No Negation is not Ambiguous Negative Ra

122226 SE Linguistics Seminar: Usage-based grammar / BA Paper WS 2013/14

No negation is not ambiguous:

negative raising

Lecturer: Priv. Doz. Dr. Mag. Gunther Kaltenböck, MA BA Paper English and American Studies A 033 612

Nico Sommerbauer Matrikelnummer: 0908059 nico.sommer@hotmail.com

List of figures

Figure 1 Frequency of

I don’t think in American English between 1810 and 2009

Figure 2 Frequency of I do not think in American English between 1810 and 2009 19 Figure 3 Frequency of not supposed to in American English between 1810 and 2009

20 Figure 4 Frequency of

21 Figure 5 Frequency of not supposed to in the GloWbE

I don’t think in the GloWbE

List of tables

Table 1 Results for

16 Table 2 Frequency of

I don’t/do not think and not supposed to in COCA and BNC

I don’t think and not supposed to in the ICE-GB

1. Introduction

The present paper deals with the issue of negation raising, from here on referred to as neg-raising. The research question for this paper is what is negation raising, how and how often do speakers use it, and what is its purpose?

Section 2 will offer the theoretical background for this paper; in subsection 2.1, we will establish a short definition of negation in order to be able to then define neg-raising in subsection 2.2. Then, we will look at some of the most widespread approaches to neg- raising in subsection 2.3, starting with the syntactic approach – the oldest approach to neg-raising – before moving on to approaches within the semantic and pragmatic paradigm of linguistics, and finally discussing the idea that neg-raising as its own independent phenomenon does not exist and is merely a manifestation of the general ambiguity of negation. Section 3 will focus on the corpus study conducted in producing this paper in an attempt to answer the question how and how often do speakers use [neg-raising]? Subsection 3.1 will give an overview of the methodology, outlining which corpora were used and to what end before presenting the results and possible interpretations in subsection 3.2. Section 4 of the present paper will then discuss why people may use neg-raising in an attempt to respond to the question what is its purpose? Due to its scope – and perhaps also due to the nature of neg-raising – this paper is not able to exhaustively answer all parts of its research question; it will, however, give some information on each part of the question. Further research especially on matters of frequency of usage and purpose of neg-raising should definitely

be conducted, and this paper may offer an attentive reader numerous ideas of what they could investigate in greater detail.

2. What is neg-raising?

In this section, we will establish a definition for and different approaches towards negative raising – from here on referred to as ‘neg-raising’. In order to do this, we must first define the term negation so as to be able to define neg-raising, as the latter is intrinsically connected to the former.

2.1 Defining negation

In order to define neg-raising, we must first define negation. Before we can look deeper into a possible definition, it has to be clarified that the term ‘negation’ is intrinsically ambiguous in the sense that it can refer to the operation of negating (something), to the operator utilized to achieve this (a sign of negation, such as not) or the result of said operation. These three elements are of course closely connected to each other. As such, I will use the term ‘negation’ to refer to the operation, like Napoli (2006: 234). Intuitively, one would probably say that negation is the reversal of an affirmative sentence, in order words there is a “functional asymmetry of affirmation and negation” (Horn 1989: 202). This is true in some cases, such as the examples in (1).

(1) a. The world is not flat.

b. He is not at home.

c. I do not owe you anything.

However, as Giora (2006: 979) points out, there are many situations in which negation is used as a mitigation device rather than to reverse the affirmative sentence. Consider the examples in (2).

(2) a. Sebastian is not tall; he is about average in height.

b. The restaurant is not far from here. It’s about a 20 minute drive away.

In these examples, the antonyms to tall and far would be short and near, respectively. But the examples in (2) do not mean Sebastian is short or the restaurant is near. Here, the negation functions as a means of scaling rather than a reversal of the affirmative.

Additionally, negation can be used as a rhetoric device to express the notion that an adjective is not quite strong enough to express the intended meaning. Consider the examples in (3).

(3) a. The food wasn’t good. It was delicious!

b. His acting wasn’t just bad, it was downright awful!

In utterances such as the ones above, negation does not act as a reversal of affirmation. While the initial statement is negated, the second part of the utterance elaborates on the speaker’s intentions, clarifying that the first part of the utterance is not rejected as incor rect, but rather as too weak to express the speaker’s opinion. In a sense, this could

be labeled as a kind of ‘phrasal intensifier’, as the speaker utilizes an entire utterance to replace a given adjective with a stronger adjective carrying the same base meaning.

Another ‘irregular’ use of negation – meaning it does not express the reversal of an affirmation – is scalar-implicature negation or denial. This means that a negated

sentence including a term for some or a similar meaning implies that the subject of the negated sentence actually applies to all rather than some. For illustration, see (4).

(4) The sun is not larger than some planets: it is larger than all planets. (Horn 1989 quoted in Davis 2011: 2549).

This is an important form of negation for our discussion of neg-raising because there is a scalar-implicature based approach towards neg-raising which we will look at in greater detail in subsection 2.3.2 with regards to Romoli (2013).

