Professional Documents
Culture Documents
General Editors
Werner Abraham Elly van Gelderen
University of California at Berkeley / Arizona State University
University of Vienna / Rijksuniversiteit
Groningen
Volume 105
Adverb Licensing and Clause Structure in English
Dagmar Haumann
Adverb Licensing
and Clause Structure in English
Dagmar Haumann
Universität Mannheim
Haumann, Dagmar.
Adverb licensing and clause structure in English / Dagmar Haumann.
p. cm. -- (Linguistik aktuell = Linguistics today, ISSN 0166-0829 ; v. 105)
Includes bibliographical references and index.
1. English language--Adverb. 2. English language--Clauses. I. Title.
PE1325.H36 2007
425--dc22 2007004708
ISBN 978-90-272-3369-1 (alk. paper)
References 414
Index 432
CHAPTER 1
INTRODUCTION AND OVERVIEW
This book investigates the factors that govern the structural integration and the
licensing of adverbs in English. The scope of investigation is restricted to the
clause (and thus to adverbs that function as modifiers within clause structure),
with emphasis being placed on the structural implementation of the relation
between the position and interpretation of adverbs.
Starting out with a survey of the problems that pertain to adverbs as an
empirical domain of investigation, this chapter provides a brief overview of
central questions concerning the syntax and semantics of adverbs.
modify or accompany the verb, they may be simple or complex and they fall
into a number of semantically defined classes, e.g. adverbs of place, time,
manner, quantity, etc. Conjunctions, finally, are defined as structuring devices
that express conjunctive, disjunctive, conditional, causal, etc. relations (cf.
Arens 1969:23ff.). Hence, the applicability of the criteria in question varies
across the word classes identified. Thus, criteria pertaining to the relative posi-
tion or function of a lexical item only figure in the definitions of prepositions
and adverbs. Morphological criteria do not come into play with respect to the
definition of prepositions and conjunctions; with respect to adverbs, only the
lack of number, gender and case is relevant, i.e. adverbs are indeclinable.
Semantic criteria figure in the definition of nouns, verbs, adverbs and conjunc-
tions, but not in the definition of participles, articles, pronouns and preposi-
tions. Abstracting away from minor modifications, such as the removal of par-
ticiples from the word class system and the recognition of adjectives as a word
class (see for example Priestly 1761) or the severing of demonstratives from
articles and the introduction of interjections, the inventory of word classes
originally postulated for Classical Greek "persists in, and in fact dominates, the
literature on grammar" (cf. Emonds 1987b:4).
The characterizations of adverbs employed in traditional grammar and stan-
dard reference grammars of English do not differ from their original charac-
terization some 2000 years ago, except for the detail that the modifying func-
tion of adverbs has been generalized to other domains, i.e. adverbs not only
modify verbal predicates (and by extension sentences), but also adjectives,
adverbs, prepositions and nominal expressions (cf. (3e) – (3h)).3 Moreover,
they may function as complements (4), i.e. as selected elements.4
Morphologically, adverbs fall into two broad classes: lexical adverbs (1a)
and derived adverbs (1b).5 The former are often, but not always, formally
identical with adjectives (e.g. late, hard), nouns (e.g. yesterday, tonight) or
prepositions (e.g. before, since), whereas the latter are derived from adjectives
by –ly-suffixation:6
(1) a. as, before, enough, even, fast, hard, here, home, late, little, now, not,
often, only, quite, right, since, soon, still, tomorrow, tonight, too, very,
well, yesterday
b. arbitrarily, agitatedly, completely, deftly, deliberately, frankly,
mentally, quickly, randomly, remorsefully, slightly, surreptitiously,
utterly, vociferously, wearily
Even though the –ly suffix is the prime morphological trait of adverbs, it does
not always signal 'adverbiality' since (a) both adjectives and adverbs may take
the same form (e.g. likely, kindly) and (b) elements ending in –ly are not always
adverbs (e.g. lovely, friendly).7, 8
INTRODUCTION AND OVERVIEW 3
Since the morphologically defined classes of lexical and derived adverbs cut
across both the semantically and distributionally defined classes of adverbs (cf.
(2) & (3)), it is hard to see how morphological properties could define the
domain of investigation (see also Delfitto 2000:14 for discussion).
Adverbs fall into a number of semantically defined classes, with manner
adverbs (2a), degree adverbs (2b), temporal adverbs (2c) and adverbs of place
(2d) constituting the core classes identified by all traditional grammarians: 9
(4) a. [She] was more likely to behave *(unpleasantly) towards her child.
b. She dressed *(elegantly) and had social graces which [he] didn't have.
c. Various members of his family have resided *(there) from time to
time. (BNC-data)
This brief survey serves to reveal that there is no single property shared by all
elements classified as adverbs that would allow for a unitary characterization
of this word class. The traditional routine of defining word classes on the basis
of a conglomeration of criteria has been challenged with the rise of structural-
ism (but see Jespersen 1924/1992:96ff. for an even earlier critique). Thus, for
4 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(8) a. Only forty miles ahead lay the beginnings of the great Pennine range.
b. I want to see that immediate family right afterwards.
(9) a. Only forty miles ahead of the previous sighting lay the beginnings of
the great Pennine range, the backbone of England.
b. I want to see that immediate family right after dinner. (BNC-data)
Underlying this proposal is the observation that the italicized expressions have
a dual status: they are headed by a nominal element, but they have adverbial,
i.e. PP or S distribution. To account for this duality, Larson (1985, 1987) pro-
poses assuming that expressions of this type are lexically marked by a special
feature, [+F], which percolates onto NP thereby signaling that the constituent
bears Oblique Case (cf. Larson 1985:606ff.), whereas Grimshaw & Bresnan
(1978:347) propose to embed the expressions under consideration within
empty-headed PPs.16
The elements in (15), which are a proper subset of the elements traditional
grammar classifies as "adverbs of quantity, degree and measure" (cf. Sweet
1891:121f.), have come to be recategorized as Deg(ree) elements and/or
Q(uantifiers):17
(2004) argue that as, so and too on the one hand and enough, less and more on
the other fall into two distinct categories, Deg and Q(uantifier), respectively.
Evidence for the latter distinction derives inter alia from the distributional
differences displayed by as, so and too vs. enough, less and more:
In the wake of Jackendoff's (1972) original proposal, the fact that not all
adverbs are equally admissible in the positions defined by the phrase structure
rules in (18) – (20) has often been taken to be a corollary of mapping from
semantics to syntax: adverbs are lexically specified with respect to semantic
8 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
structures which specify which interpretive rules may apply to them (cf.
Jackendoff 1972:69ff.). The interpretive rules, in turn, are specified with
respect to certain syntactic domains, the domain of predicate operators, VP (cf.
(18) & (19)), and the domain of sentence operators, S (20).21 Thus, for exam-
ple, carefully in (21), due to its lexical specification, is susceptible to being
addressed by the projection rules for manner adverbs and subject-oriented
adverbs and thus may occur in the syntactic domains in which the projection
rules are active, VP and S. Conversely, luckily in (22) can only be accessed by
the projection rule for speaker-oriented adverbs, whence its interpretation is
constant across positions:
The phrase structure rules (18) – (20) are indirectly (i.e. by the projection rules
addressing them) restricted to operate over semantically defined subclasses,
which is probably not surprising since elements that assume identical structural
positions receive identical interpretations.22 In light of these considerations, we
will refer to the various distributional classes of adverbs by semantic subclass
labels, e.g. speech act adverbs (e.g. frankly, honestly), evaluative adverbs (e.g.
unfortunately, luckily), modal adverbs (e.g. probably, maybe), subject-oriented
adverbs (e.g. cleverly, deftly), frequency adverbs (e.g. rarely, often), durational
adverbs (long, briefly), temporal adverbs (e.g. soon, currently) and manner
adverbs (e.g. vociferously, well).23 What is relevant is that the resultant picture
is a compartmentalized one: the traditional word class Adv does not come
together as one discrete category, but is scattered across a large set of distribu-
tionally and semantically homogeneous subclasses, i.e. "discrete categories
whose members have equal status as far as grammatical processes are
concerned" (cf. Newmeyer 2000:221).
Jackendoff's (1972) insight that there are strict correlations between the
positions adverbs may assume and the interpretations they receive in these
positions has proven equally incentive to recent syntactic and semantic analy-
ses of adverbs which, as we shall presently see, are diametrically opposed in
the sense that either the syntax of adverbs is taken to determine their semantics
or vice versa.
INTRODUCTION AND OVERVIEW 9
(23) XP (24) XP
2 2
Adv XP Adv X'
2 2
XP Adv X YP
2 2
Adv X(P) Adv Y'
2 2
X Adv Y …
The derivation of ordering restrictions and the relative order of adverbs (27) as
well as questions relating to the relative freedom of adverb placement (28) as
opposed to the placement of constituents with adverbial function (29) consti-
tute further canonical topics in the syntax of adverbs:25
(29) a. Five times a day he (*five times a day) has (*five times a day) been
(*five times a day) calling her (five times a day).
10 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
The syntax and the semantics of adverbs pose a chicken and egg problem at the
syntax-semantics interface insofar as it is not clear whether the syntax of
adverbs determines their semantics or vice versa. Thus, the following questions
are central: Does the distribution and licensing of adverbs follow from syntac-
tic or semantic principles? Are ordering restrictions among adverbs condi-
tioned by syntax or semantics? Which correspondences are there between syn-
tactic and semantic types of adverbs and between the syntactic and the seman-
tic domains accessible to certain types of adverbs?28
In the last few years, the two views on the interplay between syntax and
semantics have come to be associated with two distinct types of analysis: the
position that syntax determines semantics has come to be inseparable from the
functional specifier analysis (cf. Cinque 1999, 2004; Alexiadou 1997;
Laenzlinger 1996, 1998), while the position that semantics determines syntax
has come to be intimately tied to rather sophisticated adjunction analyses (cf.
Haider 2000, 2004; Frey & Pittner 1998, 1999; Ernst 2000, 2002, 2004a, b).29
INTRODUCTION AND OVERVIEW 11
This book is organized as follows. Chapter 2 reviews the major proposals con-
cerning the structural integration and licensing of adverbs and thus introduces
to the ongoing debate between proponents of the functional specifier analysis
and proponents of adjunction-based semantic scope theories. The predictions
made both by proponents of the functional specifier analysis and semantic
scope theorists will be tested and evaluated on the basis of empirical findings,
with clear preference being given to the functional specifier analysis.
Chapters 3 to 5 are devoted to the detailed analysis of the syntax of an
extensive range of syntactic categories of adverbs both in their relation to
clause structure and their (co)shaping of clause structure. Throughout these
12 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Notes
1
See for example Ernst (1984:15) who takes "our lack of knowledge about adverbs" to be
enough of a reason for "refrain[ing] from proposing even a working definition".
2
'Kind' relates to the distinction between lexical vs. derived nouns and 'form' to the distinction
simplex vs. compound nouns.
3
Cf. Mätzner (31880:97), Jespersen (1924/1992:96ff.), Curme (1935:73ff), Pence & Emery
(1947:6, 98, 342), Quirk et al. (91991:127ff., 449f.). Jespersen (61946) explicitly warns us
INTRODUCTION AND OVERVIEW 13
"against the etymological fallacy of taking [the Latin names of the word classes] at their face
value: [...] an adverb (Latin: adverbium) need not belong to a verb (cf. nearly ready, nearly
everybody) [...]" (cf. Jespersen 61946:69 [DH]).
4
Cf. Steinitz (1969:14f.), Jackendoff (1972:65), McConnell-Ginet (1982:164), Larson (1985:
596ff.), Alexiadou (1997:6, 109ff.), Goldberg & Ackerman (2001).
5
A third morphological type of adverbs is constituted by so-called compound adverbs (e.g.
someplace, everywhere). Cf. Sweet (1891:119), Mätzner (31880:427), Francis (1958:282ff.).
6
The taxonomy used here differs from that employed by e.g. Sweet (1891:118), Jespersen
(1924/1992:87) and Poutsma (1926:607) in that the class of derivational adverbs only contains
adverbs in –ly, not those which may be related to e.g. adjectives and nouns by conversion. Note
that, in addition to deadjectival adverbs in –ly, we find adverbs derived from adjectives and
nouns by a- prefixation (e.g. aloud, abroad) as well as adverbs derived by suffixation with e.g.
–wise, –ward(s), –ways, as in clockwise, easthward(s), sideways). Cf. Zandvoort (1957:329ff.),
Francis (1958:284) and Quirk et al. (91991:438) for dicussion.
7
Cf. Sweet (1891:188f.), Jespersen (61946:76), Pence & Emery (1947:339f.), Quirk &
Greenbaum (161986:125).
8
Conversely, deadjectival adverbs need not always bear the –ly suffix (cf. Sugioka & Lehr
(1983:295), Radford (1988:138), Zwicky (1989, 1995), Pittner (2001)):
(i) But now for the first time I became aware that each state had also its individual
prose style, made sharply evident in its highway signs. [...] Nearly all have
abandoned the adverb for the adjective. Drive Slow. Drive Safe. (John
Steinbeck Travels with Charley, p. 79f.)
Interestingly, though, we can observe that –ly, as the only formal marker distinguishing ad-
verbs for adjectives, quantifiers, etc., has been "increasingly spreading" since the Middle Eng-
lish period, even to the extent that adverbial adjectives in –ly are formed in the absence of an
adjectival adjective counterpart (cf. Killie 1998:123ff.; see also Swan 1988a:90ff.; 1988b;
Fischer 2003:9; Pittner (2001:4):
(i) Alice purposely did not look out of the window. (BNC)
(ii) [They are] forming a remorsely inhuman environment. (BNC)
Historically, according to Killie (1998), the spreading of –ly (at least with manner and degree
adverbial adjectives) is "the result of a morphological regularization process taking place
within the adverb categories themselves" (Killie 1998:130, note 17) that increases homogene-
ity. Note that Pittner's (2001) conjecture that –ly-suffixation "becomes more and more indiffer-
ent to the category of the source and is more and more applied to all sorts of categories in ad-
verbial function which are not PPs or sentential phrases" (cf. Pittner 2001:4) is in line with
Killie's (1998) argument. See also Baayen & Renouf (1996:82f.) for discussion.
The so-called "–ly tide" (cf. Killie 1998:130, note 17) also affects adverbial adjectives, e.g.
seldom, thus in (iii) & (iv), as well as quantificational elements in verbal and adjectival con-
texts, e.g. enoughly, lessly in (v) – (vii) :
(iii) [The] ropes were used seldomly to stop them crowding round the players (BNC)
(iv) And thusly did he pump up the volume. (BNC)
(v) [He] got a severe cold and therefore did not move enoughly... (www)
(vi) [She] sacrificed herself lessly so that others may be spared the horrors… (www)
(vii) [A] small, mousy-haired woman, chimed in, a littly archly. (BNC)
9
The number and nature of the identified semantic (sub)classes of adverbs is not consistent but
varies from author to author (sometimes only slightly, sometimes considerably).
10
Thus, lexical items are classified according to the positions they may assume within a given
structure and only lexical items that occupy the "same set of positions" and may be substituted
for one another without inducing "a change in the structural meaning" are members of the
14 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
same word class (cf. Fries 1952:74, 104). See also Bloomfield (1933:184ff.), Harris (51961:5,
15f.; 1964:33), Fries (1952:73ff. 188f.), Gleason (1955:129ff.), Francis (1958:236ff.).
11
Cf. Radford (1988:138), Déchaine (1993:51f.), Ramat & Ricca (1994:309), Alexiadou
(1997:198; 2002a:29) for discussion. The morphological relationship between adjectives and
adverbs is not only consistent, but also productive: if new adjectives are formed, the corre-
sponding –ly-adverbs will also be formed (cf. Radford 1988:138). Further morphology related
evidence for the categorial non-distinctness of adjectives and adverbs derives from the fact
that, in so-called –ly-less dialects, the adverb can occur without –ly (cf. note 8).
12
N.B. the synthetic formation of the comparative and superlative of derived adverbs involves
the attachment of –er and –est "to the root form of the adverb" (cf. Sugioka & Lehr 1983:295),
not to the derived adverb ((i) & (ii) from Déchaine 1993:52):
(i) She thinks quicker/* quicklier than me.
(ii) She thinks quickest/* quickiest of all.
13
See Bresnan (1973), Emonds (1976, 1985), Bowers (1975), Abney (1987), Radford (1988),
Déchaine (1993), Corver (1997) and Alexiadou (1997, 2002a) for discussion.
14
See Jespersen (1924/1992) for an early proposal along these lines. Uncovering the structural
similarities that exist between verbs and prepositions with respect to their respective comple-
ments, i.e. nominal or sentential complements or no complement at all, Jespersen
(1924/1992:88) proposes distinguishing between transitive prepositions and intransitive prepo-
sitions, thereby paving the way both for the recategorization of traditional adverbial subordi-
nating conjunctions (e.g. before, since) as transitive prepositions and for the recategorization of
traditional spatial and temporal adverbs (e.g. before, on) as intransitive prepositions. See also
Hendrick (1976), Jackendoff (1973, 1977), Emonds (1976), van Riemsdijk (1978).
15
It has often been observed that the modifier right may co-occur with (in)transitive temporal
and spatial prepositions, but not with adjectives or "true" adverbs:
(i) * They came over right often. (Emonds 1976:174)
(ii) * John drove right carefully. (Emonds 1976:174)
See also Francis (1958:287) for the qualification that adverbs that are formally identical with
prepositions allow for modification by right.
16
See also Emonds (1985:61, 227ff.; 1987a:617ff.), McCawley (1988b:586ff.) and Alexiadou
(1997:191).
17
Also included in this category are most, how and that.
18
Adverbs derived from adjectives are considered positional variants of adjectives "in a verb-
modifying rather than noun-modifying function" (cf. Emonds 1976:12; see also Emonds 1970;
Bresnan 1973; Bowers 1975; Larson 1987; Abney 1987; Radford 1988; Déchaine 1993;
Corver 1997; Alexiadou 1997, 2002a), the process under consideration lacks explanatory force
as regards the distributional diversity displayed by the elements under consideration and thus
merely shifts the descriptive burden.
19
The phrase structure rule in (18a) also introduces adverbs that participate in the strict
subcategorization of verbs, e.g. strangely:
(i) He acted {strangely, *only, *merely} towards me in Stuttgart. (BNC)
Adverbs of this type are part of the functional structure of a semantic reading which is deter-
mined at deep structure (cf. Jackendoff 1972:14f.). Note that the VP-final position is also
accessible to prepositional phrases.
20
According to Jackendoff (1972:106), sentence adverbs (alongside prepositional phrases, PP,
and parentheticals, S) are introduced in sentence-final position by (i).
(i) S → NP – Aux – VP – {(Adv), (PP), (S)}
INTRODUCTION AND OVERVIEW 15
2.1 Adverb Licensing in the Theory of Government and Binding and the
Theory of Principles and Parameters
Since within the Theory of Government and Binding and the Theory of
Principles and Parameters, henceforth GB/PPT,1 the specific rules of earlier
models of generative grammar have been supplanted by generalized principles,
such as the Extended Projection Principle (EPP) in (1) and Move-α in (2),
well-formedness is no longer statable in terms of specific phrase structure rules
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 17
and transformational rules, but seen as the result of the (interactive) evaluation
by the modules of grammar: Θ-Theory, Case Theory, Binding Theory, Control
Theory, Government Theory and Bounding Theory (cf. Chomsky 1981:5):
(4) LEXICON
d-structure Θ-Theory
move-α
Case Theory, Binding Theory,
EPP s-structure Control Theory, Government
move-α Theory, Bounding Theory
FI
Unsurprisingly, it has soon been felt that the overall system is lopsided since
the principles of grammar defined in GB/PPT are designed to address (and
ultimately license) only predicates and arguments (i.e. elements that assign or
bear a Θ-role), nominal constituents and traces (NP and wh). Thus, in (5a), to
John, he and what are licensed by being assigned a Θ-role by give. John, he
and what are licensed by virtue of bearing Case, John and he are licensed by
Principle C and Principle B, respectively, and what is licensed by the ECP and
the Subjacency Condition. (5b) constitutes a violation of the Θ-Criterion since
either the chain whati … ti or the book fails to receive a Θ-role, (5c) violates the
Case Filter since the nominal constituent John lacks Case, (5d) constitutes a
violation of Principle A since the anaphor himself is not bound in its local
domain, and, finally, (5e) constitutes a violation of the Subjacency Condition,
since what is raised out of a wh-island:
The principles of grammar are blind with respect to adverbs, since adverbs, in
general, are assumed not to bear a Θ-role and/or Case or have referential
properties in the same sense as nominal expressions (arguably, so-called bare-
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 19
NP adverbs, e.g. yesterday, that day, constitute an exception, cf. chapter 4.3.4).
Thus, neither the well-formedness of the strings in (6) and (7c) nor the ill-
formedness of the strings in (7a, b) follows from any of the principles of
grammar:
The ill-formedness of the strings in (8) follows from a violation of the Case
Theory-related Adjacency Condition on Case Assignment which requires that
no element intervene between a Case-assigner and a Case-assignee (cf. Stowell
1981:110ff.; Chomsky 1981:49ff.):
Ironically, chains of the type how carefullyi … ti and how horriblyi … ti in (9),
are subject to (at least some of) the principles of grammar, notably the ECP
and the Subjacency Condition:
Thus, the main obstacles we face with respect to adverb syntax within the
frameworks under consideration is the lack of designated (base) positions for
adverbs (especially with d-structure being conceived of as a pure representation
of Θ-structure)2 and the lack of specific licensing devices which would regulate
and sanction their distribution.
With phrase structure cut back to three phrase internal positions (i.e. head,
complement and specifier) and with adverbs neither partaking in thematic rela-
tions nor in checking relations, whence they are barred from assuming the
specifier and complement position within lexical or functional projections,
respectively, X-bar Theory as formulated in (10) does not provide for a suffi-
cient amount of structural positions to accommodate adverbs:3
20 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(11) [β … α … [β … ]]
(12) VP
2
Adv VP
2
Adv VP
2
VP Adv
2
V'
2
Adv V'
2
Adv V'
2
V' Adv
2
V
2
Adv V
2
V Adv
2.1.3). These three proposals are not only outstanding representatives of the
syntactic analysis of adverbs within early representation based models of
grammars, but they have also shaped our understanding of licensing as being
reliant on the presence of semantic and/or morphosyntactic features associated
with both the licensee and the licenser. A completely different approach to ad-
verb licensing is advocated by McConnell-Ginet (1982) and Larson (1985,
1988, 1990). According to their proposals, (at least) manner adverbs are
licensed as innermost complements of the verb (section 2.1.4).
2.1.1 Modification
Sportiche (1988) argues that the distribution of adverbs is governed by the
Adjunct Projection Principle in (13), which is taken to be the non-thematic
analogue of the Projection Principle and to establish "a direct connection
between the lexical meaning of modifiers and the syntactic configuration in
which they appear" (cf. Sportiche 1988:429):
Sportiche (1988:430) makes the point that the Adjunct Projection Principle is a
necessary but not a sufficient condition on adverb licensing because it merely
catalogues admissible adverb positions which may, in fact, never be realized. A
case in point is the position between a verb and its internal nominal argument.
Although this position, in principle, could be assumed by an adverb (14a), the
adverb is barred from occurring in this position if the internal argument of the
verb is nominal (14b), since Case-assigners and Case-assignees must be adja-
cent (cf. Stowell 1981:110ff.; Chomsky 1981:49ff.):4
V and its sisters (provided the Adjacency Condition on Case Assignment is res-
pected), as shown in (19a) and (19b), respectively ((18) & (19) adapted from
Zubizarreta 1982:38, 40):
(18) In the configurations: [γ ... α ... β], [γ ... β ... α], where
(i) γ = a projection of β
(ii) γ immediately dominates α and β
(iii) α = Adv
Adjunct Θ-roles further differ from "true" Θ-roles in that they are not assigned
at d-structure but only at (s-structure or) LF. Evidence for this assumption
derives from the orientation of subject-oriented adverbs and manner adverbs in
actives and passives. While subject-oriented adverbs always modify the surface
subject, i.e. the agent (21a) and the patient (21b), manner adverbs (21c, d),
always modify the agent, Joe:
Zubizarreta's (1982) proposal runs into both empirical and theoretical prob-
lems. The empirical problem relates to the contrast between (25) and (26)
(adapted from Jackendoff 1972:89):
(26) a. * Max quickly was carefully climbing the walls of the garden.
b. * Max cleverly has stealthily been trying to climb the walls.
A further case in point is adverbs that quantify over events, such as frequently,
twice and often:
2.1.2 Predication
Roberts (1987:73ff.) develops a predication-based approach of adverb licens-
ing which rests on the assumption that predication is not based on thematic
properties, but on the structural requirement in (29), originally proposed by
Rothstein (1985:11):10
(30) Type I: initial, Aux, VP-final (meaning change, e.g. cleverly, clumsily)
Type II: initial, Aux, VP-final (no meaning change, e.g. quickly, slowly)
Type III: initial, Aux (e.g. evidently, probably, unbelievably)
Type IV: Aux, VP-final (e.g. completely, easily, totally)
Type I adverbs like cleverly in (31a) are two place predicates selecting as their
arguments both the agentive subject John and the predicate, i.e. Infl, whereas
Type II adverbs like quickly in (31b) are one-place predicates that select either
Infl or VP. Type III adverbs like evidently in (31c) select the entire sentence,
i.e. IP, whereas Type IV adverbs like easily (31d) neither select an agent nor an
event, but are predicated of the verb ((31) adapted from Roberts 1987:72f.):
(34) [IP [NP ] [I' [I ] [VP [Vn ] [ADV {I, II, IV}] …]]]12
I: predicated of V and agent role of V
II: predicated of VP
III: predicated of V
(35) [IP [NP ] [I' [I ] [VP [VP ] [ADV {I, II} ]]]
I: predicated of V and agent role of V
II: predicated of V
(36) [IP [NP ] [I' [ADV {I, II, III}] [I ] [VP ]]] and [IP [NP ] [I' [I ] [ADV {I, II, III}]
[VP ]]]
I: predicated of V and agent role of V
II: predicated of I
III: predicated of IP
(37) [IP [NP ] [I' [I ] [VP [ADV {I, II, IV}] [Vn ]]]]
I: predicated of V and agent role of V
II: predicated of VP
III: predicated of V
select "true" arguments as their subjects13 can be shown to follow from the
selectional properties of adverbs in conjunction with the Predication Principle.
However, the overall conception of non-thematic predication is rather dubious.
If predication is severed from thematic relations and is exclusively based on
mutual command of subject and predicate, with the predicate imposing selec-
tional restrictions on the subject, the isomorphic relation between semantic
predicate-argument structures, on the one hand, and the Θ-role assigners and
Θ-role assignees, on the other hand, is lost (see also Alexiadou 1997:36;
Laenzlinger 1998:71). Moreover, Robert's (1987) analysis allows for heads,
e.g. V and I, to be selected by adverbs of Type II and Type IV and thus to
function as their subjects. His analysis also allows for one and the same predi-
cate, namely Type I adverbs, to take two subjects: a "true" argument of the
verb and V or I.
To make her system of head feature licensing work for all the distributional
classes identified by Jackendoff (1972), a compiled version of which is given
in (39), Travis (1988:291ff.) proposes three rearrangements:
First, she argues that the contrasts between the Type III adverb probably and
the Type IV adverb completely in (40) are indicative of Jackendoff's (1972)
28 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Aux position being a conflation of two positions, the Aux position and the VP-
initial position (cf. Travis 1988:291):
The splitting of the Aux position is consequential with respect to the analysis
of Type I and Type II adverbs both of which may now assume four positions,
i.e. the initial position, the Aux position, the VP-initial and the VP-final
position (Travis 1988:291):
Leaving aside Type VI and Type V adverbs, Travis (1988:293) contends that
adverbs that occur in the initial and Aux position (Type Ia, IIa and III) are
licensed by a head feature in I,15 whereas adverbs that occur in VP-initial and
VP-final position (Type Ib, IIb and IV) are licensed by a head feature in V.
Under this analysis, the so-called transportability effects adverbs display (cf.
Keyser 1968) are accounted for by a feature percolation mechanism which
ensures that the features of a head become accessible within the categorial
projection of that head, as illustrated by the arrows in (43). Analyzing adverbs
as heads that are licensed at all nodes within a given categorial projection, as
shown in (43), provides an explanation as to why phrasal adverbials, i.e. PPs
and adverbial subordinate clauses, are not transportable within the categorial
projections of V and I: these elements are XPs and as such they can be licensed
only as predicates of VP or IP:
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 29
(43) IP
2
NP I'
Ia 2
IIa I VP
III 2 Ib
V' XP IIb
2 IV
V YP
(Travis 1988:293)
As it stands, Travis' (1988) account is not tenable since it rests on the wrong
premise that adverbs are defective heads that fail to project onto a phrasal level.
To show this, we need not recur to the longstanding question of whether
adverbs take complements or not, but simply take into consideration the fact
that adverbs may be premodified:17
(46) a. [We] have quite enthusiastically lapsed into a chronic dualism. (BNC)
b. [They] will die very quickly after being brought into captivity. (BNC)
A second syntactic flaw of her analysis lies in allowing for Adv to adjoin to IP
and VP thus to violate the Structure Preserving Constraint.
Abstracting away from the problems that adhere to each of the adverb
licensing approaches discussed here, e.g. limited range of application (Spor-
tiche 1988), implementation adjunct Θ-role assignment as a concomitant of
argument Θ-role assignment (Zubizarreta 1982), severing predication from
thematic relations (Roberts 1987), phrase structure status of adverbs (Travis
1988), the three analyses have shaped our understanding of various kinds of
affinities between individual classes of adverbs and the structures they occur in
and thus, ultimately, have contributed to paving the way to an articulate adverb
licensing theory in terms of specifier-head agreement as based on the sharing
of semantic features, such as [MANNER], [AGENTIVITY], [ILLOCUTIONARY
FORCE] etc.
(47) a. The doctor carefully has examined John. (The doctor is careful)
b. John carefully has been examined by the doctor. (J. is careful)
c. The doctor examined John carefully. (The doctor is careful)
d. John was examined carefully by the doctor. (The doc. is careful)
The syntactic differences between VP-external adverbs (cf. (47a, b) & (48a))
and VP-internal adverbs (cf. (47c, d) & (48b)) are accompanied by semantic
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 31
(50) [VP [DP John] [V' [V sawi] [VP [DP Mary] [V' [V ti] [AdvP recently]]]]]
To account for the fact that manner adverbs may assume non-right peripheral
positions within VP, as in (51), Larson (1988:347ff., 386ff.; 1989; 1990:606ff.)
proposes a Light Predicate Raising Analysis, which involves the reanalysis of
the lowest V' in a given VP-shell structure as V, thus rendering the verb and
structural complement susceptible to head raising:
The first part of the restriction in (57) is formally stated as [β α [β ...]], with
both α and β having identical phrase structure status, i.e. Xmax (cf. Chomsky
1986b:7) and thus obeying the Structure Preserving Constraint. The second
requirement, i.e. the nonargument status of the maximal projection adjoined to,
follows from adjunction inducing barrierhood for L-marking, i.e. the assign-
ment of Θ-roles and Case (cf. Chomsky 1986b:13ff.).
However, the restriction in (57) does not really rid us of the problem of
unrestrictedness. Even adjunction structures as in (58), which are sanctioned by
(57), are basically unrestricted since the interaction of the adjunct with the
overall system is not defined (viz. level of representation, licensing):
(58) XP ( = nonargument)
2
adjunct XP
2
XP adjunct
2
XP adjunct
2
adjunct XP
The question arises how the relative linear orders of adverbs, verbal heads and
other elements of the clause can be accounted for. Assuming that the licensing
of manner adverbs is contingent on the presence of the head feature [MANNER]
which is available within VP, as proposed by Travis (1988:290), they are free
to either left- or right-adjoin to VP:
(59) VP
2
AdvP VP
2
VP AdvP
5
V…
However, as can be seen in (60), manner adverbs do not occur freely in pre-
and postverbal position. To rule out preverbal occurrences of manner adverbs,
one might consider imposing a directionality restriction on adjunction, e.g.
manner adverbs must be right-adjoined to VP:
34 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Such a restriction, however, is not tenable since manner adverbs are perfectly
acceptable in preverbal position in passives:
Moreover, manner adverbs may intervene between the verb and its internal
argument provided the Adjacency Condition on Case Assignment can be met,
e.g. by extraposition of the internal argument, as is arguably the case in (62a),
but not in (62b):
Problems multiply in cases in which the verb carries two head features, say
[MANNER] and [AGENT], since nothing prevents the generation of the illicit
strings in (63) by adjunction applying recursively to the left and/or right of VP:
(63) a. * They have (hesitantly loudly) performed the aria (hesitantly loudly).
b. * They have (loudly hesitantly) performed it (loudly hesitantly).
c. * It has been (hesitantly loudly) performed (hesitantly loudly).
d. * It has been (loudly hesitantly) performed (loudly hesitantly).
e. * The aria has been loudly performed hesitantly.
f. The aria has been hesitantly performed loudly.
To rule out the illicit strings, one would have to postulate an output filter stat-
ing that (a) VP-adjoined adverbs must not be adjacent and (b) adverbs licensed
by the feature [AGENT] must outscope those licensed by the feature [MANNER].
As witnessed by the contrasts between the well-formed and the illicit strings
in (64) – (66), multiple adjunction to IP is prone to run into the same problems
as multiple adjunction to VP:
(65) a. Usually she always ordered lasagna, but not last night.
b. * Always she usually ordered lasagna, but not last night.
c. * (Always usually) she (always usually) ordered lasagna.
d. * (Usually always) she (usually always) ordered lasagna.
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 35
To rule out the illicit strings in (64) and (65), and thus the structures in (67a)
and (67b), we would have to postulate some kind of filter stating that (a)
adjuncts must flank the subject (67c) and (b) that carefully and usually must
precede slowly and always, respectively.24 Under Travis' (1988) analysis, the
head feature [AGR] would be involved in the licensing of carefully, whereas the
head feature [EVENT] would be involved in the licensing of slowly, usually,
already and no longer. We could assume that [AGR]-licensed adverbs must
outscope [EVENT]-licensed adverbs, which neatly accounts for the contrasts
between (64a) and (64b, c). However, the condition that sentence adverbs must
flank the subject is too strong with respect to already and no longer, which
may be adjacent if they follow the subject (66d) and thus correspond to
structure (67b):
(67) a. IP b. IP c. IP
th th th
AdvP IP subject IP AdvP IP
th th th
AdvP IP AdvP IP subject IP
th th th
subject I' AdvP IP AdvP IP
(69) a. [IP [D he] [I' [I would] [VP [Adv carefully] [VP open the door]]]]
b. [IP [D he] [I' [I had] [VP [Adv already] [VP refrained from smoking…]]]]
Notice that in the configuration in (69) neither the [AGR]-feature, which would
license carefully as a subject-oriented adverb, nor the [EVENT]-feature, which
would license already as an event-related adverb, are accessible. That is to say
that carefully, at best, would receive a manner interpretation (but see chapter
3.2.1.1), whereas already would go unlicensed.25
36 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(72) a. CP b. CP
2 2
AdvP CP CP AdvP
2 2
C IP C IP
2 2
subject I' subject I'
(73) a. CP b. IP
2 2
C IP subject I'
2 2
subject IP I VP
2 2
AdvP IP AdvP VP
2.1.6 Summary
Taking as a point of departure both the lack of base positions for adverbs,
which is in sharp contrast with the availability and licensibility of derived
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 37
positions, and the lack of proper licensing conditions for adverbs in GB/PPT,
this section focused the viability of adjunction as a means of compensation for
the shortage of structural positions in conjunction with specific licensing con-
ditions for adjuncts as diverse as Sportiche's (1988) and Zubizarreta's (1982)
modification approaches (section 2.1.1), Roberts' (1987) predication approach
(section 2.1.2) and Travis' (1988) head feature licensing approach (section
2.1.3).
Abstracting away from the respective shortcomings of the analyses dis-
cussed, such as the limited range of applicability (Sportiche 1988), multiple
occurrences of adverbs (Zubizarreta 1982), dissociation of predication and Θ-
structure (Roberts 1987) and the (non)phrasal status of adverbs (Travis 1988), I
concentrated on the question of how and to which extent the adverb licensing
mechanisms under consideration felicitously restrain adjunction as such, i.e.
prevent the uncontrolled proliferation of structure, as well as on the question of
how and to which extent adverb licensing mechanisms under consideration
account for the positioning of adverbs relative to each other, to verbal elements
and to the verb's internal argument(s). From the discussion in section 2.1.5 it
emerged that the various licensing mechanisms do not eradicate the attachment
and linearization problems that result from adjunction, unless the overall sys-
tem is complemented with output filters which, however, are purely declarative
and thus lack explanatory force. In brief, the lack of designated base positions
for adverbs cannot be compensated for by adjunction if adjunction is not
severely restricted either by external factors or by process specific properties.
In the following, we shall see that two interrelated lines of research in syntactic
theory, notably the conception of clause structure as a cascade of functional
projections26 and the emergence of both the Minimalist Program and the Anti-
symmetry of Syntax, have paved the way to an understanding of adjunction as a
severely constrained option with special properties (cf. Chomsky 1995: 323ff.)
and ultimately led to the development of the specifier analysis of adverbs,
which essentially builds on transparent semantic relations between adverbs and
the projections they occur in (see also Alexiadou 2002b:40 for discussion).
phonetic form (π) and a logical representation (λ). The cancellation of both d-
structure and s-structure is accompanied by the disappearance of the modules
of grammar that operated on these representational levels. As can be seen from
the schematized presentation of the Minimalist Program in (74), the sole resi-
dues of the precursor model, i.e. GB/PPT, are the lexicon, the computational
system and the aforementioned interface levels PF and LF, which are charac-
terized as bare output conditions on derivations:
(74) lexicon
numeration
select
merge
move
Σ
computational system
spell-out
ΣP ΣL
PF LF move
(adapted from Alexiadou 1997:27)
Abstracting away from adverbs and adverbial constituents that bear operator
features such as [Q] or [FOC] (75), which warrant their raising to the left periph-
ery of the clause, i.e. the specifier position of a functional projection whose
head bears a corresponding strong feature, the overall system defined in the
MP has nothing to say about the base positions of adverbs and the licensing of
adverbs in their respective base positions (but see below). As Chomsky (1995)
concedes himself, "we still have no good phrase structure theory for simple
matters as attributive adjectives, relative clauses, and adjuncts of many types"
(cf. Chomsky 1995:382, note 22):
Chomsky (1995) suggests that adverbs, e.g. probably in (76), since they are not
associated with features that would trigger movement to a specifier position, be
treated as adjuncts, not as specifiers: 32
(77) TP
3
probably TP ( = T')
2
T AuxP34
2
Aux VP
# 2
has D VP ( = V')
# 6
John left already
(78) TP
3
D T'
# 3
John probably T' ( = TP at the time of adjunction)
3
T AuxP
2 2
has T Aux VP
# 2
has D VP ( = V')
# 6
John left already
(79) XP
egi i
A'-Spec X'2 (A'-Spec)
2
A-Spec X'1
2
X0 Compl (adapted from Laenzlinger 1996:109)
The generalization that emerges is that qualifier adverbs do not interfere with
the antecedent-government of non-argument traces, whereas quantifier adverbs
do.39 He further proposes incorporating the qualifier vs. quantifier distinction
into the definition of Relativized Minimality, as originally proposed by Rizzi
(1990), by interpreting 'typical A'-Spec' in clause (i) as either qualificational or
quantificational:
(84) [CP [Q combieni] [C' [C as] [AgrSP [D tu] ... [ AspP [Adv souvent] [ Asp' … [VP
[V lu ] [QP ti de livres de Chomsky]]]]]]]
blocks antecedent government of t by combien
Although both types of adverbs occupy A'-specifier positions and are licensed
under specifier-head agreement, they are subject to different licensing condi-
tions: qualifier adverbs are subject to Checking Theory, which plays a central
role in the licensing of formal features, e.g. Φ-features), whereas quantifier ad-
verbs are subject to the Adv-Criterion in (87), which figures prominently in the
licensing of elements bearing so-called peripheral features, e.g. [WH/Q],
[FORCE], [TOPIC] and [FOCUS] (cf. Chomsky 1998:22):
The crucial difference between the two modes of licensing lies in how check-
ing applies. Laenzlinger (1996) maintains that Checking Theory can be "satis-
fied by any member of the chain in the course of derivation", i.e. by a lexical
item or its trace, whereas the Adv-Criterion must be "satisfied by the head of
the chain, by the lexical element itself" (cf. Laenzlinger 1996:114). The latter
requirement entails that, once a quantificational adverb has reached its check-
ing site and checked off its features, it is frozen in place. Given these assump-
tions, the contrasts between (85) and (86) follow automatically. Although, in
both cases, the relevant features of the adverbs have been checked off at ti,
qualifier adverbs (e.g. impoliment) are free to raise on since they are subject to
44 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
licensing under Checking Theory, which allows for traces to enter into speci-
fier-head agreement relations, whereas quantifier adverbs (e.g. probablement)
are frozen in ti since they are subject to licensing under the Adv-Criterion,
which requires that only lexical items enter into a specifier-head agreement
relation.40
A much stricter version of specifiers and adjuncts is defended in Kayne's
(1994) Antisymmetry of Syntax. Kayne (1994) dismisses the standard assump-
tion that the relation between hierarchical structure and linear order is flexible
both within a given language and across languages as too little restrictive since
it allows for hierarchical representations to be associated with more than one
linear order. Central to his analysis is the assumption that phrase structure
"always completely determines linear order and consequently that if two
phrases differ in linear order, they must also differ in hierarchical structure"
(cf. Kayne 1994:3). To derive the linear order of terminal elements from the
hierarchical relations that hold between any two dominating nonterminal
elements, Kayne (1994:4ff.) redefines dominance as a reflection of the proper-
ties of linear order, i.e. dominance relations must obey transitivity, anti-
symmetry and totality. Kayne (1994:4) proposes modifying the standard,
transitive notion of c-command, which is defined in terms of "first node up"
(cf. Kayne 1994:7), as to include antisymmetry. Asymmetric c-command is
defined as in (88):
(89) K (90) K
3 3
J L J L
# 3 # 2
j M P j M N
# # # #
m p m P
#
p
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 45
With respect to the structure in (90), the first member of the nonterminal pairs
<J, M>, <J, N>, <J, P> and <M, P> is identified as asymmetrically c-
commanding the second member of the pair. Since each nonterminal node
dominates only one terminal element, the linear ordering of terminal elements,
d(A), is that in (92), with j preceding m, m preceding p and j, by transitivity,
preceding p:41
(93) L (94) P
2 2
M P M P
# 2 g 2
Q R S Q R S
# # # # g g
q r T q r T
# g
t t (Kayne 1994:15f.)
(95) X c-commands Y iff X and Y are categories and X excludes Y and every
category that dominates X dominates Y (Kayne 1994:16)
(96) P
ei
M P
2 2
Q N R S
# # # 2
q L r T U
# # #
l t V
#
v (Kayne 1994:34)
Since <M, R> is contained in A, d(A) contains the pairs of terminals <q, r>, <l,
r>, and since <R, T> and <R, U> are in A, d(A) contains <r, t> and <r, v>.
Given precedence, all terminals of the specifier, i.e. q and l, and all terminals of
the complement, i.e. r, t and v, flank the head r in such a way that the specifier,
i.e. every adjoined element, universally precedes the head, and the head
universally precedes the complement, as schematized in (97):43
Word order variation (both within a given language and across languages) is
conceived of as the result of overt leftward movement of either X or XP or X
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 47
and XP, i.e. every word order deviating from (97), which Kayne (1994:35ff.)
interprets as SVO, is taken to be derivative. Since double specifiers as in (78),
which are possible under Chomsky's (1995) and Laenzlinger's (1996, 1998)
proposals are ruled out under Kayne's (1994) analysis, i.e. adjunction is
restricted to one instance per phrase and phrases may contain at most two non-
heads, Kayne (1994:28ff., passim) is forced to postulate covert heads, i.e.
heads that do not contain phonetic features, to accommodate both moved XPs
and classical adpositional phrases, e.g. adnominal adjectives and adverbs, as
specifiers (cf. Kayne 1994:137, note 31). The licensing of specifiers, i.e. non-
complement XPs, is subject to feature checking under specifier-head agreement
(cf. Kayne 1994:29f.).44
Since, in the framework under consideration, a head allows for at most one
complement and at most one specifier, it follows that adverbs, i.e. AdvPs, must
be either complements or specifiers. With respect to the former type of
adverbs, e.g. manner adverbs, Kayne (1994:75) adopts Larson's (1988:350)
analysis, according to which AdvPs are innermost complements within VP-
shells, as indicated in (98):45
The AdvP carefully is the complement of the verb read, whose direct object is
realized as the specifier within the innermost VP, i.e. VP1, as in (98b). The
verb raises into the head of the VP-shell, i.e. V2, as (98c), thus yielding the
linear order in (98a).
Reference to AdvPs as specifiers, as in (99), is as sparse as reference to
complement AdvPs, which reduces to the case in (98):46
Since [DP Peter] occupies the specifier position of IP, [AdvP yesterday] in (99a)
cannot be adjoined to the IP, but must adjoin to the projection of a covert head
above [IP Peter danced]. Never in (99b) may be adjoined to the projection
headed by the auxiliary has since that projection has not yet been adjoined to:
(100) a. [XP [AdvP yesterday] [X' [X ] [IP [DP Peter] [I' [I ] [VP danced]]]]]
b. [YP [AdvP never] [Y' [Y hasi] [IP [DP Peter] [I' [I ti] [VP [V ti] [VP danced]]]]]
48 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
The string in (99c) is ruled out because never, as opposed to yesterday in (99a),
is an affective operator in the spirit of Klima (1964) which requires that the
head of the projection it adjoins to be filled by an auxiliary or a modal verb.
Yesterday is not an affective operator and thus cannot adjoin to an overtly
headed projection. Hence the contrast in acceptability between (99d) and (99a).
In recent years, Kayne's (1994) sketch of the analysis of adjuncts as unique
specifiers has been widely adopted and ultimately fleshed out into a syntactic
theory of adverb licensing, with Cinque's (1999) and Alexiadou's (1997) stud-
ies being of prime interest. Their analyses, as we shall see in the following sec-
tion, share a host of assumptions common to both the Minimalist Program and
the Antisymmetry of Syntax, such as antisymmetry, specifier-head agreement as
a licensing device and the assumption that the rigid linear order of adverbs is a
corollary of the functional projections constituting the clause being rigidly
ordered. However, their analyses differ with respect to the merge sites for and
the licensing of adverbs: while Cinque (1999) takes adverbs to be unanimously
merged as unique specifiers of functional projections and licensed under speci-
fier-head agreement, Alexiadou (1997) distinguishes between specifier- and
complement-type adverbs, with the former being merged as unique specifiers
and licensed under specifier-head agreement and with the latter either raising to
a designated specifier position or incorporating into the lexical verb.
(101) frankly > fortunately > allegedly > probably > once > … > perhaps > …
> possibly > usually > again > often > … > already > no longer > … >
always > … > soon > briefly > … > almost > completely > tutto > well
(adapted from Cinque 1999:106)
(102) a. Honestly, she's probably a bit too nosy. (*probably > honestly)
b. John would obviously have carefully planned his next move.
(*carefully > obviously)
c. They have probably long abandoned it. (*long > probably)
d. He foolishly may again have been trying to stay up late. (*again >
foolishly)
e. She has since already bought five hats. (*already > since)
f. He often could no longer remember his PIN. (*no longer > often)
g. They no longer could always afford a babysitter. (*always > no
longer)
h. Martians always briefly address the Senate. (*briefly > always)
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 49
(103) a. Iskreno, ja nažalost imam jako loše mišljenje o vama. (*nažalost >
iskreno) [Bosnian/Serbo-Croatian]
"Frankly, I unfortunately have a very bad opinion of you"
b. buxing tamen xianran dui ni wuhui hen shen. (*xianran > buxing)
"Unfortunately they evidently have a deep misunderstanding of
you" [Chinese]
c. Gianni ha ora saggiamente ceduto. (*saggiamente > ora)
"Gianni has now wisely surrendered" [Italian]
d. Dani 'ulay be-xoxma yitpater. (*be-xoxma > 'ulay) [Hebrew]
"Dani perhaps will intelligently resign"
e. Idiotischerweise schläft er gewöhnlich im Keller. (*gewöhnlich >
idiotischerweise) [German]
"He foolishly usually sleeps in the basement"
f. Klokken to har Per vanligvis allerede spist. (*allerede > vanligvis)
"At two o'clock has Peter usually already eaten" [Norwegian]
g. U to vrijeme nije već više ništa imala. (*više > već) [Bosnian/Serbo-
Croatian]
"At that time, she did already did not have anything any longer"
h. Ka ribërë pjsërisht gjithçka mirë Beni. (*mirë > pjsërisht)
"Beni has partially done everything well" [Albanian]
i. ta wanquan gaixie de hen hao (*hen hao > wanquan)
"He completely revised it very well" [Chinese]
j. Ha già detto tutto bene Gianni. (*bene > tutto)
"Has said everything well Gianni." [Italian]
relation between adverbs and the functional projections they occur in is that of
specifier-head agreement.49 Since hierarchical structure always completely
determines linear order in terms of transitivity, antisymmetry and totality (cf.
Kayne 1994:3), the canonical order of adverbs, as sketched in (101) – (103), is
complimentary:
both within and across languages. That is to say that deviations from the uni-
versal hierarchy in (104) are derivative. Thus, the unique canonical order of
adverbs as derived from (104), can be affected, i.e. reversed, scrambled, etc., in
(at least) the cases cited in (108) (cf. Cinque 1999:3ff. for discussion):
The examples given in (109b) – (109d) illustrate the case cited in (108a), where
the partial order loudly > again, as predicted by the hierarchical order VoiceP
>AspPrepetitive(II) (cf. (109a)), is reversed:
(108b), i.e. the case where some lower AdvP is wh-moved across some higher
AdvP, is illustrated in (110b) where the underlying partial order always >
carefully in (110a) is reversed:53
The string in (111b) illustrates the case in (108c): the chunk sempre i nostri
inviti 'always our invitations', which contains the lower adverb sempre 'always',
has raised across the higher adverb mica più 'no longer', which bears nuclear
stress and thus occurs in final position:54
Since sempre cannot have undergone any kind of operator movement (e.g. wh-
movement) and since the internal argument, i nostri inviti 'our invitations', is
not in its base position (viz. (111c)), the only possible way for deriving the
order in (111b) is to assume that a larger constituent containing sempre and the
lexical material following it, i.e. i nostri inviti, has moved leftwards, as illus-
trated in (112), which is adapted from Cinque (1999:14):55
This analysis allows for discarding VP-final positions for adverbs, since – as
Cinque (1999:22) puts it – their "existence is only an illusion created by mov-
ing lower portions of the clause around one or more AdvPs higher up in the
structure of the clause."56
The examples given in (113) illustrate case (108d), where one and the same
AdvP is base-generated in two different positions and thus receives two distinct
interpretations:
Carefully in (113a) relates to the manner in which Marvin sliced the bagels,
whereas in (113b), carefully related to Marvin's attitude in the bagel-slicing-
event. The same difference applies to cleverly in (113b, c). Supportive
evidence for the assumption that the adverbs in (113) are base-generated in
different positions derives from the fact that they can co-occur, which they
should not if they belonged to the same (semantic) class (cf. Cinque 1999:19ff.
for discussion):
The examples in (115) are instances of the case (108e) above. The sentence
adverbs in (115) assume the function of focusing adverbs or focusing particles
like those in (115b, c), i.e. they take scope over the constituent they precede:
The fact that focusing adverbs or focusing particles, together with the constitu-
ent they take scope over, undergo focus movement (116a), occur in cleft posi-
tion (116b) or intervene between a verb and its complement (116c) provides
evidence for the assumption that they form a constituent with the constituent
they take scope over, i.e. they are base-generated in positions that are different
from those positions in which sentence adverbs are base-generated:
The last case in point, (108f), is illustrated in (117), where sentence adverbs are
used parenthetically, i.e. they are set off by comma intonation:
(118) wh-moved adverbs > sentence adverbs > VP adverbs > (DPsubj) >
parenthetical sentence adverbs > (V) > complements > place, time,
manner, etc. adverbials > (modified, coordinated, focused) VP adverbs >
de-accented material (adapted from Cinque 1999:15)
license "a direct object which is affected by the predicate" (cf. Alexiadou
1997:131):58
(119) a. But it would be hard to ruin this part of the glen completely. (BNC)
b. * But it would be hard to ruin completely.
c. They surrounded the house completely. (BNC)
d. * They surrounded completely.
Temporal adverb(ial)s pattern like the verb's nominal internal arguments in that
wh-movement of these elements gives rise to crossover effects in cases in
which wh-movement proceeds across a bound pronoun (121a, b), and to weak
island violations if wh-movement proceeds across a wh-operator (121c, d):59
Third, manner adverbs may incorporate into the lexical verb. Alexiadou
(1997:177ff.) argues that, since incorporation is restricted to taking place in
government configurations, manner adverbs, pace Cinque (1999), cannot
originate in a preverbal specifier position, i.e. specVoiceP:
(126) ... [ VoiceP [Advmanner]j [Voice' Voice [VP DP [V' [V] [Adv tj ]]]]] ...
The fact that single-term complement-type adverbs may occur both preverbally
(127c) and postverbally (128) suggests that raising to specVoiceP is optional:
However, this is not true. As has already been argued by Lonzi (1990: 151;
1991:358), manner adverbs can remain in situ only if they are either
syntactically or prosodically heavy, as in (129) and (130) respectively:64
(131) VoiceP
3
AdvPi VoiceP
light/weak 3
Voice VP
3
V AdvPi / AdvP
light/weak strong/heavy
Preverbal occurrences of manner adverbs, as in (132), are clearly not the result
of overt raising to specVoiceP, but of adverb incorporation into the governing
lexical verb yielding complex verbal heads of the type [V Adv V] (cf. (133)):66
(132) a. Kalo-efaga
well-ate-1SG
"I ate well"
58 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
b. Gorgo-petaksa
fast-flew-1SG
"I flew fast"
c. I Maria kako-ferete stin adelfí tis
The Mary badly-behaves to sister hers
"Mary behaves badly to her sister"
d. I Maria kalo-dinete
The Mary well-dress-PASS:3SG
"Mary dresses well" (Alexiadou 1997:179f.)
(133) VP
3
V0 AdvP
2 #
Adv 0 V0 Adv0
The question arises how completion adverbs and temporal adverbs meet the
Licensing Principle in (134). With respect to the former, Alexiadou (1997:
142f.) merely conjectures that they are licensed in the specifier position of an
aspectual functional projection outside VP, which is involved in the checking
of the verb's aspectual features. This would explain the preverbal occurrences
of completely in (135):
(135) a. She completely ruined the evening for many of us. (BNC)
b. But marriage had completely changed Miss Caroline. (BNC)
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 59
(136) a. [It] would be hard to ruin this part of the glen completely. (BNC)
b. They surrounded the house completely. (BNC)
(137) T'
3
T VP
tp
V AdvP/PP
6
a. [+PAST] left {yesterday[+PAST]/*tomorrow[–PAST]/then[±PAST]}
b. will[–PAST] leave {*yesterday[+PAST] / tomorrow[–PAST]/then[±PAST]}
To account for the fact that temporal adverbs do not raise to specTP, as
witnessed by the ungrammatical strings in (138), Alexiadou (1997:115ff.)
follows Chomsky (1995:235, 329ff.) in assuming that (since adverbs do not
have morphosyntactic features to check) raising is restricted to temporal
features which are vital for the interpretation of the sentence, i.e. in order not to
give rise to violations of the Principle of Full Interpretation:69
(140) a. She cleverly has often been (often) drinking pineapple juice.
b. She may have often been cleverly avoiding encounters with aliens.
Problems with multiple adverb positions arise in cases in which the various
positions assumed by one adverb do not coincide with interpretational differ-
ences, i.e. the adverb receives the same basic interpretation across positions, as
shown in (145) for the subject-related adverb cleverly and the frequency adverb
frequently:71
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 61
(145) a. Cleverly she (cleverly) will (cleverly) have (cleverly) been (cleverly)
avoiding him.
b. Frequently she (frequently) has (frequently) been (frequently) talking
to strangers.
(146) [AgrSP [D she] [AgrS' [AgrS ] [ModP [Adv cleverly] [Mod' [Mod ] [TP [T will] [have
been avoiding this topic]]]]]]
This analysis gives rise to two problems: First, given that finite non-lexical
verbs, e.g. modals, undergo overt raising to AgrS, cleverly should not be able to
intervene between the subject and the finite non-lexical verb. Certainly, we
could modify the condition on verb raising to the extent that all verbal
elements, i.e. lexical and non-lexical verbs, resist overt raising to AgrS. Techni-
cally possible though the implementation of this modification is (i.e. the strong
D-feature in AgrS attracts the corresponding feature of the finite verbal element
without pied-piping its phonetic form, as is the case with finite lexical verbs), it
would cause problems with respect to strings as in (147) where the verbal clitic
'll attaches to the pronominal subject:
Second, this analysis does not provide a straightforward explanation for the
occurrence of cleverly below will (or any other non-lexical verb). As down-
ward movement is ruled out on general grounds, the partial relative orders in
(148) must be accounted for in terms of optional verb raising:
62 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Given the structure in (146), (148a) could be derived by the modal raising from
T to AgrS. The derivation of (148b) would have to involve overt raising of the
bare infinitive to T, i.e. the formation of a complex verbal head [ T [will]
[have]], and the subsequent raising of [T [will] [have]] to AgrS. Other things
being equal, the derivation of (148c) would have to involve raising of the past
participle been to the head position occupied by have. Such an analysis is
untenable not only because it rests on the assumption that the raising of non-
lexical verbs applies optionally, but also because nothing would prevent
complex verbal heads of the types [T [will] [have]] and [T [will] [[have]
[been]]] to undergo further raising operations, e.g. to the head position of CP:
(150) AgrS'
3
AgrS FP1
3
F FP2
3
F ModP
2
Adv Mod'
2
Mod TP
3
T AuxP
3
Aux AuxP
3
Aux VP
a. willi have been cleverly ti have been
b. willi havej been cleverly ti tj been
c. willi havej beenk cleverly ti tj tk
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 63
we not only would face the optionality problem (i.e. non-finite non-lexical
verbs may or may not move), but also preprogram violations of the Head
Movement Constraint since the derivation of (148b) would involve the raising
of have across the base position of the modal (ti). The derivation of (148c)
would involve a multiple violation of the Head Movement Constraint since
have raises across the base position of the modal (ti) and been raises across
both the base position of the modal (ti) and that of have (tj).73
Alternate linear orders of preverbal adverbs as in (151a) and (151b) pose
a potential problem for the specifier analysis since the interpretation of the
adverbs, again, is constant across positions:74
(152) [AgrSP [D she] [AgrS' [AgrS ] [FP [Adv {wisely/suddenlyk}] [F' [F hasi] [AspP [Adv
frequently] [Asp' [Asp ti] [{ModP/AspP [Adv tk ] [Mod'/Asp' [Mod/Asp ti ] [TP [T ti ] [VP ti
…]]]]]]]]]]
That is to say that the strings in (153) are derived by raising the structural
complement of the functional head with which the adverb under consideration
stands in a specifier-head agreement configuration across the projection host-
64 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
ing the adverb. The raised complement, as can be seen in (154) for (153a),
targets the specifier position of an additional functional projection FP that
dominates the functional projection hosting the adverb:
(154) [AgrSP [D she] [AgrS' [AgrS ] [FP [AgrOP asked him]i [F' [F ] [TP [Adv once ] [T'
[T ] [AgrOP ti ]]]]]]]
While yielding the desired result, this operation lacks a trigger. Moreover, the
postulation of the additional functional projection FP can hardly be seen as
justified by interface conditions; its justification by theory internal arguments is
dubious. Note incidentally that the adverb takes scope over AgrOP, irrespective
of whether AgrOP raises to specFP or stays put.
Snowballing is seen to also account for the mirror image order of post-
verbal adverbs in so-called inverse languages, e.g. Malagasy: According to
Rackowski & Travis (2001:121), preverbal adverbs (155a) follow the linear
orders discussed by Cinque (1999), whereas postverbal adverbs occur in
inverse order ((155b) & (156)), i.e. their linear order is the mirror image of the
linear order of preverbal adverbs ((156a – c) from Laenzlinger 2002:128):
(155) a. na(dia) > matetika > tsy > efa > mbola > tsy > V
even generally neg already still neg
b. V > tsara > tanteraka > foana > intsony >mihitsy > aza
well completely always anymore at-all though
(adapted from Rackowski & Travis 2001:121ff)
(157) NegP
th
tsy Neg'
th
Neg MihitsyP
th
Mihitsy'
th
mihitsy IntsonyP
th
Intsony'
th
intsony FoanaP
th
Foana'
th
foana TantP
th
Tant'
th
tanteraka TsaraP
th
Tsara'
th
tsara VP
Abstracting away from the fact that adverbs are seen as functional heads, the
problem we are facing with (157) is essentially the same we faced in connec-
tion with sentence-final occurrences of adverbs (cf. (153)): nothing in the sys-
tem forces us to assume this kind of raising operation, especially so since the
relative scope of adverbs is calculated at underlying structure.
Given the predominant view that lexical verbs in English do not raise in
overt syntax (cf. Pollock 1989, Chomsky 1995), neither Cinque's (1999) and
Laenzlinger's (1996, 1998) version of manner adverbs as specifiers, nor
Alexiadou's (1997) analysis of manner adverbs as (raisable) complements
allow for a straightforward account of the positions (these) adverbs assume
relative to the verb's internal argument(s). Irrespective of whether manner
adverbs originate in specVoiceP, i.e. in pre-VP position, or as innermost
complements within VP, the derivation of the strings in (158) would have to
involve a number of overt raising operations, notably operations that are
deemed unavailable in English (cf. Chomsky 1995:331f.), such as the overt
raising of V and DP to AgrO and AgrOP, respectively, as shown in (159a) for
(158a), where both the verb and its nominal internal argument, i.e. the napkin,
must raise overtly across the preverbal manner adverb; the derivation of (158b)
involves overt verb raising across aggressively (irrelevant traces omitted):
(159) a. he handedk [the napkin]i [VoiceP [Adv secretly] [Voice' [Voice tk] … [V' tk ti
to her]]]
b. they insistedk [VoiceP [Adv aggressively] [Voice' [Voice tk ] … [V' tk on his
arrest]]]
Thus we have to assume that the manner adverb raises overtly to specVoiceP
and that both V and DP raise as in (159), while [PP to her] and [PP on his arrest]
stay in their respective base positions. However, this assumption leaves us with
the problem that a non-single-term manner adverb raises overtly to specVoiceP
(cf. Alexiadou 1997:140):
Problems multiply if the manner adverb precedes both of the verb's internal
arguments:
Further, the specifier analysis predicts that adverb stacking is readily available
in pre-, post- and interverbal positions since adverbs sit in specifier positions
and thus are flanked verbal heads positions [X ], which actually could be
targeted under verb raising operations:
(164) [XP [Adv ] [X' [X ] [XP [Adv ] [X' [X ] [XP [Adv ] [X' [X ] [XP [Adv ] [X' [X ] …
]]]]]]]]
However, as has been pointed out by Costa (2000), adverb stacking often
requires that verbal heads be lexicalized (cf. (165a) vs. (165b):
Further, the fact that coordination is possible even though the adverbs belong
to different semantic classes, e.g. aspectual and manner as in (167), runs
counter to the rationale of the specifier analysis according to which elements
with different meanings occupy different specifier positions, i.e. aspectual
adverbs cannot be licensed in specVoiceP and manner adverbs cannot be
licensed in the specifier position of an aspect-related functional projection, e.g.
AspPfrequentative(I).77 Notice also that the canonical order of adverbs can be
reversed under coordination:
While it is perfectly clear that the adverbial cluster forms a constituent that
occupies specCP with the finite verb having raised to C (cf. (170)), the internal
structure of the adverbial clusters is far from clear, especially since the
bracketed constituents contain elements belonging to different semantic
classes, i.e. temporal and spatial in (169a, b) and temporal and reason in
(169c):78
(170) [CP [?P letztes Jahr im Juni … unter einem Tuch ] [C' [C hat]
last year in June … under a cloth
[AgrSP [D er] [AgrS' …]]]]
he
(171) a. He has [(much more) carefully (*than anybody else)] analyzed it.
b. He has [(much less) often (*than I (thought)] rehearsed it.
(Haider 2000:99)
2.2.3 Summary
This section focused proposals concerning the structural integration and
licensing of adverbs within the syntactic frameworks defined in the Minimalist
Program and the Antisymmetry of Syntax. We have seen that the licensing of
adverbs (even if adjunction is seen as a severely constrained option with spe-
cial properties) is still problematic since licensing in terms of specifier-head
agreement is readily available for adjuncts that have been summoned to A'-
specifier positions of functional heads that bear strong –interpretable features
(e.g. [Q] or [FOC]), though not for adverbs in their respective base positions,
which are taken to be adjoined positions, not specifier positions.80 It was shown
that Laenzlinger's (1996, 1998) proposal, which draws on Chomsky's (1995)
distinction between specifier and adjunct in terms of A- vs. A'-features, not
only allows for the identification and definition of base positions for adverbs,
but also for the licensing of adverbs under specifier-head agreement with
designated functional heads. Laenzlinger's (1996, 1998) argument that the base
positions of adverbs are genuine specifier positions is corroborated by the pres-
ence vs. absence of Relativized Minimality Effects, depending on whether rais-
70 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Following standard assumptions, Frey & Pittner (1998, 1999) take structure to
be binary branching with verbs projecting onto Larsonian (1988) shells.
Further they assume that the verb's internal arguments are generated as
specifiers of VP-shells, whereas the verb's external argument is generated as
the specifier of some designated functional projection, e.g. Bower's (1993)
PredP. Under this analysis, the lexical verb must raise overtly to the head
position of PredP in actives to license the subject in specPredP by assigning it
its external Θ-role; in passives, the lexical verb stays put. Overt subject raising
to specIP is Case-driven:
Manner adverbs, in accordance with (173a), are base adjoined to the lexical
verb or its trace whence they invariably follow the lexical verb, as shown for
nicely in (175); preverbal occurrences of nicely are not process-related, but
rather event-internal as in (176a), or proposition-related as in (176b). Frey &
Pittner (1999) argue that right-adjunction of the adverb to the verb or its trace
as in (175b) is warranted by the fact that the adverb may precede the verb's
internal argument(s), provided the latter do(e)s not require Case (cf. below)):
(175) a. John has spoken (nicely) to his mother (nicely) about the book nicely.
b. John has [[spoken1 nicely] [to his mother [[t1 nicely] [about the book
[t1 nicely]]]]] (Frey & Pittner 1999:18f.)
(176) a. It is utterly at variance with this to adopt […] the role of the teacher as
self-appointed proof-reader… (BNC)
b. [There] was still an old Kodak Pan film in the camera and he nicely
has developed it. (www)
They point out correctly that the verb's internal arguments, i.e. [PP to his
mother] and [PP about this book] in (175), are not extraposed from their base
positions to some VP-final A'-position, since both [PP to his mother] and [PP
about this book] allow for the preposition's argument to be extracted (cf. (160)
above; example from Frey & Pittner 1999:19):
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 73
Preverbal 'manner adverbs' do not specify the process denoted by the verb, but,
depending on position, either the subject's attitude towards the event (180a),
i.e. Bob was careful in removing the lid, or the speaker's attitude towards both
the event and the subject's behavior in that event (180b), i.e. it was careful of
Bob to remove the lid:
The interpretational differences between (180a) and (180b) on the one hand
and (180a, b) and (181) on the other are seen as a concomitant of carefully
figuring in three distinct semantic classes and thus being subject to three
distinct structural conditions (cf. Frey & Pittner 1999:21ff.). The only available
reading of (181) is the manner reading:
As we have seen in connection with nicely in (175), manner adverbs are base
adjoined to the lexical verb or its trace whence they invariably follow the lexi-
cal verb. Carefully in (180a) is subject to the structural condition on event-
internal adverbs in (173b) and thus may assume a position which is minimally
c-commanded by the argument it is related to. Since carefully relates to the
subject, follows the finite non-lexical verb and precedes the lexical verb in
Pred, PredP suggests itself as a potential adjunction site, as shown in (182a):
74 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(182) [IP [DP Bobk] [I' [I hasi] [AuxP [Aux' [Aux ti] [PredP [PredP [Adv carefully] [DP tj]
[Pred' [Pred removedk] [VP [DP the lid] [V' [V tk ]]]]]]]]]]
(183) [IP [DP Bobk] [I' [I ] [AuxP [Adv carefully] [AuxP [Aux' [Aux has] [PredP [DP tj] [Pred'
[Pred removedk] [VP [DP the lid] [V' [V tk ]]]]]]]]]]
However, to account for the fact that carefully in (180b) occurs between the
subject and the finite non-lexical verb, Frey & Pittner (1999) have to assume
that finite non-lexical verbs undergo raising to I only optionally. If raising
takes place, the resultant surface string is indistinguishable from that contain-
ing event-internal carefully, since the adverb follows the finite non-lexical
verb. That is to say that (180a) should be ambiguous between the proposition-
related and event-internal reading of carefully, which it is not (!) since care-
fully clearly does not express the speaker's attitude. This problem could easily
be solved if we assumed that subject-oriented carefully is base adjoined to IP
(cf. Svenonius 2002:228ff.):
(184) [IP [DP Bobk] [IP [Adv carefully] [I' [I hasi] [AuxP [AuxP [Aux' [Aux ti] [PredP [DP tj]
[Pred' [Pred removedk] [VP [DP the lid] [V' [V tk ]]]]]]]]]]]
(187) a. * With Johni 's computer hei began to write a book […].
b. * For Maryi 's brother shei was given some old clothes.
Frey & Pittner (1999) discuss multiple adverb positions in connection with
frequency adverbs which they take to combine the distributional properties of
manner and temporal adverbs, i.e. these elements may modify both processes
and events. If frequently in (188) is process-related, it must be c-commanded
by the subject and adjoined to the verb/predicate or its trace, if frequently is
event-related, it must c-command the base position of the subject, i.e. fre-
quently must be adjoined to PredP or above:
Twice in (189a) is ambiguous between the process and the event reading.
Under the process-reading, twice is within the scope of the subject-attitude
adverb intentionally, whereas, under the event-reading, twice takes scope over
intentionally. The latter case is an instance of reverse ordering. No such
ambiguity can be observed in (189b) where twice clearly c-commands the base
position of the subject:
Instead of following one of the two conceivable ways in accounting for twice
in (189a) outscoping intentionally, namely (a) right-adjunction of twice above
the projection containing intentionally or (b) intraposition à la Cinque (1999),
Frey & Pittner (1999) sketch an analysis that incorporates Chomsky's (1995)
contention that "if a shell structure is relevant at all, […] additional phrases
might be supported by empty heads below the main verb" (cf. Chomsky
1995:333). Under this scenario, the lexical verb would take as its structural
76 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(190) VP
2
V VP
2
Adv V'
2
V VP
2
Adv V'
2
V Adv
Frey & Pittner (1999:38f.) and Frey (2000:130f.) suggest that adverbs that are
generated within these semantically empty verbal projections cannot be inter-
preted in their base positions, but must be interpreted from above the base
position of the verb, i.e. in the interpretable part of the sentence. They suggest
that the interpretation of event-related adverbs in sentence-final position is
contingent on the presence of abstract modification markers in the interpretable
part of the sentence.85 Under this proviso, the wide scope reading of twice
could be accounted for by sentence-final twice hooking up to an abstract modi-
fication marker above the position occupied by intentionally. The modification
marker under consideration is subject to the conditions on event-related
adverbs, i.e. it must c-command the base position of the verb's highest argu-
ment (i.e. the subject trace in specPredP) and it must c-command the base posi-
tion of event-internal adverbials (e.g. intentionally).
