Académique Documents
Professionnel Documents
Culture Documents
Nicholas Sobin
English echo questions (EQs) such as (1) are rarely if ever discussed in recent minimalist-oriented
syntax.1
(1) Mary saw what?
When they are considered, they are usually mentioned fleetingly and in the context of pedagogi-
cally oriented discussions such as Adger 2003:352353 and Carnie 2007:342343. The rhetorical
intent in such works (which I do not fault here) is to put them aside, since they appear to counter-
exemplify general claims about question formation such as the obligatoriness of wh-movement.
However, there is much more to be said about them. They give the appearance of being far more
contrary than the works just cited suggest, and consequently they are of great interest and relevance
to analyses of question formation since they are clearly in the realm of automatic and un-
tutored knowledge, just the sort of linguistic knowledge that generative grammar has had the
aim of explaining since its inception.
The analysis to follow is oriented along the general lines of works such as Chomsky 2000
(as are Adger 2003 and Carnie 2007) and Adger and Ramchand 2005. Adger and Carnie each
suggest that EQs somehow involve a different C from that employed in non-EQ interrogatives.
My sincerest thanks to two anonymous reviewers, whose insightful comments have contributed substantially to the
present form of this article. My thanks also to those who offered substantial input to much earlier incarnations of the
article, including Jon Amastae, Noam Chomsky, Chris Collins, and Irene Heim. Any remaining errors are mine alone.
This article is dedicated to C. L. Baker, my dissertation supervisor at The University of Texas, whose pioneering
work on interrogatives and unselective binding has been fundamental to much subsequent work, including the present
piece.
1
Ginzburg and Sag (2001) offer an account of such questions within the framework of Head-Driven Phrase Structure
Grammar/Construction Grammar. Their work also deals with some of the phenomena analyzed here. Since the aim of
this article is simply to offer an analysis within Minimalist Program (MP) assumptions, a point-by-point comparison is
not offered here. However, such a comparison may prove of considerable interest as a topic for future work.
(For earlier work along similar lines, see Sobin 1990.) This is the line I will pursue here, though
as I will demonstrate, EQ formation is somewhat more complex than these works claim and at
the same time less contrary to non-EQ syntax than appearances would suggest. Section 1 recounts
key syntactic features of EQs that appear contrary to normal question formation. Section 2
considers Adgers (2003) characterization of EQs and certain evidence against it. Section 3 deals
with the derivation of non-EQ interrogatives, including the assignment of scope to wh-phrases
(also a key factor in analyzing EQs). Section 4 offers an analysis of EQs that accounts in detail
for their key syntactic features. Section 5 summarizes the analyses and offers speculations on the
composition of other Cs.
1.1 Wh-in-Situ
Though wh-movement has frequently been considered obligatory in English, syntactic EQs typi-
cally manifest wh-in-situ. (In offering examples, I will use U to designate an utterance and E to
designate an EQ response to it.) The most widely recognized instance of non-wh-movement is
found in examples like (2).
(2) a. U: Mary had tea with Cleopatra.
b. E: Mary had tea with who? ( a syntactic EQ)
Though it is also possible to say (2c), this is an example of a pseudo EQ (a syntactically normal
question). The limited EQ-character of pseudo EQs will become more apparent in later examples.
(2) c. E: Who did Mary have tea with? ( a pseudo EQ)
The wh-in-situ characteristic of syntactic EQs extends to other more dramatic cases such as
(3) and (4).
(3) a. U: Did Mary have tea with Cleopatra?
b. E: Did Mary have tea with who? ( a syntactic EQ)
c. E: *Who did Mary have tea with? ( a pseudo EQ)
d. E: *Mary had tea with who?
(4) a. U: What did Dracula drink at Marys party?
b. E: What did who drink at Marys party? ( a syntactic EQ)
c. E: *Who drank what at Marys party? ( a pseudo EQ)
REMARKS AND REPLIES 133
(3a) is a yes/no question (YNQ) and may only be echoed with a syntactic EQ that preserves the
YNQ character of (3a), namely, the syntactic EQ in (3b). An EQ with declarative syntax such
as (3d) fails. Also, a pseudo EQ response such as (3c) fails; a pseudo EQ is only appropriate for
echoing a declarative, as in (2). Similar restrictions hold in (4). Here, the syntactic EQ (4b)
preserves the wh-interrogative character of the U (a feature of EQ formation that will be made
more explicit below), even at the apparent cost of violating Superiority. And very interestingly,
the normally mandatory observance of Superiority, as manifest in (4c), is not viable for echoing
(4a).