Beukeboom, Finkenauer and Wigboldus (2010: 978) argue that a “negation bias” exists which makes it more likely for people to refer to “stereotype-consistent behavior” using affirmative sentences and to “stereotype-inconsistent behavior” by using negated

sentences. This means, for example, that if Tracey is a woman, speakers would be likely to produce a sentence such as Tracey is short, but if Tracey is a man, they are more likely to produce a sentence such as Tracey is not very tall. This difference occurs because it is a general assumption – a stereotype – that men are taller than women. Their argument is mostly built around more social stereotypes, such as relating certain professions or social groups to different levels of intelligence or education, but I suggest that this negation bias can be observed just as much in examples relating to more physical factors such as above.

Beukeboom, Finkenauer and Wigboldus (2010: 980-988) also conducted several studies with students of VU University Amsterdam, all of which confirmed their argument that stereotype-inconsistent behavior is expected to be expressed via the use of negation whereas stereotype-consistent behavior is expected to be expressed via the use of affirmation. This is interesting to the topic of this paper because part of the research question is ‘why do speakers use neg-raising?’, and pragmatic issues such as stereotyping and social biases are of relevance to this discussion.

I have stated at the beginning of this section that the term negation is ambiguous due to the fact it can refer to a negative operation, the operator used for it or the result of the operation. However, negation is also ambiguous in another sense: many negations can

be interpreted in more than one way; indeed, it can even be argued that all negated utterances are semantically ambiguous. As Kjellmer (2000: 122) points out, a sentence

such as the title of his article “no work will spoil a child” can be interpreted in more than one way; either that not working will spoil a child, or that there is no work which will spoil

a child. Kjellmer (2000: 123) refers to a study conducted by Tottie (1980) which led to a child. Kjellmer (2000: 123) refers to a study conducted by Tottie (1980) which led to

This is an argument also brought forth, for example, by Napoli (2006: 250-251). He argues that neg-raising may not be an actual phenomenon, but rather “another instance of the ambiguity of negation”. We will discuss this point in more detail in subsection

2.2 Defining neg-raising:

After establishing a definition of negation, we can now move on to defining neg-raising. In essence, neg-raising refers to the process of shifting the negative operator to a position ahead of where it ‘should’ be (hence the term raising), in other words to move it from the nested clause to the main clause. However, this is only possible with certain verbs, commonly referred to in the literature as “neg-raising predicates” (Gajewski 2005). Consider the examples in (5).

(5) a. I do not think that I will be able to make it in time.

b. Sandra did not believe she could play the song perfectly.

Now consider the generally perceived meaning of (5a) in (6a) on the one hand and what it could also be interpreted as – and what is the logically more sound interpretation – in (6b) on the other hand.

(6) a. I think I will not be able to make it in time.

b. I am not thinking about whether I will be able to make it in time or not; I have no opinion on this matter.

In other words, neg-raising moves the negative operator into a position that is technically illogical, but is intuitively comprehensive to native speakers. Some speakers would even argue that (5a) is unambiguous and can only mean (6a), despite the fact that (6b) makes more sense as an interpretation from a purely logical and grammatical standpoint.

The literature speaks of Neg-Raising Predicates (NRPs), since neg-raising is connected to certain predicate verbs which allow neg-raising, while all other verbs do not. This assumption can be found largely uncontroversially in the literature regardless of the author’s approach to and view of the subject of neg-raising. The only somewhat controversial issue occurs in the form of some verbs which may or may not allow neg- raising; this has been pointed out by Collins and Postal (2012: 3; 7) amongst others. Gajewski (2005: 12) quotes Horn’s (1978) list of NR predicates, assigned to certain semantic classes. It is reproduced below to give a general overview of the verbs relevant for the present discussion on neg-raising; note again that some of the verbs in this list may be considered to not be stri ct NRPs depending on the speaker’s dialect, and some verbs not listed may be strict NRPs in some speakers ’ dialect.

The classes of Neg-Raisers

a. [OPINION] think, believe, expect, suppose, imagine, reckon

b. [PERCEPTION] seem, appear, look like, sound like, feel like

c. [PROBABILITY] be probable, be likely, figure to

d. [INTENTION/VOLITION] want, intend, choose, plan

e. [JUDGMENT/OBLIGATION] be supposed, ought, should, be desirable, advise (Horn 1978, quoted in Gajewski 2005: 12)

To this date, no completely uncontroversial explanation as to why these verbs and not others allow neg-raising can be found in the literature. It has been proposed that there are semantic reasons behind which verbs allow neg-raising and which do not, but since some verbs in the English language that carry near-synonymous meaning differ in their To this date, no completely uncontroversial explanation as to why these verbs and not others allow neg-raising can be found in the literature. It has been proposed that there are semantic reasons behind which verbs allow neg-raising and which do not, but since some verbs in the English language that carry near-synonymous meaning differ in their

(7) a.

I don’t want that to happen again.

b. I don’t *desire for that to happen again.

(8) a . I don’t *hope for that to happen again.

b. Ger.: Ich hoffe nicht, dass das jemals wieder passieren wird.

Most speakers would agree that (7a) is a perfectly correct English sentence, while (7b) is not, despite the fact that the verbs want and desire are both part of the semantic category of verbs of intention or volition and are near-synonymous. Therefore, these two verbs offer resistance to a purely semantic reasoning for why some verbs allow neg- raising whilst others do not. (8b) means that the speaker hopes that (that) will never happen again. However, the English sentence in (8a) does not mean the same thing, even though the predicate verb hope is the exact synonym of the German verb hoffen; in fact, these verbs even share the same ancestor. This cross-linguistic variation, which has also been observed by Gajewski (2005: 90) and Horn (1989), again resists a general semantic explanation for the difference between NRPs and non-NRPs.

