On Invisible Language in Modern English

A Corpus-based Approach to Ellipsis

Evelyn Gandón-Chapela

• Éditeur: Bloomsbury Academic
• Année: 2020
• ISBN:
  9781350064539
• Édition: 1
• Sujet: Filología
• Sous-sujet: Sintaxis en lingüística

Winner: AEDEAN Leocadio Martín Mingorance Book Award for Theoretical and Applied English Linguistics 2020 This book investigates the syntactic phenomenon of ellipsis and the linguistic forces that trigger it. It presents the results of a corpus-based study which takes into account grammatical, semantic/discursive, usage-related and processing variables. Evelyn Gandón-Chapela builds upon the few empirical works on ellipsis in Present-day English to offer the first comparative analysis of ellipsis and its development throughout the recent history of the English language. Moreover, the book also provides a complex query algorithm which automatically detects and retrieves cases of ellipsis, leading to successful recall ratios, applicable to a wide range of parsed corpora.

1

Introduction

In a communication exchange, speakers may omit information when it can be inferred from the linguistic or extralinguistic context. This implies that addressees will need to decipher not only what has been said but also what has not in order to reach a full and correct interpretation. Therefore, the recipients of the information will have the task of filling in the blanks left by their interlocutor(s), or, in other words, they will need to interpret ellipsis. This entails that actual utterances as well as omissions are equally important for the success of the communication exchange, since silence is meaningful. Ellipsis is illustrated in (1):

(1) Daniel can speak five languages, but Joseph can’t speak five languages.¹

Example (1) is an instance of so-called VP Ellipsis² in which the elided verb phrase or VP (speak five languages) in the second conjunct can be retrieved from the first one, which serves as the antecedent.

Elliptical constructions do not occur freely. There are two main restrictions for them to be felicitous: the recoverability condition and the licensing condition.³ On the one hand, the recoverability condition alludes to the fact that elliptical constructions need to be recoverable from the context in which they take place, be it linguistic or extralinguistic (Quirk et al. 1985: 895ff; Aelbrecht 2010; van Craenenbroeck and Merchant 2013). For instance, the example in (2) would violate the recoverability condition if uttered out of the blue, since it would not be possible to infer its meaning due to the lack of an antecedent:

(2) *I know he will.⁴

On the other hand, the licensing condition dictates what can exactly be elided depending on the syntactic context in which ellipsis takes place (Zagona 1982; Lobeck 1995; Johnson 2001; Merchant 2001; Aelbrecht 2009, 2010). Aelbrecht (2009: 15) provides an example of this:

3) a. *Max having arrived and Morgan not having, we decided to wait.

b. Max had arrived, but Morgan hadn’t, so we decided to wait.

Although the VP could be easily retrieved from its surrounding context in the nonfinite clause, the English language only licenses the omission of the verb phrase that occurs in the finite clause ((3)b).

The mismatch between meaning (the intended message) and sound (what is actually pronounced) evinced in contexts of ellipsis poses a challenge to the traditional Saussurean concept of linguistic sign, defined as being composed of both ‘signifier’ (form) and ‘signified’ (meaning). This definition of the linguistic sign entails that every linguistic unit should have a form and a meaning for communication to be possible. However, as Merchant (2006) claims, in ellipsis there is significatio ex nihilo (‘meaning out of nothing’), that is, ‘there is meaning without form’ (Merchant 2013a: 1), which implies that the interpretation of an elliptical construction is richer than what is actually pronounced (Carlson 2002; Culicover and Jackendoff 2005; Aelbrecht 2009, 2010) and therefore speakers have no problems to understand the meaning of elliptical structures.

The following would be the criteria used for the identification of elliptical structures (Bîlbîie 2011: 129): (i) the syntax is apparently incomplete, since the material that would be necessary for the correct interpretation of a structure is missing; (ii) the elements belonging to the elliptical structure must be analysable as arguments, adjuncts or predicates of the elided material; and (iii) the interpretation of an elliptical structure is obtained by means of a linguistic or an extralinguistic context, thanks to the presence of an antecedent (either explicit or implicit).⁵ In addition, ellipsis also seems to defy Frege’s compositionality principle (Bîlbîie 2011: 129), echoed in Chomsky’s (1965: 136) words as follows: ‘The semantic interpretation of a sentence depends only on its lexical items and the grammatical functions and relations represented in the underlying structures in which they appear.’ This is so because elliptical sentences pose no problems for their actual interpretation even when some of their elements have been omitted.

