英语临界助动词构式变化之语言学研究

本文是一篇语言学论文，本研究采用基于语料库的定量研究手段，首先主要借助大型美式英语历时语料库 COHA，对所需语料进行提取及语义标注；然后借助 R 统计软件，对各类子构式的历时演变规律进行可视化呈现，并对结果进行语言学上的解读。

Chapter 1 Introduction

1.1 Research Background
In the Aristotelian tradition, language is viewed as the expression of thought by means of speech-sounds. This leads to a joint study on the two aspects of language: a logical/psychological on the one hand, and formal on the other. However, the main objective of the former is to disclose certain properties in human beings, and the actual use of language may not necessarily be controlled by logic, which means the link between form and meaning is usually arbitrary. Thus, in order to study the properties of linguistic expression proper, such an approach was soon abandoned and superseded by a historical account, culminating with the neogrammarians who seek to establish rigorous laws to sound change independent of logical or psychological factors. Later, people realized that this philological tradition alone cannot explain the native speakers’ everyday use of language, and thus a conceptual view on meaning within the whole language system gradually takes shape, innovated by F. Saussure’s ‘sign theory’ to discuss the association between ‘signifier’ (i.e.: sound image) and ‘signified’ (i.e.: concept).
In the post-Saussurean times, there is a burgeoning of schools of linguistics, and almost all of them involve the distinction between ‘form’ and ‘meaning’ despite their terminological differences. On the formal strand, scientific studies of speech sounds are established, forming branches of phonetics and phonology. On the meaning strand, semantic and pragmatic studies on different layers of meanings or different functions of language are developed, be it language-internal or language-external1. However, for one particular school of linguistics, language is usually viewed more or less one-sidedly, hence a general distinction between ‘formalism’ and ‘functionalism’ since1950s.
............................

1.2 Aims and Scope of the Study
This dissertation deals with the constructional changes of three marginal auxiliary construction subschemas which are sporadically placed on Quirk et al.’s (1985) continuum between ‘central modals’ and ‘main verbs’. Since many of its members carry modal meanings, it has a natural connection with modality, which is claimed to be ‘the most important and difficult area in English grammar’ (Palmer 1979).
The notion of modality is complex at least in twofold: on the one hand, it is closely related to tense and aspect, all of which together form the acronym “TAM” in generative grammar; on the other hand, it also takes over the gradual loss of inflectional ‘mood’ since Middle English. Thus, all the notional modal categories that are traditionally expressed by the subjunctive mood such as possibility, probability, obligation are gradually realized by modal syntactic forms, especially modal auxiliaries. So far, various attempts have been made on both diachronic and synchronic descriptions of the syntactically unified category of central modals. However, as to the relatively inconsistent and somewhat idiosyncratic category of marginal auxiliaries, it still awaits a further in-depth analysis. That is the aim of this present research: to reinterpret the time-honored topic of modality from a corpus-based quantitative approach, starting from this peripheral category of ‘marginal auxiliaries’.
The scope of this research ranges from marginal modals through semi-auxiliaries to catenatives at different schematic levels: on the one hand, the overall trend of development for each construction subschema is disclosed; on the other hand, a finer-grained analysis for each schema is conducted with three case studies: ‘need’ vs. ‘need to’, ‘be supposed to’, and ‘seem to’ vs. ‘appear to’. In the end, a tentative hierarchical network of modal construct-i-con is established, which aims to display the node-external links inside this constructional network, as well as the multiple source inheritance at different schematic levels from other external construction schemas.
...........................

Chapter 2 Literature Review

2.1 Introduction to Construction Grammar
This section starts with an introduction of some basic concepts in Construction Grammar and some shared tenets among usage-based construction grammarians. After that, a comparison is made between Construction Grammar and other major branches of linguistics to show its special features and advantages in the study of modality. In the end, three different models of constructional network are summarized to pave the way for the following discussions on the constructional changes of English Marginal Auxiliary Constructions.
2.1.1 Constructions and Construction Grammar
The term ‘construction’ can be dated back to Cicero, a Roman orator who in the 1st Century BCE first applied the Latin word constructio (where the English word ‘construction’ derives) to refer to a grouping of words (i.e.: a phrase or constituent). In Latin grammar, constructio was used as a translation version of the Greek grammatical term syntaksis (‘syntax’). Among those highly inflected languages, changes do not occur independently, but rather in a given context, namely a ‘construction’. Thus, in historical linguistics, construction is associated mainly or even exclusively with syntax (Traugott & Trousdale 2013:31). This has been attested by Priscian (c. 500 CE) who began using the word constructio as a grammatical term half a century later.
In the 13th century, the medieval linguists (known as the Modistae6) spent much of their time discussing the nature of the construction itself, defining it as “an ordering of words that agree and express a complete meaning” (Goldberg & Casenhiser, 2006). In this criterion, a construction should consist of at least two words, and one of them ‘governs’ or ‘requires’ the other. In other words, the Modistae argue that a constructionmust be grammatically well-formed to express a meaningful sentiment, with due attention to both formal (i.e.: syntactic) and functional/semantic aspects. In the middle of the 19th century, ‘construction’ was used to refer to those formulaic, fixed sequences.
...........................

