negative element in the embedded clause as in (1a) and (2a) can be interpreted as applying it to the matrix clause as in (1b) and (2b), respectively: (1) a. I wanted you not to tell them. b. I didn't want you to tell them. (2) a. I think that he has not come. b. I don't think that he has come.

Upload: others

Post on 11-Jun-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: A Transferred Negation ... - The Grammar Exchange

A Transferred Negation Interpretation from an Embedded to a Matrix Clause

Sangsoo Park (Pusan University of Foreign Studies)

Park, Sangsoo. 2003. A Transferred Negation Interpretation from an Embedded to a Matrix Clause. English Language and Linguistics 16, 271-294. In a negative complex sentence which is headed by the verb denoting the features of internalized epistemic modality and grammaticalized subjunctive mood, not- displacement is possible if and only if the P-feature Neg in the probe of a matrix and the Neg-feature in the goal of an embedded clause can be matched and identified by the operation Agree. This means that transferred negation interpretation in a negative complex sentence is guaranteed by the verb which has internalized epistemic modality and grammaticalized subjunctive mood feature. The P-feature Neg in the probe renders the negative feature in the goal active, which is able to implement the operation Agree and the internal Merge of Neg-operator triggering not-displacement. Therefore, it is reasonable to assume that the derivational process of the internal Merge of Neg-operator in an infinitival complement or a finite complement clause starts from the NegP of the embedded clause via CP and vP successively to the matrix NegP by phase to phase.

Key words: negative complex sentence, not-displacement, P-feature Neg, phase, internalized epistemic modality feature, grammaticalized subjunctive mood feature

1. Introduction

It seems that some negative complex sentences containing a negative element in the embedded clause as in (1a) and (2a) can be interpreted as applying it to the matrix clause as in (1b) and (2b), respectively:

(1) a. I wanted you not to tell them.b. I didn't want you to tell them.

(2) a. I think that he has not come.b. I don't think that he has come.

Page 2: A Transferred Negation ... - The Grammar Exchange

It is generally assumed that the sentence (b) is derived from the sentence (a) through negative transportation, or not-displacement in both cases. As an embedded negation is considered the implicature relation of a matrix negation, the former seems to deliver a somewhat more specific meaning than the latter. Interpreting a negative scope related to negative attraction, Jespersen (1917 [1937]:53) observes that there is a strong tendency in many languages to attract to the main verb a negative which should logically belong to the dependent nexus. On the topological phenomenon of the transferred negation, Bublitz (1992:553) points out that the negative transportation is applied to a large number of complex utterances containing such predicates as believe, suppose, think which denote non-factive mental activity or desire with epistemic modal meaning.Based on the Minimalist Theory (Chomsky 2000, 2001a, 2001b, 2002), this article examines the semantic and syntactic evidence for transferred negation interpretation by means of which some class of matrix predicates attract a negative element from an embedded to a matrix clause. That is, in some negative complex sentences transferred negation interpretation is possible if and only if the probe in the matrix and the goal in the embedded clause can be matched and identified by the operation Agree.The organization of this paper is as follows. In section 2, semantic and functional argument will be made over triggering negative transposition reading in negative complex sentences. The predicates that allow transferred negation interpretation are concerned with various kinds of epistemic modality. This is proved by the semantic relation of implicaure and the semantic feature of predicates which has internalized epistemic modal meaning. In section 3, syntactic evidence for negative transportation will be pursued in the negative complex

Page 3: A Transferred Negation ... - The Grammar Exchange

sentences which are headed by a set of verbs denoting grammaticalized subjunctive mood feature with epistemic modality. Syntactic rules or processes like the derivation of tag-formation, the constraint on negative polarity constructions and the condition on the interpretation of pronouns are examined to prove negative transportation. In section 4, the hypothesis that Transfer has a memory of phase-length (Chomsky 2001b:14) will be proved in the complex sentences that undergo not-displacement which is ridden by a Neg-operator. We will demonstrate that the syntactic scope of a Neg-operator observing Neg-criterion in a negative complex sentence is checked at a phase level, and therefore the internal Merge of the Neg-operator riding negator not is operated under the Phase Impenetrability Condition (PIC). The final section 5 is a summary with concluding remarks.

2. Semantic property

Although Fillmore (1963:220) first attempted to incorporate a rule for ‘transposition of not’ into the Generative Grammar of English, he did not offer syntactic evidence for it, but referred to the alleged paraphrase relation holding between (3a) and (3b). Namely, the negative element that is base-generated in the embedded clause moves to the matrix clause:1)

(3) a. John thinks that Bill doesn't like Harriet. b. John doesn't think that Bill likes Harriet.

It is semantically argued that negative force in the transported reading (3b) is perceptively weaker than that in its 1) At first Lakoff and Bolinger shared the view that the semantic evidence for negative

transportation is not as strong as it had seemed, but Lakoff said that certain facts like tag questions have come to light that appear to provide fairly conclusive evidence that this rule must exist (Lakoff 1969:140).

Page 4: A Transferred Negation ... - The Grammar Exchange

non-transported pair (3a). It means that negative transposition reading from the embedded to the matrix clause has the effect of softening down the negative force and we can thus infer that the sentence (3a) is the implicature of the sentence (3b).In connection with transferred negation interpretation, Horn (1978:132-133) proposed two general principles: one is that the negative force weakens with the distance of the negative element from the constituent with which it is logically associated, and the other is that the negative force strengthens with the degree of incorporation of the negative element.2) The former principle is at work in a negative complex sentence allowing the negative transportation, while the latter affects the operation of incorporation or morphological absorption of the negative element. A negative force generally weakens with the distance of the negative element from the constituent with which it is logically associated, whereas it is strengthened by using synthetic forms like none/no one or nothing rather than by the analytic forms like not anyone or not anything. The paradigm of the following examples shows the softening process (Horn 1978:132):

(4) a. I think John saw no one in the room.b. I think John did not see anyone in the room.c. I think John didn't see any one in the room.d. I do not think John saw anyone in the room.e. I don't think John saw anyone in the room.

