Stable Nodes and Unstable Features: A Principled Account of Variation Within Minimalism Lisa Green, Tom Roeper, J. Michael Terry Introduction We propose a simple and intuitive theory of where grammar variation happens easily and where grammar seems immutable. These patterns of grammar variation are reflected in dialect variation—within and across dialects, language acquisition, and language change. The idea is simply that 1) nodes are stable and 2) features are unstable. From those assumptions, it follows naturally that nodes that unique to a given language are hard to learn and difficult to change, while features easily exhibit dialect variation and predict surprising deviations in the acquisition path. African American English (AAE) provides a domain where these ideas are illustrated with both clarity and subtlety. That simple idea about nodes and features must be embedded in a technically rich theory, which we shall partly undertake presently. However, before we make that effort, we want to etch in the idea in such a way that it can survive particular theoretical changes. In the history of phrase-structure grammar, nodes were not features, but reflected a universal base produced by rewrite rules that determined allowable phrasal categories. Thus from the outset, nodes were in a separate class from categorical and word-based features. The claim became formulable when, under minimalism, via in part the influence of Categorical Grammar and Head-driven Phrase-structure, the definition of nodes themselves became feature-based. Nodes are composed of feature bundles, and the Heads of nodes reflected the primary features of grammar. The next step was to allow variation in the set of possible nodes selected by particular grammars, as it has been articulated in greatest depth in the cartographic approach to grammar. As in that approach, we assume that some nodes are indeed universal while others are present in particular languages, and others are subject to a kind of “splitting” into sub-nodes. As a result of those developments, it becomes possible to ask whether there is a difference in status between those features that define nodes and those features that are present on certain nodes without a defining role. This intuitive account is matched by the concept of “interpretable” and “uninterpretable” features, where the former correspond to nodes or node defining features and the latter to features that may motivate movement or agreement. While overt movement provides a visible dimension for uninterpretable features, agreement often does not, nor does covert movement. Our focus falls upon AAE and its intricate variation in auxiliary inversion, but we begin with an illustration from classic arguments about verb movement in Mainstream American English (MAE). 1. Morphological Inflection and Variation in American English It has been argued that the tense features are generated where they appear in English, under a V node, and then move to the Tense node covertly, as in (1): 1 1. TP 2 T’ 2 1 [+Tns] T VP V’ [+Head] 2 V-s [+Tns] [+Head] In German or French, by contrast, the tense affix is directly generated under the Tense node, and the verb moves up to that position. In other words, the tense feature, which subsequently moves to T, is a head feature in one configuration, but in the “wrong” position in another, not matching the head [Tense] features that occur under V (Chomsky 1995). Emonds (1980), Pollock (1989), and Chomsky (1991) developed an argument that the inflection –s was generated beneath the VP node and moved, overtly in French and German, and covertly in English, to the Tense position. The –s, then, is an inflection which carries a feature that does not represent the node under which it visibly occurs. We predict, therefore, that it is unstable and easily deleted. (See Deevy (2000) for speech error evidence.) This appears to be exactly correct, as many English dialects fail to use the inflection. Also, children often fail to use it, and disorders are defined in terms of missing inflections (e.g., 3rd singular –s). This approach, of course, reflects the heart of the argument for covert movement. Our extension to variation, then, in a sense, is carrying out the implication of the original idea. Of course, inflections linked to head nodes are also not instantly acquired, but we argue, still show much greater stability than -s and -ed in English. For instance in AAE, we find the following: 2. a) He run to school. ‘He runs to school’ b) Bruce jump the fence. ‘He jumped the fence’ We can illustrate the abstract claims with the following general concept: 3. General Concept 2 X W Z [W, Z, R] Z = stable, reflects node W = stable if moved overtly, less stable if covert R = inherently unstable 2 Thus Z is a direct reflection of the higher node. W is a feature under the wrong node, in a sense, and must move to be checked off, and R represents a feature which is a possible restriction on entries under Z, but is neither an instance of Z nor subject to movement. 