In the past, it has also been proposed that neg-raising is nothing more than an idiomatic expression, for example by Quine, who refers to neg- raising as a “quirk of English usage” (1960: 145-146). However, as Napoli (2006: 251) points out, idioms are typically confined to one language, and neg-raising can be observed in many languages (Collins and Postal 2013: 1). Furthermore, since not only negation itself, but also negative quantifiers such as nobody and adverbs such as never are capable of triggering neg- raising, there would either have to be separate idiomatic rules for each particular case or all other triggers would have to be decomposed “into some element and a negation that In the past, it has also been proposed that neg-raising is nothing more than an idiomatic expression, for example by Quine, who refers to neg- raising as a “quirk of English usage” (1960: 145-146). However, as Napoli (2006: 251) points out, idioms are typically confined to one language, and neg-raising can be observed in many languages (Collins and Postal 2013: 1). Furthermore, since not only negation itself, but also negative quantifiers such as nobody and adverbs such as never are capable of triggering neg- raising, there would either have to be separate idiomatic rules for each particular case or all other triggers would have to be decomposed “into some element and a negation that

2.3 Approaches to neg-raising

In the literature, several different approaches towards neg-raising can be found. This subsection will give an overview of the most important and widespread approaches, summarized as syntactic approach and semantic/pragmatic approach respectively. The latter includes Romoli’s (2012; 2013) scalar-implicature based approach towards neg- raising, which aims to point out flaws of the presuppositional approach offered by Gajewski (2007); the syntactic approach was first proposed by Fillmore (1963) and more recently by Collins and Postal (2012). Finally, we will expand on Napoli’s (2006) argument touched upon in subsection 2.1 that neg-raising as its own distinct phenomenon does not exist and that it is merely an instance of the general ambiguity of negation.

2.3.1 The syntactic approach

Fillmore (1963) first proposed the syntactic approach to neg-raising, coining the term negation-raising in his article. He argues that in a neg-raised sentence, the negation is actually produced in the embedded clause – and is also interpreted in it – but the negation then rises above the predicate and appears in front of it. This approach assumes that the intended meaning of a given neg-raised sentence stems from the original ‘un-raised’ sentence, and the raising itself is simply a syntactic operation that does not influence the meaning (Collins and Postal 2012: 1). It makes perfect sense that the term negation-raising or neg-raising originates from the syntactic approach, because it describes a syntactic operation. The transformational syntactic approach argues that Fillmore (1963) first proposed the syntactic approach to neg-raising, coining the term negation-raising in his article. He argues that in a neg-raised sentence, the negation is actually produced in the embedded clause – and is also interpreted in it – but the negation then rises above the predicate and appears in front of it. This approach assumes that the intended meaning of a given neg-raised sentence stems from the original ‘un-raised’ sentence, and the raising itself is simply a syntactic operation that does not influence the meaning (Collins and Postal 2012: 1). It makes perfect sense that the term negation-raising or neg-raising originates from the syntactic approach, because it describes a syntactic operation. The transformational syntactic approach argues that

After Fillmore first proposed this theory, it was largely agreed in the 1960s and 70s that neg-raising was indeed a syntactic operation producing sentence pairs which differed in their syntactic structure, but shared the same meaning because the syntactic movement was presumed to have taken place after the original sentence – along with its intended meaning – was already produced. Fillmore (1963) as well as most other defenders of the syntactic approach, such as Horn (1972), Shlonsky (1988) apply a transformational rule to neg-raising; however, some linguists have argued in favor of a non-transformational syntactic account of neg-raising, which can be applied if the author applies no

generative rule at all, but rather utilizes “a framework assuming to be a grammar to be a model- theoretic system” (Collins and Postal 2012; 1). Among them are Johnson and Postal (1980), Pullum and Scholz (2001) and Collins and Postal (2012).

Collins and Postal’s (2012: 7) non-transformational syntactic approach towards neg- raising argue that “classical NR is sensitive to syntactic islands”, referring primarily to

wh-islands, and that this fact provides strong evidence against the semantic/pragmatic approaches we will discuss in the following subsection. Collins and Postal (2012: 47-50) go on to argue that the phenomenon of never-raising cannot be properly explained by any non-syntactic approach, and that since classical neg-raising and never-raising are very similar – they suggest they are even “two instances of a single phenomenon” (50) – wh-islands, and that this fact provides strong evidence against the semantic/pragmatic approaches we will discuss in the following subsection. Collins and Postal (2012: 47-50) go on to argue that the phenomenon of never-raising cannot be properly explained by any non-syntactic approach, and that since classical neg-raising and never-raising are very similar – they suggest they are even “two instances of a single phenomenon” (50) –

2.3.2 The semantic/pragmatic approach

The semantic/pragmatic approach towards neg-raising has been the most prevalent approach in recent publications, and arguably even since Fillmore (1963) first proposed the syntactic approach, given publications such as Jackendoff (1971), Bartsch (1973) and Horn (1978). Horn is especially interesting because his earlier works (Horn 1972) argue in favor of the syntactic approach, but he later became one of the most important representatives for the semantic/pragmatic approach (Horn 1978, 1984, 1989, 1993 and more). Other defenders of this approach include Kas (1993), Zwarts (1993), Heim (2000), and more recently Gajewski (2005; 2007) and Romoli (2012; 2013).