Native speakers seem to perceive that in terms of negative force (4a) is stronger than (4b)/(4c), which is, in turn, stronger than (4d)/(4e). We can explain the degree of perceptional difference affected by the placement of a negative

2) A typical example of negative incorporation is proved with the form of a negative prefix+a positive word as in un+happy>unhappy, dis+like>dislike, etc.

Page 5: A Transferred Negation ... - The Grammar Exchange

element through entailment relation (Horn 1978:134, Huddleston and Pullum 2002:840-841). It is clear that (4d)/(4e) are true in any situation which makes (4a) or (4b)/(4c) true, but not vice versa. It implies that the internal-Neg reading of the embedded negation in (4b)/(4c) unilaterally entails the external- Neg reading of the matrix negation in (4d)/(4e).3) Between the sentences with non-contracted negative forms as in (4b)/(4d) and those with contracted negative ones as in (4c)/(4e), the former is recognized as stronger than the latter, in each case. This means that the morphological contraction of a negative element to a tense verb weakens the negative force.The predicates that permit negative transposition reading are related to epistemic modality. Of describing the semantic property of modality, we usually make a distinction on the dimension concerning the strength of commitment to the factuality or actualization of the verbalized content: weak, medium and strong commitment related to modality (Huddleston and Pullum 2002:175-177). For example, in the core epistemic modality concepts of necessity, possibility and probability, the necessity involves a strong commitment, whereas the probability a weak one.4) Let us consider the degree of strength of commitment from the following examples:

3) In this paper we use the term implicature and entailment without a special distinction in conceptual definition. Consider the following examples:

(ⅰ) She thinks that he isn't alive.(ⅱ) She doesn't think he's alive.

We can say that the sentence (ⅰ) functionally implicates the sentence (ⅱ) in the light of pragmatic presupposition, whereas the sentence (ⅰ) logically entails the sentence (ⅱ) in terms of formal semantics.

4) According to the definition of three types of modality (Palmer 1990:35-36, Huddleston and Pullum 2002:52), epistemic modality refers to the speaker's opinion that the sentence expresses the necessity, possibility or probability concerned with matters of knowledge or belief, deontic modality explains the speaker's attitude that the proposition describes the obligation, permission or prohibition performed by morally responsible agents, and dynamic modality is concerned with the ability or volition of the subject of the sentence.

Page 6: A Transferred Negation ... - The Grammar Exchange

(5) The meeting must be over now. <strong commitment> (6) The meeting may be over now. <weak commitment>(7) The meeting should be over now. <medium commitment>

The sentence (5) includes the epistemic modal verb must that shows the modal meaning of obligation: it delivers a strong commitment in that it doesn't approve meeting going on. The sentence (6) has the epistemic modal verb may, but it is interpreted as having the modal meaning of probability: it denotes a weak commitment in that it can be consistent with such continuations as, “but it's unlikely to be”. Regarding the meaning of (7), our attention is psychologically focused on the epistemic modal interpretation of possibility, “The meeting is probably over by now”. Such reading is triggered by the epistemic modal verb should with the meaning of a reasonable assumption or conclusion without certainty (Turner 1985:131, Palmer 1990:59, Warner 1993: 180, Park 2002b:31).Thus, from the modality concept classified by the degree of strength, we can infer from some negative complex sentences a significant property which is related to the medium commitment of modality: there is little difference between the external and internal negation.5) Let us examine the scope of negation in the examples (8) which show the modal property containing medium commitment:

(8) a. She isn't likely to be ready. <external negation>

5) We can distinguish sentences where the modal has scope over the negative as in (ⅰ) and the negative has scope over the modal as in (ⅱ):

(ⅰ) Jane may not have read it.(ⅱ) Jane cannot have read it.

In the example (ⅰ), the negation falls within the scope of the modal as internal negation: “It is possible that Jane didn’t read it”, whereas in the example (ⅱ), the negative element applies to the modal itself as external negation: “It is not possible that Jane read it”.

Page 7: A Transferred Negation ... - The Grammar Exchange

b. She is likely not to be ready. <internal negation>

The sentence (8a) containing external negation allows that the possibility of her being ready may be around 50 percent as well as low, but in the 50 percent case one would generally say, “She may be ready”, so that (8a) tends to be used only in the low possibility case, “She is unlikely to be ready”, and this is equivalent to the sentence with internal negation (8b) (Huddleston and Pullum 2002:177). Thus, we can deduce that in a clausal construction with medium modality a negative complex sentence semantically has the internal negation in the normal situation.In negative complex sentences, if the predicate of a matrix clause connotes internalized epistemic modal meaning as its semantic property, the negative element in the embedded clause is interpreted as applying to the matrix one as in the examples below:

(9) a. I don't want him to go into politics.b. I want him not to go into politics.

(10) a. She doesn't seem to be diligent.b. She seems not to be diligent.

(11) a. Doctor didn't advise me not to drink.b. Doctor advised me not to drink.