Overall, the rigid wh-in-situ requirement for EQs induces far more apparent syntactic idiosyn-
crasy than seen in (1) and (2), the sort of example that those offering brief discussion or analysis
of EQs normally limit themselves to. But there are yet further complications.
in the wh-marking, as partially wh-marked.2 While such expressions cannot appear in normal
non-EQ interrogatives, either raised to Spec,CP or in situ, as illustrated in (7ab), they may appear
in EQs, as in (8) and (9).
(7) a. *The what did Mary see? (non-EQ)
b. *Who saw the what? (non-EQ)
(8) a. U: Who saw the flying saucer?
b. E: Who saw the what?
(9) a. U: The Martians asked for more spaghetti.
b. E: The who asked for more spaghetti?
2
This is only a provisional characterization of this phenomenon. There is undoubtedly more to discover about it,
but it is worth mentioning here since it intersects with question formation. As far as I can see, nothing in the analysis
of question formation to be offered here would be negatively affected by further details of this phenomenon.
REMARKS AND REPLIES 135
Spec,CP. YNQs are claimed to use the same C as wh-questions (WHQs), with [uwh*] being
satisfied by merger of an expletive wh-operator into Spec,CP. Constructions such as (10), in
which both the wh-phrase and the expletive wh-operator are involved, are claimed to be blocked
semantically since the wh-phrase triggers for which x . . . semantics, and the wh-operator in
turn triggers is it true that p semantics, a conflict resulting in gibberish.
(10) *Did Mary see who? (as a normal question)
3
My thanks to the anonymous reviewers for their insights into this and other possible interpretations and ramifications
of Adgers analysis of EQs.
136 REMARKS AND REPLIES
Another observation that goes against characterizing EQs as in (13) is that EQ wh-phrases
show an animacy distinction (who vs. what).4 This should not be so if an expression rather than
a referent is being asked for, since the normal form for asking about an expression is simply
what.
Another interpretation of an analysis such as Adgers (2003) is that EQs are essentially a
pragmatic phenomenon.5 Thus, one might use the EQ-introduced wh-phrase simply to draw atten-
tion to a portion of the U being echoed for a variety of possible pragmatically determined purposes
ranging from genuine interrogation (for which x . . . semantics), to doubting what someone has
said, to trying to coerce someone into altering an utterance known to be false in a case where
the correct answer is known. However, the problem with using such possible pragmatically driven
purposes to separate EQs as pragmatic and distinct from other (syntactic) non-EQ interroga-
tives is that any interrogative can be put to such alternative pragmatic uses (e.g., Can you pass
the salt?). For purposes of this discussion, I will refer to this phenomenon of shifting away from
the ideal syntactic/semantic function of an interrogative to some other pragmatic purpose as
pragmatic intrusion.6 So, just as we see it at work in a U-EQ sequence like (16ac), we also see
it at work in a sequence like (17ac), involving a U and a non-syntactic-EQ interrogative (marked
Q).
(16) a. U: I fought off ten ninjas.
b. E: You fought off who? (meaning here that the ten ninjas part is not believable)
c. R: OK, I only fought off a Cub Scout.
(17) a. U: I fought off ten ninjas.
b. Q: Who did you fight off? (same meaning as (16b))
c. R: OK, I only fought off a Cub Scout.
Since both syntactic EQs and non-EQ interrogatives are subject to pragmatic intrusion, it is hard
to determine if one clause type is more susceptible to it than the other. But it is noteworthy
that syntactic EQs allow a broader range of interrogative-marked forms than do mainstream
interrogatives, including partially wh-marked ones, and even words with special emphasis as in
(18b).
(18) a. U: What did Dracula drink at Marys party?
b. E: What did DRACULA drink at Marys party?
c. R: Dracula!
d. R: *Dracula drank wine!
e. R: *Wine!