Among the representatives of the semantic/pragmatic approach, Bartsch (1973) is the first to claim a presuppositional approach towards neg-raising. She argues that there is a so- called ‘Excluded Middle’ which speakers assume the listener can intuitively infer to be implied, and listeners infer to be implied. An example for this ‘Excluded Middle’ is given in (9).

(9) a. I don’t believe that Mary has eaten.

b. I believe that Mary has not eaten.

c. I have an opinion on whether Mary has eaten. (EXCLUDED MIDDLE)

This means that (9b) arises in the listener’s mind as the meaning of (9a) due to the ‘Excluded Middle’ presupposition in (9c). According to Bartsch (1973), these conditions

typically apply and neg-raising takes place; however, if they do not apply, they are no longer required and the expression can be uttered regardless. Bartsch allows this to be typically apply and neg-raising takes place; however, if they do not apply, they are no longer required and the expression can be uttered regardless. Bartsch allows this to be

On the pragmatic side, Horn’s (1989) R-implicature approach is the most widespread and influential. Horn builds his theory upon the Excluded Middle implicature discussed above and introduces what he calls the R-principle. This means that the listener always reads as much as possible into any utterance; this allows neg-raising to be understood as intended by the speaker because the listener will presumably infer a contrary negation from a contradictory negation. However, Horn notes that this cannot be applied when the difference in literal meaning between a high negation and a low negation is too big, as “[t]he meaning that would result is too different from the literal meaning“ (Gajewski 2005: 23). The strength of the negation is also a factor. Horn states that with increasing distance between the negation and the negated clause, uncertainty on the side of the speaker pertaining to the negation also increases. Horn calls this the Uncertainty Principle , and argues that whenever a sentence’s meaning clashes with the Uncertainty Principle, neg-raising cannot take place because the R-implicature cannot

be applied. He goes on to state that this is the case with factives, which explains why there are no factive NRPs.

Additionally, in the semantic/pragmatic paradigm, there is the so-called scalar- implicature based approach. This approach is built o n Grice’s (1975) well-known Maxims of Communication. It argues that speakers expect others to adhere to these maxims, most notably for this context the maxim of quantity which states that if a speaker has several sentences he or she could use in a conversation, they should use the most informative and relevant one that they also believe to be true. The last part – they have to believe it to be true- is typically referred to as the maxim of quality (see Gajewski 2005: 77-78). If several sentences that are logically independent, but are informative, relevant and believed to be true are options, they should all be uttered (Romoli 2012:

199). In the context of neg-raising, this means that the scalar-implicature based approach argues that neg-raising is used by speakers who adhere to these maxims of communication if it is not less informative than any other possible version of the utterance and the speaker believes it to be true.

2.3.3 The general ambiguity of negation approach

This is, in a way, a non-approach towards neg-raising. It can be argued that neg-raising as its own independent phenomenon does not exist and is merely a symptom of the general ambiguity of negation. Napoli (2006: 250-251) argues that it may be nothing more than the accidental or ambiguous transformation of a negation from the contrary to the contradictory. In other words, ‘not believe’ is turned to ‘disbelieve’ because it makes no practical difference whether one or the other form is used or understood. Napoli’s (2006: 249) argument that it does not make any practical difference whether John in Chomsky’s classic example “John does not like mushrooms” actively dislikes mushrooms or simply does not particularly like them is rather convincing. In Napoli’s words, “[y]ou will not serve mushrooms to a friend even if he does not particularly like them rather than positively disliking them”. As such, this approach argues that neg- raising as an individual linguistic phenomenon does not exist and is nothing more than another symptom of the general ambiguity of negation and the indifference of speakers towards differentiations that lead to no practical difference. This explanation is very simple and convenient; but it can be argued that it does not solve anything and rather just removes the proverbial nametag off the phenomenon of neg-raising, which still remains in some fashion, regardless of what it may be referred to as.

3 Corpus research

3.1 Methodology

Corpus research can give us great insight into the frequency of usage of certain phrases, expressions and terms over time as well as in different geographical regions. Part of the research question of the present paper was ‘how and how often do speakers use neg- raising?’ In an attempt to answer this question, I conducted corpus research in the Corpus of Contemporary American English (henceforth referred to as COCA), the British National Corpus (BNC), the Corpus of Historical American English (COHA), the Corpus of Global Web-Based English (GloWbE) and the British Component of the International Corpus of English (ICE-GB). The COCA and the BNC are large corpora – consisting of approximately 450 million and 100 million words respectively – of contemporary English in the respective geographical region, which gives us opportunity to compare the frequency of usage of typical neg-raising phrases between Americans on the one hand and British people on the other hand. The COHA is a vast corpus of historical American English, comprised of approximately 400 million words from the years 1810 to 2009. This corpus will allow us to look at the development of the frequency of usage of common neg-raising phrases in American English over the last 200 years. At almost two billion words, the GloWbE is the largest corpus we will look at. It is composed of texts on websites from twenty different English-speaking countries, and as such allows us to compare a vast amount of data from all over the world. Finally, the ICE-GB is a much smaller corpus, comprised of only about one million words, but is nonetheless a decently sized corpus of contemporary British English which allows us comparisons with the BNC. Unfortunately, due to the construction’s complexity and the different shapes it can take, it cannot be searched for as a structure in a corpus; instead, the present study investigated two phrases that are particularly typical for neg-raising, namely

I don’t/do not think and not supposed to.