2. Stable and Unstable Features: A Feature-Based Theory We turn now to the features involved in auxiliary movement as linked to questions. We argue that some feature must move to a Force Phrase (=CP) position where a Force feature, [+yes/no Q], is present. Nevertheless, a second feature [+presuppositionality] ([PRESUP]) is optional in questions and varies across dialects, as we shall illustrate in considerable detail. The notion of presuppositionality we represent with a feature [PRESUP] although this obviously stands for a more complex analysis. The crucial point is that [PRESUP], which refers to a presupposed proposition, is not a Force feature; therefore, it is not a reflection of a node.1 Predictably, then its appearance should be variable across grammars and unstable within them. In fact, within MAE it is difficult for speakers to accept that (4a) and (4a’), for example, both have the same [-PRESUP] force: 4. a) Can she leave? a’) She can leave? b) Do they wanna go to the bookstore? b’) They wanna go to the bookstore? The (4a’, b’) forms in AAE produce yes/no interpretations via intonation, as we discuss below, and, therefore, the head feature under the Force node is satisfied. While the MAE speaker has the uninverted forms (4a’, b’) as well, they have a [+PRESUP] feature for them. In MAE (4a) is an open proposition where nothing is presupposed, while She can leave? has the same force as a tag question, in which the proposition is presupposed, then questioned: She can leave, can’t she? The tag question requires a [+PRESUP] element which is not present in the ungrammatical form in MAE: *Can she leave, can’t she?. The sentence is predictably ungrammatical because the matrix inversion presupposes an open hence unpresupposed proposition, and the tag question requires a closed, presupposed proposition. Here we give the theoretical orientation and restate our view more formally. In feature-based theory in minimalism, nodes are feature bundles with labels that define and motivate movement (Chomsky 2005, 2006; Adger and Tsoulas 2004). Our view is that a) Feature-based labels motivate movement obligatorily and are stable. b) Uninterpretable features in a probe-goal relationship are unstable. Chomsky (2006) makes the following observation about uninterpretable features: 1 The notion of proposition itself may have a variety of subtypes, including semelfactives, negative factives, and propositionality linked to implicatures. Our argument suggests that it must have a syntactic reflex in some, but not all, cases. 3 “Since the values of these features are determined by context, the simplest assumption is that they are unvalued in the lexicon, thus properly distinguished from interpretable features, and assigned their values in syntactic configurations.” We add to that claim the distinction between overt and covert movement as contributing to or subtracting from stability: If a feature is uninterpretable and covertly moved, then it will be unstable. These claims have the implication that where the syntactic configurations differ, we can expect that the behavior of uninterpretable features will differ. If they are linked to a dialect-specific node that requires overt movement, then they will be stable, otherwise not. 3. Question Formation: Unstable [-PRESUP] and Stable [wh-Q] On the notion of presuppositionality, we argue that Question Inversion can carry an uninterpretable feature [±PRESUP], which is a syntactic approximation to a link to the semantic (and/or pragmatic) system that can enforce or lift a presupposition in the ccommand domain. The distinction is sharply evident in the following example from MAE: 5. a) Why don’t you go outside? = [-PRESUP] b) How come you don’t go outside? = [+PRESUP] The sentence in (5a) opens the proposition (you go outside) via inversion, while the sentence in (5b) presupposes the proposition you don’t go outside. We argue that precisely this feature is not linked to a particular node and is subject to variation in its realization. In particular, while AAE speakers allow such sentences as (5b) to be linked to [-PRESUP], MAE speakers require the presupposition and cannot comprehend this variant. MAE speakers must regard (5b) as carrying a presupposition. 