Wh-questions in Singapore English tell us what about questions with declarative syntax ?

Bobaljik & Wurmband (2015) have recently developed a hypothesis that no language truly mixes wh-movement and wh-in-situ structures in its syntax, with seemingly optional wh-in-situ in a wh-movement language being analyzed as a question with declarative syntax. In this paper, we will present novel data from Colloquial Singapore English (CSE) which question this hypothesis. Instead of assuming that the Q-feature of the interrogative CWH head in a language must be specified in a binary manner (valued or unvalued), we will propose that this feature is underspecified in languages such as CSE. The proposed amendment is not only sufficiently restrictive to cover the type of languages predicted by B&W’s original hypothesis, but also flexible enough to accommodate languages with a mixed wh-system. We will further argue that contact-based explanations, though plausible, do not have to be taken as a reason for CSE to develop this specific trait, which could have developed under independent, non-contact situations. This position is supported by Malay and Ancash Quechua, two non-contact languages which nonetheless exhibit optional wh-in-situ like CSE.


Introduction
In their recent attempt to re-assess the important role of syntactic selection within the minimalist framework, Bobaljik & Wurmbrand (2015;hereafter B&W) propose that what appears to be optional wh-in-situ in an otherwise wh-movement language such as English is actually a question with declarative syntax (DSQ) -a declarative syntactic structure which is associated with interrogative semantics/pragmatics at the level of speech act.The authors show that this analysis is supported by the cross-linguistic generalization, tested against 10 typologically different languages, that the in-situ construction is systematically blocked as the complement of an interrogative-seeking verb such as ask and wonder.The reason, they claim, is that such a construction, being syntactically a declarative clause, cannot satisfy the syntactic selection requirement of the verb.In this short remark, we will present our new data from Colloquial Singapore English (CSE) which question the robustness of B&W's typological generalization.We will then propose a minor modification to B&W's original system to accommodate the particular wh-movement profile of this English-lexified contact variety while maintaining their essential theoretical claims.Our suggested modification, in turn, leads one to predict that in principle, there should be some other languages which develop the same wh-movement profile as CSE, but without any contact-induced grammatical changes, and hence present themselves as problematic for B&W's generalization.We will show that this prediction is indeed borne out in Malay (Cole & Hermon 1998) and Ancash Quechua (Cole 1982).
Glossa general linguistics a journal of Sato, Yosuke and Jian Gang Ngui. 2017.Wh-questions in Singapore English tell us what about questions with declarative syntax?Glossa: a journal of general linguistics 2(1): 17. 1-18, DOI: https://doi.org/10.5334/gjgl.66Sato and Ngui: Wh-questions in Singapore English tell us what about questions with declarative syntax?
Art. 17, page 2 of 18 This remark is structured as follows.In section 2, we will review B&W's generalization and their analysis thereof.In section 3, we will establish that CSE is a wh-movement language, just like standard varieties of English, but also allows in-situ wh-questions, just like Mandarin Chinese.In section 4, we will introduce our crucial examples showing that CSE allows both wh-in-situ and wh-movement options as syntactic CP complements of interrogative verbs such as ask and wonder.We conclude that the availability of the wh-in-situ strategy in this context contradicts B&W's central conjecture behind their generalization that no language truly mixes wh-movement and wh-in-situ structures.In section 5, we will propose a minor modification to B&W's original system to accommodate the wh-movement profile of CSE while essentially keeping their central theoretical assumptions intact.We will point out further that our proposed amendment leads us to predict that some other languages which have the CSE-like wh-movement profile should, in principle, exist even without any language contact, unlike CSE.We will show that Malay and Ancash Quechua facts bear out this prediction.Section 6 is the conclusion.

Two types of wh-in-situ and DSQs in wh-movement languages
B&W propose that what appears to be an optional, non-echo, in-situ wh-question in whmovement languages is a DSQ, namely, a syntactically declarative clause with an in-situ wh-phrase in focus which manages to carry interrogative force merely as a semantic/ pragmatic speech act.B&W argue that this DSQ analysis of the apparent wh-in-situ phenomenon in otherwise wh-movement languages is supported by the cross-linguistic generalization shown in (1).