3.2 Corpus study and results

3.2.1 COCA and BNC: geographical data

One of the main goals of this corpus study was to compare the frequency of usage of neg-raising between British and American English. For this purpose, I compared the results for the most typical neg-raising phrase,

I don’t/do not think, in the COCA on the one hand and the BNC on the other hand. Then, I compared the frequency of usage of the neg-raised phrase not supposed to in the two corpora in order to be able to compare whether there are similarities or differences in frequency of use; the assumption being that if the results for both are similar, conclusions can probably be drawn and applied to the phenomenon of neg-raising as a whole in British and American English; whereas, if the results were to differ significantly from each other, any conclusions may only be applied to a specific phrase rather than to neg-raising in general. Table 1 details the results, normalized to frequency of occurrence per one million words for both corpora in order to simplify comparisons. In the case of the phrase

I don’t/do not think, not only were both the contracted as well as the non-contracted form taken into consideration; additionally, the results of searching for the phrase

I don’t/do not think so were subtracted from the final numbers, as this phrase is not neg-raised. Any occurrence of a phrase such as I don’t/do not think about… was also looked at, but this phrase occurred so rarely – in fact, it does not occur at all in the BNC – that it is in no way significant.

Table 1 Results for I don’t/do not think and not supposed to in COCA and BNC

Not supposed to section

phrase

I don’t/do not think

As table 1 shows, the frequency of usage of both neg-raising phrases I don’t/do not think and not supposed to is significantly higher overall in American English than in British English. This result is very interesting, as there is nothing in the literature or intuitively that would suggest either variety to be more or less likely to utilize neg-raising. Considering the vast size of both corpora, the results can probably be regarded as rather representative of each variety, making this significant difference all the more intriguing. A closer look at the different sections of the corpora makes this issue yet more interesting: in the largest category in both corpora – Spoken -, the phrase in question actually occurs significantly more often in the BNC (349 occurrences per one million words) than it does in the COCA (276 occurrences per one million words). This is

a difference of almost a third of total occurrences, and a very similar difference in favor of the British corpus can be observed in the section Fiction, with 102.5 occurrences per one million words in the BNC and only 63 occurrences per one million words in the COCA; skimming the individual results in Fiction led to the conclusion that the vast majority of these cases are to be found in dialogues, which are of course fictional reproductions of spoken conversation and could as such be regarded as a category a difference of almost a third of total occurrences, and a very similar difference in favor of the British corpus can be observed in the section Fiction, with 102.5 occurrences per one million words in the BNC and only 63 occurrences per one million words in the COCA; skimming the individual results in Fiction led to the conclusion that the vast majority of these cases are to be found in dialogues, which are of course fictional reproductions of spoken conversation and could as such be regarded as a category

style between Americans on the one hand and the British on the other hand; perhaps American newspaper journalists tend to use more colloquial and personal language, or perhaps they simply include more direct quotations in their articles than their British colleagues do. The rather low frequency of the phrase in Academic context can be explained by the fact that the phrase in question is relatively colloquial and therefore does is not very compatible with academic language; it is also a rather vague phrase which suggests uncertainty. While expressing uncertainty is necessary in certain contexts in academic language, this task is typically accomplished through the usage of other phrases specifically suited for the academic context.

The other analyzed phrase in this context- not supposed to- generally occurs much less frequently than

I don’t/do not think in both corpora. Where the total numbers per one million words for the latter phrase are 93 in the COCA and 77 in the BNC, they are only

4.8 in the COCA and 3 in the BNC for the former. However, the general pattern of significantly higher frequency in the COCA than in the BNC still applies and as such supports the idea that Americans are more likely to neg-raise than British people. The numbers for the section Newspapers show an even larger difference in favor of the COCA, with 3.5 per one million words versus 1 per one million words in the BNC, while the numbers for the section Spoken again reflect a higher frequency in the BNC. As such, our results are largely consistent and appear to suggest the more general idea that British people neg-raise more frequently than Americans in spoken language, but American newspaper journalists either use neg-raising phrases more frequently than their British colleagues in their articles or include significantly more direct quotations involving such phrases. It should also be noted that searching for the ‘non-neg-raised’ version of not supposed to, namely supposed to not, leads to a very small number of 4.8 in the COCA and 3 in the BNC for the former. However, the general pattern of significantly higher frequency in the COCA than in the BNC still applies and as such supports the idea that Americans are more likely to neg-raise than British people. The numbers for the section Newspapers show an even larger difference in favor of the COCA, with 3.5 per one million words versus 1 per one million words in the BNC, while the numbers for the section Spoken again reflect a higher frequency in the BNC. As such, our results are largely consistent and appear to suggest the more general idea that British people neg-raise more frequently than Americans in spoken language, but American newspaper journalists either use neg-raising phrases more frequently than their British colleagues in their articles or include significantly more direct quotations involving such phrases. It should also be noted that searching for the ‘non-neg-raised’ version of not supposed to, namely supposed to not, leads to a very small number of