As will be maintained in this volume, ellipsis is indeed a complex phenomenon for any theory of grammar because of its diverse characteristics, which fall between sentence and discourse grammar (Williams 1977; Gallego 2011). On the one hand, the sentence-grammar characteristics of ellipsis would be related to the different ellipsis types that have been attested and their properties (category, internal structure and morphological restrictions). The discourse-grammar characteristics present in ellipsis, on the other hand, would make reference to the context where the elliptical phenomena take place (linguistic and extralinguistic) and the thematic structure of sentences, which are formed by both given and new information (Gallego 2011). Elliptical sentences have usually been claimed to contain significant new information in discourse, avoiding unnecessary, old information.

The rationale behind the study of ellipsis, as put forward by Bîlbîie (2011: 130), is that it is present in all natural languages but not really understood in the grammar because there is the preconceived idea that ellipsis is governed by the ‘principle of minimum effort’ and is in free distribution with its non-elliptical counterpart. First, as a representative of the former view, Bîlbîie quotes Zribi-Hertz (1986), who claims that ellipsis remains as a choice made by the language user. Another aspect usually brought up with regard to ellipsis would be the claim that the use of ellipsis is one of the reasons why languages are so ambiguous (Hendriks and Spenader 2005: 29; Bîlbîie 2011: 130). Hendriks and Spenader (2005) and Bîlbîie (2011) confront this view by defending that ellipsis cannot be reduced to the mere instantiation of the principle of minimum effort. In ellipsis, as they contend, that said principle of minimum effort derives from the interaction of two antonymous principles which had already been captured in Grice’s (1975) quantity maxim. This maxim claimed that speakers should make their contribution as informative as is required for the purpose of the communicative exchange while at the same time not making it more informative than is required. When it comes to dealing with the phenomenon of ellipsis, this amounts to saying that one can make use of ellipsis as long as our interlocutor is able to decipher our message. Therefore, only when the information can be recovered may one speaker dispense of redundant information by means of ellipsis. Hendriks and Spenader (2005: 29–30) summarize the functions of ellipsis by stating that it allows us to express things which ‘are otherwise ineffable, disambiguate discourse structure, and serve as a rapport-creating device that could be relevant to automatic dialogue systems’.

Second, as will be shown below, there are numerous examples which evince differences between elliptical sentences and their non-elliptical counterparts. Indeed, there are cases where the use of ellipsis is the only means one can employ to build a grammatical sentence or obtain a certain interpretation (Bîlbîie 2011: 130). In addition, there is a great number of examples in the literature that instantiate that while certain cases of ellipsis are grammatical, their non-elliptical counterparts would be ungrammatical due to the violation of certain syntactic restrictions (like the presence of finite VPs in cases of Gapping,⁶ an elliptical construction exemplified in (4)) or the violation of the so-called island constraints (as in the example of Sluicing⁷ in (5), where there is a locality constraint). Ever since Ross (1969), it has been found that Sluicing appears to be insensitive to syntactic islands (Merchant 2001, 2008b: 135), that is, it allows the movement of wh-phrases out of islands, as in (5) and (6) (Merchant 2001, 2013a; Boeckx 2006; Bîlbîie 2011; van Craenenbroeck and Merchant 2013):

(4) a. Robin speaks French, as well as Leslie (*speaks) German.
b. Robin speaks French, and not Leslie (*speaks) German. [Bîlbîie (2011: 131); originally in Culicover and Jackendoff (2005)]

(5) They want to hire someone who speaks a Balkan language but I don’t remember which (*they want to hire someone who speaks). [Merchant (2001: 5); originally in Ross (1969)]

Compare the elliptical example in (6)a with its non-elliptical counterpart:

(6) a. Ben will be mad...