On the face of it, this fact in itself may suggest that EQ-introduced interrogative-marked expres-
sions are subject to a broader pragmatic range, still allowing for the neutral possibility of for
4
Again, my thanks to an anonymous reviewer for pointing this out.
5
My thanks to another anonymous reviewer for noting this possibility.
6
This term is similarly employed by Ginzburg and Sag (2001:264).
REMARKS AND REPLIES 137
question formation first analyzed in Baker 1970 and later discussed in other key works such as
Pesetsky 1987.7 In this discussion, I follow Baker, Pesetsky, and others, who characterize the
scope of a wh-phrase as being derived from the complementizer that it is (somehow) associated
with.
Wh-phrases in simple root WHQs such as (20) have wide scope, or what I have termed
above root scope; a fully informative response to such a WHQ entails giving values for each
wh-phrase.
(20) Who bought what?
When a question like (20) is embedded in a declarative clause as in (21), no such response is
called for, and we can say that these wh-phrases have narrow or embedded scope; their only
role is within the embedded question. That is, they are not objects of inquiry; a question is not
being posed.
(21) I asked who bought what.
As noted and analyzed first by Baker (1970) and later by others, a surprising ambiguity arises in
sentences like (22).
(22) Who knows where Mary bought what?
This question may be answered by (23) or (24).
(23) Bill does.
(24) Bill knows where she bought the soap, Jane knows where she bought the toothpaste,
etc.
That is, what in the embedded clause in (22) may have the embedded scope expected from its
surface position, as shown by the answer in (23), or it may somehow be assigned root scope, as
shown by the answer in (24). Pesetsky (1987) (and many other authors, all following Chomsky
1976 and later works) argues that possible assignment of root scope to embedded what is due to
possible LF movement of what into the root clause. Pesetsky further argues that both movement
(overt movement as well as LF movement) and unselective Baker (1970)style binding are avail-
able processes in the derivation of questions. However, the more recent assumptions of MP ban
LF movement (see (19)), so we are again left with the question of how an in-situ wh-phrase gains
scope either within or beyond its clause of origin. The approach I will develop here relies on
expanding the role of unselective binding. I will discuss two possibilities for doing this. The first
involves establishing in C a fairly traditional unselective binder. The second involves using C
itself more directly as valuing a scope feature on wh-phrases.
7
Works such as Lasnik and Saito 1992 and Epstein 1998 deal with various aspects of scope marking assuming the
availability of an LF component. Here, I will not offer a detailed account of scope marking; rather, I will posit a gross
non-LF mechanism for the sake of sketching a general backdrop for the ensuing analysis of EQs. Many details about
scope marking of wh-phrases in non-EQ interrogatives need to be worked out; however, they do not bear directly on
scope marking in EQs.
REMARKS AND REPLIES 139
3.2.1 A Traditional Unselective Binder It is often assumed that overt movement of a wh-
phrase to Spec,CP is sufficient to establish its scope. But what of the other wh-phrases in a
multiple-wh-phrase interrogative? To begin to address this question, let us consider adjusting the
feature composition of the interrogative C as shown in (25).8
(25) C[Int, Q, uwh*, BWH ]
I will take [Int] on C as an interpretable force feature [Interrogative] (in contrast to [Decl(arative)]).
Partially following Adgers (2003) casting of these features, if [Q] triggers T-to-C movement in
root questions, then it may be claimed to be absent from C in embedded questions, which are
nonetheless interrogative. The feature [uwh*] (or [uwh] associated with EPP) plays the same role
here as in Adgers analysis. In addition to these features, this interrogative C has a binding feature,
really a binding function, [BWH ]. This function is described as in (26) and in the discussion to
follow.9
(26) BWH (XPi ) N Si
[wh]
Question formation involving (26) works as follows. Interrogative C probes for and looks at every
fully wh-marked constituent (DP/PP) in its domain. (See below for further characterizations of
these terms.) Once C has established a probe relation with every wh-phrase, the function [BWH ]
may evaluate (or not evaluate) any of these in-situ and unscoped wh-phrases. Each such evaluation
produces a scope operator Si in C which binds that wh-phrase XPi , giving the wh-phrase its scope
via this binding relation with C. If more than one wh-phrase is evaluated by [BWH ], then the
sequence of Ss produced may be conflated into a single unselective binder, as in (27).