3.2.2 COHA: historical data from 1810-2009

In this subsection, we will investigate the frequency of use of the two phrases previously looked at in American English over the last 200 years. For this purpose, the COHA was used. Figure 1 shows a largely consistent increase in frequency of use of the phrase I don’t think in American English from 1810 to 2009. If we regard the data for the non- contracted version I do not think, displayed in figure 2, we can see that there has been a development in the opposite direction: there has been a mostly consistent decrease in usage. However, this can be explained by the fact that – naturally – the percentage of spoken texts among the corpus data increases the closer we get to the present, and speakers are significantly more likely to contract in spoken language than they are in written language. More importantly, the raw numbers for I do not think are so much lower than those for

I don’t think that even if one was to simply combine the results, a significant increase would still be the outcome. While these results may suggest an increase in frequency of neg-raising over time in American English, it can be argued that since we have only investigated one particular phrase, it may only be the use of this one phrase that has increased; this argument is supported by the idea of the increasing I don’t think that even if one was to simply combine the results, a significant increase would still be the outcome. While these results may suggest an increase in frequency of neg-raising over time in American English, it can be argued that since we have only investigated one particular phrase, it may only be the use of this one phrase that has increased; this argument is supported by the idea of the increasing

Figure 1 Frequency of

I don’t think in American English between 1810 and 2009

Figure 2 Frequency of I do not think in American English between 1810 and 2009

However, a look at figure 3 reveals that there are at the very least two phrases typical for neg-raising which have experienced a marked increase in frequency of usage in American English over the last 200 years, according to the COHA: the phrase not supposed to has, in fact, increased in usage even more dramatically and consistently than

I don’t/do not think. Searching for supposed in the COHA leads to a more or less even distribution over the years, while searching for suppose even results in a gradual decrease of frequency. These results suggest that the popularity of the words suppose or supposed without the neg-raised negation is not responsible for the observed increase in frequency of the latter.

While not entirely conclusive, the consistent significant increase in frequency of both phrases strongly suggests that neg-raising has been enjoying increasing popularity in American English over the last 200 years. This is very intriguing and deserves a more detailed investigation, as well as a comparison with historical corpus data of British English. Unfortunately, these investigations would exceed the scope of the present paper; the data presented in this section, however, does offer very interesting possibilities for further research into historical development of neg-raising as well as a closer investigation of potential reasons behind this apparent increase. The latter issue is also discussed in section POSSIBLE REASONS FOR NEG-RAISING of the present paper.

Figure 3 Frequency of not supposed to in American English between 1810 and 2009

3.2.3 The GloWbE: an international web-based comparison

Searching for the phrase

I don’t think in the Corpus of Global Web-Based English, which is comprised of 1.9 billion words taken from 1.8 million web pages in 20 English- speaking countries, leads to the results in figure 4.

Figure 4 Frequency of

I don’t think in the GloWbE

The first thing to note is that, yet again, the frequency of usage in American English is higher than in British English, which corresponds perfectly with our previous results. It can also be clearly observed that the frequency of usage of this phrase is highest in American English, British English and Australian English in this order. Canadian and Singaporean English correspond almost exactly with the overall average frequency in the twenty national varieties included in the GloWbE, whereas Irish and New Zealand English – as the remaining ‘inner circle’ varieties – are to be found fairly significantly below the average frequency. However, their frequencies are still higher than those of any outer circle varieties besides Singaporean English, allowing a general proposition that neg-raising – or at the very least the usage of the phrase I don’t think – occurs more frequently in inner circle varieties than it does in outer circle varieties. This conclusion makes sense if we consider neg-raising to be a pragmatic presupposition, because such presuppositions tend to be more pronounced the more proficient a speaker – or writer – is in the language in question; however, the exact same argument could be made in favor of the syntactic approach: if the speaker is more proficient in the language, he or she is more likely to be – at least subconsciously – aware of underlying syntactic rules, and, as such, to apply them. Why Americans use it more strongly than any other variety, however, is a very intriguing question that cannot be answered in the present paper and deserves closer investigation; the same applies to the relatively low frequency in Irish and New Zealand English on the one hand and the remarkably high frequency in Singaporean English on the other hand. I can only offer the tentative suggestion that if neg-raised utterances are indeed interpreted to be more polite, this may be a reason as The first thing to note is that, yet again, the frequency of usage in American English is higher than in British English, which corresponds perfectly with our previous results. It can also be clearly observed that the frequency of usage of this phrase is highest in American English, British English and Australian English in this order. Canadian and Singaporean English correspond almost exactly with the overall average frequency in the twenty national varieties included in the GloWbE, whereas Irish and New Zealand English – as the remaining ‘inner circle’ varieties – are to be found fairly significantly below the average frequency. However, their frequencies are still higher than those of any outer circle varieties besides Singaporean English, allowing a general proposition that neg-raising – or at the very least the usage of the phrase I don’t think – occurs more frequently in inner circle varieties than it does in outer circle varieties. This conclusion makes sense if we consider neg-raising to be a pragmatic presupposition, because such presuppositions tend to be more pronounced the more proficient a speaker – or writer – is in the language in question; however, the exact same argument could be made in favor of the syntactic approach: if the speaker is more proficient in the language, he or she is more likely to be – at least subconsciously – aware of underlying syntactic rules, and, as such, to apply them. Why Americans use it more strongly than any other variety, however, is a very intriguing question that cannot be answered in the present paper and deserves closer investigation; the same applies to the relatively low frequency in Irish and New Zealand English on the one hand and the remarkably high frequency in Singaporean English on the other hand. I can only offer the tentative suggestion that if neg-raised utterances are indeed interpreted to be more polite, this may be a reason as