The assumption that higher adverb(ial)s in postverbal positions, despite
being located within the semantically empty VP, are licensed indirectly, i.e.
from within the interpretable part of the sentence, also plays a central role in
Haider's (2000, 2004) analysis of inverse order patterns of preverbal
adverb(ial)s in OV and postverbal adverb(ial)s in VO languages:
Haider's (2000, 2004) proposal rests on the assumption that the mechanisms of
adverb(ial) placement are not uniform across positions: while preverbal
adverb(ial)s are adjoined within the projection of the lexical verb, postverbal
adverb(ial)s are embedded in Larsonian VP-shells, as shown in (193) and (194)
respectively, where ∆ marks potential adverb(ial) positions:
(193) VP (194) VP
2 2
∆ ← V' V0 → VP
2 2
∆ ← V' ∆ V'
2 2
∆ ← V0 V0 → VP
2
∆ V'
2
V0 → ∆
(adapted from Haider 2000:126)
The rationale behind the two modes of adverb(ial) placement lies in the failure
of traditional adjunction analyses to account for the fact that the linear order of
the postverbal adverb(ial)s in (192) mirrors the linear order of the preverbal
adverb(ial)s in (191). That is to say that, since adjunction is an internally
unrestricted process, non-mirrored linear orders of postverbal adverb(ial)s
should be as readily available as inversely ordered preverbal adverb(ial)s:
(195) a. * She has worked [time today] [time the whole time] [space in the garden]
[process with great care] [respect on her hobby]
b. * Sie hat [respect an ihrem Steckenpferd] [process mit großer Sorgfalt]
[space im Garten] [time die ganze Zeit] [time heute] gearbeitet
The Interface Criterion in (196) requires that mapping be both monotonic and
incremental. That is to say that once a higher type of adverb(ial), e.g. a t-related
adverb(ial), has been addressed by the mapping procedure, mapping of a lower,
i.e. less specified type of adverb(ial), e.g. an e-related one, is impossible since
the t-related adverb(ial) would be placed in the domain of e-related
adverb(ial)s, whence the strings in (198c) and (198d) are illicit:
(198) a. Ms. San had seriously once thought of accepting her for a god-
daughter as she had no children of her own. (www)
b. The stoves were cheap imitations of Preways and have fortunately
long since disappeared, … (www)
c. * Ms. San had once seriously thought of accepting her …
d. * The stoves […] have long since fortunately disappeared.
Since the mapping onto semantic type domains parallels syntactic incremental-
ity, the mapping of a lower type, e.g. an e-related adverb(ial), closes off the
domain of a higher type, e.g. a t-related adverb(ial). Domain closing is an
inconvertible process, i.e. a closed-off domain cannot be reopened (cf. Haider
2000:130f.). Thus, (199) is a corollary of the Interface Criterion in (196):
(200) a. They [VP [V met] [ VP [space in the attic] [ V' [ V ] [time on Tuesday]]]]
b. * They [VP [V met] [ VP [time on Tuesday] [ V' [ V ] [space in the attic]]]]
(201) Speech-Act > Fact > Proposition > Event > Specified Event
(Ernst 2002:53)
The hierarchy of FEO types in (201) basically reflects the ordering of composi-
tional rules in the sense that operations on a given type of FEO are either type
preserving, i.e. Events may be converted into other subtypes of Events, or type
shifting, i.e. a lower FEO type may be converted into a higher FEO type, but
not vice versa.
80 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Ernst (2002:58, passim) maintains that the manner reading in (202a) is derived
from the clausal reading in (202b). The two instances of cleverly differ with
respect to the comparison classes they evoke. For agent-oriented cleverly in
(202b), the comparison class is events in general so that cleverly warrants the
positing of the property denoted by the adverb's adjectival base, clever, in the
Agent, with the degree to which that property obtains exceeding the standard
norm of cleverness displayed by Agents in Events. For the manner variant of
cleverly in (202a), the comparison class consists of all events of Sally opening
letters.88 Epistemic and evaluative adverbs combine with propositions to yield
propositions and thus are barred from occurring to the right of the base position
of an aspectual head since the latter requires an Event (203d, e). By the same
line of reasoning, speech-act adverbs are barred from following the base posi-
tion of an aspectual or modal head, i.e. speech-act adverbs operate on proposi-
tions which are not available within the domain governed by aspectual or
modal heads (203f):89
The ungrammaticality of the strings in (204c) & (204d) "results from one
adverb requiring a particular FEO at a specific point in structure, which
prevents the second adverb from fulfilling its own scope requirements" (cf.
Ernst 2002:217). Thus, for example, probably combines with a proposition to
yield a proposition and cannot occur within the scope of tactfully since tactfully
requires as its FEO an event, not a proposition.
A second huge class of adjuncts Ernst (2002) discusses in detail is that of
functional adjuncts which comprises time-related elements (e.g. now, for a
minute, still), quantificational elements (e.g. frequently, again, precisely),
focusing elements (e.g. even, just, only), sentence-negating not and adverb(ial)s
specifying clausal relations (e.g. purpose, causal, concessive, conditional, etc.).
By and large, functional adjuncts engage in focus-presupposition structures
(e.g. focusing elements) and/or quantification either over events or with respect
to completion etc. (cf. Ernst 2002:120ff.). Functional adjuncts may assume a
number of positions, in each case of which they single out different types of
entities.90 Thus, for example, only makes an assertion about Carol in (205a),
about buying in (205b) and about junk food in (205c). Again in (206a) indicates
that the speaker adds a new point, in (206b) again indicates that the state of
affairs under consideration has obtained at some point in the past and finally, in
(206c), again indicates the repetition of a (nonstative) event ((205) & (206)
adapted from Ernst 2002:121):
[+F] items are typically specifiers that are licensed under specifier-head
agreement with a head bearing relevant features. F-Dir stands for Functional
Direction, which is universally leftwards, i.e. specifiers precede heads, and C-
Dir stands for Complement Direction, which is subject to parameterization,
expressed by (in)activity. If C-Dir is inactive, as is the case in OV languages,
nonheads are licensed as left sisters ([–R]). This holds for specifiers,
complements and adjuncts alike. If C-Dir is active, as is the case in VO
languages, [–F]-nonheads within the projections of lexical heads are linearized
as right sisters [+R] (e.g. manner adverbs), whereas, in functional projections,
they are linearized as either right or left sisters (with the exception of structural
complements of functional heads which are invariably [+R]). Since the verb's
thematic arguments are generated as specifiers, they are [+F] and thus
linearized as left sister, ([–R]). As regards adverb(ial)s, Ernst (2004a:760)
presents the following patterns:
in principle, may be [±R], their realization in pre- and postverbal position is not
free, but subject to Weight Theory, as outlined in (213), which basically
restates the empirical fact that heavy constituents follow the lexical verb,
whereas light constituents precede it, as shown in (214) and (215), both taken
from Ernst (2002:172):91
(212) IP
2
AdvP IP
2
IP AdvP
2
DP I'
2
I PredP
2
AdvP PredP
2
PredP AdvP
2
Pred VP
2
VP AdvP
2
DP V'
#
V
(Ernst 2002:168)
(215) a. Ice-fishermen (*because they like it) fish because they like it.
b. The cupcakes (*over an hour before we got there) had been finished
off over an hour before we got there.
To account for the fact that elements such as hardly and just in (214) are
strictly preverbal, Ernst (2002) proposes endowing them with the feature
[+Lite] which figures in Ernst's (2002:172) linearization principle:
Recall that under Haider's (2000, 20004) analysis, the linear order in (216b)
obtains in preverbal position, while, in postverbal position, the linear order is
reversed. Under Frey & Pittner's (1998, 1999) analysis (cf. (216a)), process-
related adverb(ial)s, e.g. manner adverbs, are strictly postverbal in VO
languages, whereas under Ernst's (2000, 2002, 2004) analysis (cf. (216c)), the
same class of elements (referred to as specified event adverb(ial)s) may be both
post- and preverbal. In the latter case, they necessarily follow non-lexical verbs
and sentential negation.
In the following, we take a closer look at how individual semantic scope
analyses fare with respect to multiple and alternate adverb positioning, the
linear orders of adverbs and the verb's internal argument(s), adverb stacking,
coordination and clustering and so-called Edge Effects. It will be shown that
the picture that emerges is relatively unbalanced in the sense that the analyses
under consideration excel on the critical data to varying degrees. Moreover, it
will emerge that theoretical and empirical arguments in favor of adjunction-
based semantic scope analyses are not compelling and that the specifier analy-
sis, albeit with slight modifications, is fully compatible with the data.
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 85
Given Frey & Pittner's (1998, 1999) proposal according to which seemingly
right-adjoined adverb(ial)s, such as last Tuesday and yesterday in (219) are
base-generated within the empty verbal structure below the lexical verb and
licensed indirectly by abstract modification markers in the interpretable part of
the sentence, the data in (219) can be accounted for. That is to say that the
abstract temporal modification marker, which fulfills the structural requirement
for event-related adverb(ial)s, licenses the actual temporal expression in
sentence-final position. Thus, the relative scope relation between temporal and
spatial elements is preserved. Supplemented with appropriate conditions on
prosodic heaviness, their analysis would also account for the strings in (220):
Haider (2000, 2004) explicitly refutes the idea that syntax feeds semantics (i.e.
there are no designated base-domains) and suggests that the mirror canonical
order of postverbal temporal and spatial adverb(ial)s follows from monotonic
incrementality which requires that a semantically lower type be addressed
before a semantically higher type (i.e. space before time), with the order of
addressing reflecting the order of structural integration (cf. Haider 2004:802).
An analysis along these lines rests on the implicit assumption that the order
within the scope domains in (216) is not free, but subject to finer grained scope
relations, e.g. 'time > space', which is linearized as 'space > time' in postverbal
86 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
position (cf. (219) & (220) and (191) & (192)). Strings as in (218) do not pose
a problem for his analysis either: the preverbal temporal expression would be
left-adjoined to VP and the postverbal spatial expression would be embedded
within the empty verbal structure below the lexical verb. Given that spatial
expressions constitute a lower type than temporal expressions, they have to be
addressed first. This is possible since, in a VO language, structural composi-
tionality and incremental compositionality need not coincide (cf. above). Under
Ernst's (2002:264ff.) analysis, spatial adverb(ial)s are event-internal in very
much the same way as the verb's arguments, i.e. they are confined to occurring
within VP, whereas temporal adverb(ial)s are confined to the so-called Aux-
range, i.e. the domain delimited by TP and PredP. This automatically accounts
for (218). The derivation of the strings in (219) and (220) would have to
involve right-adjunction of the temporal adverb as determined by Weight
Theory (assuming that last Tuesday and yesterday in (219) and tomorrow and
now in (220) are [+Heavy]). Again, the spatial expressions are within the scope
of the temporal expressions (cf. Ernst 2002:156ff).
Postverbal mirror orders of adverb(ial)s, as discussed in 2.2.2 for
Malagasy, are accounted for by the same set of mechanisms:
(221) V > tsara > tanteraka > foana > intsony > mihitsy > aza
well completely always anymore at-all though
(adapted from Rackowski & Travis 2001:121)
Under Ernst's (2002) analysis, the relative linear order of postverbal adverbs in
(221) would be accounted for under right-adjunction to suitable chunks of
verbal structure, as shown in (222), with the relative scope relations between
any two postverbal adverbs following automatically since the rightmost adverb
asymmetrically c-commands the adverb to its left:
(222) [PredP [PredP [PredP [PredP [Pred ] [VP [VP [VP ] [tsara]] [tanteraka]]] [foana]]
[intsony]] [mihitsy]]
Under Frey & Pittner's (1998, 1999) and Haider's (2000, 2004) proposal, the
postverbal adverbs (excepting manner adverbs) would occupy positions within
the empty verbal structure below the lexical verb, as shown in (223).92 Recall
that, under Frey & Pittner's proposal, the mirror relative order of adverbs
follows from the relative order of abstract modification markers, [MM], in
preverbal position to which the postverbal adverbs hook up, while under
Haider's proposal, mirror linear orders follow from monotonic incrementality:
(223) [VP [MM ]1 [VP [MM ]2 [VP [MM ]3 [VP [MM ]4 [V' [V ] [ [tanteraka]4 [
[foana]3 [ [intsony]2 [ [mihitsy]1 [ ]]]]]]]]]]
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 87
However, extraposition is not an option (cf. (160) & (177) above), since extra-
posed constituents are extraction islands and the allegedly extraposed constitu-
ents in (224) are not:
As has been pointed out by Pesetsky (1995), the relative scope among
postverbal adverb(ial)s and binding data impose conflicting requirements on
syntactic structure: scope data require that structure be left-branching as in
(227), whereas binding data require that structure be Larsonian, i.e. right-
branching as in (228) (see also Frey & Pittner 1999:37, Cinque 2004:696ff. and
Haider 2004:790ff. for discussion).
Clearly, this conflict in structural requirements cannot be resolved under the
adjunction analysis since either the anaphor would c-command its antecedent,
as in (227a), or the relative scope of the postverbal adverb(ial)s would come
out wrong, as in (227b):
88 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(227) IP
qp
D I'
qp
I PredP
qp
PredP PP
qo
PredP PP
fu
P red VP
th
DP V'
gu
V PP
6
a. He gives the book to them in the garden on each other's
b. * He gives the book to them on each other's in the garden
(228) IP
2
D I'
2
I PredP
3
Pred VP
3
DP V'
3
V VP
3
PP V'
2
V VP
2
PP V'
2
V PP
6
He gives the book to them in the garden on each other's…
(adapted from Haider 2004:790; traces omitted)
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 89
(229) a. The petal was faded now but it had {once, *ONCE, *the day before
yesterday} been bright pink. (BNC)
b. The petal was faded now but it had been bright pink {*once, ONCE,
the day before yesterday}.
c. * The petal was faded now but it had once been bright pink the day
before yesterday.
(230) a. Graham will {long, * for a long time, *for two days} be remembered
for his wit. (CC)
b. Graham will be remembered for his wit {*long, for a long time, for
two days}.
c. * Graham will long be remembered for his wit {for a long time, for
two days}.
(231) a. She [TP (cleverly) [TP will [AuxP (cleverly) [AuxP have [AuxP (cleverly)
[AuxP been [PredP (cleverly) [PredP avoiding him]]]]]]]].
b. She [TP 'll [AuxP (cleverly) [AuxP have been avoiding this topic]]].
Abstracting away from Costa's (2000) claim "that the adjunction of adverbs is
parasitic on the lexicalization of the heads" (cf. Costa 2000: 23), the restric-
tions on adjunction are purely semantic, e.g. the compatibility of aspectual
properties. That is to say that the strings in (232) are syntactically possible, but
ruled out because the semantic requirements of the adverbs cannot be met.
Thus, terminative elements like no longer in (232a, b) require that a given state
of affairs not obtain any longer at reference time. This requirement cannot be
met in the context of perfective have which specifies that a given state of af-
fairs still obtains at reference time (cf. Ernst 2002:345f.). Adverbs expressing
retrospective aspect, e.g. recently in (232c, d) are parasitic on the presence of
an overt anteriority marker, i.e. past tense morphology or perfective have. This
requirement is not met in (232c, d):
(232) a. * They (no longer) have (no longer) refused our invitations.
b. * Hillary (no longer) has (no longer) been (no longer) ignoring their
advice.
c. * Hackers (recently) would (recently) access the system.
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 91
(233) AgrSP
2
D(P) AgrS'
2
AgrS …
h
ReltP
2
spec Relt'
th
Relt …
h
AspPdur
2
spec Asp'
th
Asp …
h
VP
fu
V'
eu
V VP
fh
PP V'
1
V DP
a. it had once … been bright pink
long time]
[PP for a
[PP for a
[DP the day …]
(234) a. She luckily has probably got a job. (Frey & Pittner 1999:33)
b. * She probably has luckily got a job. (Frey & Pittner 1999:33)
92 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
c. Honestly, they surely will drive us out of this house in the end.
d. * Surely, they honestly will drive us out of this house in the end.
e. Jim luckily has wisely refused the offer.
f. * Jim wisely has luckily refused the offer.
g. Gina probably has tactfully suggested that we leave.
h. * Gina tactfully has probably suggested that we leave.
((c) – (h) from Ernst 2002:127)
Under Ernst's (2002) proposal, the ill-formed strings in (234) are ruled out
since the leftmost adverb cannot meet its FEO requirements (see also (204)
above). Thus, for example, wisely and tactfully require their FEO to be an
event. This condition is met in (234e) and (234g), but not in (234f) and (234h),
where the sister constituent of the adverb represents a fact or a proposition by
virtue of containing an evaluative or an epistemic adverb, respectively. Frey &
Pittner (1999:33) merely state that, for semantic reasons, epistemic adverbs
cannot outscope evaluative adverbs (cf. (234a) vs. (234b)). Under Haider's
(2000, 2004) analysis, we would have to specify that speech act adverbs
constitute a higher semantic type than evaluative adverbs, which in turn
constitute a higher type than evidential adverbs, etc. That is to say, we
ultimately would have to come up with a hierarchy of semantic types that
parallels Cinque's (1999) hierarchy of adverbs:
(236) speech act > evaluative > evidential adverbs > epistemic > subject-related
stupidly in (235c, d), suddenly and frequently in (235e, f), and for frequently
and suddenly in (235g, h). Given Haider's (2000, 2004) Interface Criterion in
(196) and its corollary in (199), the adverbs under consideration should be of
the same semantic type whence the order of addressing is free. Interestingly,
under Frey & Pittner's (1999) analysis, only the strings in (235e) – (235h) con-
stitute a case of free ordering. Since they formally distinguish between subject-
attitude adverbs and subject-oriented adverbs, with the former being event-
related and the latter proposition-related, wisely and stupidly in (235) fall into
two distinct semantic classes and thus have distinct base positions. That is to
say that the strings in (235a) – (235d) are not an instance of free adverb order-
ing. If ordering was free, both adverbs should be able to occur in any order
after the finite non-lexical verb. However, only one order is admissible, as pre-
dicted by Frey & Pittner's (1999) analysis:95
(238) a. She wisely would frequently have cleverly avoided this topic.
b. Stupidly, she would occasionally have wisely replied to his messages.
A closer look at (235e) – (235h) reveals that the two occurrences of frequently
correspond to Cinque's (1999) distinction between high and low frequentative
aspect (cf. (104)), which is substantiated by the fact that the instances of
frequently may co-occur:96
(239) She frequently was suddenly (being) frequently rejected by the publish-
ers. (adapted from Cinque 2004:692)
Given that the positions assumed by wisely, stupidly and frequently coincide
with distinct interpretations, as witnessed by the strings in (237) to (239),
alternate orders widely discussed by opponents of the specifier analysis turn
out to be an illusion. That is to say that the ordering restrictions on the adverbs
in (235) are no less severe than those in (234).97
In light of these considerations, I would like to come back to multiple and
interverbal adverb positions:
94 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(240) a. She (cleverly) will (clev.) have (clev.) been (clev.) avoiding him.
b. He (frequently) may (freq.) have (freq.) been (freq.) eating ants.
(241) <adverb> will have <adverb> been avoiding … (cf. Cinque 2004:706)
Specifically, he argues that strings as in (242a) are derived "if the higher
instance of the AdvP is selected and if the modal crosses over it" (cf. Cinque
2004:706f.), while the derivation of (242b, c) involves the section of the lower
instance of AdvP. While (242b) corresponds to the base order yielded by low
merge, the derivation of (242c) involves the raising of the past participle been
across the adverb to a verbal head position between have and the adverb:
Although multiple merge along the lines in (241) rids us of violations of the
Head Movement Constraint as discussed in section 2.2.2, we are still left with
what seems to be optional raising of the past participle, i.e. been either stays
put, as is the case in (242b), or raises across the adverb, as in (242c). This
problem could be easily avoided if we assumed that there is an additional
merge site between the position assumed by the past participle been and the
lexical verb (cf. chapter 4.3.1.1 for detailed discussion):
Cinque (2004) conjectures that the positioning of adverbs in pre- and post-
subject position warrants an additional merge site. Basically, this additional
merge position must be higher than the Case-checking site of the subject
(DPsubj), as shown in (244) for frequently:99
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 95
If nothing moves, the linear order in (245a) obtains; (245b) is derived by the
subject raising across the adverb (cf. chapter 4.3.1.1 for detailed discussion):
As we have already seen, the derivation of (242c) involves the raising of the
past participle been across the adverb.
At first glance, it would seem that the specifier analysis loses much of its
original restrictiveness if we allow for multiple merge sites for functional
projections hosting adverbs, but this is definitely not the case since multiple
merge both respects and reflects the underlying hierarchy of functional
projections, with the functional projection hosting the frequency adverb being
universally ranked higher than the functional projection hosting the subject-
attitude adverb:
(246) a. She may have often been cleverly avoiding encounters with aliens.
b. * She may have cleverly been often avoiding encounters with aliens.
(248) She cleverly has often been (often) drinking pineapple juice.
Notice however that cleverly in (248a) precedes the finite non-lexical verb and
thus, as we shall see, is a representative member of the class of subject-oriented
adverbs. That is to say that the functional projection hosting subject-oriented
cleverly in (248) is well different from that hosting subject-attitude cleverly in
(246). Supportive evidence for this assumption derives from the fact that
subject-oriented adverbs and subject-attitude adverbs may co-occur, as in
96 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(249a), and that, if an aspectual adverb like often is present as in (249b), often
is flanked by the two types of subject-related adverbs:100
(249) a. She cleverly has been carefully answering their questions (stupidly).
b. She cleverly has often been carefully answering their questions
(stupidly).
Since Cinque (1999), as opposed to Frey & Pittner (1999), does not distinguish
between subject-attitude adverbs and subject-oriented adverbs and since his
discussion of English data is restricted to post-subject occurrences of subject-
related adverbs, which, in fact, are subject-oriented, we are safe to argue that
subject-attitude adverbs and subject-oriented adverbs are specifiers of distinct
functional projections with the former being ranked below and the latter being
ranked above the functional projection hosting frequency adverbs (cf. (246) vs.
(248) and (249b)). What these brief considerations show is that the assumption
of multiple merge sites for one and the same class of adverbs does not make
the specifier analysis less restrictive since each of the potential merge sites
must be in accordance with the overall hierarchy of functional projections
accommodating adverbs.
The argument that adverb stacking is preferred in cases in which any two
adverbs are separated by an overt verbal head can hardly be considered solid
evidence against the specifier analysis. Recall from section 2.2.2 that Costa
(2000:23) maintains that "the adjunction of adverbs is parasitic on the
lexicalization of the heads":
Rather, it would seem that the strings in (250) are instances of haplology or
Horror Aequi. That is to say that the strings under consideration are ill-formed
because the stacked elements are not distinct enough formally, i.e. they all end
in –ly.101 The assumption that haplology is at work is corroborated by the
following strings, where the stacked adverbs are formally distinct:
(251) a. Ms. San had seriously once thought of accepting her for a god-
daughter as she had no children of her own. (www)
b. It is now cleverly incorporated into the [...] design. (www)
c. Now that I'm in the prime (ahem!) of my life, I wisely no longer
assume that this will be my last computer. (www)
d. Although this particular safari company has fortunately since closed
down, there are still safari operators like this… (www)
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 97
Likewise, the fact that adverbs which clearly belong to different semantic
classes may occur in coordinate structures (cf. (252)) does not constitute
evidence against the specifier analysis since the two conjuncts may not be
adverbs, but larger chunks of structure,102 with any phonetic material below the
adverb being deleted under identity in the first conjunct:
(252) a. The king replied that she must keep her promise, so she slowly and
sadly opened the castle door. (www)
b. The current owner has recently and lovingly restored it into a holiday
home. (www)
c. She often and passionately spoke of her regard for the one room
school concept. (www)
d. They deliberately and repeatedly lied to us about 9/11 ... (www)
(253) a. … she [[slowly opened the door] [and [sadly opened the door]]]
b. … has [[recently restored it] [and [lovingly restored it]]]
An analysis along these lines would also account for the fact that otherwise
rigidly ordered adverbs (e.g. speaker-oriented adverbs) may occur in any order:
(255) a. Fortunately and honestly, this album does get better as it progresses.
b. Quite clearly, and fortunately, it has not.
c. 'The Crucible' probably and unfortunately is not going to be
overwhelmed with irrelevance too soon. (www-data)
(256) a. This has often and obviously been observed among […] parents.
b. This writer has recently and fortunately become the recipient of a
catalogue of the Obodda library… (www-data)
98 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
As has already been pointed out by Cinque (2004), the internal structure of
adverb(ial) clusters, as in (257) and (169) above, "can hardly be considered
anyone's exclusive problem" (cf. Cinque 2004:694, note 23):
(257) a. Letztes Jahr im Juni an einem Sonntag kurz vor Mittag rief er alle
an.
"Last year in June on a Sunday shortly before noon he phoned all up"
b. In der Küche neben dem Tisch auf dem Boden unter einem Tuch
fand er es.
"In the kitchen besides the table on the floor under a cloth he found it"
c. Abends wegen des Staus hat er diesen Platz gemieden.
"(In the) evening because of the (traffic) congestion has he this place
avoided"
d. Gestern im Hörsaal als der Vortrag begann hustete er wie verrückt.
"Yesterday, in the lecture room, when the lecture started, coughed he
like mad" (adapted from Cinque 2004:694)
While it follows from the V2 constraint that the adverbial clusters in (257) are
contained within a single constituent, it is far from clear how this constituent is
internally structured (cf. Haider 2000:114ff.). As has already been argued by
Steinitz (1969:47ff.), (257a) and (257b) are instances of subordination with the
lower adverbial expression, e.g. kurz vor Mittag 'shortly before noon' in (257a),
functioning as an attribute to the higher adverbial expression an einem Sonntag
'on a Sunday', which, in turn, functions as an attribute to im Juni 'in June',
which functions as an attribute to letztes Jahr 'last year'. Cinque (2004:694)
suggests that progressively specified temporal and spatial expressions as in
(257a) and (257b) be seen as merged as one constituent, which is then fronted
to the sentence-initial position. An analysis along these lines is clearly com-
patible with the specifier analysis, since the complex temporal or spatial
expression is licensed in its base position under specifier-head agreement, i.e.
within the projection of a functional head bearing temporal or spatial features.
Their being licensed in sentence-initial position involves peripheral features,
e.g. [TOPIC]. The cases illustrated in (257c) and (257d) are different since the
adverbial expressions are not semantically homogeneous: while abends 'in the
evening' in (257c) is a temporal expression, wegen des Staus 'because of the
traffic jam' is a causal expression. Likewise, gestern 'yesterday' and als der
Vortrag begann 'when the lecture began' in (257d) are temporal expressions,
whereas im Hörsaal 'in the lecture room' is a spatial expression. To account for
the fact that semantically diverse expressions in (257c) and (257d) are con-
tained in one constituent, Cinque (2004:694f.) suggests that the constituent
preceding the finite verb be viewed as a remnant verbal projection excluding
the trace of the finite verb.103 An analysis along these lines requires that each
semantic type of adverbial expression be independently licensed under speci-
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 99
fier-head agreement with a suitable functional head and that the preposed rem-
nant be licensed under specifier-head agreement with a functional head bearing
relevant peripheral features, e.g. [TOPIC] (cf. chapters 4.3.4.3 and 5.2.2 for
discussion). In general, the argument that the specifier analysis fails to cope
with adverbial clusters (cf. Haider 2000, 2004; Ernst 2002) is not convincing.
First, the criticism only relates to the external structure, i.e. the specifier status
of adverbial clusters, and second, the criticism is not accompanied by an analy-
sis of the internal structure of adverbial clusters.
Likewise, the conjecture that so-called Edge Effects can be captured under
the adjunction analysis (cf. Haider 2000:100; 2004:783), but not under the
specifier analysis is not entirely convincing. First, as Haider (2004) points out
himself, "the 'edge effect' is caused by (not yet fully understood) properties of
head initial structures" (cf. Haider 2004:782). Second, evaluative adverbs
constitute a well-known case of preverbal adverbs taking genuine post-head
complements in head-initial structures, as shown in (258).104 Third, as is
pointed out by Cinque (2004:695, note 24), in light of strings as in (259), the
generalization that preverbal adverbial projections must be head-final seems
too strong:
(258) a. [Luckily [for all of us]], it's perfectly within the law. (BNC)
b. Abrasive, but not arrogant, there is no edge to him but, [surprisingly
[for the owner of the Sun]], he has a puritanical side. (BNC)
As we have seen in the previous section, the treatment of manner adverbs is not
uniform across analyses. Thus, for example, Frey & Pittner's (1989, 1999)
analysis of manner adverbs as right-adjoined to the verb or its trace is highly
problematic since it involves XP adjunction to heads, as shown in (260b):
(260) a. John has spoken (extremely nicely) to his mother (extremely nicely)
about the book extremely nicely.
b. John has [[spoken1 extremely nicely] [to his mother [[t1 extremely
nicely] [about the book [t1 extremely nicely]]]]]
(adapted from Frey & Pittner 1999:18f.)
Moreover, it is not clear how right-adjunction would work with passives since
manner adverbs may precede the passive participle of the lexical verb:
100 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Similar problems arise under Haider's (2000, 2004) analysis. First, it is not
immediately clear how English manner adverbs in actives can be postverbal
without being realized in the so-called extraposition domain (i.e. the
semantically empty verbal structure below the lexical verb), unless there is
overt verb movement (recall that Haider 2000, 2004 refutes 'extraposition').
That is to say that we have to assume that manner adverbs are left-adjoined to
VP (yielding V') with the verb obligatorily raising across the manner adverb:
For passives, we would have to allow for the verb to either raise or stay put. In
the latter case, AdvP must be head-final ((263a, b) from www):
(263) a. The rebuke was so gently given that Arthur hardly coloured under it.
b. His nails were too nicely trimmed to belong to a man…
c. * The rebuke was so gently that Arthur hardly coloured under it given.
d. * His nails were too nicely to belong to a man trimmed.
Irrespective of genus verbi, Ernst's (2002, 2004) system provides two positions
for manner adverbs: they may be right-adjoined to VP or left-adjoined to PredP
(cf. Ernst 2002:269ff.), as illustrated in (264).105 Recall from above that pre-
verbal occurrences of carefully etc. are interpreted as subject-attitude adverbs,
not as manner adverbs:
(264) a. They have [PredP [Adv carefully] [PredP [Pred searchedi] [VP [VP ti his room]
[Adv carefully]]]]
b. His roomj has been [PredP [Adv carefully] [PredP [Pred searchedi] [VP [VP ti
tj] [Adv carefully]]]]
Ackerman 2001 for related arguments). Finally, poorly in (265c) is barred from
occurring preverbally since build is "an action verb with a volitional agent and
an individuated, affected object and a well-defined end-point" (cf. Ernst
2002:275) and thus ranges relatively high on the transitivity scale:106
(267) a. John has spoken ti tj nicely [to his mother]i [about the book]j.
b. John has spoken to his mother tj nicely [about the book]j.
However, as we have seen in connection with (160), (177) and (224), extra-
position is not an option since extraposed constituents are extraction islands
and the allegedly extraposed PPs in (267) are not:
Abstracting away for the moment from the phrase structure clashes (cf. (260b))
that arise under Frey & Pittner's (1998, 1999) analysis, their analysis needs to
be kitted out with some version of Stowell's (1981: 110ff.) Adjacency Condi-
102 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
At this point, it would seem that neither the specifier analysis (cf. section 2.2.2
for discussion) nor the adjunction analysis can handle the data under consid-
eration. As will be seen in chapter 3, supplemented with a modified account of
the internal structure of VP incorporating overt object shift, the specifier analy-
sis captures the data adequately without having to recur to stipulations along
the lines 'adjunction to agreement-related projections is prohibited' (cf.
Koizumi 1993:110; Runner 1995a: 94 for arguments along these lines).
2.3.3 Summary
This section reviewed the adjunction-based semantic scope analyses proposed
by Frey & Pittner (1998, 1999), Haider (2000, 2004) and Ernst (2000, 2002,
2004), according to which the positioning and ordering restrictions on adverbs
(ideally and exclusively) derive from independent semantic scope principles.
The analyses under consideration were shown to differ with respect to the
extent to which syntax has a word in determining semantics by defining adver-
bial base-positions (cf. especially Frey & Pittner vs. Haider), the X-bar status
of the adjunction site (cf. Frey & Pittner and Ernst vs. Haider), the direction of
adjunction (cf. Frey & Pittner and Ernst vs. Haider) and the structural integra-
tion of postverbal adjuncts (cf. Frey & Pittner and Haider vs. Ernst). The rela-
tive canonical order of adverbs follows from their respective scope require-
ments with the relative order of adverbs being relatively free within any given
domain, e.g. proposition-, event- or process-related adverbs. Section 2.3.2 read-
dressed multiple and alternate adverb positioning, the linear orders of adverbs
and the verb's internal argument(s), adverb stacking, coordination and cluster-
ing and so-called Edge Effects from a semantic scope perspective. It was
shown that the analyses under consideration do not fare equally well with the
critical data. Notwithstanding the empirical and theoretical problems adhering
to individual analyses, I have argued that adjunction-based semantic scope
analyses as such are not as compelling as they may seem at first glance. More-
over, I sketched certain adjustments to the specifier analysis (e.g. multiple
merge and specifier-head agreement by proxy, to be fleshed out and tested on a
vast empirical base in chapters 3 – 5) which allow us to account for the critical
data.
2.4 Conclusion
Starting out with a critical review of the major proposals concerning the
integration and licensing of adverbs within the Theory of Government and
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 103
Binding and the Theory of Principles and Parameters (section 2.1), the
Minimalist Program and the Antisymmetry of Syntax (section 2.2) and Seman-
tic Scope Theories (section 2.3), this chapter not only highlighted the chal-
lenges adverbs pose for syntactic and/or semantic theory, but also introduced to
the ongoing debate between proponents of the specifier analysis and semantic
scope theorists and set the scene for the analysis of adverbs presented in chap-
ters 3 to 5. The debate of whether syntax determines semantics or vice versa,
which has come to be inseparable from the question of whether adverbs should
be integrated into structure as specifiers or as adjuncts, was illustrated mainly
on the basis of the catalogue of empirical phenomena which scope theorists
take to invalidate the functional specifier analysis (sections 2.2.2 & 2.3.2).
Throughout this chapter, the focus was on the specific theoretical and empirical
problems of the analyses under consideration. Section 2.3.2 reviewed and
tested the predications made by proponents of both the specifier analysis and
semantic scope theories. A close look at the critical data revealed the follow-
ing: First, postverbal occurrences of adverbs (unless they are postverbal
because the lexical verb has moved across them) and reverse orders of post-
verbal adverbs can only be accounted for under the assumption that they are
merged into structure as specifiers and/or complements within a semantically
empty VP which is projected below the lexical verb (cf. Frey & Pittner 1998,
1999; Haider (2000, 2002), i.e. both right-adjunction as under Ernst's (2002)
analysis and snowballing are not eligible options. An analysis along these lines
was argued to be fully compatible with the specifier analysis: adverbs that are
contained within VP are licensed from within the interpretable part of the
clause by forming representational chains with their null counterparts, which
are licensed under specifier-head agreement within designated functional pro-
jections. Second, to account for multiple and interverbal adverb positions
and for alternate linear orders of preverbal adverbs reference must be made
to finer grained semantic types, i.e. distinctions in terms of 'proposition-
related', 'event-related' and 'process-related' do not suffice, since the relative
linear orders of adverb classes constituting these semantic types are far more
restricted than is claimed by scope theorists, in fact as restricted as predicted by
the specifier analysis, if supplemented with finer grained distinctions (e.g. Frey
& Pittner's 1998, 1999 distinction between subject-oriented and subject-attitude
adverbs). Third, semantic scope analyses neither offer a solution to the problem
of adverb stacking, coordination and clustering nor to the question of which
factors give rise to Edge Effects. That is to say that the empirical problems that
are adduced as evidence against the specifier analysis are not solved. More-
over, it could be shown that adverb stacking and adverb coordination do not
invalidate the specifier analysis, but the semantic scope analysis. Finally, the
positioning of manner adverbs relative to the verb's internal argument(s)
cannot be accounted for without either admitting extraposition of the verb's
internal argument(s) (cf. Haider 2000, 2004; Ernst 2002, 2004) or phrase
104 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
structure clashes (cf. Frey & Pittner 1998, 1999). Even though the solution of
this problem under the specifier analysis is still outstanding, we conclude at
this point that the arguments against the specifier analysis and in favor of the
adjunction-based semantic scope analysis are not compelling. On the whole,
semantic scope theories do not excel with respect to the called critical data
because they are not restrictive enough, as can already be seen from the fact
that they are, to varying extents, supplemented by structural conditions (cf.
Frey & Pittner 1998, 1999; Ernst 2000, 2002, 2004).
The following chapters are devoted to the detailed analysis of the syntax of
an extensive range of adverb classes in their relation to the three layers of the
clause, i.e. the lexical layer, the inflectional layer and the complementizer layer
(cf. Rizzi 1997, 2002). I present a wide array of empirical and theoretical evi-
dence in support of the claim that the licensing of adverbs within and across the
three layers of the clause is contingent on a strict one-to-one relationship
between a (given member of a) given syntactic adverb class and a designated
clausal head.
Notes
1
In the following, the Theory of Government and Binding (GB), Theory of Principles and
Parameters (PPT) are jointly referred to as GB/PPT since these two models, despite the differ-
ences that exist between them, share a multitude of general assumptions about the interaction
of the lexical specification of lexical items and the computational system.
2
In this context, the question arises whether adverbs are present at d-structure or inserted in the
course of the derivation by means of adjunction (cf. Lebeaux 1988).
3
That is to say that since adverbs are not considered thematic arguments, they are barred from
assuming the specifier- or complement-position within lexical projections (but see McConnell-
Ginet 1982; Larson 1985, 1988, 1990; Travis 1988 for divergent analyses), and since they are
not associated with Φ-features), i.e. grammatical features such as person, number, gender and
case (cf. Chomsky 1986b:24), they are barred from assuming the specifier- or complement-
position within functional projections. The qualification that adverb(ial)s are barred from
assuming the specifier position only holds with respect to base-generation, i.e. wh-raised ele-
ments (how often, when), clearly assume the specifier position within CP. See Chomsky
(1981:330) for the conjecture that [WH] be considered a Φ-feature. )
4
As Roberts (1987:74) correctly observes, the Adjacency Condition on Case Assignment is too
strong with respect to the Case-relation between Infl and the subject:
(i) It evidently must be raining.
5
As is pointed out by Laenzlinger (1998:69), frequency adverbs constitute another problem for
modification-based approaches since they modify neither V nor IP directly.
6
Cf. Emonds (1976:3, 153ff., 164ff.), Chomsky (1986b:81ff.; 1995:253). See also Laenzlinger
(1998:67f.) for criticism.
7
Notice that Zubizarreta's (1982) proposal is compatible with Higginbotham's (1985) analysis
of modifiers in terms of Θ-identification and Williams' (1987a, b; 1989; 1994) account of
thematic adjuncts (see especially Williams 1994:22ff., 166ff.).
8
Under Jackendoff's (1972:58ff.) analysis, strings as in (26) are ruled out on the basis of the
prohibition against multiple occurrences of subject-oriented adverbs.
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 105
9
This flaw in Zubizarreta's (1982) account could easily be amended by modifying the Adjunct
Θ-Criterion to the extent that an argument Θ-role must not be combined with more than one
adjunct Θ-role of a given type. This modification would account for the ungrammatical exam-
ples in (26), where the Agent Θ-role assigned by the verb climb is combined with two identical
adjunct Θ-roles, as well as for the grammatical examples in (25), where two distinct adjunct Θ-
roles are assigned to the (implicit) agent.
10
In order for (29) to apply, the projection of X must not be Θ-marked and it must be linked to
its subject under mutual c-command (cf. Rothstein 1985:11f.).
11
See Roberts (1987:72, 83) for secondary agentivity in passives.
12
Recall that VP-internal adverbs are barred from intervening between the verb and its internal
nominal argument.
13
Cf. (32) for frequency adverbs and (33) & (36) for speaker-oriented adverbs.
14
See also Jackendoff (1977:78) for arguments that adverbs do not take complements.
However, under his analysis, the (in)ability of heads to take arguments does not affect their X-
bar status.
15
Type Ia adverbs are licensed by the [AGR]-feature in I, whereas Type IIa and Type III ad-
verbs are licensed by the [E]-feature in I.
16
With respect to speaker-oriented adverbs, Travis (1988:299f.) conjectures that these could
be assumed to be licensed by a discourse feature in the head of matrix CPs.
17
Interestingly, Travis (1988:288, note 4, 300) concedes that adverbs seem to take comple-
ments ((i) & (ii) quoted from Travis 1988:287f.):
(i) Because they behave differently from other categories, adverbs are not considered
major lexical categories.
(ii) [...] heads that do not project must be licensed differently from other maximal
projections.
18
See also Stroik (1990) and Alexiadou (1997). Larson's (1985, 1988, 1990) analysis extends
to temporal, spatial and directional bare-NP adverbs and obliques like on Thursday.
19
Although VP-external adverbs are dominated by S (i) and behave like S-adverbs in that they
may precede passive be ((48b) & (ii)), they differ from S-adverbs in that they may assume the
sentence-final-position without intonational break (iii) vs. (iv) (examples adapted from
McConnell-Ginet 1982:147f.):
(i) [S [NP Mary] [Adv reluctantly] [Aux was] [VP instructed by Joan]]
(ii) Mary {probably, allegedly, apparently, unfortunately} was instructed by Joan.
(iii) Joan instructed Mary {wisely, unwillingly, obediently, knowingly}.
(iv) * Joan instructed Mary {probably, allegedly, apparently, unfortunately}.
20
These two interpretations are structurally reflected in VP-external vs. VP-internal attach-
ment of the adverb ((i) & (ii) adapted from McConnell-Ginet 1982:160):
(i) [S [NP Louisa] [Adv rudely] [VP departed]]
(ii) [S [NP Louisa] [VP [V departed] [Adv rudely]]]
21
Under this analysis, intransitive verbs like depart necessarily belong to two logical catego-
ries: the category of intransitive verbs, i.e. IV, and the category of intransitive verbs with an
ad-verb argument, i.e. IV/AD-V. Independent evidence for the latter category derives from
verb subcategorization facts (cf. McConnell-Ginet 1982:164; see also Jackendoff 1972:64ff.):
(i) Joan behaved *(rudely) to Marcia.
22
See also Larson (1988:345f., note 11, 384, note 49; 1990:621ff. ).
106 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
23
Under this analysis, the argument structures of verbs must contain adverbial Θ-roles which
are assigned only optionally. This assumption, however, runs counter to the requirements im-
posed by the Θ-Criterion (cf. Chomsky 1981:36; 1982:6).
24
Under Roberts's (1987) analysis the ill-formedness of (66b) and (66e) would follow from
selection.
25
Recall that Travis' (1988) analysis allows for adverbs to combine with heads.
26
In the wake of Pollock's (1989) Split-Infl Hypothesis according to which the two
morphosyntactic features residing in the head position of IP, i.e. [AGREEMENT] and [TENSE],
are assigned the status of autonomous functional heads, Agr and T, that project onto a phrasal
level, AgrP and TP, it has been suggested that the clause be viewed as constituted by a cascade
of functional projections, e.g. AgrSP (subject agreement), AgrOP (object agreement), AspP
(aspect), MoodP, Neg(ation)P, Pred(icate)P, VoiceP, Pass(ive)P, etc. (see Laka 1990; Belletti
1990, 1994; Ouhalla 1990, 1991; Haegeman & Zanuttini 1991; Sportiche 1992; Bowers 1993;
Roberts 1993; Alexiadou 1997; Cinque 1999 among others). Depending on analysis, the func-
tional projections are seen as rigidly ordered across languages (Cinque 1999), or the order of
the functional projections is seen as being subject to parametric variation (Ouhalla 1990, 1991).
27
Chomsky (1993:32) assumes that overt movement is triggered by greed, which is
characterized as an operation "[...] driven by morphological necessity: certain features must be
checked in the checking domain of the head, or the derivation will crash." Chomsky
(1995:261) interprets greed as Last Resort.
28
The term EPP feature derives from the Extended Projection Principle of GB/PTT, which
requires that sentences have subjects. In MP, this requirement is generalized and translates as
"the specifier position within the projection of the head they occur on be lexically filled". This
requirement is the lowest common denominator of the rather heterogeneous collection of fea-
tures.
29
Cf. Chomsky (1993:32; 1995:289ff.). See Baker (1970) for the introduction of the question
morpheme Q and the Q-universal. Notice that [Q] is taken to be a morphosyntactic feature (cf.
Chomsky 1995:289ff.; Rizzi 1996:66, 84f., note 3). Recall that Chomsky (1981:330)
conjectures that wh- could also be considered a Φ-feature). Note also that the set of criterial
features is not homogeneous in the sense that [WH/Q], [FOC] and [NEG], but not [TOP], are quan-
tificational.
30
See Chomsky (1995:232ff.) for the qualification that if "F [a given feature] is strong, then F
is a feature of a nonsubstantive [ = functional] category and F is checked by a categorial fea-
ture" [DH].
31
Raising operations, be they overt or covert, must obey economy constraints such as the
Principle of Shortest Movement and the Minimal Link Condition (cf. Chomsky 1993:14ff.;
1995:264ff., 267ff., passim).
32
This observation, obviously, does not extend to adverbs bearing operator features.
33
The distinction between category and segment, which goes back to May (1985), has been
adopted as follows (taken from Chomsky 1986b:7):
(i) [In] a structure of the form [ [β α [β…]] ], a typical adjunction structure with α adjoined
to β, α is not dominated by the category β; rather β consists of two "segments," and
category is dominated by β only if it is dominated by both of these segments. [DH]
34
I use the label Aux/AuxP to distinguish auxiliary verbs and their projections from lexical
verbs and their projections.
35
However, there is not an exact match of specifiers and A-positions and adjuncts and A'-
positions, since, for example, operator movement, e.g. wh-movement, as sketched in (i), is
unanimously conceived of as A'-movement (cf. Chomsky 1995:248):
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 107
(i) [CP [DP which book]i [C' [C [will]j [C Q]] [TP [D you]k [T' [T tj] [VP [D tk ] [V' [V buy] [DP
ti]]]]]]]
Here, [DP which book] raises from its base-position, i.e. the complement position of the verb
buy, to specCP in order for the derivation not to crash at PF, i.e. movement is forced by the
requirement that the strong Q-feature in C be eliminated before the derivation reaches spell-
out.
36
See Zagona (1988) for an early version of the specifier analysis. Adapting the main verb
analysis of auxiliary verbs (cf. Ross 1967; Kayne 1975; Emonds 1976), Zagona (1988)
suggests that adverbs occupy left- and/or right-branching specifier positions within VP:
(i) [VP1 [AdvP ] [V' [V ] [VP2 [AdvP ] [V' [V ] [VP3 [AdvP ] [V' [V ]] ([AdvP ])]3 ] ([AdvP ])]2 ] ([AdvP ])]1
Specifically, she argues that temporal and aspectual adverbs be analyzed as right-branching
specifier within the projections of both lexical and auxiliary verbs where they are licensed in
terms of selection and feature compatibility (an early version of specifier-head agreement), e.g.
temporal and aspectual features (cf. section 2.2.1).
37
According to Laenzlinger (1996:112f.), evidence for right-branching specifiers derives from
right-peripheral occurrences of non-argumental adverbs, such as discretely, often and yesterday
in (i) – (iii), which take scope over the projections they are associated with (VP, AspP and TP,
respectively), as well as from the presence or absence of Relativized Minimality effects
(adapted from Laenzlinger 1996:112):
(i) John … [TP [T' [T ] [AspP [Asp' [Asp ] [VP [V' looked at Mary]]]] [Adv yesterday]]
(ii) John … [TP [T' [T ] [AspP [Asp' [Asp ] [VP [V' met Mary]] [Adv often]]]]
(iii) John … [TP [T' [T ] [AspP [Asp' [Asp ] [VP [V' looked at Mary] [Adv discretely]]]]]
38
See also Lewis (1975) for a thorough discussion of quantificational adverbs.
39
This generalization carries over to allegedly right-branching specifiers ((i) from Laenzlinger
1996:113):
(i) * Combieni as- tu lu [ ti de livres de Ch.] {vraiment, énormément,
how muchi have-you read ti of books of Ch. {really a lot,
vraiment souvent}?
really often}?
"How many of Chomsky's books did you read {really a lot, really often}?"
40
However, as Wilder (1999:697) points out, the claim that quantifier adverbs may not move
after feature checking, is too strong since aspectual adverbs as in (i) – (iii) may be questioned,
topicalized and focalized after having checked off their features in the position of the trace ((i)
– (iii) from Wilder 1999:698):
(i) [[AdvP How often]+wh did John [tAdvP+A'-feat F0 [meet Sue]]?
(ii) [[AdvP How often]+Top John doesn't [tAdvP F0 [meet Sue]].
(iii) It was [AdvP seldom]+Foc that John [tAdvP F0 [met Sue]].
41
Notice that in (89) above, asymmetric c-command holds between the nonterminal members
of the pairs <J, M> and <J, P>. Thus, the linear ordering of terminals, d(A), is <j, m>, <j, p>.
However, since none of the nonterminal elements, i.e. M and P, asymmetrically c-commands
the other, the linear ordering of the terminal elements m and p cannot be established, whence
the structure in (89) fails the LCA.
42
Exclusion is defined as follows (cf. Kayne 1994:133, note 1; Chomsky 1986b: 9):
(i) X excludes Y if no segment of X dominates Y
43
Note that since structure necessarily is always binary branching (see also Kayne 1984a),
there may be at most two non-heads per phrase, one specifier and one complement, with the
specifier occupying a left-peripheral position which asymmetrically c-commands all nontermi-
108 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
nal categories to its right. Within this system, right-adjunction is impossible (cf. Kayne
1994:78, 117ff. for discussion).
44
Covert heads not only make available additional specifier positions but also serve as the
targets for head movement. Kayne (1994:30) maintains that "for every moved phrase, [there
must] be a distinct head to whose projection it can adjoin as specifier." Moreover, he points out
that heads have one of two qualities: they are either intrinsically contentful, e.g. lexical heads
and functional heads like T(ense) and Asp(ect), or they lack intrinsic content. Kayne (1994:30)
conjectures that AgrO instantiates the latter type of head, which he takes to be "imposed upon
phrase markers by the paucity of available adjunction sites, with this paucity following from
the present theory."
45
The example Kayne (1994:75) discusses involves leftward movement of AdvP to the
specifier of some higher VP-shell, which is triggered by the requirement that the "heavy NP"
all his old linguistics books be in final position:
(i) John reread carefully all his old linguistics books.
(ii) ... reread [[carefully i] [X0 [all ... books] [Y0 [[e] i ...
X0 corresponds to V2 and Y0 to V1 in (98). Notice that the verb reread has moved into the head
position of a yet higher VP-shell than the one which hosts the moved AdvP.
46
While never in (99b) is clearly moved from within IP, it is not clear whether yesterday in
(99a) is taken to be base-adjoined to XP or moved from within the VP, i.e. from the
complement position of the lowest V in a VP-shell (cf. (98) above, see also Larson 1988:384,
note 49).
47
The adverbs in (101) – (103) are representatives of distributional classes, e.g. frankly is a
token of the class of speech act or illocutionary adverbs also containing honestly, bluntly, con-
fidently, flatly, prosaically, sincerely, truly, etc. (cf. Greenbaum 1969:91f.) and well is a token
of the class of manner adverbs (cf. Cinque 1999, chapter 1).
48
The set of functional heads that figures in the overall architecture of the clause is
heterogeneous as it comprises both free and bound lexical items, i.e. clausal heads may be
realized as non-closing suffixes, inflectional suffixes, auxiliaries and functional particles.
Under the Mirror Principle (Baker 1985:375), the linear orders of suffixes translate into
syntactic structure.:
(i) Morphological derivations must directly reflect syntactic derivations (and vice versa).
49
As Cinque himself (1999:106) points out, the hierarchy of functional projections is "out-
rageously rich", but given that the specifier positions of these functional projections host
AdvPs "the richness should not appear so outrageous."
50
The concept of parametric variation is rejected since the phenomena that are customarily
attributed to parametric variation are captured as spell out options (cf. Cinque 1999:127, 136f.).
51
Laenzlinger (2000:113f.) proposes a modification of Cinque's (1999) analysis which rests on
the assumption that sentences, in order to be interpretable, must have the minimal internal
structure in (i) (agreement related projections omitted) with MoodP, ModP and AspP(high) and
AspP(low) being root categories that are associated with functional features which, "in the
absence of any semantically related contentful material" (Laenzlinger 2000:113) receive a
default interpretation along the lines proposed by Cinque (1999:128ff.):
(i) MoodP > ModP > TP > AspP(high) > AspP(low) > vP > VP
In cases in which the numeration contains an adverb, the functional features associated with the
root categories will be spelt out as discrete functional projection, e.g. if the numeration con-
tains a speech act adverb and evidential adverb, Moodspeech act and Moodevidential will be pro-
jected. This conception of (non)projected structure is similar to that advanced by Rizzi (1997,
2002) with respect to the left periphery (cf. chapter 5.1) in the sense that the functional projec-
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 109
tions which are always present may split under the condition that there be lexical material
present. This condition, obviously, can be met under both Merge and Move.
52
Marked values may also be assigned on the basis of morphological marking, e.g. [+PAST] vs.
[–PAST] or [+PASSIVE] vs. [–PASSIVE].
53
In this context, note that higher sentence adverbs resist wh-movement (cf. Chomsky (1986b,
1995b), Lonzi (1991), Kayne (1994), Cinque (1999)):
(i) * How { probably / courageously} has he eaten the mushrooms?
54
Cinque (1999:8) points out that virtually any VP-adverb can be in focus position provided
that it is heavy enough (with heaviness being achieved by modification, coordination or by
nuclear stress (cf. (109b) – (109d)). In addition to syntactic and prosodic means of rendering
some AdvP heavy, there are adverbs which have a morphologically heavier variant (cf. Cinque
1999:14).
55
Cinque (1999:21f.) points out that the relative scope properties of the adverbs are not
affected, i.e. più has scope over sempre in both in (111a) and in (111b), because scope is
calculated under reconstruction.
56
If more than one VP-adverb occurs in VP-final position, they are in canonical order (cf.
Cinque 1999:14):
(i) Gianni non vince le sue partite già più sempre BENE/*già sempre più BENE/* sempre già
più BENE.
"Gianni does not win his matches already any longer always well."
57
Cf. McConnell-Ginet (1982), Larson (1985, 1988, 1990), Stroik (1990) and Kayne
(1994:75) for related arguments. See also Laenzlinger (1996:40f., 66ff. passim) for the
conjecture that certain types of AdvP, e.g. temporal, spatial and manner AdvPs, have the status
of arguments and thus are generated in complement position.
58
Moreover, Alexiadou (1997) conjectures that the complement analysis of completion ad-
verbs neatly accounts for their occurring in VP-final position (but see chapter 3.2.2).
59
Cf. Kratzer (1988), Enç (1991), Stroik (1992) and Rizzi (1990) for discussion.
60
Alexiadou's (1997) analysis makes the wrong prediction that e.g. preverbal carefully is a
manner adverb (cf. chapters 3.2.1.1 & 4.3.1):
(i) Bob has carefully removed the lid.
61
Notice that, since Alexiadou's (1997:135) analysis revives Chomsky's (1965:103ff.) proposal
to relate the availability of passivization to the presence of a manner feature, [MANNER] cannot
be a purely semantic feature.
62
Internally complex AdvPs may undergo movement if they are affective operators, such as
wh-elements or negative elements (Bowers 1975:561).
63
Notice that, on closer inspection, Alexiadou's (1997:141) claim that "only bare manner ad-
verbs can appear in the specifier of VoiceP" does not hold:
(i) The message was [{extremely, very} secretly]i sent ti.
(ii) The message was [so secretly]i sent ti that even the CIA didn't find out.
64
See also Cinque 1999:8, 14 for discussion.
65
Alexiadou (1997:142) maintains that unstressed single-term manner adverbs are inherently
specified as [–FOC(US)]. See also Ernst (2002:272) for an information structure based account
of adverb placement.
66
See Rivero (1992) and Alexiadou (1997:177ff.) for a detailed discussion. But see
Smirniotopoulos & Joseph (1998:458ff., 480ff.) for arguments against argument incorporation
in Modern Greek. Smirniotopoulos & Joseph (1998) present a lexical compound analysis of the
data discussed by Rivero (1992).
110 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
67
Alternatively, we could assume that the incorporated adverb is licensed in the head position
of VoiceP which is overtly targeted by the complex lexical verb [V Adv V].
68
See also Zagona (1988:32, 65ff.). Cf. Larson (1985) for arguments that temporal bare-NP
adverbs, e.g. yesterday, tomorrow, today, that day, Sunday, now, then, etc., are lexically
marked as [+temporal].
69
Notice, incidentally, that, at least with respect to English, the analysis presented is not
consistent with Chomsky's (1995:232) analysis, according to which specTP is targeted by the
subject DP (cf. Alexiadou 1997:118ff.; Haumann 1997:125ff. for discussion).
70
Even if temporal adverbs could incorporate into the lexical verb, they would not be licensed
since the relevant temporal features reside in T, not in V.
71
Although widely discussed by opponents of the specifier analysis, we will ignore cases in
(113) & (114), where multiple occurrences coincide with clear interpretational differences.
72
Sentence-final occurrences are irrelevant here since they coincide with different interpre-
tations.
73
See Bobalijk 1999:27, Svenonius 2002:203ff., 209ff. and Ernst 2002:116ff. for criticism.
74
Cases as in (i) & (ii) and (iii) & (iv) will be ignored here since postverbal often and again in
(i) & (ii) differ in interpretation from their preverbal counterparts in (iii) & (iv), i.e the linear
orders cleverly > often and loudly > again constitute only seemingly reversed orders since
again and often are merged in position quite distinct from that assumed by their preverbal
counterparts:
(i) Martians have cleverly solved these equations often. → Situations in which Martians
have cleverly solved these equations are many.
(ii) They read her messages carefully again. → They re-read her messages carefully.
(iii) Martians often have cleverly solved these equations. → Most Martians have cleverly
solved these equations.
(iv) They again read her messages carefully. → As on previous occasions, they read her
messages carefully.
Likewise, reversed linear orders that result from wh-raising a lower adverb across a higher
adverb are insubstantial here:
(v) [How quickly]i will they usually ti respond to calls? (www)
(vi) I was thinking, [how long]i have they already ti lived together… (www)
75
Cf. Aboh (1997; 2001), Pearson (1999), Laenzlinger (2000, 2004), Rackowski & Travis
(2001). See also Koster (1999a, b, 2000), Shlonsky (2000) for generalized pied-piping analy-
ses, and Cinque (1999) for intraposition.
76
It would be interesting to see whether (postverbal) adverbs in Malagasy may be modified. If
so, the head analysis would have to be discarded on empirical grounds as well.
77
See also Ernst (2002:135ff.) for discussion and criticism.
78
Cf. Haider (2000:114ff.) for a detailed discussion.
79
See also Williams (1982:160), Emonds (1976:19), van Riemsdijk (1983:228ff.) and Haider
(2004:782f.).
80
See Chomsky (1995) for the qualification that "we still have no good phrase structure theory
for simple matters as attributive adjectives, relative clauses, and adjuncts of many types"
(Chomsky 1995:382, note 22).
81
Recall that, under Kayne's (1994) proposal, adverbs may also be merged as innermost
complements within VP.
82
In this context, 'raise to spec' does not extend to adverb raising to the left periphery, which,
of course, is available under Cinque's (1999) 'merge in spec' analysis, but relates to Alexiadou's
THE STRUCTURAL INTEGRATION AND LICENSING OF ADVERBS 111
(1997) analysis of e.g. manner adverbs as complements which may or may not raise to
specVoiceP.
83
Frey & Pittner (1999:26ff.) argue that adverbials as in (185a) and (186) are base-generated
in their surface position since they help ameliorate That-Trace Effects which arise if the
embedded subject raises across the overt complementizer that (cf. Browning 1996:252, 254;
Ernst 2002:407ff. for related arguments):
(i) *Whoi do you think that ti took it easy?
Strings of this type improve if a temporal, frame or reason adverb(ial) intervenes between that
and ti:
(ii) Whoi do you think that {yesterday, in the US, for this reason} ti refused to sing the
Marseillaise?
'Topicalized' material, i.e. genuine topics (iii) and preposed spatial, instrumental or benefactive
adverb(ial)s (iv), cannot salvage the strings under consideration:
(iii) * Who did you say that, presents, stole?
(iv) * Whoi do you think that {in the attic, with her computer, for me} wrote a poem?
84
The spatial expression in (186a) is seen to differ from that in (185b) in that it identifies the
circumstances under which Ben is an absolute dictator rather than providing a mere spatial
setting.
85
In this context, Frey & Pittner (1999:38) and Frey (2000:130f.) refer to Williams' (1986)
scope markers.
86
Haider (2000) states that, syntactically, "there are no reserved parking slots for an adverbial.
It may be parked wherever this is not explicitly forbidden by syntactic structure constraints"
(Haider 2000:96). This assumption allows for discarding Frey & Pittner's (1998, 1999)
proposal according to which the base positions for adverb(ial) classes are determined by struc-
tural conditions (cf. (173)) since their approach, in very much the same way as Cinque's
(1999), constitutes an "unsolved chicken and egg problem" with respect to interfacing: it is not
clear whether the actual position of an adverb(ial) is determined exclusively by autonomous
structural constraints or "just a function of a semantically geared choice out of the potential
positions in the syntactic structure of an expression" (Haider 2000:129).
87
Haider (2000, 2004) refers to the domain under consideration as the 'extraposition domain'.
This label is a misnomer since postverbal adverb(ial)s are not extraposed, but base-generated in
that domain (cf. Haider 2004:798).
88
Cf. Ernst (2000:37; 2002:58; 2004a:761ff.)) for detailed discussion.
89
If a speaker-oriented adverb, e.g. honestly in (i), allows for the manner reading, its "lexical
entry specifically mentions the second possibility for FEO type". That is, honestly may com-
bine with either a proposition (203c) or with a Specified Event ((i) from Ernst 2002:70). In the
latter case, the manner reading is derivative:
(i) We've been dealing with them honestly.
90
According to Ernst (2002), the positional versatility of functional adverbs is indicative of
their being licensable in any projection as long as their respective scope requirements can be
met.
91
Notice that, under Ernst's (2002) proposal, Edge Effects are a corollary of Weight Theory.
The 'endweight template' of Weight Theory is seen as a trigger for rightward movement, a.k.a.
extraposition (cf. Ernst 2002:228ff. for detailed discussion).
92
For expository ease I chose to label all verbal projections as VP.
93
This proposal is clearly inspired by analyses as different as Frey & Pittner's (1999) semantic
scope analysis, Williams' (1986) scope marker analysis and also Haegeman's (1995) discussion
of representational chains.
112 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
94
See Cinque (2004:699f.) for arguments that, say, durational adverbs, e.g. long, and dur-
ational prepositional expressions, e.g. for a long time, compete for the same specifier position,
with sentence-final occurrences of durational prepositional expressions resulting from intrapo-
sition of the VP remnant containing [DP a long time], but not the preposition for, which is
merged above VP, across the preverbal PP (cf. chapter 4.3.4.2 for discussion).
95
Frequently and occasionally are proposition-related and thus must outscope event-internal
adverbs, e.g. wisely and stupidly.
96
What Cinque (2004) clearly overlooks is the fact that the lower occurrence of frequently can
be interpreted as marking lower aspect only in passives. In actives, as in (i), the pattern in (239)
is not admissible, i.e. adverbs marking lower frequentative aspect must follow the active lexi-
cal verb as in (ii). Cf. chapter 4.3.3.1 for discussion:
(i) * The publishers frequently have suddenly (being) frequently rejecting her…
(ii) The publishers frequently have suddenly (being) rejecting her …. frequently.
97
As we shall see below, (235b, c) are also subject to rigid ordering.
98
However, as we shall see in later chapters, pre- and post-subject aspectual adverbs
frequently, as opposed to interverbal aspectual adverbs, are discourse prominent.
99
Depending on analysis, (245a) could be seen as an instance of adverb raising. Cf. chapter 5
for discussion.
100
Postverbal stupidly prevents preverbal carefully from being misinterpreted as a manner ad-
verb.
101
Cf. Rohdenburg (2003:236), Menn & MacWhinney (1984a, b), McCawley (1988a:305ff.).
102
See also Cinque (2004:689f.) for discussion.
103
See Haider (2000:784f.) for arguments that topicalization must not involve the trace of the
finite verb.
104
However, it is noteworthy that evaluative adverbs fail to take complements in non-periph-
eral positions, i.e. when they are immediately preceded or followed by a non-lexical verb:
(i) *It's [TP [luckily [for all of us]] [T' perfectly within the law]].
(ii) *It [AgrSP/TP [luckily [for all of us]] [AgrS'/T' is perfectly within the law]].
105
Adverbs that assume the latter position are not necessarily assigned the manner reading (cf.
Ernst 2002:269ff.). But see above for arguments against right-adjunction.
106
Degrees of transitivity are measured in terms of the following properties (adapted from
Ernst 2002:275):
(i) 2 or more participants, action, telicity, punctuality, volition, affirmation, realis mode,
agency, affected object, individuated object
The more properties a predicate displays, the higher it ranges on the transitivity scale (see also
Hopper & Thompson 1980).
107
Notice that under the checking accounts of Case there is "no natural place for the condition
of adjacency" (Chomsky 1995:330), since Case is assigned or checked under specifier-head
agreement within an agreement-related functional projection.
CHAPTER 3
ADVERBS WITHIN THE LEXICAL LAYER
The aim of this chapter is to establish the principles that govern the distribution
and licensing of adverbs within the lexical layer, i.e. within the shell structure
projected by active and passive verbal elements. Two aspects are central: First,
the lexical layer of the clause, which has traditionally been equated with the
thematic domain of the verb, is split by intervening functional architecture (cf.
section 3.1). The lowest domain within the lexical layer, a potentially layered
VP-shell, is the thematic domain encompassing the lexical verb and its internal
argument(s), if any. This domain is dominated by a set of object-related func-
tional projections, notably AgrOP for direct object agreement, AgrIOP for indi-
rect object agreement and TelP for telicity, the head and specifier positions of
which are targeted by overt raising operations. The lexical layer is closed off
by an outer vP-shell hosting the verb's external argument (if any). It will be
shown that the systematic disentangling of the two thematic domains of the
verb, i.e. the domain of the verb's internal argument(s) and the domain of its
external argument, not only allows for a principled account of overt object
shift, Case-checking and word order variation, e.g. 'dative shift', but also paves
the way to accommodating within the lexical layer a set of functional projec-
tions hosting adverbs. Second, adverbs in the lexical layer are merged into
structure as specifiers of designated functional projections intervening between
the set of agreement and lexical aspect related functional projections and the
lower domain projected by the lexical verb. The positions vP-internal adverbs
assume relative to each other, the lexical verb and its internal argument(s) will
be shown to be accounted for by a small number of syntactic operations: overt
raising of the lexical verb and its internal argument(s), antecedent binding of
lexical material in , adverb incorporation into the lexical verb and verb
excorporation out of a syntactically complex verbal head. The latter two
options are exclusively reserved for completion adverbs. Vital to the analysis
proposed is the assumption that the lexical layers projected by active and
passive verbal elements are not identical, neither with respect to their internal
structure nor with respect to the movement properties of the verbal element
ultimately heading the lexical layer.
complement, undergoes overt raising to the head position of the outer shell, as
shown in (1b). As Larson (1988:343) points out, V raising is not only neces-
sary with respect to linearization but also follows from Case and agreement
requirements (cf. chapter 2.1.4). Given that Case-assignment by lexical heads
is to the right, the VP-internal subject, i.e. [DP the book], would fail to receive
Case if the verb did not raise to the head of the outer vP-shell, from where it
can govern DP in specVP. In addition, V-raising is triggered by the require-
ment that V must head a projection which is governed by Infl:1
(1) a. [I' [I ] [vP [v' [v ] [VP [XP ] [V' [V ] [VP [XP ] [V' [V V] [AdvP ]]]]]]]]
b. [I' [I ] [vP [v' [v Vi] [VP [XP ] [V' [V ti] [VP [XP ] [V' [V ti] [AdvP ]]]]]]]]
Under Larson's (1988, 1990) original analysis, the verb's internal arguments are
either generated as specifiers, e.g. [DP the napkin], [PP to her] and [D me] in
(2a), (2c) and (2d), or as complements, e.g. [DP the napkin] and [D her] in (2b),
[PP about the meal] in (2c) and [DP the details] in (2d). The external argument is
invariably generated as the specifier of the highest VP-shell:2
(2) a. [vP [D He] [v' [v handedi ] [VP [DP the napkin] [V' [V ti] [PP to her]]]]]
b. [vP [D He] [v' [v handedi ] [VP [D her]j [V' [V' [V ti] [D tj]] [DP the
napkin]]]]]
c. [vP [D He] [v' [v complainedi ] [VP [PP to her ] [V' [V ti] [PP about the meal
]]]]]
d. [vP [D He] [v' [v sparedi ] [VP [D me]j [V' [V' [V ti] [D tj]] [DP the details]]]]]
Following Aoun & Li (1989), Mulder (1992), Kitagawa (1994) and others, I
assume that the verb first merges with the theme argument (irrespective of its
categorial realization), so that the structural position assumed by the goal
argument (irrespective of its categorial realization) must be higher, as
schematically illustrated in (4).3 Under this analysis, verbs with two internal
arguments project one shell for each argument: the lower VP hosts the theme
argument and the higher VP hosts the goal argument:
(4) [vP [DP AGENT ] [v' [v ] [VP [XP GOAL] [V' [V ] [XP THEME]]]]]
ADVERBS WITHIN THE LEXICAL LAYER 115
Supportive evidence for the hierarchy in (4) derives from the so-called Barss-
Lasnik-asymmetries (1986), which have always been employed as a touchstone
for VP-structure and also as the main argument in favor of assuming VP-shells
(cf. Larson 1988:351ff.):
The contrasts in acceptability derive from the requirement that anaphors (him-
self) and bound pronouns (its) must be c-commanded by their antecedents
(John and each worker, respectively), but not vice versa. This requirement is
met in (5a) and (5c), but not in (5b) and (5d). Under the assumption that the
goal argument assumes a higher position than the theme argument, as stated in
(4), the vital c-command relations between the antecedent John and the ana-
phor himself on the one hand, and the quantifier-expression each worker and
the pronoun his on the other, come as a natural consequence.
As we shall see below, strings as in (6) do not pose a problem for our analy-
sis since the anaphor contained in the goal argument is c-commanded by the
raised theme argument:
(6) I introduced [[Sue, Anne and Louise] k ... [ … [to each other's husbandi]
[ … [ tk ]]]]
(7) a. * [AgrSP [D He]k [AgrS' … [… [vP [D tk] [v' [v handedi ] [VP [PP to her] [V'
[V ti] [DP the napkin ]]]]]]
b. [AgrSP [D He]k [AgrS' … [… [vP [D tk] [v' [v handedi ] [VP [D her] [V' [V ti]
[DP the napkin ]]]]]]
c. [AgrSP [D He]k [AgrS' … [… [vP [D tk] [v' [v complainedi ] [VP [PP to her]
[V' [V ti] [PP about the meal ]]]]]]
The derivation of the surface string He handed the napkin to her from (7)
involves movement operations that are reminiscent of overt object shift, as
shown in (8). That is to say that [DP the napkin] raises overtly from the comple-
ment position within VP, across [PP to her] in specVP to some higher specifier
position, AgrOP, and [V handed] raises overtly out of vP, first to AgrO and then
across the raised DP to a higher head position X:5
116 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(8) AgrSP
2
D AgrS'
th
AgrS …
h
XP
eu
X AgrOP
ti
DP AgrO'
fu
AgrO vP
fu
DP v'
eh
v VP
2
goal PP V'
th
theme V DP
# 4
hek handedi [DP the napkin]j ti tk ti [PP to her] ti tj
The assumption that English displays overt object shift has originally been
advanced by Johnson (1991) and later adapted and modified by Koizumi
(1993), Runner (1995 a, b, 2000, 2001), Lasnik (1995, 1999a, b, 2002) and
others.6 Koizumi (1993:111f.) argues that overt raising of DP to specAgrOP
and V to AgrO is triggered by a strong accusative Case- feature in V. In order
for the strong Case-feature not to induce interface violations, the verb raises
overtly to AgrO and summons [DP the napkin] to specAgrOP so that the strong
Case-feature can be eliminated by checking under specifier-head agreement
(see also Runner 1995a:2ff.; 2001:23).7 Other things being equal, Lasnik
(1995, 1999a, b, 2002) takes the strong feature under consideration to be "an
'Extended Projection Principle feature' residing in Agr, hence the same feature
that drives overt subject raising" (cf. Lasnik 1999b:203).8
Before turning to discuss supportive evidence for the claim that object shift,
in fact, is overt in English, I shall briefly comment on some of the conse-
quences of this analysis.
Given that the presence of a strong (Case- or) EPP-feature in AgrO triggers
both overt DP and V raising, the analysis sketched in (8) should carry over to
the derivation of the strings in (9), where both the theme and the goal argument
are realized as D(P)s:
ADVERBS WITHIN THE LEXICAL LAYER 117
That is to say that both D(P)s undergo overt object shift, as shown in (10),
where AgrIOP stands for indirect object agreement:9
(10) AgrSP
2
D AgrS'
th
AgrS …
g
XP
ru
X AgrIOP
tu
D AgrIO'
ru
AgrIO AgrOP
eh
DP AgrO'
fu
AgrO vP
2
D v'
fh
goal v VP
fh
theme D V'
1
V DP
# 4
hem sparedi [D me]k ti [DP the details]j ti [D tm] ti [D tk] ti [DP tj ]
(12) [vP [spec ext. arg.] [v' [v ] [AgrIOP … [AgrOP … [VP int. arg(s). ]]]]]
(13) a. [v' [v handedi ] [AgrOP [DP the napkin]j [AgrO' [AgrO ti ] [VP [PP to her] [V'
[V ti [DP tj ]]]]]]
b. [v' [v handedi ] [AgrIOP [D her]k [AgrIO' [AgrIO ti ] [AgrOP [DP the napkin]j
[AgrO' [AgrO ti ] [VP [D tk ] [V' [V ti [DP tj ]]]]]]
The first piece of evidence for overt object shift derives from so-called ECM
(Exceptional Case Marking) constructions as in (14), where the phonetically
realized subject of a nonfinite clausal complement precedes a matrix clause
adverb.12 The structure underlying the well-formed strings in (14) is that given
in (16) below.