(27) Si , Sj, . . . Sn N Si, j, . . . , n
If a wh-phrase is not immediately evaluated, then a higher interrogative C with [BWH ] may
evaluate it and give it a wider scope. An interrogative C can only satisfy its [uwh*]/EPP feature
by raising a wh-phrase that is scope-bound by [BWH] and that, following the MLC, is the nearest
wh-phrase. Since this raised wh-phrase has been assigned a scope by [BWH ] of C, it cannot raise
further; long wh-movement will apply only to unscoped wh-phrases raised to the Spec,CP
of noninterrogative Cs.10 (In languages such as Ukrainian, where all wh-phrases are raised, perhaps
8
Chomsky (2000), offering a different feature architecture, proposes that C with an uninterpretable feature [Q]
probes and matches a wh-phrase with interpretable [Q] and uninterpretable [wh]. The wh-phrase is active until its uninter-
pretable [wh] is checked and deleted, allowing for the possibility of multiple movements. The possibility of such an
uninterpretable wh-feature on a wh-phrase is pursued here in the feature valuation analysis of interrogatives to be offered
later.
9
Adger and Ramchand (2005) offer a view of what may be construed as scope assignment to items in situ in relative
clause formation in terms of -abstraction. While relative clauses allow only a single item to be relativized, questions
allow multiple items to be questioned. Given the complexities that this fact would entail, in this article I employ the more
classic descriptive mechanism of coindexation binding in the traditional unselective binder analysis.
10
I will assume that declarative Cs bear a gratuitous EPP feature and have no scope assignment function, so that
long movement is facilitated and the wh-phrase involved is still open for scope binding by a higher CWH .
140 REMARKS AND REPLIES
[uwh*] is not checked/deleted/resolved unless all of the wh-phrases it finds are raised.11 For
languages like Chinese that do not exhibit wh-movement, the interrogative C may have [BWH]
and lack [uwh*]/EPP. Though this picture of movement is very unrefined at this point, it is
consistent with the MP view that differences in the feature composition of lexical items account
for or explain crosslinguistic differences in derivation. Of course, many details remain to be
worked out.)
As sketched in section 1.2, the term fully wh-marked constituent (DP/PP) refers to wh-
expressions such as what, which book, with what, and where, expressions in which wh-marking
is either Determiner-related (in a DP) or subsumes an entire argument or adjunct expression (as
with a PP). I will refer to other interrogative-marked expressions such as the what, where the D
head is not apparently subsumed in the wh-marking, as partially wh-marked. Though such expres-
sions may appear in EQs, they cannot appear in normal non-EQ interrogatives, either raised to
Spec,CP or in situ, as illustrated in (28ab).12
(28) a. *The what did Mary see? ( (7a))
b. *Who saw the what? ( (7b))
At least from a descriptive standpoint, it is possible to attribute the nonappearance of partially
wh-marked expressions in non-EQ interrogatives to the complementizer: they are invisible to
C[Int, Q, uwh*, BWH] and hence cannot get a scope.
Next, by the term domain of C used earlier (the domain within which C probes wh-phrases
for possible scope binding), I refer to a region within the c-command domain of C. (I will
abbreviate C[Int, Q, uwh*, BWH ] as CWH .) When the CWH of a root interrogative goes on the
hunt for unscoped wh-phrases in lower clauses, there appears to be a two-TP limit on such a
search. This is indicated by the fact that the only response to (29) is (30a); (30b) seems odd at
best.
(29) [ CP Who CWH1 [ TP says [ CP that [ TP Bill knows [ CP where CWH2 [ TP Mary bought
what . . .
(30) a. Jane does.
b. ?*Jane says that Bill knows where Mary bought the towels, Zelda says that Bill
knows where Mary bought the sheets, . . .
11
See Richards 1998 and later Pesetsky and Torrego 2001 for more detailed analysis and discussion of multiple
wh-movement. In most recent accounts that I am aware of, it is the interrogative complementizer that is viewed as the
trigger for such movement.