3.1 Yes-No Questions We can illustrate the distinction in different types of features in our framework with yesno questions: 6. ForceP (=CP) 2 Spec, Force Force’ [+YNQ] 2 Force TP [+YNQ] 2 T’ 2 T 1 AUX [+YNQ] [±PRESUP] 4 Both the [+YNQ] feature and the [+PRESUP] feature move, but only the [+YNQ] feature satisfies the Force node in the derivation for the sentence Can she leave? (7). 7. ForceP (=CP) 2 Spec, Force Force’ [+YNQ] 2 Force TP 1 2 can T’ [+YNQ] 2 [+PRESUP] T 1 AUX 1 (can) A second method for satisfying the [+YNQ] on the Force node is through rising intonation, which one can see as a marker of covert question movement. It exists in both MAE and AAE, but with different presuppositional characteristics. There is no auxiliary inversion in the derivation for She can leave? (8). 8. ForceP (=CP) 2 Spec, Force Force’ [+YNQ] 2 Force TP 1 2 [+YNQ] T’ [+PRESUP] 2 T 1 AUX 1 can [+YNQ] [+PRESUP] While MAE utilizes intonation to satisfy [+YNQ], which triggers covert movement with [+PRESUP], AAE uses intonation with [-PRESUP], also. Cheng and Rooryck (2000) have also argued that intonation carries an uninterpretable feature to capture wh-in situ in French. In summary, we have both [+PRESUP] and [-PRESUP] in AAE with both readings in (9b) and only [+PRESUP] in MAE (9bii), so AAE has a wider range of variation than MAE: 9. a. Can she leave? [-PRESUP] b. She can leave? i) ‘Can she leave?’ [-PRESUP] ii) ‘She can leave, can’t she?’ [+PRESUP] 5 3.2 Wh-Questions Stable nodes are linked to features which must be checked off and exhibit overt movement. A classic case, of course, is wh-movement, which we illustrate here again with examples from AAE: 10. a) Why can she have that book? a’) Why she can have that book? ‘Why can she have that book?’ b) *Can she have that book why? b’) *She can have that book why? c) Why don’t you go outside? c’) Why you don’t go outside? ‘Why don’t you go outside?’ While auxiliary inversion does not obligatorily occur, wh-movement is obligatory in AAE as it is in MAE, as illustrated in (10 b, b’). The sentences in (10) show that whmovement is obligatory. It is not sufficient to invert the auxiliary (10b), nor is it possible to have a non-inverted auxiliary and a non-moved wh-word (10b’). The sentences show that the problem is that the wh-word is still in situ. The claim here is that the stable feature in wh-constructions is wh-Q, which is in Spec, ForceP. 11. ForceP 2 why[wh-Q] Force’ 2 Force TP 1 2 (can) T’ [+YNQ] 2 [+PRESUP] T 1 (can) [+YNQ] [+PRESUP] The [wh-Q] feature moves overtly. 4. Embedded Question Inversion and [-PRESUP] We have given one example of variation in question inversion in AAE, which is linked to features on a node, not to the features that define a node. In addition to matrix question inversion, AAE also has embedded question inversion, as in the examples in (12) (from Green (2002)): 12. a) It’s gonna ask you [do you /if you wanna make a transfer]. ‘It’s going to ask you if you want to make a transfer’ (reference to automatic teller machine) b) We trying to see [can we/if we can work out March].2 2 In some environments in AAE, the auxiliary be (and copula) are not overtly expressed. 6 ‘We are trying to determine whether we can work out [the dates for] March’ c) I wonder [do it/if it be like the water we drink].3 ‘I wonder if it generally like the water we drink’ As in matrix yes-no questions, we posit a [-PRESUP] feature on AUX and on the Force node in embedded questions: 13. VP 1 V’ 2 V ForceP = (CP) 2 Spec, ForceP Force’ 2 Force [+YNQ] [-PRESUP] 2 TP 2 TP 2 T 1 AUX T’ [+AUX] [-PRESUP] Not only is embedded inversion in AAE permissible, it is obligatory if, in its absence, Force (=C) would be empty. Consider the ungrammatical instances of non-inversion in (14): 14. a) *It’s gonna ask you [__ you DO wanna make a transfer].4 b) *We trying to see [__ we can work out March]. c) *I wonder [__it DO be like the water we drink]. The [-PRESUP] feature cannot be checked by intonation in embedded clauses. The different ways of checking the [-PRESUP] feature (by auxiliary movement and intonation) in matrix clauses led to variation, but the checking options are not available in embedded clauses, so there is no variation or optionality in embedded question inversion. The checking options for features in matrix and embedded questions are summarized in the chart below: 3 This sentence contains aspectual (or habitual) be, which indicates that an eventuality recurs (Green 2000). 4 The stressed auxiliary do is included in T in (14a, c) to show that do support has occurred, but auxiliary inversion has not. 7 15. Summary of Inversion in Matrix and Embedded Questions in AAE Question-type Matrix Question Embedded Question [-PRESUP] Checking Options 1. AUX to Force movement (e.g., Can she leave?) 2. Intonation (e.g., She can leave?) 1. AUX to Force movement (e.g., I wonder can she leave.) 