Under analyses that take object shift to be an LF operation, i.e. an instance
of covert movement, the ill-formed strings in (15) cannot be accounted for:
(14) a. I have found Bobi recently [to be ti morose]. (Postal 1974: 146)
b. I suspect himi strongly [to be ti a liar]. (Authier 1991: 729)
ADVERBS WITHIN THE LEXICAL LAYER 119
(16) v'
ru
v AgrOP
2
D AgrO'
2
AgrO XP
tu
AdvP X'
fu
X VP
rh
V TP
rh
T VP
2
V AP/DP
# 5
a. foundj [D Bob]i recently tj tj to be [ti morose]
b. suspectj [D him]i strongly tj tj to be [ti a liar]
The second piece of evidence for overt object shift derives from pseudo-
gapping constructions, as in (17), where either the verb or the verb and other
VP-internal material is elided (see also Lasnik 1995, 1999a,b, 2002):13
(17) a. If you don't believe me, you will believe the weatherman.
b. Kathy likes astronomy, but she doesn't like meteorology.
c. The DA proved Jones guilty and the Assistant DA will prove Smith
guilty.
d. ? John gave Bill a lot of money, and Mary will give Susan a lot of
money. (adapted from Lasnik 2002:194)
Central to Lasnik's (2002) analysis is the idea that ellipsis is a PF deletion phe-
nomenon that applies to the entire VP, i.e. to the verb and to "other material"
that, for reasons to be made explicit, has not vacated VP. The ellipsis structure
in (18) differs from the non-elliptical structures considered so far, e.g. (13)
above, in that the verb (here in boldface) does not undergo overt raising to the
head position of vP:
120 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(18) vP
tu
D v'
tu
v AgrOP
3
DP AgrO'
2
AgrO VP
2
V XP
# 5
[D you]j will tj [DP the weatherman]i believe [DP ti]
(19) a. * Al called Pat a lunatic, and George did insult Pat (too).
b. * You will believe the weatherman. (out of the blue)
((a) from Smith (2001:179))
Under this analysis, the contrasts in (20) are attributed to the fact that X likes
winning votes is also instantiated in the matrix clause in (20a), with X
corresponding to George, whereas X did insult Pat is not instantiated in the
matrix clause in (20b). Since (20c) is an out-of-the-blue utterance, there is no
context in which X will believe Y could be instantiated.
Finally, coordinate structures as in (22) provide evidence for overt verb rais-
ing and indirect evidence for overt object shift. Following Munn (1987), I take
coordination to involve a functional projection, here represented as :P (cf.
Koster 1999a). Further, I assume with Kayne (1994), Rijkhoek (1998), Bianchi
(1999) and others that the first conjunct is realized in the specifier position of
:P, thus asymmetrically c-commanding the second conjunct in the complement
position. Crucially, :P is invisible in the sense that it "behaves as if it were the
constituent in its specifier position" (cf. Rijkhoek 1998:123) and, thus, can be
merged freely into structure:
(22) a. I handed [:P [the napkin to her] [:' [: and] [the plate to him]].
b. He handed [:P [her the napkin] [:' [: and] [him the plate]].
c. He complained [:P [to her about the meal] [:' [: and] [to him about the
wine]].
122 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Clearly, the coordinated constituents in (22) contain the verb's internal argu-
ments, but not the verb. That is, the coordinated constituents in (22a) and (22b)
must be larger than just VP. By hypothesis, (22a) involves AgrOP coordination,
(22b) AgrIOP coordination and (22c) VP coordination, as illustrated in (23):16
(23) a. [v' [v handed] [:P [AgrOP the napkin to her]] [:' [: and] [AgrOP the plate to
him]]]]
b. [v' [v handed] [:P [AgrIOP her the napkin] [:' [: and] [AgrIOP him the
plate]]]]
c. [v' [v complained] [:P [VP to her about the meal] [:' [: and] [VP to him
about the wine]]]]
This analysis automatically gives rise to the question of how the raised verb in
v is associated with the second conjunct.17 The seemingly uncomplicated
answer to this question would be that the verbal element in the second conjunct
is PF deleted under identity with the overt form originating in the first con-
junct, as illustrated in (24):18
(24) a. [v' [v handedi] [:P [AgrOP [DP the napkin]j [AgrO' [AgrO ti] [VP [PP to her] [V'
[V ti] [DP tj]]]] [:' [: and] [AgrOP [DP the plate]k … [VP [PP to him] [V' [V
handed] [DP tk]]]…]]]]
b. [v' [v handedi] [:P [AgrIOP [D her]k [AgrIO' [AgrIO ti] [AgrOP [DP the napkin]j
[AgrO' [AgrO ti] [VP [D tk] [V' [V ti] [DP tj]]]]]]] [:' [: and] [AgrIOP [D him]m …
[AgrOP [DP the plate]o … [VP [D tm] [V' [V handed] [DP to ]]]…]…]]]]
c. [v' [v complainedi] [:P [VP [PP to her] [V' [V ti] [PP about the meal]]] [:' [:
and] [VP [PP to him] [V' [V complained] [PP about the wine]]]]]]
that the verb is deleted in situ, as illustrated in (25), where Θ stands for the
extracted external Θ-role of the verb, which is assigned to specvP:19
(25) a. [AgrSP [D I]i [AgrS' ... [:P [vP [D ti ] [v' [v handedj] [AgrOP [DP the napkin]k
[AgrO' [AgrO ti] [VP [PP to her] [V' [V tj] [DP tk]]]]]]] [:' [: and] [vP [D he ] [v'
[v Θ ] [AgrOP [DP the plate]o [AgrO' [AgrO ] [VP [PP to him] [V' [V handed ]
[DP to]]]]]]]]]…]]
b. [AgrSP [D he]i [AgrS' ... [:P [vP [D ti ] [v' [v handedj] [AgrIOP [D her]m [AgrIO'
[AgrIO tj] [AgrOP [DP the napkin]k [AgrO' [AgrO tj] [VP [D tm] [V' [V tj] [DP
tk]]]]]]]]] [:' [: and] [vP [D he ] [v' [v Θ ] [AgrIOP [D him]p [AgrIO' [AgrIO ]
[AgrOP [DP the plate]o [AgrO' [AgrO ] [VP [D tp] [V' [V handed] [DP
to]]]]]]]]]]]…]]
c. [AgrSP [D he]i [AgrS' ... [:P [vP [D ti ] [v' [v complained j] [VP [PP to her] [V' [V
tj] [PP about the meal]]]]] [:' [: and] [vP [D he ] [v' [v Θ] [VP [PP to him] [V'
[V complained] [PP about the wine]]]]]]] … ]]
However, if we take into consideration the differences that exist between the
prepositional elements (and their projections) in (26) and lexical prepositions
(and their projections), the picture hitherto drawn turns out to be inadequate.
The fact that prepositional elements as in (26) are often referred to as
grammatical prepositions, grammaticalized prepositions, semantically empty
prepositions or Case-prepositions already points at the functional rather than
lexical categoryhood of the items under consideration.20 In the following, I
assume with Rauh (1996,186ff.; 1997: 147ff.; 2002b:16ff.), Bittner & Hale
(1996:5ff.) and Hale & Keyser (no date:23) and others that the prepositional
forms in (26) are spell-outs of inherent Case and as such an integral part of the
nominal projection, as shown in (27), where KP stands for Kase Phrase:21, 22
At this point, two of the arguments Rauh (1996, 1997a, b, 2002b) presents in
support of the KP analysis of Case-prepositions are relevant. First, the phonetic
form of the prepositional elements in (26) is determined, or 'selected', by the
124 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
lexical heads which take the constituents under consideration, i.e. KPs, as their
internal arguments, as shown in (28).23 No such 'phonetic form selection' can be
observed with verbs, e.g. put in (29), that take prepositional internal arguments:
(29) He put his wallet [{under, behind, near, on, into} the pillow cases].
Second, the constituents under consideration, i.e. KPs, pattern like D(P)s with
respect to binding facts, not like PPs, as the contrasts in (30) and (31) show (cf.
Rauh 1996:189; 2002b:17 for discussion):
Under the assumption that the goal argument in (26a, b), (30b) and (31c) as
well as both the theme (26b) and (31b) correspond to KP, i.e. a nominal projec-
tion, rather than to PP as in (29), (30c) & (31d), the constituents under
consideration are subject to Case-checking under specifier-head agreement.
Under Rauh's (1996, 2002b) analysis, the prepositional items heading KP
are "double-faced" in the sense that they are "in part already Case forms, in
part still prepositions" (cf. Rauh 2002b:18), whence they fulfill a dual function:
they are spell-outs of inherent Case and they assign structural Case to their DP
complement.24 However, the assumption that the prepositional items under
consideration are "double-faced" should not lead us to assume that KP, a
nominal projection, is not subject to Case-checking under specifier-head
agreement. I claim that the checking of inherent Case, as realized by preposi-
tional items, like structural Case and inherent Case, i.e. the so-called 'dative', is
checked under specifier-head agreement within an agreement-related func-
tional projection dominating VP, as in (32).25 However, the proposed analysis
wrongly predicts (33) to be well-formed:
(32) [V' [V handed]i [AgrIOP [KP to her]k [AgrIO' [AgrIO ti ] [AgrOP [DP the napkin]j
[AgrO' [AgrO ti ] [VP [KP tk ] [V' [V ti ] [DP tj ]]]]]]]]
ADVERBS WITHIN THE LEXICAL LAYER 125
In the following, I shall assume that [DP the napkin] raises across [KP to her] in
specAgrIOP to some higher specifier position within the verbal projection.
Before turning to the nature of this functional projection and the triggers of
overt DP and V to this additional functional layer, I would like to briefly focus
on the differences in interpretation between double object constructions of the
type V-D(P)goal-DPtheme, as in (34), and double object constructions of the type
V-DPtheme-KPgoal, as in (35):
It has often been argued that strings as in (34) and (35) differ with respect to
aspectual properties, i.e with respect to telicity: in so-called 'dative shifted'
structures, as in (34), the events are delimited by D(P)goal receiving DPtheme, e.g.
her friend receiving a rubber frog, whereas in (35), the event is delimited at the
moment DPtheme is affected, e.g. the rubber frog is sent.26 Building on Zagona's
(1994) analysis, Sato (1995) proposes accounting for the differences in event
delimitation, i.e. for the temporal differences between (34) and (35), in terms of
two discrete subevents, "E1 and E2, which correspond to the initial process and
the result-state of the process, respectively" (cf. Sato 1995:101). Under this
analysis, the result-state, i.e. E2, is associated with an aspectual functional pro-
jection, which I choose to re-label Tel(icity)P:
The telic verb raises to the head position of TelP to have its aspectual features
checked and summons DPtheme or KPgoal to specTelP in order to delimitate the
event.27, 28 Under Sato's (1995) analysis, Case-checking of [DP her friend] in
(37a) and [DP a rubber frog] in (37b) takes place under specifier-head agree-
ment in TelP ((37) adapted from Sato 1995:102f.).29
(37) a. [v' [v sent]i [TelP [DP her friend]j [Tel' [Tel ti ] [VP [DP tj ] [V' [V ti ] [DP a
rubber frog]]]]]]
b. [v' [v sent]i [TelP [DP a rubber frog]j [Tel' [Tel ti ] [VP [KP to her friend ] [V'
[V ti ] [DP tj ]]]]]]
126 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
In the following, I assume with Adger & Tsoulas (2000:14) that aspectual fea-
tures and Case-features are not checked under specifier-head agreement within
one and the same functional head. That is to say that aspectual features are
checked under specifier-head agreement within TelP and Case-features are
checked under specifier-head agreement within AgrOP and AgrIOP, with TelP
dominating AgrIOP, as in (38):
(38) v'
ru
v TelP
ru
DP Tel'
tp
Tel AgrIOP
eu
KP AgrIO'
th
Agr IO AgrOP
tu
DP AgrO'
th
AgrO VP
1
KP V'
1
V DP
# 4
a. senti [DP a frog]j ti [KP to her friend ]k ti [DP tj] ti tk ti tj
b. senti [DP her friend]j ti ti [DP a frog]k ti tk ti tj
This analysis not only allows us to maintain that Case-prepositions are func-
tional heads in the nominal projection and as such an integral part of the
extended nominal projection, KP, which is subject to Case-checking under
specifier-head agreement, but it also allows us to account for the fact that
DPtheme precedes KPgoal.
3.1.2 Summary
Based on the assumption that lexical verbs taking more than one internal argu-
ment (e.g. a theme and a goal argument) project one shell for each argument, I
argued that the verb always merges first with the theme argument. The deri-
vation of strings in which the theme argument linearly precedes the goal argu-
ment was shown to involve overt raising of both the direct theme and the
indirect goal argument to the specifier positions of AgrOP and AgrIOP, respec-
ADVERBS WITHIN THE LEXICAL LAYER 127
tively. Overt raising of both the direct and the indirect object was shown to be
triggered by the requirement that the Case-features involved be checked.30
Since the configuration that results from overt object shift gives rise to Mini-
mality violations, I suggested assuming with Koizumi (1993) that the thematic
domain of the verb be split as to accommodate the agreement-related func-
tional projections. Under this analysis, the external and the internal argument(s)
of the verb do not originate within the same minimal domain, i.e. the external
argument occupies the specifier position of vP and the internal arguments are
within VP. Overt verb raising (through AgrO and AgrIOP) to the head position
of vP was argued to be triggered by a strong Θ-feature residing in v.31 In sec-
tion 3.1.1, I suggested some refinements of the analysis as to account for the
relative linear orders of the direct and indirect object in (non)dative-shifted
double object constructions. As a first refinement, I suggested that the indirect
object in a non-dative shifted structure be categorized as a nominal constituent
(corresponding to Rauh's 1996, 1997, 2002b KP), which – in order to get fully
licensed – must raise overtly to an appropriate specifier position to have its
Case checked. The second refinement concerns the positioning of the indirect
object in dative shifted double object constructions. Drawing on earlier work
by Zagona (1994) and Sato (1995), I suggested that the differences between
dative shifted and non-dative shifted structures be accounted for in terms of
telicity, with telicity being associated with a functional projection, TelP, dis-
tinct from both AgrOP and AgrIOP.32 These refinements were shown to properly
account for the relative orders and the licensing of nominal constituents in so-
called double object constructions.
3.2 The Distribution and Licensing of Adverbs within the Lexical Layer
Starting out with the empirical fact that so-called manner adverbs (manner ad-
verbs proper, degree of perfection adverbs and means-domain adverbs) are
confined to the postverbal position in actives, whereas, in passives they may
either follow or precede the lexical verbs (sections 3.2.1.1 – 3.2.1.5), I provide
an analysis which ultimately rests on the assumption that the relative linear
orders of the adverbs and the verbal elements under consideration result from
structural differences in the lexical layers projected by active and passive ver-
bal elements and, closely related to that, from the movement properties dis-
played by active and passive verbal heads (cf. section 3.2.1.5). The analysis is
fleshed out in section 3.2.1.5.1 as to incorporate the positioning of manner
adverbs relative to the verb's internal argument(s). Section 3.2.1.5.2 focuses on
the vP-boundedness of the adverbs under consideration. Specifically, it will be
shown that there is no derivational relationship between the set of adverbs
considered here and their homophonous counterparts in the inflectional layer,
e.g. subject-related adverbs or agent-oriented adverbs in actives. Section
3.2.1.5.3 deals with agent-oriented adverbs in passives and the syntax of the so-
called by-phrase. Section 3.2.1.5.4 addresses the co-occurrence of vP-internal
128 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(39) a. The lid has been (carefully) opened carefully (by Bob).
b. The spider has been (deliberately) squashed deliberately (by Joe).
c. The cake has been (beautifully) decorated beautifully (by Paul).
d. The invitations were (electronically) sent electronically (by Tom).
3.2.1.1 Manner Adverbs. I follow Frey & Pittner (1999), Pittner (1999, 2000a,
2004) and Frey (2000) in assuming that 'true' manner adverbs are restricted to
following the lexical verb, as in (41), and that seeming preverbal occurrences
of manner adverbs in actives, as in (42), are, in fact, instances of subject-
related adverbs:
(43) a. Bob removed the lid carefully. ( ≠ Bob removed the lid, being careful
in doing so.)
b. Bob carefully removed the lid. ( = Bob removed the lid, being careful
in doing so.)
Under the analysis adopted here, manner adverbs are restricted to occurring
within the lexical layer, vP, whereas subject-related adverbs originate within
the inflectional layer, i.e. they are not derivationally related to 'true' manner
adverbs (pace Alexiadou 1997:134ff.). The strict local segregation of manner
adverbs and subject-related adverbs reflects the semantic distinction between
process-related adverbs and event-internal or event-related adverbs, respec-
tively (cf. Frey & Pittner 1999; Pittner 1999, 2000a, 2004; Frey (2000).
Supportive evidence for the distinction between the homophonous forms
under consideration in terms of manner adverbs and subject-related adverbs
derives from the fact that not all manner adverbs may assume the preverbal
position (cf. (44a, b)) and that the two types of adverbs may co-occur (even if
they are homophones as in (44c, d)):35
In passives, manner adverbs may follow and precede the passive participle:
3.2.1.1.1 All's Well That Ends Well.36 It has often been argued that manner
adverbs, such as harshly and carefully in (46) and well, slow, hard and fast in
(47) are tied to the postverbal position:37
(48) We've figured out the content of all the exam questions, but we haven't
worded them yet. (Ernst 2002:273)
To account for the fact that hard, well and quick are strictly postverbal, Ernst
(2002) suggests that these elements be "intrinsically (lexically) marked
[+Heavy]" (cf. Ernst 2002:274; 2004:770f.) and thus forced to occur in post-
verbal position.39 However, in light of the discussion in the previous section,
lexical marking of these elements is superfluous since manner adverbs are
restricted to occurring within the lexical layer, vP. This is not to say that these
elements are not special. But they are only special in so far as they cannot be
construed as subject-related adverbs (cf. chapter 4.3.1), whence they fail to
precede the lexical verb in actives. Strictly speaking, harshly, carefully, ele-
gantly, well and hard are perfect manner adverbs since, in passives, they may
either precede or follow the lexical verb:40
(49) a. [She]'s been told he's being harshly treated (harshly) … (CC)
b. The ground had been well prepared (well). (www)
c. [Rural] areas have been hard hit (hard) by the recession. (BNC)
The fact that slow and fast do not occur in preverbal position in passives, as
shown in (50), does not come as a surprise since –ly-less variants of manner
adverbs, in general, are confined to the postverbal position (cf. Sugioka & Lehr
1983:295; Radford 1988:138; Pittner 2001):
(51) a. * It rained
reluctantly, intentionally, deliberately,
b. * The bomb exploded
willingly, obediently
c. * This book reads
Abstracting away for a moment from the fact that strings as in (52a) are
ambiguous between the two readings, i.e. Joe is reluctant, willing or obedient,
or the doctor is, agent-oriented adverbs in passives either precede or follow the
participle, as shown in (56) and (57), and thus behave like manner adverbs:
3.2.1.5 The analysis. In a recent article, Caponigro & Schütze (2003) argue that
active participles and finite lexical verbs assume a higher position than passive
participles. Evidence for this assumption derives from the (non)admissibility of
degree of perfection adverbs in pre-participle position in actives and passives,
as already discussed.44
Since degree of perfection adverbs pattern like manner adverbs and means-
domain adverbs with respect to the positions they assume relative to active and
passive verbal elements, Caponigro & Schütze's (2003) proposal should apply
to the three subclasses under consideration as well as to agent-oriented adverbs
in passives (cf. chapter 3.2.1.2 for agent-oriented adverbs in actives).
The authors maintain that active participles and finite lexical verbs raise to
the head position of VoiceP, whereas passive participles do not. Following
Cinque (1999), they argue that the functional head Voice comes in two
variants: VoiceAct(ive) and VoicePass(ive), with only the former being associated
with a strong V-feature that attracts the active participle or finite verb: 45
(65) VoicePPass
ti
VoicePPass AgrOP
[–strong] 2
DPi AgrO'
2
AgrO µP
5
Adv µ'
5
µ VP
5
V'
5
V ti
Now, under the assumption that passive verbs, i.e. unaccusatives in general,
project an AgrOP whose specifier position is targeted by the nominal internal
argument (on its way to the sentential subject position), the passive participle
must raise to AgrOP in order for DP in specAgrOP and Adv in specµP to be
equidistant from the trace in the complement position of V.
However, as a result of V raising to AgrO, the adverb occurs postverbally:46
(66) a. [D The message]i was [AgrOP [DP ti ] [AgrO' [AgrOP sentj] [µP [Adv secretly]
[µ' [µ tj] [VP [V' [V tj ] [DP ti ]]]]]]]
b. [D The slogan]i was [AgrOP [DP ti ] [AgrO' [AgrOP parrotedj] [µP [Adv
obediently] [µ' [µ tj] [VP [V' [V tj ] [DP ti ]]]]]]]
c. [DP The aria]i was [AgrOP [DP ti ] [AgrO' [AgrOP performedj] [µP [Adv
horribly] [µ' [µ tj] [VP [V' [V tj ] [DP ti ]]]]]]]
d. [D His nose]i was [AgrOP [DP ti ] [AgrO' [AgrOP removedj] [µP [Adv
surgically] [µ' [µ tj] [VP [V' [V tj ] [DP ti ]]]]]]]
At first glance, it would seem that Caponigro & Schütze's (2003) analysis
could be maintained if we assumed that the adverbs under consideration origi-
136 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
nated in specVoiceP, which would give us the desired linear order (67). Notice
however, this analysis cannot be reconciled with Caponigro & Schütze's (2003)
analysis according to which the finite lexical verbs and active participles
cannot raise beyond the head position of VoiceP. That is to say the adverb
would necessarily, but illicitly precede the active verbal element (68):
(67) a. [DP The message]i was [VoiceP [Adv secretly] [Voice' [VoiceP ] [AgrOP [DP ti ]
[AgrO' [AgrOP sentj] [VP [V' [V tj] [DP ti ]]]]]]]
b. [DP The slogan]i was [VoiceP [Adv obediently] [Voice' [VoiceP ] [AgrOP [DP ti ]
[AgrO' [AgrOP parrotedj] [VP [V' [V tj] [DP ti ]]]]]]]
c. [DP The aria]i was [VoiceP [Adv horribly] [Voice' [VoiceP ] [AgrOP [DP ti ] [AgrO'
[AgrOP performedj] [VP [V' [V tj] [DP ti ]]]]]]]
d. [DP His nose]i was [VoiceP [Adv surgically] [Voice' [VoiceP ] [AgrOP [DP ti ]
[AgrO' [AgrOP removedj] [VP [V' [V tj] [DP ti ]]]]]]]
(68) a. * They (have) [VoiceP [Adv secretly] [Voice' [VoiceP sentj] [AgrOP [DP the
message]i [AgrO' [AgrOP tj] [VP [V' [V tj] [DP ti ]]]]]]]
b. * She (has) [VoiceP [Adv horribly] [Voice' [VoiceP performedj] [AgrOP [DP the
aria]i [AgrO' [AgrOP tj] [VP [V' [V tj] [DP ti ]]]]]]]
c. *We (have) [VoiceP [Adv surgically] [Voice' [VoiceP removedj] [AgrOP [DP his
nose]i [AgrO' [AgrOP tj] [VP [V' [V tj] [DP ti ]]]]]]]
Before turning to show that the positional differences between active and
passive verbal elements as well as the relative orders of the adverbs and active
and verbal elements follow automatically under a modified version of the Split
VP Hypothesis, I would like to comment on the status of AgrOP in passive
constructions (cf. (64b) & (65)).
As we have seen in section 3.1, overt object shift is triggered by a strong
Case-feature in AgrO. The fact that passive verbs, i.e. unaccusatives in general,
fail to assign Case to their nominal internal argument (whence the nominal
internal argument raises to the sentential subject position) casts doubt on the
assumption that passive verbs project an AgrOP, as in (64b) & (65) above.
However, there is evidence for the presence of a functional projection, FP,
dominating passive verbs.47 The first piece of evidence derives from expletive-
associate structures, where the nominal internal argument precedes the passive
participle (data from CC):
(69) a. There have been [FP [DP 67 matches]i [F' [F drawnj] [VP [V tj] [DP ti]]]].
b. There have been [FP [DP enough people]i [F' [F hurtj] [VP [V tj] [DP ti]]]].
(70) a. [DP The cookies]k have been [FP [QP [Q all] [DP tk]]i [F' [F eaten j] [VP
[V tj ] [QP ti]]]]
b. [DP The pages]k have been [FP [QP [Q all] [DP tk]]i [F' [F tornj] [VP
[V tj ] [QP ti]]]]
While there is ample evidence for the presence of an extra functional projection
within the extended projection of a passive verb, there is no evidence for this
functional projection to correspond to AgrOP, the domain in which Case-
checking of the direct object takes place.
In two fairly recent articles, Belletti (1998:12f.; 2001:2) suggests that Case-
checking and passive participle agreement be seen as taking place in two
different functional projections because the checking of (accusative) Case
proceeds completely independently from the presence or absence of participial
agreement morphology.48 The two functional projections involved in the two
checking configurations are AgrOP for Case and AgrPrtP for passive participle
agreement:49
In the following, I will show that the Split VP Hypothesis allows us to account
for the distributional differences between active and passive verbal elements as
well as for the placement of the adverbs relative to the verbal elements under
consideration.
Under the analysis I am going to propose, the internal structure of the lexi-
cal layers projected by active and passive verbal elements differ with respect to
the presence vs. absence of the vP-layer and thus with respect to the movement
properties of active and passive verbal elements. Moreover, there is no place
for the notorious VoiceP, whose specifier position may or may not be overtly
targeted by the adverbs under consideration, since the active-passive
dichotomy is reflected by the absence vs. presence of the functional projection
Pass(ive)P.
As we have seen in section 3.1, active verbal elements project a vP the head
of which hosts a strong Θ-feature that attracts the matching feature of V, i.e.
the external argument contained in the argument structure of the verb, by drag-
ging along the phonetic form of V, which is necessary for PF convergence.
Since the argument structures of passive verbs contain an implicit agent argu-
ment, but not a genuine external argument, they do not project onto vP, i.e. the
upper boundary of the lexical layer of passive verbs is not delineated by vP, but
by AgrPrtP. The structures of the lexical layers projected by active and passive
verbs are given in (73) and (74), respectively.50 If we say nothing further, the
adverbs under consideration will occur strictly postverbally, as is required for
actives, but not for passives:
In actives, both the nominal internal argument and the lexical verb raise
overtly. The respective movement operations are triggered by a strong Case-
ADVERBS WITHIN THE LEXICAL LAYER 139
(75) a. [D The message]i was [AgrPrtP [DP ti ] [AgrPrt' [AgrPrtP sentj] [µP [Adv
secretly] [µ' [µ tj] [VP [V' [V tj ] [DP ti ]]]]]]]
b. [D The slogan]i was [AgrPrtP [DP ti ] [AgrPrt' [AgrPrtP parrotedj] [µP [Adv
obediently] [µ' [µ tj] [VP [V' [V tj ] [DP ti ]]]]]]]
c. [DP The aria]i was [AgrPrtP [DP ti ] [AgrPrt' [AgrPrtP performedj] [µP [Adv
horribly] [µ' [µ tj] [VP [V' [V tj ] [DP ti ]]]]]]]
d. [D His nose]i was [AgrPrtP [DP ti ] [AgrPrt' [AgrPrtP removedj] [µP [Adv
surgically] [µ' [µ tj] [VP [V' [V tj ] [DP ti ]]]]]]]
The fact that the adverbs in (78) intervene between the raised nominal internal
argument in specAgrPrtP and the lexical verb clearly shows that the passive
participle cannot have raised to the head position of AgrPrtP, but must have
stayed put.51 This, in turn, means that the head position of AgrPrtP is not
140 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
endowed with strong verbal features for abstract passive participle agreement,
whence the raising of the passive verbal element in (75) and (77) cannot be
seen as preventing interface violations.
In the remainder of this section, I would like to concentrate on the seem-
ingly optional raising of the passive participle across the adverb in specµP.52
The seemingly alternative orders of the adverbs under consideration and the
passive participles can be accounted for in terms of focalization, with focus (in
the relevant sense) being "defined in terms of the discourse notion of presup-
position: that is, the focus is the nonpresupposed part of the sentence" (cf.
Zubizarreta 1998:1). The information shared by the participants is referred to
as the background.53 Focus has traditionally been marked by the feature [F],
which is not to be confused with the peripheral feature [ FOC] (cf. Chomsky
1998:22; Rizzi (1997: 269ff.). That is to say that while [FOC] is an operator
feature that is subject to checking within FocP, [F] is a pragmatic feature that
can be associated with any node in surface structure, as shown in (79), where
the context questions, which roughly correspond to the presupposition, serve to
elicit the focus domain in terms of dominance by [F]:
All and only constituents that are dominated by [F] count as focus. Thus, for
example, the focus domain in (79a) is restricted to [DP the pie] (minimal focus),
whereas in (79b), the focus domain is constituted by [vP ate the pie] (non-
minimal focus), and in (79c) by the entire clause [ CP John ate the pie] (maximal
focus). Constituents which neither carry the feature [F] nor are dominated by a
constituent carrying the feature [F] are background constituents that convey
participant-shared information.
In the raising case, the adverb is focalized and thus constitutes new
information, e.g. the manner in which the message was sent or the means by
which his nose was removed. In the non-raising case, the action denoted by the
verb is focalized, whereas the manner in which or the means by which the
action is performed are backgrounded:54
(80) a. the messagei was [AgrPrtP [DP ti ] [AgrPrt' [AgrPrt sentk] [µP [Adv secretly] [µ'
[µ tk ] [VP [V' [V tk ] [DP ti ] ]]]]]]
b. the messagei was [AgrPrtP [DP ti ] [AgrPrt' [AgrPrt ] [µP [Adv secretly] [µ'
[µ [VP [V' [V sent ] [DP ti ] ]]]]]]
This assumption is corroborated by the fact that the adverbs under consider-
ation cannot be stressed if in pre-participle position, i.e. the participle raises
overtly across the adverb to the head position of AgrPrtP:
ADVERBS WITHIN THE LEXICAL LAYER 141
The fact that manner adverbs, degree of perfection adverbs and means-domain
adverbs may head internally complex structures, as shown in (82), suggests
that syntactic complexity restrictions are not as severe as prosodic restrictions,
unless syntactically complex AdvPs are not head-final:
At first glance, strings as in (83) constitute a problem for our analysis since, as
we have seen in chapter 3.1.1, KPs are nominal internal arguments that, just
like DPs, are subject to Case-checking under specifier-head agreement with an
appropriate functional head.
To cut a long story short, postulating additional functional projections that
could be interspersed between vP and Agr(I)OP and/or between AgrIOP and
AgrOP, as indicated in (84), is not desirable since, in addition to giving rise to
the question about the nature of FP and the trigger for adverb raising, it falls
short of ruling out the illicit strings in (85):56
(86) He handedk [DP the napkin]i … [AgrIOP [KP her]m [AgrIO' [AgrIO tk] [AgrOP [DP
ti] [AgrO' [AgrO tk] [µP [AdvP secretly] [µ' [µ tk ] [VP [KP tm] [V' [V tk ] [VP [DP ti]
[V' [V tk] … ]]]]]]]]]]
ADVERBS WITHIN THE LEXICAL LAYER 143
(88) µP
tu
Adv µ'
th
µ VP
th
KP V'
rh
V VP
th
DP V'
2
V
2
KP
# 5
He handedk the napkini secretly tk tk ti tk e to her
Under this scenario, is licensed by the (trace of the) lexical verb handed,
and KP is structurally licensed as the complement of the empty verbal head .
However, since the head of is semantically empty, it cannot formally
license KP. I assume that the formal licensing of KP takes place indirectly, i.e.
by forming a chain with a null antecedent pro in specAgrIOP. I take pro in (89)
to be formally licensed by carrying a Θ-role assigned by V, by being in
specifier-head agreement configuration with a head that carries strong Φ-
features) (i.e. AgrIO) and by being coindexed with an overt subsequent [KP to
her], which allows for the recovery of the semantic content of pro: 61
144 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(89) v'
ei
v TelP
3
DP Tel'
1
Tel AgrIOP
2
KP AgrIO'
fu
AgrIO AgrOP
2
DP AgrO'
eh
AgrO µP
2
Adv µ'
2
µ VP
th
KP V'
th
V VP
th
DP V'
th
V
1
KP
# 4
handedk [DP the napkin]j tk [KP prom] tk tj Adv tk tm tk tj tk e to her
The presence of pro is not only required for the indirect licensing of the overt
subsequent to her, but also for full interpretation. That is to say that pro
buddies as the nominal internal goal argument of the verb handed which, in
order not to induce interface violations, raises to specAgr IOP to have its Case
checked. By virtue of forming a representational chain with pro, [KP to her]
bears case. The requirement that [KP to her] be formally licensed by an ante-
cedent also prevents the proliferation of empty structure below the lexical verb,
i.e. lexical material within can only be licensed if it can be associated with
a null antecedent in the "interpretable part of the sentence."
ADVERBS WITHIN THE LEXICAL LAYER 145
Obviously, this analysis gives rise to two interrelated questions: why are DP
internal arguments barred from occurring within (cf. (90)) and which
conditions warrant the 'seeming' extraposition of internal arguments, i.e. their
occurring within ?
In reply to the first question, it is not the case that DP internal arguments are
barred from following the adverb. As shown in (91), the order adverb > DP is
admissible if the adverb is defocalized and the DP focalized by either carrying
a heavy accent or by being syntactically complex:62
At first glance, the strings in (97), where silently, quickly and carefully appear
in sentence-initial position, seem to invalidate the claim that manner adverbs
are vP-bound:
Were the sentence-initial adverbs in (97) and (98) in fact raised manner
adverbs, the unacceptability of the strings in (98) would follow from Mini-
mality violations, i.e. on its way to the sentence-initial position, the manner
adverb would have to have raised across a subject-related adverb. However,
given the fact that the strings in (99) are perfectly well-formed despite the
alleged manner adverb having raised across a subject-related adverb, Mini-
mality violations cannot be the key to the ill-formedness of the strings in (98).
That is to say that the strings in (99) should be as bad as those in (98) since the
alleged manner adverb would have to have raised across a subject-related
adverb (irrespective of the position the latter assumes):
Thus, the contrasts between the illicit strings in (95) and (96) and the well-
formed strings in (97) are deceptive because the adverbs in (97) are subject-
related homophones of manner adverbs, not manner adverbs proper (cf.
Emonds 1976:157ff. for related arguments). The sentence-initial adverbs in
(97) are subject-oriented adverbs that are not only distinct from manner
adverbs, but, as will be seen in chapter 4.3.1, also from subject-attitude adverbs
as in (99). As we have seen at the outset of section 3.2.1.1, manner adverbs do
not qualify the subject, but subject qualification is what is at stake here, as can
be read off the paraphrases in (100):
Supportive evidence for the claim that seeming sentence-initial manner adverbs
are, in fact, subject-related adverbs again derives from the fact that adverbs like
loudly, dimly and unfairly in (101) are barred from this position:
Finally, the fact that calmly, clumsily and carefully in sentence-initial position
do not block the presence of a genuine manner adverb within vP shows that the
elements under consideration cannot be manner adverbs:
henceforth [+PASS, +AG].66 That is to say that the feature [+PASS, +AG] should
be vital for the licensing of both agent-oriented adverbs and the by-phrase.
Given that agent-oriented adverbs in passives are vP-bound, they are automati-
cally within the same minimal local domain as the implicit agent argument viz.
[+PASS, +AG].67 The feature [+PASS, +AG] is also vital for the licensing of the
by-phrase by means of which the implicit argument is optionally lexicalized:
(105) AgrPrt'
eo
AgrPrt µP
3
Adv µ'
rh
µ ...
h
V'
ru
V
2
PP
5
a. was deliberately besmirched e by the KGB
b. has been encouragedk deliberately tk tk e by strategists
[+PASS, +AG] [+PASS, +AG]
also that the licensing of the by-phrase within is contingent on the presence
of the feature [+PASS, +AG], which is present on every link in the participle
movement chain.
A potential problem for the analysis presented consists in Case-checking of
the nominal expressions denoting the implicit agent, i.e. [DP the KGB] in (104).
The verbal head, , being semantically empty, can hardly be argued to project
an AgrOP the specifier position of which could be targeted for reasons of Case-
checking. Moreover, by cannot be felicitously categorized as an element of
category K since the phrase headed by by is not selected by the lexical verb, i.e.
by the KGB is not an internal argument of the lexical verb. On the other hand,
by does not make a good lexical preposition either. Thus, to account for the
licensing of [DP the KGB] we have to assume that by as an a-thematic
prepositional element projects its own functional architecture:68
(106) [FP [F byj] [AgrOP [DP the KGB]i [AgrO' [AgrO tj] [PP [P tj] [DP ti]]]]]
3.2.1.5.4 Co-Occurrence and Linear Order: A Residual Problem and Its Solu-
tion. Under the analysis proposed, the fact that means-domain adverbs may co-
occur with both manner adverbs and degree of perfection adverbs, as shown in
(107a, b) and (108a, b), is neither expected nor desired. However, the relative
order of the elements under consideration is fixed, i.e. means-domain adverbs
must follow manner adverbs and degree of perfection adverbs, as shown in
(107c, d) and (108c, d). Moreover, as can be seen in (107e, f) and (108e, f), the
elements under consideration may not be adjacent. The illicit strings in (109)
suggest that manner adverbs and degree of perfection adverbs cannot co-occur
which, in turn, suggests that they compete for the same structural position:
However, the assumption that manner adverbs and degree of perfection ad-
verbs sit in the same structural slot is not tenable on semantic grounds because
manner and degree of perfection expressions in fact may co-occur. This is the
case in (110), where the manner expression is phrasal 'in an X manner' (and
thus realized within VP, i.e. manner adverbs are merged in a higher position
than degree of perfection adverbs (cf. below):69
(115) [AgentP [Adv ] [Agent' [Agent ] [µP [Adv ] [µ' [µ ] [DegPerfP [Adv ] [DegPerf' [DegPerf ]
[MeansP [Adv ] [Means' [Means ] [VP ]]]]]]]]]
Under this analysis, the structure of the active strings in (107a) and (108a) is
given in (116). Notice that both [KP to everybody/to their new surroundings]
and the lower of the two adverbs, i.e. electronically and surgically, are realized
within , but licensed from outside (cf. section 3.2.1.5.1):70, 71
152 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(116) AgrIOP
tu
KP AgrIO'
eh
AgrIO AgrOP
2
DP AgrO'
fu
AgrO µP/ DegPerfP
th
Adv µ'/DegPerf'
fo
µ//DegPerf MeansP
eg
Adv Means'
1
Means VP
th
KP V'
th
V VP
th
DP V'
th
V
th
KP
th
Adv
# 4
a. [KP pro]m ti [DP tj] tk secretly tk e tk tm tk tj tk [KP ] e electronically
b. [KP pro]m ti [DP tj] tk perfectly tk e tk tm tk tj tk [KP ] e surgically
Underlying the strings in (110) is the structure in (117), where the complex
manner expression, here given as PP, is realized within VP:
(117) a. [v' [v singsk] [µP [Adv ] [µ' [µ ] [DegPerfP [Adv beautifully] [DegPerf' … [V' [V tk]
[ VP [ V e] [PP in a slow manner]]]]]]]]
b. [AgrPrt' [AgrPrt ] [µP [Adv ] [µ' [µ ] [DegPerfP [Adv beautifully] [DegPerf' … [V'
[V written] [ VP [ V e] [PP in a bizarre way]]]]]]]]
ADVERBS WITHIN THE LEXICAL LAYER 153
The passive strings in (112a), (112c) and (112e) correspond to the structure in
(118) (irrelevant details omitted),72 whereas the strings in (113a) and (113c)
correspond to the structure in (119). In all cases considered, verb raising
proceeds as outlined in the previous sections (irrelevant details omitted):
(118) a. has been [AgrPrt' [AgrPrt ] [AgentP [Adv deliberately] [Agent' [Agent writtenk]
[DegPerfP [Adv poorly] [DegPerf' … tk … ]]]]]
b. has been [AgrPrt' [AgrPrt ] [AgentP [Adv deliberately] [Agent' [Agent ] [DegPerfP
[Adv poorly] [DegPerf' … written … ]]]]]
c. has been [AgrPrt' [AgrPrt writtenk] [AgentP [Adv deliberately] [Agent' [Agent tk ]
[DegPerfP [Adv poorly] [DegPerf' ... tk … ]]]]]
(119) a. has been [AgrPrt' [AgrPrt ] [AgentP [Adv expertly] [Agent' [Agent removedk ]
[MeansP [Adv surgically] [Means' ... tk … ]]]]]
b. has been [AgrPrt' [AgrPrt ] [AgentP [Adv expertly] [Agent' [Agent ] [MeansP [Adv
surgically] [Means' ... removedk …]]]]]
The structure of the string in (112b), where the agent-oriented adverb follows
the degree of perfection adverb, contains VP which hosts the agent-oriented
adverb, as shown in (120):
(120) has been [AgentP [Adv ] [Agent' [Agent ] [DegPerfP [Adv poorly] [DegPerf' [DegPerf ] [VP
[V written] [ VP [ V e] [Adv deliberately]]]]]]]
3.2.1.6 Summary. I have shown that the positioning of manner adverbs, degree
of perfection adverbs and means-domain adverbs relative to active and passive
verbal elements (cf. sections 3.2.1.1 – 3.2.1.5) follows from the interplay of the
following factors (cf. section 3.2.1.5). First, the internal structure of the lexical
layer differs across the projections of active and passive verbal heads. While
active verbal heads project onto vP, the specifier position of which hosts the
external argument, passive verbal elements do not project onto vP. Second, the
lexical layers projected by active and passive verbal heads differ with respect
to the nature of the agreement-related functional projection, AgrOP in the case
of transitive active verbal heads and AgrPrtP in the case of passive verbal heads:
whereas AgrOP is involved in Case-checking, AgrPrtP is involved in the
checking of participle agreement (cf. Belletti 1998:12f.; 2001:2), which is
abstract in English. Third, active and passive lexical verbs display different
movement properties: while active verbs obligatorily raise overtly to (and
154 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
through) AgrO, passive verbs are not subject to obligatory raising to AgrPrtP.
Overt raising of the passive participle to AgrPrtP was shown to be discourse
based, i.e. if the adverb is focalized, the participle vacates its base position and
if the participle is focalized, it stays put (wide scope reading).
In section 3.2.1.5.1, the focus was on manner adverbs in actives that
seemingly assume positions other than the VP-final position, notably positions
preceding the verb's internal argument(s). Instead of assuming that the
positioning of manner adverbs in these cases is contingent on the presence of
further functional projections preceding AgrOP and/or AgrIOP, I suggested that
nominal internal arguments that follow manner adverbs be phonetically
realized as non-extraposed structural complements within an empty-headed
verbal projection, , which is projected below the base position of the lexical
verb.73 Nominal internal arguments that are realized within , i.e. outside the
lexical layer proper, must be licensed by a null antecedent, pro, within the lexi-
cal layer whose presence is also required for full interpretation: pro is assigned
a Θ-role and pro has to have its Case checked under specifier-head agreement
with an appropriate Agr-head. The realization of a nominal internal argument
outside the lexical layer was argued to be conditioned by the focalization of the
nominal internal argument and the defocalization of the adverb. That is to say
that the nominal internal argument, in order to be focalized, must be in a right-
peripheral position which, for reasons of Case-checking, cannot be its base
position: the realization of the nominal internal argument within resolves
this conflict. In section 3.2.1.5.2, I presented evidence for the assumption that
manner adverbs, degree of perfection adverbs and means-domain adverbs, as
well as agent-oriented adverbs in passives are vP-bound, i.e. they do not leave
the lexical layer. In section 3.2.1.5.3, I sketched an analysis for by-phrases in
passives which relies on the assumption that by-phrases are optionally merged
into structure as complements within an empty-headed verbal projection, with
the licensing of the by-phrase being contingent on the feature [+PASS, +AG]
present on every link of the participle movement chain. Finally, I focused on
the co-occurrence of the adverb types discussed in sections 3.2.1.1 – 3.2.1.5
and suggested splitting µP into four discrete functional projections: AgentP,
µP, DegPerfP and MeansP. The admissible linear orders of manner adverbs,
agent-oriented adverbs, degree of perfection adverbs and means-domain ad-
verbs relative to the lexical verb and its internal argument(s), if any, were
shown to follow from the different raising properties of verbs in actives and
passives and from the realization of lexical material within .
(122) a. The work of a damaged area of the brain is sometimes partially taken
over by another area. (CC)
b. He squeezed it slightly. (CC)
c. But if you cut our alcohol completely, you cut out this risk. (CC)
d. Ultimately, Microsoft products will integrate your PC's desktop
environment totally with the Internet. (CC)
e. [He] is wise enough to see that it would ruin him utterly. (BNC)
Moreover, completion adverbs – not only in passives ((123) & (124)), but also
in actives ((125) & (126)) – may both follow and precede the lexical verb:75
(126) a. [The] complete sentence, which you only partially quoted, makes
clear that … (CC)
b. The Nationwide decision will slightly affect the rate at the Cheltenham
Gloucester. (CC)
c. [Questioned] by the police, he completely lost his head… (CC)
d. Those who assimilate must totally break with the past. (CC)
e. He […] utterly rejects the idea of present suffering… (CC)
156 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Notice further that completion adverbs are barred from the inflectional layer of
the clause:
It is generally held that completion adverbs interact with the aspectual, i.e. telic
properties of the verbs involved: they specify the degrees to which a direct
object is affected by the action denoted by the verb.76
Instead of assuming with Laenzlinger (1998:42) and Cinque (1999: 100ff.)
that completion adverbs originate in the specifier position of a designated
functional category Asp(ect)P within the inflectional layer of the clause, I
follow Alexiadou (1997:132) in assuming that completion adverbs originate
within the lexical layer which corresponds to VP under her analysis. This
assumption is based on the fact that the presence of completion adverbs is
sanctioned by the presence of corresponding lexical aspectual, i.e. telic features
on the verbs they are construed with. As for structure, two scenarios are
possible: completion adverbs integrate into structure as either complements
(128) or as specifiers (129). Given that all raising operations within vP are
overt, both structures automatically account for the postverbal occurrences of
completion adverbs as in (123) and (125), but not necessarily for the preverbal
occurrences (cf. (124) & (126)), especially not for preverbal occurrences in
actives since the lexical verb obligatorily raises to the head of vP:
(129) [vP' ... [AspPcompletive [Adv ] [Asp' [Asp ] [VP [V] [DPtheme ]]]]]
(131) [AspPcompletive [Adv ADVi] [AspP' ... [vP ... [VP [V ] [Adv ti ]]]]]
(132) [VP [DP ] [V' [V ] [... [VP [V [Adv ADVi] [V V]] [Adv ti ]]]]]
(133) a. [Those who] commit themselves totally to their careers will succeed.
b. [He] decided to devote himself completely to healing [...] (BNC-data)
This leaves us with the incorporation analysis in (132) which not only would
account for preverbal occurrences of completion adverbs in both passives and
actives, but also, if kitted out with conditions on excorporation, for completion
adverbs intervening between the verb's internal arguments. Before going into
details, I would like to come back to the specifier analysis in (129). As it
stands, it runs into almost the same set of problems as the analysis in (131).
That is to say that preverbal occurrences of completion adverbs in actives can-
not be accounted for since the lexical verb obligatorily raises across the adverb
and it is not clear how completion adverbs come to intervene between the
verb's internal arguments as in (133).
As already mentioned, adverb incorporation is a promising lead. However, it
is not clear that we need to analyze completion adverbs as structural comple-
ments. Given that "even under a Larsonian analysis of VP, complements can be
merged as specifiers" (Cinque 2004:691) and given that nothing prevents the
incorporation of a single-term element from a specifier into a c-commanding
verbal head, we can give up the rather artificial asymmetry between comple-
ment- and specifier-type adverbs and assume that completion adverbs, like all
other types of vP-internal adverbs, are merged as specifiers:
(134) [AgrO' [AgrO [V [Advk] [Vi ]]] [AspPcompletive [Adv tk ] [Asp' [Asp ti ] [VP [V ti ]
[DPtheme ]]]]]
Recall from section 3.2.1.5 that, in passives, manner adverbs, degree of perfec-
tion adverbs and means-domain adverbs, depending on whether they are focal-
ized or not, follow or precede the passive participle. The postverbal occurrence
of these elements was seen as the result of verb raising to AgrPrt, i.e. across the
adverb in specµP, as triggered by the defocalization of the verb and the focal-
ization of the adverb. It would seem that the same mechanism is at work with
pre- vs. postverbal completion adverbs in passives (cf. below):
158 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(135) a. This ruini has been [AgrPrtP [DP ti] [AgrPrt' [AgrPrt ] [AspPcompl. [Adv partially]
[Asp' [Asp ] [VP [V rebuilt] [DP ti]]]]]]
b. This ruini has been [AgrPrtP [DP ti] [AgrPrt' [AgrPrt rebuiltk] [AspPcompl. [Adv
partially] [Asp' [Asp ] [VP [V tk] [DP ti]]]]]]
If we say nothing further, an analysis along these lines fails to explain how
partial strings consisting of a completion adverb and a passive participle, but
not a completion adverb on its own, come to precede other vP-internal adverbs:
(137) AgrPrt'
qp
AgrPrt µP
1
Adv µ'
1
µ AspPcompletive
1
Adv Asp'
1
Asp VP
4
V DP
was [V [completely]o [ruined]]k delib. tk to tk tk ti
This analysis carries over to actives, as shown in (138). Unless focalized (or
syntactically complex), the completion adverb incorporates into the lexical
verb forming a complex verbal head, [V [Adv ] [V ]], which obligatorily raises to
the head position of vP. Slight modifications are necessary for us to account for
ADVERBS WITHIN THE LEXICAL LAYER 159
the fact that completion adverbs may be separated from the lexical verb as is
the case in (139), where the completion adverb precedes a nominal internal
argument of the verb which, in turn, is preceded by the verb itself:
(138) vP
qo
D v'
qp
v AgrOP
ri
DP AgrO'
fu
AgrO AspPcompletive
1
Adv Asp'
1
Asp VP
1
V DP
# 4
he [V [utterly]o rejects]k [DP the idea…]i tk to tk tk ti
(139) a. But in retirement, she's devoting herself completely to her other self.
b. Journalists are able to devote themselves utterly to the search for
information… (www-data)
The position the adverbs assume (139) is between the direct and indirect object
of the verb. As we have seen in chapter 3.1.1, vPs headed by double object
verbs, in addition to projecting AgrOP and AgrIOP, also project the aspect
related functional projection Tel(icity)P whose specifier position is targeted by
the nominal constituent that delimits the event, i.e. DPtheme in (139).77, 78 As can
be seen in (140) below, the only available position for the completion adverbs
is the head position of TelP. Since both the presence of completion adverbs and
the presence of TelP are contingent on the lexical aspectual properties of the
verbs involved, and since the presence of completion adverbs is also condition-
al on the presence of an affected object, I assume that completion adverbs
excorporate out of the complex verbal head [V [Adv ] [V ]], i.e. they are left
behind in Tel, after the verb has checked its lexical aspectual features with DP
in specTelP, as schematically indicated in (141). That is to say that completion
adverbs can only excorporate out of [V [Adv ] [V ]] in a position in which the
verb can check off its telicity feature (details omitted):
160 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(140) [v' [v devoting ] [TelP [DP herself]i [Tel' [Tel completely] [AgrIOP [KP to her
other self] [AgrIO' …ti …]]]]]
(141) v'
ru
v TelP
wi
DP Tel'
ru
Tel AgrIOP
2
KP AgrIO'
1
AgrIO AgrOP
1
DP AgrO'
1
AgrO AspP completive
1
Adv Asp'
1
Asp VP
1
KP V'
1
V DP
# 4
[V ]z [DP ]m [V [Adv ]o [V tz]]k [KP ]i tk tm tk to tk ti tk tm
These patterns are expected under the analysis presented here. MADM adverbs
are generated in preverbal position. In (142a) – (145a), the MADM adverb is
focalized, whence the defocalized complex verbal element [V [Adv ] [V ]] raises
across the preverbal specifier position hosting these adverbs to the head posi-
tion of AgrPrtP. In (142b) – (145b), the completion adverb is focalized and thus
does not incorporate into the verb, which stays put. The strings in (142d) –
(145d) exemplify the case in which the verb is focalized, whence the comple-
tion adverb has incorporated into the verb, and the newly formed complex verb
162 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
[V [Adv ] [V]] stays put. The illicit cases in (142c) – (145c) cannot be the result
of overt raising operations: to derive them from the underlying order 'MADM
adverb > participle > completion adverb', the completion adverb would have to
have raised autonomously to AgrPrt, which, given Categorial Identity and the
Head Movement Constraint, is impossible. To derive (142e) – (145e) from the
underlying order 'MADM adverb > participle > completion adverb', we would
have to allow for V', i.e. the verb and its unincorporated completion adverb
complement, [V' [V ] [Adv ]], to raise to AgrPrt, which is excluded by the
Uniformity Condition on Chains. The well-formedness of (142f) – (145f) is
expected since completion adverbs, as we have seen above, do not incorporate
if they are focalized; the verb raises on its own across MADM adverbs to AgrPrt,
thus stranding the completion adverb in situ. However, a cautionary note is in
order here. The felicitousness of the pattern is highly dependent on semantic
and pragmatic factors.83 First, not all choices of completion adverbs yield satis-
factory results:
(147) a. [You] will partially look carefully for your health… (www)
b. [It] even bypass them entirely quite easily…(www)
c. I was able with a lot of patience to separate the external […] plastic
sheet very carefully partially. (www)
ADVERBS WITHIN THE LEXICAL LAYER 163
Again, these patterns are expected: (147a) – (149a) are derived from the
underlying orders in (150) by the completion adverb incorporating into the
verb and by the complex verb subsequently raising across the preverbal adverb
to v:86
3.2.2.2 Summary. In this section I argued that completion adverbs enter the
structure as complements of V which, depending on whether they are focalized
or not, follow or precede the lexical verbal element. The linear order
'completion adverb > lexical verb' was shown to be the result of adverb incor-
poration into a lexical verbal element (this is only possible in cases on non-
complexity). In passives, the postverbal occurrence of completion adverbs is a
clear case of non-incorporation. In actives, the postverbal occurrence of
completion adverbs comes in two flavors: either the completion adverb is right-
adjacent to the verb or right-adjacent to the internal nominal argument of the
verb.87 In the former case, the adverb is in its base position, in the latter, the ad-
verb is stranded in Tel/AgrO as a consequence of verb excorporation. The
conditions regulating the co-occurrences of completion adverbs and manner
adverbs were shown to follow from the analysis presented.
will become clear in the course of this section, there is ample evidence that
pure domain adverbs originate within vP.
The main characteristics of pure domain adverbs lies in their function of
restricting the domain with respect to which a given predicate is interpreted:90
Bellert (1977:347ff.) argues that the semantic function of pure domain adverbs
is identical to that of a restrictive universal quantifier, in the sense that the
proposition is true in the domain denoted by the adverb. Thus, in (151a), for
example, the truth of the proposition Callow has missed a golden opportunity
is interpreted relative to the domain of stylistics, whence the pure domain
adverb may occur within the scope of interrogative operators (152a),
counterfactual operators (152b) and sentential negation (152c):
as the following contrasts show (see also Ernst 1984:39ff.; Frey 2000:128 for
discussion):
Second, pure domain adverbs are barred from occurring between any two non-
lexical verbs. Again, this is a position freely assumed by sentence adverbs, as
the following contrasts show (cf. Ernst 1985:171ff.):
Third, pure domain adverbs (159), but not sentence adverbs (160), may precede
interrogative operators (examples from Ernst 1985:171):
Pure domain adverbs have a slightly wider distributional range than means-
domain, degree of perfection and manner adverbs.91 They may both precede
and follow the lexical verb in both passives and actives (cf. (161) – (164)),
whereas means-domain, degree of perfection and manner adverbs, in actives,
are restricted to following the lexical verb (cf. section 3.2.1 for discussion).
Thus, the crucial data are those in (163):
(161) a. Also in that country the unions have been financially supported by the
political parties… (BNC)
b. We are legally allowed to fight when protecting ourselves…(BNC)
c. [He] had been physically and mentally scarred by the accident. (BNC)
d. The Left was psychologically equipped for the reality of war. (BNC)
166 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Notice that, since pure domain adverbs denote absolute properties (i.e they are
non-gradable), the properties cannot be questioned:
For the time being, let us assume that the sentence-initial occurrence of pure
domain adverbs involves raising to the complementizer layer (but see chapter
5.2.2 for a modified analysis).92
Pure domain adverbs obligatorily precede vP-internal adverbs, as shown in
(167) and (168) for the degree of perfection adverbs in actives and passives
(recall that vP is projected only in actives):
To account for these empirical facts, I suggest situating pure domain adverbs in
the specifier position of an additional functional projection, Dom(ain)P, which
is sandwiched between AgrOP and (split) µP in actives and between AgrPrtP
and (split) µP in passives:93
(169) a. [v' [v affectedk] [AgrOP [D him] [AgrO' [AgrO tk] [DomP [Adv psychologically]
[Dom' [Dom tk] [µP [Adv badly] [µ' [µtk] [VP tk]]]]]]]]
b. [v' [v was] [AgrPrtP [DP ] [AgrPrt' [AgrPrt ] [DomP [Adv psychologically] [Dom'
[Dom ] [µP [Adv badly] [µ' [µ ] [VP affected ]]]]]]]]
As already mentioned, pure domain adverbs differ from manner, degree of per-
fection and means-domain adverbs in that they may precede the lexical verb in
actives. Since lexical verbs in actives obligatorily raise to the head position of
vP to check off their strong Θ-feature, the question arises how they come to
occur in preverbal position. Unless we want to assume that DomP can be
merged in two different positions, i.e. one within vP/AgrPrtP as in (169) and
one outside the lexical layer, as schematically indicated in (170), we have to
establish a derivational relation between the pre- and postverbal occurrence of
domain adverbs in both actives and passives:
Before turning to see how pre- and postverbal occurrences of domain adverbs
can be accounted for derivationally, I would like to very briefly present an
empirical argument against multiple merge as in (170a). Evidence against
merging DomP outside the lexical layer derives from expletive-associate con-
structions, as in (171). The pure domain adverb financially (f-ly) fails to pre-
cede [DP crooks] in specvP in actives as well as [DP families] in specAgrPrtP in
passives (cf. (171a) & (171b)). In both actives and passives, DomP must be
merged within the lexical layer, as shown in (171c) – (171f). Notice that (171d)
is ruled out because both the active verbal element and its nominal internal
argument failed to undergo overt raising to their respective targets:
168 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
In section 3.2.2, we have seen that pre- and postverbal occurrences of comple-
tion adverbs (in passives and actives) are derivationally related by adverb in-
corporation into the lexical verb. I assume that this analysis carries over to pure
domain adverbs, i.e. pure domain adverbs in both actives and passives may
incorporate into the lexical verb if warranted by defocalization. Compare (172)
with the pure domain adverb in situ vs. (173) with the pure domain adverb
incorporated into the lexical verb.
(172) a. Could youi ... [vP [D ti ] [v' [v copek] … [DomP [Adv financially] [Dom' [Dom
tk] … [VP [V tk]]]]]]
b. Theyi are... [AgrPrtP [D ti] [AgrPrt' [AgrPrt supportedk] … [DomP [Adv
financially] [Dom' [Dom ] … [VP [V tk ] [DP ti]]]]]]
ADVERBS WITHIN THE LEXICAL LAYER 169
(173) vP/AgrPrtP
wp
DP v'/AgrPrt'
ri
v/AgrPrt AgrOP
2
DP AgrO'
1
AgrO DomP
th
Adv Dom'
1
Dom …
g
VP
fg
V DP
#4
a. [D she]i ti [V [f-ly]o [supports]]k [D him]p tk to tk tk tp
b. [DP they]p were [V [f-ly]o [supported]]k to tk tk tp
3.2.3.1 Summary. Starting out with a brief survey of the distributional differ-
ences displayed by pure domain adverbs and genuine sentence adverbs on the
one hand, and the similarities displayed by pure domain adverbs and genuine
vP-internal adverbs on the other, I argued that pure domain adverbs originate in
the specifier position of DomP, which is sandwiched between AgrOP/AgrPrtP
and the set of functional projections hosting manner, degree of perfection and
means-domain adverbs. The fact that the elements under consideration have a
somewhat wider distributional range within the lexical layer than manner,
degree of perfection and means-domain adverbs was accounted for in terms of
incorporation into the lexical verb, with incorporation being warranted in con-
texts in which the pure domain adverb is to be defocalized.
(176) a. The balls of yarn that have been fading in the attic are starting to
resemble something like a sweater. (www)
b. A rabbit had been sleeping under the tree. (www)
c. [Police] arrested Mitchell walking on Telegraph Avenue (www)
(178) a. The balls of yarn that have been fading in the attic are starting to
resemble something like a sweater. (www)
b. Amazingly Gaston still hasn't seen some correlation between playing
five meters behind the line and loosing... (www)
c. The parrots have been (*there) seen there… (www)
d. We met right here.
ADVERBS WITHIN THE LEXICAL LAYER 171
The assumption that the preposition raises overtly to Deg is vital with respect
to (179a, b): had the preposition failed to raise, the resulting order would be an
illicit postpositional one (but see chapter 4.3.4.1 for temporal postpositions):
(181) a. [We are daily users] of communications devices, all of which are
carefully identified here (*carefully). (www)
b. Before an experimental agent can be evaluated …, it is carefully
studied in the laboratory. (www)
Instead of assuming that spatial expressions are merged into structure as the
complement of the lexical verb (cf. Larson 1985, 1988, 1990), I suggest they
be located within the empty verbal structure below the lexical verb, but
licensed from within the lexical layer. Since there are no preverbal occurrences
of spatial expressions, the identification of their long-distance licenser must be
based on other evidence. One such piece of evidence is provided by Frey &
Pittner's (1999:26) examples in (186):
The fact that the extraction of the spatial expression gives rise to crossover
effects is indicative of its originating below the subject, to be precise, within
the c-command domain of the base position of the subject (cf. Frey & Pittner
1999:25ff.). Thus, I tentatively suggest that the licensing projection for spatial
expressions, SpaceP, be sandwiched between vP and all object-related func-
tional projections (if projected) in actives, and above AgrPrtP in passives. By
transitivity, SpaceP dominates all functional projections hosting vP-internal
adverbs. Under this analysis, spatial expressions within VP are licensed by
hooking up to an empty spatial operator, OPspa:
(187) v'
2
v SpaceP
2
OPspa i Space'
2
Space {TelP, Agr IOP, AgrOP} in actives
{AgrPrtP} in passives
h
…
h
V'
2
V VP
2
V DegPi
ADVERBS WITHIN THE LEXICAL LAYER 173
This analysis not only accounts for the surface linear orders in (181) – (185),
but also for the fact that him and John in (188) can be interpreted as coreferen-
tial. The absence of Principle C Effects follows from the fact that the licensing
and interpretation site of spatial expressions, specSpaceP, c-commands the
bound pronoun in specAgrOP (cf. chapter 2.3.1 for discussion):
(190) a. At home, she seldom found the same tranquility to read… (www)
b. In Keni 's own garden hei has Hare Bell, which he admits he didn't
plant. (www)
c. * The man, in the attic, could have died years ago.
Spatial expressions in the left periphery follow the complementizer (191a), but
precede focalized constituents (191c):
(191) a. It must be said [ForceP that [?P in Britain [AgrSP the new public library
authorities ... have in many cases failed to capitalize on the
opportunities for better stock provision which the larger units were
supposedly able to achieve]]]. (BNC)
174 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
b. Nevertheless, [?P in France [FocP not only [Foc are] [AgrSP there ethnic
and cultural communities originating basically in the former colonies
[…], but there are also communities whose origins are found in the
regions of France […]]]]. (www)
(192) a. [ForceP [Force' [Force ] [SceneP At home [Scene' [Scene ] [FocP [Foc' [Foc ] [FinP she
seldom found…]]]]]]]
b. [ForceP [Force' [Force that ] [SceneP in Britain [Scene' [Scene ] [FocP [Foc' [Foc ] [FinP
authorities failed ]]]]]]]
c. [ForceP [Force' [Force ] [SceneP in France [Scene' [Scene ] [FocP not only [Foc' [Foc
are ] [FinP there communities]]]]]]]
3.2.4.2 Summary. Starting out with a brief survey of the distribution of spatial
expressions relative to the lexical verb, its internal arguments (if any) and all
other classes of vP-internal adverbs, I argued that spatial expressions are
invariably realized within VP, but are licensed from within the lexical layer
under chain formation with the expletive spatial operator in specSpaceP, which
was shown to be sandwiched between vP and the entire set of object-related
functional projections in actives, and above AgrPrtP in passives. Left-
peripheral occurrences of spatial expressions were argued not to be derivation-
ally related to postverbal/sentence-final occurrences, but to be integrated in the
left periphery under merge in the specifier position of a special functional
projection, SceneP.
3.2.5 Conclusion
The aim of this chapter was to propose and defend an analysis of adverbs in the
lexical layer which, at the same time, is highly restrictive in the sense that
adverbs are generated in unique specifier positions and allows for an account of
ADVERBS WITHIN THE LEXICAL LAYER 175
the various linear orders adverbs, lexical verbs and their arguments (if any)
occur in. In section 3.1, I presented the general structural assumptions about
the lexical layer underlying my analysis. Capitalizing on the idea that the verb
and its nominal internal argument(s) undergo overt raising, I suggested adopt-
ing Koizumi's (1993) Split VP Hypothesis according to which the external and
the internal argument(s) of the verb do not originate within the same minimal
domain, but are separated by intervening functional projections, notably AgrOP
and AgrIOP for direct and indirect object agreement. Overt raising within the
lexical layer was shown to be triggered by Case-requirements and the require-
ment that the lexical verb check off its strong Θ-feature. To account for the
relative linear orders of the direct and indirect object in (non)dative-shifted
double object constructions, I suggested two refinements (section 3.1.1). First,
I suggested that the indirect object in a non-dative shifted structure be treated
on a par with 'regular' nominal constituents which, for reasons of Case-check-
ing, must raise overtly to an appropriate specifier position. Second, I suggested
that a third functional projection, TelP, be interspersed between vP and VP.
The postulation of TelP, and thus the dissociation of telicity-related and Case-
related features, was shown to pave the way to an account of the differences
between dative shifted and non-dative shifted structures.
Starting out with the observation that manner adverbs, degree of perfection
adverbs and means-domain adverbs are restricted to following the lexical ver-
bal element in actives, but not in passives (sections 3.2.1.1 – 3.2.1.5), I devel-
oped a detailed analysis of the lexical layer which rests on the following inter-
related assumptions: (a) the adverbs under consideration occupy a unique
specifier position of a designated preverbal functional projection within the
lexical layer, µP, (b) the lexical layers projected by active and passive verbal
elements differ with respect to their internal makeup and (c) active and passive
verbal elements display different movement properties (section 3.2.1.5). I pro-
vided evidence that the lexical layers projected by active and passive verbal
elements not only differ with respect to the nature of the agreement-related
functional projections involved (AgrOP in actives vs. AgrPrtP in passives), but
also with respect to presence vs. absence of the vP-layer in actives and
passives, respectively. The presence of an external argument in the verb's argu-
ment structure was shown not only to be vital for the projection of the vP-layer,
but in defining the movement properties of lexical verbal elements. Since only
active verbal elements are associated with argument structures that contain an
external argument and since the strong Θ-feature in v must be checked off for
reasons of PF convergence, they are subject to obligatory overt raising to the
highest verbal head position and thus, across the adverbs in specµP so that the
adverbs unavoidably are in postverbal position. Since the argument structures
of passive verbs lack an external argument, they do not project onto vP but
only onto the agreement-related functional projection, AgrPrtP, which is the
checking domain for abstract participle agreement. Taking expletive-associate
176 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
constructions as a diagnostics, I have shown that overt raising (or the lack of it)
in passives is not subject to the requirement that morphosyntactic features be
checked, but subject to discourse requirements. The passive verbal element
raises overtly to AgrPrt (and thus across the adverbs under consideration) if the
adverb is focalized; if the verb is focalized, it stays put. To properly account for
the placement of manner adverbs relative to the verb and its nominal internal
arguments, especially the placement between the verb and its internal argu-
ment(s), I suggested to take advantage of the empty-headed verbal structure,
, which may be projected below the base position of the lexical verb
(section 3.2.1.5.1). Specifically, I argued that – in cases in which the adverb
intervenes between the verb and its internal argument – the internal argument is
realized within , but licensed from without, namely by a null antecedent,
pro, within the lexical layer, whose presence is also required for reasons of
convergence at the LF interface. The realization of the verb's internal argument
outside the lexical layer proper, i.e. within , was shown to be contingent on
discourse factors: the internal argument of the verb can only be focalized if it
occurs in a right-peripheral position. Since manner adverbs occupy a unique
preverbal specifier position, specµP, and since active verbal elements – on their
way to v – pass through Agr(I)OP obligatorily summoning the nominal internal
argument to specAgr(I)OP, the internal argument cannot be focalized since it is
inevitably followed by the manner adverb. I suggested that the realization of
the internal argument within resolves the conflict between syntactic neces-
sity and discourse requirements. After a brief discussion of the vP-boundedness
of the adverbs under consideration (section 3.2.1.5.2) and the status and
licensing of the so-called by-phrases in passives (section 3.2.1.5.3), I focused
on the co-occurrence of manner adverbs, degree of perfection adverbs and
means-domain adverbs, which constitutes a problem for the analysis of these
elements as residing in specµP. A close look at the data revealed that admissi-
ble co-occurrences of the adverb types under consideration are both limited and
highly restricted with respect to the linear order of any two elements so that the
analysis presented can be maintained with a slight modification to the extent
that µP, the designated preverbal functional projection hosting the adverbs
under consideration, be split into four discrete, intrinsically ordered functional
projections which accommodate the adverbs under consideration. In section
3.2.2, I provided a detailed analysis of completion adverbs which differ
syntactically from manner adverbs, degree of perfection adverbs and means-
domain adverbs in that they may precede the active verbal element. I argued
that the interaction of completion adverbs with the aspectual, i.e. telic proper-
ties of the verb licenses adverb incorporation into the lexical verb. Preverbal
occurrences of completion adverbs, in both actives and passives, were shown
to result from incorporation into the lexical verb (yielding [V [V ] [Adv ]]), as
triggered (or blocked) by discourse requirements and/or syntactic complexity
of the completion expression. Evidence in support of the analysis presented
ADVERBS WITHIN THE LEXICAL LAYER 177
was shown to derive from the fact that completion adverbs, together with the
verb, may precede both the preverbal specifier-type adverbs and the verb's
nominal internal argument and also from the fact that completion adverbs may
intervene between the verb's direct and indirect object. The derivation of the
latter scenario was shown to involve the excorporation of the lexical verb out
of [V [V ] [Adv ]]. I showed that the availability of excorporation is intimately
tied to the accessibility of telic features in the head position of TelP. The analy-
sis was shown to make the right predictions with respect to the co-occurrence
of completion adverbs with manner adverbs, agent-oriented adverbs, degree of
perfection adverbs and means-domain adverbs, MADM adverbs, in both passives
and actives. The fact that completion adverbs and MADM adverbs may occur in
either order if they flank the passive participle was shown to follow from the
interplay of (non)incorporation of the completion adverb and the
(de)focalization of the MADM adverb. Incorporation of the completion adverb
and focalization of [V [Adv ] [V]] was shown to be responsible for the linear
order 'completion adverb > MADM adverb' in pre-participle position. The illicit
linear order 'MADM adverb > completion adverb' in preverbal position was
shown to be underivable since it would involve a violation of both Categorial
Identity and the Head Movement Constraint, while the illicit linear order 'com-
pletion adverb > MADM adverb' in post-participle position was shown to be
excluded by Uniformity Condition on Chains. The derivation of the linear order
'MADM adverb > completion adverb', although syntactically readily available,
was shown to be subject to pragmatic constraints. In actives, the possible array
of co-occurrences of the adverb types under consideration is restricted by the
fact that MADM adverbs fail to occur preverbally. Again, the analysis presented
accounts for all and only the admissible co-occurrence patterns. The linear
order 'completion adverb > verb > MADM adverb' was shown to involve the
incorporation of the completion adverb and the subsequent raising of [V [Adv ]
[V]] across the MADM adverb to v, whereas the derivation of the linear order
'verb > completion adverb > MADM adverb' requires the excorporation of the
completion adverb in the head position of TelP. Finally, the linear order 'verb >
MADM adverb > completion adverb' was shown not to involve incorporation but
only verb raising to the head position of v. Section 3.2.3 was devoted to the
syntax of pure domain adverbs which – contrary to traditional reasoning – were
shown to originate as specifiers of the functional projection DomP within the
lexical layer of the clause. Starting out with the observation that pure domain
adverbs have a wider distributional range than MADM adverbs, i.e. pure domain
adverbs may occur within both the complementizer, I argued that pure domain
adverbs are not vP-bound in the relevant sense. To account for the fact that the
elements under consideration may both precede and follow the lexical verb in
actives and passives, but are barred from assuming positions higher than
highest position that lexical verbs may assume, I suggested that pure domain
adverbs, just like completion adverbs, incorporate into the lexical verb if
178 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Notes
1
Independent evidence for overt verb raising derives from coordination data. The fact that the
conjuncts in (i) & (ii) contain the verb's internal arguments, but not the verb, is indicative of
overt verb raising (cf. Larson 1988:345, note 11; 1990: 596ff.; Jackendoff 1990b:439ff.; John-
son 1991:485ff.; Runner 1995a:34f., 92; 2000:262; 2001:27, 30):
(i) He handed [[the napkin to her] and [the plate to him]].