12
One reviewer notes possibly problematic expressions like the underlined one in the non-EQ interrogative The host
of which party did Dracula insult?, where the highest D is not subsumed in interrogative marking. To my own ear, this
is pretty marginal. However, assuming that it is possible, this expression contains as the key element dictating its interroga-
tive status the sort of expression that has been characterized above as fully wh-markednamely, the expression which
party. Thus, it appears similar to other expressions containing a fully wh-marked constituent such as With which hand
did Dracula sign the letter?, which can be pied-piped to Spec,CP in a non-EQ interrogative.
REMARKS AND REPLIES 141
Thus, although what of (22) may be bound by the C in either clause, the CWH1 in (29) appears
unable to reach what more than two TPs away, and so (30b) is not a possible response to (29).
Now, returning to the question of why (28ab) are ungrammatical, one possible explanation
is that these sentences violate a semantic interface requirement for interpretability such as (31).
(31) The Scope Requirement
All interrogative-marked constituents must have a scope (Baker 1970, Pesetsky
1987) in the sense of being associated with some interrogative C.
That is, an interrogative-marked constituent is only interpretable if it is scope-bound by a particular
C. Otherwise, it cannot be fully interpreted as something actually being asked about, or not being
asked about; it lacks scope in Bakers and Pesetskys sense. This requirement would hold for all
interrogative-marked expressions, including fully and partially wh-marked constituents. Thus, the
sentences in (28) fail for lack of a scope assignment to their partially wh-marked DPs. They are,
to use the common term, gibberish. This analysis thus establishes binding relations with a
traditional unselective binder present in C very much like the ones discussed and illustrated
in Pesetsky 1987:100.
3.2.2 Unselective Binding through Feature Valuation There is also the possibility of a somewhat
different and perhaps simpler analysis based on the mechanism of feature checking through
valuation (see Chomsky 2000, Adger 2003, Adger and Ramchand 2005). For example, the [uCase]
feature of a DP is argued to be unvalued and in need of a Case value. This is assigned to it by
a c-commanding probe with a Case to assign; the [uCase] feature of the DP is thus valued and
checked in situ. Perhaps an interrogative-marked constituent also bears such a feature that requires
valuation. Following Chomskys claim that wh-phrases bear an uninterpretable feature [wh], what
if this feature is [uwh: ], and it is uninterpretable because it lacks a scope value (again, in Bakers
and Pesetskys sense), that is, a connection to or identification with some C? Lets suppose that
C in WHQs simply has the feature composition in (32).
(32) C[Int, Q, uwh*]
In this proposal, C assigns a scope value to a wh-phrase by assigning its own label as the requisite
value for [uwh: ], effectively binding the wh-phrase to itself. As in the earlier proposal, C probes
every wh-phrase in its domain, and it may value (or not) any of the wh-phrases in its domain.
However, the [uwh*]/EPP feature of interrogative C can only be satisfied by raising to its Spec,CP
a wh-phrase that is both scope-valued and nearest to C (again following the MLC). A wh-
phrase left in situ with an unvalued [uwh: ] may receive a value from a higher C (as discussed
earlier) by being assigned the label of that higher C. Also, as in the earlier proposal, a wh-phrase
raised to a noninterrogative Spec,CP that will undergo long movement must not have a scope
value. Wh-movement cannot raise from Spec,CP a wh-phrase that has been assigned a scope; a
higher interrogative C can only see and hence affect unscoped wh-phrases to which it might give
its label. The different valuation possibilities would explain the scope ambiguity of what observed
by Baker (1970) in sentences like (22) (repeated here), as shown in (33).
142 REMARKS AND REPLIES
13
The proposal that English interrogative C may probe all wh-phrases in its domain is now supported by a couple
of observations. First, this must be true in (e.g.) Slavic languages, which move all wh-phrases, so perhaps it is universal.