(1) DSQ/wh-in-situ generalization: If a language has wh-movement (to Spec,CP), then wh-movement is obligatory in indirect questions.
Equivalently: If a wh-movement language allows 'optional' wh-in-situ, the in-situ construction is blocked in selected questions.(B&W 2015: 13) The reason why the generalization holds under B&W's system is as follows.They assume that for all languages, a syntactic interrogative clause involves an interrogative complementizer, C wh .They further assume that the relevant C in a wh-movement language is endowed with a certain featural property which triggers overt wh-movement into its specifier.To fully appreciate their system, consider the syntactic derivations shown in (2a, b) as a point of illustration.
( B&W implement overt wh-movement in a wh-movement language within the Reverse Agree framework (Wurmbrand 2014), which essentially states for our present purposes that an unvalued feature of a given lexical item is valued by the corresponding valued feature of another lexical item only if the latter c-commands the former: see Wurmbrand (2014) andB&W (2015: 21) for further technical details on feature valuation and Agree.In (2a), the C WH head is introduced into the syntactic workspace with an unvalued but interpretable Q-feature.Given the Reverse Agree framework, then, this unvalued feature can only be valued by a c-commanding wh-phrase endowed with the valued Q-feature.This valuation requirement is correctly satisfied only if there is overt wh-movement, as depicted in (2a), yielding a syntactically interrogative clause, i.e. a CP typed as iQ: wh; otherwise, the Q-feature of the C head would remain unvalued, causing the derivation to crash.The CP so typed can subsequently be merged with an interrogative verb such as ask and wonder, which syntactically selects this particular type of CP, as shown in (2b). 1  In other words, an interrogative-seeking predicate cannot be merged with a CP which is not appropriately typed as iQ: wh via overt wh-movement, for this merger operation is blocked by syntactic selection.B&W hypothesize that DSQs are syntactically declarative CPs without C WH , with their interrogative interpretation as a wh-question being computed in the semantics/pragmatics component due to the inherently interrogative meaning/function of wh-expressions.The syntactic structure associated with DSQs, thus, is as shown in (3). 1 An anonymous reviewer asks how the required c-command relationship can be met within the Reverse Agree framework when the moving wh-phrase is internally complex, as in which book or what kind of fruit, in which case the wh-word itself cannot c-command the C head to valuate its unvalued Q-feature.We simply assume that in such a case, the valued Q-feature of the wh-word may percolate onto its dominating DP so that the entire wh-phrase may be endowed with the Q-feature.This way, the newly derived "lexical item" now c-commands the C head, as desired.
Consequently, DSQs cannot occur as syntactic interrogative complements of an interrogative-seeking predicate.Note that there is no comparable obligatory selectional requirement imposed on matrix interrogative CPs.This gives rise to potential optionality with respect to overt wh-movement within a matrix clause.Under B&W's system, overt whmovement takes place in a matrix clause if headed by the C wh endowed with the unvalued Q-feature, as shown in (2a), whereas the DSQ option obtains in the clause if headed by the declarative C head, as shown in (3).It follows then that the in-situ construction (namely, DSQs), though possible in principle in matrix questions, is systematically blocked in selected questions in a wh-movement language, as explicitly stated in (1).B&W show that the generalization in (1) is supported in the following 10 languages they surveyed: English, German, Dutch, Icelandic, American Sign Language, Brazilian Portuguese, Spanish, Russian, Polish, and French.To illustrate the generalization using (standard varieties of) English, it has been reported in the literature (Postal 1972;Ginzburg & Sag 2000;Huddleston & Pullum 2002;Pires & Taylor 2007) that English speakers may use information-seeking, non-echo, in-situ wh-questions in several well-demarcated pragmatic circumstances such as legalistic questions, quiz-show questions, and requests for further information to flesh out situationally salient contexts.Some examples of the insitu construction in Standard English are shown in (4).