(ii) He handed [[her the napkin] and [him the plate]].
2
See Larson (1988:375) for an analysis of spare-type verbs as unaccusatives.
3
Cf. Grimshaw (1990:10ff.), Jackendoff (1990a:246ff.), Hale & Keyser (1991:37ff.), Koizumi
(1993:129ff.), Radford (1997:369ff.). But see den Dikken (1995:117ff.) and Baker (1996:19ff.)
for arguments that the theme argument, in underlying structure, is higher than the goal argu-
ment.
4
Alternatively, if V first merges with the goal argument, the theme will be higher:
ADVERBS WITHIN THE LEXICAL LAYER 179
(i) [vP [DP AGENT ] [v' [v ] [VP [XP THEME] [V' [V ] [XP GOAL]]]]]
In this case, the ill-formedness of (iii) & (iv) has to be accounted for:
(ii) [AgrSP [D He]k [AgrS' … [… [vP [D tk] [v' [v handedi ] [VP [DP the napkin] [V' [V ti] [PP to her
]]]]]]
(iii) * [AgrSP [D He]k [AgrS' … [… [vP [D tk] [v' [v handedi ] [VP [DP the napkin] [V' [V ti] [D her ]]]]]]
(iv) * [AgrSP [D He]k [AgrS' … [… [vP [D tk] [v' [v complainedi ] [VP [PP about the meal] [V' [V ti] [PP
to her ]]]]]]
5
Notice that this analysis challenges the hitherto made assumption that both lexical verbs and
object DPs in English, due to the weakness of the morphosyntactic features involved, fail to
undergo overt raising.
6
See also, for example, Harley (1995), Radford (1997), Harley & Noyer (1998), Baltin (2000,
2001), Adger & Tsoulas (2000). Cf. Pesetsky (1989) for an earlier analysis of overt verb rais-
ing in English.
7
Recall that Chomsky (1995) takes the subscripts S and O on Agr to be mere mnemonics, i.e.
"there is only one element Agr, a collection of Φ-features)" (cf. Chomsky (1995:121)). On the
basis of the fundamental identity of Agr heads, i.e. Agr S and AgrO, we assume that the AgrOP,
like AgrS, hosts a strong D-feature which must be checked overtly by DP raising to specAgrP.
That is to say that the strong D-feature resides in AgrO rather than in V, as under Koizumi's
(1993) analysis.
8
Cf. Chomsky (1995) for the asssumption that Extended Projection Principle, EPP, "plausibly
reduces to a strong D-feature in [T]" (Chomsky 1995:232).
9
See also Koizumi (1993:124ff.) for arguments that, in the so-called Dative Construction as in
(i), [DP Bill] raises overtly to an agreement-related functional projection dominating AgrOP.
Koizumi (1993) takes this functional projection to be ΩP, which is related to terminative aspect
(adapted from Koizumi 1993:125):
(i) [v' [v gavej] [ΩP [DP Billi] [Ω' [Ω tj ] [AgrOP [DP a bookk] [AgrO' [AgrO tj ] [VP [DP ti ] [V' [V tj ] [DP tk
]]]]]]]]
10
Cf. Koopman & Sportiche (1985; 1991), Speas (1986), Fukui & Speas (1986), Contreras
(1987), Kuroda (1988), Sportiche (1988), Burton & Grimshaw (1992), Woolford (1991),
McNally (1992), Chomsky & Lasnik (1993).
11
Empirical evidence for the Split VP Hypothesis derives from the observation that, under the
VP-Internal Subject Hypothesis according to which the subject originates in a position lower
than AgrOP, the illicit string in (i) is predicted to be well-formed:
(i) * The meni will have givenk [AgrOP [DP a bookj ] [AgrO' [AgrO tk] [VP [QP all ti] [V' [V tk] [VP [DP
tj ] [V' [V tk ] [PP to John]]]]]]] (adapted from Koizumi 1993:131)
12
See also Kayne (1984b:114f.), Johnson (1991:587), Bowers (1993:632), Koizumi
(1993:117), Runner (1995;74ff., passim).
13
As Lasnik (1999a:158) points out, "even the best instances of Pseudogapping are somewhat
degraded." Lasnik (1999a:164ff.) argues that multiple object shift, as in (i), clearly results in
degradation:
(i) * John gave Bill a lot of money, and Mary will give Susan a lot of money.
14
See also Chomsky (1995:264) for comments on feature raising, as triggered by strong fea-
tures, without pied-piping.
15
However, Lasnik (1999a:161; 1999b:204) locates the strong Θ-feature in V, not in v. But see
Lasnik (1999b:213) for alternatively locating the strong feature in v.
16
Cf. Larson (1988:345, note 11; 1990:624ff.), Jackendoff (1990b:439ff.), Johnson
(1991:485ff.) for VP coordination and Runner (1995:34f., 92; 2000:262; 2001:27, 30) for AgrP
coordination.
180 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
17
See Larson (1990:624ff.), Johnson (1991:485ff.) and Runner (1995:34f., 92; 2001:27, 30)
for analyses under which overt verb movement proceeds Across The Board:
(i) [v' [v handedi] [:P [AgrOP [DP the napkin]j [AgrO' [AgrO ti] [VP [PP to her] [V' [V ti] [DP tj]]]] [:' [:
and] [AgrOP [DP the plate]k [AgrO' [AgrO ti] [VP [PP to him] [V' [V ti] [DP tk]]]]]]]]
(ii) [v' [v handedi] [:P [AgrIOP [D her]k [AgrIO' [AgrIO ti] [AgrOP [DP the napkin]j [AgrO' [AgrO ti] [VP [D
tk] [V' [V ti] [DP tj]]]]]]] [:' [: and] [AgrIOP [D him]m [AgrIO' [AgrIO ti] [AgrOP [DP the plate]o [AgrO'
[AgrO ti] [VP [D tm] [V' [V ti] [DP to]]]]]]]]]]
(iii) [v' [v complainedi] [:P [VP [PP to her] [V' [V ti] [PP about the meal]]] [:' [: and] [VP [PP to him]
[V' [V ti] [PP about the wine]]]]]]
18
Cf. Jackendoff (1990b:439ff.) for an analysis of VP-coordination involving gapping.
19
In the second conjunct, there is no AgrSP projected since deleted D(P)s do not require Case.
20
Cf. Williams (1989:448), Emonds (1985:224ff.; 1987a:617f.), Rauh (1993a, b, 1995, 1996,
1997, 2002a, b), Rooryck (1993), Zwarts (1997), van Riemsdijk (1998:27), Koopman
(2000:247ff.).
21
Note that Bittner & Hale's (1996) analysis radically differs from Rauh's (1993a; 1996;
2002b) in that KP is an integral part of extended nominal projections carrying either a marked
structural Case or inherent Case, whereas 'a nominal in the unmarked Case (nominative) is K-
less, a bare DP or NP' (cf. Bittner & Hale 1996:5f.).
22
But see Radford (no date, note 7): "It is far from clear that it makes sense to say that (e.g.)
inherent dative/genitive case is encoded via the preposition to/of, for the obvious reason that
the complement of to/of surfaces with accusative (not dative/genitive) case."
23
For a discussion of the selectional requirements on prepositional items as imposed by the
governing lexical head see also Grimshaw (1991:3), Rauh (1993b:262ff.), Bittner & Hale
(1996:6), Neeleman (1997:114f.), van Riemsdijk (1998:30), Koopman (2000: 247f.). As has
been argued by van Riemsdijk (1998:26ff.), Grimshaw's (1991:3) analysis, under which prepo-
sitions, in general, are taken to figure in the extended nominal projection, fails to account for
(i), where the alleged extended nominal projection, i.e. PP, is not selected (data from van
Riemsdijk 1998:27):
(i) Before the war, life was much better.
Moreover, the fact that prepositions may be intransitive, as in (ii), is problematic for Grim-
shaw's (1991) analysis (data from van Riemsdijk 1998:27):
(ii) He stayed inside for three years.
These data clearly show that Grimshaw's (1991) analysis of prepositions as functional heads
within the extended nominal projection is not tenable for prepositions other than Case-preposi-
tions. Cf. Riemsdijk (1998:27ff.) for further arguments against Grimshaw's (1991) analysis.
24
Rauh (1996:189f.; 2002b:18) maintains that structural Case as assigned by Case-preposi-
tions is checked under specifier-head agreement within KP, i.e. by covert raising of DP from
the complement to the specifier position of KP. Notice, that – within the frameworks of The
Theory of Government and Binding and The Theory of Principles and Parameters – an analysis
along these lines would constitute a violation of Stowell's (1981) Case Resistance Principle,
according to which "Case must not be assigned to a category bearing a Case-assigning feature"
(cf. Stowell 1981:146). To avoid this problem, we could assume with Koopman (2000:247ff.)
that semantically empty prepositions, such as Case-prepositions, project functional architec-
ture. See also van Riemsdijk (1998) for arguments that prepositions "constitute a category sui
generis" in the sense that the "relevant phenomena, i.e. the external distribution and the internal
syntax of their projections, hold of 'functional PPs' and 'lexical PPs' alike" (van Riemsdijk
1998:27). Under such an analysis, the internal structure of KP, i.e. nominal constituents con-
taining a Case-preposition, would be as in (i), where the Case-preposition is labeled P:
(i) [V' [V complain] [FP [F ] … [PP [P about] [DP the meal]]]]
ADVERBS WITHIN THE LEXICAL LAYER 181
This analysis, in conjunction with the overt object shift analysis advanced here, Case-preposi-
tions would have to be dominated by at least two functional projections. The functional projec-
tion immediately dominating P would have to be an agreement-related projection, AgrOP, in
which Case is checked under specifier-head agreement. This projection would have to be
dominated by a functional projection the head of which is overtly targeted by P, as shown in
(ii):
(ii) [V' [V complain] [FP [F aboutj] [AgrOP [DP the meal]i [AgrO' [AgrO tj ] [PP [P tj ] [DP ti ]]]]]]
This analysis gives rise to at least two questions, which I leave for further research: first, what
is the nature of the second functional projection and second, what is the trigger for overt P-
raising to the head position of this functional projection?
25
Supportive evidence for the claim that to her and about the meal in (i) & (ii) vacate VP
derives from pseudogapping data as in (iii) – (vii) (cf. 3.1 for coordination data):
(i) He handed [DP the napkin] [KP to her].
(ii) He complained [KP to her] [KP about the meal].
(iii) He handed [DP the napkin] [KP to her], but he didn't hand [DP the knife] [KP to him].
(iv) If you talk [KP to my neighbors], I will talk [KP to yours].
(v) ? Tom complained [KP to Sue] [KP about the dirt] and Bob complained [KP to Kim] [KP
about the dirt].
(vi) ? Al complained [KP about the dirt] [KP to Tom] and Bob complained [KP about the dirt]
[KP to Tom]
(vii) She gave a book clumsily [KP to him] and he gave a spoon gracefully [KP to her].
26
Cf. Koizumi (1993:125), Sato (1995:100ff.), Solà (1996:227), Adger & Tsoulas
(2000:12ff.). See also Culicover (1982), Tenny (1987), Pustejovsky (1988, 1992) and Travis
(1991).
27
Sato (1995) assumes that the theme argument is higher than the goal argument.
28
Sato's (1995:102f.) analysis of double object verbs is more complex. He follows Keyser &
Roeper (1992) in assuming that "verbs in English are associated with a category-neutral
abstract clitic position" (Sato 1995:102), which – in dative shifted structures – is either spelt
out as to and licensed as the head of PP (here KP), or to is not phonetically realized and
licensed by the head of TelP.
29
Cf. Solà (1996:227) for arguments that the extended verbal projection in English – due to the
lack of agreement morphology – does not contain an AgrOP, but an aspectual phrase,
Tel(icity)P, instead.
30
Cf. Pesetsky (1989), Johnson (1991), Harley (1995), Radford (1997), Harley & Noyer
(1998), Baltin (2000, 2001), Adger & Tsoulas (2000), Runner (1995, 2000, 2001), Lasnik
(1995, 1999a, b, 2002) for arguments.
31
Cf. Lasnik (1999a:161; 1999b:204).
32
See also Adger & Tsoulas (2000:14) for the dissociation of aspectual and Case-features.
33
Recall from chapter 2.3.2 that adjunction analyses fail to account for this asymmetry.
34
See also Emonds (1976) for the qualification that manner adverbs in preverbal position do
"not always seem to be an exact paraphrase of the postverbal manner [adverb]" (Emonds
1976:157).
Notice that, according to Cinque's (1999:106) classification of adverbs, slowly – alongside
quickly and fast – counts as an aspectual adverb located in the specifier position of AspPcelerative.
This functional projection is split into AspPcelerativeI and AspPcelerativeII, with the former, i.e.
quickly in (42b), assuming a relatively high position and the latter, i.e. quickly in (41b), a rela-
tively low position. Evidence for this analysis derives from the fact that slowly and quickly in
182 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
pre- and postverbal position are subject to interpretational differences ((i) & (ii) from Thoma-
son & Stalnaker 1973:200):
(i) He has been slowly testing some bulbs.
(ii) He has been testing some bulbs slowly.
In (i), slowly modifies the entire testing procedure as being carried out slowly. That is to say
that the person testing the bulbs is slow in the overall process (due to extended coffee breaks,
though individual bulbs could have been tested rather quickly (Thomason & Stalnaker
1973:200), whereas in (ii) slowly modifies the process of testing bulbs, i.e. every single bulb is
tested slowly (the person testing the bulbs could have quickly accomplished the entire testing
procedure).
35
See also Ernst (1984:240, 272ff.; 2004a: 770ff.) and Pittner (2000a) for discussion.
36
Apologies to Shakespeare.
37
See Jackendoff (1972:64), Alexiadou (1997:130), Cinque (1999:101f.) and Ernst
(2002:274).
38
Cf. Steinitz (1969:20ff.) and Goldberg & Ackerman (2001) for related arguments.
39
Cf. Alexiadou (1997:130) for related arguments.
40
Well is special in so far as it is ambiguous between a quality reading, as in (i) & (ii), and a
quantity reading, as in (iii) & (iv) (data from Cattell 1998:54, 61):
(i) The ball was well fielded by the bowler.
(ii) The ball was fielded well by the bowler.
(iii) The shirts were well advertised by the shop.
(iv) The shirts were well advertised [(*well)] by the shop. [my insertion, DH]
See also Cinque (1999:101ff.) for the distribution of bene 'well' in passives.
41
See Jackendoff (1972:82ff.), Bellert (1977:339ff.), McConnell-Ginet (1982:145ff.), Roberts
(1987:81ff.), Ernst (2002:106ff.) for discussion.
42
The fact that beautifully, horribly, etc. in (58) are systematically barred from preceding the
finite lexical verb and (non)-finite non-lexical verbs strongly suggests that they do not have
subject-related counterparts.
43
As has been the case with degree of perfection adverbs, means-domain adverbs are barred
from preceding finite lexical verbs as well as all non-lexical verbs, i.e. they do not have sub-
ject-related counterparts:
(i) * They surgically (have) obtained brain tissue via biopsy.
(ii) * They telepathically (have) communicated.
However, Ernst (2002:282) gives the following example in support of his claim that means-
domain adverbs may precede the lexical verb:
(iii) In this film, straight arrow FBI agent Travolta surgically trades faces with arch enemy
[…] Cage.
44
Caponigro & Schütze (2003:297) do not mention that degree of perfection adverbs may
follow the passive participle. We shall come back to this issue below.
See also Blight (2000) for related arguments. Under Blight's (2000:4f.) analysis, degree of
perfection adverbs are either left- or right-adjoined to VP, which – since passive verbs do not
undergo overt syntactic movement – accounts for the pre- and postverbal occurrence of the
adverbs under consideration. The fact that the adverbs are restricted to occurring postverbally
in actives is seen as a consequence of the active verb raising overtly to the head position of the
functional projection dominating VP, i.e. PredP as under Bowers' (1993) analysis.
45
Caponigro & Schütze (2003) attribute the fact that all finite lexical verbs and active partici-
ples precede their nominal internal argument to the presence of VoiceP in all active clauses.
ADVERBS WITHIN THE LEXICAL LAYER 183
46
Caponigro & Schütze (2003) do not address postverbal degree of perfection adverbs in
passives (cf. note 44 and below).
47
See among others Koizumi (1993:135ff.), Lasnik (1995:626ff.), Chomsky (1995:341ff.),
Radford (1997:454ff.), Caponigro & Schütze (2003:293).
48
Belletti (1998:12f.) conjectures that the severing of Case-checking from the checking of
other ϕ-features could be extended to other AgrPs, e.g. AgrSP.
49
Belletti (1998:12f.; 2001:2) labels this head AgrPstPrtP. See also Friedemann & Siloni (1997)
for arguments that participle agreement is checked in a specific agreement projection, AgrPP.
50
A seeming alternative to this solution would be to assume that passive verbs project onto vP,
but that the Θ-feature residing in v is weak. This analysis, basically, would also allow for pos-
iting PRO in specvP in passives (cf. Ernst 2002:107, passim). However, given that PRO as a
nominal element is subject to Case-checking, we would have to assume an additional agree-
ment projection above vP.
51
The fact that the adverbs cannot precede both the passive participle and the nominal internal
argument clearly shows that the nominal internal argument has vacated its base position:
(i) * There have been telepathically messages sent.
(ii) * There have been horribly arias performed.
(iii) * There has been surgically a nose removed.
52
Under Ernst's (2002:276) analysis, the postverbal degree of perfection adverb would take
immediate scope over the trace of the verb.
53
See, among others, Jackendoff (1972:230), Jacobs (1983:19ff.; 1984:103ff.), Rochemont
(1986:129), Rochemont & Culicover (1990:19ff.), König (1991a:7), Bayer (1996:15f.;
1999:56), Erteschik-Shir (1997:11).
54
See also Lonzi (1990:151; 1991:358), Alexiadou (1997:135ff.), Cinque (1999:8, 14), Ernst
(1984:240, 272ff.) for a discourse-based analysis of manner adverb placement.
55
But see Koizumi (1993:109ff.) and Runner (1995a:93f.) for a base-adjunction analysis under
which adverbs are merged with VP (or any projection of V) ((i) adapted from Runner
(1995a:93)):
(i) [vP [D Cindy] [v' [v showedi] [AgrOP [DP her book]j [AgrO' [AgrO ti ] [VP [Adv quietly ] [VP [PP
to the boys ] [V ' [Adv quietly ] [V ' [V ti ] [DP tj ]]]]]]]]]
(ii) [vP [D He] [v' [v complainedi] [VP [AdvP vociferously] [VP [PP to her] [V' [AdvP vociferously]
[V' [V ti] [PP about the meal ]]]]]]
While this analysis provides a straightforward account of double complement constructions, as
in (83), it fails to account for the ungrammatical string in (iii), where the adverb immediately
precedes D(P), unless we follow Koizumi (1993:110) and Runner (1995a:94) in assuming that
adjunction to agreement-related projections is illicit:
(iii) [vP [D He] [v' [v handedi ] [AgrOP [AdvP quietly] [AgrOP [DP the napkin]j [AgrO' [AgrO ti] [VP [PP to
her] [V' [V ti] [DP tj ]]]]]]]]
56
Notice that, in the framework adopted here, adjunction to VP, as shown in (i), is not an op-
tion, since "for every moved phrase, [there must] be a distinct head to whose projection it can
adjoin as specifier" (Kayne 1994:30):
(i) [vP [D ] [v' [v ] [VP [AdvP ]i [VP [DP/KP] [V' [V ] [VP [AdvP ]i [VP [DP/KP] [V' [V ] [AdvP ti ]]]]]]]]]
57
This follows from the Head Movement Constraint according to which the verb, on its way to
the highest verbal head position, passes through all intervening verbal head positions.
58
The irrelevance of adjacency of a Case-assigner and a Case-assignee is most obvious in (i)
& (ii), where an AdvP intervenes between the subject and the finite verb (see also Roberts
1987:74; Johnson 1991:579f.; Koizumi 1993:101ff.; Runner 1995a:45ff.):
(i) He deliberately lit his cigarette clumsily.
184 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
78
If the lexical verb takes only one argument, the telicity features are represented in AgrO.
79
Cf. Ernst (2000:45f.; 2002:286) for non-syntactic factors restricting the co-occurrence
options for the adverb types under consideration.
80
Corpora abound with strings as in (i) & (ii), which I take not to involve adverb coordination,
but coordination of some higher chunk of structure, i.e. AgrPrtP in (i) and AgrPrtP or VP in (ii):
(i) [Fine] hair, fibers, and scraps of paper, which are impossible to be removed by a
vacuum cleaner, can be easily and completely removed. (www)
(ii) [They are] able to repeat verbatim a spoken message using silent articulation that is
clear and accurate so as to be completely and easily understood… (www)
Coordination of just the adverbs is impossible since the elements under consideration originate
in different positions and are subject to different licensing conditions, whence they cannot
jointly precede active verbal elements, as shown in (iii):
(iii) * This vac {easily and completely/completely and easily} removes even hair.
81
On the basis of strings as in (i), where poorly written has a compound reading, e.g. the most
poorly written paper, one might wish to argue that completion adverbs and degree of perfection
adverbs may occur in pre-participle position:
(i) The paper was partially poorly written.
82
(145c) is acceptable if read as 'his nose will have to be removed by the exclusive application
of surgical means'.
83
As has also been argued by Ernst (2002:288, passim) in connection with the co-occurrence
of the adverbs under consideration, semantic and pragmatic factors may be severe so as to
outrank syntactic factors. A case in point is (i), where the completion adverb seems to have
independently moved to AgrPrt:
(i) His nose can be removed entirely (only) surgically, but not telekinetically.
The structure of (i) would be one in which the internally complex constituent (only) surgically,
but not telekinetically is realized within , but licensed from specMeansP.
84
Agent-oriented adverbs in actives are not relevant here since they originate outside vP. If,
however, the agent-oriented adverb in actives is realized in , we get the following:
(i) Investors completely ruined the house deliberately.
(ii) * Investors ruined the house completely deliberately. ( = the house is completely ruined)
(iii) * Investors ruined the house deliberately completely.
85
Notice that strings as in (i) are irrelevant here since carefully is a subject-related adverb (i.e.
qualifies the attitude of the unexpressed subject of the imperative):
(i) Carefully partially lift (only enough to insert the loop) the lid to your bacterial plate and
gently rub the loop over the surface of the plate. (www)
86
In addition, the derivation of (147a) involves the realization of the KP internal argument in
the empty VP structure below the base position of the lexical verb.
87
Under this analysis, completion adverbs, like manner adverbs, degree of perfection adverbs
and means-domain adverbs, are restricted to occurring in the lexical layer.
88
Ernst (1985) states that "in the case of the biggest anomaly – that of sentential adverbs
occurring inside the VP – we are in fact dealing with sentence-modifiers" (Ernst 1985:172).
89
See, among others, Bellert (1977:347ff.), Quirk & Greenbaum (161986:211) and Cinque
(1999:11ff.) for the sentence adverb status of the elements under consideration.
90
See Bellert (1977:347ff.), Ernst (1984:39ff.; 1985; 2002:260ff.) and Cinque (1999: 11ff.).
Pure domain adverbs are also often referred to as 'viewpoint adverbs' (cf. Quirk & Greenbaum
16
1986:211), 'frame adverb(ial)s' (cf. Parsons 1990:211ff.) or 'adverbs of setting' (cf. Ramat &
Ricca 1994:313).
91
See also Ernst (2004a:770ff.) for discussion.
186 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
92
See Ernst (2000) for the assumption that pure domain adverbs freely prepose across other
adverbs (from Ernst (2000:40)):
(i) Physically and psychologically, Jordan definitely can really work you over.
93
But see Ernst (2000:42; 2002:284; 2004b:119f.) for the qualification that the order of pure
domain adverbs and manner adverbs is free. I assume that linear orders of the type domain
adverb > manner adverb, as in (i), result from merging the domain adverb into structure within
(data from Ernst 2000:42):
(i) Emerging democracies evolve [rather slowly] [politically].
94
Under non-shell analyses of prepositions, AgrOP would dominate PP, and both DP-raising to
specAgrOP, and P-raising to AgrO would take place at LF (cf. Rooryck 1993:2; Haumann
1997:206, 211ff.; 2002:28ff.; Rauh 1996:191, 195; 2002b:9, 18).
95
The lexical layer projected by predicative spatial prepositions contains an outer shell, pP, the
specifier of which hosts the preposition's external argument, as shown in (ii) for (i):
(i) The bats are in the attic.
The nominal internal argument of the in, i.e. [DP the attic], raises overtly to specAgrOP to have
its Case checked under specifier-head agreement. The preposition raises via AgrO to the head
position of the outer shell to assign its Θ-feature to the constituent in specpP, [DP the bats], and
then further to the head position of DegP. The external argument of the preposition, [DP the
bats], must vacate its base position and raise to an appropriate specifier position within the
extended verbal projection, specAgrSP, for reasons of Case-checking:
(ii) [AgrSP [DP The bats]o [AgrS' [AgrS are] … [DegP [Deg' [Deg in k ][pP [DP to][p' [p tk] [AgrOP [DP the
attic]i [AgrO' [AgrO tk][PP [P tk] [DP ti]]]]]]]]]]
96
DegP roughly corresponds to Zwarts' (1992:202) RP, Napoli's (1993:217) Int(ensifier)P,
Koopman's (1993; 2000) PlaceP and DegP or Rauh's (1996, 2002b) R(aum)P and
Ext(ension)P.
97
Cf. Cinque (1990:93), Frey & Pittner (1999:26) for discussion.
CHAPTER 4
ADVERBS WITHIN THE INFLECTIONAL
LAYER, MOSTLY
Modal verbs and auxiliary verbs differ with respect to inflectional properties.
While the inflectional paradigms of auxiliaries contain finite and non-finite
forms, the inflectional paradigms of modals contain only finite forms (cf.
Roberts 1985; 1993). As a consequence, modal verbs (but not auxiliary verbs)
are precluded from occurring in the complement position of any verbal head
and after the infinitival marker to:2
(5)
have
be
bepass
verbs
lexical
modals
As can be seen from (5), the paradigms of auxiliaries differ with respect to
whether they contain a past and a present participle (see also Warner
1993:35ff.). The fact that the paradigms of auxiliary verbs do not contain a
passive participle derives from the fact that non-lexical verbs lack an argument
structure.
The paradigm of the auxiliary have contains neither a past participle nor a
present participle. While there is no morphological form of have corresponding
to past participle, the extant form having, as illustrated in (6a), does not have
the distribution of a present participle, i.e. it is not governed by be. In (6b),
having is governed by be, but having, in this case, is the present participle of
the homophonous lexical verb have:
(7) a. [He] had been planning that warning for some time. (BNC)
b. [They] use quality as an excuse for not being on time. (BNC)
190 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Given our standard background assumption that lexical items enter syntax fully
inflected and, at some point in the derivation, must have their morphosyntactic
features checked, and taking into account the various morphological forms
contained within verbal paradigms and the differences that exist across verbal
paradigms, as shown in (5), each auxiliary, i.e. have, be and bepass, and each
participle should represent a functional category of their own. Thus, the struc-
ture underlying the string in (9), in addition to the head positions into which
modal and auxiliary verbs are merged, must contain functional head positions
which serve as checking sites for all and only those verbal elements that have
morphosyntactic features to check, i.e. finite verbal elements and the participial
forms of auxiliary and lexical verbs ((9) from Emonds 1976:209):
(9) Our houses might have been being undermined for a long time.
As we have seen in chapter 3.2.1.5, the checking site for passive participles is
AgrPrtP, which is contained within the lexical layer, i.e. the thematic domain
projected by the lexical verb. Since non-lexical verbs are a-thematic heads,
their paradigms do not contain a passive participle which would warrant the
postulation of an adequate checking site within the inflectional layer (or,
absurdly, allowing for lowering into the lexical layer).6 Since the paradigms of
both lexical verbs and the auxiliary verbs bepass and be, but not have, contain
participial forms which need checking, it seems reasonable to assume that the
checking sites are located between the functional projection headed by have
and that headed by bepass. As can be seen in (10), there is a correlation between
the 'morphological richness' of the paradigms of verbal elements and the posi-
tion they assume within hierarchical structure, i.e. the fewer forms the para-
digm of a given verbal elements (potentially) contains, the higher its position.
ModP is headed by modal verbs. Relt stands for "Relative Tense" as lexical-
ized by have which introduces an additional temporal reference point relative
to which the event is located in time (cf. Solà 1996:227). Perf(ective)P and
Prog(ressive)P, are the checking sites for the past and the present participles,
respectively. Due to the distinct grammatical properties of the phonological
form be I assume two distinct functional categories, Aux and be. While the
head of AuxP hosts the aspectual auxiliary be (7a), the head of beP hosts the
passive auxiliary bepass (8a, b):
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 191
(10) TP
2
T ModP
2
Mod ReltP
2
Relt PerfP
2
Perf AuxP
2
Aux ProgP
2 lexical
Prog beP layer
2
be AgrPrtP
2
AgrPrt VP
4
participle
participle
participle
passive
bepassive
present
lexical
modal
[±fin]
have
verb
past
be
to
We have seen in chapter 3.2.1.5 that the lexical verb cannot raise beyond the
head position of vP in actives and the head position of AgrPrtP in passives.
Moreover, we have seen that overt raising in passives is not conditioned by a
strong feature in AgrPrt which would pied-pipe the phonetic form of the passive
participle, but by focalization. Given the uncontroversial assumption that the
lexical verb, irrespective of genus verbi and irrespective of its finiteness
properties, cannot leave the lexical layer, the checking of the morphosyntactic
features defining the present and the past participle must be achieved under
pure feature raising to the head position of ProgP and PerfP, respectively. Since
non-lexical verbs are not domain-bound, I assume that the checking of the
morphosyntactic features defining the present and the past participle involves
overt raising. That is to say that the present participle of bepass raises to the
head position of ProgP and the past participle of both bepass and be raises to the
head position of PerfP. The various covert and overt raising operations are
schematically indicated in (12) for the strings in (11), with the dotted arrows
indicating pure feature movement:
(12) TP
2
T ModP
2
Mod ReltP
2
Relt PerfP
2
Perf AuxP
fu
Aux ProgP
fi lexical layer
Prog beP
tu
be AgrPrtP/vP
5
a. wouldi ti have beenj tj blown
b. arek tk [PROG] leading
c. hadm tm [PERF] gained
d. wask tk beingj tj done
e. hadm tm beenj tj taken
f. hadm tm beenj tk [PROG] crying
In the remainder of this section, I will show that ReltP serves a double func-
tion: it accommodates the perfect auxiliary have and it introduces an additional
temporal reference point relative to which the event denoted by vP or AgrPrtP is
located in time.
As is pointed out by Solà (1996:227), have in the head position of ReltP
conveys the meaning that the E(vent time) precedes the R(eference time), i.e.
the event denoted by [vP t killed the spider] in (13) is interpreted as having
taken place prior to the point in time denoted by have:
Solà (1996:227) further conjectures that have has a null counterpart. In this
case, the default value is conveyed: E and R are congruous (E, R).7 The
assumption that ReltP is present in all clauses allows us to account for the fact
that temporal adverbs introduce temporal reference points with respect to
which a given event is located in time. This assumption is also at the heart of
Zeller's (1994) syntactic analysis of tense. He argues that reference time, R,
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 193
can be lexicalized by temporal adverbs which are located in the specifier posi-
tion of the functional projection ZP (for 'Zeit', event time) which is sandwiched
between TP and vP. Under his analysis, reference time, R, and speech time, S,
are represented as specifiers of ZP and TP, respectively ((14) adapted from
Zeller 1994:228ff.):8
Zeller (1994:228ff., 257ff.) maintains that the temporal location of events rela-
tive to S is mediated through R, which, in the default case, is not lexicalized
but represented by PRO, which is contextually controlled.9 However, if specZP
is lexicalized by e.g. once, the event denoted by AgrPrtP is interpreted relative
to the point in time denoted by once:
(15) TP
2
S T'
2
T ZP ( = E)
tu
Adv Z'
2
Z beP
2
be AgrPrtP
6
North Rona wasi once ti ti populated but is now deserted (BNC)
Abstracting away from labels and potential difficulties,10 I will assume that
Solà's (1996) ReltP fulfills exactly the functions of Zeller's (1994) ZP. That is
to say that ReltP is always projected with specReltP providing a syntactic posi-
tion R as either contextually inferred or realized by temporal adverbs (cf.
chapter 4.3.4).
4.1.1 Summary
In this section, I showed that the internal make-up of the inflectional paradigms
of non-lexical verbs and the necessity that morphosyntactic features of parti-
cipial elements be checked warrants the interrelated assumptions that modal
and auxiliary verbs head their own projections and that there are additional
functional projections the head positions of which are overtly targeted by the
participles of auxiliary verbs (provided the paradigm of the auxiliary verb con-
tains a participle). Drawing on work by Solà (1996) and Zeller (1994), I capi-
talized on the idea that clauses, irrespective of whether or not they contain
194 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
have, project ReltP the specifier position of which serves as the site where
reference time is encoded.
It is generally held that English has two syntactically distinct negative markers:
not and n't, with not being located in specNegP and n't in Neg:12, 13
(17) a. She could [NegP not [Neg [Neg ] [TP eat strawberries]]].
b. She has [NegP not [Neg [Neg ] [TP eaten strawberries]]].
(18) a. She could [NegP [Neg [Neg n't] [TP eat strawberries]]].
b. She has [NegP [Neg [Neg n't] [TP eaten strawberries]]].
Arguments in favor of analyzing not as the specifier of NegP rather than its
head derive from the lack of Minimality violations (cf. Rizzi 1990). If not
occupied the head position of NegP, not would either block the raising of the
finite non-lexical verb to AgrS or, if the Head Movement Constraint were
circumnavigated, not would count as a closer potential governor for the trace
left behind in T.14, 15
N't in (18) does not block the raising of auxiliary and modal verbs to Agr,
since n't is a phonetically reduced element that needs to attach to – or
incorporate into – a lexical carrier, i.e. the finite non-lexical verb that (on its
way to AgrS) passes through Neg. Evidence for incorporation of n't into the
finite non-lexical verb derives from the fact 'negative' verbal elements assume a
head position in matrix alternative questions:
In the context of sentential negation, the Affect Criterion in (20) must be met in
overt syntax, i.e. the configuration in (21) must obtain prior to LF:
(23) a. ? Whoi did you say that we will not/won't meet ti at the opera?
b. * Howi did she suggest that he should not/shouldn't drive ti?
In the following, I assume with Williams (1994:168, 174) that haven't, aren't,
couldn't, etc. are genuine lexical items, i.e. morphological products, which I
take to be selected from the lexicon and merged into their respective base
positions, i.e. the head positions of ModP, ReltP, AuxP and beP (cf. section 4.1
for discussion). This move not only is fully compatible with the current
framework, but also allows us to treat n't as a 'syncategorematic' derivational
suffix. N't, as opposed to run-off-the-mill derivational suffixes, e.g. –ness, –ly,
–able, etc. in (25), does not change the category of the element it attaches to
((25) adapted from Scalise 1988:230):
196 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Rather, n't resembles so-called evaluative suffixes, e.g. –ino, –etto, –one, etc.,
which, according to Scalise (1986:131ff.; 1988:233ff.) are 'transparent' with
respect to categorial information (as the paraphrases show, evaluative affixes
change the semantics of the base):
Under the assumption that the requirement that n't combine with finite non-
lexical verbs is expressed in the lexical entry of that suffix, the internal
structures of negative auxiliary and modal verbs, such as haven't, aren't and
couldn't, can be given as in (29), where the categorial features of the non-head
are percolated onto the dominating node:
The analysis proposed does not affect the analysis of negative sentences as
containing a NegP if we assume that the head of NegP – by virtue of the
negative non-lexical verbs, on their way to Agr and beyond, passing through
Neg – is endowed with the negative feature [NEG], thus entering into a
configuration in which the Affect Criterion in (20) above can be satisfied in
overt syntax. Notice that this analysis is compatible with both Rizzi's
(1997:317f.) proposal, according to which the [NEG]-feature is realized on T (or
any other functional head position within the extended verbal projection into
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 197
4.2.2 Summary
In this section, I focused on the functional projection NegP and the analysis of
sentence-negating not and n't, both of which traditional grammar classifies as
negative adverbs.21 With respect to not, I followed the standard routine of ana-
lyzing this element as an overt negative operator in specNegP (cf. Haegeman
1995; Rowlett 1995; Rizzi 1996) which is subject to meeting the Affect
Criterion in overt syntax. As for the analysis of n't, I suggested haven't,
couldn't, etc. be considered lexically negative verbal elements that are merged
into their respective base positions, e.g. [Relt have] and [Mod modals]. This
analysis was shown to be compatible with the analysis of negative sentences as
containing a NegP, since the negative verbal elements under consideration, by
passing through Neg, endow the head of NegP with the negative feature [ NEG]
so that the Affect Criterion can be satisfied in overt syntax.
This move, however, is highly undesirable since it would involve overt raising
operations, which, on top of being optional, would have to illicitly proceed
across verbal head positions, as is the case in (36b), where a lower verbal head,
have, raises across the trace of a higher verbal head [Mod t] and [T t]. The more
rightward the adverb occurs, the more head-crossing operations must have
taken place. Thus, the derivation of (36c) would involve the raising of have
across [Mod t] and [T t] as well as the raising of been across [Relt t], [Mod t] and [T
t].31 Clearly, these problems would not arise under the adjunction analysis
since cleverly could be adjoined to any of the functional projections
constituting the inflectional layer, as shown for the strings in (35):32
(37) a. [AgrS' [AgrS hasi ] [TP [Adv cleverly] [T' [T ti] [ReltP [Adv cleverly] [Relt' [Relt ti]
[PerfP [Adv cleverly] [Perf' [Perf beenj] [AuxP [Adv cleverly] [Aux' [Aux tj] [ProgP
[Adv cleverly] [Prog' … Ving ]]]]]]]]]]]
b. [AgrS' [AgrS willi ] [TP [Adv cleverly] [T' [T ti] [ModP [Adv cleverly] [Mod'
[Mod ti] [ReltP [Adv cleverly] [Relt' [Relt have] [PerfP [Adv cleverly] [Perf' [Perf
beenj] [AuxP [Adv cleverly] [Aux' [Aux tj] [ProgP [Adv cleverly] [Prog' …Ving
]]]]]]]]]]]]]
(39) a. [AgrS' [AgrS has]i [NegP not [Neg' [Neg ti] [TP [Adv cleverly] [T' [T ti] [ReltP [Rel
ti ]]]]]]]]
b. [AgrS' [AgrS has]i [NegP [Adv cleverly] [NegP [not] [Neg' [Neg ti] [TP [T ti] [ReltP
[Relt ti ]]]]]]]
c. [AgrS' [AgrS has]i [TP [Adv cleverly] [T' [T ti ] [NegP not [Neg' [Neg ti] [ReltP [Relt
ti ]]]]]]]
Since the analysis in (39b) would leave us with two A'-specifiers within NegP,
which is clearly an undesirable state of affairs, the system will have to make do
with the analysis in (39c). Notice that (39c) allows for the Affect Criterion to
be met in overt syntax since has, on its way to AgrS, passes through Neg and
thus enters into a specifier-head agreement configuration with not.
It has often been observed that sentential negation preceding the subject-
attitude adverb, as in (40), is better with contracted negation than with not (see
also Ernst 2002:105). The preference for contracted negation in (40a) over
negation by not in (40b) follows from the potential ambiguity of (40b) between
sentence and constituent negation. That is to say that the structure of (40b) can
be either that in (41a) or in (41b):
(41) a. [AgrS' [AgrS has]i [NegP [NegP not [Neg' [Neg ti] [ReltP [Adv cleverly] [Relt' [Relt ti
…]]]]]]
b. [AgrS' [AgrS has]i [ReltP [Adv not cleverly] [Relt' [Relt ti …]]]]
(42) [AgrS' [AgrS hasn't]i [NegP Op [Neg' [Neg ti] [TP [T ti] [ReltP ti …]]]]]
204 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(45) a. [AgrS' [AgrS ] [TP [Adv cleverly] [T' [T hasi] [ReltP [Relt' [Relt ti] [PerfP [Adv
cleverly] [Perf' [Perf beenj] [AuxP [Adv cleverly] [Aux' [Aux tj] [ProgP [Prog' [Prog ]
[vP Ving ]]]]]]]]]]]]
b. [AgrS' [AgrS willi] [TP [Adv cleverly] [T' [T ti] [ModP [Mod' [Mod ti] [ReltP [Relt'
[Relt have] [PerfP [Adv cleverly] [Perf' [Perf beenj] [AuxP [Adv cleverly] [Aux'
[Aux ] [ProgP [Prog' [Prog tj] [vP Ving ]]]]]]]]]]]]]]
constitute compelling evidence against the specifier analysis and in favor of the
adjunction analysis (see also chapter 2.3.2):
With the higher instance of cleverly being chosen, (46a) would be derived from
(47) by raising of the modal across the adverb. This is the trivial case. The
derivation of (46b) and (46c) involves the selection of the lower instance of
cleverly. While (46b) corresponds to the base order in (47), i.e. nothing moves,
the derivation of (46c) involves the raising of the past participle been across the
adverb to the head position of PerfP in our system. True, the postulation of two
merge sites for cleverly rids us of violations of the Head Movement Constraint
since have does not have to move to end up in a position preceding cleverly
and been does not have to raise across any verbal head to end up in a position
preceding cleverly. As it stands, multiple merge as in (47) still forces us to
assume that been may optionally raise to Perf. Again, this problem could be
avoided if we assumed that there is an additional merge site for the adverb, as
in (48) which translates into the structure in (52) below:36
Since the results yielded by the multiple merge structure in (52) and the
adjunction structure in (45) are compatible, elucidation as to which analysis is
more appropriate on a larger scale should come from adverb co-occurrence
within the inflectional layer, as in (49) – (51):37
(52) AgrS'
2
AgrS AttP*
2
Adv Att'
2
Att TP
2
T ModP
tu
Mod ReltP
3
Relt AttP*
3
Adv Att'
2
Att PerfP
th
Perf AuxP
1
Aux AttP*
1
Adv Att'
1
Att ProgP
1
Prog vP
4
willi (c-ly) ti ti ti have (c-ly) beenj tj (c-ly) Ving
No such filter is needed under the specifier analysis since the relative order of
adverbs is determined by the underlying hierarchy of functional projections,
with the functional projection hosting the frequency adverb universally being
ranked higher than the functional projection hosting the subject-attitude adverb
(cf. Cinque 1999:106, passim). At this point, it might be objected that the
strings in (54) constitute counterevidence to this claim since cleverly felicitous-
ly precedes often:
However, cleverly in (54a) precedes the finite non-lexical verb and thus, as we
shall see in the following section, is a representative member of subject-
oriented adverbs which are generated outside the inflectional layer. That is to
say that the strings in (54) do not constitute counterevidence to the specifier
analysis because the functional projection hosting subject-oriented cleverly is
different from that hosting subject-attitude cleverly. This assumption is
corroborated by the fact that subject-oriented adverbs and subject-attitude
adverbs may co-occur, as in (55a), and that, if an aspectual adverb like often is
present as in (55b), often is flanked by the two types of subject-related
adverbs:38
with (49) and (54), that subject-attitude adverbs and subject-oriented adverbs
are specifiers of distinct functional projections with the former being ranked
below aspectual adverbs, e.g. Cinque's (1999) Aspfrequency and Asp(im)perfect, as in
(49) and (51), and the latter being ranked higher than aspectual adverbs:
account for: since the subject and the modal or finite auxiliary are in a
specifier-head agreement relation within AgrSP (59), there is obviously no
structural position the adverb could possibly assume:40
(59) AgrSP
qp
D AgrS'
ei
AgrS …
u
vP
ru
D v'
# 6
a. Shei cleverly has … been … ti avoiding this topic
b. Shei cleverly will … have been … ti avoiding this topic
In the following, I will briefly contrast three analyses which seek to account for
the fact that subject-oriented adverbs may intervene between the subject and
the finite non-lexical verb.41
The first type of analysis involves adjunction of the adverb to the modal or
finite auxiliary, as illustrated in (60):42
(60) a. [AgrSP [D shei ] [AgrS' [AgrS [Adv cleverly] [AgrS has ]] … been … [vP [D ti ]
[v' avoiding this topic]]]]
b. [AgrSP [D shei ] [AgrS' [AgrS [Adv cleverly] [AgrS will ]] … have been … [vP
[D ti ] [v' avoiding this topic]]]]
As has been argued by Fukui & Takano (1998:37ff., passim), root adjunction
structures are not asymmetric enough to linearize. To circumvent this problem
they suggest that head-adjoined preverbal adverbs be seen as additional
specifiers:
(61) a. [AgrSP [D shei ] [AgrS' [Adv cleverly] [AgrS' [AgrS has ]] … been … [vP [D ti ]
[v' avoiding this topic]]]]]
b. [AgrSP [D shei ] [AgrS' [Adv cleverly] [AgrS' [AgrS will ]] … have been … [vP
[D ti ] [v' avoiding this topic]]]]]
oriented adverbs may be modified, as shown in (62), clearly shows that head
adjunction is not an option:43, 44
(63) a. [TP [D shei ] [TP(=T') [Adv cleverly] [T' [AgrS has ]] … been … [vP [D ti ] [v'
avoiding this topic]]]]]
b. [TP [D shei ] [TP(=T') [Adv cleverly] [T' [AgrS will ]] … have been … [vP
[D ti ] [v' avoiding this topic]]]]]
(64) a. [TopP [D Shei] [Top' [Top ] [?P [Adv cleverly] [?' [? ] … [AgrSP [D ti ] [AgrS'
[AgrS has ] … been … [vP [D ti] [v' avoiding this topic]]]]]]]]
b. [TopP [D Shei] [Top' [Top ] [?P [Adv cleverly] [?' [? ] … [AgrSP [D ti ] [AgrS'
[AgrS will ] … have been … [vP [D ti] [v' avoiding this topic]]]]]]]]
The analyses in (63) and (64) are equivalent in the sense that they rely on
topicality, i.e. discourse prominence of both the subject-oriented adverb and
the subject. The choice of (63) over (64), or vice versa, should be consequential
with respect to the analysis of subject-oriented adverbs in pre-subject position,
which is customarily, though inadequately, labeled TopP (cf. Rizzi 2002):
Second, genuine topics, but not preposed adverbs can precede an interrogative
operator (cf. (68) & (69) and (70) & (71), respectively):
(72) [ForceP [Force' [Force ] [TopP* [Top' [Top ] …[FocP [Foc' [Foc ] [PromP* [Prom' [Prom ] [TopP*
[Top' [Top] [FinP]]]]]]]]]]]
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 213
(73) a. [PromP [Adv Cleverlyi] [Prom' … [AgrSP she ti has been avoiding …]]]
b. [PromP [Adv Calmlyi] [Prom' … [AgrSP she ti deactivated the bomb]]].
(74) a. [SoP [Adv Cleverly] [So' [So ] … [AgrSP she has been avoiding…]]].
b. [SoP [Adv Calmly] [So' [So ] … [AgrSP she deactivated the bomb]]].
This assumption not only allows us to keep apart raising sites and merge sites,
but also allows us to account for the fact that subject-oriented adverbs are
inherently discourse related, whereas [PROM] is only an additional peripheral
feature on e.g. raised aspectual adverbs (cf. section 4.3.3):52
(75) a. [PromP [Adv Ofteni] [Prom' … [AgrSP they would ti change on the move]]].
b. [PromP [Adv Frequentlyi] [Prom' … [AgrSP they would ti meet us …]]]
(BNC-data)
(77) a. [SoP [Adv Foolishly] [So' … [TopP [D he k] [Top' … [PromP [AdvP ofteni]
[Prom' … [AgrSP tk would ti wear his pink hairpiece …]]]]]]].
b. [PromP [Adv Ofteni] [PromP' … [TopP [DP studentsk] [Top' … [PromP [AdvP
alreadyj] [Prom' … [AgrSP tk have ti tj received counseling]]]]]]].
214 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
c. * [TopP [D Hek] [Top' … [SoP [Adv foolishly] [So' … [TopP [D tk] [Top' …
[PromP [AdvP ofteni] [Prom' … [AgrSP tk would ti wear his hairpiece]]]]]]]]].
d. * [TopP [DP Studentsk] [Top' … [PromP [Adv ofteni] [PromP' … [TopP [DP tk]
[Top' … [PromP [AdvP alreadyj] [Prom' … [AgrSP tk have ti tj received
counseling in high school]]]]]]]]].
Under the analysis in (64), cleverly and calmly in (67) are clearly merged into
structure in their surface position, while, under the analysis in (63), both raising
to specPromP and merge in specSoP should be possible (cf. (73) vs. (74)).
Proponents of the adjunction analysis favor (74) over (73). This is not to say
that they subscribe to merge. What is relevant here is that the position assumed
by subject-oriented adverbs in pre-subject position is not a derived position (cf.
Browning 1996:252, 254; Frey & Pittner 1999: 26ff.; Ernst 2002:407ff.).53
Their argument is based on the observation that the presence of adjuncts, as
opposed to that of genuine arguments, alleviates the so-called That-Trace
Effect which arises in cases where the embedded subject is raised across the
overt complementizer that, but not if the embedded subject is raised across the
covert complementizer, as shown in (78).54 The fact that the topicalization of
arguments (79a) unlike the preposing of adjuncts (79b) does not salvage the
structure is accounted for in terms of the distinction between substitution
(argument raising) and adjunction (adverb positioning):55
(79) a. * Whoi do you think that [the boss]k ti avoided tk on Friday afternoon?
b. Whoi do you think that [Adv cleverly] ti avoided the boss on
Friday afternoon?
However, the Split CP System, as advanced by Rizzi (1997), allows for a more
elegant solution since arguments and discourse-prominent adverb(ial)s occupy
different structural positions, i.e. specTopP vs. specPromP, specFocP and, as I
assume, specSoP (cf. chapter 5.2.2 for discussion). In the former case the
specifier position of the lower CP shell, i.e. Rizzi's (1997, 2002) FinP, hosts a
null operator which instantiates long-distance agreement between the [+AGR]-
feature and raised subject, but not the subject trace, whence licensing in terms
of proper government fails (cf. Rizzi 1997:313ff.). In the latter case, no
operator is involved so that the [+AGR]-feature in Fin, irrespective of whether
cleverly sits in specPromP or in specSoP, can properly govern the subject trace
(details omitted):57
(82) a. Luckily for me, usually the train takes a half an hour stop … (www)
b. Politically, ONLY RECENTLY has much information regarding this
oppression been recorded… (www)
c. At the time, I foolishly did not realize that all of you did not want to be
spoken to in rational terms. (www)
d. But even he didn't realize how much the girls would enjoy making
music together and how successful they soon would become. (www)
e. Fortunately, soon after she arrived, her pups multiplied and saved the
day. (www)
(83) a. Stupidly, she'd forgotten it in the back of the fridge, months maybe,
Thanksgiving to Christmas. (www)
b. Carefully she'd memorized the list before coming here. (www)
c. Deftly they'll program themselves to our repeated needs… (www)
d. "Promise me you'll never cut it." Willingly, she'd made that promise.
She would never cut her hair. (www)
Given that cliticization is conditional on the adjacency of the clitic and its host,
i.e. there must be no offending traces or PRO intervening between the subject
pronoun and the clitic (viz. the illicit strings in (84)), the subject-oriented
adverbs in (83) cannot have originated between the subject and the finite verbal
element and raised to specPromP since the trace of the adverb, contrary to fact,
would block the cliticization of 's, 'd, 'll etc. onto the subject pronoun (85):
We have also seen that strings as in (87) and (88) are not ambiguous between
the two readings. In (87), the adverb can only be construed with the structural
subject, whereas, in (88), it can only be construed with the implicit argument,
i.e. the logical subject:
(89) a. [I have] my own suspicions that my father was being [{has been, has
to be}] deliberately poisoned. (CC)
218 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
b. Because that information (the phone number) is being [{has been, has
to be}] voluntarily conveyed to a third party… (www)
c. On many sides it is being [{has been, has to be}] reluctantly admitted
that law enforcing agencies […] have fallen down … (www)
d. Currently this burden is being [{has been, has to be}] willingly borne
by too few persons … (www)
Thus the critical case is that in (86), where the adverb immediately precedes
the passive participle and is immediately preceded by the finite passive
auxiliary. Since subject-oriented adverbs are in specSoP (section 4.3.1.2), they
simply cannot follow the finite passive auxiliary in the head position of AgrSP
(recall that subject raising to specTopP does not trigger verb raising to Top).
That is to say that the adverb in (86) under the subject-related reading must be
a subject-attitude adverb (section 4.3.1.1) contained in a high-merged AttP:
(90) [AgrS' [AgrS wasi ] [AttP* [Adv reluctantly] [Att' [Att ti ] [TP [T ti ] [beP [be ti ] [AgrPrtP
examined ]]]]]]
The fact that the adverbs in (89) cannot be construed with the passive subject
corroborates the analysis presented in chapter 3.2.1.2, according to which
agent-oriented adverbs in passives are within the lexical layer:
(91) a. [DP my father]i was [beP [be being ] [AgrPrtP [D ti ] [AgrPrt' [AgrPrt ] [µP [Adv
deliberately] [µ' [µ ] [VP poisoned ti ]]]]]]
b. [DP that information]i is [beP [be being ] [AgrPrtP [D ti ] [AgrPrt' [AgrPrt ] [µP
[Adv voluntarily] [µ' [µ ] [VP conveyed ti ]]]]]]
4.3.1.4 Summary. In this section I followed Frey & Pittner (1999) and Pittner
(1999, 2000a, 2004) in assuming that subject-related adverbs fall into two dis-
tinct syntactic classes: subject-attitude adverbs and subject-oriented adverbs,
with the former being confined to occurring within the scope of finite non-lexi-
cal verbal elements and the latter being restricted to taking scope over the finite
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 219
(94) a. He found something interesting there not long ago and so did she.
b. I would force her not to marry anyone.
c. Not many hours earlier had he spoken with someone else.
d. Not even two years ago could you enter without paying.
Starting out with the empirical fact that inherently negative adverbs, even
though they are not in a structural position from which they could take senten-
tial scope, section 4.3.2.1 explores the licensing of never, seldom, scarcely,
hardly in terms of pure [NEG]-feature raising to specNegP. Section 4.3.2.2 is
concerned with the analysis of constituent negation and the syntax of negated
constituents. It will be shown that constituent-negating not is a non-projecting
syncategorematic head whose sole contribution to structure lies in the 'inver-
sion' of the properties denoted by the constituent in its scope (this is a property
also displayed by negative prefixes, e.g. in-/un-). In section 4.3.2.2.1, we shall
address the longstanding puzzle of why certain negated constituents, if they
occur in the left periphery of the clause, may attain the status of negative
operators and thus give rise to sentential negation. The final section summa-
rizes the main results.
(95) a. Bonnie and Clyde would never get caught, would they?
b. They would seldom eat oysters, not even fresh ones.
c. Jerry would scarcely open the door to Tom, would he?
d. Dracula could hardly wait, not even till sunset.
(97) a. There was never any need to explain things to him. (BNC)
b. There is seldom any need for anything more complex than simple
ticks on diary pages. (BNC)
c. There was scarcely any time for a cigarette before we were hustled
over to the equipment store. (BNC)
d. All through this time he hardly had any painkillers. (BNC)
In section 4.2, we have seen that sentential negation involves the presence of
the functional projection NegP within the inflectional layer. Clearly, since
never, seldom, scarcely and hardly carry the feature [NEG] and thus are
negative operators, they should associate with NegP so that the Affect Criterion
can be met.
However, since never, seldom, scarcely and hardly display a wider and
different distributional range than sentence-negating not ((98) & (99)), they can
hardly be argued to occupy specNegP and thus to qualify as negative operators
in overt syntax.61 The adverbs under consideration may occur in sentence-
initial position (98a) – (98d) and they may precede the finite lexical verb (98e)
– (98h):
(100) a. I ... [NegP [NEG] [Neg' [Neg [NEG] ] … [TP [T ] [AspP* [Adv never[NEG]] [Asp' felt so
happy]]]]]
b. She ... [NegP [NEG] [Neg' [Neg [NEG] ] … [TP [T ] [AspP* [Adv seldom[NEG]] [Asp'
complained]]]]]
(102) a. [FocP [Adv neverk ] [Foc' [Foc hadi ] … [AgrSP [QP so many children] [AgrS' …
[NegP [Neg ti ] [AspP [Adv tk] [Asp' developed weak bladders ]]]]]]]
b. [FocP [Adv hardlyk ] [Foc' [Foc hadi ] … [AgrSP [DP the rains] [AgrS' … [NegP
[Neg ti] [AspP [Adv tk] [Asp' stopped ]]]]]]]
Since the finite non-lexical verb, on its way to Foc, has passed through the
head position of NegP, it is endowed with the [NEG]-feature (cf. Rizzi
1996:74). Thus, FocP, by virtue of accommodating both the negative constitu-
ent and the verbal element with the [NEG]-feature, is the structural domain in
which the Affect Criterion is satisfied in overt syntax. The fact that the pre-
posed negative constituent is a negative operator which overtly enters into a
specifier head configuration with a modal or auxiliary verb specified as [ NEG]
gives rise to the question whether negative inversion structures contain a NegP.
Haegeman (1995) and Rizzi (1996) argue that negative sentences in general
contain a NegP. In negative inversion structures, the head of NegP contains a
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 223
[NEG]-feature which the modal or auxiliary verb picks up on its way to the head
position of FocP, i.e. the verbal elements under consideration pass through the
head of NegP. Rizzi (1997:317f.) maintains that the [NEG]-feature is generated
in the head position of TP, i.e. negative inversion structures do not contain a
NegP. While Haegeman's (1995:181f.) and Rizzi's (1996:74) proposals allow
for a uniform treatment of negative sentences as extended verbal projections
containing a NegP, Rizzi's (1997:317f.) proposal allows for minimizing struc-
ture, i.e. NegP is projected if and only if it serves as the local domain in which
the Affect Criterion is satisfied.
Before turning to discuss constituent negation which, as we shall see, may
or may not induce sentential negation in preposing contexts, I would like to
comment on a potential problem for the analysis of inherently negative adverbs
presented in this section. The relevant data are those in (99a) – (99d) above
(repeated as (103)), where never, seldom, scarcely and hardly intervene
between the subject and the finite non-lexical verb, i.e. in a position that clearly
dominates NegP:
Given that the subject occupies the specifier position of AgrSP and the finite
non-lexical verb the position of AgrSP, the adverbs cannot have raised to their
actual position from below NegP simply because there is no specifier position
between specAgrSP and AgrS which could be targeted. At this point, I would
like to come back to subject-topicalization as discussed in connection with the
analysis of subject-oriented adverbs in section 4.3.1.2. That is to say that the
subject, after checking off its agreement features, vacates specAgrSP and raises
to specTopP to check off its [TOPIC] feature, thereby crossing the inherently
negative adverb in specFocP. Under this analysis, both the negative adverbs
and the subject carry peripheral features ([NEG] and [TOP], respectively) and
both the negative adverbs and the subject target specifier designated specifier
positions in the complementizer layer, specFocP and specTopP, respectively:
224 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(104) TopP
2
D Top'
2
Top FocP
2
Adv Foc'
2
Foc AgrSP
1
D AgrS'
2
AgrS NegP
rh
OP Neg'
rh
Neg TP
th
T AspP
2
Adv Asp'
2
Asp ModP
6
a. shei neverk wouldj ti tj tk tj tj tk tj …have [vP ti done it]
b. shei seldomk wouldj ti tj tk tj tj tk tj … [vP ti say sorry]
[TOP] [NEG] [NEG] [NEG][NEG] [NEG]
4.3.2.2 Constituent Negation. As has already been pointed out by Klima (1965:
306ff.), sentences containing negated constituents (105) are not necessarily
negative sentences, as the application of the diagnostic tests discussed in the
previous sections shows ((105) from Klima 1964:306):
(106) a. * Not long ago there was rain falling, was there?
b. * Not many hours earlier he had spoken…, had he?
c. * Not long ago there was any rain falling.
d. * Not many hours earlier he had spoken with anybody else.
e. * Not long ago was there rain falling.
f. * Not many hours earlier had he spoken with someone else.
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 225
(107) a. ? Whati did they say not long ago that the parcel weighed ti?
b. ? Whyi did they say not long ago that Bill would be fired ti?
(108) a. * Whati did they not say that she should do ti?
b. * Whyi did they not say that she left ti?
(109) [not Y] X
(111) a. [DegP [AP not long] [Deg' [Deg ] … [P' [P before]k [AgrOP [DP the party]i
[AgrO' [AgrO tk] [PP [P tk] [DP ti ]]]]]]]
b. [DegP [DP not two hours]i [Deg' [Deg ago]k … [P' [P tk] [AgrOP [DP ti] [AgrO'
[AgrO tk] [PP [P tk] [DP ti ]]]]]]]
(112) XP
3
not XP
Morphologically complex adverbs of the form [Adv in-/un- Adv] "are not truly
negative adverbs, in that the scope of negation is confined to the word in which
it occurs" (cf. Williams 1994:172). However, despite the differences that exist
between not and in-/un- (head vs. non-head, free vs. bound lexical item), the
effect these elements have in their respective domains, i.e. syntax and morpho-
logy, is identical, i.e. both not and in-/ un- endow the dominating node with
negative semantics:
(116) a. Not far from here they have found the corpse.
b. * Not far from here they have found it and neither has she.
c. * Not far from here have they found the corpse.
(118) * DegP
rp
notAP Deg'
2 2
not AP Deg …
h
P'
6
a. [AP not long] before the party
b. [DP not two hours]i ago k ... … tk ti
4.3.2.2.1 A Puzzle and Its Solution. In addition to the clear cases of negative
constituents inducing sentential negation, and in addition to the clear cases of
negative constituents not inducing sentential negation, as discussed in the pre-
vious section, there are cases where negative constituents may or may not in-
duce sentential negation, and thus (not) give rise to subject auxiliary inversion
((119a) adapted from Klima 1964:307; (119b) from Rudanko 1980: 357):68
(119) a. Not even two years ago {could you/you could} enter without paying.
b. In no clothes {does Mary look/Mary looks} attractive.
As can be seen in (120), only those notXPs which give rise to subject auxiliary
inversion may license indefinite quantifiers and positive tags:
228 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(120) a. Not even two years ago {could you/*you could} enter any bar.
b. In no clothes {does Mary look/*Mary looks} attractive, does she?
(121) a. Not even two years ago {* could you/you could} not enter without
paying.
b. In no clothes {*does Mary/Mary does} not look attractive.
(122) …
h
PP
fu
P notXP
tu
not XP
Under the assumption that percolation may or may not take place if notXP is an
internal argument, but is blocked otherwise, we are faced with the question of
why percolation applies only optionally.69 Moreover, under the assumption that
both [DP not even two years] in (123a) and [AP not long] in (123b) originate as
internal arguments of ago, the contrast in acceptability is unexpected ((123)
from Klima 1964:307, 306):
(123) a. Not even two years ago could you enter without paying.
b. * Not long ago was there rain falling.
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 229
(124) a. [DegP [DP Not many hours] [Deg' … [AP earlier]]] he had spoken...
b. [DegP [AP Not far] [Deg' … [PP from here]]] they have found…
*percolation
(125) a. [DegP [QP not even two years] i [Deg' [Deg ago]k [pP [p tk] [AgrOP [QP ti] [Agr'
[AgrO tk] [PP [P tk] [QP ti]]]]]]]
b. [DegP [Deg ] [pP [p ink] [AgrOP [DP no clothes]i [Agr' [AgrO tk] [PP [P tk] [DP
ti ]]]]]]]
Since [AP not long] in (123b) is not a nominal internal argument, no AgrOP is
projected, i.e. raising of [AP not long] proceeds directly to specDegP:
(126) [DegP [AP [not] [AP long]]i [Deg' [Deg agok] … [PP [P tk ] [AP ti ]]]]
The illicit string in (129) patterns with the illicit strings in the non-inverted
structures in (120) above and they are in sharp contrast with genuine negative
operators in situ as in (130):
4.3.2.3 Summary. In this section, the focus was on the inherently negative
quantificational-aspectual adverbs never, seldom, scarcely and hardly, and on
constituent-negating not. It was shown that inherently negative adverbs, which,
like negative verbal elements (e.g. shouldn't, aren't) and not in specNegP, in-
duce sentential negation and warrant the projection of NegP. The elements
under consideration were shown to stay in their respective base positions below
NegP (section 4.3.3 for analysis) with only their [NEG]-feature raising to
specNegP. Under the analysis presented, the Affect Criterion is satisfied in
overt syntax. The same holds for inherently negative adverbs under inversion.
The properties of constituent negation and the syntax of negated constituents
were at the heart of section 4.3.2.2. It was shown that negated constituents are
invariably of the format [XP not [XP]], with not being a non-projecting head
whose sole contribution to the entire constituent is a negative feature which
does not percolate beyond the constituent not takes scope over. The final sec-
tion provided a solution for the longstanding puzzle created by negated
constituents which may or may not attain operator status and thus induce sen-
tential negation (viz. in no clothes vs. not long ago), depending on the associa-
tion of the constituent with additional peripheral features, i.e [ FOC] vs. [PROM].
(131) a. And once caught they usually remain customers for life. (BNC)
b. [Wintering] birds have usually gone by late March…(BNC)
232 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(134) a. [He] rarely goes out because he seldom feels well. (BNC)
b. Belief can rarely be made clear and distinct. (CC)
c. His picture would have rarely been requested. (www)
d. It has been rarely using its battery. (www)
(136) a. Mr Norrie rarely goes out because he seldom feels well. (BNC)
b. You could rarely get a straight answer from any of them. (BNC)
c. Sheringham will rarely have had a quieter game. (CC)
d. [You] will often be printing at minimum aperture. (CC)
e. Whales can often be seen off the coast. (CC)
f. [They] have seldom strayed more than a few miles… (BNC)
g. They're often seeking a new support group… (CC)
h. I have rarely heard him sound more relaxed. (BNC)
i. I have often been thinking of you… (www)
j. I've often been told I come over. (CC)
k. [It] is more often being used as a defence [sic!] … (BNC)
l. It has been rarely using its battery. (www)
m. A man of genius has been seldom ruined but by himself. (www)
n. [The] definition that had been rarely articulated. (CC)
o. These phrases are being often abused by Western politicians… (www)
d. continuative:
[I] tried to move away, but still [I] was paralysed.
e. (im)perfect:
Never had so many children developed weak bladders.
f. retrospective:
Lately she seems a little low in spirits. (BNC)
g. proximative:
Soon we went in to our seats. (BNC)
h. durative: Briefly Jaq entertained the notion that he was being
tested. (BNC)
i. generic: Characteristically he has given it all away. (BNC)
j. prospective: Almost she disliked herself. (BNC)
((e) from BNC)
Given the data in (136) and given the movement operations non-lexical verbs
undergo, the following pattern emerges, where • marks the positions assumed
by frequency adverbs:
(142)
AgrS
be
Perf
v
Aux
Prog
Relt
Mod
subject
• [fin] [fin]
• modal t • VB
• have[+fin] t • VBD
• modal t • have • VBD
• be[+fin] t • VBG
• modal t • be • VBG
• have[+fin] • been t • VBG
• be[+fin] t • VBN
• modal t • be • VBN
• have[+fin] • been t • VBN
• be[+fin] t • being t • VBN
VB=bare infinitive, VBD=past participle, VBG=present participle, VBN=past participle
(143) a. They will often have been described in the literature. (www)
b. [They} often will have been born in this country… (www)
c. I have often been thinking of you… (www)
d. [They] often have been doing volunteer work … (www)
e. [Diagnosis] will have often been confirmed by a Doctor. (www)
f. [Boris's son] could be often seen dancing… (www)
g. [The team] could have been often led by a pensioner. (www)
h. These towels could have been being often used to wipe up all kinds of
liquids. (www)
i. [He] had been often being mistaken for his twin brother. (www)
j. Nepali workers are being often deprived of their rights. (www)
236 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(144)
be
T
Perf
AspP
Mod
Aux
Prog
Relt
a. willi often ti have been
b. often will have been
c. havei often ti been
d. often have been
e. willi havej often ti tj been
f. couldi bej often ti tj
g. couldi havej beenk often ti tj tk
h. couldi havej beenk beingm often ti tj tk tk tm tm
i. hasi beenj often ti tj being
j. isi beingj often ti tj tj
(145) a. [AgrS [Agrs willi ] [AspP* [Adv often] [Asp' [Asp ti] [TP [T ti] [ModP [Mod ti] [ReltP
[Relt have] [AspP* [Adv often] [Asp' [Asp ] [PerfP [Perf beenj] [AuxP [Aux tj] [ProgP
[Prog beingk] [beP [be tk] [AspP* [Adv often] [Asp' [Asp ] … ]]]]]]]]]]]]]]
b. [AgrS [Agrs hasi ] [AspP* [Adv often] [Asp' [Asp ti] [TP [T ti] [ReltP [Relt ti] [AspP*
[Adv often] [Asp' [Asp ] [PerfP [Perf beenj] [AuxP [Aux tj] [ProgP [Prog beingk] [beP
[be tk] [AspP* [Adv often] [Asp' [Asp ] … ]]]]]]]]]]]]]
c. [AgrS [Agrs isi ] [AspP* [Adv often] [Asp' [Asp ti] [TP [T ti] [ReltP [Relt ti] [AspP*
[Adv often] [Asp' [Asp ti] [AuxP [Aux ti ] [ProgP [Prog beingk] [beP [be tk] [AspP*
[Adv often] [Asp' [Asp ] … ]]]]]]]]]]]]
d. [AgrS [Agrs isi ] [AspP* [Adv often] [Asp' [Asp ti] [TP [T ti] [ReltP [Relt ti] [AspP*
[Adv often] [Asp' [Asp ti] [AuxP [Aux ti ] [AspP* [Adv often] [Asp' [Asp ]
…]]]]]]]]]]]
merge site for the high occurrences in (145b) – (145d)). As far as I can tell,
nothing hinges on that.
In the following, rather than discussing for every aspectual subclass all the
valid options of adverb placement, I shall briefly point out some cases of non-
admissibility of aspectual adverbs which, as has also been argued by Ernst
(2002:327ff.), derive from semantic mismatches between the adverbs and the
verbal elements under consideration, rather than from syntactic restrictions.
Since habitual adverbs are subject to the requirement that the (parts of)
events they take scope over cover "a fairly large time interval" (cf. Ernst
2002:351), they are barred from occurring in relatively low positions within the
inflectional layer, because the time intervals as denoted by the participles in
(146) "are often too short" for the habitual adverb to quantify over:
Terminative adverbs require that a given state of affairs not obtain any longer
at reference time, while continuative adverbs require that an event be unbound
in the sense that a given state of affairs must obtain at reference time and at
some point in time preceding reference time (cf. Ernst 2002:345f.). Both
scenarios are incompatible with perfective have:
(148) a. * They {no longer/still} have {no longer/still} refused our invitations.
b. * Hillary has been {no longer/still} ignoring their advice.
c. * The president has been {no longer/still} insulted.
d. * Al would have {no longer/still} waited for us.
The contrasts between (149) and (150) suggest that adverbs expressing retro-
spective aspect are parasitic on the presence of an overt anteriority marker,
such as past tense morphology on either the lexical or the auxiliary verb (cf.