Second, the limitation of wh-phrases that move to Spec,CP to fully wh-marked ones also extends to wh-phrases in multiple
WHQs that remain in situ, that is, ones that could have moved to Spec,CP but for Superiority, locality, and so on. If C
is the key factor in limiting movable wh-phrases to fully wh-marked ones, as proposed here, then it is simplest to propose
that C relates to all wh-phrases in its domain and imposes the same limitation in all cases.
REMARKS AND REPLIES 143
14
EQs in fact provide interesting evidence against Rizzis (1997) split CP analysis for English. See Sobin 2009.
144 REMARKS AND REPLIES
As a final observation here, given that the TP elements of a U are not frozen in position in
a corresponding EQ but the CP elements are, we get some interesting evidence that subject wh-
phrases indeed move to Spec,CP. Consider the sentences in (41) and (42).
(41) a. U: Who was spotted by the Martians?
b. E: Who was spotted by who?
c. E: *Who spotted who?
(42) a. U: Who did the Martians spot?
b. E: Who did who spot?
c. E: *Who was spotted by who?
As in (39) and (40), active and passive can fluctuate between U and EQ when the major elements
affected are not in the frozen CP. If one of them is, then given Comp freezing, active/passive
fluctuation should no longer be possible. Since this is what we see in (41) and (42), we are led
to the conclusion that subject wh-phrases are located in Spec,CP.
Having sketched out Comp freezing, we are now in a position to more precisely characterize
the derivation of EQs.
15
Also, the surface form but not the basic content of C may vary somewhat. For instance, a complementizer that
in a U may be echoed with a that , and vice versa, but the basic [wh] character of C must remain intact. Refer to the
earlier works on EQs for further details.
REMARKS AND REPLIES 145
16
The feature [uwh(*)] is not needed on C to obtain for which x . . . semantics, since in Adgers (2003) analysis
(as discussed earlier), both YNQs and WHQs have this feature.
17
A wh-phrase must have a scope to be interpretable, and it may bear either [wh] or [i-m] to meet this requirement.
Either feature will allow the wh-phrase to get a scope from the correct scope-assigning C. A mismatch between this
feature and the type of C will simply result in a crash, since the wh-expression will fail to get a scope.
146 REMARKS AND REPLIES
The indices on the interrogative-marked phrases here are simply those of the interrogative constitu-
ents themselves that [BWH ] of CWH and [BEQ ] of CEQ have created scope binders for. Any
interrogative-marked phrase needs a scope (per (31)). A fully wh-marked phrase may get a scope
by bearing either [wh] or [i-m]. CWH will scope-bind only the former, and CEQ only the latter.
The fact that CEQ always takes the U with its CP frozen as its complement entails that any wh-
phrase that originally had root scope in the U loses it in the EQ, where it is now in a subordinate,
embedded position.
In addition to the scope facts and the ability of EQs to scope-bind all interrogative-marked
constituents, this analysis explains the other seemingly contrary characteristics of EQs described
in section 1. In apparent cases of Superiority violations (as above) or verb movement without
wh-movement, the normal restrictions actually are observed in forming the EQ. Since CWH is
probing and raising only [wh]-bearing phrases, and CEQ is probing and raising only [i-m]-bearing
phrases, at no point is a restriction like Superiority ever actually violated.
4.2.2 Binding through Feature Valuation in Echo Questions Following the alternative view of
scope assignment that involves feature valuation from C, we might say that CEQ has the feature
composition in (47).
(47) CEQ[Int, ui-m]
The feature [Int] indicates that CEQ is interrogative; in its pragmatically unskewed use, it tries to
elicit a WHQ-like response. The uninterpretable feature [ui-m] is what triggers CEQ as a probe
for any interrogative-marked expressions. CEQ does not have an EPP aspect or [Q] and cannot
trigger any raising. As noted earlier, CEQ assigns scope to a broader range of interrogative-marked
expressions than does CWH . All interrogative-marked phrases introduced in the EQ, including
wh-phrases, here bear an unvalued feature, [ui-m: ], which also requires a scope value in order
for the interrogative-marked expression to be interpretable. So, while CWH is limited to valuing
[uwh: ], CEQ values [ui-m: ]. Thus, an EQ like (18b), repeated here, would have roughly the
structure in (48), and one like (4b), also repeated here, would have the structure in (49).