2. Complementizer in Force (e.g., I wonder if she can leave.) 3. *Intonation (e.g., *I wonder she can leave.) 5. Negative Inversion and the Stable [NegFoc] Feature Another construction that exemplifies a stable feature on a node is negative inversion (NI) in AAE. We will argue that a dialect-specific, but stable node is required to capture these facts, following Green (2007). 16. Don’t nobody be working on the 10th floor. ‘Not a single person is usually working on the 10th floor. These negative concord constructions, which are characterized by an initial negated auxiliary followed by a negative indefinite DP, superficially resemble interrogatives but are interpreted as declaratives. In addition, the preposed negated auxiliary widens the domain of quantification of the negative indefinite DP (along the lines of widening in Kadmon and Landman (1993)). The NI structures in AAE are generated in a Rizzi-type (2004) split CP construction, in which the CP-node is split into a separate FocusP and ForceP (Green 2007). The analysis here is that the negated auxiliary is attracted to Focus (=C) by a negative focus feature [NegFoc] that defines the FocusP node and that is linked to a widened domain reading, and the negative indefinite DP is in Spec, TP. 8 17. ForceP 2 2 Force Force’ FocusP 2 Spec, Focus Focus’ [NegFoc] 2 Focus TP 1 2 don’t DP T’ [+NegFoc] 2 nobody 1 T (don’t) This [NegFoc] feature analysis accounts for negative scope and the order of the negated auxiliary followed by a negative indefinite DP, and it also explains why this type of inversion only occurs with negated auxiliaries. Inversion of the negated auxiliary is obligatory in the presence of the [NegFoc] feature. The prediction, then, is that NI constructions and non-inversion negative constructions are not always variants. That is, if the [NegFoc] feature is present, then inversion will necessarily apply. The only way to check the feature is by inversion. There is no optional movement giving way to (18a). Compare the non-inversion construction in (18a) to the NI construction in (16), repeated as (18b): 18. a) Nobody don’t be working on the 10th floor. ‘Nobody is usually working on the 10th floor’ b) Don’t nobody be working on the 10th floor. One might advocate an account of the sentences in (18) as variants, and of (18b) as being generated by the optional movement of the negated auxiliary (18a) to the head of FocusP. However, such an analysis would be analogous to optional subject-auxiliary inversion in yes-no questions. A strong argument against the optional inversion analysis is that the NI and non-inversion constructions can differ, such that the NI construction has a widening effect in the terms of the any analysis in Kadmon and Landman (1993): “In an NP of the form any CN, any widens the interpretation of the common noun phrase along a contextual dimension” (p. 361). They note that the generic sentences in (19) state a general rule about owls, but the sentence in (19b) conveys less tolerance of exceptions to the general rule: 19. a) An owl hunts mice. b) Any owl hunts mice. (p. 361) NI constructions have a similar effect, so nobody in the non-inversion construction (19a) may very well refer to ‘not many people,’ but nobody in the NI construction (18b) refers to an even smaller set. Given the two different readings, the non-inversion and NI constructions are not taken to be variants of each other. 9 An indirect argument for the [NegFoc] analysis is based on reports about the NI and non-inversion constructions in other varieties of English. For instance, Foreman (1999), in his discussion about NI in West Texas English, reports that not all non-inverted constructions are acceptable. While this claim has not been made for AAE, it is still an important one to consider. If the NI and non-inversion constructions were variants, it would seem that the corresponding non-inverted constructions would be acceptable in at least some environments. This observation is directly linked to the claim that NI constructions differ in semantics/pragmatics from the ‘corresponding’ non-inversion constructions. The [NegFoc] movement analysis raises questions about the landing site of the preposed auxiliary, especially in instances of embedded NI introduced by a complementizer: 20. a) The teachers know (that) can’t nobody leave. ‘The teachers know that not a single person can leave’ b) The teachers know *(if) can’t nobody leave. ‘The teachers know if not a single person can leave’ Under the cartography of syntactic structures as proposed in Rizzi (2004) which is assumed here, the complementizer (that, if) and the negated auxiliary ([NegFoc]) have separate CP positions. The