Intriguingly, entering the phrase not supposed to into the GloWbE leads to very different and thus unexpected results, as can be seen in figure 5.

Figure 5 Frequency of not supposed to in the GloWbE

While the overall numbers are significantly smaller than those for

I don’t think – which is consistent with our previous findings in the other corpora – the total numbers are still high enough to be considered statistically relevant due to the vast size of the GloWbE. The graph shows a vastly higher frequency in Nigerian English than it does in any other variety; in fact, its frequency is more than twice as high as the average frequency across all twenty national varieties. The only other varieties that are significantly above the average are Pakistani English, American English and – to a lesser degree – Singaporean English. These results are quite unexpected and a clear counterargument to the previously proposed argument that neg-raising may simply occur more often in inner circle varieties due to higher English proficiency. The only inner circle variety significantly above the average for not supposed to is US American English; Canadian English is just barely above the average, while all other inner circle varieties are even I don’t think – which is consistent with our previous findings in the other corpora – the total numbers are still high enough to be considered statistically relevant due to the vast size of the GloWbE. The graph shows a vastly higher frequency in Nigerian English than it does in any other variety; in fact, its frequency is more than twice as high as the average frequency across all twenty national varieties. The only other varieties that are significantly above the average are Pakistani English, American English and – to a lesser degree – Singaporean English. These results are quite unexpected and a clear counterargument to the previously proposed argument that neg-raising may simply occur more often in inner circle varieties due to higher English proficiency. The only inner circle variety significantly above the average for not supposed to is US American English; Canadian English is just barely above the average, while all other inner circle varieties are even

I don’t think, as previously mentioned, which means that our previous observations are somewhat confirmed in this regard. American

English being above the other inner circle varieties also corresponds with our previous findings; however, the rest of figure 5 contradicts our previous data quite profoundly. The extremely high frequencies of occurrence in Pakistani and especially Nigerian English are particularly intriguing; perhaps there are certain cultural or cross-linguistic reasons at work here, such as an expression in the native languages of these peoples which shares

a similar structure with not supposed to, making it an attractive phrase for native speakers of the languages in question. However, any possible reasons behind this can only be guessed at in this paper. The fact that the phrase not supposed to is used less

frequently in all inner circle varieties except for North America (where Canada’s frequency is barely above the average while that of the US is about 30% above the average) is also intriguing. Due to the nature of the GloWbE data, it is difficult to make any conclusions of overall usage based only on web-based data; however, this difference is significant enough to be taken under closer scrutiny. Based on only this data, it seems that US Americans specifically use the phrase very frequently, and the somewhat higher frequency of Canadian usage compared to frequency in the other inner circle varieties could perhaps be explained through the geographical and cultural proximity of Canada to the US. However, why US Americans would use this phrase significantly more often in web-based texts than their British, Australian and New Zealand contemporaries can only be guessed at; there may be cultural reasons behind this discrepancy, or it can be traced back to a qualitative difference in sources of the texts offered by the GloWbE. Unfortunately, these sources cannot be examined and as such, a conclusive argument cannot be given in the present paper. Yet again, further research into the matter may be useful.

3.2.4 ICE-GB: further data on contemporary British English

The ICE-GB, as previously mentioned, is much smaller than the other corpora considered in this study; it consists of approximately one million words of contemporary

British English. Among these one million words, the phrase

I don’t think occurs 224 times. The phrase not supposed to only occurs six times in total in ICE-GB. Table 2 offers a detailed breakdown of occurrences.

Table 2 Frequency of

I don’t think and not supposed to in the ICE-GB

I don’t think

I don’t think so

Not supposed to

Spoken

Spoken Written 207

The occurrences of

I don’t think so are listed as they need to be subtracted from the total frequencies of

I don’t think, as I don’t think so is not neg-raised. This means that we end up with 177 occurrences of spoken

I don’t think as part of a neg-raised sentence and 16 occurrences of written

I don’t think as part of a neg-raised sentence in ICE-GB. These findings are consistent with my previous findings in the BNC detailed in table 1. In both corpora,

I don’t think occurs significantly more frequently in spoken English than it does in written English, and the same is true for not supposed to.

I don’t think also generally occurs much more frequently than not supposed to does, which is again a result consistent between both British corpora.