(150a) and (150b)), or aspectual have, as in (150c):
(151) a. [AgrS [D they] [AgrS' [AgrS ] [AspP [Adv habitually] [Asp' [Asp ] [TP [T ] [vP
spoke Egyptian Arabic]]]]]] (BNC)
b. [AgrS [D they] [AgrS' [AgrS 'll i ] [AspP [Adv still] [Asp' [Asp ti] [TP [T ti] [ModP ti
be scratching their heads]]]]]] (CC)
c. [AgrS [D he] [AgrS' [AgrS had i] [AspP [Adv always] [Asp' [Asp ti] [TP [T ti] [ ReltP
ti worked for the benefit of his country]]]]]] (BNC)
d. [AgrS [D they ] [AgrS' [AgrS are i] [AspP [Adv usually] [Asp' [Asp ti] [TP [T ti]
[AuxP ti thinking 'what did I do wrong'?]]]]]] (CC)
e. [AgrS [D I] [AgrS' [AgrS was i] [AspP [Adv again] [Asp' [Asp ] [TP [T ti] [beP ti
struck by the timelessness of the … Plain]]]]]] (BNC)
(152) a. [AgrS [D Tranquillisers] [AgrS' [AgrS will i] [TP [T ti] [ModP [Mod ti] [ReltP [Relt
have] [AspP [Adv often] [Asp' [Asp ] [PerfP been initiated by doctors ]]]]]]]
b. [AgrS [DP Griffiths] [AgrS' [AgrS had i] [TP [T ti] [ReltP [Relt ti] [AspP [Adv
characteristically] [Asp' [Asp ] [PerfP taken the existing Coalition ]]]]]]]
c. [AgrS [DP the fingertip] [AgrS' [AgrS is i] [TP [T ti] [AuxP [Aux ti] [AspP [Adv no
longer] [Asp' [Asp ] [ProgP being pressed]]]]]]]
((a) = (143e), (b) & (c) from CC)
(153) habitual > repetitive > frequentative > celerative > anterior > terminative
> continuative > (im)perfect > retrospective > proximative > durative >
prospective
b. Usually, the dog never stirs in the morning until I appear. (BNC)
c. Once again they were still talking at 01:00! (www)
d. The tornado again lifted briefly and touched down again in downtown
Carterville. (www)
e. Dot found to her surprise that, just as she often could no longer
remember what her father's face looked like… (BNC)
f. [Direct] contact with animals by infected humans often cannot always
be established. (www)
g. To lose weight quickly has gained much popularity recently.
h. I think it might be worth while to quickly review a little more briefly
some of the history …(www)
i. But he already no longer hears them. (BNC)
j. Their joint contribution would have warranted a commemorative
book, had not one by chance bearing their names already been
recently released. (BNC)
k. I no longer can always get up the stairs … (www)
l. When entering a new URL while a page is loading, we no longer
briefly revert to showing the URL … (www)
m. We still always felt we could make it to state, though. (www)
n. He hadn't thought so, but his insides still ached briefly. (BNC)
o. A rubber band that is kept under tension always will soon lose its
elasticity and snap. (www)
p. But always briefly freeze the mixture before coating with chocolate.
q. He also recently briefly served as President and CEO of Aspirian Inc.
r. Recently I almost bought a watch that I've been meaning to get for
sometime. (www)
s. Walker would soon briefly join his boss under the public microscope.
t. His voice got hoarser and hoarser, and pretty soon I almost had to
strain to catch the words. (www)
((p) from BNC, (q) & (s) from www)
However, four things need to be considered. First, certain adverbs cannot co-
occur on semantic grounds within the inflectional layer.84 Thus, for example,
continuative still cannot be in the scope of already and no longer. Both already
and no longer require that the activity, process or state it takes scope over be
completed at reference time, whereas still requires unboundedness, i.e. the non-
completion of an activity or process at reference time (cf. (155a) & (155d)).
Recently and already fail to co-occur since they both require completion at
reference time (cf. (155b)). Aside from the incompatibility of no longer and the
perfect auxiliary, the same holds for no longer and recently (cf. (155e)). (155c)
and (155f) are ruled out because the requirements of already and no longer
clash with those of soon: while already and no longer require completion at
reference time, soon requires that the activity, process or state arise in the near
240 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
future. Since still requires that the event hold at reference time, it cannot take
into its scope retrospective and prospective adverbs (cf. (155g) & (155h)).
Finally, (155i) is ruled out because of the retrospectivity–prospectivity clash
induced by recently and soon:
Second, aspectual adverbs tend not to cluster within the inflectional layer. Co-
occurrences of aspectual adverbs are best if one of them occurs within the
complementizer layer (156) or after the lexical verb (157); adjacent occur-
rences as in (158) are relatively rare:
(158) a. They are so bright, their explosions so powerful, that they often briefly
outshine the 100 billion stars of the rest of the galaxy… (www)
b. But he already no longer hears them. (BNC)
c. Walker would soon briefly join his boss under the public microscope.
((c) from www)
Third, sequences of the type not > aspectual adverb as in (159) are ambiguous
between constituent negation and sentential negation. However, since pitch
accent most naturally falls onto the adverb, preference is given to the contras-
tive notAdv-reading, e.g. not usually, but often, not often, but sometimes, etc.:
(159) a. The lift was not usually operated after six in the evening. (BNC)
b. Many knitters will not often come across patterns that take up the
whole working memory. (BNC)
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 241
Interestingly, strings of the type not > aspectual adverb are high-frequent in
counterfactual conditionals. Since the counterfactual operator cancels out
negation, the clause containing the adverb clearly is not negative:
(161) a. [He] prophesied another war between Egypt and Syria, as if the
Romans would not again have intervened. (BNC)
b. [He] would be surprised if I did not at least briefly mention the
position in the London borough of Barnet. (BNC)
c. If that martinet had not been already married, he would have married
Sophia, … (BNC)
d. [If] the IMF were working effectively, five countries wouldn't have
recently gotten into financial trouble […]. (www)
(162) a. The lift was not usually operated after six in the evening.
-> The lift was operated after six in the evening.
-g> The lift was not operated after six in the evening.
b. Biotech won't soon replace 'conventional' breeding.
-> Biotech will replace 'conventional' breeding.
-g> Biotech won't replace 'conventional' breeding.
Fourth, reverse orders can be found with a seemingly higher adverb being
realized in a lower, notably the postverbal position ((164) from BNC):86
(164) a. She still woke often in the night full of strange forebodings…
b. I sat down and Katrina soon joined me again.
c. But he was already kissing her again, more hungrily this time.
d. They can always start again and it doesn't matter.
Reverse orders as in (164) are often taken to constitute solid evidence against
the specifier analysis (cf. Ernst 2000; 2002). As we shall see in the following
section, this is not the case. Before turning to present evidence that these data
are fully reconcilable with the specifier analysis, I would like to summarize
what has been said so far. Syntactically, the positioning of aspectual adverbs is
virtually free. Unless ruled out by semantic restrictions, any single adverb may
assume any of the interverbal positions within the inflectional layer (136) as
well as the pre- and post-subject position which we have identified as spec-
PromP (137). The seemingly distributional freedom of any single aspectual ad-
verb within the inflectional layer (barring semantic restrictions (146) – (150))
was attributed to the functional projections (AspP) hosting the adverbs under
consideration being mergible in four different positions ((145), (151) & (152)).
The co-occurrence and the relative order of aspectual adverbs was argued to be
essentially governed by the underlying hierarchy of functional heads identified
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 243
(165) a. habitual: Reorders can go out, usually, within two weeks. (www)
b. celerative: [Come] sit on your potty, quickly. (BNC) 87
c. anterior: This has been briefly discussed already. (BNC)
d. terminative: Come: let us wait no longer. (CC)
e. continuative: [They] were with him still. (www) 88
f. (im)perfect: I will pray for him always. (www)
g. retrospective: We've had a couple of bomb threats lately. (BNC)
h. proximative: I hope he comes through to see me soon. (BNC)
i. durative: He looked up briefly. (BNC)
j. generic: He explains how each of these organs behaved
characteristically, … (www)89
k. prospective: When this discovery is finally made, then will
physicists feel that they have solved, almost at least, the
mystery of matter. (www)
(166) a. Lawyers often love to tell you about how good they are. (BNC)
b. People rarely asked for your invitation card. (BNC)
c. In fact, in situations that allow more than two choices, and use both
terms, young children seldom make errors at all. (BNC)
(167) a. Most lawyers love to tell you about how good they are.
b. Few people/not many people asked for your card.
c. Few/not many young children make errors at all.
(168) a. Lawyers will tell you (quite) often how good they are.
b. People asked rarely for more.
c. Children meet seldom at night.
(169) a. Situations in which lawyers will tell you … are are many.
b. Situations in which people asked for more were few.
c. Situations in which children meet at night are few.
Similar observations can be made with respect to again and twice. Although
the contrasts are not as sharp as with often, rarely and seldom, it is clear that
again and twice relate to different entities in pre- and postverbal position:
(170) a. On the night before his death he again said he would commit suicide.
b. Fears soon began to be expressed that wines from the Midi too would
again find their way back into Champagne cellars.
c. It had been switched off after two senior doctors had twice run tests on
the patient before declaring him dead.
d. The president had twice vetoed different versions of the plan.
(BNC-data)
Preverbal again expresses repetition in the sense that there is a previous event
of the same or of a similar type, e.g. saying that one would commit suicide
(170a), whereas preverbal twice 'counts' the frequency with which two events
of the same type recur, e.g. senior doctors running tests (170c). In postverbal
position, these adverbs express the repetition of the action or process denoted
by vP on one single occasion ((171a) & (171c, d)). Postverbal again in (171b)
is special insofar as again has a restitutive reading, i.e. it expresses the restitu-
tion of a prior state in which the wines were in the Champagne cellars. This
reading is available only in postverbal position.90
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 245
Pre- and postverbal occurrences of frequency and repetition adverbs are not
only similar to pre- and postverbal occurrences of carefully, quietly, rapidly,
quickly with respect to interpretational differences, but also with respect to
derivational unrelatedness, i.e. pre- and postverbal repetitive and frequentative
adverbs may co-occur:
However, co-occurrence is not the only piece of evidence supporting the claim
that there is no derivational relationship between pre- and postverbal occur-
rences of the adverbs under consideration. Seldom and rarely are customarily
classified as inherently negative adverbs and thus as inducing sentential nega-
tion. Tagging clearly shows that only preverbal occurrences of seldom and
rarely in (173) induce sentential negation: they are compatible with the 'and
neither'-tag, but not with the 'and so'-tag. The reverse situation obtains with
postverbal occurrences of seldom and rarely in (174). That is to say that post-
verbal seldom and rarely do not induce sentential negation:
(173) a. We seldom hear the tolling of a bell to signify disaster, and neither/
*so do they. (BNC)
b. Belief can rarely be made clear and distinct, and neither/ *so can
disbelief. (CC)
(174) a. We met seldom during the first six months …, and *neither/so did
they.
b. Her eyes blinked rarely, and *neither/so did mine.
It might be objected that seldom and rarely in (174) fail to induce sentential
negation because they are too deeply embedded to take propositional scope.
However, taking into consideration the strings in (175), we see that this
objection overshoots the mark since the nominal constituents nobody and
nothing – in spite of not being in structural position from which they could
take propositional scope – induce sentential negation (cf. section 4.3.2.1 for
discussion; (175) adapted from Huddleston 1984:420):
That is to say that postverbal rarely and seldom do not give rise to the projec-
tion of NegP, whence they cannot take sentential scope by having their [ NEG]-
feature extracted. In this respect, they are similar to negated constituents, e.g.
in no clothes, under no circumstances.
The fact that rarely, seldom, often and twice pattern like manner adverbs,
degree of perfection adverbs and means-domain adverbs ((176) – (178))
suggests that the natural habitat of process-modifying repetitive and frequen-
tative adverbs is the lexical layer: in actives, under the intended reading,91 they
invariably follow the lexical verb (176), whereas, in passives, they either
follow or precede the passive participle of the lexical verb ((177) & (178)):
(177) a. [They] are in excellent order and had been used rarely. (CC)
b. Because it bites seldom we are unprepared for it. (CC)
c. [This] particular store had been robbed often. (CC)
d. [The] deadline had been extended twice. (CC)
However, this nice picture gets blurred if we take into consideration the strings
in (179), (182) and (185), where frequency adverbs may or may not co-occur
with other vP-internal adverbs:
Under the assumption that postverbal often, rarely, seldom and twice make up a
distributional class, i.e. sit in the same structural slot, the contrasts between
often, rarely, seldom on the one hand and twice on the other are not expected.
However, the illicit sequences in (179) improve as soon as we replace the
adverbs by semantically equivalent expressions of the type Q-times:
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 247
(180) a. Just after reading that poem a few times carefully, Schubert set it to
music ...
b. There's a big difference between the way something feels after you've
played it that many times "perfectly" and not having had done it.
c. The […] experts met several times electronically.
d. Always proofread your final draft carefully many times to avoid typos
or misspelled words or other errors.
e. This has worked perfectly many times on another machine.
f. [They] met three times during July to November and communicated
electronically various times between the meeting times.
(www-data)
In the following, I will take the acceptable strings in (180) as a diagnostics for
structure. The fact that carefully, perfectly and electronically in (180a) – (180c)
are focalized is suggestive of their being realized within the empty verbal
structure below the lexical verb. That is to say that they are licensed from
within the lexical layer (cf. chapter 3.2.1.5). In (180d) – (180f), these adverbs
sit in their respective licensing positions, i.e. specµP for carefully and perfectly,
and specMeansP for electronically. The structure in (181), where FreqP is the
functional projection hosting the frequency adverbs under consideration,
accounts for both linear orders, i.e. (180a) – (180c) and (180d) – (180f):
(181) a. [v' [v proofreadk] [AgrOP [DP your drafti] [AgrO' [AgrO tk] [µP [Adv carefully]
[µ' [µtk] [FreqP [QP many times] [Freq' [Freq tk] [VP [DP ti] [V' [V tk ]]]]]]]]]]
b. [v' [v readingk] [AgrOP [DP that poemi] [AgrO' [AgrO tk] [µP [Adv e] [µ' [µtk]
[FreqP [QP many times] [Freq' [Freq tk] [VP [DP ti] [V' [V tk] [ [ ] [Adv
carefully ]]]]]]]]]]]
(183) a. [These books may] have been read carefully several times, but the
spine will still be very clean although slightly bent. (www)
248 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
b. Real-time translation has been done poorly many times, but never
done well as far as we know. (www)
c. During that time the plant may have been cultivated mechanically five
times or more plus being hand hoed at least once. (www)
d. You've been asked many times politely to sod off. (www)
e. His shirts have been ironed many times perfectly. (www)
f. Evolution has been observed many times experimentally. (www)
(184) a. [DP the plant]i may have been [AgrPrt' [AgrPrt cultivatedk] [µP/MeansP [Adv
mechanically] [µ'/Means' [µ/Means tk] [FreqP [QP five times or more] [Freq' [Freq
tk] [VP [DP ti] [V' [V tk]]]]]]]
b. [DP Evolution]i has been [AgrPrt' [AgrPrt observedk] [µP/MeansP [Adv e]
[µ'/Means' [µ/Means tk] [FreqP [QP many times] [Freq' [Freq tk] [VP [DP ti] [V' [V tk]
[ [ ] [Adv experimentally]]]]]]]]]
Interestingly, with often, rarely, seldom and twice in pre-participle position the
patterns in (185) emerge:
Moreover, replacing often, rarely, seldom and twice by Q-times yields only
marginally acceptable results:
??
(186) a. My bookshelves have been Q-times {carefully/perfectly/
telepathically} dusted.
??
b. My bookshelves have been Q-times dusted {carefully/
perfectly/telepathically}.
Under our analysis, the grammaticality judgments in (186) are expected. Since
Q-times is generated in a position below that occupied by carefully, perfectly or
telepathically, the derivation of the strings in (186) would have to involve the
realization of both the higher adverbs and the passive participle within the
empty verbal structure below the base position of the passive participle, with
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 249
the verb either assuming a higher or lower position than the seemingly
extraposed adverb, as shown in (187) for many times carefully:
(187) µP
1
Adv µ'
1
µ FreqP
th
QP Freq'
1
Freq VP
1
DP V'
2
V
ti
Adv
fu
Adv
#
a. have been e Q-times carefully dusted
b. have been e Q-times dusted carefully
Two things are going wrong here. First, given that the empty verbal structure
must be structurally licensed by the verb (or its trace), it is not clear how the
participle would get licensed in the head position of Second, even if it
could, it would be odd for the participle to get realized within since, barring
defocalization, it may stay in situ (cf. chapter 3).
The non-availability of twice and Q-times as opposed to the availability of
often, rarely, seldom (cf. (185) & (186)) suggests that often, rarely, seldom are
not within the lexical layer, but within the inflectional layer. This assumption is
corroborated by the strings in (188), where often, rarely and seldom are left-
adjacent to the participle and where there is an additional, i.e. postverbal
occurrence of a frequency adverb:
Under this analysis, the non-admissibility of twice and Q-times could be related
to their not felicitously occurring after been:
?
(189) a. [The departure] had been twice postponed twice.
?
b. [The departure] had been Q-times postponed twice.
This leaves us with the contrasts between the illicit sequences in (179) and
(182) and the licit ones in (180) and (183). Since often, seldom and rarely are
semantically equivalent with expressions of the type Q-times (which for Q = 1
is lexicalized as once and for Q = 2 as twice), the contrast cannot be attributed
to semantics having gone astray. Likewise, it is hard to see which syntactic
constraints could be at work. Since I do not have a solution at this point and
since, if one searches long enough and hard enough, at least some of the
'missing' strings can be found (cf. (190)), I tentatively suggest that the contrasts
are contingent on lexical choice and/or prosodic restrictions:
(190) a. If you drive fast often, the under panel bends and after a long time, it
just falls off (www)
b. 96% of patients were judged to snore loudly "often" or "always" by
their bedpartners… (www)
As we have seen in connection with (171a, b), postverbal again can have either
a repetitive or a restitutive reading. While again in (191) has only the repetitive
reading, again in (192) is ambiguous between the two readings. That is to say
that the vP in (192a) reads as either opened the lines a second, third, etc. time
(repetitive) or as reopened the lines (restitutive). The two agains will be
discussed in turn:
(192) a. [They] opened the lists again for public submission,… (www)
b. How to Put Humpty Together Again. (www)
c. [Hook] up the battery again, … (www)
(195) v'
eu
v AgrOP
2
D(P) AgrO'
2
AgrO µP/MeansP
fo
Adv µ'/Means'
fu
µ/Means RepP
2
Adv Rep'
1
Rep VP
1
D V'
1
V
1
Adv
a. publishk tk electronically tk again tk ti tk
b. filek [them]i tk e tk again tk ti tk electr.
Since the same pattern emerges with co-occurrences of repetitive again and
frequency adverbs (cf. (196) & (197)), it seems reasonable to assume that RepP
is merged into structure below FreqP (pace Cinque 1999):92
(198) a. These have been carefully sifted again and again in a search for clues
as to the nature of the Star of Bethlehem. (www)
b. [It] has been beautifully recorded again by Craker, and the program
[…] is very good indeed. (www)
c. The applicability of the parallel port for connecting to different
systems has been experimentally verified again. (www)
(200) a. The present edition has been again carefully revised. (www)
b. For six years now we do visit us rotative and this year we've been
again perfectly looked after! (www)
c. And in some cases they've been again manually recounted, in some
counties I think as many as four or maybe even five times by the end
of all the process. (www)
Repetitive again may appear in postverbal position, with either the manner,
degree of perfection and means-domain adverb preceding the passive participle
(cf. (198)) or following the passive participle but preceding again (cf. (199)),
and repetitive again may be left-adjacent to manner, degree of perfection and
means-domain adverbs in preverbal position (cf. (200)). The first two patterns
are easily accounted for under the assumption that the participle raises to the
head position of µP/MeansP, as in (198), or to the head position of AgrPrtP, as
in (199). However, given that the participle, in principle, may stay in situ, the
nonavailability of strings as in (201) is unexpected:
(205) a. Wow, thanks for telling me! I would have completely missed it again.
b. Once the magazine to whom you have sold first serial rights
publishes the piece, you own it completely again. (www-data)
(206) a. "Our side has been completely vindicated again," said Hank Zumach,
a plaintiff in the suit. (www)
b. [Marx's most important] work has never been performed completely
again (let alone recorded) in his lifetime. (www)
254 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(211) a. I also agree with you about the tone now I have re-read it again.
b. [Maybe] I didn't subscribe correctly, so I have re-done it again today.
c. Carefully repeat the experiment again.
d. Based on my […] discussions over the last year or two, I realized how
this theme has recurred again and again.
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 255
Also notice in this context that repetitive again (212), but not restitutive again
(213), may iterate. Trivially, both the unacceptability of (210) under the restitu-
tive reading and the unacceptability of (213) follow from the semantics of
restitutive again, namely the undoing of the effects of a preceding opposite
process:
(212) a. Oh, the doctors and nurses had assured her again and again that she
was fine and would soon be fit and well again. (BNC)
b. I have tried it again and again. (www)
(213) a. * To activate the settings, close the browser and open it again and
again.
b. * Then open the switch for a couple of seconds, and close it again and
again.
(214) a. * Scared out of her wits, she got up, closed all the windows and got
back to bed. Five minutes later, he got up and opened them again
slightly.
b. * Roughly speaking, excorporation means that after merging, the
adverb and verb separate again completely.
(215) a. Tighten the cutter again slightly and rotate again. (www)
b. Release the breath again completely – deep breathing should be done
in a slow and steady manner. (www)
The same kind of incompatibility can be observed with the restitutive prefix
re–, but not with the repetitive prefix re–:
The incompatibilities sketched are not surprising since 'restitution' implies the
complete reversal of a process, i.e. the undoing of the effects of a given proc-
ess, e.g. closing the window, and the restoration of a given prior state of affairs,
e.g. the window was open, by definition, must be complete (but see below).
However, since this cannot mean that restitutive again competes for the same
structural position as completion adverbs, i.e. specAspPcompletive (cf. chapter
3.2.2), we need an additional functional projection within the lexical layer.
256 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Ignoring for the moment the question of the relative position of the functional
projection hosting restitutive again, let us dub this functional projection RestP
whose specifier position hosts restitutive again or abstract restitution marker,
say , in cases in which restitution is not lexicalized by again but by the pre-
fix. Let us further assume that in the latter case the restitution marker is
coindexed with re– (so as to signal that this structural position is already been
'taken', i.e. that the relevant features are already present on the lexical verb):
(218) RestP
2
Adv Rest'
2
Rest …
h
VP
6
again shut the door
i [[re]i [shut]] the door
(219) v'
wp
v AgrOP
2
DP AgrO'
1
AgrO …
h
AspP completive
1
Adv Asp'
1
Asp VP
1
V DP
# 4
a. rejects k this idea o tk utterly tk tk to
b. [v [Adv utterly] [V rejects]] k this idea o tk tk tk to
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 257
This mechanism, I think, can also shed light on the alternation between the
restitutive prefix re- and the restitutive adverb again: re-prefixation amounts to
the defocalization of 'restitution' and the focalization of the verb's internal
argument, i.e. re-prefixation in (220b) can be seen as morphologically
mimicking the effects of syntactic incorporation (219b):
(220) v'
ro
v AgrOP
fi
DP AgrO'
th
AgrO …
h
RestP
2
Adv Rest'
2
Rest …
h
VP
th
V DP
# 4
a. shut k the dooro tk again tk tk to
b. [[re]i [shut]]k the dooro tk i tk tk to
Two things are important in this context.97 First, the verbs under consideration
take incremental objects so that the restitution of a prior state of affairs can
refer to both the entire object, to the entire set of objects and/or to its compo-
nent parts (each of which is completely restored):
Second, and I have no explanation to offer for this fact, the presence of com-
pletion adverbs is fully sanctioned only in passives. As can be seen in (223),
completely and entirely are barred from both preceding and following the
active verbal element. A slightly different situation obtains with partly and
partially. While systematically barred from preceding the active verbal
element, partly and partially may occur in postverbal position, if preceded by
only:
Again, the restitution of a prior state of affairs is complete, albeit only with
respect to component parts of the affected incremental object.
There is one more detail I would like to address. As we have seen in chapter
3.2.2.1, completion adverbs and MADM adverbs may co-occur in actives if the
completion adverb assumes the preverbal position (225a) or if both completion
adverb and the MADM follow the lexical verb (order irrelevant (cf. (225b, c)):
(228) a. This door has been (*carefully) opened (*carefully) again carefully.
b. It has been (*correctly) installed (*correctly) again correctly.
c. Your card will be (*electronically) filled (*electronically) again
electronically.
(229) [AgrOP/AgrPrtP [AgrO'/AgrPrt' [AgrO/ AgrPrt ] … [AspPcompletive [Adv ] [Asp' [Asp ] [RestP
[Adv ] [Rest' [Rest ] [VP ... ]]]]]]] (AgrOP for actives, AgrPrtP for passives)
proach, von Stechow (1996) proposes that the projections headed by resultative
and causative verbs be syntactically represented as in (230). That is to say that
the adjectival state predicate 'the door open' is embedded as the complement of
an abstract verbal head specified as BECOME, whose projection, in turn, is em-
bedded as the complement of an abstract verbal head specified as CAUSE:
(230) [VP [V CAUSE ] [VP [V BECOME ] [AP [DP the door ] [A open ]]]]
Von Stechow (1996) maintains that restitutive again forms a (small clause-
like) constituent with the adjectival small clause:100
Under von Stechow's (1996) analysis, the VP-final occurrence of again comes
as a natural consequence of both [DP the door] and [AP open] overtly targeting
specAgrP and {VBECOME, VCAUSE, etc.}, while the linear orders of again and
MADM adverbs in (227) and (228), with again obligatorily preceding the MADM
adverb, do not since MADM adverbs are merged above VP. Thus, von Stechow's
(1996) analysis only derives the illicit linear orders:
(233) a. * [v' [v openedk] [AgrOP [DP the doori] [AgrO' [AgrO tk ] [µP [Adv carefully] [µ'
[µ tk ] [VPCAUSE [V tk ] [VPBECOME [V tk] [AP [Adv again] [AP [DP ti ] [A tk
]]]]]]]]]]
b. * the doori has been [v' [v openedk] [AgrOP [DP ti] [AgrO' [AgrO tk ] [µP [Adv
carefully] [µ' [µ tk ] [VPCAUSE [V tk ] [VPBECOME [V tk] [AP [Adv again] [AP
[DP ti ] [A tk ]]]]]]]]]]
If again headed a small clause taking the adjectival state predicate, i.e. [the
door open], as its subject, as in (234), the unexpected linear orders of again and
MADM adverbs in (227) and (228) could be accounted for under the assumption
that, after A-raising to VBECOME and VCAUSE, the AdvP remnant containing the
[DP the door] and again raises to specAgrOP (in actives) or to specAgrPrtP (in
passives), as shown in (235). The latter case would also have to involve the
stranding of again in specAgrPrtP since [DP the door] has to raise to specAgrSP:
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 261
(234) [VP [V CAUSE ] [VP [V BECOME ] [AdvP [AP [DP the door ] [A open ]] [Adv
again ]]]]
(235) v'
qp
v AgrOP
3
DP AgrO'
1
AgrO µP
1
Adv µ'
1
µ VPCAUSE
1
V VPBECOME
gh
V AdvP
#4
a. She openedk [the door again]i tk carefully tk tk tk tk
b. [The door]o has been openedk [to again]i tk carefully tk tk tk tk
Even though this analysis could be reconciled with the requirement that again
be within the scope of BECOME (cf. von Stechow 1996; Ernst 2002:279ff.) –
notably by open raising to and incorporating into VBECOME and by [V open
VBECOME] undergoing subsequent raising to v in actives and AgrPrtP in passives,
whence BECOME assumes a relatively high position thus being able to take
scope over again – it would force us to assume that restitutive again is a syn-
tactic predicate taking a small clause argument. Moreover, remnant raising of
[AdvP [AP [DP the door] [A t]] [Adv again]] to specAgrOP in actives and spec-
AgrPrtP in passives clearly lacks a trigger (likewise the obligatory stranding of
[AdvP [AP [DP t] [A t]] [Adv again]] in specAgrPrtP in passives). That is to say that
the postulation of these operations (together with a ban on [DP the door] vacat-
ing AdvP and leaving it behind in its base position) is nothing more than a re-
statement of the still unexplained empirical fact that restitutive again needs to
be in the immediate vicinity of the affected object in actives and the participle
in passives.
functional projections for which there are multiple merge sites within the
inflectional layer (cf. (145)), whose (in)accessibility to certain types of aspec-
tual adverbs was shown to be subject to semantic restrictions. Multiple occur-
rences of aspectual adverbs were shown to follow from the underlying hierar-
chy of functional heads, as identified by Cinque (1999), with certain
combinations being prohibited on semantic grounds. It emerged that aspectual
adverbs, which preferably occur between AgrSP and (NegP)TP, tend not to
cluster within the inflectional layer whence the higher of two aspectual adverbs
is often found to immediately precede or follow the subject, i.e. realized in the
complementizer layer. In section 4.3.3.1, the focus was on seemingly reverse
orders of aspectual adverbs with a higher or seemingly higher adverb realized
in postverbal position. It was shown that reverse orders pose only apparent
problems for the specifier analysis. Reverse orders comply in one of two ways
with the specifier analysis: either a genuine higher adverbs is realized within
, but licensed from within the inflectional layer, with the actual licensing
position being subject to the hierarchy of functional projections, or the reverse
order is an illusion created by homophony/polysemy, as was shown to be the
case with postverbal adverbs of frequency and repetition. Evidence for this
assumption derives from the well-known interpretational differences between
pre- and postverbal adverbs of frequency and repetition,101 from the co-occur-
rence of the two types of adverbs of frequency and repetition and from the fact
that postverbal seldom and rarely, as opposed to their preverbal counterparts,
do not induce sentential negation. On the basis of these findings, I have
suggested that postverbal adverbs of frequency and repetition be analyzed as
vP-internal adverbs, i.e. as specifiers of functional projections within the lexi-
cal layer (FreqP > RepP), which are merged into structure between µP/MeansP
and VP. Although the analysis proposed accounts for the placement of the ad-
verbs under consideration and their co-occurring with manner, degree of per-
fection and means-domain adverbs, it left us with some loose ends, notably the
contrasts between often, rarely and seldom vs. twice and Q-times in actives
with respect to adjacent manner, degree of perfection and means-domain ad-
verbs, and the reverse situation in passives. Finally, I concentrated on the
analysis of restitutive again which is restricted to co-occurring with change-of-
state verbs and which is tied to the postverbal position. I showed that restitutive
again is incompatible with verbs carrying the restitutive prefix re- and that
both restitutive again and verbs prefixed with restitutive re- are incompatible
with completion adverbs (e.g. completely, slightly), unless the verb takes an
incremental object. On the basis of the linear orders of restitutive again, com-
pletion adverbs, MADM adverbs I suggested that the functional projection host-
ing restitutive again, RestP, be sandwiched between AspPcompletive and VP.
Further, drawing on the comparison between the incorporation vs. stranding of
the completion adverb on the one hand, and the alternation of restitutive again
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 263
and the restitutive prefix re-, I argued that re-prefixation mirrors syntactic
incorporation and thus serves the defocalization of the adverb.
appearances, temporal expressions within the inflectional layer and the right-
periphery, i.e. in postverbal position, are subject to licensing under specifier-
head agreement within ReltP (cf. chapter 4.1) which, in the case of head-final
temporal expressions, may be instantiated in situ, i.e. specReltP may host
elements such as once, since or long ago. In the case of head-initial temporal
expressions, specifier-head agreement is instantiated 'long distance': the tempo-
ral expression is realized within the empty verbal structure governed by the
lexical verb (or its trace) and forms a representational chain with an expletive
operator in specReltP. This analysis will be shown to carry over to head-final
temporal expressions in postverbal position. In section 4.3.4.3, it will be argued
that temporal expressions in the left periphery of the clause (a) do not have a
'dating' function, but rather identify circumstances "in which actions or states
obtain" (cf. Geis 1987:177) and (b) are not derivationally related to temporal
expressions licensed under either direct or 'long distance' specifier-head agree-
ment within ReltP. The distribution of these elements relative to the subject
and relative to the inmates of the functional projections constituting the
complementizer layer, i.e PromP, TopP, FocP and Force(Fin)P, warrants the
postulation of an additional functional projection whose specifier position hosts
the temporal expressions under consideration. A summary is presented in sec-
tion 4.3.4.4.
Tomorrow, yesterday and once are inherently specified with respect to
tense: tomorrow is inherently specified as [–PAST] and thus cannot be con-
strued with verbs bearing a [+PAST]-specification. The reverse situation obtains
with yesterday and once which are inherently specified as [+PAST]:104
by any NP having such an N as its head, and assigns an Oblique Case to the NP
it labels" (cf. Larson 1985:606ff.):107
(243) DP[+F]
2
D N[+F]
# #
that day[+F]
∅ yesterday[+F]
(adapted from Larson 1987:251)
The analysis in (244) allows us to treat today, that day, next year, etc. on a par
with expressions like in March, at night, in the morning, on Monday, etc., all of
which contain overt prepositions and all of which are identical in distribution
with today, that day, next year, etc. and also with full-fledged prepositional
constructions headed by a transitive preposition, e.g. during intervals, a long
time after I went to bed, etc. (compare (240) – (242), (245) – (247) and (248) –
(250)):
(246) a. The final axe is expected to come after Birt takes over as the new
director general in March. (BNC)
b. Some of us […] work at night and sleep during the day. (BNC)
c. In fact, I don't usually shave in the morning… (BNC)
d. She will give reasons for her decision […] on Monday. (BNC)
This analysis not only guarantees that the nominal expressions under consid-
eration receive Case, but also sheds light on their distributional properties:
today, tomorrow and yesterday are structurally complex expressions and as
such are barred from preverbal positions.109
The assumption that structural complexity is what is at stake with respect to
the ill-formed strings in (242) and (247) is corroborated by the fact that single-
term temporal expressions, e.g. once, are perfectly acceptable both within the
inflectional layer ((251a) – (251c)) and in peripheral positions, as in (251d) –
(251g) ((251) from BNC):110
(251) a. It is difficult to imagine that this was once Britain's first electric main
line.
b. The petal was faded now but it had once been bright pink.
c. She once said, 'We're like two schoolboys!'
d. Once she said: 'I don't know what I should do…'
e. Once there was a wayside inn, now a pathetic ruin.
f. I was a soldier once.
g. I touched his chest, remembering that I had wanted to desert him once.
(253) a. Now she saw the terra incognita of [his] bedroom. (BNC)
b. And now there is Christopher Taylor who is currently being taught
Berber by Professor Neil Smith and Dr Ianthi Tsimpli… (BNC)
c. Then, gradually, her ideas would come together … (BNC)
d. Then she turned her head deliberately aside from them, touched her
heels against the chestnut's gleaming flanks and moved away. (BNC)
Now and then in pre-subject position and within the inflectional layer are un-
ambiguously temporal if accompanied by right:111
After(wards), before and since are intransitive prepositions which are inter-
preted relative to contextually known entities. The elements under considera-
tion are lexically specified with respect to the [±POST (ERIORITY)] relation they
introduce. That is to say that the event under consideration is interpreted as [–
POST(ERIOR)] relative to reference time introduced by after(wards) and since,
whereas the event is interpreted as [+POST] relative to the reference time intro-
duced by before. Being prepositional in nature, after(wards), before and since
most naturally occur in postverbal position, where they freely alternate with
their transitive counterparts and with temporal expressions of the type yester-
day and in the morning (cf. (256) – (258)):112
(256) a. Whatever had happened before or was to happen after, it is clear that
the population [...] was growing... (BNC)
b. In the time that has elapsed since, the receiving units are virtually
unchanged... (BNC)
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 269
As opposed to all other temporal expressions, after, before and since (but not
afterwards) are bad in pre-subject position (cf. (259)), unless accompanied by
either pre-head material, as in (260), or an internal argument as in (261):113
(260) a. Not long after, Lexandro also tumbled – not down the inside of the
heat sink, but certainly parallel to it. (BNC)
b. Many years before, he had contemplated writing a book of essays
about his childhood experience… (BNC)
c. Ever since, she has been delighting deaf people she encounters by
communicating in their own language. (BNC)
(261) a. After [DP that] England always looked the likely winners. (BNC)
b. After [AgrSP she had swept the room], she scrubbed the floor. (BNC)
c. Before [DP that] he was a successful history and art student at the De
Montfort University in Leicester. (BNC)
d. Before [AgrSP she could answer] there was a loud banging on the front
door. (BNC)
e. Since [DP that day] there have been no killings … (BNC)
f. Since [AgrSP he had been commissioned in ... 1858], he had 'almost
wholly devoted' himself to the building … (BNC)
(263) a. Since he will afterwards have a perfectly clear memory of all that has
occurred … (BNC)
b. Sometimes this involves heavy conflict, and he may ever after be
sensitive to the threat posed by powerful women who try to dominate
him. (BNC)
c. He long afterwards said he was not financially able to hire help to
improve his farm, … (www)
d. He long after wrote of it as "the place to which he owed his first
breathing," … (www)
(264) a. So many locations that would before have been snapped up by the
professionals will be available to the amateur. (BNC)
b. EIB money has never before been offered to Eastern Europe. (BNC)
c. Yussuf had been put in the cells to cool off and Owen had not long
before been down to see him. (BNC)
d. He long before was struggling from my hands. (www)
(265) a. Many houses have since been built in and near Back Lane. (BNC)
b. America has ever since remained a staunch bastion of manualism.
c. Also the blowpipes will long since have been thrown away. (BNC)
d. He long since has kissed that money goodbye. (www)
((b) from BNC)
Since the vast majority of so-called temporal adverbs are prepositional projec-
tions and more often than not barred from occurring within the inflectional
layer, a closer look at the syntax of temporal prepositions is in order so as to
shed light on the complexity issue and the (non)availability of prepositional
expressions within the inflectional layer.
(266) a. I had picked […]a place that I had visited many years before. (BNC)
b. I want to see that immediate family right afterwards. (BNC)
c. Her body was found in the car six days after she went missing. (BNC)
d. Everybody waited to eat until the priest arrived. (BNC)
e. Until the onset of his final illness, some ten days before his death,
there is no evidence that his health was failing. (BNC)
f. Coffee is available during intervals. (BNC)
While the lexical layer projected by intransitive prepositions, e.g. before and
afterwards in (266a, b), consists of just the preposition as in (267),116 that pro-
jected by transitive prepositions contains at least the preposition and its com-
plement, as is the case with after and until in (266a, b) which take sentential
internal arguments, AgrSP, as shown in (268):
(267) DegP
ei lexical layer
QP Deg'
tu
Deg P
# #
[QP many years] beforek tk
[Adv right] afterwardsk tk
(268) DegP
ei lexical layer
QP Deg'
ti
Deg PP
2
P AgrSP
# 6
[QP six days] afterk tk she went missing
untilk tk the priest arrived
Since before and during in (266c, d) take a nominal internal argument which is
subject to Case-checking, they project an AgrOP whose specifier and head
position are overtly targeted:
272 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Second, the specifier position of DegP may not only host degree modifiers, as
in (266), but may also be targeted by PP-internal material under move (cf.
below).117
In connection with the projections of spatial prepositions in chapter 3.2.4, I
conjectured that overt P-to-Deg-raising is warranted by linearization require-
ments, i.e. to prevent illicit postpositional orders which would arise if the
preposition stayed in AgrO, as shown in (271) for (266c, d). At first glance it
would seem that (271) properly accounts for genuine postpositional construc-
tions headed by ago:118
(270) a. [AgrOP [QP three weeks]i [AgrO' [AgrO ago k ] [PP [P tk] [QP ti ]]]]
b. [AgrOP [QP a couple of years]i [AgrO' [AgrO agok ] [PP [P tk] [QP ti ]]]]
(271) DegP
ro
QP Deg'
ro
Deg AgrOP
wu
DP AgrO'
fu
AgrO PP
1
P DP
# 4
a. * [QP some ten day] [DP his death]i beforek tk ti
b. * [DP intervals]i duringk t k ti
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 273
Second, under the Split PP Analysis defended here, both the preposition and its
nominal internal argument raise overtly to specAgrOP and AgrO, respectively.
Thus, after the respective movement operations have taken place, the linear
order within PP is non-distinct from the regular prepositional order (cf. (271)):
(273) a. [p' [p agok ] [AgrOP [QP three weeks]i [AgrO' [AgrO tk] [PP [P tk] [QP ti ]]]]]
b. [p' [p afterk ] [AgrOP [DP my trip]i [AgrO' [AgrO tk] [PP [P tk] [QP ti ]]]]]
One might be tempted to argue that [QP three weeks], [QP a couple of years ago]
and [AP long] are no arguments at all, but degree modifiers that are generated
outside the outer shell, i.e. within the functional structure dominating the lexi-
cal layer. Under this analysis, ago would be an intransitive preposition. Ana-
lyzing [QP three weeks], [QP a couple of years ago] and [AP long] as degree
specifiers would also neatly account for the ungrammatical strings in (274):
(276) DegP
3
QP Deg'
fu
Deg AgrOP
fu
QP AgrO'
fu
AgrO PP
1
P QP
# 4
a. [AP long]i ago k tk tk ti
b. [QP three weeks]i ago k ti tk tk ti
By the same line of reasoning, long in (277) is analyzed as the raised internal
argument of since:
(277) a. Also the blowpipes will long since have been thrown away. (BNC)
b. He long since has kissed that money goodbye. (www)
(278) a. * [He] got out the box of cigars he had not touched long since that
dinner. (BNC)
b. * It's got worse long since he took up golf again. (BNC)
Thus, the derivation of the strings in (277) is analogous to that of (276a), with
long raising to specDegP and since to Deg.
I would like to come back to temporal expressions as in (279) and suggest
the internal structure of these expressions is essentially identical to the struc-
ture headed by transitive temporal prepositions taking a nominal internal argu-
ment, e.g. two days before the hearing, during intervals, etc. Clearly, the tem-
poral expressions in (279) differ from those in (266c, d) and (269) in that the
preposition heading the entire construction is not inherently temporal (cf.
(279a) – (279c)) or even phonetically null, as in (279d) – (279f):
(279) a. Some of us have to work at night and sleep during the day. (BNC)
b. Will we plan tomorrow's work in the morning? (BNC)
c. She will give reasons for her decision in public on Monday. (BNC)
d. Today I have had a reply to a parliamentary question… (BNC)
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 275
(280) a. at, *in, *on, *Ø {night, noon, dawn, sunset; 5pm, 2am}
b. in, *at, *on, *Ø {May; 1900, 1901; the morning, the afternoon }
c. on, *in, *at, *Ø {Monday; May 1; Christmas Eve }
d. *on, *in, *at, Ø {today, yesterday, that day, next year}
(281) PP
ru
P[+TEMP] …
h
N[+TEMP]
The nominal expressions, e.g. [DP last night], [DP June], raise overtly to spec
AgrOP for reasons of Case-checking:
(283) a. [DegP [QP some time] [Deg' [Deg Øk ] [AgrOP [DP last night]i [AgrO' [AgrO tk ]
[PP [P tk ] [DP ti ]]]]]]
b. [DegP [QP early] [Deg' [Deg ink ] [AgrOP [DP June ]i [AgrO' [AgrO tk ] [PP [P tk ]
[DP ti ]]]]]]
276 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
In the following, the focus will be on the licensing of temporal adverbs and
temporal prepositional projections within and without the inflectional layer and
on the interpretation of temporal expressions outside the inflectional layer, i.e.
in right- and left-peripheral positions.
The fact that prepositional projections headed by ago (cf. (285)) may occur in
exactly the same position within the inflectional layer as once and as intransi-
tive prepositions, clearly shows that the distributional discrepancies that exist
among temporal expressions result from the (non)head-final status of the
temporal expression. A further case in point is 'postpositional' since in (286):
(285) a. The second bedroom had long ago been turned into an office. (BNC)
b. They had long ago come to terms with that sorrow. (BNC)
c. He had long ago decided never to say a word about it… (BNC)
(286) a. [The] blowpipes will long since have been thrown away. (BNC)
b. My nose had long since lost all feeling... (BNC)
c. Any normal person would long since have collapsed. (BNC)
(287) a. [AgrSP [AgrS' [AgrS hadk] [TP [T tk] [ReltP [Adv once ] [Relt' [Relt tk] [PerfP been
bright pink]]]]]]
b. [AgrSP [AgrS' [AgrS isk] [TP [T tk] [ReltP [DegP right now] [Relt' [Relt tk] [ProgP tk
making a go]]]]]]
c. [AgrSP [AgrS' [AgrS wouldk] [TP [T tk] [ModP [Mod tk] [ReltP [DegP before] [Relt'
[Relt ] [PerfP been snapped up]]]]]]]
d. [AgrSP [AgrS' [AgrS havek] [TP [T tk] [ReltP [DegP since] [Relt' [Relt tk] [PerfP been
built in …]]]]]]
e. [AgrSP [AgrS' [AgrS hadk] [TP [T tk] [ReltP [DegP long ago] [Relt' [Relt tk] [PerfP
abandoned the goal]]]]]]
(288) [TP [T willk] [ModP [Mod tk] [ReltP Op [Relt' [Relt ] … [vP [D ] [v' [v give] …
[ VP [ V ] [DegP next year]]]]]]]]
This assumption not only accounts for the fact that temporal expressions in
postverbal position take scope over the entire event, but also for the fact that
temporal expressions may not co-occur unless they are in an inclusion relation
or, as we shall see below, belong to two different classes:125
(289) a. * Illicit smokers were long ago fined yesterday for taking a puff.
b. * The petal was faded now but it had once been bright pink in the
evening.
c. * Jean is right now making a go of a job today.
d. Every year in October we say it is the finest thing… (BNC)
e. Don't forget Central Newsweek on Sunday at half past twelve with
Ann Davis, ... (BNC)
f. The next day at noon, Aurangzeb and the other young princes
escorted Dara through the palace … (BNC)
The analysis presented allows for a uniform licensing account for temporal
expressions that occur within the inflectional layer, i.e. are phonetically real-
ized in specReltP, and for temporal expressions that occur in postverbal posi-
tion either because they are not head-final or because they are focalized. The
assumption that postverbal temporal expressions inhabit VP, but are licensed
by forming a representational chain with an expletive operator in specReltP
also allows us to discard costly raising operations which would be needed if
postverbal adverbs factually occupied the same structural position as their pre-
verbal counterparts. A case in point is snowballing as discussed and discarded
in chapter 2.3 or Cinque's (2001a) antisymmetric account of prepositional
projections (see also Kayne 2002), under which prepositions are merged into
structure independently of their respective complements. The latter originate in
specifier positions of functional projections above VP (i.e TempP, LocP and
IOP), but below the prepositional layer constituted by P TempP > PLocP > PDatP,
the head positions of which host temporal, spatial and 'dative' prepositions,
respectively, and the specifier positions of which are overtly targeted, as shown
in (292) below for in (291) (both adapted from Cinque 2001a):
proceeds across PLocP since spatial expressions in English are always post-
verbal. Finally, how do genuine temporal adverbs like once fit into the overall
scheme: do they originate in specTempP or in the head position of PTempP?
(292) PTempP
th
PTemp'
th
PTemp PLocP
th
PLoc'
th
PLoc PDatP
th
PDat'
th
PDat AgrIOP
th
DPobj AgrIO
th
AgrIO TempP
2
DP Temp'
th
Temp LocP
2
DP Loc'
th
Loc IOP
2
DP IO'
th
IO VP
#
V
#
on in to Friday the garden John talk
I would like to conclude this section with a brief look at the structure of the
string in (291) under the analysis of spatial and temporal expressions devel-
oped in the present study. In (293), the lexical verb raises to the head position
of the outer shell where it assigns its external Θ-role to [DP I] under specifier-
head agreement. [DP I] and [KP to John] raise for reasons of Case-checking to
280 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
their respective checking sites. Both the temporal expression [DegP on Friday]
and the spatial expression [DegP in the garden] are realized within VP, but
bound from outside by their respective operators, i.e. OPspa in specSpaceP
within the lexical layer and OPtmp in specReltP within the inflectional layer.
Despite following the spatial expression, the temporal expression [DegP on
Friday] takes scope over [DegP in the garden]. Under the analysis proposed, the
relative scope of temporal and spatial expressions is not calculated within VP,
but within the interpretable part of the clause. That is, since OPtmp in specReltP
outscopes OPspa in specSpaceP, the temporal expression outscopes the spatial
expression:
(293) ReltP
2
OPtmp …
h
vP
2
D v'
tu
v SpaceP
2
OPspa Space'
2
Space AgrIOP
2
KP AgrIO'
1
AgrIO …
h
VP
1
KP V'
2
V
2
DegP
tu
DegP
5
Io OPtmp to talkedk OPspa tk [KP to John]m tk tm tk e [DegP on Friday]
the garden]
[DegP in
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 281
However, the assumption that the temporal expressions in (294) have a 'dating'
function just like their postverbal counterparts has often been challenged and
with it, implicitly, also the assumption that sentence-initial temporal expres-
282 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(297) a. Last year, Congress wisely defeated the president's energy bill…
b. On Sunday I foolishly left my window open during the day… (www)
(298) a. [On Rosai 's birthday] shei took it easy. ≠ [[On Rosai 's birthday]k shei
took it easy tk]
b. [Before Johni left], hei did the dishes. ≠ [[Before Johni left]k hei did the
dishes tk]
In chapter 3.2.4.1, we have argued that the functional projection hosting spatial
scene-setting expressions is SceneP, which is located between ForceP and
FocP because spatial scene-setting expressions follow the complementizer in
Force and precede raised elements in specFocP. The data in (299) and (300)
suggest that SceneP hosting temporal expressions also sits between ForceP and
FocP:
(299) a. The story of King Herod tells [ForceP that [SceneP on his birthday [he
made a supper for his lords… (www)
b. Sonny Boy Williamson said [ForceP that [SceneP once, [he was a wise
man who taught me so much… (www)
(300) a. [SceneP At Christmas [FocP NOT ONLY FANCY HOTELS [AgrSP are
overbooked]]].
b. She said that [SceneP last year [FocP BROCCOLI [AgrSP he hated (*it)
most]]].
(301) a. [SceneP After parties [TopP he [PromP always [AgrSP would drink milk]]]].
b. I was afraid [ForceP that [SceneP tonight [TopP he [PromP still [AgrSP could be
unwell]]]]].
Note that the illicit strings in (302) are not to be confused with the well-formed
ones in (303), where the temporal expression does not identify the circum-
stances in which the states of affairs denoted by the sentences obtain, but rather
serves to locate the referent of the subject in time, i.e. they are secondary
predicates:133
(303) a. The visit in April was followed by quality checks to verify the
qualifications of the Cyprus Institute of Marketing. (www)
b. Then, the chastity until marriage will be much more acceptable for
disabled nymphets. (www)
As has been the case with spatial expressions, the co-occurrence of temporal
expressions with a 'scene-setting' function and those with a 'dating' function is
subject to temporal inclusion, i.e. the time interval denoted by the higher tem-
poral expression must be able to include the time interval denoted by the lower
temporal expression (examples form Geis 1987:179):
In the remainder of this section, I would like to briefly come back to the issue
of mixed adverbial clusters, raised in chapters 2.3.1 and 2.3.2. As can be seen
in (305) and (306), scene-setting expressions may be mixed in the sense that
they contain spatial and temporal elements, with the order of spatial and tem-
poral elements being free:
(305) a. Under the trees in summer we played 'shop', using seeds or nuts for
groceries and bits of broken crockery for money. (www)
b. In Tacoma a few months ago, I heard a writer sighing that he was
having a hell of a time getting plots. (www)
c. Behind the iron curtain decades ago there were [sic!] research
wherein blind children were able to "see" colors… (www)
(306) a. Yesterday in Seoul, Mr. Powell met with visiting Australian Foreign
Minister Alexander Downer… (www)
b. Two weeks ago in Vancouver I received […] a badge with a motif of
children freed from the horrors of landmines. (www)
c. Just last week in Denver, we found an Army soldier on leave from
Korea…(www)
the fact that complex scene-setting expressions of the type under consideration
may co-occur with both temporal and spatial expressions, as shown in (307),
precludes a remnant raising analysis along the lines proposed by Cinque
(2004:694f.):
(307) a. Under the trees in summer we played 'shop' behind the bench every
Sunday afternoon, … (cf. (305a))
b. Yesterday in Seoul, Mr. Powell met with […] Alexander Downer at
the opera at 8 p.m. (cf. (306a))
Moreover, had raising taken place, it would have been two operations, one
affecting the spatial expression and one affecting the temporal expression. An
analysis along these lines not only would obscure the fact that the italicized
strings in (305) – (307) represent one constituent, but also would fail to account
for the absence of Minimality effects.
Since there are no sequential restrictions on spatial and temporal expres-
sions contained in a spatio-temporal scene-setting cluster, it would seem that
the internal structure of the italicized strings in (305) – (307) is one of
coordination. However, since the elements under consideration belong to
different semantic subclasses, genuine coordination is out:134
To account for the fact that the expressions under consideration set a complex
spatio-temporal scene which, semantically speaking, is the intersection of two
domains against which the rest of the clause is interpreted, I suggest a make-
shift analysis of the italicized strings in (305) – (307) in terms of 'relativiza-
tion'. The structure is essentially that we assume for coordinate structures, i.e.
relatives are conceived of as second conjuncts in asymmetric conjunction
structures (cf. Koster 1999a), with the empty-headed :P being a Boolean
operator indicating set intersection (cf. Koster 1999a:4):
(309) a. [SceneP [:P [DegP Under the trees] [:' [: ] [DegP in summer]]] [AgrSP we
played 'shop']]]]
b. [SceneP [:P [DegP Yesterday] [:' [: ] [DegP in Seoul]]] [AgrSP Mr. Powell met
with … Alexander Downer]]]]
Whichever analysis will prove adequate for the internal structure of complex
scene-setting expressions, their external structure is clear: they are merged in
specSceneP.
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 285
cational phrases raised to specDegP, e.g. long in the context of ago and since.
Temporal expressions such as in the morning, on Monday, yesterday and next
year were shown to have the same internal structure as temporal expressions
headed by a transitive temporal preposition taking a nominal internal argument.
To account for the fact that these expressions, in spite of not being headed by a
temporal preposition, receive a temporal interpretation, I suggested that the
temporal feature of the nominal expression incorporates into the preposition.
This assumption was shown to be corroborated by the modificational properties
of the expressions under consideration: by virtue of the preposition containing
the temporal feature and overtly raising to Deg, degree modifiers can be
licensed under specifier-head agreement. On a larger scale, the analysis
presented not only accounts for the distributional and interpretational differ-
ences between temporal adverbs with a 'dating' function and those with a
'scene-setting' function, but also allows for a uniform licensing account of the
former – irrespective of whether they occur in specReltP in 'person' or 'by
proxy', so to speak. A further advantage of the present analysis was argued to
lie in its simplicity: by merging head-initial and focalized temporal expressions
in VP and allowing them to form a representational chain with an expletive
reference time operator in specReltP, we can not only elegantly account for the
fact that temporal expressions may not co-occur (unless they are semantically
nested) since the actual temporal expression and the expletive operator in
specReltP are of the same kind, i.e. they both are markers of reference time, but
we also avoid getting caught between the Scylla of prepositional attractors and
the Charybdis of remnant movement, as under Cinque's (2001a) proposal.
However, as the contrast between (311a) and (311b) shows, the licensing of
focusing particles is contingent on the presence of a focus associate, i.e. a
prosodically prominent element, the so-called focus associate: 137
As opposed to the full pronoun him, the pronoun it is a weak, i.e. unstressable
pronoun and thus cannot function as a focus associate for the focusing particle.
Bayer (1999) maintains that the unacceptability of (311b) follows directly from
the semantics of focusing particles which requires "a non-trivial set such that
there is an ALTERNATIVE to the focused element" (cf. Bayer (1999:60)).138
Only full pronouns like him in (311a) have (discourse-relevant) alternatives,
e.g. with respect to gender, person and deixis:
The vital condition that focusing particles be associated with a focused element
is met in (313), where even associates with an adjacent, stressed DP (313a) or
with a stressed subconstituent of an adjacent DP ((313b) – (313e)):139
(315) a. [DP JOHN] even [vP [v gave] his daughter a new bicycle].
b. John even [vP [v gave] HIS daughter a new bicycle].
c. John even [vP [v gave] [DP his DAUGHTER] a new bicycle].
d. John even [vP [v gave] his daughter [DP a NEW bicycle]].
e. John even [vP [v gave] his daughter [DP a new BICYCLE]].
f. John even [vP [v GAVE] his daughter a new bicycle].
(adapted from Jackendoff 1972:248)
The puzzle focusing particles pose for syntactic analysis lies in their combina-
torial versatility (cf. (310)) which seems to be only restricted by the require-
ment that they associate with some prosodically prominent element (cf. (311) –
(315)) to which there exists a contextually relevant set of alternatives (cf.
(313)). The fact that focusing particles may or may not be adjacent to their
focus associate, as shown for even in (313) and (315), gives rise to the question
of constituency. Whereas movement data as in (316) suggest that focusing
particles form a constituent with the maximal projections they take scope over
(see also Bayer 1999:60ff.; 1996:14ff.),141 data as in (317), where the focusing
particle is not adjacent to its bound focus (FOCUS), seem irreconcilable with
this assumption (cf. Jacobs 1983:46, 101):
(317) a. [They] [only [vP teach [DP the SPELLING of the word]] … (BNC)
b. She [even [vP fancied [DP the Duke [KP of [DP EDINBURGH]]]]]. (BNC)
c. He [also [vP sent Kaptan [PP OUT OF THE ROOM]]. (BNC)
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 289
(318) He neither sent Jim a rubber frog nor Tim a dead squirrel, he only1 sent
[1Kim1] [1a box of dust bunnies1]
(319) a. [It] requires [only [ForceFinP that some … people behave in this way]].
b. Fen gave [only [DP a grunt of acknowledgement]].
c. [Even [DP May]] was so cold.
d. I sent this to the principal, who gave it [only [KP to the workers]].
e. [I] sent [only [DP a civil servant]] to the preliminary discussions.
f. [Better] to depend [even [KP on dubious private money]] than a
criminal state.
g. On rare occasions, [they] perform the activities [only [Adv poorly]].
Technically, Jacobs' (1983) proposal could easily be integrated into the analy-
sis of verbal projections defended in this study, if we allowed for adjunction to
apply within the lexical layer. That is to say that the focusing particle could be
analyzed as attaching to any verbal subconstituent within vP hosting the focus
associate and also to AgrSP, as shown in (321) for the strings in (319):
290 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(321) a. [v' [v requiresk] [VP [only] [VP [V tk] [ForceFinP that …]]]]
b. [v' [v gavek] [AgrOP [only] [AgrOP [DP a grunt of acknowledgement]i [AgrO'
[AgrO tk] [VP [V tk] [DP ti]]]]]]
c. [AgrSP [even] [AgrSP [DP May]i [AgrS' [AgrS was] … [QP [Q' [Q so] [AP [DP ti]
[A' [A cold]]]] …]]]
d. [v' [v gavek] [TelP [DP it]i [Tel' [Tel tk] [AgrIOP [only] [AgrIOP [KP to the
office workers]j [AgrIO' [AgrIO tk] [AgrOP [D ti] [AgrO' [AgrO tk] [VP [KP tj] [V' [V
tk] [D ti]]]]]]]]]]]
e. [v' [v sentk] [TelP [only] [TelP [DP a civil servant]i [Tel' [Tel tk] [AgrIOP [KP to
the preliminary discussions]j [AgrIO' [AgrIO tk] [AgrOP [D ti] [AgrO' [AgrO tk]
[VP [KP tj] [V' [V tk] [DP ti] ]]]]]]]]]]
f. [v' [v dependk] [AgrIOP [even] [AgrIOP [KP on dubious private money]j
[AgrIO' [AgrIO tk] [VP [V tk] [KP tj]]]]]]
g. [v' [v performk] [AgrIOP [DP the activities]j [AgrIO' [AgrIO tk] [µP [only] [µP
[Adv poorly] [µ' [µ tk] [VP [V tk] [DP tj]]]]]
(322) a. [AgrSP [ Nur] [AgrSP [DP Paul]] [AgrS' [AgrS schickte] sie
Only Paul sent she
einen Plastikfrosch]]].
a rubber frog
"Only to Paul did she send a rubber frog"
b. [AgrSP [ Sogar] [AgrSP [DP tote Eichhörnchen]
Even dead squirrels
[AgrS' [AgrS schickte]… sie ihm]]].
sent she him
"Even dead squirrels she sent him"
As for the strings in (314), Jacobs' (1983) analysis predicts them to be perfectly
well-formed since the focusing particle takes scope over the entire sentence
and thus should be able to associate with any focused element within its scope
(cf. Bayer 1996:22f. for discussion). Further problems of Jacobs' (1983)
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 291
(323) a. [Only [when Helen and Tony told him that Helen expected a baby
before Christmas]] did he seem to come out of his stupor of misery.
b. [Only [in books]] had she seen buildings as magnificent as those
which rose up before her, and she had never, anywhere, seen so many
people.
Recall that only constituents that bear the affective feature [ FOC] may raise to
specFocP and thus induce inversion, i.e. raising of the finite non-lexical verb to
Foc.142 Clearly, since the feature [FOC] resides in the focusing particle only, not
in its focus associate (whence the strings in (324) are ill-formed), this mecha-
nism cannot be accounted for under the assumption that the focusing particle is
a co-constituent of the verbal projection:
(324) a. * [When Helen and Tony told him that Helen expected a baby] did he
seem to come out of his stupor of misery.
b. * [In books] had she seen buildings as magnificent as those which rose
up before her.
(325) XPq
3
PRT q XP
(326) In a structure [XP PRT XP], PRT has +operator status if XP is a possible
scope domain for PRT; otherwise PRT has −operator status. (Bayer
1996:56)
Bayer (1996:57; 1999:55f., 65ff.) argues that scoping takes place in a special,
semantically motivated functional projection, PrtP, the head of which invaria-
bly takes as its complement a verbal projection, which contains the verb, its
internal argument(s) and the external argument viz. the subject and thus con-
stitutes a Complete Functional Complex in the spirit of Chomsky (1986a:169;
1986b:15). PrtP demarcates the domain of quantification which, under the Split
VP Analysis defended in this study, is vP:
(328) a. Annai could [PrtP [Prt even] [vP ti ESCAPE FROM PRISON].
b. ANNAi could [PrtP [Prt even] [vP ti escape from prison].
c. Annai could [PrtP [Prt even] [vP ti ESCAPE from prison].
d. Annai could [PrtP [Prt even] [vP ti escape FROM PRISON].
(adapted from Bayer 1999:53)
However, the structural analysis in (327) and (328), where focusing particles
are taken to be projecting heads, seems to cancel out the analysis of focusing
particles in (325) above, according to which these elements are non-projecting
heads which take phrasal complements of any categorial format ((310) &
(319)). This inconsistency is resolved by the following stipulation, which
correlates the (non)operator status of focusing particles to their (in)ability to
project 'categorial features':
(329) PRT being a Minor Functional Head cannot project categorial features
unless it occupies an operator position; in this case, PRT heads the func-
tionally defined phrase PrtP. (Bayer 1999: 63)
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 293
As for the question of how focusing particles that combine with e.g. sentential,
prepositional, adjectival or adverbial constituents, as in (330), associate with
PrtP, i.e. how they attain propositional scope, Bayer (1996, 1999) proposes an
LF raising analysis, which we shall briefly look into:
LF raising of [DP only syntax] to specPrtP, as in (331), activates the head fea-
tures in Prt under specifier-head agreement, so that the relevant features of only
are realized in the head position of PrtP, which is a +operator position. As a
consequence, the copy of only in [DP only [DP syntax]] in specPrtP as well as its
base occurrence in the internal argument position of V, which is a −operator
position, can be deleted:148
(332) [Hem would [PrtP [DP only [DP syntax]] [Prt' [Prt ] [vP [tm ] [v' [v studyj]
[AgrOP [DP only [DP syntax] [AgrO' [AgrO tj] [VP [V tj] [DP only [DP
syntax]]]]]]]]]]
While the structure in (333a), where the entire vP is chosen as the focus associ-
ate of only, converges as it is, the structure in (333b), where only [DP syntax] is
294 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
chosen as the focus associate of only, requires LF raising of the focus associate
to specPrtP and its subsequent reconstruction in its base position:
(334) [Hem would [PrtP [DP syntax] [Prt' [Prt only] [vP [tm ] [v' [v studyj] [AgrOP [DP
syntax] [AgrO' [AgrO tj] [VP [V tj] [DP syntax]]]]]]]]]
The mechanisms sketched also apply in cases in which the focus associate
assumes a vP-external position in overt syntax, e.g. specAgrSP in (335a) or
specFocP in (335b):
While the focusing particle in (335a) is merged into structure as the head of
PrtP and hence has +operator status, the focusing particle in (335b) achieves
+operator status by [XP PRT [XP]]-raising through specPrtP (on its way to
specFocP), thereby activating the head features in Prt (represented as only in
(336b)). In (336a), overt raising is restricted to the focus associate:
(336) a. [AgrSP [D Anna]i [AgrS' [TP [T could] [PrtP [D ti ] [Prt' [Prt even] [vP [D ti ] [v'
escape from prison ]]]]]]]
b. [FocP [DP Only George]j [Foc' [Foc would m] [AgrSP [D we] i [AgrS' [AgrS tm] …
[PrtP [DP tj ] [Prt' [Prt only] [vP [D ti ] [v' [V invite k] … [VP [V tk] [DP tj
]]]]]]]]]]
Even though Bayer's (1996, 1999) analysis allows for an elegant solution of the
problem of focus association, it suffers from the stipulation that focusing parti-
cles in pre-vP position project onto PrtP:
(337) a. But remember they [PrtP [PrtP only] [vP teach the spelling of the word,
not an understanding of its meaning]].
b. He may [PrtP [PrtP even] [vP refuse to search for any answer at all]]…
c. She [PrtP [PrtP even] [vP fancied the Duke of Edinburgh]].
d. He [PrtP [PrtP also] [vP sent Kaptan [PP out of the room]]. (BNC-data)
A further problem with his analysis relates to the assumption that Prt invariably
takes as its complement a Complete Functional Complex (cf. Chomsky (1986a:
169; 1986b:15)), which we have identified as vP. Bayer (1996, 1999) does not
discuss data as in (338), where the focusing particle assumes a relatively high
position within the inflectional layer ((338) from www):
(338) a. They would [[only] [ReltP [Relt have] had access to the source and
destination of e-mails and phone calls]].
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 295
b. [Duisenberg] said the effects of any further interest rate move now
would [[only] [beP [be be] felt in a year's time]].
c. [It] would [[only] [AuxP [Aux be] making common cause with the
Taliban…]].
d. I have [[only] [PerfP [Perf been] skiing for three years]].
e. My memory is [[only] [ProgP [Prog being] recognized as half size]]?
Notice that only cannot be taken to 'float' within the inflectional layer since its
scope, once it has attained operator status, is fixed for once and for all.149 Thus,
PrtP must assume a higher position than the pre-vP position, i.e. the comple-
ment of Prt only minimally consists of vP. This assumption is corroborated by
the strings in (339), where the focusing particle forms a constituent with
adverbs based in the inflectional layer:
The fact that constituents of the type [[Prt] [Adv ]] cannot precede the finite non-
lexical verb, as shown in (340), clearly shows that the PrtP must be located
below AgrSP the head of which hosts the finite verbal element:150
(340) a. * ... it [[even] [Adv cleverly]] has harnessed the power of that current It-
phrase…
b. * … he [[even] [Adv intentionally]] has defaulted on some school loans.
c. * This first ray potency [[only [Adv lately]] has been experienced…
d. * He [[even] [Adv briefly]] had mentioned the possibility…
The fact that focusing particles cannot outscope sentential negation (341), but
may well be within the scope of sentential negation (342), suggests that PrtP is
located below NegP, as shown in (343):151
(341) a. * They would only not have had access to the source…
b. * I have only not been skiing for three years.
c. * It does [even cleverly] not harness the power…
d. * He would [even briefly] not mention the possibility…
296 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(343) [AgrSP [AgrS' [AgrS ] [NegP [Neg' [Neg ] [PrtP [Prt' [Prt ]]]]]]]
The amendment relating to the structural position of PrtP is not meant to imply
that focusing particles that take into their scope chunks of the extended verbal
projection are generated in the head position of PrtP (and thus count as pro-
jecting heads). That is to say that focusing particles taking chunks of the
extended verbal projection as their complement form a constituent with that
complement, as shown in (344):
(344) a. She [vP [even] [vP fancied the Duke of Edinburgh]]. (BNC)
b. They would [ReltP [only] [ReltP [Relt have] had access...
c. I have [PerfP only [PerfP [Perf been] skiing for three years]]].
In the remainder of this section, rather than following Bayer's (1996, 1999) LF
licensing account of focusing particles, I would like to suggest that the formal
licensing of focusing particles, i.e. association with focus, takes place in overt
syntax. Specifically, I assume that constituents of the type [XP Prt XP] are
licensed as members of a representational expletive-associate chain. Under this
analysis, which is clearly inspired by Haegeman's (1995) treatment of postver-
bal negative constituents (cf. section 4.3.2), [XP Prt XP] is associated with an
expletive operator (Op) residing in specPrtP, as shown in (345), with PrtP
being dominated by AgrSP (or NegP (if projected), cf. (343)). As we have seen
in connection with the discussion of constituency, the focusing particle 'perco-
lates' its semantic, i.e. quantificational features, which Bayer (1996:15), for
short, represents as q, to the constituent containing it. The quantificational fea-
ture q is not only represented on the focusing particle and on [XP Prt XP], but
also triggers the projection of purely semantically motivated PrtP. Under the
assumption that the formation of the representational chain containing the
expletive operator in specPrtP and the associate [XP Prt XP] involves the quan-
tificational feature q, this feature is present on every member of the chain so
that [XP Prt XP], by virtue of being coindexed with the expletive operator on
specPrtP, enters into a specifier-head agreement relation with the head of PrtP:
(345) [TP [T ] [PrtP OPi [Prt' [Prt ] … [XP Prt [XP ]]i … ]]]
[q] [q] [q]
Under the analysis presented, bound focus is syntactically uniform in the sense
that the focusing particle takes as its complement the constituent representing
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 297
or containing its focus associate, i.e. there is no need to assume that the projec-
tion properties of preverbal focusing particles are different from those that
combine with DP, KP, QP, etc.:
(346) PrtP
2
Op Prt'
2
Prt …
a. … would OPi [ReltP only have had access]]i
b. … would OPi [beP only be felt in a year's time]]i
c. … may OPi [vP even refuse to search]]i
d. … has OPi [ReltP [AdvP only lately]]i been experienced]]
e. … OPi [vP gave [DP only a grunt]]i]
f. … to OPi [vP perform the activities [AdvP only poorly]]i]
Rather than assuming that the quantificational subject DPs as in (347a) and
constituents raised to specFocP as in (347b) form a representational chain with
an expletive operator in specPrtP, I assume that (on their way to the sentential
subject position) they have passed through specPrtP thus entering into a speci-
fier-head agreement relation with Prt:152
(347) a. [AgrSP [DP Even May]k [AgrS' [AgrS was] … [PrtP [DP tk] [Prt' [Prt ] [QP [Q so]
[AP cold [DP tk]]]]]]]
b. [FocP [ForceFinP Only when Helen and Tony told him that Helen expected
a baby before Christmas]k [Foc' [Foc did i] [AgrSP [D he] [AgrS' [AgrS ti] [PrtP
[ForceFinP tk] [Prt' [Prt ] seem to come out of his stupor of misery tk]]]]
There is one more case to consider, namely the occurrence of focusing particles
in 'postposition':
Since focusing particles under the favored analysis are complement taking
heads, it should be possible for their complement, i.e. their focus associate to
strand the focusing particle in situ (if warranted by stylistic factors which, at
the moment, are not clear to me).153
Under the assumption that only in (348a) takes [PP on Sundays] as its
complement, the PP-final occurrence would follow automatically after overt
raising of [P on] and [DP Sundays] to the head position of pP and specAgrOP,
respectively:
298 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(349) [pP [p on]k [AgrOP [DP Sundays]i [AgrO' [AgrO tk] [PP [only] [PP [P tk] [DP ti]]]]]]]
Likewise, the vP-final occurrence of even can be seen as the result of [QP some
flowers] raising overtly to spec-AgrOP:
(350) [vP [D I] [v' [v bring]k [AgrIOP [D you]j [AgrIO' [AgrIO tk] [AgrOP [QP some flow-
ers]i [AgrO' [AgrO tk] [VP [D tj] [V' [V tk] [QP [even] [QP ti ]]]]]]]]]]
Again, scoping takes place indirectly, i.e. via chain formation of [PP [only] [PP
[P tk] [DP ti]]] and [QP [even] [QP ti]] with the expletive operator in specPrtP.