2.2 A Construction Grammar Approach to Language Change
Since its foundation, Construction Grammar has been mainly applied to the synchronic study of language, from peripheral to central linguistic phenomena, with its main objective to find out what speakers know when they know a language and to describe such knowledge as accurately as possible (Goldberg 2003:219). However, in order to fully explain those highly unpredictable and idiosyncratic constructions such as modal auxiliary constructions, a synchronic description alone is not enough, for ‘each construction has its own history’ (Hilpert 2013; Traugott 2014). Therefore, in recent years, construction grammar has witnessed a diachronic turn, namely to study the constructional changes at all levels of linguistic structure, from allomorphy through syntax to discoursal units, as a reflection of language change.
The basic assumption is that different constructions will change in mutually related ways: one construction may encroach the territory of another, or fall out of use, leaving a functional gap that is subsequently filled by other constructions. Likewise, a new construction from another language may be borrowed to fill in the sparsely populated grammatical space between existing constructions (Hilpert 2013). An attendant assumption is that constructions compete for territory, thus striving to be evenly dispersed across grammatical space. However, as Hilpert (2013:4) himself argues, there is another possibility that one construction’s development of greater usage patterns may not necessarily result in the demise of another. Thus, grammatical change is not a ‘zero-sum game’, and it could be that the paradigm of constructions is expanding as a whole. This is the starting point of this current study: to explore the constructional changes of different subschemas within the network of marginal auxiliary constructions, as well as the interrelations of different micro-constructions within each subschema to judge whether those micro-constructions are developing independently, or in lock-step under a construction subschema.
..........................
Chapter 3 Methodological Foundations ........................ 93
3.1 A Corpus-based Quantitative Research.................... 93
3.1.1 Corpus linguistics and diachronic corpora used in this study ·················· 93
3.1.2 Corpus-based approach vs. corpus-driven approach ················· 99
Chapter 4 Marginal Modal Constructions (MM.Cxn) ...................... 117
4.1 DARE, NEED and OUGHT (TO) as Prototypical Members ....................... 117
4.1.1 Previous studies on infinitive complementation ························ 117
4.1.2 Historical development and present-day use of dare (to), need (to), and ought (to) ··············· 120
Chapter 5 Semi-auxiliary Constructions (SA.Cxn) ........................... 165
5.1 Semi-auxiliaries in a Broader Context .......... 165
5.2 A Prototype-based Fuzzy Set Model for the Development of Semi-auxiliary Constructions ................ 172

Chapter 6 Catenative Constructions (CA.Cxn)

6.1 Previous Research on Catenatives
In comparison with marginal modals and semi-auxiliaries, the category of catenative verbs is only recognized until quite recently and is relatively under-researched ─ it merely takes up two pages in Quirk et al.’s (1985) monumental work, with even no formal definition but only a brief description instead that the term ‘catenative’ alludes to the ability of these verbs to be concatenated in sequences of nonfinite constructions, such as “Our team seems to manage to keep on getting beaten”. In fact, the propensity to form chain-like structures is not only confined to catenative verbs, but is also typical of semi-auxiliaries and main verbs followed by nonfinite clauses as objects. The typical catenative constructions listed in Quirk et al. (1985) include appear to, come to, fail to, get to, happen to, manage to, seem to, tend to and turn out to. Such constructions have meanings relevant to aspect and modality, but are syntactically nearer to main verb constructions (e.g.: expect (to), attempt (to)) than semi-auxiliaries are, and the major difference is that such catenative verbs are in no way related to transitive verb constructions (i.e.: verb + direct object/prepositional object), as is shown in the two examples below.

..........................

Chapter 7 Conclusion

7.1 Major Findings of the Present Research
This dissertation mainly deals with three subschemas of English marginal auxiliary construction, namely: Marginal Modal Construction, Semi-auxiliary Construction and Catenative Construction, each with a case study to supplement the overall trend of development at schematic level. The major findings are summarized as follows, in line with the four research questions raised in Chapter 1.3:
(1) Marginal Modal Constructions:
For marginal modal constructions, as is visualized in the linguistic motion chart, it is shown that this whole construction schema is moving towards the to-infinitive end as an indication of deauxiliarization, except a slightly opposite trend in dare (to) which may be due to some frequently used frozen idioms such as ‘I dare say’. In terms of want (to), or the so-called ‘emerging modal’ in literature, at the beginning of the 19th century, it displayed a strong overlap with ought (to) in its frequency distribution, but afterwards they were gradually separated due to the latter’s sharp decrease in frequency. Instead, the role of ought (to) was taken over by need (to). This might be a tentative explanation for the semantic change of want to expressing weak deontic necessity of advice, especially with a second person singular subject. From a formal perspective, those emerging “want + bare-inf.” examples may be due to the effect of constructional contamination by paradigmatic analogy with another micro-construction “help (to) V”, which also allows an intervening NP and a to- vs. bare-infinitive alternation. However, since most of ‘want + bare-inf.’ examples occur in English-based Pidgins/Creoles in fiction genre, whether this emerging usage will encroach onto the territory of mainstream AmE still remains to be further observed.
reference(omitted)