complementizer (that, if) is generated in ForceP, and the negated auxiliary inverts to the Focus head. The presence of the separate positions leads to a crucial prediction of this account: the two nodes should be compatible. We turn to that prediction in a moment. It is interesting to consider the embedded NI constructions in light of the [-PRESUP] feature that is associated with the T and Force nodes. The analysis of yes-no questions was that auxiliary inversion in those constructions was triggered by a [-PRESUP] feature; however, NI is triggered by a [NegFoc] feature. The prediction is that both features should be able to appear together because they are associated with different nodes. The data support this claim: 21. a) The teachers know [(that) can’t[NegFoc] nobody leave]. b) The teachers know [if[-PRESUP] can’t[NegFoc] nobody leave]. c) *The teachers know [(if)[-PRESUP] can’t[NegFoc] nobody leave]. 22. a) The teachers don’t know [(that) can’t[NegFoc] nobody leave]. b) The teachers don’t know [if[-PRESUP] can’t[NegFoc] nobody leave]. c) *The teachers don’t know [(if)[-PRESUP] can’t[NegFoc] nobody leave]. The sentences in (21) and (22) show that the [-PRESUP] and [NegFoc] features are satisfied differently. Inverted auxiliaries can satisfy the [-PRESUP] feature in yes-no questions, but they do not seem to be able to satisfy the [-PRESUP] feature in NI constructions. The feature is satisfied by the complementizer if. For instance, the sentences in (21c) and (22c) are ungrammatical with an optional if because it is needed to satisfy the [-PRESUP] feature. Those sentences strongly suggest that the negated auxiliary cannot satisfy the [NegFoc] feature and then move to Force to satisfy the 10 [-PRESUP] feature. It is not clear what rules this out, unless the negated auxiliary must stay in FocusP because it is satisfying a feature that defines a node. In the account of questions above, it was noted that wh-movement is obligatory, but the movement of the auxiliary to C in matrix questions is optional. Wh-movement and negated auxiliary inversion in wh-NI constructions are both obligatory because the whfeature and the [NegFoc] feature define the nodes with which they are associated, respectively. 23. a) Why don’t[NegFoc] nobody read those books? ‘Why is it that not a single person reads those books?’ b) Why nobody don’t read those books? ‘Why doesn’t anybody read those books?’ These wh-question negative inversion constructions also provide evidence that negated auxiliary inversion is due to a negative feature, not to a [YNQ] feature. In the sentence in (24a), while what undergoes wh-movement, the auxiliary does not undergo question inversion. The sentence is acceptable in AAE because there is an additional position (Focus) for the auxiliary, but the corresponding sentence in MAE is ungrammatical: 24. a. AAE: What don’t[NegFoc] nobody want? ‘What is it that not a single person wants?’ b. MAE: ?*What does nobody want? The claim is that there is no negative auxiliary inversion in MAE. Sobin (2003) presents a non-movement analysis of negative inversion in MAE, as in Never again will he eat raw spaghetti., which has also received a wh-question account in previous analyses. In brief, the wh- analysis is that the negative phrase undergoes movement similar to wh-movement, and the auxiliary also undergoes movement. He notes that whereas children acquire wh-question formation relatively early, there is no indication of NI in early child language development. According to Sobin, such acquisition patterns suggest that there are differences between question inversion and NI and that the latter should not be analyzed as resulting from inversion or auxiliary movement. In considering NI in the acquisition of AAE, we find that child AAE speakers seem to acquire NI relatively late. While developing 3-, 4-, and 5-year-old AAE-speaking children are able to produce and comprehend negative concord constructions such as that in (25), they do not seem to have NI:5 25. They don’t have no training wheels. (T086, 4;6) ‘They don’t have training wheels’ The analysis that we have put forward helps to explain why children might acquire NI later, even after they have acquired auxiliary inversion and wh-questions. It might be that 5 The negative concord data are based on comprehension and production tasks with 3-, 4-, and 5-year-olds in Southwestern Louisiana. 