Diachronic and Comparative Syntax

Ian Roberts

• Publisher: Routledge
• Year: 2018
• ISBN:
  9781315310558
• Edition: 1
• Topic: Filología
• Subtopic: Lingüística

This book brings together for the first time a series of previously published papers featuring Ian Roberts' pioneering work on diachronic and comparative syntax over the last thirty years in one comprehensive volume. Divided into two parts, the volume engages in recent key topics in empirical studies of syntactic theory, with the eight papers on diachronic syntax addressing major changes in the history of English as well as broader aspects of syntactic change, including the introduction to the formal approach to grammaticalisation, and the eight papers on comparative syntax exploring head-movement, the nature and distribution of clitics, and the nature of parametric variation and change. This comprehensive collection of the author's body of research on diachronic and comparative syntax is an essential resource for scholars and researchers in theoretical, comparative, and historical linguistics.

Part I
Diachronic Syntax

1
Agreement Parameters and the Development of English Modal Auxiliaries^*

Ian Roberts

1. Introduction

1.1 General Background

In Modern English there is a syntactically and morphologically definable subclass of verbs, the modal auxiliaries. These verbs¹ differ from other verbs (main verbs) with respect to the following criteria among others (cf. Jespersen, 1909–49; Palmer, 1974; Pullum and Wilson, 1977):

• (1)
  a.	Inversion:	Must they leave?
  		*Leave they?
  b.	Negation:	They cannot walk.
  		*They walk not.
  c.	Agreement:	*He mays, musts, wills, cans, etc.
  d.	Non-finite forms:	*He has (?)might (etc.) to do it.
  		*They are canning to do it.
  		*They might could do it.^2

This list of properties, while not exhaustive, suffices to establish the distinction between the two classes of verbs.

This distinction did not exist at an earlier stage of the language. On this matter we quote Visser (1963–73):

Originally, … they [the modals – IGR] were not function-words, but full or independent notional verbs that did not differ syntactically in any way from other full verbs. Thus they could regularly be construed with direct objects: “ic can eow” (= “I know you”), “ic sculde tyn þusend punda” (= “I had to pay ten thousand pounds”), “eall þæt he ahte” (= “all that he possessed”). Since infinitives were nouns, the relation between them and the verbs shall, can, etc., to which they were joined must originally have been the same as that between a direct object and a full verb, so that there was structurally no difference in this respect between ‘he can manigfealdan spræce’ [= “he knows many languages”—IGR] and ‘he can sprecan [= “he can speak”—IGR].’

(Visser, 1963–73, p. 548)

The sameness in syntactic distribution of these two classes in Middle English is shown in the examples in (2), where we can see that negation and various processes of V-fronting³ affected both modals and main verbs in the same fashion:⁴

• (2)
  1. Inversion:
     • (i) Al her cariage was stole be the Frenshmen, so mote they nedes go home on fote
       All their conveyance was stolen by the Frenchmen; so they had to go home on foot.
       (c. 1464 Capgrave Chronicle of England: V 1694)
     • (ii) Wilt thow ony thinge with hym?
       Do you want him for anything?
       (1470–85 Malory Morte d’Arthure III, iii, 102, V 559)
     • (iii) Than longen folk to goon on pilgrimages Then people want to go on pilgrimages.
       (c. 1386 Chaucer General Prologue Canterbury Tales, 12)
  2. Negation:
     • (i) ʒif ʒe wollnot to haue mercy of God If you don’t want God’s mercy.
       (c. 1450 Mirk’s Festial 285: V 1177)
     • (ii) Thy godfadirs wyff thow shalt not take You shall not take your godfather’s wife.
       (c. 1450 Idley Instructions 2 a. 1757: V 1489)
     • (iii) A blynde man kan nat juggen wel in hewis A blind man cannot judge colours well.
       (c. 1387 Chaucer Troilus 2, 21: V 1624)
     • (iv) He ne held it noght He did not hold it.
       (Mossé, 1952, p. 112)
     • (v) My wyfe rose nott
       My wife did not get up.
       (Mossé, 1952, p. 112)

Example (3) shows that Middle English (ME) (and some Early Modern English (ENE)) modals had non-finite forms:

• (3)
  • (i) I shall not konne answere
    I will not be able to answer.
    (c. 1386 Chaucer Canterbury Tales B 2902: V 1649)
  • (ii)Cunnyng no recour in so streit a neede
    Knowing no recourse in so desperate a need.
    (c. 1439 Lydgate Fall of Princes 7, 1346: V 1650)
  • (iii) if we had mought conuenient come together
    If we had been able to meet conveniently.
    (c. 1528 St. Thomas More Works 107, 86: V 1687)
  • (iv) if he had wolde
    If he had wanted to.
    (1525 Ld. Berners, Froiss. II, 402: V 1687)

We will see later that the ME modals were in fact distinct with respect to their agreement properties. This distinction will be crucial in what follows.