It is logically clear that the sentence (b) is the implicature of sentence (a) in each of the paired examples. This means that if there is relatively little difference in the logical meaning of the sentences between the matrix and embedded negation, the matrix lexical verb has the semantic property of the epistemic modality with the medium strength. Given that the negation with narrow scope is the implicature of the negation with wide one, a negative complex sentence where an implicature relation

Page 8: A Transferred Negation ... - The Grammar Exchange

applies between the embedded and matrix clause permits transferred negation interpretation.The transferred negation interpretation illustrated in (9)-(11) is rather widespread in real language use. There are a good number of verbs and adjectives that permit negative transposition reading in negative complex sentences containing a non-finite as well as finite embedded clauses. A sample of the predicates allowing transferred negation interpretation is listed and classified into semantic groups (Horn 1978:187, Quirk, Greenbaum, Leech and Svartvik 1985:1033-1034, Huddleston and Pullum 2002:840):

(12) a. Wanting: choose, intend, mean, plan, want, ...b. Advice: advise, advisable, be meant, be supposed, had better,

recommend, suggest, ...c. Probability: likely, necessary, possible, probable, supposed, ...d. Opinion: anticipate, believe, calculate, expect, feel, figure, guess,

imagine, predict, reckon, suppose, think, ... e. Perception: appear, feel, look, seem, sound, ...

The predicates included in this group generally connote the semantic feature that is related to the meaning of psychological judgement, non-factive mental state or activity, or desire.Considering that modality is preliminarily defined as the opinion and attitude of the speaker (Lyons 1977:452) and is prototypically concerned with the speaker's opinion or attitude toward the factuality or actualization of the situation expressed by the rest of the clause (Huddleston and Pullum 2002: 173), the linguistic expression of the modality includes lexical modals, past tense, the residual irrealis were, non-default clause types such as imperatives and interrogatives as well as modal auxiliaries. Among these linguistic expressions, lexical modals cover verbs like insist, permit, require, adjectives like

Page 9: A Transferred Negation ... - The Grammar Exchange

necessary, possible, likely, adverbs like perhaps, necessarily, probably, certainly, surely, and nouns like possibility, necessity, permission. Such lexical modals began to be used in discourse to tone down assertion, which is performed by reflecting the semantic property of the internalized epistemic modality with the medium strength of commitment. Therefore, the predicates denoting internalized epistemic modality with the strength of medium commitment as in (12) permit transferred negation interpretation. In sum, the semantic property of predicates allowing transferred negation interpretation can be stipulated as follows:

(13) Only predicates having internalized epistemic modal feature permit transferred negation interpretation in negative complex sentences.

The semantic condition (13) demonstrates that the predicates included in (12) are transparent regarding the scope of negation whenever they are used in the negative complex sentences. The reason is that the semantic feature of epistemic modality has to do with non-factive knowledge which qualifies speaker's medium commitment to the truth of the proposition uttered. The predicates having the internalized epistemic modality that is related to non-factive verbs of mental state (Green 1974:49) generally serve to qualify the speaker's attitude toward propositions and interlocutors alike (Bulitz 1992: 559). The predicates that show the semantic property of psychological judgment, non-factive mental state or activity, or desire with internalized epistemic modality usually permit transferred negation interpretation from the embedded clause with neutral, unmarked and basic internal-negation to the matrix clause with non-neutral, marked and external-negation that is perceived as a little bit qualified and softened.

Page 10: A Transferred Negation ... - The Grammar Exchange

3. Syntactic property

In the early Generative Grammar, negative transportation was generally taken to be a cyclic, lexically governed and structure-preserving rule which extracts a negative element from a lower clause and raises it one clause up over a predicate marked to allow the rule application (Horn 1989:312), but it is now considered that a negator is merged in the embedded clause and then its scope is extended to the matrix clause by internal Merge. The operation of negative transportation is diachronically connected with the development of epistemic modals from a morphological system into a more grammaticalized modal auxiliary system; the development of psychological judgement verbs like expect, suppose, think into a kind of hedges, or the medium commitment of assertion; the loss of multiple negation and the resultant grammaticalization of implicit negatives like any, ever, yet, at all; the change of position in the placement of certain negative elements including not; and the rise of subject raising and the concomitant loss of impersonal constructions (Fisher 1998:56). Under certain context a negative element tends to be closely associated with other words. In OE, for example, the negative particle ne could be contracted with the verbs that follow it,6) and the contraction as such is applied to verbs having grammatical function like epistemic modal rather than to verbs having purely lexical meaning (Plank 1984:329). The difference in the strength of the ties between the negatives and modal auxiliaries on the one hand and lexical verbs on the other seems to have been caused by the categorial similarity that is mainly concerned with grammatical functions. In a further step of

6) Historically the negative element ne could be cliticized onto following verbs, adverbs or pronouns, which had an initial vowel or with loss of initial pronunciation [w] or [h] as in ne woldon>noldon.

Page 11: A Transferred Negation ... - The Grammar Exchange

development, several negated modals with contracted forms such as nolde/ne wolde, nat/newat, nyllan/ne willan, can't, mustn't, shouldn't began to gain some degree of independent meaning which became different from the meaning of positive counterparts. Such grammaticalization from preterite-present verbs to modal auxiliaries helped strengthening the syntactic connection between negative elements and modal auxiliary verbs and resulted in triggering the application of negative transportation. By virtue of negative concord,7) no(n) with ne was used regularly in the negative clause of OE. The loss of multiple negation caused only matrix verbs to become negated and the negator in the embedded clause to get lost, and it consequently triggered negative transportation. The change that started with the decline of multiple negation and went on with the successive weakening of one negator has led to the reanalysis of the negative element that is no longer the specifier, but the head of NegP (Fisher, Kemenade, Koopman and Wurff 2000: 314-315). As the negative concord in multiple negation was lost diachronically, one of the negation either in the matrix or in the embedded clause became extinct, and this change automatically made it possible that a negative complex sentence headed by the matrix verb denoting internalized epistemic modality began to allow negative transportation. It is proved by 7) The rule of negative concord which is a by-product of Neg-criterion means that a

logically negative sentence can contain more than one negative marker. Consider the following sentences:

(ⅰ) He didn't know anything.(ⅱ) He knew nothing.(ⅲ) He didn't know nothing.