11 stable split nodes—in this case split CP—may be acquired later; however, once the split node is acquired, it is stable. We have distinguished the feature variation connection and the node nonvariation connection, as summarized below: 26. Feature-Variation Connection and Node-Non-Variation Connection Inversion-Type Question Inversion Wh-Questions Negative Inversion Feature [PRESUP] [Wh-Q] [NegFoc] Defines a Node No Yes Yes Variation Yes No No The link is that variation is associated with unstable features on a node, not with stable features that define a node. 6. Can CP’s Carry Different Features?: Acquisition Across Languages We assume, of course, that the tree skeleton in all grammars is universal. The core properties of a high CP (carrying Force) and VP have invariant properties which, we can assume, are not learned. Nevertheless, there remains the question of the exact feature content of such nodes. A discussion in recent years has arisen over whether CP carries Tense (as argued by Pesetsky and Torrego (2001) and Koster (2003)). Koster and Chomsky suggest that the tense under TP may be a way station, reflecting partial movement, for a universal move to CP. In particular, it has been argued that the CP in German carries Tense as a property of the node itself. How does our approach connect to such claims? We argue that the feature content of nodes can vary. If the Tense feature is linked to the CP as a reflection of its head properties, then we predict that overt movement will be obligatory, just like whmovement. If in German it is correct that Tense is linked to CP, then V2 is predictably obligatory. Thus, while V2 appears to be like inversion in English, it actually involves obligatory overt movement because a Tense head feature is involved. If CP has Tense as the head feature, then V2 in German CP [+Tense]: 27. CP 2 Spec, CP C’ 2 C [+Tns] 1 V [+Tns] If children realize the presence of Tense on CP, then movement to CP should be stable in wh-questions. It is not the [PRESUP] feature that is motivating movement, but the Tense feature. In fact, Roeper (1973) revealed a dramatic contrast between German and English acquisition both in naturalistic data and in experimentation. There are no reported cases of the form (*” = unreported in naturalistic data), and the unmoved case is quite rare: 28. a) *” Was er kann tun/ what he can do 12 b) rare: Was er tun kann/what he do can In English, by contrast, where the wh-word satisfies the [Q] feature, and auxiliary inversion marks the non-obligatory [PRESUP] feature, we find extensive variation over many years in all children reported upon (Brown (1973), deVilliers (1991)). Conroy (2006) also points to the possibility that meaning differences may arise: “what he can do” “why we are daddies” “why that’s a little piece of foil” Non-inversion for what/where/when is almost universal in the early years. For why, noninversion is common until 6 years. This is predictable on our account. One naturally asks why about something that is presupposed. This is generally true for other wh-questions; however, it is not always true. Thus we have a contrast between (29a, b): 29. a) Who did he see? possible answer: “nobody” b) Who was it that he saw? not possible: “nobody” While (29b) might allow the “nobody” answer under strange circumstances, all speakers sense that (29b) has a presupposition that someone was seen which is not present for (29a). The hypothesis, then, emerges that English-speaking children who use the expression “what he can do” presuppose that he can do something, while the adult requires the inverted form what can he do? to drop this presupposition and possibly allow the answer “nothing.” It follows that children who fail to carry out inversion may have a slightly different meaning in mind. We conclude that German has a stable Tense feature in CP while English has an unstable [PRESUP] feature. Conclusion Our proposals have a number of deep implications which we will point toward but not resolve. In a word, the architecture of trees may be engaged in interesting ways. Proposals by Frampton (2004), Giorgi and Panesi (1997), about Feature-sharing and Multi-dominance (Gärtner, among others), suggest an interaction between nodes which reflect head features and those which are subject to agreement relations under ccommand. (See Green and Roeper (to appear) for discussion.) We can summarize our claims with respect to acquisition and variation with two strong, related claims: 1) Non-universal nodes are difficult to acquire. 