( 4 The Mandarin Chinese-type pattern is consistent with the generalization in (1), for Mandarin Chinese (as well as Turkish and Japanese, for that matter) is not a wh-movement language, to begin with.Given B&W's system outlined above, it must be the case that, in a genuine wh-in-situ language such as Mandarin Chinese, the dependency between the interrogative C wh and a wh-phrase should be satisfied without overt wh-movement, either through covert wh-movement (Huang 1982;Pesetsky 2000) or unselective binding (Pesetsky 1987;Tsai 1994; see also Reinhart 1997 for an alternative in-situ analysis in terms of choice functions).Under the former analysis, wh-movement takes place covertly, but otherwise, the syntactic derivation proceeds in the same way as overt wh-movement in (2a).Under the latter analysis, we may assume that the C WH head is introduced into the syntactic workspace with a valued Q-feature, so that overt wh-movement will not be triggered by the valuation requirement and hence cannot take place on grounds of economy.In this case, then, the whole CP headed by the relevant C head is interpreted as a syntactically interrogative question by means of unselective binding, according to which the valued Q-feature of the C head binds (another valued Q-feature contained within) an in-situ wh-phrase.This way, the CP with an in-situ wh-phrase in a wh-in-situ language is appropriately typed as an interrogative clause without overt wh-movement so that it can further merge with an interrogative-selecting predicate, as attested in (6c).In this paper, we will assume the unselective binding approach to interrogative clause typing for true wh-in-situ languages such as Mandarin Chinese and return to some empirical evidence for this approach in section 5.2.
It should be clear from the exposition above that B&W's analysis argues that apparent cases of in-situ wh-questions in an otherwise wh-movement language are all DSQs, a point supported by their novel, cross-linguistically robust observation that such questions, though in principle available and facilitated under various contexts in matrix questions, are systematically prohibited once they are selected by interrogative-seeking matrix predicates.We have also observed that other languages such as Mandarin Chinese allow the in-situ wh-construction in selected questions, the reason being that the Q-feature of the C wh head in this type of language is valued, so that the syntactically interrogative CP can be created by unselective binding instead of overt wh-movement in languages such as English.Given these two observations, B&W's generalization in (1) entails an explicitly typological prediction that "[i]n their syntax languages are either wh-movement or wh-in-situ (more accurately, wh-in-focus) but no language (that we know of) truly mixes both constructions."(B&W 2015: 14). 2 To phrase this point in formal terms within B&W's framework, the generalization in ( 1) embodies an all-or-nothing claim regarding the availability of the two types of the C wh head, namely that a language can only ever have either the C with the unvalued Q-feature (as in Standard English) or the C with the valued Q-feature (as in Mandarin Chinese). 3n the next two sections, we will show that the wh-question paradigm in CSE is problematic for B&W's generalization and conclude that B&W's all-or-nothing claim regarding the availability of the different types of the same interrogative C wh head in a single language does not survive close scrutiny.

Wh-questions in CSE: Optional wh-in-situ in an otherwise wh-movement language
It has been well documented in the growing body of literature on CSE (Chow 1995;Bao 2001;2015;Kim et al. 2009;Yeo 2010;2011;Sato 2013) that this English-lexified contact variety permits both wh-movement and in-situ wh-questions.Given that regular wh-movement is a prerequisite for sluicing, the grammaticality of ( 12) indicates that CSE has wh-movement.The two considerations above, therefore, show that CSE is an overt wh-movement language on a par with Standard English and other languages discussed by B&W.
4 Wh-in-situ questions in CSE as a genuine challenge for B&W's generalization We have established in section 3 that CSE is a wh-movement language which also has the option of in-situ wh-questions.The presence of overt wh-movement in this variety means under B&W's system that it occurs in order to type a CP as a syntactic interrogative question (iQ: wh), with apparent in-situ wh-questions being instantiations of DSQs.B&W's system, therefore, predicts that such questions in CSE, being DSQs, should be blocked as interrogative complements of interrogative verbs.This prediction is not borne out.