The change that involved the development of the subclass of modals and its consequences will be the object of study in this article. The paper is organized as follows: the remainder of this introduction will be devoted to giving theoretical background to our account. In particular, we will present those aspects of Government Binding theory that are important here: government and the theory of thematic roles. We also propose a condition on the distribution of verbs. This condition leads to the postulation of two kinds of agreement system: one syntactic and one morphological. We take these two agreement systems to represent parameters of Universal Grammar (UG), and consider the development of English auxiliaries to be an instance of a shift in the value of this parameter from a morphological system to a syntactic system. In section 2, after discussing the syntax and morphology of ME modals (section 2.2), we go on to consider the causes of the parametric shift: the loss of subjunctive inflection (section 2.3) and the general loss of inflections marking verb agreement (section 2.4). In section 2.5, we describe the parametric shift in more detail and point out some of its effects. Finally, in section 2.6 we briefly consider how root modals in ENE and present-day English fit into our account. In conclusion, we compare our account with those of Lightfoot (1974, 1979) and Steele et al. (1981), criticizing Lightfoot’s Transparency Principle and suggesting a way of viewing syntactic change in terms of a parameter-setting model of acquisition.

Thus the paper has three distinct but related goals. First, as a contribution to Government Binding theory, we propose and motivate the condition on verbs. Second, the paper is meant to contribute to our knowledge of the history of English by bringing together well-known facts in a novel way. Third, the paper exemplifies an approach to diachronic syntax adumbrated in Lightfoot (1979), where syntactic change is explicitly related to aspects of acquisition.

1.2 Theoretical Background

We assume the framework of Government Binding theory (henceforth GB theory), as in Chomsky (1981, 1982). The expansion of S is as in (4):

• (4) S ® NP INFL VP

GB theory consists of a small number of autonomous subsystems of principles. Of these subsystems, the most important in this paper is the theory of thematic relations (θ -theory). We briefly sketch the main points of θ -theory below. Before doing so, however, we introduce and define the notion of government, as this will be central in what follows.

1.2.1 Government

The definition of Government is as in (5):

• (5)α governs γ in a configuration like [_β … γ … α … γ …] where:
  • (i)α = X⁰ (a lexical element)
  • (ii)where φ is a maximal projection, if φ dominates γ, then either φ dominates α, or φ is the maximal projection of γ
  • (iii)α c-commands γ.
  (from Belletti and Rizzi, 1981, p. 12).

Definition (5) means that a head α governs a node γ if and only if α c-commands γ and α c-commands no maximal projection which dominates γ except possibly the maximal projection of γ.⁵ C-command is defined in (6):

• (6)α c-commands β iff the minimal maximal projection dominating α also dominates β.

Definitions (5) and (6) together define an upper limit to government. It is impossible for a head (α) to govern a node (γ) which is outside the minimal maximal projection dominating α. So α cannot govern γ in (7):

• (7)
  

On the other hand, if γ is located within the minimal maximal projection (φ) dominating α, three situations are possible:

• (i)φ is the minimal maximal projection dominating γ, as in (8):
  • (8)
    
• (ii)φ is the minimal maximal projection dominating the maximal projection of γ:
  • (9)
    
• (iii)γ is more deeply embedded inside φ than in either (8) or (9):
  • (10)
    

In all of (8), (9) and (10), α c-commands γ. In (8) and in (9) α also governs γ. In (10), however , γ is too deeply embedded within φ to be governed by α. We can see from this discussion that government is a more restrictive version of c-command.

More concretely, if in (7–10) α = V, φ = VP and γ = NP or N⁰, we have the following configurations:

• (7′)
  
• (8′)
  
• (9′)
  
• (10′)
  

In (7′), NP is the subject of V. In (8′, 9′), NP is the object of V. In (10′), NP is not a complement of V. So we see that government is intimately bound up with complementation. In fact, Chomsky (1981, p. 51) proposes that only positions governed by α can be subcategorized by α.

Now consider certain proposals in the recent literature on morphology (cf. Lieber, 1980; Selkirk, 1982; Williams, 1981):

• (11) In a word of the form [_w Stem +Af]:
  1. Af subcategorizes for the stem;
  2. Af heads W.

The Affix -al, for example, makes Nouns into Adjectives. This is captured by saying that -al subcategorizes for a Noun. Thus -al cannot attach to Stems which are not Ns:

• (12)*out_P + al_A
  *carry_V + al_A
  *fortunate_A + al_A.

In addition, (11b) says that when -al attaches to a stem, the resulting word will be of category A. This is illustrated in (13):

• (13)[transformation + al]_A
  [industry + al]_A
  [nation + al]_A.

So the Affix-Stem relation is a kind of head-complement relation.

Then we might suggest that in (11), Af governs Stem; (11) would then be a case of (8), where α = Af, and γ = Stem. The proposal that Affixes govern Stems is immediately supported by (i) the fact that Affixes do not govern outside the Word—this is the analog of (7), where φ = Word; and (ii) the fact that Affixes cannot ‘see into’ the structure o...

Table of contents

• Cover
• Title
• Copyright
• Contents
• Preface
• Acknowledgments
• List of Contributors
• PART I Diachronic Syntax
• PART II Comparative Syntax
• Index