The example (ⅱ) is derived from (ⅰ) by the incorporation of negative element which has the effect of moving [NEG] down onto an indeterminate constituent. The negative concord illustrated in (ⅲ) can then be seen as a copying rule whereby [NEG] is not lowered onto indeterminate constituent but copied onto it leaving [NEG] associated with auxiliary verb (Haegeman 1995:297).

Page 12: A Transferred Negation ... - The Grammar Exchange

widely admitted phenomenon that the position of NPIs is determined by c-command condition between the negative element not and NPIs in negative complex sentences: subject NPIs may not be triggered by not, whereas object can be (Hoeksema 2000:123). This means that a negative element cooccurring with NPIs is merged in the embedded clause and then optionally undergoes the syntactic operation of negative transportation.In OE and ME there occurred impersonal constructions which had oblique arguments and lacked a nominative subject. The disappearance of impersonal constructions from the late ME was ascribed to the loss of inherent Case and to the fact that the subject position became obligatory. Accordingly, as subject raising occurred with raising predicates such as appear, seem, be certain, be likely, they seem to have gained a syntax-driven epistemic modal property from pure impersonal verbs. The raising predicates having epistemic modal meaning now permit a negative element to be placed next to them, whether or not it logically belongs in the matrix clause (Fischer 1998:69).8)In the history of English language, as the morphological inflections of verbs for marking subjunctive mood were leveled and lost, not only grammaticalized modal auxiliaries, but also a set of verbs denoting non-factive mental state or activity like believe, suppose, think, etc. and expressing desire like intend, want, etc. came to fill up the confused morphological

8) The innovation of modal raising, for instance, a can't seem to-construction is related to the disappearance of non-finite modals (Jack 1978:70). In the can't seem to-construction, the modal auxiliary semantically belongs to the complement clause in the underlying structure, exemplified in (ⅰ):

(ⅰ) It seems that he can't do the work.(ⅱ)*He seems not to can do the work.(ⅲ) He can't seem to do the work.

As can be seen from the difference of grammaticality between (ⅱ) and (ⅲ), a modal auxiliary must sneak into the finite matrix clause along with the negative element.

Page 13: A Transferred Negation ... - The Grammar Exchange

subjunctive system. This change is proved by the fact that mood reflects the grammaticalization of modality within the inflectional system of verbs (Huddleston and Pullum 2002:172). As far as English is concerned, the diachronic change of mood has eliminated mood marker from the inflectional system of verbs, and resultantly it was replaced by the analytic property of some verbs along with the divergence of modal auxiliaries. This means that the extent of subjunctive mood expressed by inflections was reduced and the supplementary grammaticalized subjunctive mood feature9) came to be realized in a set of verbs having epistemic modal meaning as the complementary syntactic mechanism. It is worth pointing out that negative complex sentences which show the form from subjunctive to indicative and accordingly have the converted meaning from positive to negative, constitute themselves the typical evidence for a change.Based on such diachronic analysis, we propose that some class of verbs denoting grammaticalized subjunctive mood feature with epistemic modality permit negative transportation. The syntactic property of predicates allowing negative transportation can be stipulated as follows:

(14) Only predicates having grammaticalized subjunctive mood feature with epistemic modality permit the extension of negative scope from embedded to matrix clauses in negative complex sentences.

The matrix predicates observing the syntactic condition (14) can be read in two possible ways (Lakoff 1969:140, Lakoff 9) Given that grammaticalization is taken to be a semantic tendency for an item with a full

lexical meaning to be a bleached over time and to come to be used as a grammatical function (Lightfoot 1999:200), the meaning of grammaticalized in the “grammaticalized subjunctive mood feature” implies that the subjunctive sense of items as in (12) is identified not through the lexical meaning of each word itself, but through the grammatical and syntactic feature of each word whose meaning is interpreted as a compositional part of the whole sentence.

Page 14: A Transferred Negation ... - The Grammar Exchange

1970:148, Bublitz 1992: 557):

(15) She doesn't think her husband's alive.(16) a. It is not the case that she thought that her husband was alive.

b. She thought that her husband was not alive.

When (15) is interpreted as in (16a), the negative element not has the matrix clause as its scope, but when (15) is interpreted as in (16b), its scope is restricted to the embedded clause. Considering the fact that the logical meaning of (15) is derived from (16b), it is reasonable to suggest that the negator not is base-generated in the embedded clause and is subsequently transferred to the matrix clause for negative transposition reading. The evidence for negative transportation can be argued from the formation of tag question as in (17) (Lakoff 1969:142-145, Horn 1978:153-154, Fischer 1998:57):

(17) a. You don't think that he has paid, do you?b. You don't think that he has paid, has he?

Given the rule of tag-formation, the matrix clause must be negative for the positive tag to occur in the sentence (17a), whereas it is the embedded clause that is tagged and thus the embedded clause must be inherently negative in the sentence (17b). So it could be argued that the negator in the (17b) has been derived from the embedded to the matrix clause through internal Merge. This analysis presupposes that a negative element is base-generated in the embedded clause and subsequently transferred to the matrix clause. The syntactic argument over the position and scope of NPIs is related to the patterning of the relatively restricted NPIs as in (18)-(21):

Page 15: A Transferred Negation ... - The Grammar Exchange

(18) a. I think he doesn't want to see anyone.b. I don't think he wants to see anyone.