Examples (13)(14) show that both overt wh-movement and in-situ structures are available as an interrogative CP complement of the verb wonder in CSE.These two structures are also freely available when they occur as a syntactic complement of another verb, ask, as witnessed by the grammaticality of (15).'John asked who the rice was for.' As stated in section 2, B&W's theory of selection prohibits DSQs from being selected by interrogative verbs.Following the theory, then, the in-situ wh-questions in the (b)examples of (13-15) are not DSQs.Indeed, there are two independent arguments against the DSQ analysis of the examples.First, B&W point out that in-situ wh-phrases in a whmovement language, as DSQs, may occur within an embedded clause headed by the noninterrogative, declarative complementizer that, as shown in ( 16).
Thus, if the in-situ wh-questions in (13-15) were DSQs, we would predict that they should occur as CPs headed by the same complementizer.This is not the case, as shown in (17ac), however.These examples show that in-situ wh-phrases cannot occur in this context.Note that this co-occurrence restriction cannot be attributed to the Doubly-Filled Comp Filter (see section 3), for the wh-phrases are not in (cf.I know that the rice is for Mary).
The second argument against the DSQ analysis of in-situ wh-questions in CSE comes from the complete lack of any semantic-pragmatic restriction imposed on the use of such questions in this English variety.We have seen in section 2 that the in-situ construction in standard varieties of English, or DSQs in B&W's theory, is only available in a range of special semantic-pragmatic contexts such as quiz-show questions and legalistic questions.Such a characterization, however, does not hold true for the corresponding in-situ whquestions in (13-15) and (7-10), for the existing literature on CSE wh-questions observe, and our native speaker consultants of CSE (cf.notes 4 and 5) indeed confirm, that both wh-movement and in-situ constructions are equally acceptable without imposing any additional semantic-pragmatic context to facilitate one variant over the other.We take this property of CSE in-situ questions to indicate that they are bona fide wh-in-situ configurations. 7In section 5.2, we will present further arguments for the wh-in-situ analysis of in-situ wh-questions in CSE.
The conclusion that CSE is a wh-movement language which allows the genuine whin-situ option in selected questions, therefore, is problematic for B&W's generalization that no language truly mixes genuine wh-in-situ and wh-movement structures.Recall that this generalization was a corollary of their feature-based system of wh-question licensing through syntactic selection.Given the Reverse Agree framework, overt wh-movement must take place in a wh-movement language such as Standard English because the unvalued Q-feature of the C WH head must be valuated by the valued Q-feature of a c-commanding wh-phrase.As implied from the generalization in (1), B&W's theory critically assumes that a language can have either the C wh with the unvalued Q-feature or the C wh with the valued Q-feature.Due to this all-or-nothing assumption, all instances of apparent in-situ wh-questions in a wh-movement language must be analyzed as DSQs in B&W's theory.More generally, then, their generalization would predict that no language can truly conflate wh-in-situ and wh-movement structures.We therefore conclude that the availability of the true wh-in-situ construction in the CSE examples in (13-15) presents a genuine challenge for B&W's typological generalization. 6We thank an anonymous reviewer for encouraging us to check the availability of the declarative complementizer to co-occur with embedded in-situ wh-questions in CSE to independently exclude the DSQ analysis.Note that we used the verb know in (17), a verb which optionally selects an interrogative complement, because obligatorily interrogative-seeking verbs such as wonder and ask cannot co-occur with the declarative complementizer, whether or not the CP selected by such a verb contains an in-situ wh-phrase, as shown in (ia, b).