2) Corollary: Nodes are difficult to change. For instance, particles appear in two positions in English, which seems redundant: a) I picked the baby up. b) I picked up the baby. 13 By our hypothesis, they should be stable and in fact have changed little over hundreds of years. In addition, there is evidence that they are not acquired as easily as argument structure. (See Snyder and Roeper (2004)). We have argued in behalf of a simple distinction between stable nodes and unstable features, which has deep theoretical and empirical implications. We have illustrated these concepts with both a classic example, the tense marker –s verb raising, and with original examples from AAE, which involve different types of inversion. These theoretical claims receive direct support from acquisition, dialect variation, and patterns of disorders and processing. We take these diverse empirical sources to be the strongest evidence in behalf of the theory. The overarching argument, however, can be seen as a natural extension of current minimalist theory, in which tree structure has become more abstract and feature bundles more refined. References Adger, David and George Tsoulas (2004) Circumstantial Adverbs and Aspect, University of York ms. Brown, Roger (1973) A First Language. Cambridge: Harvard University Press. Cheng, Lisa and Johan Rooryck (2000) Types of Wh-in-situ” Leiden University ms. Chomsky, Noam (1991) Some notes on economy of derivation and representation. In R. Friedin (ed.), Principles and Parameters in Comparative Grammar. Cambridge: MIT Press. Chomsky, Noam (1995) The Minimalist Program, Cambridge: MIT Press. Chomsky, Noam (2005) Three Factors in language design, Linguistic Inquiry 36, 1-22. Chomsky, Noam (2006) Approaching UG from Below. In Uli Sauerland and HansMartin Gärtner (eds.), Interfaces + Recursion = Language?: Chomsky’s Minimalism and the View from Syntax-Semantics. Berlin: Mouton de Gruyter. Conroy, Anastasia (2006) The Semantics of How Come: A Look at How Factivity Does It All. In N. Kazanina, U. Minai, P. J. Mohanan, and H. L. Taylor (eds.) University of Maryland Working Papers in Linguistics, Volume 14, College Park. deVilliers, J. (1991) Why Questions. In Tom Maxfield and Bernadette Plunkett (eds.), The Acquisition of wh, University of Massachusetts Occasional Papers in Linguistics, Amherst, MA. Deevy, P. (2000) Agreement checking in comprehension: Evidence from relative clauses. Journal of Psycholinguistic Research 29, 69-79. University 14 Emonds, Joseph (1980) Word order in generative grammar, Journal of Linguistic Research 1:33-54. Foreman, John (1999) Syntax of negative inversion in non-standard English. In Kimary Shahin, Susan Blake, and Eun-Sook Kim (eds.), The Proceedings of the Seventeenth West Coast Conference on Formal Linguistics. Stanford: CSLI. Frampton, John (2004) Copies, Traces, Occurrences, and all that: Evidence from Bulgaria multiple wh-phenomena, Northeastern University ms. Gärnter, Hans-Martin (2002) Generalized Transformations and Beyond: Reflections on Minimalist Syntax. Berlin: Academie Verlag. Giorgio, Alessandra and Fabio Pianesi (1997) Tense and Aspect: From Semantics to Morphosyntax. Oxford: Oxford University Press. Green, Lisa J. (2002) African American English: A Linguistic Introduction. Cambridge: Cambridge University Press. Green, Lisa (2000) Aspectual be-type constructions and coercion in African American English, Natural Language Semantics 8, 1-25. Green, Lisa (2007) Negative inversion and negative focus in African American English, University of Massachusetts ms. Green, Lisa and Thomas Roeper (2007) The acquisition path for tense-aspect: Remote past and habitual in child African American English, Language Acquisition 14, 269-313. Kadmon, Nirit and Fred Landman (1993) Any, Linguistics and Philosophy 16, 353-422. Koster, Jan (2003) Pied Piping and the Word Orders of English and Dutch” University of Groningen ms. Pollock, Jean-Yves (1989) Verb-movement, universal grammar and the structure of IP, Linguistic Inquiry 20:365-424. Rizzi, Luigi (ed.) (2004) The Fine Structure of CP and IP: The Cartography of Syntactic Structures, Volume 2. Oxford: Oxford University Press. Pesetsky, David and Esther Torrego (2001) T-to-C movement: Causes and consequences. In M. Kenstowicz (ed.), Ken Hale: A Life in Language. Cambridge, MA: MIT Press. Roeper, T. (1973) Theoretical implications of word order, topicalization and inflections in 15 German language acquisition." In Charles A. Ferguson & Dan Slobin (eds.), Studies in Child Language Development. New York: Holt, Rinehart and Winston. Roeper, Thomas and Lisa Green (to appear) Node-labels and features: stable and unstable dialects and variation in Acquisition” Yearbook of Variation. Snyder, William (1995) Language Acquisition and Language Variation: The Role of Morphology. Doctoral dissertation, MIT, distributed by MIT Working Papers in Linguistics, Cambridge, MA. Snyder, William and Thomas Roeper (2004) Learnability and recursion across categories. In: A. Brugos, L. Micciulla and C. Smith (eds.) BUCLD 28: Proceedings of the 28th Annual Boston University Conference on Language Development. Somerville, Massachusetts: Cascadilla Press. 16