3.2.5 Conclusion

Due to the structural complexity of neg-raising, the present corpus study has looked at two typical neg-raising phrases:

I don’t/do not think and not supposed to. Data from COCA, the BNC, the GloWbE and ICE-GB suggest that in contemporary English, Americans use these phrases – and therefore presumably neg-raising – more frequently than other varieties of English. However, some anomalies which contradict this proposition were also observed; according to GloWbE, the outer circle varieties of Pakistani English and Nigerian English use the phrase not supposed to more frequently than any of the inner circle varieties. In contrast to this, the data for

I don’t think in the

GloWbE show that all inner circle varieties use the latter phrase more frequently than all outer circle varieties except for Singaporean English. These results are quite intriguing and could be an interesting subject for future studies. An examination of the same phrases in the COHA has led to the conclusion that both phrases have enjoyed an almost entirely consistent increase in usage; however, this can be at least partially explained by the fact that the percentage of spoken texts compared to written texts also increases gradually from 1810 to 2009 in the COHA, and these phrases are more typical for spoken than for written English; this last argument is supported by the results of the COCA, BNC and ICE-GB, which all show significantly higher frequencies in spoken than in written English.

4 Possible reasons for neg-raising

This final section of the present paper will discuss possible reasons for why speakers use neg-raising. These will be mostly my own ideas, with additional arguments or support from the literature.

The first reason I would like to propose is one that has been touched upon several times already in this paper: neg-raising may be used as a politeness marker. Politeness is, of course, a pragmatic issue and as such, this interpretation is an argument in favor of neg- raising as a pragmatic phenomenon. It is uncontroversial that a sentence such as (10a) is considered to be more polite than (10b). Taken at face value, both sentences carry the same meaning. However, the neg-raised version in (10a) moves the negation from the nested clause to the main clause. In other words, the part of the utterance that is negated is now the clause which includes the personal pronoun I and no longer the one that includes the personal pronoun you. While the speaker expresses the same idea on

a detonational level in both (10a) and (10b), the connotation changes: the phrase I don’t think pragmatically implies a greater uncertainty than

I think that […] negation. This implied greater uncertainty is an uncertainty on the side of the speaker, not the listener. Therefore, the speaker weakens his/her judgment of the listener, thereby committing a I think that […] negation. This implied greater uncertainty is an uncertainty on the side of the speaker, not the listener. Therefore, the speaker weakens his/her judgment of the listener, thereby committing a

a. I don’t think your statement is correct.

b. I think your statement is incorrect.

This argument is certainly intriguing; however, I would like to propose a counterargument. In political speeches, neg-raised phrases can frequently be heard, such as the following example taken from a speech by US President Barack Obama.

I don't think that's a satisfactory response. (obamaspeeches.com)

Dokumen yang terkait

AKIBAT HUKUM PENGGANTIAN NAMA ANAK TERHADAP LEGALITAS STATUS HUKUM ANAK (Studi Penetapan Pengadilan Negeri No : 11/PDT.P/2015/PN.BPP)

1 43 85

PENGARUH LAMA PENUTUPAN PINTU PERLINTASAN KERETA API TERHADAP TUNDAAN DAN PANJANG ANTRIAN KENDARAAN (Studi kasus pada perlintasan kereta api di JPL No 24 Jl Raya Karangsono – Sukorejo)

2 27 2

Gambaran Sanitasi Pada Kereta Api Ekonomi, Bisnis dan Eksekutif (Studi pada PT.Kereta Api Indonesia Daerah Operasi IX Jember) Description Of Sanitation In Economic, Business and Executive Railway (Studies in IX Jember Regional Operations The Indonesian Ra

1 13 7

JAR AK AT AP P UL P A T E RHAD AP T E P I I N S I S AL GI GI I NSI S I VU S S E NT RA L P E RM AN E N RA HAN G AT AS P AD A S UB RA S DE UT ROM E L AY U ( T in j au an L ab or at o r is d an Radi ol ogis )

0 35 16

KAJIAN YURIDIS COUNCORSUS (PERBARENGAN) PERKARA ISBATH NIKAH DENGAN CERAI GUGAT DI PENGADILAN AGAMA JEMBER (Studi Putusan Pengadilan Agama Jember No : 364/Pdt.G/2006/PA.Jr) Tanggal 23 Maret 2006

0 11 16

Perlindungan Hak Kesehatan Narapidana Dalam Pandangan Hukum Positif dan Hukum Pidana Islam di Lembaga Pemasyarakatan Kelas IIA Pemuda Tanggerang (Analisis Yuridis UU No 12 Tahun 1995)

0 17 0

Implementasi Kebijakan Peraturan Daerah Kota Bandung No 26 Tahun 2009 Tentang Kesetaraan Dan Pemberdayaan penyandang Cacat

0 8 1

On not learning about the mind from lite

0 1 42

ANALISIS PERTANGGUNGJAWABAN PIDANA PELAKU TINDAK PIDANA PENIPUAN PENERIMAAN CALON PEGAWAI NEGERI SIPIL PEMDA PROVINSI LAMPUNG (Studi Putusan No 859/Pid.B/2012/PN TK)

1 19 51

THE PERSISTENCE OF EARNINGS AFTER THE ADOPTION OF IFRS (The Study is to Investigate on Banking Sector at The Indonesia Stock Exchange) PERSISTENSI LABA SESUDAH ADPOSI STANDAR AKUNTANSI INTERNASIONAL (IFRS) (Studi Empiris pada Perbankan yang Terdaftar di B

1 26 42