From mcginnis at ucalgary.ca Wed Dec 1 16:21:06 1999 From: mcginnis at ucalgary.ca (Martha McGinnis) Date: Wed, 1 Dec 1999 09:21:06 -0700 Subject: Carson Schutze: my current work in DM-spirited morphology Message-ID: Hi everyone, At Martha's encouragement, I'm posting a little summary of things I'm working on or thinking about that involve morphology "in the spirit of DM". What I mean is, they don't rely on much/any of the sort of technicalia that have been discussed here recently (those paradigms scare me!), but they do crucially rely on having a morphological component of the DM sort, namely, one that is post-syntactic, (mostly?) interpretive rather than filtering, uses underspecification, and is able to deal with genuinely syntactic representations, relations, etc. These ideas are in various states of development, between half-baked and burned to a crisp, but comments on any/all would be most welcome. Those that are written up, I will be happy to send to people. 1) Default case. I have a manuscript on this, arguing that the right notion of default case can explain lots of otherwise puzzling facts about morphological case marking both in English and crosslinguistically. (And that most other notions of default case in the literature are nonsensical.) The account can be seen as working out and further supporting the basic view of default case from Marantz 1991 (ESCOL). In a nutshell, morphology provides a last-resort way to spell out nominals that fail to acquire case features by case assignment, concord, or any other "syntactic" means. Languages vary in which forms they pick for this purpose, and perhaps even choose individual items idiosyncratically from different parts of the paradigm. This explains why even very close dialects differ in their case marking choices in certain corners of the syntax, and perhaps more importantly, allows even the English pronominal case system to be seen as behaving in a pretty "well behaved" way, relative to "rich" case languages. 2) Agreement and its "maximization". Those familiar with my dissertation will know that a central proposal was a cross-derivational syntactic principle called "Accord Maximization", which governed the insertion of case and agreement features into a syntactic tree. (Those not familiar with my dissertation, MITWPL will be happy to sell you a copy right away! :-) The power of the cross-derivational comparisons was always a worry with this proposal, however, so I'm now working out how Accord Maximization could be implemented entirely in the morphological component, on the view (again from Marantz 91) that all of morphological case and agreement might be introduced there, i.e., postsyntactically (in the narrowest sense of syntax). The trick is to make Vocabulary Insertion do a bit more work than it standardly does. Specifically, when figuring out which item to insert in a slot for an agreement affix, say, V.I. needs to go hunt for an eligible set of features to express (eligibility being determined in terms of syntactic notions like command and locality). Since V.I. tries to insert the most highly specified items first, it will always mark agreement if possible, but if there are no eligible things to agree with, a default affix will be inserted. I think this captures the generalization, "If you can agree you must, but if not it's OK", which characterizes the data I was looking at, and which is very hard to implement in Minimalist feature-checking terms. 3) "Semantically empty lexical heads." This is in manuscript form as well. The central claim is that not just affixes but also lexical heads (N,V,A, maybe P) have default vocabulary items, and that taking seriously their status as lacking encyclopedic meaning but still being genuinely lexical leads to some interesting new accounts of well-known facts involving "pro forms" ('one', 'so') and expressions of predication structures ('be', 'with'). The account crucially invokes and distinguishes several factors that can trigger the use of these empty heads, including the need for a syntactic attachment site and the need for clitics to be morphologically supported. Further factors come into play to trigger insertion of 'be', and these factors are distinguished from those that invoke dummy 'do', which is treated as of category Mood/Modal rather than V, thereby capturing the intuition that both 'be' and 'do' are dummies, but are not interchangeable. 4) The English auxiliary system. These are ideas that are like raw dough, having not even risen for the first time yet, combined with aspects of the account of the aux system in section 5.2 of my dissertation (which I believe is the most detailed working out of that within DM). There are many pieces to this story, of which I'll mention just a couple. One is to make crucial use of the claim that external arguments are projected by a "little v" that heads a syntactic projection higher than the object Case position. If verbs (V) have to hook up with little v by overt head movement, we can explain many otherwise paradoxical facts about the relative position of nonsubject NPs and verbs in paradigms like "They had been forecasting a storm." vs. "There had been a storm being forecast(ed)." I claim that "a storm" is in the same position in both sentences, and that properties of v, V, and affixes arrange the various participles around that position. A second part of this story is the treatment of (auxiliary) 'have' and 'be', which (following the aforementioned paper) I view essentially as V heads with no semantic content. I think one can argue that this tack provides an explanation for why just these two verbs can "raise over negation" when finite in English--in fact, no verb can raise over English Neg, but 'have' and 'be' can be "generated" above Neg when they are finite, because of their semantic emptiness--an implication (with important modifications) of Pollock's old idea. The general picture that comes out of this line of thinking is that English has way more V-raising than is generally assumed. A side conclusion is that its impersonal passives (exemplified above) are in no way weird, relative to other languages, contra what Chomsky has said recently. Thanks for listening! ---------------------------------------------------------------------- ---------- Carson T. Schutze Department of Linguistics, UCLA Email: cschutze at ucla.edu Box 951543, Los Angeles CA 90095-1543 U.S.A. Office: Campbell Hall 2224B Deliveries/Courier: 3125 Campbell Hall Campus Mail Code: 154302 Web: http://www.humnet.ucla.edu/people/cschutze Phone: (310)825-9887 Messages: (310)825-0634 Fax: (310)206-8595 From mcginnis at ucalgary.ca Wed Dec 1 16:21:06 1999 From: mcginnis at ucalgary.ca (Martha McGinnis) Date: Wed, 1 Dec 1999 09:21:06 -0700 Subject: Carson Schutze: my current work in DM-spirited morphology Message-ID: Hi everyone, At Martha's encouragement, I'm posting a little summary of things I'm working on or thinking about that involve morphology "in the spirit of DM". What I mean is, they don't rely on much/any of the sort of technicalia that have been discussed here recently (those paradigms scare me!), but they do crucially rely on having a morphological component of the DM sort, namely, one that is post-syntactic, (mostly?) interpretive rather than filtering, uses underspecification, and is able to deal with genuinely syntactic representations, relations, etc. These ideas are in various states of development, between half-baked and burned to a crisp, but comments on any/all would be most welcome. Those that are written up, I will be happy to send to people. 