(19) a. I believe his wife isn't a spy anymore.b. I don't believe his wife is a spy anymore.

(20) a. I claimed that he wouldn't arrive until midnight.b.*I don't claim that he would arrive until midnight.

(21) a. I realized that she didn't love him any longer.b.*I didn't realized that she loved him any longer.

The grammaticality of the sentences in (18)-(21) shows that NPIs like any, ever, yet, either, at all, till/until can appear freely in a negative embedded clause, but they are allowed only by the negated matrix predicates that meet the condition (13) and (14). Since the negation cooccurring with some other class of verbs in the matrix clause cannot trigger the NPI of the embedded clause, the derivation of such sentences as (20b) and (21b) are blocked. This implies that a negative element like not must be merged in the embedded clause due to the fact that the first syntactic constraint between the negation and the NPI applies within the same clause. Negative transportation is thought to be a minor rule in that the possibility of the rule application is qualified by verbs like believe, suppose, think, etc., but not by verbs like claim, know, realize, etc. The evidence for negative transportation in some negative complex sentences can be proved by the syntactic interaction of the negative raising and NPIs: negative polarity items normally require a tauto-clausal negative trigger as in (18a)-(21a). We can infer from this observation that some grammatical relation must exist which codes the dependency between a negative element in the matrix and a negative polarity item in the embedded clause.Sentences like (22) also provide fairly persuasive evidence that there is a rule of negative transportation (Lindholm 1969:154):

Page 16: A Transferred Negation ... - The Grammar Exchange

(22) a. I don't think Bill paid his taxes and Mary is quite sure of it.b. I think Bill didn't pay his taxes and Mary is quite sure that he

didn't.

The pronoun it in the example (22a) is interpreted as having the meaning “he didn't (=Bill didn't pay his taxes).” This implies that before the pro-form it was substituted for the constituent “that he didn't”, the negative element not was merged in the embedded clause as in (22b). Such analysis clearly proves that the negative element not was displaced from the embedded to the matrix clause after the complement of sure was replaced by the pro-form it.In sum, it is quite reasonable to propose that some class of verbs of having the internalized epistemic modality began to connote grammaticalized subjunctive mood feature in the history of the English language, and that the loss of inflectional subjunctive mood marker and complementary introduction of the grammaticalized subjunctive mood feature to the class of verbs denoting internalized epistemic modality rendered the operation of negative transportation possible in negative complex sentences.

4. Not-displacement

Judging from the viewpoint of Minimalist Theory, not-displacement ridden by the internal Merge of Neg-operator, of which derivation proceeds from the hypothesis of Neg-criterion (Haegeman 1995:106), can be checked by the operation Agree. In order to make the interpretation of long-distance negation work properly, it should be assumed that there exists some feature Agree relation between a real negative element in the functional category of the matrix clause and the copied element in the embedded clause.

Page 17: A Transferred Negation ... - The Grammar Exchange

In the previous section, it was proposed that the application of not- displacement which was possible along with diachronic linguistic changes came to be guaranteed by the matrix verb denoting the semantic feature of epistemic modality as in (13) and the syntactic feature of grammaticalized subjunctive mood as in (14). If there is an interpretive dependency between a matrix negation and the semantic proposition of a complement clause, the operation Agree must hold between the probe in the matrix and the goal in the embedded clause. Thus the feature value related to not-displacement is determined by the operation Agree under the condition given in (23):

(23) A class of verbs which have the exceptional features of internalized epistemic modality and grammaticalized subjunctive mood satisfy the syntactic condition of not-displacement in a negative complex sentence.

The condition in (23) is elicited from the fact that some grammatical relation must hold between a negative element in the matrix clause and negative meaning which is expressed via an implicit negative force in the embedded clause (Hornstein and Uriagereka 2002:127) through the long-distance syntactic relation of feature Agree.10) This means that if the NegP that observes Neg-criterion (Haegeman 1995:106) is selected by T in the matrix clause where the condition (23) is met, the Neg-feature in the matrix clause must be valued with the related feature in the embedded clause under the syntactic operation Agree and be transferred to the phonological component by Transfer. Considering the circumstances that not-displacement can be applied only in some negative complex sentences, we propose that a verb allowing transferred negation

10) The negative element spelled out in the functional category of matrix does not negate its own clause, but agrees into the embedded clause, which demonstrates that the embedded clause preserves its negative meaning.

Page 18: A Transferred Negation ... - The Grammar Exchange

interpretation should be merged in the matrix clause, and it categorizes for the form of its complement.11) Given the two kinds of semantic conditions applied at C-I between argument structure associated with external Merge and scope and discourse-related properties with internal Merge (Chomsky 2001b:8), we can exemplify the latter case in negative complex sentences permitting not-displacement. The syntactic and semantic property of negative complex sentences which we have analyzed so far suggests one way of looking at generalization of triggering negative transportation: ECM verbs12) and raising predicates allow not- displacement. Let us examine the following examples:

(24) a. I think that John isn't intelligent.b. I don't think that John is intelligent.

(25) a. I think John not to be intelligent.b. I don't think John to be intelligent.

(26) a. It seems that Harold isn't honest.b. It doesn't seem that Harold is honest.

(27) a. Harold seems not to be honest.b. Harold doesn't seem to be honest.