Finally, strings as in (351), where the focusing particle occurs between the
subject and the finite non-lexical verb, pose a potential problem for the analysis
presented:154
(351) a. The villagers initially thought they only would be questioned and they
sat on the ground as ordered. (www)
b. If your friend has this service she only can receive SMS messages
from another… (www)
c. We only are staffed to deliver equipment. (www)
d. In his testimony he indicated that he only had been asked to advise as
to whether Canadians could own TCP, not provide advice on the full
set of transactions. (www)
Under one interpretation, the focusing particle takes the pronominal subject as
its focus associate. In this case, the problem lies in the derivation of the linear
order pronoun > only from [D only pronoun]. In the cases under consideration,
only can also take as its focus associate the lexical verb, which is not a problem
as such. The problem lies in only preceding the finite non-lexical verb in AgrS,
i.e. only assumes a higher position than it ought to (recall that focusing parti-
cles cannot outscope negation). At the moment, I have no solution to this
problem.
vene between verbal heads within the functional layer (338) and also may form
constituents with adverbs that are home-based within the inflectional layer
(339), I have argued that PrtP assumes a relatively high position within the
inflectional layer, albeit below NegP (viz. (341) vs. (342)). Instead of assuming
with Bayer (1996, 1999) that focusing particles in pre-vP position are project-
ing heads, while focusing particles taking complements other than vP are not, I
maintained that the projection properties of focusing particles are uniform
across constructions in the sense that they are non-projecting, but complement-
taking heads. This move also allows for a uniform treatment of scoping in
terms representational chains involving an expletive operator in specPrtP and
the constituent that represents or contains the bound focus. Under the analysis
presented, both the interpretation and the licensing of focusing particles take
place in overt syntax.
(352) a. [The author] has carelessly not made this distinction. (www)
b. We had foolishly not used contraception, just the once, when I
conceived. (www)
300 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
c. If they can't wisely evaluate and use information, it affects all their
coursework. (www)
d. He certainly wouldn't have calmly pulled a gun and forced the other
man [out]. (www)
(353) a. * While moving we discovered the excellent back way in, which I
have cleverly since blocked with the big bookcase.
b. * […] Angel had foolishly long ago shed his potentially sexy Irish
accent for a standard American L.A. accent.
c. While moving we discovered the excellent back way in, which I
have since cleverly blocked with the big bookcase. (www)
d. […] Angel had long ago foolishly shed his potentially sexy Irish
accent for a standard American L.A. accent. (www)
e. She may have thought he would patiently wait until Augustus died
and then take over. (www)
f. She had calmly slit her wrists two days ago, and that was the very
last thing she had ever done. (www)
(354) a. Then they leave them there, in among her own kittens, and when she
returns the chances are that [AgrSP [D she] [AgrS' [AgrS willi] [AttP [Adv
calmly] [Att' [Att ti] [TP [T ti] [ModP [Mod ti] [ReltP lie down and let all the
kittens feed from her without examining them in detail]]]]]]]. (BNC)
b. I'd try to take control … by asking her a question and [AgrSP [D she]
[AgrS' [AgrS wouldi] [AttP [Adv cleverly] [Att' [Att ti] [TP [T ti] [ModP [Mod ti]
[ReltP respond. (www)
(355) a. [AgrSP [DP The author] [AgrS' [AgrS hasi] [AttP [Adv carelessly] [Att' [Att ti]
[NegP [not] [Neg' [Neg ti] [TP [T ti] [ReltP ti made this distinction]]]]]]]].
b. [AgrSP [D we] [AgrSP [AgrS' [AgrS hadi] [AttP [Adv foolishly] [Att' [Att ti] [NegP
[not] [Neg' [Neg ti] [TP [T ti] [ReltP ti used contraception, just the once,
when I conceived]]]]]]]].
(358) habitual > repetitive > frequentative > subject-oriented > celerative >
anterior > terminative > continuative > perfect > retrospective > proxi-
mative > durative > generic > prospective
302 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(359) a. I have even carefully provided the justification for drawing this
conclusion. (www)
b. Moreover, the police […] have even violently intervened in
demonstrations when it is unnecessary. (www)
(360) a. They will even be carefully studying the actual cell itself again.
b. You may opt for a Minimal installation if you will only be
passively viewing the meeting. (www-data)
(361) a. Ruduse discovers that Lirpa has been able to recover and reactivate
the tracking device – but has carelessly only recovered one part of it.
b. I'm carefully not even thinking about titles for the next books.
(www-data)
All three cases are expected under the analysis defended: the functional pro-
jection hosting subject-attitude adverbs has three merge sites at their disposal,
i.e. 'AgrSP > AttP* > TP > … > ReltP > AttP* > PerfP > AuxP > AttP* >
ProgP', and focusing particles are licensed by an expletive operator residing in
specPrtP, which is located immediately below TP (cf. section 4.3.1.1). The
functional projection hosting the subject-attitude adverb, i.e. AttP, is merged
between AgrSP and TP in (361) and between AuxP and ProgP in (360). The
case in (359) is only relevant in so far as the constituent [AttP Prt AttP] sits
below the licensing site for the focusing particles which also is an admissible
position for subject-attitude adverbs.
In section 4.3.3, we have argued that the functional projections hosting
aspectual adverbs can be merged into structure between AgrSP and TP (and
thus above or below NegP), between ReltP and PerfP, between AuxP and
ProgP and, though quite markedly, between beP and the lexical layer.
The highest merge position accounts for the fact that individual members of
all aspectual adverb classes may precede and follow sentence-negating not:157
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 303
(362) a. The lift was not usually operated after six in the evening. (BNC)
b. The sun will not always shine on you. (BNC)
c. The senator was again not looking at me but at the sea, and his voice
was immensely sad. (BNC)
d. Unless they are called as witnesses, victims are often not told when a
case is coming to court. (BNC)
Since we have already established that aspectual adverbs may precede, but not
follow subject-attitude adverbs (cf. (356)), we now turn to the co-occurrence
options for aspectual adverbs and temporal expressions. Given that temporal
expressions within the inflectional layer invariably sit in specReltP with ReltP
being consistently merged into one and the same position, and given that
aspectual adverbs may be merged above and below ReltP, both orders, i.e.
'aspectual adverb' > 'temporal expression' and 'temporal expression' > 'aspec-
tual adverb', should be available. As can be seen in (363), the linear order
'temporal expression' > 'aspectual adverb' is attested for individual members of
all aspectual classes:
The linear order 'aspectual adverb' > 'temporal expression' is not attested:
Sequences as in (365) are not quite what they seem. Generally in (365a) takes
narrow scope over afterwards and thus induces as contrastive reading, e.g.
generally, but not always. Again in (365b) also seems to take scope only over
afterwards. Often, always, only recently, soon and long in (365c) – (365g) read
like degree modifiers of the temporal expressions, e.g. many times before
(365c), ever since (365d), not long afterwards, almost immediately afterwards
(365e, f) and a long time afterwards (365g). Almost in (365h) modifies long
since rather than expressing prospectivity. Finally, still in (365i) has the
adversative reading:
As has been the case with subject-attitude adverbs, there are three possible
scenarios in which aspectual adverbs may co-occur with focusing particles: (a)
by forming a constituent (cf. (366)), (b) by being contained within a constituent
also containing focusing particle (cf. (367)) and (c) by preceding the constitu-
ent that contains the focusing particle (cf. (368)). Again, all expected patterns
are attested:158
(366) a. But above all, engineers are only slowly acknowledging that many
Western solutions […] are not the right answer for many parts of the
Third World. (BNC)
b. Although the medal has only recently been created, it was a proud and
memorable day when we went to the Polish Embassy in London to
receive it. (BNC)
(367) a. I believe I was still sleeping in a crib when I was treated to my first
closedown on our CBS affiliate, WSAW – it may even have still been
known as WSAU then. (www)
b. The three final stages of modification in Fig. 1.2 will only be briefly
mentioned here. (BNC)
(368) a. So yours are already only being paid time and a half? (BNC)
b. He's almost even enjoying the pain (BNC)
We have already seen that temporal expressions may co-occur with subject-
attitude adverbs and with aspectual adverbs within the inflectional layer only if
they precede them (cf. (353), (363) & (364)). Since both temporal adverbs and
sentence-negating not occupy unique, i.e fixed positions within the inflectional
layer, and since the functional projection hosting temporal expressions sits
below the functional projection hosting sentence-negating not, temporal
expressions are expected to occur within the scope of not, but not vice versa.
This expectation is borne out. Notice that the strings in (370) are unacceptable
only under the intended reading, not under the constituent-negating reading:159
(369) a. The decisions made in 1968 and 1969 by Heseltine […] opened a door
which could not afterwards be closed. (BNC)
b. We should also note, that the Pope […] has not long ago publicly
expressed the legitimate right of Judaism and Islam to co-exist with
Christianity. (www)
c. I am unaware of any other ISP that hasn't long ago closed this
security and privacy breach. (www)
d. Be grateful you are not right now suffering that sickening sinking
feeling that you are not, in fact, protecting America from any sort of
marauding terrorists. (www)
306 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(370) a. * The decisions made in 1968 and 1969 by Heseltine opened a door
which could afterwards not be closed.
b. * [The pope] has long ago not publicly expressed the legitimate right
of Judaism and Islam to co-exist with Christianity.
c. * Be grateful you are right now not suffering…
(372) a. The current owners would therefore be dispossessed of their land and
only afterwards be given a right to challenge the legitimacy of the
process.
b. Somehow, even in Seattle which was even then known as a melting
pot, a transplanted Kentucky Hillbilly was considered lower class.
(373) a. Some of these experiments have since been even reproduced. (www)
b. [This] announcement came as a startling confirmation of what had
before been only suspected. (www)
The reason we do not find the following presumably derives from the fact that
only (if not taking the subject as its focus associate) assumes a higher position
than it ought to as regards licensing (cf. chapter 4.3.5):
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 307
Leaving out sentence-negating not and focusing particles, which have a virtu-
ally free distribution, the partial linear orders in (375), can be translated into
the overall order presented in (376) and illustrated in (377):
(377) a. He had long since always cleverly been mentally rehearsing his texts.
b. He would afterwards rarely have willingly been financially supporting
them.
c. He has right then already foolishly been emotionally involved.
d. Many of those who would before have only seldom actively been
socially misbehaving are now indulging in being impish.
The nonadmissibility of the orders in (378) does not have a semantic source
since exactly these orders obtain in cases in which the temporal expression is
realized within , as shown in (379) and (380):
(380) a. [The] vacuum pumps are typically shut down just after the chamber is
filled with H2 because they are no longer needed then. (www)
b. [It] was suitable for the ceramics fired between 750° to 950 °C,
because they have to be quickly chilled afterwards. (www)
The fact that (379) and (380) are perfectly well-formed, can mean one of two
things: either the subject-attitude adverb and the aspectual adverb are merged
below the functional projection hosting the temporal expression, i.e. ReltP, or
they are merged above ReltP in which case they would sit higher than the
expletive reference time operator. Needless to say, the second option is far
more interesting. And, what's more, it can be shown to be actually available:
since ReltP invariably sits below NegP (cf. (369) vs. (370)) and since both sub-
ject-attitude adverbs and aspectual adverbs may precede not (cf. (352) &
(362)), they assume a higher position than ReltP, the specifier position of
which hosts the expletive reference time operator:
(381) a. The author has carelessly not made this distinction long ago.
b. The Professor has cleverly not avoided me since.
c. The sun will not always shine on you afterwards.
d. Clearly, money that would often not be available now if it had to
come from the public purse is sometimes found from private sources.
((d) from www)
Given these considerations, the fact that temporal expressions within the
inflectional layer do not tolerate other adverbs to their left cannot be attributed
to either syntactic or semantic restrictions. At the moment I do not have an
answer to the question of which factors force the temporal expression to show
up in postverbal position in cases in which the functional projections hosting
subject-attitude adverbs or aspectual adverbs are merged above ReltP.160
The findings presented in this section clearly invalidate the scope theorists'
claim that the relative orders of adverbs within a given domain are free unless
they give rise to semantic clashes or scope violations (cf. Haider 2000; Ernst
2000, 2002).
Ironically, the empirical facts adduced in support of this claim (cf. (382) –
(385)) support our analysis since, in all cases considered, one of the adverbs
does not occur within the inflectional layer, but either within (on Friday
(382a), at 4:00 AM most nights (383a), at that time (385a)) or within the
complementizer layer (on Friday (382b), at 4:00 AM most nights (383b), occa-
sionally in (384a), stupidly (384b), at that time (385b)) (examples gleaned from
Ernst 2002:358ff.):161 162
(383) a. Recently, Vivian has again been waking up at 4:00 AM most nights.
b. [At four/recently] Vivian again performed her new composition.
(384) a. Occasionally, Dan would stupidly forget to tell Security that he was
going out.
b. Stupidly, Dan would occasionally forget to tell Security that he was
going out.
4.3.7 Conclusion
The aim of this chapter was to provide a uniform account for the distribution
and licensing of adverbs within the inflectional layer as contingent on speci-
fier-head agreement within designated functional projections. The main chal-
lenge for a uniform analysis of elements as diverse as subject-related adverbs,
inherently negative adverbs, aspectual adverbs, temporal expressions and
focusing particles was shown to lie in the distributional versatility displayed by
these elements relative to both (non)finite verbal elements populating the
inflectional layer and sentential negation, whose syntax was central in sections
4.1 and 4.2, respectively. Subject-related adverbs (section 4.3.1) were shown to
fall into two distinct syntactic classes, subject-attitude adverbs and subject-
oriented adverbs, with only the former being inmates of the inflectional layer
and thus susceptible to follow finite non-lexical verbs and to precede or follow
sentential negation and nonfinite non-lexical verbs (section 4.3.1.1). The obsta-
cles that the distributional versatility of these elements creates for the 'tradi-
tional' specifier analysis, e.g. optional raising operations and violation of the
Head Movement Constraint, were argued to evaporate once we admit that the
functional projection hosting subject-attitude adverbs may be merged into more
than one position provided they are merged only once and provided the overall
hierarchy of adverbs is respected. It could be shown that the adjunction analy-
sis, unless supplemented with output filters, cannot properly account for the
data. Subject-oriented adverbs (section 4.3.1.2) were shown not to be deriva-
tionally related to subject-attitude adverbs and to be merged above Agr SP. The
elements under consideration were argued to invariably sit in specSoP, which
may be crossed by a topical subject. The analysis proposed was shown to be
superior to the adjunction analysis on both empirical and theoretical grounds: it
explains why phonetically reduced non-lexical verbs may cliticize on the sub-
ject and it allows for a uniform treatment of subject-oriented adverbs in pre-
and post-subject position. Section 4.3.2 was hybrid in nature since it focused
on both inherently negative aspectual adverbs (also figuring prominently in
310 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(386) a. The petal was faded now but it had once been bright pink. (BNC)
b. The petal was faded now but it had been bright pink {once, the day
before yesterday, etc.}
c. * The petal was faded now but it had once been bright pink the day
before yesterday.
These observations carry over to all preverbal adverbs that have an expletive
counterpart which is vital for the licensing of postverbal occurrences of the
preverbal 'adverb(ial)' (be it non-head final or focalized), as shown in the
following for habitual, repetitive, frequency, (im)perfective, durational and
pure domain adverb(ial)s, which, as we have seen in chapter 3.2.3, have their
licensing site within the lexical layer:
(387) a. People in their thirties are generally trying to impress the boss. (CC)
b. People in their thirties are trying to impress the boss {generally, in
most cases}
c. * People in their thirties are generally trying to impress the boss
{generally, in most cases}
(388) a. The periodic 'Fed bashing' which one has come to expect of the
incumbent political administrations has again resurfaced. (BNC)
b. The periodic 'Fed bashing' has resurfaced {once more, yet again}.
c. * The periodic 'Fed bashing' has again resurfaced {once more, yet
again}.
The final section of this chapter was devoted to the co-occurrence and linear
orders of adverbs within the inflectional layer. It was shown that subject-
related adverbs and (inherently negative) aspectual adverbs, which enjoy dis-
tributional freedom when they are either on their own or co-occurring with
sentence-negating not and/or focusing particles, are extremely rigidly ordered
if they co-occur within the inflectional layer, with (inherently negative) aspec-
tual adverbs invariably preceding subject-related adverbs. Both subject-related
adverbs and all classes of (inherently negative) aspectual adverbs invariably
follow temporal expressions, which, in turn, invariably follow sentential nega-
tion:
(394) a. [I] must use the available evidence and clues to piece together a series
of events that often have long since passed. (aspectual > temporal)
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 313
b. She had calmly slit her wrists two days ago, and that was the very last
thing she had ever done. (subject-attitude > temporal)
c. She had wisely thawed the cake, slowly. (subject-attitude > aspectual)
Notes
1
See among others Emonds (1976), Zagona (1988), Ouhalla (1990), Koopman & Sportiche
(1991), Roberts (1993), Solà (1996), Cinque (1999), Haumann (2000), Ernst (2002).
2
Another difference between modal and auxiliary verbs relates to agreement morphology.
While modal verbs lack agreement morphology altogether, auxiliary verbs, like lexical verbs,
display rudimentary agreement inflection in the present tense paradigm (see also Roberts
1993:242, 317, passim; Warner 1993:33ff.).
3
The fact that there are subsets of lexical verbs whose paradigms, for example, do not contain
a passive participle is not relevant to the present discussion since (non)passivizability of a
lexical verb does not affect its categorial status, namely V.
4
Being, like having in (6), is restricted to occurring in gerundive constructions.
5
There are, of course, occurrences of being with a verbal complement, as in (8), but note that
constructions that contain be with verbal complement are invariably passives.
6
In light of these considerations, I refute Roberts' (1987, 40) claim that there is only one
participle form which, depending on whether it is governed by have or be, is spelled-out as a
past participle or as a passive participle (cf. Haumann (2000:165ff.)).
7
The idea of representing syntactically the "three temporal entities with respect to which a
clause is interpreted" (Zagona 1988:66), i.e. E(vent time), R(eference time) and S(peech time),
is not novel (see also Enç 1987; Zagona 1988, 1992; Stowell 1993; Hornstein 1990; Zeller
1994). Notice, however, that these analyses differ with respect to the assumption of whether R
is present in all clauses (cf. Hornstein 1990; Solà 1996; Zeller 1994), or only in clauses that
contain have (cf. Zagona 1988).
8
Under his analysis, the head of TP hosts the finiteness specification in terms of [+fin], and the
head of ZP the tense specification in terms of [+PAST] (cf. Zeller 1994:228ff.).
9
One might be tempted to assume that now, which is typically assumed to denote speech time,
is located in specTP. However, under Zeller's (1994) analysis, specTP is reserved for PRO,
which, in matrix clauses, is coreferential with S and, in embedded clauses, controlled by and
coindexed with the event argument of the matrix verb. The establishing of sequence of tense is
achieved via coindexation and control.
10
With respect to the syntax of perfect tenses in German, Zeller (1994:279) assumes two ZPs,
with the higher one corresponding to reference time and the lower to event time. The functional
projection hosting the perfect auxiliary is sandwiched between these two ZPs.
11
In retrospect, sentential negation, as induced by not, can be considered as a catalyst for the
Split IP Hypothesis according to which agreement and tense are represented as distinct syntac-
tic heads. It should be noted that under Pollock's (1989) original proposal TP assumes a higher
position than AgrP (however, Pollock (1989:384, note 19) admits that the relative order of TP
and AgrP is subject to parametric variation). (16) corresponds to the structure employed by
Belletti (1990, 1994), Ouhalla (1991), Haegeman (1995), Rowlett (1995), Zanuttini (1996),
Rizzi (1996) and others.
12
Languages differ with respect to the realization of sentential negation as the specifier or/and
the head of NegP (cf. Pollock 1989:414; Ouhalla 1990:191ff.; 220; Haegeman 1995:117ff.,
126ff., 165ff.; Schafer 1995:151; Vikner 1995:20, 144; Rizzi 1996:73; Haegeman & Zanuttini
1996:123ff. for discussion).
314 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
13
Cf. Kayne (1989), Haegeman (1995), Rowlett (1995), Zanuttini (1996:182ff.). But see
Zanuttini (1996) for arguments that not is "an adverbial element which can be adjoined to, or
perhaps occur in the specifier position of, any maximal projection" (Zanuttini (1996:192).
14
Pollock (1989) assumes that not occupies the head position of NegP, but conjectures that a
"possibly preferable solution would be to analyze these adverbs as specifiers of a NegP with an
empty head serving as the landing site for Verb Movement" (Pollock 1989:405, note 36).
15
But see Williams (1994:169f.) for arguments that sentence-negating not is a head which has
"one feature of its own, [–tense]" (cf. Williams 1994:169). The [–tense]-feature accounts for
the fact that not is restricted to combining with a nonfinite verbal projection, as captured by (i),
and it also accounts for the fact that the projection of not, i.e. notP, fails to function as the
predicate of a matrix clause, as shown in (ii) ((i) – (iii) taken from Williams 1994:170, 169):
(i) not: [–tense]; ___ XP
(ii) * John [not left].
(iii) John must (not) have (not) been (not) thinking.
16
(20) is a recent and more formal version of Klima's (1964:312ff.) Affect-attraction.
17
See also Haegeman (1995:190) for discussion.
18
Cf. Haegeman (1995:272) and Rizzi (1996:67) for discussion.
19
Evaluative affixes, as opposed to n't, are also systematic exceptions to the Unitary Base
Hypothesis (cf. Aronoff 1976) in that they combine freely with nouns and adjectives.
20
This is also captured by Lieber's (1980) Feature Percolation Conventions (FPC) in (i) & (ii),
which are illustrated in (iii) and (iv), respectively ((i) & (ii) quoted from Spencer 1991:203ff.):
(i) FPC II: The features of an affix are passed to the first dominating node which
branches.
(ii) FPC III: If a branching node receives no features by FPC II, then it is labelled by the
next lowest labelled node.
(iii) [[strong]A [ly]Adv ]Adv
(iv) [[alberò]N] [ino]]N
21
Cf. Mätzner (31880:426) and Sweet (1891:123).
22
For reasons to be made clear, I choose the term 'subject-related' instead of 'subject-oriented'.
23
Recall from chapter 3.2.1.1 that adverbs like cleverly in preverbal position in actives, as in
(i), do not receive a 'manner' interpretation:
(i) She cleverly avoided this topic.
The reputed ambiguity between subject-related adverbs and adverbs with alleged manner inter-
pretation disappears in the context of sentential negation. First of all, only subject-related ad-
verbs may precede sentential negation:
(ii) She is cleverly not avoiding this topic.
(iii) * She is not cleverly avoiding this topic (ok under the constituent negation reading of not).
Second, as has been argued by Bellert (1977:339ff.), negative sentences containing subject-
related adverbs presuppose the negated sentence without the adverb, as in (iv), whereas nega-
tive sentences containing manner adverbs presuppose the corresponding affirmative sentence
without the manner adverb, as in (v):
(iv) She has not been cleverly avoiding this topic. → She has not been avoiding this topic.
(v) She has not been answering the questions cleverly. → She has answered the questions.
24
Cf. Cinque (1999:71ff., 89f., 203, note 33), Laenzlinger (2000:113f.). See also Zubizarreta
(1982), Roberts (1985), Travis (1988) for arguments that subject-related adverbs are related to
root modality.
25
On the basis of (i) & (ii), Svenonius (2002) argues that a "head movement account is inade-
quate, since heads cannot generally cross other head positions. If all of the auxiliaries in [(i)
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 315
DH] occupy lower positions, and if they are all taken to have moved in [(ii) DH], then each one
will have to have crossed at least two other head positions" (Svenonius 2002:208):
(i) Howard foolishly may have been trying to impress us.
(ii) Howard may have been foolishly trying to impress us.
26
For the status of subject-related adverbs as sentence adverbs, ad-VPs or VP-external adverbs
see Jackendoff (1972:49ff.), Bellert (1977:340ff.), McConnell-Ginet (1982:145ff.), Travis
(1988:289ff.), Sportiche (1988:431ff.), Laenzlinger (1998:88ff.), Ernst (2002:54ff.), Frey &
Pittner (1999:22ff.), Pittner (2000a).
27
See also Ernst (2002:105) and Pittner (1999:113f.) for discussion. As we shall see in section
4.3.1.2, subject-related adverbs in post-subject position as in (33) are marked (see also Ernst
(2002:105, 399ff.)).
28
Ernst (2002:105) points out that the variant in (32c) in which the subject-related adverb
follows n't is better than the variant with not. We shall come back to this issue in section
4.3.1.1.
29
This analysis entails that subject-attitude adverbs as in (32) may precede and follow senten-
tial negation (see also Ernst 2002:105).
30
But see Jackendoff (1972:76) for the judgment that in strings as in (i), where a so-called S-
adverb follows two or more non-lexical verbs the adverb, are unacceptable:
(i) She will have cleverly been (cleverly) avoiding this topic.
31
Cf. Bobalijk (1999:27), Svenonius (2002:203ff.), Ernst (2002:116ff.) for criticism.
32
Recall that, since base adjunction always gives rise to the creation of specifier positions and
thus to the extension of the checking domain of a given head, the base-adjoined adverb can
enter into a specifier-head agreement relation within a given functional head. Cf. Chomsky
(1995:235ff.) and Ernst (2002:399ff.) for discussion. Also, recall that adjunction structurally
differs from substitution only in that it forms a two-segment category and that specifiers and
adjuncts display different properties which relate to the A/A' distinction (cf. Chomsky
1995:248).
33
Evidence for the assumption that NegP may assume more than one position, e.g. below TP,
derives from the scope interactions between negation and other elements as well as from the
fact that sentences may contain more than one NegP (with the individual negative elements
canceling each other out), as shown in (i) (taken from Cinque 1999:126):
(i) He couldnae have no been no working (Hawick Scots)
'It is impossible that he had been out of work'
34
The array of possible adjunction sites is narrowed down in the context of sentential negation:
if cleverly follows not, as in (38a), it cannot be adjoined to TP and if cleverly precedes not, as
in (38b), cleverly must be adjoined to TP.
35
Note that adverbs like cleverly or foolishly as opposed to intentionally or voluntarily are
impossible to come by in terms of root modality. But see Cinque (1999) for the conjecture that
"manner adverbs used in the 'subject-oriented' way may fall into the class [of Mod ability]"
(Cinque 1999:203, note 33). The fact that subject-attitude adverbs that are said to correlate
with volitional modality, e.g. intentionally in (i) – (iii), assume a number of positions, would
seem to require the postulation of different merge sites ((i) – (iii) from www):
(i) [Some] children will intentionally have accidents or wet their beds on purpose.
(ii) Here is a clearer picture […] than we could have intentionally tried to create.
(iii) This brochure has been intentionally removed from this site.
36
At first glance, it would seem that the specifier analysis loses much of its original restrictive-
ness if we allow for multiple merge sites for functional projections hosting adverbs, but, as we
shall see below, this is not the case.
316 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
37
Native speaker's judgments vary with respect to the admissibility of epistemic adverbs (e.g.
probably) within the scope of non-finite aspectual auxiliaries, as in (50a). Cf. chapter 5.2.1.4
for discussion.
38
The presence of postverbal stupidly prevents preverbal carefully from being misinterpreted
as a manner adverb.
39
See also Cinque (1999:19f.; 2004:692f.) for discussion of data similar to (54b).
40
Notice incidentally that the same problem arises under analyses that deny the existence of
AgrSP, e.g. analyses according to which the subject occupies specTP (cf. Chomsky 1993:21ff.;
1995:234, 248ff.; Ernst 2002:399ff.; Svenonius 2002:213ff., 228ff.).
41
Other analyses of subject-oriented adverb entail (i) a predicate-argument structure with the
adverb taking both the subject and the proposition as its arguments (Jackendoff 1972:70ff.), (ii)
a higher abstract predicate act which is modified by the adverb and which takes the proposition
as its argument (McConnell-Ginet 1982:172ff.) or (iii) a parallel structure analysis (Sportiche
1988:431f.).
42
Here, we could also assume that adjunction takes place in the base position of the modal or
auxiliary, i.e. the complex verbal head raises to AgrS.
43
Notice that the root adjunction analysis in (60) also fails to account for the fact that
parentheticals, e.g. as we all know in (i), may intervene between the adverb and the modal or
finite auxiliary (see also Ernst 2000:295):
(i) She cleverly, as we all know, has been avoiding this topic.
44
Nakamura (2000) conjectures that quite could be either head-adjoined to the complex verbal
element or to the adverb. See also Sadler & Arnold (1994) for an analysis of attributive adjec-
tives along these lines.
45
Notice that under Ernst's (2002) analysis TP conflates TP and AgrSP, i.e. in the system
employed in this study, TP should read AgrSP.
46
Under Ernst's (2002:399) analysis, [TOPIC] is a secondary feature on tin very much the same
sense as [NEG] under Rizzi's (1997:317f.) analysis (cf. section 4.3.2). See also Svenonius
(2002:228) for discussion.
47
See also Cinque (1999:115) for the conjecture that subject raising may either be triggered by
some feature in the higher functional projection, e.g. the feature [TOPIC], or by scope require-
ments, e.g. 'the base position of subject-oriented adverbs must be minimally c-commanded by
the subject' (cf. Frey & Pittner 1999:24). In the latter case, the subject-oriented adverb must
originate in spec?P (cf. below for discussion).
48
The same asymmetry between subject-oriented and subject-attitude adverbs arises in cases
in which the subject is focalized:
(i) * A WOMAN cleverly has been avoiding this topic, not a man.
(ii) * SEVEN WOMEN carefully have deactivated the bomb, not three men.
(iii) A WOMAN has cleverly been (cleverly) avoiding this topic, not a man.
(iv) SEVEN WOMEN have cleverly been (cleverly) deactivating the bomb, not three men.
49
In earlier models of generative grammar, raising to the left periphery was characterized as
adjunction-based CP-recursion (cf. Vikner 1994a, b, 1995; Watanabe 1992; Authier 1992;
Culicover 1993).
50
See Hoekstra (1992:194), Müller & Sternefeld (1993:486), Kiparsky (1995:140ff.) for ad-
verb preposing involving TopP.
51
Rizzi (2002) maintains that Mod subsumes "all the features licensing adverbs, i.e. the whole
of Cinque's hierarchy, and the dedicated Mod positions in the left periphery which make ad-
verbs 'prominent' (cf. Rizzi 2002:19).
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 317
52
Subject-oriented adverbs are discourse-related in the sense that they express the speaker's
judgment about both the subject and the action under consideration (cf. Quirk & Greenbaum
16
1986:245), not the subject's attitude (see also Frey & Pittner 1999; Pittner 1999, 2000a,
2004).
53
But see Svenonius (2002:228ff.) for analyzing adverbs in pre-subject position as adjoined
above the subject within the functional projection hosting the subject. See also Laenzlinger
(2000:118ff.) for arguments that adverbs in pre-subject position occupy the outer, i.e. A'-speci-
fier position within the functional projection hosting the subject.
54
See also Bresnan (1977), Culicover (1993), Watanabe (1992).
55
Frey & Pittner (1999:26ff.) and Pittner (2000a) maintain that instrumental, spatial and
benefactive adverb(ial)s, as in (i) – (iii), as opposed to temporal, frame and reason adverb(ial)s,
as in (iv) – (vi) pattern with raised arguments in the sense that they are raised to the pre-subject
position and thus do not help to remedy the structure. This is illustrated in (i) and (ii) for spatial
and temporal adverbials (adapted from Frey & Pittner 1999:26):
(i) * Whoi do you think that [in Ben's office]k ti lay on his desk tk?
(ii) Whoi do you think that [on Rosa's birthday] ti took it easy?
However, a cautionary note is in order with respect to Frey & Pittner's (1999:26f.) claim that
the adverbial in (i) is raised to TopP, whereas the adverbial in (ii) is base-adjoined to TopP.
First, intervening adverb(ial)s only ameliorate That Trace Violations, i.e. the strings under
consideration are not fully acceptable. Second, for most native speakers, there is no discernible
difference in acceptability between (i) and (ii). Third, Frey & Pittner's (1999:26f.) distinction
between raising and base-adjunction with respect to the types of adverb(ial)s under considera-
tion is problematic with respect to strings as in (iii), where the allegedly base-adjoined con-
stituent [DP on Saturdays] precedes the raised constituent [PP on Domplatz].
(iii) He said that [DP on Saturdays], [PP on Domplatz], you can get stuffed olives.
56
Cf. Ernst (2002:410f.) for discussion.
57
Cf. Rizzi (1997:310ff.) and Laenzlinger (1998:112ff.).
58
The fact that the referents of the structural subjects in passives containing strings à la 'non-
finite be > {reluctantly, etc.}' tend to be inanimate (cf. (89b) – (89d) & (i) – (iii)) or non-hu-
man (cf. (iv)) may be taken as supportive evidence for the claim that the adverbs are agent-
oriented:
(i) The downside of the Budget were the additional duties on fuel, […] but these had to be
reluctantly accepted in view of environmental considerations. (BNC)
(ii) Nothing could be willingly given up […] (BNC)
(iii) Sometimes a certain piece of data may be deliberately omitted. (BNC)
(iv) Every generation of pig and chicken must be deliberately domesticated … (BNC)
59
See McConnell-Ginet (1982:148) for the argument that the ambiguity of strings as in (86)
should be attributed to adverbial homonymy, e.g. reluctantly VP vs. reluctantly S.
See also Frey & Pittner (1999:22ff.) and Frey (2000:118ff.) for discussion. Frey & Pittner
(1999:22ff.) suggest that the structures corresponding to the ambiguous strings in (86) differ
with respect to the adjunction site of the adverb, i.e. if the adverb is adjoined to V', it is con-
strued with the implicit agent, whereas, if adjoined to the projection of be, it is construed with
the passive subject.
60
This property is also shared by inherently negative modals (e.g. can't, won't) and auxiliaries
(e.g. wasn't, hadn't). See section 4.2.1 for discussion.
61
Under Pollock's (1989:405) analysis, the contrasts in (98) derive from not and inherently
negative adverbs occupying different structural positions, the head position of NegP and the
VP-initial positions, respectively. But see Pollock (1989:405, note 36) for the conjecture that
318 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
inherently negative adverbs had probably been better taken to occupy the specifier position of
an empty-headed NegP. See Bouchard (1995:37) for the distributional properties of not.
62
See Haegeman (1995:185f.) for an early account of 'covert feature checking' in terms of
operator replacement at LF. Haegeman (1995) suggests that – since the post-verbal negative
constituent [DP nothing] in (i) cannot raise to specNegP (cf. (ii)) – (i) be analyzed as in (iii),
where specNegP hosts an expletive, i.e. a non-overt operator ((i) from Huddleston (1984:420),
(iii) adapted from Haegeman (1995:186)). Since non-overt operators, like any other non-overt
category, are subject to identification by an overt and contentful element, the identification of
the non-overt constituent in specNegP is achieved by forming a chain with the overt and con-
tentful negative DP internal argument of the verb solves:
(i) It solves [DP nothing].
(ii) [AgrP It [NegP [ nothingi ] [Neg' [Neg ] [TP … [VP solves ti ]]]]]
(iii) [NegP OP[NEG]i [Neg [NEG]] [TP … [VP solves [DP nothing[NEG]i ]]]]
At LF, the negative associate of the expletive operator raises to specNegP and replaces the
operator (cf. Haegeman 1995:186). Note that the necessity that the non-overt negative operator
be identified also explains why (iv) is ruled out: there is no negative constituent the expletive
negative operator could form a chain with, whence the expletive negative operator cannot be
identified and NegP cannot be licensed ((iv) from Haegeman 1995:187):
(iv) * He [NegP OP [Neg 0] said this]
63
To account for the fact that constituent-negating not in not long ago in (105a) has narrow
scope, Klima (1964) proposes extending the phrase structure rule for Time (adapted from
Klima 1964:308):
(i) Time → (neg) long {ago, after, before}
See also Lasnik (1972:30) for arguments that not "is generated, by the phrase structure rules,
within the adverbial in which it appears [e.g. (ii)]":
(ii) not long ago, not long after, not long before, not far from here, not far away, not infre-
quently, not unexpectedly, not unnaturally, not unreasonably, not uncommonly, not
unjustly, not surprisingly
64
Constituent-negating not shares three properties with evaluative affixes (cf. chapter 4.2.1 for
discussion): not is 'transparent' with respect to 'categorial information', takes multiple comple-
ments and changes the semantics of the base.
65
Williams (1994:172) uses the term 'inverted' to describe the semantic contribution of nega-
tive prefixes such as in- and un-.
66
Rudanko (1980:356) argues that not far away and not infrequently in (113) do not constitute
negative expressions in the strictest sense since the expressions have non-negative paraphrases.
Haegeman (1995:272), on the basis on Rudanko's (1980) analysis, argues that adverbials such
as not infrequently and not unreasonably are parentheticals, whereas not long ago, not long
after, etc. are referential expressions. She further argues that "[a]n adjunct like not long ago
refers to a specific moment in time, while an adjunct like not often [a genuine operator] does
not" (Haegeman 1995:273).
67
Had the negative feature of [AP not far] percolated onto DegP or any higher, we would
expect that constituent to license an and neither tag and to induce inversion after having raised
to specFocP. As can be seen from the illicit strings in (116), these expectations are not borne
out.
68
According to Rudanko (1980:350f.), the inverted structures in (119) differ prosodically from
their uninverted counterparts in that only the uninverted structures contain an intonational
break. See also Jackendoff (1972:364ff.) for discussion.
69
As we have seen in (118), specifier-type notXPs fail to percolate their negative feature onto
the dominating projection. Haegeman (1995) points out that "[t]he fact that NEG-percolation is
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 319
blocked from the specifier position cannot be due to a generalized ban on percolation from
specifiers" (Haegeman 1995:281), as witnessed by the strings in (i) & (ii), where the wh-
feature on how long and how far percolates onto DegP (taken from Haegeman 1995:281):
(i) How long after (the party) did it happen?
(ii) How far away does she live?
In connection with her discussion of pied-piping in West Flemish, Haegeman (1995:283) con-
jectures that affective features of the type [NEG] can satisfy the Affect Criterion within the
negated constituent, i.e. the negated constituent does not attain operator status. The reverse
situation obtains in cases where the Affect Criterion is not satisfied within the negated constitu-
ent, i.e. the negative feature percolates onto "the maximal projection which is turned into a
negative operator" (Haegeman 1995:283).
70
This proposal is inspired by Bayer's (1996:58f., 1999:67f.) "activation account" of operator
features on focusing particles (cf. chapter 4.3.5; see also Bayer 1996:174ff. for a discussion of
dynamic agreement).
71
Notice that the unacceptability of (129b) is unaccounted for under the assumption that the
negative feature on no(t)XP percolates optionally within the extended preposition projection.
72
Dormant [NEG]-features do not give rise to the projection of NegP, not even under inversion
(cf. (119), (120) & (127)).
73
At this point, it is not clear to me whether this analysis requires that [NEG]-feature be gener-
ated in the head position of TP or whether the raising of the auxiliary or modal verb to the head
of FocP could be interpreted as Last Resort. I leave this issue for further research.
74
This section has greatly benefited from discussions with Stefan Engelberg.
75
'Outer aspect', i.e. Smith's (1991, 2004) 'viewpoint aspect' and Dahl's (1985) 'grammatical
aspect', focuses on the "different ways of viewing the internal temporal constituency of a situa-
tion" (cf. Comrie 1976:3), e.g. viewing the event as a whole (perfective) or its internal structure
(imperfect/progressive) (see also Alexiadou 1997: 86ff.; Cinque (1999:83ff.).
76
Time related aspectual adverbs are 'aspectual' in the narrow sense in that they encode a
"temporal relation between two events, one of which is linked to reference-time, and the other
is of the same sort as the first and must have a specific relation to it" (Ernst 2002:341).
77
Cf. Cinque (2001b) for a modified analysis of aspect-related functional heads/ projections
and aspectual adverbs.
78
As we have seen in chapter 3.2.2, completion adverbs do not occur within the functional
layer of the clause (cf. (i) – (iii)), but are confined to the lexical layer, where they originate as
complements of lexical verbs into which they incorporate unless they are focalized (or syntac-
tically heavy):
(i) * The car has {slightly, completely, totally, utterly} been damaged.
(ii) * The car will {slightly, completely, totally, utterly} be damaged.
(iii) * They will {slightly, completely, totally, utterly} have damaged the car.
79
Recall from chapter 3.2.1 that the postverbal occurrences of these adverbs receive a manner
interpretation:
(i) Nigel would […] retie his shoelace slowly. (BNC)
80
Not all members of the class under consideration may occur in all positions, e.g. rarely and
seldom do not combine with present participles of non-lexical verbs.
81
Postverbal occurrences of so-called celerative adverbs do not pose a problem; they are
manner adverbs unless commaed-off, as in (i):
(i) Well, come on, bring him in to the fire, quickly! (BNC)
82
The status of (143h) is marginal.
320 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
83
Thus, for example, have in (144e), (144g) & (144h) would have raised across [Mod t], been
would have raised across [Relt t] and [Mod t] in (144g) & (144h), and across [Relt t] in (144i), and
being in (144h) would have raised across [Aux t], [Perf t], [Relt t] and [Mod t].
84
It should also be noted that, if viewed on a larger scale, preverbal again and still preceding
other aspectual adverbs often fail to receive an aspectual interpretation, but are interpreted as
'however' and 'in spite of', respectively:
(i) Again they are often the unseen … victims of crime. (www)
(ii) This again will soon change as technology increases. (www)
(iii) Needless to say we still recently sent Terry to take pictures of both. (www)
(iv) We still soon decided we should get something done.
85
Notice incidentally that Strings of the type n't + aspectual auxiliary + aspectual adverb are
dispreferred (presumably the adverb cannot be readily construed as event-related (www-data):
(i) [As] he is a small man and he wouldn't have often had to stand so far back…
(ii) You can't have ALREADY finished your dinner.
(iii) I wouldn't have always had the confidence to handle that.
(iv) [But] we haven't been characteristically tracking them in the district.
86
Manner homophones of aspectual adverbs are irrelevant here:
(i) Helen spoke clearly, slowly, and loudly. (BNC)
87
This string does not provide any counterevidence to the claim that postverbal quickly and
slowly unless set off intonationally as in (165b) are manner adverbs, as in (i). Thus, (165b) is
interpreted as a command that the time that elapses between speech time and event time be
short, while slowly in (i) specifies the process of retying his shoelace taking some time:
(i) Nigel would […] retie his shoelace slowly. (BNC)
88
Notice that still has a manner adverb homophone which, of course, occurs in postverbal
position.
(i) She stood perfectly still.
89
Notice also that characteristically in (165j) cannot be interpreted as a manner adverb since
the manner adverb how is wh-moved.
90
See also McCawley 1968; Dowty 1979:261ff.; Pittner 2000b; Ernst 2002:255ff.; 2004:733ff.
91
Read as event-modifiers, preverbal rarely, seldom, often, twice and again are perfectly
acceptable.
92
However, since again expresses repetition, and since the number of repetitions has no built-
in upper limit, we may consider the option of analyzing the frequency adverbs in (196) as
modifiers of again. Under this analysis, the linear order in (197) could be taken to reflect the
underlying order again > frequency adverb, which would be in line with Cinque's (1999) pro-
posal.
93
The fact that repetitive again hardly ever co-occurs with frequency adverbs in passives, and
then only in preverbal position, may be taken as further supportive evidence for the claim that
preverbal again is not within vP:
(i) The place formerly selected for going into action had been again twice reconnoitered
during the wait (www)
94
The order 'completely > again' may be reversed if completely is focalized:
(i) Just download it again completely. (www)
95
Cf. Ernst (2002:277ff.; 2004a:773f.), Pittner (2000b:365), von Stechow (1996), Kamp &
Roßdeutscher (1994:195).
96
Cf. Pittner (2002b:336) for adjunction to the verb and Ernst (2002:277ff.; 2004a:773) for
adjunction to VP.
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 321
97
Also, the relative frequency of the linear orders, i.e. 'completion adverb > re-V' (221a-d) vs.
're-V > completion adverb' (221e-h) is important. A quick search of the internet (using google)
yielded 30.200 matches for 'been {partly, partially, halfway, entirely, completely} {rebuilt,
restored}' and only '462 matches for 'been {rebuilt, restored} {partly, partially, halfway,
entirely, completely}'.
98
Notice that, when co-occurring with verbs carrying the restitutive prefix, MADM adverbs
trivially follow their regular distributional patterns both in actives and in passives:
(i) She has (*carefully) reopened the door carefully.
(ii) I have (*correctly) reinstalled the scanner correctly.
(iii) You can (*electronically) refill your card electronically.
(iv) This door has been (carefully) reopened carefully.
(v) The scanner has been (correctly) reinstalled correctly.
(vi) Your copy card will be (electronically) refilled electronically.
99
Both the linear orders, i.e. 'V > MADM V >again' and 'MADM adverb > V > again' are
perfectly acceptable under the repetitive reading of again (cf. (198) & (199)).
100
Under his analysis, repetitive again assumes a structural position outside VP. See Ernst
(2002:279ff.) for low right-adjunction of restitutive again and high left- or right-adjunction for
repetitive again.
101
Cf. Thomason & Stalnaker (1973:200), Andrews (1983), Cinque (1999:26ff.), Delfitto
(2000:44ff.) for discussion.
102
The analysis presented in this section is compatible with the assumption that temporal ad-
verbs are referential expressions that fix the reference of their anaphoric associate, namely
tense (cf. Enç 1987; Partee 1973, 1984; Larson 1985; Zeller (1994).
103
But see Haegeman (2002) for sentence medial placement of complex temporal expressions
in journalistic prose.
104
See also Zagona (1988:65f.), Zeller (1994:228ff.), Alexiadou (1997:102ff.), Haumann
(1997:126ff.) and Cinque (1999:87f.) for discussion.
105
Cf. Larson (1985) for arguments that temporal bare-NP adverbs, e.g. yesterday, tomorrow,
today, that day, Sunday, now, then, etc., are lexically marked as [+temporal] (see also Zagona
1988:32, 65ff.).
106
Tomorrow and yesterday also allow for quantifiers ((i) from BNC):
(i) It was as if she stood in a noisy limbo; all the yesterdays had gone as if they had never
been and all the tomorrows were no more than a tantalising promise.
107
Under his analysis, bare-NP adverbs are genuine arguments of the lexical verb which are
freely assigned an adverbial Θ-role in accordance with the Adverbial Θ-Role Assignment Prin-
ciple in (i) (taken from Larson 1985:606):
(i) Assign an adverbial Θ-role to α, where α is any phrase.
(i) is taken to apply optionally under the proviso that the inherent semantics of the assigner and
the assignee match, e.g. a temporal adverbial Θ-role may only be assigned to a constituent α if
α itself is temporal.
108
To make the analysis in (244) work under the Case-checking approach, we have to assume
that PP is dominated by an agreement-related functional projection, the specifier position of
which is targeted by the nominal expression (cf. below).
109
See also Jackendoff (1977:73), Emonds (1985:61, 227ff.; 1987a), Alexiadou (1997: 191).
Cf. Ernst (2002:334) for a weight-theoretic approach to the sentence-final placement of tempo-
ral DPs.
322 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
110
Notice that the phonetic form /wns/ is ambiguous between the temporal adverb and the
frequency adverb, which can be conjoined with e.g. twice (cf. section 4.3.3; see also McCawley
1988b:586, note 7):
(i) We went to see him once (or twice).
Notice also that sentence-initial once has a concessive variant:
(ii) Once she had managed to enter that cupboard successfully without any fear or panic,
she was to telephone me for another appointment. (BNC)
111
Now and then are purely temporal if they function as internal arguments of temporal
prepositions:
(i) Did I forget things before now? (BNC)
(ii) Our golf had been perfectly OK until now, we all agreed. (BNC)
(iii) No it was hard work after then. (BNC)
(iv) I was stuck in hospital for three weeks and since then have been shuffling around
behind that thing. (BNC)
112
Afterwards does not have a transitive counterpart.
113
On the whole, strings as in (260) are relatively rare, except for the type in (i), where never
before occurs in specFocP (cf. section 4.3.2):
(i) Never before had he gone so far and labelled himself an alcoholic. (BNC)
114
Ernst (2002) points out that temporal expressions within the inflectional layer "are some-
times hard to find in English due to weight constraints" (Ernst (2002:504, note 40)). However,
this argument is somewhat beside the point with respect to intransitive prepositions and genu-
ine temporal adverbs, e.g. once, currently.
115
The qualification 'adverbial' is of prime relevance since the projections of both spatial and
temporal prepositions heading predicative spatial and temporal expressions is structurally dif-
ferent from those of adverbially used prepositional projections (cf. chapter 3.2.4, note 95). In
the latter case, the lexical layer contains an outer shell, pP, the specifier position of which hosts
the preposition's external argument. Thus, the structure of the lexical layer projected by after in
(i) is (ii), the overall structure for (i) is given in (iii):
(i) The party was two weeks after Christmas.
(ii) [pP [DP the party] [p' [p afterk] [AgrOP [DP Christmas]i [AgrO' [AgrO tk] [PP [P tk] [DP ti ]]]]]]
(iii) [AgrSP [DP The party]o [AgrS [AgrS was] … [DegP [QP two weeks] [Deg' [Deg after k] [pP [DP to] [p'
[p tk] [AgrOP [DP Christmas] i [AgrO' [AgrO tk] [PP [P tk] [DP ti ]]]]]]]]]]
But see Haumann (1997:197ff.; 1999:84ff.; 2002:35ff.) for the assumption that temporal
prepositions heading the adverbial expressions in (266) take the matrix clause as their external
argument which, however, has to vacate its base position (specPP) and raise to the specifier
position of the functional projection ∆P dominating DegP. The overt raising of the external
argument to spec∆P was argued to be triggered by the requirement that sentence type informa-
tion, which is relevant for both interpretation and selection, be available. In this respect, ∆P is
similar to Rizzi's (1997:281) ForceP within the extended verbal projection.
116
But see Koopman (1993:13f., 42f.) for an analysis of intransitive prepositions as transitive
prepositions taking a [DP pro]-internal argument. Under this analysis, at least before in (266a)
would project an AgrOP which then would be targeted by overt raising of both the preposition
and [DP pro]. This analysis is attractive in many respects, especially since it would allow for the
syntactic representation of the implicit entity relative to which the event is located in time. That
is to say that pro could be construed as relating to "some entity salient in discourse" (Rizzi
1986:513; see Steinitz 1969:153 for an early proposal along these lines). However, it is not
entirely clear how pro could be licensed since licensing is seen as being contingent on the
recovery of the Φ-features) of pro (cf. Rizzi 1986:518). In this context, Adger & Ramchand's
(2001:3f.) discussion of the morphology and distribution of prepositions in Scottish Gaelic is
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 323
illuminating. They show that only prepositions which inflect for person and gender may license
pro (i); uninflected prepositions (ii) must take an overt nominal internal argument:
(i) leatha pro
with-3FS pro 'with her'
(ii) fo bh` ord
under table-DAT 'under a table'
A further shortcoming of the analysis of English intransitive prepositions as transitive preposi-
tion with a [DP pro] internal argument lies in the fact that not all prepositions that take nominal
internal arguments may also occur intransitively – seemingly or factually (see also Haumann
1997:194; 2002:32):
(iii) *We couldn't get hold of him until pro.
(iv) * He snored during pro.
117
Recall that the raison d'être for DegP derives from the Principle of Full Interpretation, i.e.
Deg is a quantificational head that must bind a variable and the preposition's referential argu-
ment must be bound.
118
In previous work, under a non-shell analysis of PP, I have argued that ago is associated
with strong features which must be checked against those of their nominal internal argument.
Thus P-raising to AgrO and raising of the nominal internal argument to specAgrOP come as a
natural consequence (cf. Haumann 1997:206; 2002:29).
119
As we have seen in section 4.3.2.2, the projections of adjectives that qualify as internal
arguments of ago also contain a QP.
120
See Corver (1997:137ff.) for an analysis of sequences of the type "measure phrase + abso-
lute adjective" along these lines.
121
See Haumann (1997:218ff.) for the nonavailability of modifiers with since.
122
This proposal is reminiscent of Rauh's (1996:192; 2002b:19f.) conjecture that so-called
grammatical prepositions, e.g. in or on in expressions like in progress and on duty, are lexically
complex prepositions containing an incorporated N.
123
Under Alexiadou's (1997:102ff.) analysis, temporal adverbs are licensed in specTP (cf.
Haumann 1997:126ff. for discussion).
124
See also Frey & Pittner (1999), Frey (2000) and Haider (2000) for arguments that temporal
prepositional projections are merged into structure as complements of semantically empty
verbal heads that project below the base position of the lexical verb. But see Alexiadou
(1997:102ff.) for an LF raising analysis of temporal expressions which she takes to originate in
the complement position of V. See also McConnell-Ginet (1982), Larson (1985, 1987, 1988,
1990), Enç (1987, 1991), Kratzer (1988), Rizzi (1990)) and Ernst (2002:338f., passim) for the
assumption that temporal adverb(ial)s may be adjoined anywhere above VP, this includes
right-adjunction to TP for sentence-final occurrences.
125
See also Geis (1987:179ff.), Alexiadou (1997:125, note 24), Cinque (2004:700ff.).
126
Sentences that do not contain temporal expressions are interpreted relative to some con-
textually specified reference point.
127
Cf. Kuno (1975), Geis (1987), Cinque (1990:90), Delfitto (2000:44) and Poletto (2002:
222ff.).
128
Given our analysis in the preceding sections, the interpretation of temporal expressions in
postverbal position cannot be any different from the interpretation they receive when occurring
within the inflectional layer (if not prohibited on the basis of non-head-finalness).
129
See also Cinque (1999:13, passim), Delfitto (2000:36f.) and Ernst (2002:337f.).
130
Under the first reading, on Friday … at noon is a semantically nested construction.
324 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
131
See also Cinque (1990:93) and Frey & Pittner (1999:26) for discussion .
Notice that R-expressions contained in the temporal expressions under consideration cannot be
coreferential with the pronoun, if the temporal expressions originated in postverbal position:
(i) * Shei took it easy on Rosai 's birthday.
(ii) * Hei did the dishes before Johni left.
132
But see Bianchi (2000) for an analysis of 'time adverbials', i.e. temporal PPs, under which
sentence-final and sentence-initial occurrences of the elements under consideration are seen as
derivationally related. Central to her analysis is the assumption that temporal prepositions
occupy a head position within the left periphery with the 'internal argument' of the preposition
being realized as the specifier of the projection governed by the preposition, e.g. FocP or TopP.
The matrix, so to speak, is represented as FinP. The linear order 'matrix > subordinate
clause/temporal PP' is conceived of as involving FinP raising across the functional projection
headed by the temporal preposition. Her analysis bears similarities to both Cinque's (2001a)
antisymmetric account of prepositional projections (cf. (292)) and Haumann's (1997, 1999,
2002) analysis of adverbial subordination.
133
See also McCawley (1988b:587) for discussion.
134
The fact that the strings in (308a, b) are ill-formed also precludes an analysis in terms of
SceneP coordination cum deletion in the first conjunct, as schematically indicated in (i) for
(308a):
(i) * [:P [SceneP Under the trees [AgrSP we played 'shop']] [:' [: and] [SceneP in summer [AgrSP we
played 'shop']]]]
Notice that an analysis along these lines is also irreconcilable with the assumption that under
the trees in summer is a constituent.
135
See, for example, Sweet (1891:121ff.), Quirk & Greenbaum (161986:211), König
(1991a:33f.; 1991b), Jacobs (1983:1; 1984) and Bayer (1999:60). Other focusing particles (e.g.
also, just, too) display more limited combinatorial properties.
136
Recall from section 4.3.1.2 that ForceFinP, as used in (310a), is the projection of Force and
Finiteness- related features in a non-split CP.
137
Cf. König (1991a:10), Jacobs (1983:8ff.; 1984:115ff.), Bayer (1996:15f.; 1999:56).
138
See also Jackendoff (1972:352ff.), Jacobs (1983:17ff.), Bayer (1996:51ff.; 1999:56ff.),
McCawley (1996).
139
See also Rochemont (1986:129), Rochemont & Culicover (1990:19ff.), Iatridou & Embick
(1994:11) and Bayer (1996:14) for related arguments.
140
The contextually relevant sets of alternatives are the same as in (313).
141
Notice that (316a) is special insofar as the moved constituent [only at parties] is in a
structural focus position, i.e specFocP. See Rochemont (1986:127ff.), Rochemont & Culicover
(1990:17ff.), Rooth (1999:240f.) for a distinction of intonational and constructional focus.
142
Cf. Klima (1964:311ff.) for arguments that only is an affective operator.
143
Notice incidentally that adjunction analysis of focusing particles cannot be considered an
option since they combine freely with arguments (cf. (319a) – (319f). See, among others,
Chomsky (1986b:6; 1995:331), Williams (1987a:437), Grimshaw (1994:22ff.) for arguments
against adjunction to argumental XPs.
144
See also König (1991a:15), who maintains that "[the] category 'focus particle' is thus no
more than a convenient abbreviation for a number of syntactic and semantic properties."
145
Semantic differences, such as the presence vs. absence of an affective feature, not only
have an impact on the interpretation of XP but also on its distribution ((i) & (ii) adapted from
Bayer (1996:14)):
ADVERBS WITHIN THE INFLECTIONAL LAYER, MOSTLY 325
157
(362a, b) are ambiguous between the sentence-negating and the constituent-negating
reading of not.
158
The co-occurrence possibilities for aspectual adverbs and focusing particles are restricted
by the nonavailability of a set of alternatives, e.g. *even/only again, *even/only no longer,
*even/only soon, *even/only typically. In some cases, single-membered sets of alternatives can
be created by constituent-negating not: not even soon, not even typically.
159
Notice that, whereas (369b) is potentially ambiguous between sentential and constituent
negation, (i) only has the constituent negation reading ((i) from www):
(i) [The] audience has not long before heard Kent repeatedly label Oswald "knave."
160
Maybe there is a ban on the 'overpopulation of the inflectional layer'.
161
With respect to (384), it should be noted that Ernst (2002) does not distinguish between
subject-attitude adverbs (384a) and subject-oriented adverbs (384b).
162
Ernst (2002:369) also presents (i), which my informants reject outright:
(i) George deliberately had at that time gone back to an abstract style of painting.
Ernst (2002) points out that temporal expressions within the inflectional layer "are sometimes
hard to find in English due to weight constraints" (Ernst (2002:504, note 40)), which, as we
have seen in section 4.3.4, is not true.
163
We have also seen that temporal expressions may co-occur when they are in an inclusion
relation.
CHAPTER 5
ADVERBS WITHIN THE COMPLEMENTIZER
LAYER, MOSTLY
(2) a. [ForceFinP [DP what]i [ForceFin' [ForceFin do] [AgrSP I like ti ]]]
b. … [ForceFinP [DP what]i [ForceFin' [ForceFin ] [AgrSP I like ti ]]]
c. … [ForceFinP [OP ] [ForceFin' [ForceFin if ] [AgrSP I like ants]]]
← [WH/Q] [+FINITE] →
(3) [ForceP [Force' [Force] [TopP* [Top' [Top] [FocP [Foc' [Foc] [TopP* [Top' [Top] [FinP]]]]]]]]]
(adapted from Rizzi 1997:297)
and FocP.4 This division of labor between ForceP and FinP is most evident in
subordinate contexts, as in (4b) and (5b):
(4) ForceP
2
Force'
3
Force TopP
tu
DP Top'
2
Top Fin
2
Fin'
rh
Fin AgrSP
6
a. [DP spiders]i 0 I don't like ti
b. said that [DP spiders]i 0 she didn't like ti
← [DECL] [+FINITE] →
(5) ForceP
2
Force'
ti
Force FocP
to
DP Foc'
tu
Foc Fin
2
Fin'
2
Fin AgrSP
# 6
a. [DP only Tom]i wouldk tk we tk invite ti
b. said that [DP only Tom]i wouldk tk we tk invite ti
← [DECL] [+FINITE] →
In a non-split Force-Finiteness System, as in (1b) and (6b), that and its null
counterpart are specified as both [+DECL, +FIN], whereas, in a split Force-
Finiteness System, [+DECL] is spelt-out as that and [+FIN] is not lexicalized.
Despite the configurational similarities that exist between the two, topic-
comment structures as in (4) and focus-presupposition structures as in (5) are
fundamentally different in nature.6 As shown in (8), topic-comment structures
differ from focus-presupposition structures in that they allow for the extraction
of a DP carrying the feature [TOPIC] across a bound pronoun. The extraction of
a DP carrying the feature [FOCUS] across a bound pronoun gives rise to weak
crossover effects:7
Bare quantifiers, such as nessuno 'no one', resist topicalization, but allow for
focalization (examples from Rizzi 1997:289):
Topic is recursive, focus is not (11); topics are compatible with a wh-operator,
foci are not (12):
(11) a. She said that, Jo, for no obvious reason at all, had turned into a frog.
b. * [ONLY THIS BOOK] [TO TOM] I gave (not to Tim).
Rizzi (1997) maintains that the differences sketched in (9) – (10) "can be traced
back to one basic distinction: focus is quantificational, topic is not" (cf. Rizzi
1997:291).
That is to say that, in focus-presupposition structures, the trace left behind
by the focalized constituent is a variable that needs to be bound by a quantifi-
cational operator, i.e. the focalized constituent, whereas, in topic-comment
structures, the trace left behind by the topicalized constituent does not have the
status of a variable, but that of a so-called null constant which must be bound
by an empty anaphoric operator (OP) which, in turn, takes the topicalized con-
stituent as its antecedent, as schematically indicated in (13a) and (13b), respec-
tively (cf. Rizzi 1997: 291ff.):
The differences in (8) derive from the fact that only quantificational elements,
such as focus and wh-operators, give rise to weak crossover effects. The differ-
ences in (9) also follow from the (non)quantificational status of constituents
under consideration: since focalized elements are quantificational, they must
bind a variable. However, since a resumptive pronoun does not qualify as a
variable, (9b) gives rise to a violation of the Principle of Full Interpretation in
terms of vacuous quantification. The unacceptability of (10b) derives from the
same scenario: (bare) quantifiers, by definition, are quantificational expressions
that must bind a variable which is present in focus-presupposition structures,
but not in topic-comment structures.
Before turning to discuss the positions targeted under adverb raising, I
would like to briefly comment on Rizzi's (1999, 2002) proposal to have inter-
rogative operators targeting either specForceP or specFocP. So far, we have
tacitly assumed that ForceP either hosts the declarative complementizer that or
interrogative elements such as the interrogative complementizer if or raised
interrogative operators, e.g. what. Reasonable as this assumption is with
respect to sentence type information and selectability, there is evidence that
interrogative operators are not necessarily overtly realized in specForceP, but
in specFocP (cf. Rizzi 1999, 2002). First, interrogative operators and foci com-
pete for the same structural position, whence they cannot co-occur (irrespective
of linear order):
(16) [ForceP [Force [Q]] [TopP [DP regrets] [Top' [Top ] [FocP [DP how many[Q]] [Foc'
[Foc do [Q]] … you think Sinatra had?]]]]]
Adverb raising to specFocP differs from adverb fronting to the left periphery,
as in (18), in very much the same way as argument raising to specFocP and
specTopP differ (cf. above). Moreover, as can be seen from the contrasts
between (19) and (20), "the ban on adverb crossing is selective" (cf. Rizzi
2002:11), i.e. adverb raising to specFocP may proceed across a higher adverb,
whereas adverb fronting may not:
Whereas, in his (1997) paper, Rizzi maintained that adverb preposing as in (18)
targets specTopP – an idea that has been widely adopted in the relevant litera-
ture – he now convincingly argues against analyzing adverb preposing as topi-
calization (cf. Rizzi 2002:14ff.). Despite the fact that the preposed adverbs in
(18) share with genuine topics the property of being discourse-prominent, they
cannot felicitously be argued to assume specTopP. The reason for discarding
the specTopP analysis of adverbs is twofold.10 First, adverbs do not make good
topics on interpretive grounds given that topics represent background informa-
tion, i.e. information readily available from the discourse context:
(24) a. The violini they saw [him ease ti out of the case].
b. The violini they said [that he eased ti out of the case].
(25) a. * Carefullyi, they saw [him ease the violin out of the case ti].
b. * Carefullyi they said [that he eased the violin out of the case ti].
On the basis of these contrasts, Rizzi (2002:16) argues that preposed adverbs in
the complementizer layer occupy the specifier position of a designated func-
tional projection below the functional projection hosting interrogative opera-
tors, i.e specFocP. Since modification is "the substantive relation between an
adverb and the structure it relates to", Rizzi (2002:18) labels this functional
projection Mod(ifier)P which, like TopP, is recursive.12 As the label ModP is
334 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
already reserved for the functional projection hosting modal verbs (cf. section
4.1) and since the functional motivation for projections in the complementizer
layer that host 'preposed' adverbs is "that they make the moved adverb promi-
nent" (cf. Rizzi 2002:19), I choose to use the label Prom(inence)P:
(26) [ForceP [Force' [Force ] [TopP* [Top' [Top ]…[FocP [Foc' [Foc ] [PromP* [Prom' [Prom ] [TopP*
[Top' [Top ] [FinP ]]]]]]]]]]] (adapted from Rizzi 2002:18)
Notice that PromP, as opposed to TopP, invariably sits below FocP whence
preposed adverbs, but not genuine topics, fail to precede interrogative operators
(compare (23) vs. (15) & (22)).
Moreover, since raising to specPromP, as opposed to raising to specTopP,
does not involve the presence of an empty anaphoric operator (cf. Rizzi
1997:313ff.), the fact that elements in specPromP, but not those in specTopP,
ameliorate That-Trace Effects (cf. (27a) vs. (27b)), which basically arise if the
embedded subject is raised across the overt complementizer that, receives a
straightforward explanation (see also chapter 4.3.1.2 for discussion):
(27) a. Whoi do you think [ForceP [Force' [Force that] [PromP* [Adv secretlyk] [Prom'
[Prom ] [TopP* [Top ] [FinP [Fin 0] [AgrSP ti tk ate cake?]]]]]]]
b. * Whoi do you think [ForceP [Force' [Force that] [TopP*[DP cakek ][Top' [Top ]
[FinP Opi [Fin' [Fin 0] [AgrSP ti ate tk ?]]]]]]]
The licensing of the subject trace fails in (27b) because the null anaphoric
operator in specFinP instantiates long-distance agreement between the [+AGR]-
feature in Fin and the raised subject so that the [+AGR] feature does not qualify
as a proper governor for the trace in specAgrSP. Raising to specPromP does not
involve the presence of an anaphoric operator in specFinP, whence the [+AGR]-
feature in Fin can properly govern the subject trace.
5.1.1 Summary
This section reviewed Rizzi's (1997, 2002) proposal according to which the left
periphery of the clause, i.e. the Force-Finiteness System which represents the
interface between the content of the proposition and the "articulation of
discourse" (cf. Rizzi 1997:283), is subject to splitting if the Topic-Focus Sys-
tem is activated, i.e. if the complementizer layer is targeted by raising from
within the inflectional and/or lexical layer. If no raising takes place, the sole
functional projection constituting the complementizer layer is ForceFinP in
which case both the Force- and the Finiteness-related features are realized in
one single head position. If constituents are focalized, topicalized or made
prominent otherwise, the Force-Finiteness System splits in order to accommo-
date the raised elements. In this case, the monolithic ForceFinP splits into two
distinct functional projections, ForceP and FinP, marking the upper and the
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 335
lower boundary of the complementizer layer and thus looking outside and
inside the clause. Depending on the nature of the peripheral feature borne by
the raised constituent, i.e [FOCUS], [TOPIC] or [PROM], they target specFocP,
specTopP or specPromP. From the set of elements under consideration, only
focalized constituents are quantificational, i.e. the trace left behind counts as a
variable which, in order not to violate FI, must be bound. Constituents raised to
specTopP are construed with both their trace and an empty anaphoric operator,
while constituents raised to specPromP are construed with only their trace.
(31) a. Clearly, there are further topics [on his agenda]. (BNC)
b. Obviously there was a future in this business, … (BNC)
(35) a. Fortunately things definitely take a turn for the better... (www)
b. * Definitely, things fortunately take a turn for the better...
c. * Fortunately definitely things take a turn for the better...
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 337
(36) a. He certainly would probably like to do so, and his ambitions are
obvious… (www)
b. * He probably would certainly like to do so…
c. * He certainly probably would like to do so …
(37) a. Fortunately, most of America wisely does not accept that the
inconsistent et al are guilty of what happened in New York and DC.
b. * Wisely, most of America fortunately does not accept that…
c. The scanner probably only seldom drew something close to the
500mA rating of its regular plugpack. (www)
d. * The scanner only seldom probably drew close to …
e. Evidently you have not figured how to use the "link" button yet.
f. * You have not evidently figured how to use the "link" button…
g. Luckily they had long ago mastered spells of protection against
normal weapons. (www)
h. * They had long ago luckily mastered spells of protection…
i. It was evidently hastily formed – an afterthought. (www)
j. * It was hastily evidently formed – an afterthought.
The fact that illocutionary adverbs fail to occur within the scope of relative
operators (40), conditional operators (41) and interrogative operators (in both
root and embedded contexts (42)) allows us to posit that illocutionary adverbs
are parasitic on Force-related features:20
(40) a. * Demelza, whose mother frankly is her only relative, wanted a gran so
much she advertised for one in a Jobcentre. (BNC)
b. * On the way down I fell over a man hiding in a dark corner, who
roughly ran away immediately. (BNC)
(41) a. * If she knew a thing ought to be done she honestly would really push
hard. (BNC)
b. * Had she not been so downhearted briefly Ruth would have enjoyed
herself. (BNC)
(42) a. * Did you honestly [speaking] expect me to leave a job half done?
b. * Why did Mrs Jenks roughly insist on giving them to him for
breakfast?
c. * I wonder if frankly that would have helped. (BNC)
d. * I asked whether briefly he thought he had the same difficulties as his
father. (BNC-data)
The latter observation that illocutionary adverbs fail to occur within the scope
of interrogative operators allows us to posit that illocutionary adverbs are para-
340 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
According to van Gelderen (2001:6), the fact that illocutionary adverbs and the
complementizer that are mutually exclusive (44) provides supportive evidence
for locating illocutionary adverbs in ForceP. She takes the ban on (44b, d) to
derive from Koopman's (2000:337ff.) Generalized Doubly Filled Comp Filter,
which states that the specifier position and the head position of a functional
projection cannot be simultaneously lexicalized:23
(47) a. Honestly, HIS PASSWORD (*honestly) you should have asked for,
not his PIN-code.
b. Briefly, ONLY WITH USE OF INTRAVENOUS CONSCIOUS SEDATION,
(*briefly) access into the pelvic collection was attained... (www)
(48) a. Seriously, last month (*seriously) she said she was doing ok.
b. Briefly, nowadays (*briefly) I am working as a Priest… (www)
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 341
Illocutionary adverbs, as shown in (49) – (51), may assume positions other than
the sentence-initial position, i.e. specForce(Fin)P. Ignoring for the moment the
fact that the distributional range of honestly-type adverbs is wider than that of
briefly-type adverbs, illocutionary adverbs may occur in post-subject position,
after the finite non-lexical verb and in postverbal position ((49a) from BNC):25
As can be seen in (52) and (53), illocutionary adverbs are barred from follow-
ing sentential negation as well as nonfinite non-lexical verbs (if read as an
aspectual and a subject-attitude adverb, respectively, briefly and seriously are
acceptable; see also Ernst 2002:99):
(53) a. * They could have briefly been (briefly) entertaining this stupid idea.
b. * They will have seriously been (seriously) claiming that …
Let us begin with the easy part. Illocutionary adverbs in postverbal position as
in (51) are realized within the by-now familiar empty structure, where they
are licensed by forming a representational chain with the expletive assertion
operator in specForce(Fin)P:26
Given that illocutionary adverbs have the status of assertive operators which
may be spelt out in specForce(Fin)P or in , the unacceptability of (49b) and
(50b) should not come as a surprise. But the well-formedness of (49a) and
(50a) does since the illocutionary adverb, quite unexpectedly, follows the
subject and precedes and follows the finite non-lexical verb in AgrSP (49a) and
(50a). While (49a) can be reconciled with the basic assumption that the illocu-
342 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
tionary adverb is realized within the complementizer layer, albeit not where it
should be, (50a) is a clear case of realization within the inflectional layer.
In this context, the contrast between the grammatical strings in (49a) and
(50a) and the ungrammatical strings in (49b) and (50b) is quite revealing since
only honestly-type illocutionary adverbs have subject-related counterparts, i.e.
subject-oriented adverbs that precede the finite non-lexical verb in AgrSP (55)
and subject-attitude adverbs that follow the finite non-lexical verb in AgrSP
(56).27 That is to say that, in (55), the agentive subject is considered frank,
honest or serious with respect to what she or he is doing. In (56), the adverbs
under consideration express the agentive subject's attitudes towards what she or
he is doing:
(58) a. I am frankly surprised that this home truth hasn't hit home before.
b. He had honestly believed he could take the place …
c. [For sixteen years], I have seriously used only one machine.
(BNC-data)
Notice that Force(Fin)P must split for the complementizer layer to be able to
accommodate both the adverb in specSoP in (59a, b) and the subject in spec-
TopP (59a):
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 343
(59) ForceP
2
Op Force'
1
Force TopP
2
D Top'
2
Top SoP
2
Adv So'
2
So AgrSP
2
D AgrS'
2
AgrS AttP*
2
Adv …
a. Op[ASS] Shei frankly[ASS] ti enjoyed this
b. Op[ASS] I amk frankly[ASS] tk surprised
(60) a. * (Frankly) she frankly wisely would enjoy the extra publicity.
b. * (Honestly) she honestly carefully had transcribed all the tapes.
c. * (Seriously) he seriously cleverly has withheld all the evidence.