(18) a. U: What did Dracula drink at Marys party?
b. E: What did DRACULA drink at Marys party?
(48) [ CP[ CEQ Int, ui-m] [ CP what[uwh: CWH ] [ CWH did[Int, Q, uwh*]] [ TP DRACULA[ui-m:
CEQ ] drink what at . . .
(4) b. E: What did who drink at Marys party?
(49) [ CP[ CEQ Int, ui-m] [ CP what[uwh: CWH ] [ CWH did[Int, Q, uwh*]] [ TP who[ui-m: CEQ ]
drink what at . . .
All EQ-introduced interrogative-marked expressions must be assigned a scope value by
CEQ , though this need not be stipulated. If scope assignment is optional for CEQ as it is for CWH ,
then the only EQs that will not crash are those in which all EQ-introduced interrogative-marked
expressions are scope-valued by CEQ .
REMARKS AND REPLIES 147
Finally, English EQs must be root clauses. They cannot be embedded, as seen in (50).18
(50) *Bill asked what did who drink at Marys party?
Descriptively, it appears that interrogative-complement-taking verbs like ask are limited to select-
ing a normal interrogative CP complement (a WHQ or a YNQ), guaranteeing that all EQs are
root clauses. This may in turn be due to a discourse factor: EQs are immediate responses to a
preceding utterance, something it is hard to envision the embedded EQ of (50) as being, and so
(50) is impossible.
18
My thanks to an anonymous reviewer for this example. Ginzburg and Sag (2001) also make note of this fact.
148 REMARKS AND REPLIES
References
Adger, David. 2003. Core syntax. Oxford: Oxford University Press.
Adger, David. 2006. Remarks on minimalist feature theory and Move. Journal of Linguistics 42:663673.
Adger, David, and Gillian Ramchand. 2005. Merge and Move: Wh-dependencies revisited. Linguistic Inquiry
36:161193.
Baker, C. L. 1970. Notes on the description of English questions: The role of an abstract question morpheme.
Foundations of Language 6:197219.
Carnie, Andrew. 2007. Syntax. 2nd ed. Oxford: Blackwell.
Chomsky, Noam. 1976. Essays on form and interpretation. Amsterdam: North-Holland.
Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Step by step: Essays on minimalist syntax
in honor of Howard Lasnik, ed. by Roger Martin, David Michaels, and Juan Uriagereka, 89156.
Cambridge, MA: MIT Press.
Epstein, Samuel David. 1998. Overt scope marking and covert verb second. Linguistic Inquiry 29:181227.
Ginzburg, Jonathan, and Ivan A. Sag. 2001. Interrogative investigations: The form, meaning, and use of
English interrogatives. Chicago: University of Chicago Press.
Lasnik, Howard, and Mamoru Saito. 1992. Move . Cambridge, MA: MIT Press.
Pesetsky, David. 1987. Wh-in-situ: Movement and unselective binding. In The representation of (in)definite-
ness, ed. by Eric J. Reuland and Alice G. B. ter Meulen, 98129. Cambridge, MA: MIT Press.
Pesetsky, David, and Esther Torrego. 2001. T-to-C movement: Causes and consequences. In Ken Hale: A
life in language, ed. by Michael Kenstowicz, 355426. Cambridge, MA: MIT Press.
Richards, Norvin. 1998. The Principle of Minimal Compliance. Linguistic Inquiry 29:599629.
Rizzi, Luigi. 1997. The fine structure of the left periphery. In Elements of grammar, ed. by Liliane Haegeman,
281337. Dordrecht: Kluwer.
Sobin, Nicholas. 1978. On echo questions in English. In Papers from the 1977 Mid-America Linguistics
Conference, ed. by Donald Lance and Daniel Gulstad, 247259. Columbia: University of Missouri,
University Extension Division.
Sobin, Nicholas. 1990. On the syntax of English echo questions. Lingua 81:141167.
Sobin, Nicholas. 2009. Echo questions and split CP. In Dislocated elements in discourse: Syntactic, semantic,
and pragmatic perspectives, ed. by Benjamin Shaer, Philippa Cook, Werner Frey, and Claudia
Maienborn, 95113. New York: Routledge.