In order to make interpretation of long-distance negation possible in the examples from (24a)-(27a) to (24b)-(27b), 11) Descriptively the term form refers to words, word classes, phrases, clauses or sentences.

We suggest that the syntactic property of a lexical verb that decides the form of its complements is called complement form feature which is composed of the syntactic features of a verb along with subcategorizational feature.

12) Negative complex sentences with infinitival complements introduced by ECM verbs typically show the evidence of not-displacement:

(ⅰ) I don't expect him to go there.(ⅱ) I don't advise you to go there.

The sentences like (ⅰ) and (ⅱ) usually have the meaning that the speaker expects or advises someone not to go there, and such semantic interpretation licenses the syntactic operation of negative transportation to be reasonable.

Page 19: A Transferred Negation ... - The Grammar Exchange

respectively, it must be ensured as syntactic evidence that there is feature Agree between the probe that renders the Neg-feature in the matrix Neg position active and the goal that manifests negative force in the embedded clause. In the negative complex sentences like (24)-(27), the Neg-feature induces the operation Agree in terms of the internal Merge of Neg-operator, whose structural condition is guaranteed by the matrix verb containing the features related to internalized epistemic modality and grammaticalized subjunctive mood. Given PIC, as the faculty of language takes the scopal property to be edge phenomenon and internal Merge to be successive- cyclic, the Neg-operator in the embedded NegP-spec moves through the phonological edge of phases to the matrix NegP-spec observing Neg-criterion that applies universally at narrow syntax (Haegeman 1995:111). Considering that raising and ECM constructions have the selectional relation V-Tdef, while control structures and finite clauses have the relation C-Tcomp, a negative complex sentence which is realized as the raising or ECM construction permits the internal Merge of Neg-operator triggering not-displacement as shown in the configurational structures (28)-(29):

(28) a. I think John not to be intelligent. ⇒b. I do not think John to be intelligent.c. [TP1 I do [NegP Neg-operatori [Neg not][vP ti [vP [v think [VP ... ][TP2

John ti to be intelligent]]]]]](29) a. Harold seems not to be honest. ⇒

b. Harold does not seem to be honest.c. [TP1 Harold does [NegP Neg-operatori [Neg not][vP ti [vP [v seem [VP ... ]

[TP2 ... ti to be honest]]]]]]

The Neg-operator inducing not-displacement moves to the NegP-spec of the matrix clause through each phase structure

Page 20: A Transferred Negation ... - The Grammar Exchange

of (28c) and (29c) respectively, where the matrix lexical verb carries such exceptional features as internalized epistemic modality and grammaticalized subjunctive mood. The matrix Neg guaranteed by the verb satisfying the condition (23) in negative complex sentences has the occurrence OCC (Chomsky 2001b:10)13) as the P-feature of the peripheral system of negative force that yields a new negative scopal property. It triggers not-displacement ridden by the Neg-operator from the embedded to the matrix clause. In (28c) and (29c), the feature OCC is checked by the internal Merge of Neg-operator as the OCC performs the discourse- related function of providing new scopal interpretation. The internal Merge of Neg-operator triggering not-displacement in (28) and (29) by phase starts from the domain of TP2 and moves to the matrix NegP-spec position. This means that the P-feature Neg that is coded by the features of internalized epistemic modality and grammaticalized subjunctive mood which are contained in the matrix verb matches and identifies the Neg-feature in the embedded clause.As a diagnostic device for not-displacement, we must turn to an alternative way of looking at generalization of triggering negative transportation: ECM verbs and raising predicates which select either an infinitival complement or a finite clause permit the internal Merge of Neg-operator. This implies that the matrix lexical verb in a negative complex sentence permits not-displacement that observes Neg-criterion if and only if the condition (23) is satisfied. As the core functional category C expresses the force and mood (Chomsky 2000: 102), the complementizer that in the negative complex sentences provides an escape hatch for the internal Merge of Neg-operator between the probe in the Neg and the goal in the original

13) The feature-OCC that performs the function of generalized ECP is selected by the head H only if that yields new scopal or discourse-related properties.

Page 21: A Transferred Negation ... - The Grammar Exchange

position. Therefore, it seems to be reasonable to assume that the internal Merge of Neg-operator in (30) and (31) is carried out by phase to phase, starting from the domain of TP2 via CP2 and vP successively, to the matrix NegP:

(30) a. I think that John is not intelligent. ⇒b. I do not think that John is intelligent.c. [TP1 I do [NegP Neg-operatori [Neg not][vP ti [vP [v think [VP ... ][CP2 [CP2

ti [C2 that][TP2 John is ti intelligent]]]]]]]](31) a. It seems that Harold is not honest. ⇒

b. It does not seem that Harold is honest. c. [TP1 It does [NegP Neg-operatori [Neg not][vP ti [vP [v seem [VP ... ][CP2

[CP2 ti [C2 that]][TP2 Harold ti is honest]]]]]]]]

New scopal interpretation is given by the operation Agree from the domain of TP2 via CP2 and vP to NegP in the course of internal Merge of the Neg-operator; hence the syntactic operation in (30) and (31) do not violate PIC. The OCC of P-feature in C2 is coded by the complementizer that which not only has mood feature but also is selected by the matrix verb denoting the features of internalized epistemic modality and grammaticalized subjunctive mood that allows not-displacement ridden by the internal Merge of Neg-operator.Let us examine negative complex sentences like (34) and (35) whose heads are the non-factive verbs of mental state, but which are superficially similar to control structures and finite clauses:

(32) a. I wish not to see you again. ⇒b. I don't wish to see you again.

(33) a. I hope never to see you anymore. ⇒b. I never hope to see you anymore.