(61) a. * He had frankly been foolishly assuming he could get away with it.
b. * He had honestly been cleverly trying to fool us.
c. * I have seriously been carefully considering this proposal.
(62) a. Frankly, they (frankly) have never carefully explained things to us.
b. Honestly, we (honestly) have carelessly been assuming he was right.
c. Seriously, you (seriously) shouldn't have foolishly tried to fix it with
cello tape.
344 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(65) a. [It] is frankly better to find the ways and means of presenting yourself
somehow, than not perform at all. (BNC)
b. It would honestly seem a bit bizarre to me if it did. (www)
c. It has seriously become the highlight of my week. (www)
The fact that honestly-type illocutionary adverbs, but not purely subject-related
adverbs (e.g. wisely, cleverly, carelessly, etc.) may co-occur with non-agent
subjects can be related to the feature [ASS] which accounts for the licensing of
honestly-type adverbs.30 By (almost) the same line of reasoning, the fact that
roughly-type illocutionary adverbs fail to assume non-peripheral positions can
be accounted for: they are reliant on agentivity, i.e. a speaker, but they lack
subject-related counterparts.
The detail that honestly-type illocutionary adverbs fail to occur in any posi-
tion below NegP or ReltP (cf. (52) and (53)) only partially derives from their
status as assertive operators which must take scope over the entire proposition.
Given that the adverbs under consideration can take propositional scope
even if they are not realized in, but licensed from within specForce(Fin)P (cf.
(54) & (59)), why should they fail to take propositional scope if they occur in
positions lower than NegP or ReltP? In other words, why is it not possible for
the adverb to hook up to the assertion operator in specForce(Fin)P, as in (67)?
The answer to this question is simple: the illocutionary adverb would be within
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 345
the scope of the negative operator in specNegP or/and within the scope of the
reference time operator on specReltP:
(67) ForceP
rh
Op Force'
th
Force …
g
AgrSP
th
… AgrS'
fu
AgrS NegP
fu
Op[NEG] Neg'
1
Neg …
h
ReltP
2
Op[TEMP] Relt'
2
Relt AttP*
ti
Adv …
a. Op[ASS] … won'tk Op[NEG] tk Op[TEMP] seriously[ASS] claim
b. Op[ASS] … willk tk Op[TEMP] have seriously[ASS] been
* claiming
non-lexical verb (71) and the sentence-final position (72), but are barred from
occurring within the scope of interrogative operators (73), counterfactual
operators (74) and sentential negation (75):33
(70) a. Tom and I would remain friends and he oddly enough would be
living only three hours south of me... (www)
b. The manual hopefully stops the nasty letters ... (CC)
(72) a. I don't think you're gonna get any snow, luckily. (BNC)
b. It was Dr Entwistle who had prodded her more firmly in that
direction, oddly enough. (BNC)
(74) a. * Had he had more self-esteem, he luckily could have taken a joke.
b. * If you had been smart, you ideally would have eaten only half of it.
The reason for evaluative adverbs being barred from the scope of interrogative,
counterfactual and negative operators, as in (73) – (75), derives from their
status as what we may wish to call factive operators ([ FACT]) in the sense that
the constituent they take scope over represents a fact, or, in Fregean (1918)
terms, a proposition which is true: 34, 35
(76) The fact that he cannot take a joke is {unfortunate, surprising, amazing,
odd, etc.}
(77) a. Oddly, in 1998, (*oddly) Byrd cited them as a reason for Clinton to
action against Hussein. (www)
b. Luckily, in May…, (*luckily) I was invited to the NIH. (www)
(80) a. Police said the Toyota rammed three cars [...], but that fortunately no
one had been injured. (BNC)
b. I believe of course that ideally chimps should live freely … (BNC)
(81) a. Resolutely, … she opened her new copy of the Church Times which
normally she would have looked forward to reading on her journey
home. (BNC)
b. It's a good thing because da Gama again made the mistake of giving
cheap gifts to the local sheik, who fortunately wasn't offended. (www)
(83) a. Fortunately, most of America wisely does not accept that [they] are
guilty of what happened in New York and DC … (www)
b. Luckily, I wisely had my wife drive the 4Runner… (www)
(84) a. Frankly, not surprisingly, they did not even utter a single word of
condemnation. (www)
b. Briefly, I would ideally like all my bets to be singles... (www)
Given that evaluative adverbs must take scope over true propositions, i.e. facts,
given that the feature [FIN] is vital for the fixing of truth values and given that
evaluative adverbs follow that, raised relative pronouns and topicalized con-
stituents (cf. (80)), FinP – the lowest functional projection within the comple-
mentizer layer – suggests itself as the domain over which evaluative adverbs
take scope.40 Notice that, since topicalization may proceed across evaluative
adverbs (cf. (82) & (85)), in which case specFinP hosts an empty operator
forming a chain with the topicalized constituent, specFinP cannot be taken to
accommodate evaluative adverbs, as shown in (86):41, 42
(85) a. Tom and I would remain friends and he oddly enough would be
living only three hours south of me now… (www)
b. The grammar, luckily, is not all that difficult … (www)
c. And the scramble, incredibly, has not ended. (www)
(86) TopP
3
D(P) Top'
3
Top FinP
ri
*Adv Fin'
ok
Op tu
Fin AgrSP
6
temptationi *fortunately/Opi she could resist ti
the scramblei *incredibly/Opi has not ended ti
(87) [EvalP [Adv Fortunately] [Eval' [Eval ] … [TopP [QP most of Americai] [Top' [Top ]
[SoP [Adv wisely] [So' [So ] [FinP Opi [Fin' [Fin ] [AgrSP ti does not]]]]]]]]]
Low placement of evaluative adverbs is independently ruled out by the fact that
evaluative adverbs may not occur within the scope of scene-setting expressions
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 349
and focalized constituents (cf. (77), (78)). Thus, the functional projection host-
ing evaluative adverbs must sit higher than SceneP and FocP.
Since evaluative adverbs may follow the declarative complementizer that
(80) as well as relative operators in specForceP (81), the domain in which
EvalP is located can be narrowed down to the stretch between ForceP and
SceneP (assuming that SceneP dominates FocP).43 The positioning of evalua-
tive adverbs above SceneP and FocP, as shown in (88), straightforwardly
accounts for the strings under consideration (cf. (77), (78), (80) & (81)), but
leaves us with the question of how strings are derived in which the evaluative
adverb follows the topicalized subject, as in (85):
(88) ForceP
2
D Force'
1
Force …
h
EvalP
2
Adv Eval'
2
Eval SceneP
2
DegP Scene'
rh
Scene FocP
2
PP Foc'
2
Foc ...
… that ideally [FinP chimps should…]
… which normally [FinP she would …]
oddly in 1998 [FinP Byrd cited them]
luckily even with [FinP the mystique…]
changes
Given that the linear order 'subject > evaluative adverb' is only possible in the
absence of FocP, and given that TopP may be introduced recursively, we
assume that EvalP may be dominated by TopP the specifier position of which
may host the topicalized subject, as shown in (89):44
(89) a. [TopP [D he] [Top' [Top ] [EvalP [Adv oddly enough] [Eval' [Eval ] [FinP would
be living … south of me]]]]]
350 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
b. [TopP [DP The grammar] [Top' [Top ] [EvalP [Adv luckily] [Eval' [Eval ] [FinP is
not all that difficult]]]]]
(90) a. The suspect, unfortunately (for us), cleverly has avoided incriminating
herself.
b. Elsa, strangely (for security), foolishly did not control the monk's
suitcase.
c. Furman, fortunately for us, carefully (had) jotted down all that
occurred to his observation in the elementary condition and progress
of his native city. (www)
In these cases, since one and the same constituent can be topicalized only once,
EvalP and SoP in (90) cannot be separated by an intervening TopP, i.e. raising
of the topicalized subject must proceed in one swoop across both types of
adverbs, as shown in (91):
(91) [TopP [DP The suspecti] [Top' [Top ] [EvalP [Adv unfortunately (for us)] [Eval'
[Eval ] [SoP [Adv cleverly] [So' [So ] [FinP ... [AgrSP ti has avoided incriminating
herself]]]]]]]]
Under the analysis presented, the presence of evaluative adverbs warrants the
complementizer layer to split. As already mentioned, factivity feeds on
finiteness and is only compatible with declarative force: only constructs bear-
ing the features [FIN] and [DECL] can have the status of true propositions. As
factive operators, evaluative adverbs are barred from occurring within the
scope of interrogative, counterfactual and negative operators, but not from
occurring within the scope of declarative and assertive operators (that, illocu-
tionary adverbs). Trivially, the positioning of EvalP immediately below ForceP
(and TopP) accounts for the fact that evaluative adverbs take scope over FinP,
whose head bears the feature [FIN] .
In the remainder of this section, I would like to concentrate on the distribu-
tion and licensing of these elements in non-left peripheral positions, i.e. the
position following the finite non-lexical verb, as in (92), and the sentence-final
position, as in (93). Underlying the analysis of (92) and (93) is the assumption
that spec EvalP (alternatively to hosting an overt factive operator, i.e. an
evaluative adverb) may host an expletive factive operator with which the actual
evaluative adverb forms a representational chain:
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 351
(93) a. We're above all that kind of petty vanity, unfortunately. (BNC)
b. They seemed to find the place as fascinating as we did, luckily. (BNC)
c. It was Dr Entwistle who had prodded her more firmly in that
direction, oddly enough. (BNC)
d. Within a minute or so you have one fully rejuvenated disk drive,
hopefully. (BNC)
(94) [EvalP [Op[FACT] [Eval' ... [FinP ... [V' [V ] [ [ e ] [Adv hopefully[FACT]]]]]]]]
Given that factivity ultimately feeds on finiteness, I assume that the evaluative
adverbs that occur within the inflectional layer, as in (92), are realized in
specTP, where they take scope over the finiteness feature in T. However, in
order to be fully licensed, evaluative adverbs must be in construction with an
expletive factive operator [ FACT] in specEvalP, as schematically illustrated in
(95):45
(95) a. [EvalP Op[FACT] [Eval' [Eval ] ... [FinP ... [AgrSP [DP she] [AgrS' [AgrS hasi] [TP
[Adv unfortunately[FACT]][T' given credence …]]]]]]]
b. [EvalP Op[FACT] [Eval' [Eval ] ... [FinP ... [AgrSP [DP they] [AgrS' [AgrS mighti] [TP
[Adv ideally[FACT]] [T' like to live in Knightsbridge]]]]]]]
(96) a. [Luckily [for all of us]], it's perfectly within the law. (BNC)
b. [Strangely [for one who has been so successful]], Charlie has never
been inundated by offers from English trainers. (BNC)
352 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(97) a. 'TWAS built on marshy land, fortunately for us,' Simon told her.
b. You came to in the middle of it, unfortunately for you…
(98) a. [TopP The book [EvalP [surprisingly [for potential readers]] was not on
sale]].
b. He bought a new lawnmower [ForceP which [EvalP [fortunately [for his
neighbors]] he was not able to operate]].
c. She said [ForceP that [EvalP [fortunately [for him]] the liquid was not
poisonous]].
(99) a. * It's [TP [luckily [for all of us]] [T' perfectly within the law]].
b. * It [AgrSP/TP [luckily [for all of us]] [AgrS'/T' is within the law]].
c. * She had [TP [unfortunately [for the patient]] [T' forgotten the
anaesthesic]].
b. * She [AgrSP/TP [hopefully [for the next patient]] [AgrS'/T' will
remember]].
(102) a. You can definitely improve your chances of finding a job. (CC)
b. He'd clearly had his hair brushed properly by [him]. (CC)
Like illocutionary and evaluative adverbs, evidential adverbs must not follow
scene-setting expressions (104), focalized constituents (105) and non-finite
non-lexical verbs (106):49
(104) a. Apparently, last year, (*apparently) they put off taking photos until
the end ... (www)
b. Clearly, in 1999, (*clearly) the Web was a US-centric information
space. (www)
(108) a. Check out PrcView which apparently has a logging feature. (www)
b. They gave us a story which obviously wasn't correct. (www)
They may also co-occur with subject-oriented adverbs (110) and they may
follow both illocutionary and evaluative adverbs (111):
354 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
On the basis of the data considered so far, we can posit that the functional pro-
jection hosting evidential adverbs, EvidP, is sandwiched between EvalP and
the set of SceneP and FocP (if projected), as shown in (116) on page 355 for
(111b), (108b), (107a), (111d), (104b), (100a) and (105a).
Exploiting the multiple availability of TopP, the strings in (101a), (109a)
and (110a) are accounted for by the following structure:
(112) a. [TopP [DP she] [Top' [Top ] [EvidP [Adv evidently] [Evid' [Evid ] [TopP [Top' [Top ]
[PromP [Adv ] [Prom' [Prom ] [FinP had no concept of books]]]]]]]]]
b. [TopP [DP spiders] [Top' [Top ] [EvidP [Adv admittedly] [Evid' [Evid ] [TopP [Top'
[Top ] [PromP [Adv ] [Prom' [Prom ] [FinP I'm afraid of (them)]]]]]]]]]
c. [EvidP [Adv apparently] [Evid' [Evid ] [TopP [D you] [Top' [Top ] [PromP [Adv
wisely] [Prom' [Prom ] [FinP have your browser set]]]]]]]
(115) a. Had he taken her to hospital in time, she definitely wouldn't have died.
b. If she had read the text carefully, she certainly would have spotted
all the typos.
(116) ForceP
3
Adv/D Force'
3
Force …
g
EvalP
2
Adv Eval'
2
Eval ...
g
EvidP
2
Adv Evid'
2
Evid SceneP
2
DegP Scene'
2
Scene FocP
2
QP …
a. frankly obviously it is …
b. which obviously affects …
c. that apparently there…
d. luckily for me apparently everyone…
e. clearly in 1999 the Web…
f. evidently she had …
g. evidently so hard was it to…
forms" (cf. Ernst 2002:104, 479, note 18), as in (117), and thus may occur
within the scope of interrogative, counterfactual and negative operators:
Under the assumption that evidential adverbs are to be analyzed as some kind
of factive operator, they must be kitted out with an additional feature that dis-
tinguishes them from evaluative adverbs and allows them (a) to co-occur with
evaluative adverbs and (b) to be outscoped by interrogative, counterfactual and
negative operators. In the following, I will take this difference to be encoded
by [FACT] for evaluative adverbs vs. [@FACT] for evidential adverbs. Since
evidential adverbs, like evaluative adverbs, require that the proposition they
take scope over be true and since the fixing of truth values is parasitic on the
presence of the feature [FIN], I suggest that evaluative adverbs that follow the
finite non-lexical verb, are realized in specTP (a position which is also avail-
able to evaluative adverbs), as shown in (120) for (102) and in (121) for (113)
and (114). Under both scenarios, the adverb is subject to indirect licensing, i.e.
via chain formation with an expletive [@FACT]-operator in specEvidP:
(120) a. [EvidP Op[@FACT] [Evid' [Evid ] … [FinP ... [AgrSP [D you] [AgrS' [AgrS cani]
[TP [Adv definitely[@FACT] [T' [T ti ] … [improve]]]]]]]]
b. [EvidP Op[@FACT] [Evid' [Evid ] … [FinP ... [AgrSP [D He] [AgrS' [AgrS 'di]
[TP [Adv clearly[@FACT] [T' [T ti ] … [had… ]]]]]]]]
(121) a. [ForceP Op [Force' [Force hasi ] … [EvidP Op[@FACT] [Evid' … [FinP … [AgrSP
[D she] [AgrS' [AgrS ti ] [TP [Adv obviously[@FACT] ] [T ti ] [VP cleaned the
table ?]]]]]]]]]
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 357
b. [ForceP Whatj [Force' [Force didi ] … [EvidP Op[@FACT] [Evid' … [FinP … [AgrSP
[D he] [AgrS' [AgrS ti ] [TP [Adv definitely[@FACT] ] [T ti ] [VP do right
tj? ]]]]]]]]]
c. [EvidP Op[@FACT] [Evid' … [FinP … [AgrSP [D she] [AgrS' [AgrS hasn'ti ] [NegP Op
[Neg' [Neg ti][TP [Adv obviously [@FACT] ] [T ti ] [VP cleaned the table ]]]]]]]]]
d. [EvidP Op[@FACT] [Evid' … [FinP … [AgrSP [D they] [AgrS' [AgrS didi ] [NegP not
[Neg' [Neg ti][TP [Adv obviously [@FACT] ] [T ti ] [VP avoid questions ]]]]]]]]]
(123) a. Probably, some band members will contribute more than others.
→ It is probably true that some … will contribute …
b. Possibly she had pulled her stitches, or had developed an infection…
→ It is possibly true that she had pulled her stitches…
c. Maybe we even passed each other in the supermarket.
→ It may be (the case) that we even passed each other… (BNC-data)
(125) a. After that she reckoned that [if she hadn't moved on to somewhere
bigger, she'd probably be stuck here forever reading out the latest
sheep prices at six o'clock every morning].
358 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(126) a. [I] think you wouldn't probably have gone along with them…
b. In Saint-Moritz we were able to do things we couldn't possibly
arrange during the season.
c. I won't maybe have as much physical contact. (www)
(127) a. Were he alive today he would undoubtedly have bemoaned the fact
that he was the frequent target of what he would have probably called
'scribblers'.
b. They also inexplicably turned down a deal with another cable network
that could have possibly provided live, or much closer to live,
coverage.
c. So perhaps a lot of the times but I would have maybe suggested that,
that we could have done and presented that as part of the package.
Epistemic adverbs follow the declarative complementizer that (131) (pace van
Gelderen 2001:8), relative operators in specForceP (132) and topicalized con-
stituents (133) ((131) & (132) from www):
(131) a. Minich said that probably many of the codes people are concerned
about will be rejected by the committee.
b. [He] said that possibly the webmaster could provide return email
addresses.
c. My gay friend said that maybe God made him gay to keep the
population in control.
(132) a. Their terminals in the dorsal nucleus come into relation with
neurons whose axons probably descend into the spinal cord…
b. Is this the same lady whose dinner possibly made you ill?
c. Light against of Urobos, the snake, whose bite maybe illustrates the
self-afflicted pain of becoming whole again.
As we have already seen in the introductory section, epistemic adverbs may co-
occur with illocutionary, evaluative and evidential adverbs, provided they
follow them and provided they are not adjacent to them:
(135) a. Probably next summer we'll come and play Phoenix… (www)
b. Possibly next summer, we will come back.
c. Maybe next summer we'll bring his mom […].(www)
However, a closer look at (135) reveals that the epistemic adverb takes scope
only over the scene-setting expression, not the entire proposition, as can be
read off the following paraphrases:
(138) a. Next week, probably the overriding factor will be the currency. (www)
b. And next week, possibly we could hear from the FAA…(www)
c. Next week, maybe we can start on the base. (www)
In the following, I take the fact that epistemic adverbs may occur freely within
the scope of interrogative operators and counterfactual operators (cf. (124) &
(125)) in conjunction with the fact that they fail to precede any lexical material
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 361
(142) a. [AgrSP [DP he] [AgrS' [AgrS] [EpiP [Adv probably] [Epi' [Epi ] [TP drank the
clubhouse dry]]]]]
b. [AgrSP [D she] [AgrS' [AgrS would] [EpiP [Adv possibly] [Epi' [Epi ] [TP enjoy ...
gardening]]]]]
Under this analysis, both sentence-initial occurrences and the linear order
'subject > epistemic adverb > finite non-lexical verb' would be derivative, with
the epistemic adverb raising to specPromP and the subject either staying put
(143a) or raising across the adverb in specPromP to specTopP (143b):
(143) a. [PromP [Adv Possiblyi] [Prom' [Prom ] [FinP … [AgrSP [D she] [AgrS' [AgrS had]
[EpiP [Adv ti ] [Epi' pulled her stitches]]]]]]]]]
b. [TopP [D hej ] [Top' [Top ] [PromP [Adv probablyi] [Prom' [Prom ] [FinP … [AgrSP
[D tj ] [AgrS' [AgrS would] [EpiP [Adv ti ] [Epi' not like … splashes]]]]]]]]]
Under the assumption that epistemic adverbs originate within the inflectional
layer, and thus below subject-oriented adverbs, the fact that probably and
possibly fail to precede subject-oriented adverbs (140) follows from Minimali-
ty, i.e. in order to precede subject-oriented adverbs, epistemic adverbs such as
probably and possibly would have to raise across the subject-oriented adverb
which, then, would qualify as a closer (potential) antecedent governor for the
trace left behind by the epistemic adverb in specEpiP than the epistemic adverb
itself (cf. below for maybe in (140c)). An additional problem lies in the target
for the epistemic adverb, ?P (cf. (146) on page 362). The position to which the
epistemic adverb would have to raise cannot be specPromP since PromP sits
below SoP. ?P in (146) cannot be identified as FocP since the epistemic adverb
is not focused. 57
Speakers, who accept epistemic adverbs within the scope of sentential nega-
tion, clearly prefer contracted negation over not (presumably because the
sequence 'not > epistemic adverb' is potentially ambiguous between constituent
negation and sentential negation):
(144) a. [You wouldn't probably have gone along with them. (BNC)
b. In Saint-Moritz we were able to do things we couldn't possibly
arrange during the season. (BNC)
362 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(145) a. [AgrSP [D you] [AgrS' [AgrS wouldn'ti ] [EpiP [Adv probably] [Epi' [Epi ti ] [NegP
Op[NEG] [Neg' [Neg ti ] [TP have gone along with them...]]]]]]]
b. [AgrSP [D we] [AgrS' [AgrS couldn'ti ] [EpiP [Adv possibly] [Epi' [Epi ti ] [NegP
Op[NEG] [Neg' [Neg ti ] [TP arrange during the season]]]]]]]
(146) ?P
tu
Adv ?'
th
? TopP
2
D Top'
1
Top SoP
2
Adv …
h
AgrSP
2
D AgrS'
2
AgrS EpiP
2
Adv Epi'
6
a. * Probablyi hej clumsily tj had ti tried to open the box.
b. * Possiblyi shej foolishly tj had ti pulled her stitches.
As for the somewhat marginal status of (130a, b), at this point, I cannot but
conjecture that epistemicity, i.e. the speaker's judgment on the truth of the
proposition, comes in too late in the discourse for the hearer to be integrated in
the processing of the sentence, i.e. the status of sentence-final epistemic ad-
verbs is that of an afterthought.
So far, maybe has been exempted from our discussion. The reason for this
lies in the empirical fact that maybe, as opposed to probably and possibly, pref-
erably occurs in sentence-initial position and not in its base position within the
inflectional layer.58 The answer to the question of why maybe is repelled from
its base position and raised across the subject lies in its more overt 'verbal'
character: as a secondary adverb formed by univerbation of the modal verb
may and the bare infinitive be, maybe is too similar to strings of the type
"modal be", as can also be read off the obvious misspellings in (148), where
maybe is conceived of as a verbal head (i.e. the product of univerbation):
(148) a. They maybe not exactly what you have to offer, … (www)
b. The much publicised event has sparked speculation that he maybe
under pressure from his captors… (www)
c. [It] probably maybe still there, I couldn't say. (BNC)
d. She maybe in a desperate hurry! (BNC)
(149) a. [A] survey would most probably reveal the English text as the most
commonly possessed book in most schools. (BNC)
b. But Gould had quite possibly thanked Darwin for the wrong
present. (BNC)
c. * And if they were in a hurry they would {most, quite, very} maybe
paint them all with what they called paint oil… (BNC)
(150) a. And some years you'd probably maybe get a few pounds back and
sometimes you would ... maybe have to pay … (BNC)
364 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
b. And I think that's an area that we often overlook, and it's probably
maybe not wise to overlook it, … (www)
c. [This] is something you should possibly maybe check… (www)
d. Yeah, I didn't realize that I should possibly maybe press tofu before
cooking… (www)
These findings allow us to posit that maybe originates as the head of EpiP:60
(151) [AgrSP [D She] [AgrS' [AgrS maybei ] [EpiP [Epi' [Epi ti ] … [in a hurry ]]]]]
This analysis also accounts for the fact that (140c), where maybe precedes the
subject-oriented adverb, is marginally acceptable (cf. (140c) vs. (140a, b) &
(146)): since maybe is a head (and thus cannot raise but to a head position) and
since subject-oriented adverbs are not, the latter cannot interfere with the trace
left behind by the former and thus there is no way in which Minimality could
be violated. However, as before, the question of the target remains open:
(152) ?P
3
Adv ?'
2
? TopP
fu
D Top'
2
Top SoP
2
Adv So'
2
So …
g
AgrSP
2
D AgrS'
rh
AgrS EpiP
2
Adv Epi'
2
Epi …
#
a. Maybei theyj wisely tj were ti watching…
b. * Probablyi hej clumsily tj had ti tried to …
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 365
Under the assumption that the accommodation of lexical material within the
empty verbal structure projected below the lexical verb must respect
Uniformity (cf. Chomsky 1995:253), the analysis of maybe as a head would
also account for the highly marginal status of (130c). That is to say that maybe
as a head is barred from assuming a specifier position within . Inserting
maybe into head position within would corrupt the idea of being
semantically empty-headed. Probably and possibly are not affected by these
considerations: structurally, though probably not pragmatically, they make
perfect -inmates.61
(153) a. * He had frankly been foolishly assuming that he would make it.
b. She will hopefully have cleverly refrained from smoking.
c. John would obviously have carefully planned his next move.
d. She would possibly have carefully avoided this topic.
e. * He had foolishly been frankly assuming that he would make it.
f. * She will cleverly have hopefully refrained from smoking.
g. * John would carefully have obviously planned his next move.
h. * She will carefully have possibly avoided this topic.
(154) EvalP
2
Op[FACT] …
g
EvidP
2
Op[@FACT] …
h
AgrS'
2
AgrS EpiP
ti
Adv …
g
TP
3
Adv …
h
ReltP
2
Op[TMP] Relt'
2
Relt AttP
1
Adv …
5
a. Op[FACT] …will hopefully[FACT] Op[TMP] have cleverly
b. Op[@FACT] …would obviously[@FACT] Op[TMP] have carefully
c. …would possibly Op[TMP] have carefully
(155) a. Hopefully she will cleverly refrain from smoking in her office.
b. Obviously John will carefully plan his next move.
c. {Possibly, maybe} she has carefully avoided this topic.
d. Frankly he (frankly) had been foolishly assuming this.
(156) a. Ms. San had seriously once thought of accepting her for a god-
daughter as she had no children of her own. (www)
b. The stoves were cheap imitations of Preways and have fortunately
long since disappeared, … (www)
c. Merhi is also as terrible as ever, but he's clearly long ago given up
any idea that he could be anything but. (www)
d. If not for religion's historical, zealous denial of science, research,
and progressive study, society would possibly long ago have found
cures…(www)
e. * Ms. San had once seriously thought of accepting her ...
f. * The stoves … have long since fortunately disappeared…
g. * [He]'s long ago clearly given up any idea ...
h. * [Society] would long ago possibly have found cures …
(157) a. More than a decade after the Web came into the world, online
publishing is frankly still too challenging… (www)
b. Only Harry Moore who will unfortunately soon be leaving,
questioned the wisdom of this action. (www)
c. Once this step has been made we can evidently no longer treat
pronouns as coreferential with their antecedents. (www)
d. [They] have probably long abandoned it. (www)
e. * [Online] publishing is still frankly too challenging ...
f. * [These] features had only recently fortunately become familiar.
g. * [We] can no longer evidently treat pronouns as coreferential…
h. * [They] have long probably abandoned it.
Notice that, in the cases under consideration, the ban on the adjacency of sen-
tence adverbs (cf. Jackendoff 1972:87ff.) is lifted. Given the data in (156) and
(157), it would seem that the adjacent placement of the adverbs under
consideration is warranted by formal distinctness, with only the speaker-
oriented adverbs being of the –ly-type.63 However, formal distinctness cannot
be the key since, in (158), both the speaker-oriented adverbs and the aspectual
adverbs are of the –ly-type:
(158) a. For me, Knezevic is merely a metaphor for strange behavior, which is
obviously quickly forgotten here… (www)
b. Good will, which is necessary for that has unfortunately recently been
absent from the Muslim-Bosniac side. (www)
c. This attitude has unfortunately frequently been accepted. (www)
d. The varnish has obviously recently been attended to. (www)
Rather, it seems that the ban on the adjacent placement of formally (near)-
identical elements may be overridden in cases in which the elements clearly
368 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(159) a. I think the Engineering faculty has probably recently developed some
form of airship. (www)
→ … it is probable that the faculty has recently developed…
→ … it was probably recently that the faculty has developed…
b. The fact that such a small dog has probably frequently displayed
aggression […] has been overlooked… (www)
→ It is probable that the dog has frequently displayed…
→ It was probably frequently that this dog displayed…
c. However, anyone who's seen it has probably quickly come to the
conclusion that a lot of it is, well, pretty mediocre. (www)
→ It is probable that somebody has quickly come to this…
→ It was probably quickly that somebody has come to this…
Abstracting away from adjacency, the patterns in (156) – (158) are expected
under our analysis. Since speaker-oriented adverbs within the inflectional layer
are restricted to occurring within the domain delimited by AgrSP and TP they
automatically precede temporal expressions in specReltP. Since the system
provides multiple merge sites for the functional projections hosting aspectual
adverbs and since speaker-oriented adverbs must take scope over aspectual
adverbs, the aspectual adverbs in (157) must be merged below TP, as illus-
trated in (161) on page 369.
Indirect specifier-head agreement between illocutionary, evaluative and
evidential adverbs that occur within the inflectional layer and their respective
expletive operators within the complementizer layer allows for a uniform
licensing account of the elements under consideration. In conjunction with the
analysis of temporal expressions as being licensed in (or from within) spec-
ReltP and the analysis of the functional projections hosting subject-attitude ad-
verbs (AttP) and aspectual adverbs (AspP) as being mergible in different posi-
tions within the inflectional layer, the present analysis allows for an imple-
mentation of the empirical fact that speaker oriented adverbs precede temporal,
aspectual and subject-attitude adverbs irrespective of whether the former are
realized within the complementizer layer or the inflectional layer:
(160) illocutionary > evaluative > evidential > … > epistemic > temporal >
aspectual > subject-attitude > …
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 369
(161) ForceP
2
OP[ASS] …
g
EvalP
2
Op[FACT] …
g
EvidP
3
Op[@FACT] …
g
AgrS'
6
AgrS AttP
2
Adv …
g
EpiP
3
Adv …
g
TP
3
Adv …
h
ReltP
2
Op[TMP] …
g
AspP
1
Adv
a. Op[ASS] had seriously[ASS] once
b. Op[FACT] have fortunately[FACT] long since
c. Op[@FACT] 's clearly[@FACT] long ago
d. would possibly long ago
e. Op[ASS] is frankly[ASS] still
f. Op[FACT] had fortunately[FACT] recently
g. Op[@FACT] can evidently[@FACT] no longer
h. have probably long
Seemingly reversed orders as in (162) and (163) do not invalidate the analysis:
370 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(163) a. This year they are clearly avoiding the same mistake. (www)
b. Because I am really into both activism and school and am so psyched
to have these two worlds collide, because usually they are
unfortunately quite far apart. (www)
In (165) – (167), the higher adverb occurs within the complementizer layer of
the clause and the lower adverb either within the complementizer layer ((165c,
f) & (166a, b, c)) or within the inflectional layer ((165a, b, d, e), (166d) &
(167)):
(165) a. Well, honestly, yes they are, unfortunately most of their hits never
make it to our dear PCs. (www)
b. Briefly, I would ideally like all my bets to be singles… (www)
c. Frankly, I certainly don't know. (www)
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 371
(167) a. Well, I'm not surprised; she certainly has probably attracted a ton of
other college guys. (www)
b. It most definitely could possibly do that. (www)
(168) a. * Honestly, unfortunately most of their hits never make it to our dear
PCs.
b. * Frankly, certainly I don't know.
c. * Seriously, probably you will think so when you read this…
(170) Certainly she probably has attracted a ton of other college guys. (www)
(171) a. The print, unfortunately, could definitely use a full restoration. (www)
b. Your spider unfortunately has probably passed away with the
coming of winter… (www)
to complying to the Horror Aequi Principle (cf. section 5.2.1.4), lies in the
realization of the higher adverb in sentence-final position, i.e. within VP.
Notice incidentally that, as a consequence of the sentence-final placement
certainly and unfortunately in (172), the linear orders 'evaluative > evidential'
and 'evidential > epistemic' are reversed. However, despite occurring in sen-
tence-final position, unfortunately and certainly take scope over certainly and
probably. This is a corollary of the expletive factive operators [ FACT] and
[@FACT], which license the sentence-final evidently and certainly, taking scope
over certainly, i.e. [@FACT] and probably, i.e. [EPI], as shown in (173):
(173) a. [EvalP Op[FACT] [Eval' [Eval ] … [EvidP certainly[@FACT] [Evid' [Evid ] … [EpiP
OP[EPI] [Epi' [Epi ] … [V' [V ] [ [ ] [Adv unfortunately[FACT]]] ... ]] ... ]] ... ]]
b. [EvidP OP[@FACT] [Evid' [Evid ] … [EpiP probably[EPI] [Epi' [Epi ] … [V' [V ] [
[ ] [Adv certainly[@FACT]]] ... ]] ... ]] ... ]]
(174) a. Well, honestly speaking, maybe you and Colin just shouldn't be
friends,… (www)
b. Frankly speaking, maybe I will have to make the right choice very
carefully. (www)
(175) a. Basically we all know that Ares wants Xena to go back to being a
ruthless warlord, which unfortunately for him, probably isn't going to
happen. (www)
b. Receiving the torch from Anna and continuing correspondence with
our TASIS Hellenic, TASIS Cyprus, TASIS France and summer
program alumni is Frances Lovett, who is Scottish but who has lived
"down south" for over 20 years and who, fortunately for us, definitely
plans to stay! (www)
c. Luckily for me, maybe it was after work hours on this beginning of the
weekend. (www)
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 373
(176) a. * Seriously, last month {luckily, evidently} he had lost his wallet.
b. * Luckily, last week, evidently he had not yelled at his boss.
The only viable cases are those in which a scene-setting expression or a focal-
ized constituent is followed by an epistemic adverb:
(178) {Seriously, hopefully, evidently}, next week, he possibly will tell us more.
(179) {Frankly, unfortunately, apparently}, NOT EVEN BEER would they proba-
bly sell.
With the restriction that in the context of illocutionary adverbs the preceding
topic must be an agentive subject, members of all classes of speaker adverbs
may precede and follow topicalized constituents. That is to say that speaker-
oriented adverbs may co-occur within the complementizer layer if separated by
topicalized constituents:
(180) a. Honestly, this man, they {hopefully, evidently, probably} will never
invite him again.
b. Unfortunately, her husband, she {obviously, possibly} has shot him.
c. Certainly, the polka-dotted tie, maybe you should not wear it tonight.
(181) a. Honestly, this man, hopefully, they {clearly, probably} will not invite
him again.
374 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(182) [XP Adv1 [X' … [{SceneP, FocP, TopP} Adv2 [{Scene', Foc', Top'} … [TopP Adv3 [Top' …
[FinP will [[t1] [[ t2] [[t3 ]]]]]]]]]
However, what is conjoined in these cases is not two adverbs that belong to
two different classes, e.g. illocutionary and evaluative (183a), evaluative and
evidential (183b), and evidential and epistemic in (183c), but larger chunks of
structure,68 arguably two ForcePs, with any phonetic material below the adverb
being deleted under identity in the first conjunct:69
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 375
(184) :P
qp
ForceP :'
2 fu
Adv Force' : ForceP
1 fu
Force … Adv Force'
g th
EvalP Force …
2 g
Adv Eval' EvalP
1 rh
Eval … Adv Eval'
g th
EvidP Eval …
3 #
Adv Evid' EvidP
2 eh
Evid … Adv Evid'
# th
FinP Evid …
# g
AgrSP FinP
g g
… AgrSP
# g
TP …
1 #
Adv … EpiP
eh
Adv Epi'
2
Epi …
frankly and ideally, [AgrSP he should …
luckily and obviously, [AgrSP I have …
Op[@FACT] definitely and possibly [TP vote …
Evidence for an analysis along these lines derives from the fact that reverse
orders of speaker-oriented adverbs are possible:
(185) a. Fortunately and honestly, this album does get better as it progresses.
b. Quite clearly, and fortunately, it has not.
376 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
terms of chain formation, i.e. the specEpiP hosts an expletive operator which
binds the epistemic adverb in VP. The fact that maybe is even less acceptable
in sentence-final position than e.g. probably was argued to derive from the
requirement that be empty-headed. In section 5.2.1.5, I readdressed the
placement of speaker-oriented adverbs within the inflectional layer with the
focus being on their co-occurring with subject-attitude adverbs, temporal and
aspectual adverbs. I showed that both the various co-occurrence options and
the strict linear orders can be accounted for under the proposed analysis.
Speaker-oriented adverbs occupy a restricted number of positions within the
inflectional layer (either by being parasitic on both an expletive operator within
the complementizer layer and functional architecture within the inflectional
layer, i.e. specTP and specAttP, or by being merged within the inflectional
layer, i.e. specEpiP, all of which take into their scope the functional projections
hosting subject-attitude adverbs, temporal and aspectual adverbs.72 Seemingly
reversed orders were shown to be contingent on the realization of speaker-
oriented adverbs in VP or the realization of temporal and aspectual expressions
within the complementizer layer. In both cases, the scope properties of the
elements involved were shown to be respected. The final section was devoted
to the co-occurrence of speaker-oriented adverbs within the complementizer
layer with emphasis being placed on the avoidance of adjacency. Basically,
speaker-oriented adverbs, provided they respect the underlying relative order,
may occur within the complementizer layer whenever they can be separated by
intervening lexical material, e.g. topicalized constituents, scene-setting expres-
sions and focalized constituents. The latter two options are only available in
cases in which the second adverb is an epistemic adverb since members of all
other classes are barred from following scene-setting expressions and focalized
constituents within the complementizer layer.73 Depending on the type of the
higher of two adverbs, the following strategies of adjacency avoidance are
available: illocutionary adverbs occurring with speaking may be followed by
an evaluative, evidential or epistemic adverb, and evaluative adverbial expres-
sions that are not head final may be followed by an evidential or epistemic
adverb. Coordinating conjunctions were shown not to be eligible 'separators'
since the conjuncts are not adverbs, but clausal projections.
In fact, the elements under consideration can leave vP only if they are focused
or questioned:75, 76
(187) a. SO LOUDLYi did they cry ti, that they woke Ulysses. (www)
b. SO RELUCTANTLYi were the corpses examined ti by the pathologist that
her boss fired her.
c. ONLY HORRIBLYi did Madonna perform 'American Pie' ti.
d. NOT EVEN SURGICALLYi can we remove your nose ti.
e. SO COMPLETELYi was the maximalist agenda defeated ti in Geneva that
one might almost assume its proponents would now give up. (www)
(188) a. How loudlyi did Shakespeare want his actors to talk ti, and how broad
did their gestures need to be? (www)
b. How reluctantlyi were the corpses examined ti …?
c. How horriblyi did Madonna perform 'American Pie' ti?
d. How completelyi (= to which extent) did you understand the text ti?
Raised vP-adverbs occur in specFocP (recall from section 5.1 that interrogative
and focused constituents compete for the same position):
(189) a. [FocP [Adv So loudlyi] [Foc' [Foc didk] … [FinP … [AgrSP [D they ] [AgrS'
[AgrS tk ] … [vP cry ti that they woke Ulysses]]]]]]
b. [FocP [Adv How horriblyi] [Foc' [Foc didk] … [FinP … [AgrSP [D Madonna ]
[AgrS' [AgrS tk ] … [vP perform 'American Pie' ti ? ]]]]]]
As opposed to all other vP-adverbs, pure domain adverbs freely occur within
the complementizer layer, i.e. they need not be focused and they cannot be
questioned or accompanied by so. If they are focused (FOCUSED), they raise to
specFocP (190a, b). With respect to non-focused pure domain adverbs (190c,
d) the question arises whether they have raised to a position within the
complementizer layer, say specSceneP, or whether they are merged into that
position (cf. below):77
380 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(190) a. FINANCIALLY they have the ability to raise the money, but they are by
no means out of the woods. (www)
b. PSYCHOLOGICALLY they have hurt us badly, but we have no option but
to […] overcome that in these two matches. (www)
c. Psychologically (speaking), they have a positive effect on our
emotions… (www)
d. Medically (speaking) it's impossible for someone who has lost their
sight due to glaucoma to get it back. (BNC)
(191) a. [FocP [Adv PSYCHOLOGICALLYi ] [Foc' … [AgrSP [D they ] [AgrS' have hurt us
ti badly ]]]]
b. [SceneP [Adv Medicallyi] [Scene' … [AgrSP [D it ] [AgrS' 's ti impossible ]]]]
(192) a. SO LOUDLYi did they often cry ti, that they woke Ulysses.
b. How reluctantlyi are corpses currently examined ti by the pathologist?
c. PHYSICALLYi they are often ti limited to the amount of information that
will fit onto a computer screen. (www)
Since the adverbs under consideration (arguably with the exception of pure
domain adverbs, but see below) do not prepose freely, I chose aspectual
adverbs to illustrate Relativized Minimality effects. As can be seen from the
contrasts between (193) and (194), aspectual adverbs which are merged below
epistemic and temporal adverbs are barred from raising across the latter if they
target specPromP (193), but not if they target specFocP (194):
Ernst (2000:40) presents the example in (195) to illustrate that pure domain
adverbs (as opposed to e.g. manner adverbs) freely prepose across other
adverbs (e.g. evidential adverbs):78
(195) Physically and psychologically, Jordan definitely can really work you
over.
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 381
I am not convinced this is the case. First, non-focused pure domain adverbs in
sentence-initial position function as scene-setters, i.e. we are talking about the
physical and psychological domain. Second, sentence-initial pure domain
adverbs, as opposed to e.g. sentence-initial aspectual adverbs (197), may pre-
cede interrogative operators in specFocP:79
(196) a. Physically, did you feel you might be ready for back to back 1:16s?
b. Emotionally, do you think it's possible to help someone who is
reluctant to help themselves?
c. Mentally do you find it more difficult getting into an event like this?
d. Financially, how much can your organization expect to gain and how
will the savings accrue?
e. Botanically, what is the pod?
f. Physically, what is happening in each case? (www-data)
(198) a. In the last year, have you used force to get money or expensive things
from another person?
b. Tomorrow, Sunday, would you care to come to Paris in the afternoon
or evening (not in the morning)?
c. When you were in Tiddim in 1999, did you hear about the USDA?
d. Today, what did you do?
e. In 1999, what did you think the blogosphere would look like in 2004?
f. During summer, where will you be living? (www-data)
An easy way out would be to allow for SceneP to be merged in two positions,
i.e. either directly above µP within the lexical layer (cf. chapter 3.2.3) or
directly above FocP within the complementizer layer. This move, however,
would obscure two facts: pure domain adverbs do not have the status of sen-
tence adverbs (cf. chapter 3.2.3) and they do not have a scene-setting function
when realized in the lexical, i.e. botanically and legally in (201a) and (201b) do
not set the scene for botany or laws:
Also, we would have to explain why only ScenePs whose specifier position
hosts a pure domain adverb may be subject to multiple merge, i.e. ScenePs
hosting temporal expressions may assume only the highest of the three posi-
tions:
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 383
(202) a. Yesterday, illicit smokers were fined for taking a puff … (BNC)
b. * Illicit smokers were yesterday fined for taking a puff.
c. * Illicit smokers were fined for taking yesterday a puff.
In chapters 4.3.2 and 4.3.3, we have seen that aspectual adverbs may assume
two positions within the complementizer layer: specFocP and specPromP, with
specFocP being accessible to focused and inherently negative adverbs ((207) &
(208)) and specPromP being accessible to aspectual adverbs lacking the periph-
eral features [FOC] and [NEG] (cf. (209)):82
(208) a. SELDOM has a bone been hyped as much as this one. (BNC)
b. NEVER had so many children developed weak bladders in so short a
period. (BNC)
(210) a. [FocP [Adv ONLY RECENTLYi ] [Foc' [Foc hadk ] [FinP ... [AgrSP [D it ] [AgrS'
[AgrS tk ] ... [AspP [Adv ti] [Asp' become clear to Microsoft that it had
become an industry leader ]]]]]]]
b. [FocP [Adv SELDOMi ] [Foc' [Foc hask ] ... [FinP ... [AgrSP [DP a bone ] [AgrS'
[AgrS tk ] ... [AspP [Adv ti] [Asp' been hyped as much … ]]]]]]]
c. [PromP [Adv Ofteni ] [Prom' [Prom ] ... [FinP ... [AgrSP [D our intuition] [AgrS'
[AgrS tk ] ... [AspP [Adv ti] [Asp' warns us … ]]]]]]]
At first glance, the data in (212) seem to invalidate the claim that the underly-
ing order of aspectual adverbs may only be reversed if the lower adverb is
quantificational in the relevant sense (i.e. susceptible to targeting specFocP),
since soon clearly lacks quantificational features:
(212) a. You wash your new sweats. They shrink. You wash them again. They
shrink more. Pretty soon, they no longer fit. (www)
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 385
However, the fact that soon is time related allows us to treat soon on a par with
the sentence-initial temporal expressions in (213) as a scene-setting adverb,
which, in turn, allows us to maintain that soon is merged in specSceneP, i.e. the
sentence-initial occurrence of soon in (212) is not derivative: 85
(213) a. Yesterday, illicit smokers were fined for taking a puff… (BNC)
b. Right now we have arithmetic to do. (BNC)
c. During intervals, coffee is available.
Supportive evidence for this assumption derives from strings as in (214) with
two occurrences of soon. The first soon sets the scene (i.e. 'this is about the
near future'), whereas the second soon relates to the time that elapses between
reference time and event time:
(214) a. Soon she would soon be in a deep trance, listening only to his
commands. (www)
b. Ismail also hoped that soon peace would soon be realized in both the
Sudan and Somalia… (www)
c. Ron was not willing to keep up the pace, soon they would soon hire a
new basses from outside San Fran... (www)
Thus, sentence-initial soon in (212) and (214) is analyzed on a par with genu-
ine left-peripheral temporal expressions which, as we have seen in chapter
4.3.4.3, are merged in specSceneP (no dating function!):
(215) a. [SceneP [DegP Yesterday,] [Scene' [Scene ] … [AgrSP smokers were fined]]].
b. [SceneP [DegP Pretty soon,] [Scene' [Scene ] … [AgrSP they no longer fit]]].
c. [SceneP [DegP Soon,] [Scene' [Scene ] … [AgrSP she would soon be in
trance]]].
Given the skeleton of the complementizer layer in (216) and given the target
options for adverbs that originate within the lexical or inflectional layer, the
partial linear orders in (217) are expected:86, 87
386 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
(216) ForceP[ASS] > EvalP[FACT] > EvidP[@FACT] > SceneP > FocP{[Q], [NEG], [FOC]} >…
SoP > PromP* > FinP
(218) a. [SoP Foolishly, she {*today, evidently, unfortunately, frankly} will not
show up].
b. [SceneP Yesterday, she {*evidently, unfortunately, frankly} had seen a
rainbow].
c. [SceneP Politically, they {*evidently, unfortunately, frankly} will benefit
from this].
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 387
(219) a. [PromP Again she {*stupidly, next time, clearly, luckily, honestly}
would settle for pasta].
b. [PromP Probably, he {*stupidly, next time, clearly, luckily, honestly}
had tried to open the box].
c. [SoP Stupidly, {*so often, yesterday, obviously, luckily, honestly} {had
she, she had} not answered the phone..]
d. [FocP So badly had she {*last time, obviously, luckily, honestly} been
singing that he switched off his hearing aid.
Deviations from the overall linear order in (216) can be observed in cases in
which a higher adverb may be realized outside the complementizer layer, i.e.
within the inflectional layer or within VP:89
(221) a. Obviously, they (obviously) had fortunately not received nasty letters.
b. Obviously, they (obviously) had not received nasty letters, fortunately.
c. Next week, she will {obviously, hopefully} not be here.
d. Next week, she will not be here, {obviously, hopefully}.
e. Cleverly, Joe (?clev.) has {evidently, fortunately} withheld this detail.
f. Cleverly, John (cleverly) has withheld this detail, {evidently,
fortunately}.
(223)
specEvalP
specSceneP
specFocP
specSoP
specTP
specPromP
VP
Notice that the strings in (224) do not represent reverse orders. In sentence-
final position, both yesterday and on Sunday have a dating function, not a
scene-setting function (cf. (224a, b)). The post-auxiliary occurrence of cleverly
in (224c) and the post-participle occurrence of carefully in (224d) are accept-
able as subject-attitude adverbs, not as subject-oriented adverbs (cf. chapter
4.3.1 for discussion):
5.2.2.1 Summary. Starting out with adverb raising from the lexical and inflec-
tional layer to the complementizer layer as being subject to Relativized Mini-
mality, this section focused on the co-occurrence and overall linear order ad-
verbs within the complementizer layer, as defined by the partial hierarchy in
(225):
(225) ForceP > EvalP > EvidP > SceneP > FocP > … SoP > … PromP* > FinP
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 389
With the specifier positions of ForceP, EvalP, EvidP, SceneP and SoP being
pure merge sites, only specFocP and specPromP qualify as targets for adverb
raising. While vP-adverbs are restricted to raising to specFocP, adverbs origi-
nating within the inflectional layer, depending on their association with periph-
eral, i.e quantificational features, may target specFocP or specPromP, with the
latter potentially being recursive. In conjunction with the stipulation that non-
focused pure domain adverbs are scene-setting expressions which are merged
in specSceneP, (225) was shown to account for all and the only admissible
linear orders of adverbs within the complementizer layer (cf. (218) – (220)).
Deviations from the overall order were shown to be possible only in cases in
which a speaker-oriented adverb is realized within the inflectional layer or
within the VP, i.e. in sentence-final position (cf. (221) & (222)).
5.3 Conclusion
This chapter pursued two interrelated goals: first, the distribution and licensing
of speaker-oriented adverbs and second, the co-occurrence of (non)raised ad-
verbs within the left periphery of the clause. Underlying my analysis was the
earlier stated assumption that the Rizzian (1997, 1999, 2002) complementizer
layer of the clause, as recapitulated in section 5.1, is subject to splitting not
only under raising, but also under merge (cf. chapter 4.3.1.2 for subject-
oriented adverbs and chapter 4.3.4.3 for scene-setting temporal expressions).
Starting with a brief review of the four distributional classes of speaker-
oriented adverbs, the co-occurrence options and the linear orders they may
occur in, I argued that illocutionary, evaluative and evidential adverbs are
merged as specifiers of designated functional projections in the higher part of
the complementizer layer (sections 5.2.1.1 – 5.2.1.3), whereas epistemic ad-
verbs are merged in the specifier position of a designated functional projection
within the inflectional layer (5.2.1.4). The allocation of the former to the com-
plementizer layer and the latter to the inflectional layer was shown to be
warranted by their distribution relative to (non)raised constituents within the
complementizer layer, i.e. lexical material in specSceneP, specFocP, specSoP
and specPromP and the absence vs. presence of Relativized Minimality effects.
The specifier position of ForceP(Fin)P was shown to be the merge site for illo-
cutionary adverbs which I suggested be analyzed as assertive operators whose
presence is contingent on Force-related features, such as [ASSERTIVE] and
[DECLARATIVE]. The analysis was shown to straightforwardly account for the
following facts: illocutionary adverbs take into their scope the entire proposi-
tion, they are barred from occurring within the scope of interrogative, counter-
factual and negative operators, and they can neither follow lexical material that
is merged within the complementizer layer (e.g. that or scene-setting expres-
sions) nor lexical material that is raised to the complementizer layer (e.g.
focalized or topicalized constituents). My analysis of both the post-subject oc-
currence and non-left peripheral occurrences of illocutionary adverbs relied on
390 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
5.2.1.3) were shown to be similar to evaluative adverbs in the sense that their
presence is contingent on the proposition being true, they differ from evalua-
tive adverbs in both interpretation (they express degrees of certitude) and dis-
tribution (they may occur within the scope of interrogative, counterfactual and
negative operators). To account for the 'factive-ish' status of evidential adverbs
I posited the feature [@FACT], which like [FACT], feeds on finiteness. The
analysis of evidential adverbs in post-subject position, in the position following
the finite non-lexical verb and in sentence-final position is identical to that of
evaluative adverbs: the linear order 'subject > evidential adverb' results from
the subject raising across the evidential adverb in specEvidP to the specifier
position of a high TopP, the linear order 'non-lexical verb[FIN] > evidential
adverb' is subject to both chain formation with an expletive operator in spec-
EvidP and the head of TP being specified as [FIN] and, finally, the sentence-
final occurrence is reliant on chain formation. In section 5.2.1.4, I argued that
epistemic adverbs, as opposed to all other subclasses of speaker-oriented
adverbs, are genuine inmates of the inflectional layer. Evidence for this analy-
sis was shown to derive from the fact that these elements systematically fail to
precede (non)raised lexical material in the complementizer layer, such as
scene-setting expressions, focalized constituents, subject-oriented adverbs and
raised adverbs. Since epistemic adverbs cannot be felicitously accommodated
within the functional projections constituting the inflectional layer and since
epistemic adverbs are tied to a unique position between AgrSP and TP (or
NegP, if projected), I posited the functional projection EpiP which hosts epis-
temic adverbs in either its specifier position (probably) or in its head position
(maybe). The status of maybe as a head was shown to follow from its overtly
deverbal nature (whence maybe may buddy as a modal and co-occur with e.g.
probably), its failure to support modifiers and the fact that maybe tends to be
'repelled' from the inflectional layer in cases in which genuine verbal heads are
present. Under my analysis of epistemic adverbs as inmates of the inflectional
layer, left-peripheral occurrences of these elements are necessarily derivative,
i.e. the linear order 'epistemic adverb > subject' involves the raising of the ad-
verb across the subject to (spec)PromP, and the linear order 'subject > epis-
temic adverb > non-lexical verb[FIN]' involves raising of the subject across the
raised adverb to specTopP. Sentence-final occurrences of epistemic adverbs
were again accounted for in terms of chain formation, i.e. the specEpiP hosts
an expletive operator which binds the epistemic adverb in VP. The differences
in phrase structure status between e.g. probably and maybe were shown to be
the key to the low acceptability of maybe in sentence-final position. Sections
5.2.1.5 – 5.2.1.6 served as touchstones for the analysis of speaker-oriented ad-
verbs presented in sections 5.2.1.1 – 5.2.1.4. In section 5.2.1.5 I concentrated
on the realization of speaker-oriented adverbs within the inflectional layer with
special emphasis being placed on their co-occurring with and their relative
ordering with respect to adverbs that are merged within the inflectional layer,
392 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
i.e. subject-attitude adverbs, temporal and aspectual adverbs. It was shown that
both the admissible co-occurrences and the strict linear orders of the elements
under consideration follow under the analysis of illocutionary, evaluative and
evidential adverbs as licensed from within the complementizer layer as well as
under the analysis of epistemic adverbs as merged within the inflectional layer.
Seemingly reversed orders of the adverb classes under consideration were
shown to be contingent on either the realization of speaker-oriented adverbs
within the non-interpretable domain of the clause in VP or the raising of tem-
poral, aspectual and pure domain expressions within the complementizer layer.
Section 5.2.1.6 focused on the co-occurrence of speaker-oriented adverbs
within the complementizer layer. It was shown that the co-occurrence of
subject-oriented adverbs within the complementizer layer is not only restricted
by the underlying hierarchy of functional projections, but also by the require-
ment that co-occurring speaker-oriented adverbs be not adjacent. Adjacency
was shown to be circumnavigable by the interspersion of merged or raised
lexical material, e.g. scene-setting expressions, topicalized or focused constitu-
ents preceding raised epistemic adverbs, the participle speaking following illo-
cutionary adverbs or the prepositional restrictors of evaluative and evidential
adverbs. Taking into account both merged and raised adverbs, section 5.2.2
addressed the co-occurrence and linear orders of adverbs within the comple-
mentizer layer on a larger scale. It was shown that the overall linear order of
adverbs within the complementizer layer is restricted by the intrinsically rigid
order of the functional projections constituting the complementizer layer (226)
in conjunction with Rizzi's (2002:10ff., 19ff.) version Relativized Minimality
accounts for the selectiveness of adverb crossing:
(226) ForceP > EvalP > EvidP > SceneP > FocP > … SoP > … PromP* > FinP
Notes
1
As Rizzi (1997:284) points out, "we should think of finiteness as the core IP-related
characteristics that the complementizer system expresses", with languages varying as to how
much inflectional information is replicated in Fin.
2
The asterisk notation on TopP indicates that TopP may be recursive (cf. below).
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 393
3
Rizzi (1997:287) proposes assuming that movement into specTopP and specFocP is Crite-
rion-driven, i.e. movement into specTopP is triggered by the Topic-Criterion and movement
into specFocP by the Focus-Criterion.
4
Compare the following (adapted from Rizzi 1997:313):
(i) I think that next year, John will win the prize.
(ii) * I think next year, that John will win the prize.
5
The null variant is optionally also specified as [+AGR] whence it qualifies as a proper
governor for subject traces (cf. section 4.3.1.2).
6
Cf. Rizzi (1997:289ff.) for a detailed discussion of the differences between focus and topic
replicated here for English.
7
Cf. Rizzi (1997:289).
8
Cf. Cinque (1990:63) and Rizzi (1997:290) for discussion.
9
According to Rizzi (1999:3), Italian allows for embedded interrogative operators to be pre-
ceded by topics.
10
But see Rizzi (2002:16f) for the conjecture that preposed adverbs "can become topics in
special contextual conditions, i.e. when they have been mentioned in the immediately previous
discourse" (cf. Rizzi (2002:16f.)).
11
Cf. Ernst (2002:411f.) and Rizzi (2002:25, note 11).
12
Rizzi (2002) maintains that Mod subsumes "all the features licensing adverbs, i.e. the whole
of Cinque's hierarchy, and the dedicated Mod positions in the left periphery which make ad-
verbs 'prominent' (cf. Rizzi 2002:19).
13
By postulating that the functional projections hosting subject-oriented adverbs, SoP (cf.
chapter 4.3.1.2), and temporal scene-setting expressions, SceneP (cf. chapter 4.3.4.3), are
genuine inmates of the complementizer layer, we have already implicitly challenged both the
assumption that adverbs in the complementizer layer are necessarily raised and that the Force-
Finiteness System only splits under raising.
14
Bellert (1977:341ff.) also discusses frequency adverbs, e.g. often in (i), domain adverbs, e.g.
linguistically in (ii), and conjunctive adverbs, e.g. hence in (iii), as instances of so-called
speaker-oriented adverbs (Bellert's 1977:341, 348f. examples):
(i) John óften comes here.
(ii) Linguistically this example is interesting, but logically it is not.
(iii) Hence this statement is a theorem.
See also Schreiber (1971:84), Cinque (1999:13, 33f.), Laenzlinger (1998:42) and Ernst
(2002:69ff.) for finer grained analyses of so-called speaker-oriented adverbs.
15
Subject-oriented adverbs, as discussed in chapter 4.3.1.2, have speaker-oriented ingredient
in that they evaluate the subject participant's behavior.
16
As we shall see in section 5.2.1.4, this restriction does not hold for epistemic adverbs.
17
Cf. Quirk & Greenbaum's (161986:242ff.) distinction between attitudinal disjuncts and style
disjuncts.
18
See, for example, Greenbaum (1969) and Schreiber (1971) for arguments that illocutionary
adverbs are remnants of deleted higher performative sentences, e.g. I tell you honestly that I
despise peas.
19
Further members of this class are: bluntly, confidently, flatly, incidentally, prosaically, sin-
cerely, truly, etc. (cf. Greenbaum 1969:91f.; Swan 1988a:67 for a wider selection of illocution-
ary adverbs).
20
See also Ernst (2002:99) for discussion. Notice that, under the non-intended subject-related
reading of honestly, (42a) is perfectly well-formed.
394 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
21
See also Ernst (2002) for arguments "that assertions involve a covert operator with (roughly)
the content "Speaker expresses [Proposition]" (Ernst 2002:71). See Travis (1988:290) for an
early proposal along these lines.
22
Under Cinque's (1999:84) analysis, according to which the functional projection that hosts
illocutionary adverbs, MoodPspeechact, ranks highest within the inflectional layer (presumably
between AgrSP and TP), both the sentence-initial and the sentence-final occurrence of illocu-
tionary adverbs would have to be accounted for in terms of syntactic movement. Cinque
(1999:84) conjectures that illocutionary adverbs, since they may precede topicalized and
focalized constituents, may raise to specForceP.
Although it clearly is possible to posit that this functional projection can be merged above and
below TP to allow for the orders 'subject > (illocutionary adverb >) modal/auxiliary[+fin] > (illo-
cutionary adverb)', the higher occurrence will always be problematic since the subject and the
finite non-lexical verb, contrary to agreement requirements, are not contained within the same
functional projection, AgrSP. Irrespective of this problem, the analysis would fail to account for
the fact that sentence-initial illocutionary adverbs, despite having reputedly raised from the
post-subject position, do not block the cliticization of phonetically reduced modal and auxiliary
verbs onto the pronominal subject (cf. chapter 4.3.1.2 for discussion):
(i) Frankly, you'd be crazy not to want to go and do that. (BNC)
(ii) Honestly she's so unreliable that girl. (BNC)
(iii) But seriously I'm against this budget because of what is, what it doesn't contain
rather than what it does contain. (BNC)
Moreover, it is not clear which factors would trigger the raising operation under consideration
since illocutionary adverbs would already be formally licensed in MoodPspeechact. With respect
to sentence-final occurrences of illocutionary adverbs Cinque (1999) would either have to
readmit right-adjunction, i.e. traditional extraposition, or allow for the entire structural com-
plement of Moodspeechact to raise across the adverbs contained in its specifier position. Alterna-
tively, to circumnavigate right-adjunction, Cinque could posit Laenzlinger-style (1996, 1998)
right-branching specifiers.
23
With respect to data as in (i) – (iii), where the matrix contains a verb of 'saying' and the
matrix and embedded subject are coreferential, i.e. the embedded subject can be construed as
the speaker, one might wish to argue that illocutionary adverb, in fact, may follow that:
(i) Responding to a question the congressman said that frankly he does not see any
possibility of discussion ... (www)
(ii) They said that, honestly, their office does not prescribe anything stronger and
that codeine plus Motrin 800 should be enough. (www)
(iii) Mr. Musallam said that briefly, there are 2 regions involved in the cleanup
process ... (www)
24
See also van Gelderen (2001). But see Cinque (1999:11ff., 53ff.) and Laenzlinger (2002:6)
for the assumption that illocutionary adverbs occupy specMoodspeechact, which is the highest
functional projection within the inflectional layer (cf. below for criticism).
25
(50b) is ruled out only under the intended reading, as the aspectual adverb briefly is fine in
this position.
26
This analysis also explains why sentence-final adverbs may take propositional scope.
27
Both types of illocutionary adverbs have manner homophones (cf. (i) & (ii)) and briefly has
an aspectual homophone (iii) (cf. Greenbaum 1969; Ernst 2002:70ff.):
(i) She spoke roughly to him.
(ii) They spoke frankly about their problems.
(ii) Could we briefly stop here?
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 395
28
See Ernst (2002:99) for the assumption that honestly-type, but not roughly-type adverbs, are
lexically specified as optionally carrying "their own, lexically encoded assertion operator with
them".
29
See also chapter 4.3.1 for the co-occurrence of subject-oriented adverbs and subject-attitude
adverbs.
30
See Ernst (2002:99) for a related, but different, proposal.
31
Further members of this class are: absurdly, appropriately, conveniently, disgustingly,
incredibly, ironically, mercifully, paradoxically, regrettably, remarkably, shockingly, tragi-
cally, understandably, etc. (cf. Swan 1988a:32f. for a more comprehensive list of evaluative
adverbs).
32
Evaluative adverbs like oddly and strangely have homophonous manner adverbs, while
unfortunately and amazingly do not (cf. (i) & (ii) vs. (iii) & (iv)):
(i) [She was greeted by] a young man who looked at her oddly. (BNC)
(ii) [He] said Tim was breathing strangely. (CC)
(iii) * [She was greeted by] a young man who looked at her unfortunately.
(iv) * [He] said Tim was breathing amazingly.
Ernst (2002:76) labels the former dual evaluatives and the latter pure evaluatives.
33
Cf. Bellert (1977:342ff.), Ernst (2002:100).
34
See also Schreiber (1971); Huang (1975) and Swan (1988b) for the qualification that evalua-
tive adverbs are factive.
35
This is not entirely true since, in the case of ideally, hopefully, etc., "the truth value of the
relevant proposition is unknown (though hoped to be true)" (Ernst 2002:77):
(i) Ideally, you should eat all that has been suggested at each mealtime. (BNC)
(ii) Hopefully, coming out will make him a better performer. (CC)
Irrespective of this, ideally, hopefully, etc. are subject to the same restrictions: they fail to occur
in the scope of interrogative operators, sentential negation and aspectual auxiliaries. The cases
in (ii) & (iii) are equally problematic since surprisingly and unfortunately are evaluative ad-
verbs that take scope over facts. The presence of would and should, however, precludes the
propositions from being interpreted as facts:
(ii) Kawaguchi would somewhat surprisingly admit to early 'experiences'. (CC)
(iii) So should you unfortunately die your family would receive cash benefits. (CC)
I have no solution to offer. However, it may be noteworthy at this point that factivity is not
only a problem with evaluative adverbs, but also with so-called factive verbs. As has been
argued by Karttunen (1971), so-called factive verbs fall into two classes: full factives (regret,
confess in (iv) & (v)) and semi-factives (discover, know (vi) & (vii)):
(iv) We regret that we are unable to answer multiple queries … (BNC)
(v) I confess that I have my doubts. (BNC)
(vi) [He] will discover that there is no Author (BNC)
(vii) But at least you know that Phil likes you. (BNC)
The major difference between full factives and semi-factives is that full factives, if in the scope
of negation or possibility (viii) & (ix), trigger the presupposition that the proposition is true,
whereas semi-factives do so only under negation (x), not under possibility (xi):
(viii) We don't regret that we are unable to answer … -> we are unable to answer
(ix) We might regret that we are unable to answer … -> we are unable to answer (x)
[He] will not discover that there is no author -> there is no author
(xi) [He] might discover that there is no author -g> there is no author
36
However, we find cases, in 'legalese' mostly, as in (i) & (ii), where an evaluative adverb
follows sentential negation:
396 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
49
The judgment that evidential adverbs are barred from following scene-setting expressions is
valid only within the complementizer layer: if the evidential adverb is realized within the
inflectional layer, it may follow the scene-setting expression:
(i) My main pain was that, last year, Apple had clearly not thought out it's [Sic!]
new UI. (www)
(ii) Last year, Nyaga had apparently run out of legal options. (www)
Admittedly, there are cases in which the underlying order 'evidential adverb > scene-setting
expression' is reversed within the complementizer layer, but they are significantly lower in
number:
(iii) In 1999, apparently, you need to be more socially responsible in your portrayals
of high school horn-dogs. (www)
(iv) Last year, obviously things didn't work out like we wanted them to. (www)
Strings as in (v) & (vi) are irrelevant since the temporal expression is argumental:
(v) Last year definitely had more 'storm' feeling to it. (www)
(vi) Last year definitely opened my eyes," he said. (www)
50
But see Bellert (1977:342ff.) for arguments that evaluative adverbs, while admissible within
the scope of counterfactual operators (i), are barred from occurring within the scope of inter-
rogative operators (ii) ((i) & (ii) adapted from Bellert 1977:344f.):
(i) If John had not been sick, he would {evidently, certainly, surely, definitely}
have done it.
(ii) *{Will/has} John {evidently, certainly, surely, definitely} come?
The arguments Bellert (1977) adduces with respect to the asymmetry between (i) & (ii) are not
conclusive: (i) is acceptable since counterfactuals do not require that the proposition they take
scope over be a fact (cf. Bellert 1977:244f.); (ii) is ruled out since evidential adverbs "take the
truth value of their sentences as their argument" (cf. Bellert 1977:344) and the truth of a propo-
sition cannot simultaneously be asserted and questioned (for the same reason, evidential ad-
verbs should also be barred from occurring within the scope of sentential negation (cf. Bellert
1977:343). Interestingly, however, Bellert (1977) admits that evidential adverbs may occur in
interrogatives if they have "an additional meaning component" (cf. (iii)), e.g. the implication
that "the matter under consideration has not been definitely settled before" (Bellert 1977:344):
(iii) Has John definitely made up his mind?
51
See also Cinque (2004:688, note 13) for arguments that Ernst's (2002:104) analysis allows
for the derivation of (119).
52
Epistemic adverbs are also referred to as modal adverbs (cf. Bellert 1977:343ff.; Ernst
2002:73ff).
53
Cf. Bellert (1977:344) for discussion. But see Jackendoff (1977:84ff.) for the claim that
"speaker-oriented adverbs" do not felicitously occur in "subject-auxiliary inversion contexts".
54
Ernst (2002) takes strings as in (126) to be exceptional and suggests that epistemic adverbs
that may be within the scope of negation are marked as to allow for "an interpretation along the
lines of 'even on the widest set of possibilities', so that [the adverb] is a sort of domain modifier
of [certain modals]" (Ernst 2002:479, note 17).
55
Epistemic adverbs immediately preceding subject-oriented adverbs take narrow scope just
over the subject-oriented adverb, in which case they cannot be modified and thus are heads
similar to focusing particles (as we shall see, maybe does not support modifiers at all) :
(i) (*{Very, Quite, Most}) probably wisely, he chose the matches in no particular
order. (www)
(ii) The Minister, (*{very, quite, most}) possibly wisely, did not think it worth while
to devote a line to the objections I put forward on Second Reading... (www)
(iii) And maybe foolishly, he expected things to go on as they were. (www)
398 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
56
Alternatively, one could assume that epistemic adverbs are merged into specModP headed
by an epistemic modal. To account for the fact that the presence of epistemic adverbs is not
contingent on the presence of an epistemic modal (cf. cases in which no modal is present), we
would have to assume that ModP is always projected. This is not a problem as such. However,
under this analysis, epistemic adverbs would always be within the scope of negation, since
NegP is located above ModP. Under the analysis presented, the affinity of epistemic adverbs
and epistemic modals would be accounted for by epistemic modals raising through the head
position of EpiP.
57
Moreover, raising to specFocP does not induce Minimality violations (cf. Rizzi 2002:10ff.,
19f.), i.e. if the epistemic adverb in (146) had targeted specFocP, the result should be well-
formed.
58
A quick search of the BNC for the pattern 'she > epistemic adverb' yielded 202 hits for
probably, 44 for possibly and only 5 for maybe.
59
As Rohdenburg (2003:244, note 18) points out, the term horror aequi goes back to Brug-
mann (1909, 1917/1920) (see also Menn & MacWhinney 1984a, b; McCawley 1988a:305ff.).
60
I assume that maybe in (151) buddies as a modal and thus raises to the head position of
AgrSP :
(i) Well I maybe not see the problem … (www)
(ii) Sorry, I maybe not post the message to this mail list. (www)
61
Speakers who accept (130c) probably do so by analogy with (130a, b).
62
Notice that (156e) is acceptable under the degree of perfection reading of seriously.
63
That is to say that the strings under consideration comply with the Horror Aequi Principle
(cf. Rohdenburg (2003:236)).
64
Notice also that the vast majority of strings of the type (159) involves probably.
65
Recall that the co-occurrence of temporal expressions and speaker-oriented adverbs within
the complementizer layer (with the exception of epistemic adverbs (cf. (i)) is restricted to the
linear order 'speaker > temporal expression':
(i) A dozen years ago, New Hampshire was probably the last place the first
President Bush wanted to run a campaign. (www)
66
Theoretically, this pattern should also be available with illocutionary adverbs:
(i) When I came back from America to Switzerland ten years ago, I was, frankly
speaking, shocked. (www)
However, since illocutionary adverbs that occur within the inflectional layer occupy specAttP,
they tend to receive a subject-attitude interpretation rather than the speaker oriented interpreta-
tion if preceded by temporal expressions:
(ii) Only a short time ago she had frankly thrown her arms around his neck. (www)
67
See Costa (2000:20f.) for arguments that the stacking of adverbs is often dependent on the
presence of a lexically filled head.
68
See also Cinque (2004:689f.) for discussion.
69
Cf. chapter 3.1 for coordination structures as being asymmetric and involving the projection
:P (the first conjunct occupies the specifier position and the second conjunct the complement
position).
70
specTP was chosen since factivity, ultimately, feeds on finiteness.
71
Evidential adverbs may follow both illocutionary adverbs and evaluative adverbs.
72
The fact that illocutionary adverbs within the inflectional layer are barred from co-occurring
with subject-attitude adverbs follows from their competing for the same structural slot.