(  Art. 17, page 9 of 18 5 Wh-questions in CSE and the feature specification of the C wh head We have shown that CSE, an overt wh-movement language like standard varieties of English, allows the Chinese-style genuine wh-in-situ configuration as a syntactic complement selected by interrogative verbs, contrary to the prediction made by B&W's theory.The data introduced in those sections show that CSE grammar permits both wh-movement and wh-in-situ, in matrix clauses (7a, b, 8a, b, 9a, 10a) or in embedded clauses (9b, 10b, 13a, b, 14a, b, 15a, b).The question now is how CSE grammar can produce these two syntactic options, the co-existence of which should be blocked within a single grammar under B&W's theory.In this section, we will suggest a minor technical modification to B&W's theory to accommodate the CSE profile while still keeping their essential theoretical claims to maintain their empirical coverage.More specifically, we will propose to drop their assumption that the C wh head in a language can only have one type of feature structure, i.e., either the valued Q or the unvalued Q, and argue instead that in some languages such as CSE, the C head may well be underspecified with respect to its value, so that its Q-feature may enter the derivation either valued or unvalued.We will show that this underspecification approach also correctly predicts the co-existence of wh-movement and wh-in-situ options in selected questions in Malay (Cole & Hermon 1998) and Ancash Quechua (Cole 1982), two other languages which are also shown to be problematic for B&W's typological generalization/analysis.

A micro-parametric feature underspecification in CSE: C WH [u/vQ]
To accommodate the CSE wh-paradigm within B&W's overall system, we propose here that the C WH head in CSE may enter the syntactic derivation with either the valued or the unvalued Q-feature.Our proposed modification to B&W's theory is schematically illustrated in Table 1.Here, [uQ] stands for an unvalued Q-feature.
[vQ] stands for a valued Q-feature.[u/vQ] indicates that the Q-feature may be either unvalued or valued.Finally, [Q] stands for a Q-feature which has no room for any valuation in the syntactic workspace, i.e., a Q-feature which is neither unvalued nor valued.B&W's generalization was designed to cover the typological profile of the two types of languages mentioned in the first two rows.In languages such as Standard English, the C WH [uQ] requires overt wh-movement into [Spec,CP].All in-situ wh-questions are DSQs and hence are not embeddable as syntactic complements of interrogative verbs.In languages such as Mandarin Chinese, the C WH [vQ] results in unselective binding, so that the dependency between the head and the wh-phrase is satisfied in-situ, yielding genuine wh-in-situ configurations which can meet the selectional requirement of interrogative verbs.
Our central proposal here is that the valuation of C WH does not need to be binary: the head can also come with either the valued or unvalued Q-feature, as indicated in the third row (i.e., C WH [u/vQ]), contrary to B&W's assumption that a language can have only one featural-type of the interrogative C wh .Expectedly, this type of language can combine properties from English-type and Chinese-type languages so that overt wh-movement and wh-in-situ emerge as two equally available mechanisms to create syntactically interrogative CPs.There is the fourth logical possibility shown in the final row (i.e., C WH [Q]), where the C WH is endowed with no value slot.We believe that this technical possibility is excluded on an independent ground: under the reasonable assumption that all languages have "wh-questions" as a universal semantic template, the C head with no appropriately specified Q-feature would not be able to send any proper instructions to the syntactic computation to derive such questions.This would then result in the violation of the LF legibility condition, formulated one way or another, to the effect that C HL must prepare an articulate syntactic structure to be appropriately mapped onto the interrogative whquestion in the semantic interface component.However, we choose to remain agnostic regarding the availability of DSQs in this hypothetical language type.According to B&W, DSQs are declarative syntactic structures which have interrogative semantics due to the inherent interrogative meaning/function of wh-phrases.As such, this type of language may still retain DSQs, which are then not embeddable in selected questions.
One question remains in our proposed amendment (as well as in B&W's system, for that matter): whether Mandarin Chinese and CSE allow DSQs in matrix interrogative sentences, in addition to genuine wh-in-situ. 8Recall that the DSQ option should be universally available, as it is based on the inherent interrogative force of wh-words.Consequently, nothing should, in principle, prevent these two languages from utilizing this option in matrix clauses, which are not required to be typed as interrogative through unselective binding.However, it seems plausible to believe that neither CSE nor Mandarin Chinese actually makes use of DSQs as an alternative in-situ strategy, given the undeniable fact that matrix wh-questions in these languages are not subject to special semantic-pragmatic constraints which characterize DSQs in a wh-movement language.We nonetheless leave an in-depth investigation of the trade-off relation between unselective binding and DSQs in genuine wh-in-situ languages for another occasion as an important topic for future research.