(34) a. I wish that I will not see you again. ⇏

Page 22: A Transferred Negation ... - The Grammar Exchange

b. I do not wish that I will see you again.c. [TP1 I do [NegP Neg-operatori [Neg not][vP ti [vP [v wish [VP ... ][CP2 [C2

that][TP2 I will ti see you again]]]]]]](35) a.. I hope that I will not see you anymore. ⇏14)

b. I do not hope that I will see you anymore. c. [TP1 I do [NegP Neg-operatori [Neg not][vP ti [vP [v hope [VP ... ][CP2 [C2

that][TP2 I will ti see you anymore]]]]]]]

Contrary to the paired sentences in (32) and (33), there is a semantic difference between (34a) and (34b), (35a) and (35b), respectively. This means that whereas not-displacement is possible from (32a) to (32b) or from (33a) to (33b), it is not possible from (34a) to (34b) or from (35a) to (35b). It can be inferred from the grammaticalty that some verbs selecting non-finite complement clauses allow not-displacement, but other verbs taking finite complement clauses do not permit it. We suggest that it is clearly related to the typical form of complements which a matrix verb subcategorizes for, and the structural information as such must be compatible with the syntactic condition that some verbs allows the internal Merge of Neg-operator, while others do not. The blocking effect of the internal Merge of Neg-operator from (34a) to (34b) and from (35a) to (35b) can be explained principally by the PIC. When we try to apply the internal Merge of Neg-operator to 14) Horn (1978:151) considers hope-construction similar to wish-structure, while Green

(1974: 50) insists that under ordinary circumstances hope doesn't permit negative-raising with the negative particle not, but it does permit negative-raising if the negative is never:

(ⅰ) a. I don't hope I see you tomorrow. ⇏ b. I hope I don't see you tomorrow.(ⅱ) a. I never hope I have to live in Boston. ⇒

b. I hope I never have to live in Boston.

The difference of grammaticality between (ⅰ) and (ⅱ) seems to be caused by the facts that never is the incorporated form of an NPI contrary to not, or that some class of verbs like hope is in the transition period expressing subjunctive mood from residual morphological system to syntactic paradigm.

Page 23: A Transferred Negation ... - The Grammar Exchange

the sentences like (34) and (35), the feature interpretation and evaluation in (34c) and (35c) is operated between NegP and TP2, skipping the strong phase CP2. The guiding principle for PIC that the domain of TP2 is not accessible to the operation Agree at CP2 is due to the fact that the complementizer that in C2 as in (34) and (35) does not have mood feature which must agree to that of the matrix lexical verb.The hypothesis that some class of verbs have the features of internalized epistemic modality and grammaticalized subjunctive mood is also related to the syntactic property of mandative verb constructions. Mandative verbs such as demand, require, propose select a finite complement clause headed by a bare infinitival verb in order to carry out the grammatical function of subjunctive mood. This implies that subjunctive mood feature in a mandative verb construction is matched and identified by the operation Agree between the mandative verb in the matrix clause and the bare infinitival verb in the embedded clause. It is really suggestive to propose that mandative verbs select a complement clause with a bare infinitival verb which has the default tense feature of zero morpheme,15) and that ECM verbs select a complement clause with a to-infinitive verb16) which has unrealized future tense feature. The fact as such shows that mandative and ECM verb constructions share the same syntactic property in that they both take their complement clauses headed by a non-finite verb denoting analytically grammaticalized subjunctive mood feature that is licensed by the mandative and ECM verbs.

15) The similar hypothesis that bare infinitive verbs have defective tense feature of zero morpheme was proposed by Park (2002a:146) in the diachronic approach to bare infinitival constructions.

16) As is known, the particle to has undertaken the process of grammaticalization from a preposition of direction which assigns a dative Case to its complement in Old English to a grammatical marker which expresses unrealized future tense in Present-day English

Page 24: A Transferred Negation ... - The Grammar Exchange

5. Conclusion

In this article we have examined the semantic and syntactic evidence of not-displacement that is guaranteed by some class of matrix verbs permitting transferred negative interpretation. When the verb of a negative complex sentence connotes the features of internalized epistemic modality and grammaticalized subjunctive mood, the operation of the internal Merge of Neg-operator triggering not-displacement is possible if and only if the P-feature Neg in the probe of matrix and the Neg-feature in the goal of the embedded clause can be matched and identified by the operation Agree.Some verbs which allow negative transposition reading is licensed by internalized epistemic modality, which reflects the semantic property of non- factive verbs of mental state and serves to qualify the speaker's attitude toward propositions. A class of verbs denoting the semantic feature of epistemic modality simultaneously contain grammaticalized subjunctive mood feature as their syntactic property. In the history of English language, not only grammaticalized modal auxiliaries but also non-factive verbs of mental state came to fill up the confused inflectional subjunctive system as the morphological inflections of verbs were leveled and lost for marking subjunctive mood. In negative complex sentences, the process of matching and identifying Neg- feature which is coded by internalized epistemic modality and grammaticalized subjunctive mood feature in the matrix clause induces the operation Agree, and determines the value of the Neg-feature that activates the operation of not-displacement ridden by internal Merge of Neg-operator. As a diagnostic device for not-displacement, we can employ the condition (23): A class of verbs which have the exceptional features of internalized epistemic modality and

Page 25: A Transferred Negation ... - The Grammar Exchange

grammaticalized subjunctive mood satisfy the syntactic condition of not-displacement in a negative complex sentences. If ECM verbs and raising predicates selecting either an infinitival complement or a finite complement clause, meet the condition (23), they permit not-displacement. Given that the core functional category C expresses the force and mood, the complementizer that in the negative complex sentences headed by the verbs with the features of internalized epistemic modality and grammaticalized subjunctive mood provides an escape hatch for the internal Merge of Neg-operator from the embedded to the matrix clause. Therefore, it is natural to assume that the derivational process of the internal Merge of the Neg-operator triggering not-displacement in an infinitival complement or a finite complement clause starts from the NegP of the embedded clause and proceeds, via CP and vP successively, to the matrix clause NegP by phase to phase.To conclude, non-factive verbs of mental state denoting internalized epistemic modality as their implicit modal indicator and grammaticalized subjunctive mood feature as their analytic mood marker concomitantly render not- displacement possible in negative complex sentences by way of the syntactic process of matching and identifying Neg-feature by the operation Agree.