1) Default case. I have a manuscript on this, arguing that the right notion of default case can explain lots of otherwise puzzling facts about morphological case marking both in English and crosslinguistically. (And that most other notions of default case in the literature are nonsensical.) The account can be seen as working out and further supporting the basic view of default case from Marantz 1991 (ESCOL). In a nutshell, morphology provides a last-resort way to spell out nominals that fail to acquire case features by case assignment, concord, or any other "syntactic" means. Languages vary in which forms they pick for this purpose, and perhaps even choose individual items idiosyncratically from different parts of the paradigm. This explains why even very close dialects differ in their case marking choices in certain corners of the syntax, and perhaps more importantly, allows even the English pronominal case system to be seen as behaving in a pretty "well behaved" way, relative to "rich" case languages. 2) Agreement and its "maximization". Those familiar with my dissertation will know that a central proposal was a cross-derivational syntactic principle called "Accord Maximization", which governed the insertion of case and agreement features into a syntactic tree. (Those not familiar with my dissertation, MITWPL will be happy to sell you a copy right away! :-) The power of the cross-derivational comparisons was always a worry with this proposal, however, so I'm now working out how Accord Maximization could be implemented entirely in the morphological component, on the view (again from Marantz 91) that all of morphological case and agreement might be introduced there, i.e., postsyntactically (in the narrowest sense of syntax). The trick is to make Vocabulary Insertion do a bit more work than it standardly does. Specifically, when figuring out which item to insert in a slot for an agreement affix, say, V.I. needs to go hunt for an eligible set of features to express (eligibility being determined in terms of syntactic notions like command and locality). Since V.I. tries to insert the most highly specified items first, it will always mark agreement if possible, but if there are no eligible things to agree with, a default affix will be inserted. I think this captures the generalization, "If you can agree you must, but if not it's OK", which characterizes the data I was looking at, and which is very hard to implement in Minimalist feature-checking terms. 3) "Semantically empty lexical heads." This is in manuscript form as well. The central claim is that not just affixes but also lexical heads (N,V,A, maybe P) have default vocabulary items, and that taking seriously their status as lacking encyclopedic meaning but still being genuinely lexical leads to some interesting new accounts of well-known facts involving "pro forms" ('one', 'so') and expressions of predication structures ('be', 'with'). The account crucially invokes and distinguishes several factors that can trigger the use of these empty heads, including the need for a syntactic attachment site and the need for clitics to be morphologically supported. Further factors come into play to trigger insertion of 'be', and these factors are distinguished from those that invoke dummy 'do', which is treated as of category Mood/Modal rather than V, thereby capturing the intuition that both 'be' and 'do' are dummies, but are not interchangeable. 4) The English auxiliary system. These are ideas that are like raw dough, having not even risen for the first time yet, combined with aspects of the account of the aux system in section 5.2 of my dissertation (which I believe is the most detailed working out of that within DM). There are many pieces to this story, of which I'll mention just a couple. One is to make crucial use of the claim that external arguments are projected by a "little v" that heads a syntactic projection higher than the object Case position. If verbs (V) have to hook up with little v by overt head movement, we can explain many otherwise paradoxical facts about the relative position of nonsubject NPs and verbs in paradigms like "They had been forecasting a storm." vs. "There had been a storm being forecast(ed)." I claim that "a storm" is in the same position in both sentences, and that properties of v, V, and affixes arrange the various participles around that position. A second part of this story is the treatment of (auxiliary) 'have' and 'be', which (following the aforementioned paper) I view essentially as V heads with no semantic content. I think one can argue that this tack provides an explanation for why just these two verbs can "raise over negation" when finite in English--in fact, no verb can raise over English Neg, but 'have' and 'be' can be "generated" above Neg when they are finite, because of their semantic emptiness--an implication (with important modifications) of Pollock's old idea. The general picture that comes out of this line of thinking is that English has way more V-raising than is generally assumed. A side conclusion is that its impersonal passives (exemplified above) are in no way weird, relative to other languages, contra what Chomsky has said recently. Thanks for listening! ---------------------------------------------------------------------- ---------- Carson T. Schutze Department of Linguistics, UCLA Email: cschutze at ucla.edu Box 951543, Los Angeles CA 90095-1543 U.S.A. Office: Campbell Hall 2224B Deliveries/Courier: 3125 Campbell Hall Campus Mail Code: 154302 Web: http://www.humnet.ucla.edu/people/cschutze Phone: (310)825-9887 Messages: (310)825-0634 Fax: (310)206-8595