CSE wh-in-situ and unselective binding: Non-wh-variables and island-insensitivity
In this section, we will provide two pieces of evidence in favor of the unselective binding analysis of wh-in-situ questions over the covert wh-movement alternative in CSE.This point is important for our present purposes because the Chinese-like properties of CSE wh-questions crucially depend on the availability of its [vQ] feature, which, in turn, triggers unselective binding.Under the covert wh-movement analysis, on the other hand, the Q-feature would necessarily be specified as [uQ] in CSE so that a CP may be properly typed as a syntactic interrogative complement.In that case, then, our underspecification approach in Table 1 would not stand.
The first piece of evidence for the unselective binding analysis in CSE comes from a non-interrogative use of wh-words in CSE.According to the unselective binding approach, an interrogative Q-element is assumed to existentially quantify all in-situ wh-phrases in a sentence as individual wh-variables (Heim 1982).Such an analysis, then, is supported if wh-words in CSE may also be used as variables bound by some other operator different from the interrogative Q-element/operator.Examples (18a-c) show that wh-variables in CSE such as what, who, and where can occur as individual variables bound by a non-interrogative operator in the context of universal quantification.This observation, however, is difficult to capture or mysterious at best under the (covert) wh-movement analysis of wh-in-situ in CSE. (18) a. I what also never eat.'For all x, I didn't eat x.' b.I who also never meet.'For all x, I didn't meet x.' c.I where also never go.'For all x, I didn't go to x.' d. *I why also never cheat.'For all x, I didn't cheat because of x.' Note that the adverbial wh-word why cannot be used as a variable in this context, as shown by the ill-formedness of (18d).This pattern also makes sense because why denotes a function over a higher-order entity, not an individual, and hence cannot be interpreted through unselective binding.Now, given this observation, our current analysis predicts that why-questions in CSE can only be licensed through the C WH with the unvalued Q-feature, namely via overt wh-movement, as in Standard English, as opposed to other nominal wh-questions headed by who, what and where, which may occur in-situ (recall (7-10)).This prediction is indeed borne out in ( 19).
b. *Mary must leave early why?
The second piece of evidence for the unselective binding analysis of wh-in-situ in CSE is concerned with the island-insensitivity of a wh-in-situ phrase within the CP complement of an interrogative-seeking matrix predicate.Recent works on the syntax of wh-questions in CSE (Kim et al. 2009;Yeo 2010;2011;Sato 2013) point out that CSE allows not only full wh-movement and wh-in-situ, as shown in (20a) and (20c), respectively, but also partial wh-movement, as shown in (20b).
( If this example involved covert wh-movement of what into the specifier of the intermediate CP selected by the matrix verb, it would result in ungrammaticality due to the violation of the complex DP island constraint, for we have just seen in ( 21b) that covert whmovement is sensitive to the same constraint.The grammaticality of ( 22), on the other hand, naturally falls into place if the wh-phrase is interpreted in-situ through unselective binding, because this mechanism triggers no island violations, as shown in (21c).The observations made here thus support our unselective binding analysis of wh-in-situ in CSE.