ADVERBS WITHIN THE COMPLEMENTIZER LAYER, MOSTLY 399
73
If realized within the inflectional layer or within VP, illocutionary, evaluative and evidential
adverbs may follow setting expressions and focalized constituents (cf. below).
74
See also Cinque (1999:110ff.), Koster (2000:4) and Rizzi (2002:10ff., 19f.).
75
Since means-domain adverbs are not gradable, they cannot be accompanied by so and they
cannot be questioned.
76
Exclamatives are a further case in point:
(i) Oh how loudly she did yell and how loudly she did call… (www)
(ii) How reluctantly she examined the corpse!
(iii) How horribly Madonna performed 'American Pie'.
(iv) How completely you've changed!
77
See Ernst (2004b:111ff.) for a brief discussion of topic reading of sentence-initial pure
domain adverbs.
78
Recall from chapter 3 that Ernst (2002:40), in connection with (i), brings into play Relativ-
ized Minimality, i.e. he takes calmly to be a manner adverb that has raised across a subject-
related adverb:
(i) * Calmly(,) he cleverly answered all their questions.
However, his analysis rests on the erroneous assumption that manner adverbs (in the absence
of higher adverbs) may freely prepose, i.e. he takes calmly in (ii) to be a manner adverb:
(ii) Calmly(,) he answered all their questions.
Strings as in (iii) clearly show that calmly must be subject-related:
(iii) Calmly(,) he answered all their questions quickly.
79
Ernst (1985:171) contrasts the well-formed strings as in (196) with (i) & (ii):
(i) * Tragically, did the attempt oil the enemy stronghold fail?
(ii) * Possibly, what are the implications of that speech?
However, these examples compromise the otherwise adequate observation that preposed 'sen-
tence adverbs' (cf. (197)) fail to precede interrogative operators since tragically, as an evalua-
tive adverb, is confined to assertive contexts and since possibly fails to raise beyond
specPromP.
80
Strings as in (i) are irrelevant here since maybe takes narrow scope over the pure domain
adverb, i.e. the better approach is maybe within the domain of psychology:
(i) Maybe psychologically, they've got a better approach to it,… (www)
81
I have suggested that the [NEG]-feature borne by a negated constituent is activated in the
presence of an additional [FOC] feature.
82
The latter position is also taken in (i) & (ii) with the subject having raised across the adverb:
(i) She again would choose homebirth in future. (www)
(ii) She recently had been taking medication to help her sleep. (www)
83
Simple adverb preposing is illicit:
(i) * Often i they usually ti drink?
(ii) * Long i they have already ti lived together in the carriage house.
84
(211a) is potentially ambiguous between the high and the low reading of often.
85
Similar observations can be made with respect to already (www-data):
(i) Already, they often feel shunned and unwanted, like second-class citizens.
(ii) The correct answer may only exist retrospectively and already we no longer may
be the masters of our destiny.
86
Recall from section 5.2.1.1 that illocutionary adverbs and subject-oriented adverbs (order
irrelevant) cannot co-occur:
(i) * Frankly she wisely had avoided this topic.
(ii) * Wisely she frankly had avoided this topic.
400 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
87
Notice that seriously in (217i) has subject-orientation.
88
In cases in which PromP is recursive (cf. (i) & (ii), where both an epistemic and an aspectual
adverb are preposed) the order of the adverbs is dictated by Minimality:
(i) They probably quite often will find these things stupid.
(ii) * They quite often probably will find these things stupid.
89
As has been pointed out to me by Jennifer R. Austin, (221a, b) & (221e, f) are only accept-
able under the proviso that fortunately and evidently be maximally backgrounded with the pitch
accent falling on the final syllable of the lexical verb. Moreover, she informs me that fortu-
nately and evidently somehow seem not to take scope over obviously and cleverly, i.e. it seems
that the speaker considers the receiving of nasty letters fortunate, not the obvious receiving of
nasty letters (cf. (221a, b)). Likewise, with respect to (221e, f), it seems that the speaker
considers the withholding of details evident or fortunate.
CHAPTER 6
GENERAL CONCLUSION
(6) [vP [spec ext. arg.] [v' [v ] [AgrIOP … [AgrOP … [VP int. arg(s). ]]]]]
(7) a. [v' [v senti] [TelP [DP a frog]j [Tel' [Tel ti] [AgrIOP [KP to him]k [AgrIO' [AgrIO ti]
[AgrOP [DP tj] [AgrO' [AgrO ti] [VP [KP tk] [V' [V ti] [DP tj]]]]]]]]]]
b. [v' [v senti] [TelP [D him]j [Tel' [Tel ti] [AgrIOP [D tj ] [AgrIO' [AgrIO ti] [AgrOP [DP a
frog] [AgrO' [AgrO ti] [VP [KP tk] [V' [V ti] [DP tj]]]]]]]]]]
GENERAL CONCLUSION 403
Second, the lexical layers projected by active and passive verbal elements
differ with respect to their internal makeup and active and passive verbal
elements display different movement properties (other traces omitted). In
actives, the lexical verb is unanimously attracted to v, which hosts a strong Θ-
feature that must be checked off for reasons of PF convergence, whereas in
passives, verb raising is conditioned by discourse requirements, i.e. the passive
participle raises to AgrPrtP if it needs be defocalized, otherwise it stays put:
(8) a. [vP [DP ] [v' [v Vi ] … [AgrIOP [AgrIO [AgrIO ti] [AgrOP [AgrO' [AgrO ti] … [VP [V'
[V ti]]]]]]]]]
b. [AgrPrtP [AgrPrt' [AgrPrt e] … [VP [V' [V V ]]]]] or
[AgrPrtP [AgrPrt' [AgrPrt Vi] … [VP [V' [V ti]]]]] (if V is defocalized)
Third, the adverbs under consideration as well as low frequency adverbs, low
adverbs of repetition and restitutive again (cf. chapter 4.3.3.1) occupy unique
specifier positions of designated and rigidly ordered preverbal functional
projections (9a) that are merged into structure between- AgrOP and VP in
actives (9b) and between AgrPrtP in passives (9c):
(9) a. SpaceP > AgentP > DomP > µP > DegPerf >
here deliberately universally loudly slightly
MeansP > FreqP > RepP > AspPcompl > RestP
manually rarely again completely again
b. [vP [TelP [AgrIOP [AgrOP [SpaceP [AgentP [DomP [µP [DegPerfP [MeansP [FreqP [RepP
[AspPcompl [RestP [VP]]]]]]]]]]]]]]]
c. [AgrPrtP … [SpaceP [AgentP [DomP [µP [DegPerfP [MeansP [FreqP [RepP [AspPcompl
[RestP [VP ]]]]]]]]]]]
Word order variation within the lexical layer proper, i.e. the domain that is
delimited by V and vP in actives and V and AgrPrtP in passives, was shown to
be the result of a small number of syntactic operations, notably overt object
shift and verb raising across preverbal adverbs or adverb incorporation into V
and excorporation out of V in Tel for completion adverbs. My account of non-
canonical orders of adverbs (also in relation to other inmates of the lexical
layer) built on the supposition that there exists a semantically empty-headed
verbal projection below the base position of the lexical verb, VP:4
(10) a. [vP [SpaceP [TelP [AgrIOP [AgrOP [AgentP [DomP [µP [DegPerfP [MeansP [FreqP [RepP
[AspPcompl [RestP [VP [V [VP [V' [V ] [VP …]]]]]]]]]]]]]]]]]
b. [SpaceP [AgrPrtP … [AgentP [DomP [µP [DegPerfP [MeansP [FreqP [RepP [AspPcompl [RestP
[VP ] [V [VP [V' [V ] [ VP …]]]]]]]]]]]]]]]
404 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
I have shown that lexical material whose natural habitat is the lexical layer (cf.
(9)) may be realized within VP on the condition that it be properly licensed
from within the lexical layer of the clause. Long-distance licensing was shown
to be contingent on 'antecedent binding' in the sense that elements that are
realized within VP form representational chains with their null counterparts
which are licensed under specifier-head agreement within designated func-
tional projections, thereby preventing the proliferation of empty structure
below the lexical verb (irrelevant projections omitted).5 The realization of
lexical material within VP was shown to be warranted by discourse
requirements, i.e. (de)focalization and/or prosodic restrictions:
(11) a. Thisi has been [AgrPrtP [DP ti ] [AgrPrt' [AgrPrt ] [AgentP [Adv e[+AG]] [Agent'
[AgentP ] [DegPerfP [Adv poorly] [DegPerf' [DegPerf ] [VP [DP ti] [V written] [VP
[ V' [ Adv deliberately]]]]]]]]]]
b. He handedk [TelP [DP the napkin]i [Tel' [Tel tk] [AgrIOP [KP pro]m [AgrIO'
[AgrIO tk] [AgrOP [DP ti] [AgrO' [AgrO tk] [µP [AdvP secretly] [µ' [µ tk ] [VP [KP tm]
[V' [V tk ] [VP [DP ti] [V' [V tk] [VP [V' [ to her ]]]]]]]]]]]]]]]
(12) a. [vP [SpaceP [TelP [AgrIOP [AgrOP [AgentP [DomP [µP [DegPerfP [MeansP [FreqP [RepP
[AspPcompl [RestP [VP ]]]]]]]]]]]]]]]
b. [SpaceP [AgrPrtP … [AgentP [DomP [µP [DegPerfP [MeansP [FreqP [RepP [AspPcompl
[RestP [VP]]]]]]]]]]]
Moreover, the assumption that for each adverb(ial) merged into VP there is a
corresponding null counterpart, which is subject to licensing under specifier-
head agreement in the interpretable part of the clause, was shown to allow us to
neatly account for the complementary distribution of head final AdvPs and
larger syntactic constructions with adverbial function and focalized adverbial
elements, as shown in (13) for pure domain adverb(ial)s, and also for reversed
canonical orders of elements, as shown in (14c) and (14f) for temporal and
spatial expressions:6
GENERAL CONCLUSION 405
(14) a. Ninety-one cases have since been diagnosed in the United Kingdom.
b. * Ninety-one cases have been diagnosed since {here, in the UK}.
c. Ninety-one cases have been diagnosed in the U K {since 1999, since
the launch of the research program}.
d. [Millet] and sorghum are currently harvested in the north.
e. * Millet and sorghum are harvested {in the north, there} currently.
f. [Millet] and sorghum are harvested in the north {after sunset, during
term break}. ((a) & (d) from www)
The analysis presented accounts for all and only the admissible co-occurrence
patterns of elements that are licensed (from) within the lexical layer of the
clause. In addition, the canonical order of vP-adverbs may be reversed under
raising to the complementizer layer. As we have seen in chapter 3.2.1.5.2, only
vP-internal adverbs that are endowed with the peripheral feature [ FOC] or [Q]
may leave vP:
Chapter 4 met the challenge of providing a uniform account for the distribution
and licensing of adverb classes as diverse as subject-related adverbs, inherently
negative adverbs, various classes of aspectual adverbs, temporal expressions
and focusing particles. The central idea, again, was that the distribution and
licensing of said adverb classes within the inflectional layer is contingent on
specifier-head agreement within designated and rigidly ordered functional pro-
jections:
(16) EpiP > NegP > PrtP > ReltP > AspP* > AttP
probably not only now AspP* carefully
(17) AspP habit > AspP rep > AspP freq > AspP celerative > AspP ant > AspP term >
usually again often quickly already no longer
AspP cont > AspP (im)perf > AspP retro > AspP prox > AspP dur > AspP prospect
still never lately soon briefly almost
The set of functional projections in (16) is preceded by FinP and AgrSP and
split by functional projections the head position of which are the merge sites
for non-lexical verbs (ModP for modals, AuxP for be and beP for passive be)
and the checking sites for participial features of all lexical and the non-lexical
verbs be and bepass (cf. chapter 4.1). The head of PerfP is the checking site for
past participles and the head of ProgP the checking site for present participles
(recall that the checking site for passive participle is the head of AgrPrtP within
the inflectional layer of the clause). ReltP was shown to be special insofar as its
head is the merge site of the perfective auxiliary have and its specifier position
the licensing site for temporal expressions:8
(18) FinP > AgrSP > EpiP > NegP > TP > PrtP > ModP > ReltP > AspP* >
AttP > PerfP > ProgP > AuxP > be
(19) a. She may have often been cleverly avoiding encounters with aliens.
b. She has always been foolishly trying to convince them.
c. * She may have cleverly been often avoiding encounters with aliens.
d. * She has foolishly been always trying to convince them.
(20) a. She would not [PrtP [ OPi ] [Prt' [PrtP ] … [[only]i eat BEANS]]].
b. She may [PrtP [ OPi ] [Prt' [PrtP ] … [[even]i [REFUSE the offer]]].
c. He has [PrtP [ OPi ] [Prt' [PrtP ] … [[only]i [RECENTLY joined the club]]].
(21) a. She cleverly has often been avoiding trouble. (subject-oriented >
aspectual)
b. She has often been cleverly been avoiding trouble. (aspectual >
subject-attitude)
c. * She has cleverly been often avoiding trouble. (subject-attitude >
aspectual)
(24) a. [PromP [Adv Often i] [Prom' [Prom ] … [AgrSP they would ti change on the
move]]]. (BNC)
b. [PromP [Adv Frequently i] [Prom' [Prom ] … [AgrSP they would ti meet us at
the … gate]]] (BNC)
c. [PromP Already [TopP I [PromP no longer have to stuff my bra, I am
thrilled with the results so far]]. (www)
(25) a. [SoP Foolishly [TopP he [PromP often would wear his pink hairpiece in
public]]].
b. [SceneP Yesterday [AgrSP illicit smokers were fined for taking a puff]].
408 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
c. [SceneP At the time, [TopP I [SoP foolishly did not realize that all of you
did not want to be spoken to in rational terms]]]. (www)
((b) from BNC)
Again, the assumption that for each adverb(ial) merged into VP there is a
corresponding null counterpart, which is licensed under specifier-head
agreement in the interpretable part of the clause (here the inflectional layer),
allows us neatly to account for the complementary distribution of e.g. single-
term habitual adverbs and prosodically and/or syntactically heavier habitual
expressions:
(26) a. People in their thirties are generally trying to impress the boss. (CC)
b. People in their thirties are trying to impress the boss {GENERALLY, in
most cases}
c. * People in their thirties are generally trying to impress the boss
{GENERALLY, in most cases}
(27) ForceP > TopP > EvalP > TopP > EvidP > SceneP >
frankly amazingly clearly in 1066
FocP > TopP > SoP > TopP > PromP
only horribly[FOC] wisely often[PROM]
how loudly[Q]
(29) a. * I know that frankly she should be concerned. (*frankly > that)
b. She said that apparently there was pressure… (*apparently > that)
c. He says that last year each SARS victim … (*last year > that)
d. He realised that foolishly he'd managed to … (*foolishly > that)
((b) & (d) from www)
(32) a. Luckily for me, usually[PROM] the train takes … a stop (*[PROM] >
luckily)
b. A few years ago, he frequently[PROM] heard gunshots… (*[PROM] > a
few years ago)
c. Foolishly he often[PROM] would wear his pink hairpiece. (*[PROM] >
foolishly)
((a) & (b) from www)
410 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
The interrelated assumptions that the complementizer layer has the structure in
(27) and that illocutionary adverbs, evaluative adverbs, evidential adverbs,
scene-setting expressions and subject-oriented adverbs are merged as specifiers
within the complementizer layer are corroborated by the absence of Minimality
Effects in (31a, b) and (32), which would have arisen had the raised elements in
specFocP and specPromP been crossed by the sentence-initial adverb(ial)s.
The fact that speaker oriented adverbs, all of which I analyzed as operators,
i.e. illocutionary adverbs are assertive operators ([ASS]), evaluative adverbs are
factive operators ([FACT]) and evidential adverbs are a special kind of factive
operators ([@FACT]), may assume non-canonical positions, such as the post-
subject position (33), positions within the inflectional layer (34) and the
sentence-final position (35), were accounted for as follows:14
emerges, with the phrases set in bold face representing the licensing sites of
adverbs,15 with adverbs (irrespective of being merged or raised) entering into a
transparent semantic relationship with their licensing functional heads:
(36) complementizer ForceP > TopP > EvalP > TopP > EvidP > SceneP >
layer FocP > TopP > SoP > TopP > PromP >
FinP > AgrSP > EpiP > NegP > TP > PrtP > ModP >
ReltP > AspPhab. > AspPrep. > AspPfreq. > AspPcel. >
inflectional
AspPant. > AspPterm. > AspPcont. > AspP(im)perf. >
layer
AspPret. > AspPprox. > AspPdur. > AspPpros. > AttP >
PerfP > ProgP > AuxP > beP
lexical vP > SpaceP > TelP > AgrIOP > AgrOP > AgentP >
layer DomP > µP > DegPerfP > MeansP > FreqP >
RepP > AspPcompletive > RestP > V >
non-interpret-
table domain VP*
Throughout this study, it has been assumed with Cinque (1999) that each
functional projection has its own specific semantic interpretation16 and that
adverbs enter into transparent semantic relations with the head of the functional
projection they occur in. In this context, note that the set of functional
projections whose specifier positions have been identified as merge sites of
adverbs, by and large, matches the set of semantic subclasses that have been
populating works on grammar since traditional grammar:
At the outset of this study, it was shown that the elements constituting the
traditional word class Adv do not come together as one discrete category, i.e. a
category defined in its relation to syntax, but are scattered across a large set of
discrete categories which are identified on the basis of the distributional
properties of the lexical items constituting them. Wanting an elegant way of
referring to these categories, I employed semantically primitive labels such as
those in (37). Importantly, neither the fact that the hierarchy of functional
projections reflects semantic properties nor the choice of labels are meant to
imply that the syntax of adverbs follows from their semantics.
Notes
1
This proposal was inspired by analyses as different as Frey & Pittner's (1999) semantic scope,
Williams' (1986) scope marker and also by Haegeman's (1995) representational chains.
2
Cf. Koizumi (1993), Runner (1995, 2000, 2001) and Lasnik (1995, 1999a,b, 2002). See also
Pesetsky (1989), Johnson (1991), Harley (1995), Radford (1997), Harley & Noyer (1998),
Baltin (2000, 2001) and Adger & Tsoulas (2000).
3
See Adger & Tsoulas (2000:14) for related arguments.
4
Cf. Chomsky (1995), Frey & Pittner (1999:38f.), Frey (2000:130f.) and Haider (2000:125;
2004:802) for related assumptions.
5
The by-phrase in passives was shown to be realized within VP, but licensed from within the
lexical layer by the features [+PASS, +AG] on the passive participle (cf. chapter 3.2.1.5.3)):
(i) thisi has been [AgrPrtP [DP ti ] [AgrPrt' [AgrPrt encouragedk] [AgentP [Adv deliberately] [Agent'
[AgentP [+PASS, +AG]] [VP [DP ti] [V tk ] [ VP [V' [PP by strategists]]]]]]]]]]
6
But see Cinque (2004:699f.) for arguments that the elements under consideration in fact com-
pete for the same specifier position and that sentence-final occurrences are derivative, i.e. the
result of VP-remnant intraposition.
7
Recall that, under Cinque's (1999:106, passim) proposal, subject-oriented adverbs are sand-
wiched between Aspfreq and AspPcelerative. Recall further that Cinque does not distinguish
between subject-oriented adverbs and subject-attitude adverbs.
8
See Solà (1996:227) for the status of have as the head of ReltP.
9
Cf. König (1991a:10), Jacobs (1983:8ff.; 1984:115ff.), Bayer (1996:15f.; 1999:56).
10
See Bayer (1996, 1999) for an LF licensing account.
11
See (i) & (ii) where both types of subject-related adverbs and aspectual adverbs co-occur:
(i) She cleverly has often been cleverly avoiding trouble. (subj.-oriented > aspectual
> subj.-attitude)
(ii) John {twice/often/rarely} knocked on the door {twice/often/rarely} (Cinque 1999:27)
Note that (iii) is perfectly well-formed since carelessly is a subject-oriented adverb, not a sub-
ject-attitude adverb:
(iii) They carelessly had again been making nasty comments.
12
As we have seen in chapter 5.1, interrogative operators and foci compete for the same struc-
tural position (cf. Rizzi 1999, 2002).
GENERAL CONCLUSION 413
13
Recall that the two sets of features, i.e. [WH/Q] or [FOC] and [PROM], are not equally distrib-
uted among the classes of adverbs that originate in the lexical and the inflectional layer: while
adverbs that originate in the inflectional layer may bear either set of features and thus assume
either specFocP or specPromP (the latter barring Minimality violations), adverbs that originate
in the lexical layer resist non-quantificational raising (cf. chapters 3.2.1.5.2 and 5.2.2).
14
Illocutionary adverbs come in two flavors: honestly-type adverbs, which specify the content
of the proposition, and roughly-type adverbs, which specify the form of the proposition (cf.
Bellert 1977:349ff.). Note that the following qualifications all relate to the former type.
15
Recall that, in passives, the upper end of the lexical layer is demarcated by AgrPrtP, not by
vP. Moreover, TelP is not projected. Depending on whether the direct or indirect object is
passivized, the lexical layer does not contain AgrOP or AgrIOP.
16
Cf. Chomsky (1995) for arguments that the "[p]ostulation of a functional category has to be
justified, either by output conditions (phonetic and semantic interpretation), or by theory inter-
nal arguments" (cf. Chomsky 1995:240).
REFERENCES
Abney, S.P. 1987. The English Noun Phrase in Its Sentential Aspects. PhD Disserta-
tion, MIT.
Aboh, E.O. 1997. From the Syntax of Gungbe to the Grammar of Gbe. PhD Disserta-
tion, University of Geneva.
Aboh, E.O. 2001. Object shift and verb movement in Gbe. Generative Grammar in
Geneva 2: 1-13.
Adger, D. and G. Ramchand 2001. Phases and interpretability. West Coast Conference
on Formal Linguistics 20: 1–14.
Adger, D. and G. Tsoulas 2000. Aspect and lower VP-adverbials. In Adverbs and
Adjunction, A. Alexiadou and P. Svenonius (eds.), 1-18. Potsdam: Universität
Potsdam.
Alexiadou, A. 1997. Adverb Placement: A Case Study in Antisymmetric Syntax.
Amsterdam: John Benjamins.
Alexiadou, A. 2002a. On the status of adverb in a grammar without a lexicon. In Das
Adverb – Zentrum und Peripherie einer Wortklasse, Schmöe, F. (ed.), 25-42.
Wien: Edition Praesens.
Alexiadou, A. 2002b. The syntax of adverbs: puzzles and solutions. Glot International
6(2/3): 33-54. At: http://www.linguistlistplus.com/glot/PDF/vol6/GLOT6_2-.pdf.
Alexiadou, A. 2004. Adverbs across frameworks. In Taking up the Gauntlet. Adverbs
across Frameworks, Alexiadou, A. (ed.), 677-82.
Andrews, A. 1983. A note on the constituent structure of modifiers. Linguistic Inquiry
14(4): 695-7.
Aoun, J. and Y.A. Li. 1989. Scope and constituency. Linguistic Inquiry 20(3): 141-72.
Arens, H. 1969. Sprachwissenschaft. Der Gang ihrer Entwicklung von der Antike bis
zur Gegenwart. Vol. I. Von der Antike bis zum Ausgang des 19. Jahrhunderts.
Frankfurt/Main: Athenaeum.
Aronoff, M. 1976. Word Formation in Generative Grammar. Cambridge, MA: The
MIT-Press.
Austin, J.R., S. Engelberg and G. Rauh. 2004. Current issues in the syntax and seman-
tics of adverbials. In Adverbials. The Interplay between Meaning, Context and
Syntactic Structure, J.R. Austin, St. Engelberg and G. Rauh (eds.), 1-44.
Amsterdam: Benjamins.
Authier, J.-M. 1991. V'-governed expletives, Case Theory and the Projection Princi-
ple. Linguistic Inquiry 22(4): 721-40.
Authier, J.-M. 1992. Iterated CPs and embedded topicalization. Linguistic Inquiry
23(2): 329-36.
REFERENCES 415
Baayen, H.R. and A. Renouf 1996. Chronicling the Times: productive innovation in
an English newspaper. Language 72(1): 69-96.
Baker, C.L. 1970. Notes on the description of English questions: the role of an abstract
question morpheme. Foundations of Language 6(2): 197-219.
Baker, M. 1985. The Mirror Principle and Morphosyntactic Explanation. Linguistic
Inquiry 16(3): 373-415.
Baker, M. 1988. Incorporation. A Theory of Grammatical Function Changing.
Chicago: The University of Chicago Press.
Baker, M. 1996. On the structural positions of themes and goals. In Phrase Structure
and the Lexicon, J. Rooryck and L. Zaring (eds.), 7-34. Dordrecht: Kluwer.
Baltin, M.R. 2000. Implications of Pseudo-Gapping for Binding and The Representa-
tion of Information Structure. Ms., New York University.
Baltin, M.R. 2001. The Interaction of Ellipsis and Binding: Implications for the
Sequencing of Principle A. Ms., New York University.
Barss, A. and H. Lasnik 1986. A note on anaphora and double objects. Linguistic
Inquiry 17: 347-54.
Bartsch, R. 1972. Adverbialsemantik. Frankfurt/Main: Athenaeum.
Bayer, J. 1996. Directionality and Logical Form. On the Scope of Focusing Particles
and Wh-in-situ. Dordrecht: Kluwer.
Bayer, J. 1999. Bound focus or how can association with focus be achieved without
going semantically astray? In The Grammar of Focus, G. Rebushi and L. Tuller
(eds.), pp.55-82. Amsterdam: Benjamins.
Beghelli, F. and T. Stowell 1997. Distributivity and negation: the syntax of each and
every. In Ways of Taking Scope, A. Szabolcsi (ed.), 71-107. Dordrecht: Kluwer.
Bellert, I. 1977. On semantic and distributional properties of sentential adverbs. Lin-
guistic Inquiry 8(2): 337-50.
Belletti, A. 1990. Generalized Verb Movement: Aspects of Verb Syntax. Turin:
Rosenberg & Sellier.
Belletti, A. 1994. Verb positions. In Verb Movement, D. Lightfoot and N. Hornstein
(eds.), 19-40. Cambridge: Cambridge University Press.
Belletti, A. 1998. Agreement Projections. To appear in: The Handbook of Syntactic
Theory. M. Baltin and C. Collins (eds.) Oxford: Blackwell.
Belletti, A. 2001. (Past) Participle Agreement. Ms., University of Siena.
Belletti, A. and L. Rizzi (eds.) 1996. Parameters and Functional Heads. Oxford:
Oxford University Press
Bianchi, V. 1999. Consequences of Antisymmetry. Headed Relative Clauses. Berlin:
Mouton de Gruyter.
Bianchi, V. 2000. On time adverbials. Rivista di Linguistica 12(1): 77-106.
Bittner, M. and K. 1996. The structural determination of case and agreement.
Linguistic Inquiry 27(1): 1-68.
Blight, R. 2000. VP ellipsis, predicate fronting, and verb positions in English. MIT
Working Papers in Linguistics 37: 1-13.
Bloomfield, L. 1933. Language. New York: Holt, Rinehart & Winston.
416 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Bobalijk, J.D. 1999. Adverbs: the hierarchy paradox. Glot International 4(9/10): 27-8.
Bouchard, D. 1995. The syntax of sentential negation in French and English. In
Negation and Polarity, D. Forget, P. Hirschbühler, F. Martineau and M.-L.
Rivero (eds.), 30-52. Amsterdam: Benjamins.
Bowers, J. 1975. Adjectives and adverbs in English. Foundations of Language 13(1):
529-62.
Bowers, J. 1993. The syntax of predication. Linguistic Inquiry 24(4): 591-656.
Bresnan, J. 1973. Syntax of the comparative clause constructions in English. Linguis-
tic Inquiry 4(3): 275-343.
Bresnan, J. 1977. Variables in the theory of transformations. In Formal Syntax, P.
Culicover, Th. Wasow and A. Akmajian (eds.), 157–196. New York: Academic
Press.
Browning, M.A. 1996. CP recursion and 'that-t' effects. Linguistic Inquiry 27(2): 37-
255.
Brugman, C. 1986. Sisterhood is more powerful than you thought: scopal adverb
placement and illocutionary force. In Papers from the Parasession on Pragmatics
and Grammatical Theory at the 22nd Regional Meeting of the Chicago Linguistic
Society, A.M. Farley, P.T. Farley and K.-E. McCullough (eds.), 40-53. Chicago,
IL: Chicago Linguistic Society.
Brugmann, K. 1909. Das Wesen der lautlichen Dissimilationen. Abhandlungen der
philologisch-historischen Klasse der Königlich-sächsischen Gesellschaft der
Wissenschaften 27: 141–178.
Burton, S. and J. Grimshaw 1992. Coordination and VP-internal subjects. Linguistic
Inquiry 23(2): 305-13.
Caponigro, I. and C.T. Schütze 2003. Parameterizing passive participle movement.
Linguistic Inquiry 34(2): 293-308.
Cattell, R. 1998. The English modifier well, In The Clause in English. P. Collins and
D. Lee (eds.), 53-65. Amsterdam: Benjamins.
Chomsky, N. 1957. Syntactic Structures. The Hague: Mouton.
Chomsky, N. 1965. Aspects of the Theory of Syntax. Cambridge, MA: The MIT Press.
Chomsky, N. 1981. Lectures on Government and Binding. Berlin: Mouton de Gruyter.
Chomsky, N. 1982. Some Concepts and Consequences of the Theory of Government
and Binding. Cambridge, MA: The MIT Press.
Chomsky, N. 1986a. Knowledge of Language. Its Nature, Origin, and Use. New York:
Praeger.
Chomsky, N. 1986b. Barriers. Cambridge, MA: The MIT Press.
Chomsky, N. 1989. A generalization of X-bar theory. In Studia Linguistica et Orienta-
lia Memoriae Haim Blanc Dedicata, Wexler, P., A. Borg and S. Somekh (eds.),
86-93. Wiesbaden: Otto Harrassowitz, 1989.
Chomsky, N. 1992. A Minimalist Program for Linguistic Theory. [MIT Occasional
Papers in Linguistics 1]. Cambridge, MA: The MIT Press.
REFERENCES 417
Chomsky, N. 1993. A minimalist program for linguistic theory. In The View from
Building 20. Essays in Honor of Sylvain Bromberger, K. Hale and S. Keyser
(eds.), 1-52. Cambridge, MA: The MIT Press.
Chomsky, N. 1994. Bare Phrase Structure. [MIT Occasional Papers in Linguistics 5].
Cambridge, MA: The MIT Press.
Chomsky, N. 1995. The Minimalist Program. Cambridge, MA: The MIT Press.
Chomsky, N. 1998. Minimalist Inquiries: The Framework. Ms. MIT
Chomsky, N. and H. Lasnik 1977. Filters and control. Linguistic Inquiry 8(3): 425-
504.
Chomsky, N. and H. Lasnik 1993. The theory of Principles and Parameters. In Syntax:
An International Handbook of Contemporary Research, J. Jacobs, A. von
Stechow, W. Sternefeld and T. Vennemann (eds.), 506-69. Berlin: Walter de
Gruyter.
Cinque, G. 1990. Types of A'-Dependencies. Cambridge, MA: The MIT-Press.
Cinque, G. 1999. Adverbs and Functional Heads. A Cross-Linguistic Perspective.
Oxford: Oxford University Press.
Cinque, G. 2001a. Complement and Adverbial PPs: Implications for Clause Structure.
Paper presented at 12. Wuppertaler Linguistisches Kolloquium, 16.–17.11.2001.
Cinque, G. 2001b. 'Restructuring' and the order of aspectual and root modal heads. In
Current Studies in Italian Syntax. Essays offered to Lorenzo Renzi, G. Cinque and
G. Salvi (eds.), 137-55. Amsterdam: Elsevier.
Cinque, G. 2004. Issues in adverbial syntax. In Taking up the Gauntlet. Adverbs
across Frameworks, A. Alexiadou (ed.), 683-710.
Comrie, B. 1976. Aspect. Cambridge: Cambridge University Press.
Contreras, H. 1987. Small clauses in Spanish and English. Natural Language and Lin-
guistic Theory 5(2): 225-44.
Corver, N. 1991. Evidence for DegP. Proceedings of Proceedings of the North East
Linguistic Society 21: 33-47.
Corver, N. 1997. Much-support as a Last Resort. Linguistic Inquiry 28(1): 119-164.
Costa, J. 1996. Adverb positioning and V-movement in English: some more evidence.
Studia Linguistica 50(1): 22-34.
Costa, J. 2000. Adverbs as adjuncts to non-universal functional categories: evidence
from Portuguese. In Adverbs and Adjunction, A. Alexiadou and P. Svenonius
(eds.), 19-32. Potsdam: Universität Potsdam.
Culicover, P. 1982. Syntax. New York: Academic Press.
Culicover, P. 1993. Evidence against ECP accounts of the THAT-T EFFECT. Linguistic
Inquiry 24(3): 557-61.
Curme, G. 1935. English Grammar. New York: Barnes and Noble Books.
Dahl, Ö. 1985. Tense and Aspect Systems. Oxford: Blackwell.
de Swart, H. 1993. Adverbs of Quantification: A Generalized Quantifier Approach.
New York: Garland.
Déchaine, R. 1993. Predicates Across Categories. Towards a Category-neutral Syntax.
PhD Dissertation, University of Massachusetts.
418 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Frey, W. and K. Pittner 1998. Zur Positionierung der Adverbiale im deutschen Mittel-
feld. Linguistische Berichte 176: 489-534.
Frey, W. and K. Pittner 1999. Adverbialpositionen im deutsch-englischen Vergleich.
In Sprachspezifische Aspekte der Informationsverteilung, M. Doherty (ed.), 14-
40. Berlin: Akademie Verlag.
Friedemann, M.-A. and T. Siloni 1997. Agrobject is not Agrparticiple. The Linguistic
Review 14(2): 69-96.
Fries, C.C. 1952. The Structure of English. An Introduction to the Construction of
English Sentences. New York & Burlingame: Harcourt, Brace & World, Inc.
Fukui, N. and M. Speas 1986. Specifiers and projection. MIT Working Papers in
Linguistics 8: 128-72.
Fukui, N. and Y. Takano 1998. Symmetry in syntax: merge and demerge. Journal of
East Asian Linguistics 7(1): 27-86.
Geis, M. 1987. On situation adverbs. Working Papers in Linguistics 35: 177-85.
Gleason, H.A. 1955 (21961). Introduction to Descriptive Linguistics. New York: Holt,
Rinehart & Winston.
Goldberg, A. and F. Ackerman 2001. The pragmatics of obligatory adjuncts. Lan-
guage 77(4): 798-814.
Greenbaum, S. 1969. Studies in English Adverbial Usage. Miami, FLA: University of
Miami Press.
Grimshaw, J. 1990. Argument Structure. Cambridge, MA: The MIT Press.
Grimshaw, J. 1991. Extended Projection. Ms., Brandeis University.
Grimshaw, J. 1994. Projection, Heads, and Optimality. Ms., Rutgers University.
Grimshaw, J. and J. Bresnan 1978. The syntax of free relatives in English. Linguistic
Inquiry 9(3): 331-391.
Guéron, J., H.-G. Obenauer and J.-Y. Pollock (eds.) 1984. Grammatical Representa-
tion. Dordrecht: Foris.
Haegeman, L. 1995. The Syntax of Negation. Cambridge: Cambridge University Press.
Haegeman, L. 2002. Sentence-medial NP-adjuncts in English. Nordic Journal of Lin-
guistics 25(1): 79-108.
Haegeman, L. and R. Zanuttini 1991. Negative heads and the Neg-Criterion. The
Linguistic Review 8(2/4): 233-52.
Haegeman, L. and R. Zanuttini 1996. Negative Concord in West Flemish. In Para-
meters and Functional Heads, A. Belletti and L. Rizzi (eds.), 117-79. Oxford:
Oxford University Press.
Haider, H. 2000. Adverb placement – convergence of structure and licensing. Theo-
retical Linguistics 26(1/2): 95-134.
Haider, H. 2004. Pre- and postverbal adverbials in OV and VO. In Taking up the
Gauntlet. Adverbs across Frameworks, A. Alexiadou (ed.), 779-807.
Hale, K. and J. Keyser (no date). On the Double-Object Construction. Ms., MIT.
Hale, K. and J. Keyser 1991. On the Syntax of Argument Structure. [Lexicon Project
Working Papers 34]. Cambridge, MA: MIT, The Center for Cognitive Science.
Harley, H. 1995. Subjects, Events and Licensing. PhD Dissertation, MIT.
420 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Harley, H. and R. Noyer 1998. Mixed nominalizations, short verb movement and ob-
ject shift in English. Proceedings of the North East Linguistic Society 28: 143-
157.
Harris, Z. S. 1964. Distributional structure. In The Structure of Language. Readings in
the Philosophy of Language, J.A. Fodor and J.J. Katz (eds.), 33-49. Englewood
Cliffs, NJ: Prentice Hall, Inc.
Harris, Z.S. 51961. Structural Linguistics. Chicago: The University of Chicago Press.
Haumann, D. 1997. The Syntax of Subordination. Tübingen: Niemeyer.
Haumann, D. 1999. Adverbial subordination: main clauses as specifiers. In Anglis-
tentag: Proceedings of the Conference of the German Association of University
Teachers of English. Vol. XIX, F.W. Neumann and S. Schülting (eds.), 77-97.
Trier: Wissenschaftlicher Verlag Trier.
Haumann, D. 2000. Aspects of the Verbal Complex in English. In Interkulturelle
Thessaloniker Analysen, K. Dorfmüller-Karpusa and K. Vretta-Panidou (eds.),
161-70. Frankfurt/Main: Peter Lang.
Haumann, D. 2002. The projections of temporal prepositions. In Perspectives on
Prepositions, H. Cuyckens and G. Radden (eds.), 25-45. Tübingen: Niemeyer.
Haumann, D. 2004. Degree Phrases versus Quantifier Phrases in prenominal and pre-
verbal positions: a hybrid explanation for some distributional asymmetries. In
Adverbials. The Interplay between Meaning, Context, and Syntactic Structure,
J.R. Austin, S. Engelberg and G. Rauh (eds.), 167-203. Amsterdam: Benjamins.
Hendrick, R. 1976. Prepositions and the X-bar Theory. In Proposals for Semantic and
Syntactic Theory, J. Emonds (ed.), 95-122. Los Angeles: UCLA Papers in Syn-
tax.
Higginbotham, J. 1985. On semantics. Linguistic Inquiry 16(4): 547-93.
Hoekstra, T. 1992. On the parameterization of functional projections in CP. Proceed-
ings of the North East Linguistic Society 23, 191-204.
Hopper, P. and S. Thompson 1980. Transitivity in grammar and discourse. Language
56(2): 251-99.
Hornstein, N. 1990. As Time Goes By. Cambridge, MA: The MIT-Press.
Huang, S.-F. 1975. A Study of Adverbs. The Hague: Mouton.
Huddleston, R. 1984. Introduction to the Grammar of English. Cambridge: Cambridge
University Press.
Iatridou, S. and D. Embick 1994. Conditional inversion. Proceedings of the North
East Linguistic Society 24, 189-203.
Jackendoff, R. 1972. Semantic Interpretation in Generative Grammar. Cambridge,
MA: The MIT Press.
Jackendoff, R. 1973. The base rules for prepositional phrases. In A Festschrift for
Morris Halle, S. Anderson and P. Kiparsky (eds.), 345-56. New York: Holt,
Rinehart & Winston, Inc.
Jackendoff, R. 1977. X'-Syntax: A Study of Phrase Structure. Cambridge, MA: The
MIT Press.
Jackendoff, R. 1990a. Semantic Structures. Cambridge, MA: The MIT Press.
REFERENCES 421
Koizumi, M. 1993. Object agreement phrases and the split VP hypothesis. In Papers
on Case and Agreement I, C. Philips and J. Bobalijk (eds.), 99-148. MIT Work-
ing Papers on Linguistic.
König, E. 1991a. The Meaning of Focus Particles. A Comparative Perspective. Lon-
don: Routledge.
König, E. 1991b. Gradpartikeln. In Semantik/Semantics: An International Handbook
of Contemporary Linguistic Research, A. von Stechow and D. Wunderlich (eds.),
786-803. Berlin: Walter de Gruyter.
Koopman, H. 1993. The Structure of Dutch PPs. Ms., U.C.L.A.
Koopman, H. 2000. The Syntax of Specifiers and Heads. New York: Routledge.
Koopman, H. and D. Sportiche 1985. Θ-Theory and Extraction. GLOW Newsletter 14:
57-8.
Koopman, H. and D. Sportiche 1991. The position of subjects. Lingua 85(2): 211-58.
Koster, J. 1999a. Empty Objects in Dutch. Ms., University of Groningen.
Koster, J. 1999b. The word orders of English and Dutch: Collective vs. individual
checking. In Groninger Arbeiten zur germanistischen Linguistik, W. Abraham
(ed.), 1-42. Groningen: University of Groningen.
Koster, J. 2000. Pied-piping and the word orders of English and Dutch. Proceedings of
the North East Linguistic Society 30: 415-426.
Kratzer, A. 1988. Stage-level and individual-level predicates. In Genericity in Natural
Language. Proceedings of the 1988 Tübingen Conference, [SNS-Bericht]. M.
Krifka (ed.), 247-284. Tübingen, University of Tübingen.
Kuno, S. 1975. Conditions for verb phrase deletion. Foundations of Language 13(2):
161-75.
Kuroda, S.-Y. 1988. Whether we agree or not: A comparative syntax of English and
Japanese. In Japanese Syntax, W. Poser (ed.), 103-43. Stanford: Center for the
Study of Language and Information.
Laenzlinger, C. 1996. Adverb syntax and phrase structure. In Configurations. Essays
on Structure and Interpretation, A.-M. Di Sciullo (ed.), 99-127. Somerville:
Cascadilla Press.
Laenzlinger, C. 1998. Comparative Studies in Word Order Variation. Adverbs, Pro-
nouns, and Clause Structure in Romance and Germanic. Amsterdam: Benjamins.
Laenzlinger, C. 2000. More on adverb syntax and phrase structure. In Adverbs and
Adjunction, A. Alexiadou and P. Svenonius (eds.), 103-32. Potsdam: Universität
Potsdam.
Laenzlinger, C. 2002. A feature-based theory of adverb syntax. Generative Grammar
in Geneva 3: 67-105.
Laenzlinger, C. 2004. A feature-based theory of adverb syntax. In Adverbials. The
Interplay between Meaning, Context and Syntactic Structure, J.R. Austin, St.
Engelberg and G. Rauh (eds.), 205-52. Amsterdam: Benjamins.
Laka, I. 1990. Negation in Syntax: On the Nature of Functional Categories and Projec-
tions. PhD Dissertation, MIT.
Larson, R. 1985. Bare-NP adverbs. Linguistic Inquiry 16(4): 595-621.
REFERENCES 423
Larson, R. 1987. Missing prepositions and the analysis of English free relative clauses.
Linguistic Inquiry 18(2): 239-66.
Larson, R. 1988. On the double object construction. Linguistic Inquiry 19(3): 335-91.
Larson, R. 1989. Light predicate raising. Lexicon Project Working Papers 27.
Cambridge, MA: MIT.
Larson, R. 1990. Double objects revisited: reply to Jackendoff. Linguistic Inquiry
21(4): 589-632.
Lasnik, H. 1972. Analyses of Negation in English. PhD Dissertation, MIT.
Lasnik, H. 1995. LAST RESORT and ATTRACT F. Proceedings of the Sixth Annual
Meeting of the Formal Linguistics Society of Mid-America: 62-81.
Lasnik, H. 1999a. Minimalist Analysis. Oxford: Blackwell.
Lasnik, H. 1999b. On feature strength. Linguistic Inquiry 30(2): 197-217.
Lasnik, H. 2002. Feature movement or agreement at a distance? In Dimensions of
Movement, A. Alexiadou, E. Anagnostopoulou, S. Barbiers and H.-M. Gaertner
(eds.), 189-208. Amsterdam: Benjamins.
Lebeaux, D. 1988. Language Acquisition and the Form of the Grammar. PhD Disser-
tation, University of Massachusetts.
Lewis, D. 1975. Adverbs of quantification. In Formal Semantics of Natural Language,
E.L. Keenan (ed.), 3-15. Cambridge: Cambridge University Press.
Lieber, R. 1980. On the Organization of the Lexicon. PhD Dissertation, MIT.
Lonzi, L. 1990. Which adverbs in Spec, VP? Rivista di Grammatica Generativa 15:
141-60.
Lonzi, L. 1991. Il Sintagma Avverbiale. In Grande Grammatica Italiana Consul-
tazione II, L. Renzi and G. Salvi (eds.), 341-412. Turin: Rosenberg & Sellier.
Mätzner, E. 31880. Englische Grammatik. Berlin: Weidmannsche Buchhandlung.
May, R. 1985. Logical Form. Its Structure and Derivation. Cambridge, MA: The MIT
Press.
McCawley, J.D. 1968. The role of semantics in a grammar. In Universals in Linguistic
Theory, E. Bach and R. Harms (eds.), 124–169. New York: Holt, Rinehart &
Winston.
McCawley, J.D. 1983. The syntax of some English adverbs. In Papers from the Nine-
teenth Regional Meeting of the Chicago Linguistic Society, A. Chukerman, M.
Marks and J.F. Richardson (eds.), 263-82. Chicago: Chicago Linguistic Society.
McCawley, J.D. 1988a. The Syntactic Phenomena of English. Vols. I & II. Chicago:
The University of Chicago Press.
McCawley, J.D. 1988b. Adverbial NPs: bare or clad in see-through garb? Language
64(3): 583-90.
McCawley, J.D. 1996. The focus and scope of only. In Discourse and Meaning:
Papers in Honor of Eva Hajicová, B.H. Partee and P. Sgall (eds.), 171-193.
Amsterdam: Benjamins.
McConnell-Ginet, S. 1982. Adverbs and Logical Form. Language 58(1): 144-84.
McNally, L. 1992. VP coordination and the VP-Internal Subject Hypothesis. Linguis-
tic Inquiry 23(2): 336-41.
424 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Pittner, K. 2000a. Verschiedene Arten der Art und Weise. Zu ihrer Positionierung im
Deutschen und Englischen. Linguistik Online 6(2) At http://www.linguistik-
online.de/2_00/pittner.html.
Pittner, K. 2000b. Position and interpretation of adjuncts: process, event and wieder
'again'. In Approaching the Grammar of Adjuncts, C. Fabricius-Hansen, E. Lang
and C. Maienborn (eds.), 203-16. Berlin: ZAS.
Pittner, K. 2001. Between Inflection and Derivation: Adverbial Suffixes in English
and German. Paper presented at 12. Wuppertaler Linguistisches Kolloquium, 16.–
17.11.2001.
Pittner, K. 2004. Where syntax and semantics meet: Adverbial positions in the Ger-
man middle field. In Adverbials. The Interplay between Meaning, Context and
Syntactic Structure, J.R. Austin, St. Engelberg and G. Rauh (eds.), 253-87.
Amsterdam: Benjamins.
Poletto, C. 2002. The left-periphery of V2 Rhaetoromance dialects: a new view on V2
and V3. In Syntactic Microvariation, S. Barbiers, L. Cornips and S. van der Kleij
(eds.), 214-42. Amsterdam: Meertens Institute Electronic Publications in
Linguistics (MIEPiL). At: http://www.meertens.knaw.nl/books/synmic/index.
html.
Pollock, J.-Y. 1989. Verb movement, universal grammar and the structure of IP. Lin-
guistic Inquiry 20(3): 365-424.
Postal, P. 1974. On Raising: One Rule of English Grammar and Its Traditional
Implications. Cambridge, MA: The MIT Press.
Poutsma, H. 1926. A Grammar of Late Modern English. Groningen: P. Noordhoff.
Priestly, J. 1761. Rudiments of English Grammar. In 1967-69 English Linguistics 210
(facsimile reprints), R.C. Alston, (ed.). Menston, Scolar Press.
Pustejovsky, J. 1988. The Geometry of Events. [Lexicon Project Working Papers 24].
Cambridge, MA: MIT, The Center for Cognitive Science.
Pustejovsky, J. 1992. The syntax of event structure. In Lexical and Conceptual
Semantics, B. Levin and S. Pinker (eds.), 47-82. Oxford: Blackwell.
Quirk, R. and S. Greenbaum 161986. A University Grammar of English. London:
Longman.
Quirk, R., S. Greenbaum, G. Leech and J. Svartvik 91991. A Comprehensive Grammar
of the English Language. New York: Longman.
Rackowski, A. and L. Travis 2001. V-initial languages: X or XP movement and adver-
bial placement. In The Syntax of Verb Initial Languages, A. Carnie and E.
Guilfoyle (eds.), 117-41. Oxford: Oxford University Press.
Radford, A. (no date). NP Shells. Ms., University of Essex. At: http://privatewww.
essex.ac.uk/~radford/PapersPublications/npshells.htm.
Radford, A. 1988. Transformational Grammar. Cambridge: Cambridge University
Press.
Radford, A. 1997. Syntactic Theory and the Structure of English. A Minimalist
Approach. Cambridge: Cambridge University Press.
426 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Rizzi, L. 1999. On the Position of "Int(errogative)" in the Left Periphery of the Clause.
Ms., Università di Siena.
Rizzi, L. 2002. Locality and left periphery. To appear in Structures and Beyond. The
Cartography of Syntactic Structures. Vol. III, A. Belletti (ed.). Oxford: Oxford
University Press.
Roberts, I. 1985. Agreement parameters and the development of English modal auxil-
iaries. Natural Language and Linguistic Theory 3(1): 21-58.
Roberts, I. 1987. Representation of Implicit and Dethematized Subjects. Dordrecht:
Foris.
Roberts, I. 1993. Verbs and Diachronic Syntax. A Comparative History of English and
French. Dordrecht: Kluwer.
Rochemont, M. 1986. Focus in Generative Grammar. Amsterdam: Benjamins.
Rochemont, M.S. and P.W. Culicover 1990. English Focus Constructions and the
Theory of Grammar. Cambridge: Cambridge University Press.
Roeper, T. 1987. Implicit arguments and the head-complement relation. Linguistic
Inquiry 18(2): 267-310.
Rohdenburg, G. 2003. Cognitive complexity and horror aequi as factors determining
the use of interrogative clause linkers in English. In Determinants of Grammati-
cal Variation in English, G. Rohdenburg and B. Mondorf (eds.), 205-249. Berlin:
Mouton de Gruyter.
Rooryck, J. 1993. Prepositions and Minimalist Case-marking. Ms., Leiden University.
Rooth, M. 1999. Association with focus or presupposition? In Focus. Linguistic, Cog-
nitive, and Computational Perspectives, P. Bosch and R. van der Sandt (eds.),
232-44. Cambridge: Cambridge University Press.
Rosch, E. 1978. Principles of categorization. In Cognition and Categorization, E.
Rosch and B.B. Lloyd (eds.), 27-48. Hillsdale, NJ: Erlbaum.
Ross, J. 1967. Excerpts from CONSTRAINTS ON VARIABLES IN SYNTAX*. In On Noam
Chomsky: Critical Essays, G. Harman (ed.), 165-201. New York: Doubleday.
Rothstein, S. 1985. The Syntactic Forms of Predication. PhD Dissertation, MIT.
Rothstein, S. 1991. Heads, projections, and category determination. In Views on
Phrase Structure, K. Leffel and D. Bouchard (eds.), 97-112. Dordrecht: Kluwer.
Rowlett, P. 1995. Jespersen, negative concord and A'-Binding. In Negation and
Polarity, D. Forget, P. Hirschbühler, F. Martineau and M.-L. Rivero (eds.), 323-
39. Amsterdam: Benjamins.
Rudanko, J. 1980. Towards a description of negatively conditioned operator inversion
in English. English Studies 58: 348-59.
Runner, J. 1995a. Noun Phrase Licensing and Interpretation. PhD Dissertation,
University of Massachusetts.
Runner, J. 1995b. Overt and LF object positions in English. Proceedings of the 6th
Annual Meeting of the Formal Linguistics Society of Mid-America, 35-46.
Runner, J. 2000. The external object hypothesis and the case of object expletives. In
University of Rochester Working Papers in the Language Sciences 1, K.M.
428 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
Stroik, T. 1992. On the distribution of temporal and locative NP adverbials. The Lin-
guistic Review 9(3): 267-84.
Sugioka, Y. and R. Lehr 1983. Adverbial –ly as an inflectional affix. In Papers from
the Parasession on the Interplay of Phonology, Morphology and Syntax, J.F.
Richardson, M. Marks and A. Chukerman (eds.), 293-300. Chicago: Chicago
Linguistic Society.
Svenonius, P. 2002. Subject positions and the placement of adverbials. In Subjects,
Expletives, and the EPP, P. Svenonius (ed.), 201-42. Oxford: Oxford University
Press.
Swan, T. 1988a. Sentence Adverbials in English. Oslo: Novus.
Swan, T. 1988b. The development of sentence adverbs in English. Studia Linguistica
42(1): 1-17.
Sweet, H. 1891 (1968). A New English Grammar. Logical and Historical, Part I.
Oxford: Clarendon Press.
Taglicht, J. 1984. Message and Emphasis. On Focus and Scope in English. London:
Longman.
Tancredi, C. 1992. Deletion and Logical Form. PhD Dissertation, MIT.
Tenny, C. 1987. Grammaticalizing Aspect and Affectedness. PhD Dissertation, MIT.
Thomason, R. and R. Stalnaker 1973. A semantic theory of adverbs. Linguistic Inquiry
4(2): 195-220.
Travis, L. 1991. Derived Objects, Inner Aspect, and the Structure of VP. Paper pre-
sented at NELS 22, Newark, Delaware.
Travis, L. 1988. The syntax of adverbs. McGill Working Papers in Linguistics –
Special Issue on Comparative Germanic Syntax: 280-310.
van Gelderen, E. 2001. The force of ForceP. Southwest Journal of Linguistics 20(2):
107-20.
van Riemsdijk, H. 1978. A Case Study in Syntactic Markedness: The Binding Nature
of Prepositional Phrases. Dordrecht: Foris
van Riemsdijk, H. 1983. The case of German adjectives. In Linguistic Categories:
Auxiliaries and Related Puzzles. Vol. I, F. Heny and B. Richards (eds.), 223-52.
Dordrecht: Reidel.
van Riemsdijk, H. 1998. Categorial feature magnetism: The endocentricity and dis-
tribution of projections. Journal of Comparative Germanic Linguistics 2(1): 1-48.
Vikner, S. 1994a. Scandinavian object shift and West Germanic scrambling. In Studies
on Scrambling: Movement and Non-Movement Approaches to Free Word-Order
Phenomena, N. Corver and H. van Riemsdijk (eds.), 487-517. Berlin: Mouton de
Gruyter.
Vikner, S. 1994b. Finite verb movement in Scandinavian embedded clauses. In Verb
Movement, D. Lightfoot and N. Hornstein (eds.), 117-47. Cambridge: Cambridge
University Press.
Vikner, S. 1995. Verb Movement and Expletive Subjects in the Germanic Languages.
Oxford: Oxford University Press.
430 ADVERB LICENSING AND CLAUSE STRUCTURE IN ENGLISH
von Stechow, A. 1996. The different readings of wieder 'again': a structural account.
Journal of Semantics 13(2): 87-138.
Warner, A.R. 1993. English Auxiliaries. Structure and History. Cambridge:
Cambridge University Press.
Watanabe, A. 1992. Larsonian CP recursion, factive complements and selection. Pro-
ceedings of the North East Linguistic Society 23: 523-37.
Wilder, C. 1999. Review of "Christopher Laenzlinger, Comparative studies in word
order variation: adverbs, pronouns and clause structure in Romance and Ger-
manic". Studies in Language 23(3): 684-99.
Williams, E. 1981. On the notions 'lexically related' and 'head of a word'. Linguistic
Inquiry 12(2): 245-74.
Williams, E. 1982. Another argument that passive is transformational. Linguistic
Inquiry 13(1): 160-63.
Williams, E. 1984. Grammatical relations. Linguistic Inquiry 15(4): 639-73.
Williams, E. 1986. A reinterpretation of the functions of LF. Linguistic Inquiry 17(2):
262-299.
Williams, E. 1987a. NP trace in theta theory. Linguistics and Philosophy 10(4): 433-
47.
Williams, E. 1987b. Implicit arguments, the binding theory and control. Natural Lan-
guage and Linguistic Theory 5(2): 151-80.
Williams, E. 1989. The anaphoric nature of Θ-Roles. Linguistic Inquiry 20(3): 425-56.
Williams, E. 1994. Thematic Structure in Syntax. Cambridge, MA: The MIT Press.
Woolford, E. 1991. VP-internal subjects in VSO and nonconfigurational languages.
Linguistic Inquiry 22(3): 503-40.
Zagona, K. 1988. Verb Phrase Syntax. Dordrecht: Foris.
Zagona, K. 1994. Perfectivity and temporal arguments. In Issues and Theory in
Romance Linguistics, M.L. Mazzola (ed.), 523-46. Washington, D. C.: George-
town University Press.
Zandvoort, R.W. 1957. A Handbook of English Grammar. London: Longman.
Zanuttini, R. 1996. On the relevance of tense for sentential negation. In Parameters
and Functional Heads, A. Belletti and L. Rizzi (eds.), 181-207. Oxford: Oxford
University Press.
Zeller, J. 1994. Die Syntax des Tempus. Zur strukturellen Repräsentation Temporaler
Ausdrücke. Opladen: Westdeutscher Verlag.
Zubizarreta, M.L. 1982. On the Relationship of the Lexicon to Syntax. PhD Disserta-
tion, MIT.
Zubizarreta, M.L. 1998. Prosody, Focus, and Word Order. Cambridge, MA: The MIT
Press.
Zwarts, J. 1992. X'-Syntax X'-Semantics. On the Interpretation of Functional and
Lexical Heads. Utrecht: OTS Dissertation Series.
Zwarts, J. 1997. Lexical and functional properties of prepositions. In Lexikalische und
Grammatische Eigenschaften Präpositionaler Elemente, D. Haumann and St.
Schierholz (eds.), 1-18. Tübingen: Niemeyer.
REFERENCES 431
Other sources:
British National Corpus (BNC), Oxford: Oxford University Computing Services, 13
Banbury Road, Oxford. (1995). At: http://sara.natcorp. ox.ac.uk/lookup.html.
Collins Cobuild Corpus Concordance Sampler (CC). At: http://cobuild.collins.co.uk/
form.html.
Steinbeck, John 1962. Travels with Charley. In Search of America. New York, NY:
Viking Penguin.
INDEX
105 Haumann, Dagmar: Adverb Licensing and Clause Structure in English. 2007. ix,€435€pp.
104 Jeong, Youngmi: Applicatives. Structure and interpretation from a minimalist perspective. 2007.
viii,€140€pp.
103 Wurff, Wim van der (ed.): Imperative Clauses in Generative Grammar. Studies in honour of Frits
Beukema. viii,€339€pp.€+€index. Expected May 2007
102 Bayer, Josef, Tanmoy Bhattacharya and M.T. Hany Babu (eds.): Linguistic Theory and South
Asian Languages. Essays in honour of K. A. Jayaseelan. x,€282€pp.€+€index. Expected May 2007
101 Karimi, Simin, Vida Samiian and Wendy K. Wilkins (eds.): Phrasal and Clausal Architecture.
Syntactic derivation and interpretation. In honor of Joseph E. Emonds. 2007. vi,€424€pp.
100 Schwabe, Kerstin and Susanne Winkler (eds.): On Information Structure, Meaning and Form.
Generalizations across languages. x,€557€pp.€+€index. Expected March 2007
99 Martínez-Gil, Fernando and Sonia Colina (eds.): Optimality-Theoretic Studies in Spanish
Phonology. vi,€548€pp.€+€index. Expected March 2007
98 Pires, Acrisio: The Minimalist Syntax of Defective Domains. Gerunds and infinitives. 2006. xiv,€188€pp.
97 Hartmann, Jutta M. and László Molnárfi (eds.): Comparative Studies in Germanic Syntax. From
Afrikaans to Zurich German. 2006. vi,€332€pp.
96 Lyngfelt, Benjamin and Torgrim Solstad (eds.): Demoting the Agent. Passive, middle and other
voice phenomena. 2006. x,€333€pp.
95 Vogeleer, Svetlana and Liliane Tasmowski (eds.): Non-definiteness and Plurality. 2006. vi,€358€pp.
94 Arche, María J.: Individuals in Time. Tense, aspect and the individual/stage distinction. 2006. xiv,€281€pp.
93 Progovac, Ljiljana, Kate Paesani, Eugenia Casielles and Ellen Barton (eds.): The Syntax of
Nonsententials. Multidisciplinary perspectives. 2006. x,€372€pp.
92 Boeckx, Cedric (ed.): Agreement Systems. 2006. ix,€346€pp.
91 Boeckx, Cedric (ed.): Minimalist Essays. 2006. xvi,€399€pp.
90 Dalmi, Gréte: The Role of Agreement in Non-Finite Predication. 2005. xvi,€222€pp.
89 Velde, John R. te: Deriving Coordinate Symmetries. A phase-based approach integrating Select, Merge,
Copy and Match. 2006. x,€385€pp.
88 Mohr, Sabine: Clausal Architecture and Subject Positions. Impersonal constructions in the Germanic
languages. 2005. viii,€207€pp.
87 Julien, Marit: Nominal Phrases from a Scandinavian Perspective. 2005. xvi,€348€pp.
86 Costa, João and Maria Cristina Figueiredo Silva (eds.): Studies on Agreement. 2006. vi,€285€pp.
85 Mikkelsen, Line: Copular Clauses. Specification, predication and equation. 2005. viii,€210€pp.
84 Pafel, Jürgen: Quantifier Scope in German. 2006. xvi,€312€pp.
83 Schweikert, Walter: The Order of Prepositional Phrases in the Structure of the Clause. 2005.
xii,€338€pp.
82 Quinn, Heidi: The Distribution of Pronoun Case Forms in English. 2005. xii,€409€pp.
81 Fuss, Eric: The Rise of Agreement. A formal approach to the syntax and grammaticalization of verbal
inflection. 2005. xii,€336€pp.
80 Burkhardt, Petra: The Syntax–Discourse Interface. Representing and interpreting dependency. 2005.
xii,€259€pp.
79 Schmid, Tanja: Infinitival Syntax. Infinitivus Pro Participio as a repair strategy. 2005. xiv,€251€pp.
78 Dikken, Marcel den and Christina M. Tortora (eds.): The Function of Function Words and
Functional Categories. 2005. vii,€292€pp.
77 Öztürk, Balkız: Case, Referentiality and Phrase Structure. 2005. x,€268€pp.
76 Stavrou, Melita and Arhonto Terzi (eds.): Advances in Greek Generative Syntax. In honor of Dimitra
Theophanopoulou-Kontou. 2005. viii,€366€pp.
75 Di Sciullo, Anna Maria (ed.): UG and External Systems. Language, brain and computation. 2005.
xviii,€398€pp.
74 Heggie, Lorie and Francisco Ordóñez (eds.): Clitic and Affix Combinations. Theoretical perspectives.
2005. viii,€390€pp.
73 Carnie, Andrew, Heidi Harley and Sheila Ann Dooley (eds.): Verb First. On the syntax of verb-
initial languages. 2005. xiv,€434€pp.
72 Fuss, Eric and Carola Trips (eds.): Diachronic Clues to Synchronic Grammar. 2004. viii,€228€pp.
71 Gelderen, Elly van: Grammaticalization as Economy. 2004. xvi,€320€pp.
70 Austin, Jennifer R., Stefan Engelberg and Gisa Rauh (eds.): Adverbials. The interplay between
meaning, context, and syntactic structure. 2004. x,€346€pp.
69 Kiss, Katalin É. and Henk van Riemsdijk (eds.): Verb Clusters. A study of Hungarian, German and
Dutch. 2004. vi,€514€pp.
68 Breul, Carsten: Focus Structure in Generative Grammar. An integrated syntactic, semantic and
intonational approach. 2004. x,€432€pp.
67 Mišeska Tomić, Olga (ed.): Balkan Syntax and Semantics. 2004. xvi,€499€pp.
66 Grohmann, Kleanthes K.: Prolific Domains. On the Anti-Locality of movement dependencies. 2003.
xvi,€372€pp.
65 Manninen, Satu Helena: Small Phrase Layers. A study of Finnish Manner Adverbials. 2003. xii,€275€pp.
64 Boeckx, Cedric and Kleanthes K. Grohmann (eds.): Multiple Wh-Fronting. 2003. x,€292€pp.
63 Boeckx, Cedric: Islands and Chains. Resumption as stranding. 2003. xii,€224€pp.
62 Carnie, Andrew, Heidi Harley and MaryAnn Willie (eds.): Formal Approaches to Function in
Grammar. In honor of Eloise Jelinek. 2003. xii,€378€pp.
61 Schwabe, Kerstin and Susanne Winkler (eds.): The Interfaces. Deriving and interpreting omitted
structures. 2003. vi,€403€pp.
60 Trips, Carola: From OV to VO in Early Middle English. 2002. xiv,€359€pp.
59 Dehé, Nicole: Particle Verbs in English. Syntax, information structure and intonation. 2002. xii,€305€pp.
58 Di Sciullo, Anna Maria (ed.): Asymmetry in Grammar. Volume 2: Morphology, phonology, acquisition.
2003. vi,€309€pp.
57 Di Sciullo, Anna Maria (ed.): Asymmetry in Grammar. Volume 1: Syntax and semantics. 2003.
vi,€405€pp.
56 Coene, Martine and Yves D’hulst (eds.): From NP to DP. Volume 2: The expression of possession in
noun phrases. 2003. x,€295€pp.
55 Coene, Martine and Yves D’hulst (eds.): From NP to DP. Volume 1: The syntax and semantics of noun
phrases. 2003. vi,€362€pp.
54 Baptista, Marlyse: The Syntax of Cape Verdean Creole. The Sotavento varieties. 2003.
xxii,€294€pp.€€(incl.€CD-rom).
53 Zwart, C. Jan-Wouter and Werner Abraham (eds.): Studies in Comparative Germanic Syntax.
Proceedings from the 15th Workshop on Comparative Germanic Syntax (Groningen, May 26–27, 2000).
2002. xiv,€407€pp.
52 Simon, Horst J. and Heike Wiese (eds.): Pronouns – Grammar and Representation. 2002. xii,€294€pp.
51 Gerlach, Birgit: Clitics between Syntax and Lexicon. 2002. xii,€282€pp.
50 Steinbach, Markus: Middle Voice. A comparative study in the syntax-semantics interface of German.
2002. xii,€340€pp.
49 Alexiadou, Artemis (ed.): Theoretical Approaches to Universals. 2002. viii,€319€pp.
48 Alexiadou, Artemis, Elena Anagnostopoulou, Sjef Barbiers and Hans-Martin Gärtner
(eds.): Dimensions of Movement. From features to remnants. 2002. vi,€345€pp.
47 Barbiers, Sjef, Frits Beukema and Wim van der Wurff (eds.): Modality and its Interaction with the
Verbal System. 2002. x,€290€pp.
46 Panagiotidis, Phoevos: Pronouns, Clitics and Empty Nouns. ‘Pronominality’ and licensing in syntax.
2002. x,€214€pp.
45 Abraham, Werner and C. Jan-Wouter Zwart (eds.): Issues in Formal German(ic) Typology. 2002.
xviii,€336€pp.
44 Taylan, Eser Erguvanlı (ed.): The Verb in Turkish. 2002. xviii,€267€pp.
43 Featherston, Sam: Empty Categories in Sentence Processing. 2001. xvi,€279€pp.
42 Alexiadou, Artemis: Functional Structure in Nominals. Nominalization and ergativity. 2001. x,€233€pp.
41 Zeller, Jochen: Particle Verbs and Local Domains. 2001. xii,€325€pp.
40 Hoeksema, Jack, Hotze Rullmann, Víctor Sánchez-Valencia and Ton van der Wouden
(eds.): Perspectives on Negation and Polarity Items. 2001. xii,€368€pp.
39 Gelderen, Elly van: A History of English Reflexive Pronouns. Person, Self, and Interpretability. 2000.
xiv,€279€pp.
38 Meinunger, André: Syntactic Aspects of Topic and Comment. 2000. xii,€247€pp.
37 Lutz, Uli, Gereon Müller and Arnim von Stechow (eds.): Wh-Scope Marking. 2000. vi,€483€pp.
36 Gerlach, Birgit and Janet Grijzenhout (eds.): Clitics in Phonology, Morphology and Syntax. 2001.
xii,€441€pp.
35 Hróarsdóttir, Thorbjörg: Word Order Change in Icelandic. From OV to VO. 2001. xiv,€385€pp.
34 Reuland, Eric (ed.): Arguments and Case. Explaining Burzio’s Generalization. 2000. xii,€255€pp.
33 Puskás, Genoveva: Word Order in Hungarian. The syntax of Ā-positions. 2000. xvi,€398€pp.
32 Alexiadou, Artemis, Paul Law, André Meinunger and Chris Wilder (eds.): The Syntax of
Relative Clauses. 2000. vi,€397€pp.
31 Svenonius, Peter (ed.): The Derivation of VO and OV. 2000. vi,€372€pp.
30 Beukema, Frits and Marcel den Dikken (eds.): Clitic Phenomena in European Languages. 2000.
x,€324€pp.
29 Miyamoto, Tadao: The Light Verb Construction in Japanese. The role of the verbal noun. 2000.
xiv,€232€pp.
28 Hermans, Ben and Marc van Oostendorp (eds.): The Derivational Residue in Phonological
Optimality Theory. 2000. viii,€322€pp.
27 Růžička, Rudolf: Control in Grammar and Pragmatics. A cross-linguistic study. 1999. x,€206€pp.
26 Ackema, Peter: Issues in Morphosyntax. 1999. viii,€310€pp.
25 Felser, Claudia: Verbal Complement Clauses. A minimalist study of direct perception constructions.
1999. xiv,€278€pp.
24 Rebuschi, Georges and Laurice Tuller (eds.): The Grammar of Focus. 1999. vi,€366€pp.
23 Giannakidou, Anastasia: Polarity Sensitivity as (Non)Veridical Dependency. 1998. xvi,€282€pp.
22 Alexiadou, Artemis and Chris Wilder (eds.): Possessors, Predicates and Movement in the
Determiner Phrase. 1998. vi,€388€pp.
21 Klein, Henny: Adverbs of Degree in Dutch and Related Languages. 1998. x,€232€pp.
20 Laenzlinger, Christopher: Comparative Studies in Word Order Variation. Adverbs, pronouns, and
clause structure in Romance and Germanic. 1998. x,€371€pp.
19 Josefsson, Gunlög: Minimal Words in a Minimal Syntax. Word formation in Swedish. 1998. ix,€199€pp.
18 Alexiadou, Artemis: Adverb Placement. A case study in antisymmetric syntax. 1997. x,€256€pp.
17 Beermann, Dorothee A., David LeBlanc and Henk van Riemsdijk (eds.): Rightward Movement.
1997. vi,€410€pp.
16 Liu, Feng-Hsi: Scope and Specificity. 1997. viii,€187€pp.
15 Rohrbacher, Bernhard Wolfgang: Morphology-Driven Syntax. A theory of V to I raising and pro-
drop. 1999. viii,€296€pp.
14 Anagnostopoulou, Elena, Henk van Riemsdijk and Frans Zwarts (eds.): Materials on Left
Dislocation. 1997. viii,€349€pp.
13 Alexiadou, Artemis and T. Alan Hall (eds.): Studies on Universal Grammar and Typological
Variation. 1997. viii,€252€pp.
12 Abraham, Werner, Samuel David Epstein, Höskuldur Thráinsson and C. Jan-Wouter Zwart
(eds.): Minimal Ideas. Syntactic studies in the minimalist framework. 1996. xii,€364€pp.
11 Lutz, Uli and Jürgen Pafel (eds.): On Extraction and Extraposition in German. 1996. xii,€315€pp.
10 Cinque, Guglielmo and Giuliana Giusti (eds.): Advances in Roumanian Linguistics. 1995. xi,€172€pp.
9 Gelderen, Elly van: The Rise of Functional Categories. 1993. x,€224€pp.
8 Fanselow, Gisbert (ed.): The Parametrization of Universal Grammar. 1993. xvii,€232€pp.
7 Åfarlí, Tor A.: The Syntax of Norwegian Passive Constructions. 1992. xii,€177€pp.
6 Bhatt, Christa, Elisabeth Löbel and Claudia Maria Schmidt (eds.): Syntactic Phrase Structure
Phenomena in Noun Phrases and Sentences. 1989. ix,€187€pp.
5 Grewendorf, Günther and Wolfgang Sternefeld (eds.): Scrambling and Barriers. 1990. vi,€442€pp.
4 Abraham, Werner and Sjaak De Meij (eds.): Topic, Focus and Configurationality. Papers from the 6th
Groningen Grammar Talks, Groningen, 1984. 1986. v,€349€pp.
3 Abraham, Werner (ed.): On the Formal Syntax of the Westgermania. Papers from the 3rd Groningen
Grammar Talks (3e Groninger Grammatikgespräche), Groningen, January 1981. 1983. vi,€242€pp.