References

Bublitz, W. 1992. Transferred Negation and Modality. Journal of Pragmatics 18, 551-577.

Chomsky, N. 2000. Minimalist Inquiries: The Framework. Martin, R., D. Michaels and J. Uriagereka (eds.). Step by Step, 89-156. Cambridge, Mass.: MIT Press.

_____. 2001a. Derivation by Phase. Kenstovicz, M. (ed.). Ken Hal Life in,

Page 26: A Transferred Negation ... - The Grammar Exchange

1-52. Cambridge, Mass.: MIT Press._____. 2001b. Beyond Explanatory Adequacy. MIT Occasional Papers in

Linguistics 20. Cambridge, Mass.: MIT._____. 2002. On Nature and Language. Belletti, A. and L. Rizzi. (eds.).

Cambridge: Cambridge University Press.Fillmore, C. J. 1963. The Position of Embedding Transformations in a

Grammar. Word 19, 208-231.Fischer, O. 1998. On Negative Raising in the History of English. van Ostade,

I. T-B., G. Tottie and W. van der Wurff. (eds.). Negation in the History of English, 55-100. New York: Mouton de Gruyter.

_____, A. van Kemenade, W. Koopman, and W. van der Wurff. 2000. The Syntax of Early English. Cambridge: Cambridge University Press.

Green, G. M. 1974. Semantics and Syntactic Regularity. Bloomington, Indiana: Indiana University Press.

Haegeman, L. 1995. The Syntax of Negation. Cambridge: Cambridge University Press.

Hoeksema, J. 2000. Negative Polarity Items: Triggering, Scope, and C-Command. L. Horn and Y. Kato. (eds.). Negation and Polarity: Syntactic and Semantic Perspectives, 115-146. Oxford: Oxford University Press.

Horn, L. R. 1978. Remarks on Neg-Raising. Cole, P. (ed.). Syntax and Semantics Vol. 9: Pragmatics,129-220. New York: Academic Press.

Horn, L. R. 1989. A Natural History of Negation. Chicago and London: The University of Chicago Press.

Hornstein, N. and J. Uriagereka. 2002. Reprojections. Epstein, S.D. and T. D. Seely. (eds.). Derivation and Explanation in the Minimalist Program, 106-132. Malden, Mass.: Blackwell.

Huddleston, R. and G. K. Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press.

Jack, G. B. 1978. Negation in Later Middle English Prose. Archvum Linguisticum 9, 58-72.

Jespersen, O. 1917. Negation in English and Other Languages. [Reprinted in Selected Writings of Otto Jespersen, 3-151, Tokyo: Senjo, 1937].

Page 27: A Transferred Negation ... - The Grammar Exchange

Lakoff, G. 1970. Pronominalization, Negation and the Analysis of Adverbs. Jacobs, R. and P. Rosenbaum. (eds.). Readings in English Transformational Grammar, 145-165. Waltham, Mass.: Ginn and Company.

Lakoff, R. 1969. A Syntactic Argument for Negative Transportation. Binnick, R. I., A. Davidson, G. M. Green and J. L. Morgan. (eds.). Papers from the Fifth Regional Meeting of the Chicago Linguistic Society, 140-147. Chicago: University of Chicago.

Lightfoot, D. 1999. The Development of Language: Acquisition, Change, and Evolution. Malden, Mass.: Blackwell.

Lindholm, J. M. 1969. Negative-raising and Sentence Pronominalization. Binnick, R. I., A. Davison, G. M. Green and J. L. Morgan. (eds.). Papers from the Fifth Regional Meeting of the Chicago Linguistic Society, 148-158. Chicago: University of Chicago.

Lyons, J. 1977. Semantics. Cambridge: Cambridge University Press.Palmer, F. R. 1990. Modality and English Modals. 2nd ed. London and New

York: Longman.Park, S. 2002a. A Diachronic Approach to Bare Infinitives. Studies in Modern

Grammar 28, 129-150._____. 2002b. A Diachronic Analysis of the Grammaticalization of English

Modal Auxiliaries. Studies in Modern Grammar 30, 1-39.Plank, F. 1984. The Modals Story Retold. Studies in Language 8, 305-364.Quirk, R., S. Greenbaum. G. Leech and J. Svartvik. 1985. A Comprehensive

Grammar of the English Language. London and New York: Longman. Turner, K. 1985. Categorization, Meaning, and Change in the English Modal

System. Doctoral dissertation. University of Alabama.Warner, A. 1993. English Auxiliaries: Structure and History. Cambridge:

Cambridge University Press.

Department of EnglishPusan University of Foreign Studies55-1, Uam-dong, Nam-guBusan 608-738, Korea

Page 28: A Transferred Negation ... - The Grammar Exchange

051-640-3052, [email protected]: October 10, 2003Accepted: November 20, 2003