Feature underspecification or contact dynamics? Evidence from Malay and Ancash Quechua
In this section, we would like to investigate whether the CSE-type optional wh-in-situ is an exclusive by-product of contact-induced grammatical changes or instead due to a purely mechanical feature-level adjustment with respect to the (under-)specification of the C WH head.We will argue in this subsection that contact-based explanations, though plausible, do not have to be taken as a reason for CSE or any other language to develop this specific trait, which could have developed under independent, non-contact situations. 9This position is supported by Malay and Ancash Quechua -two non-contact languages which nonetheless exhibit optional wh-in-situ like CSE and hence present themselves as further challenges for B&W's typological generalization.It has been a consensus view in the literature on CSE (Bao 2001;2015;Bao & Lye 2005;Sato 2011;2013;2014; see also many other references cited therein) that its syntax has received non-negligible, systemic, contact-induced influences from various regional varieties of Chinese, including Hokkien, Cantonese, Mandarin and, to a lesser extent, vernacular varieties of Malay (Bazaar Malay and Baba Malay), in diverse syntactic areas such as pro-drop/argument ellipsis, bare conditional structures and, importantly for our purposes, the syntax of wh-questions.Given the merits of the Chinese substratist explanation amply documented in the CSE literature, one may well suggest, with some degree of confidence, that the availability of both the wh-in-situ and wh-movement options in this particular English variety owes itself to grammatical changes brought about by constant contact pressures imposed by Standard English (the lexifier language) and Sinitic (the substrate language) within its dynamic, endogenous contact ecology.Indeed, as we saw in (6), Mandarin Chinese uses wh-in-situ in both matrix and embedded questions.The contrast between (23a) and (23b) further demonstrates that, in fact, this option is the only grammatical choice available in selected questions in Mandarin Chinese.In (26a), the wh-word apa 'what' is successfully bound as a non-wh-variable by the existential quantifier expressed in the form of reduplication.Similarly, in (26b), the same whword is bound by the existential quantifier -pun 'also'.The ability of certain wh-words to serve as non-wh-variables is, then, correctly captured by the unselective binding analysis of wh-in-situ in Malay.Recall further from section 5.2 that the adverbial wh-phrase why in CSE cannot be licensed in-situ via unselective binding, as shown in (18d), because it denotes a function over a higher-order entity, not an individual.Accordingly, this whphrase must be licensed by overt wh-movement, as triggered by the valuation requirement of the unvalued Q-feature of the C WH head.Given this observation, our analysis also correctly predicts that, as first noted by Cole & Hermon, the Malay equivalent of CSE why, kenapa, not only lacks any function as a non-wh-variable but also undergoes obligatory overt wh-movement, as shown in (27-28), respectively.( 27 The Malay data reported above thus indicate that the underspecification of the valuation requirement on the Q-feature of the C WH head (i.e., [u/vQ]) may arise in a given language without any contact-induced grammatical changes, though it could have been triggered in the particular case of CSE.
Ancash Quechua exhibits the same profile with CSE and Malay with respect to optional wh-in-situ, and hence raises another challenge to B&W's generalization.Cole (1982) observes that this Quechuan variety allows overt wh-movement and wh-in-situ. 11rom our careful reading of Cole's data concerning wh-questions in this variety and his descriptions thereof, there is no evidence to suggest that the in-situ construction is subject to any special semantic-pragmatic restrictions, which has been shown to constrain the use of in-situ wh-questions/DSQs in an overt wh-movement language.Given this observation, what is notable is that not only the overt wh-movement but also the wh-in-situ structures can be used as a syntactic interrogative complement selected by an interrogative verb such as tapuy 'to ask', as illustrated by the grammaticality of both (29a) and (29b  Cole (1982) further points out that in Ancash Quechua, overt wh-movement is subject to island constraints whereas wh-in-situ is not.This contrast is illustrated below with the complex DP island. (30) Ancash Quechua (Cole 1982: 42-43) a. *Ima-ta-taq i qam kuyanki [ DP t i suwa-q nuna-ta]?what-acc-q you love steal-nom man-acc '*What do you love the man who stole?' b.
Qam kuyanki [ DP ima-ta suwa-q nuna-ta]?you love what-acc steal-nom man-acc 'You love the man who stole what?' Reflecting the then-standard theoretical assumption within the Extended Standard Theory (see Huang 1982, among others) that S-Structure movement, not LF movement, is subject to Subjacency, Cole (1982) himself took the absence of the island effect in (30b) as an indication that covert wh-movement involved there is not subject to island effects.We have a different interpretation of the relevant fact from Cole's, however.In section 5.2, we have presented independent evidence from CSE that partial wh-movement, an instance of covert wh-movement, does obey island effects: see also Cole & Hermon (1998) for supporting evidence for the same conclusion from partial wh-movement in Malay.Given this independent evidence, we believe that the island-insensitivity of wh-in-situ in Ancash Quechua, as shown in (30b), actually supports the unselective binding analysis over the covert wh-movement alternative.With this being said, we must leave a more careful investigation of other potential evidence for the unselective binding analysis of wh-in-situ in Ancash Quechua for another occasion due to the inaccessibility of further data and information on wh-questions in this Quechuan variety.At any rate, to the extent that this analysis holds true, the preliminary survey conducted here of the wh-question paradigm in this language lends further credence to our view that the C WH in a language can be underspecified and come with either value with respect to its Q-feature, independently of whether its grammar is shaped in contact situations.

Conclusion
In this remark, we have presented novel examples showing that CSE permits optional whin-situ in both matrix and embedded interrogative clauses.We have argued that the availability of the wh-in-situ configuration in selected questions in CSE poses an empirical challenge to B&W's otherwise cross-linguistically robust generalization that no language may truly mix English-type overt wh-movement and Chinese-type wh-in-situ structures.We have presented a minor modification to B&W's system to capture the genuine optionality of wh-movement in CSE, according to which the C WH head in this variety may enter the syntactic derivation with either the valued Q-feature or the unvalued Q-feature.Although it is reasonable to claim that this underspecification in CSE may have arisen from its intense contact with English (lexifier) and Chinese (substrate), we have also suggested that the existence of contact roots does not have to be taken as a necessary condition for a given language to develop genuine wh-movement optionality in its grammatical system because nothing in principle seems to block the logical possibility that the Q-feature may enter the syntactic workspace with either value.To support this suggestion, we have documented preliminary data showing that Malay and Ancash Quechua exhibit the same wh-movement profile as CSE, despite the fact that they are not contact languages.

should put this stuff where. c. *Umbridge asked Trelawney she's been in the post how long.
illustrate the availability of these two options with what and where in both matrix and cross-clausal contexts.4 a. What you think she buying ah? 'What do you think she is buying?' b.You think she buying what ah? 'What do you think she is buying?' (10) Yeo (2010: 9) a.Where you think I go? 'Where do you think I went?' b.You think I go where?'Where do you think I went?'There are two arguments showing that wh-phrases in CSE undergo regular overt whmovement into [Spec,CP], as in Standard English.One argument comes from the so-called Doubly-Filled Comp Filter (Chomsky & Lasnik 1977), a surface constraint which blocks the co-occurrence of a wh-phrase and a complementizer within the same CP region.The effect of this filter is illustrated in (11).(11) *Mary know [ CP what i [ C′ {that/if/whether} [ TP he buy t i ]]]?
Sato and Ngui: Wh-questions in Singapore English tell us what about questions with declarative syntax?

Table 1 :
Valuability of the Q-Feature of the C WH Head.
What is noteworthy about wh-questions in CSE is that the partial wh-movement derivation is blocked in CSE when an island boundary intervenes between the surface position of the partially moved wh-phrase and the specifier of the matrix scopal CP.Consider examples (21a-c) to illustrate this point.John like [ DP the man that think Mary eat t i ]? b. *John like [ DP the man that think what i Mary eat t i ]? c. John like [ DP the man that think Mary eat what]?(21a) is ungrammatical because overt wh-movement of what crosses the complex DP island boundary.What is interesting is that (21b) is ungrammatical despite the fact that the partially moved wh-phrase does not seem to cross the relevant boundary, unlike in (21a).The ungrammaticality of (21b) falls into place, however, if we assume that the whphrase in the intermediate specifier of CP undergoes covert wh-movement into the specifier of the matrix interrogative CP, triggering the island violation.Given this result, the grammaticality of the in-situ variant shown in (21c), in turn, indicates that wh-in-situ in CSE does not involve any syntactic wh-movement, either overt or covert, but is licensed instead by some in-situ mechanism such as unselective binding.Keeping this point in mind, consider now example (22), where the wh-phrase what remains in-situ within the interrogative CP complement of the interrogative-selecting matrix predicate wonder.CP C WH Mary like [ DP the student who ate what]]?
) a. What i John think Mary eat t i ?b.John think what i Mary eat t i ?c. John think Mary eat what? ).