From mcginnis at UCALGARY.CA Thu Feb 26 19:13:12 2004 From: mcginnis at UCALGARY.CA (Martha McGinnis) Date: Thu, 26 Feb 2004 12:13:12 -0700 Subject: hello, DM-Listers! Message-ID: Dear DM-Listers, Whoa, has this list been quiet! I guess everyone is too busy doing research to post queries. I'd like to take this opportunity to encourage you to update your fellow listers on what you've been working on, morphologically speaking. Don't be shy! I know some of you have been writing busily away. Also, please take a moment to raise any questions you might have. Here's my contribution... I've recently been working on morphosyntactic feature geometries, a la Harley & Ritter (see their 2002 Language paper). This work has got me thinking about where morphosyntactic feature geometries should fit into the grammar. H&R don't focus on this, because they want to be as theory-neutral as possible. But once you start to think of it, it's not completely obvious. Let's begin with the observation that morphosyntactic feature geometries can be somewhat unspecified. For example, a language with a special "dual" number category would have a special featural representation (with both [Minimal] and [Group] specified), while the plural number category would have only [Group] specified. On the other hand, a language without a special dual category would also represent dual number as just [Group]. So how does a given grammar determine the relation between representations and interpretations? It seems as though the answer is this: it uses the most specific morphosyntactic category compatible with the intended meaning. If there's a [Minimal]+[Group] category, it uses that for the dual; if not, it uses just [Group]. On the other hand, if the meaning is plural, [Minimal] can't be used, so only [Group] is. Now, this sounds a lot like Vocabulary competition in DM, which inserts the most specific Vocabulary item compatible with the fully specified syntactic/semantic features of the syntactic node (which in turn determine the meaning). Suppose then that H&R's privative feature geometries are morphosyntactic representations of Vocabulary items. This is more or less Bonet's approach, if I understand it correctly. If so, then how are the features of *syntactic nodes* represented? (Bonet's dissertation leaves this issue somewhat open.) For example, how do we know that the plural representation can't be used to represent the dual meaning in a language that has a special dual category? Does the syntactic node have a binary-featured representation, like [+Group, -Minimal]? If not, then what? Does the semantics fill in default feature values like [-Minimal]? I'm sure some of you have thought this through much more carefully than I've been able to, as yet. Please let me know what you think. References welcome! All the best, Martha From mcginnis at UCALGARY.CA Thu Feb 26 19:25:02 2004 From: mcginnis at UCALGARY.CA (Martha McGinnis) Date: Thu, 26 Feb 2004 12:25:02 -0700 Subject: morphosyntactic feature geometries Message-ID: Dear DM-Listers, [Please forgive re-posting: it seems some mailservers are filtering out messages with "hello" in the title.] Whoa, has this list been quiet! I guess everyone is too busy doing research to post queries. I'd like to take this opportunity to encourage you to update your fellow listers on what you've been working on, morphologically speaking. Don't be shy! I know some of you have been writing busily away. Also, please take a moment to raise any questions you might have. Here's my contribution... I've recently been working on morphosyntactic feature geometries, a la Harley & Ritter (see their 2002 Language paper). This work has got me thinking about where morphosyntactic feature geometries should fit into the grammar. H&R don't focus on this, because they want to be as theory-neutral as possible. But once you start to think of it, it's not completely obvious. Let's begin with the observation that morphosyntactic feature geometries can be somewhat unspecified. For example, a language with a special "dual" number category would have a special featural representation (with both [Minimal] and [Group] specified), while the plural number category would have only [Group] specified. On the other hand, a language without a special dual category would also represent dual number as just [Group]. So how does a given grammar determine the relation between representations and interpretations? It seems as though the answer is this: it uses the most specific morphosyntactic category compatible with the intended meaning. If there's a [Minimal]+[Group] category, it uses that for the dual; if not, it uses just [Group]. On the other hand, if the meaning is plural, [Minimal] can't be used, so only [Group] is. Now, this sounds a lot like Vocabulary competition in DM, which inserts the most specific Vocabulary item compatible with the fully specified syntactic/semantic features of the syntactic node (which in turn determine the meaning). Suppose then that H&R's privative feature geometries are morphosyntactic representations of Vocabulary items. This is more or less Bonet's approach, if I understand it correctly. If so, then how are the features of *syntactic nodes* represented? (Bonet's dissertation leaves this issue somewhat open.) For example, how do we know that the plural representation can't be used to represent the dual meaning in a language that has a special dual category? Does the syntactic node have a binary-featured representation, like [+Group, -Minimal]? If not, then what? Does the semantics fill in default feature values like [-Minimal]? I'm sure some of you have thought this through much more carefully than I've been able to, as yet. Please let me know what you think. References welcome! All the best, Martha From dan.everett at MAN.AC.UK Thu Feb 26 19:52:40 2004 From: dan.everett at MAN.AC.UK (Daniel L. Everett) Date: Thu, 26 Feb 2004 19:52:40 +0000 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: > Does the syntactic node have a binary-featured > representation, like [+Group, -Minimal]? If not, then what? Does > the semantics fill in default feature values like [-Minimal]? > This is an interesting issue. In their new book on agreement, Wechsler & Zlatic (2003, The many faces of agreement, CSLI) address some correspondences between semantics and morphology. Not specifically what Martha asks about, but an interesting discussion of the issues in any case. Their distinction between INDEX (semantic-based agreement) and CONCORD (morphological feature-based agreement - actually morphosyntactic features in their system) seems quite useful and could be relevant here as well. One would, ceteris paribus, expect number (in their system number is found in both CONCORD and INDEX) to match between morphology and semantics. So I wonder if this would predict different semantic representations for languages with/without 'dual'? As for the syntax-morphology connection, wouldn't 'unification' predict that the morphology would either (i) fill in missing values or that (ii) the syntax nodes unspecified for certain features could allow some features to play a morphological role, but no syntactic one (i.e. that feature-matching between syntax and morphology would not be forced if there is no incompatibility)? Or is (ii) ruled out in DM? Besides compositional analysis of morphosyntactic nodes, is anyone looking at agreement from this Harley-Ritter perspective? This would seem an ideal area to look for answers to Martha's queries. -- Dan ------------------------------------------ Daniel L. Everett Professor of Phonetics & Phonology Postgraduate Programme Director Department of Linguistics The University of Manchester Oxford Road Manchester, UK M13 9PL http://ling.man.ac.uk/info/staff/de Fax: 44-161-275-3187 Office: 44-161-275-3158 From rnoyer at BABEL.LING.UPENN.EDU Thu Feb 26 23:22:38 2004 From: rnoyer at BABEL.LING.UPENN.EDU (Rolf Noyer) Date: Thu, 26 Feb 2004 18:22:38 -0500 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Hi Martha et. al, Strangely coincidental: I too was looking through and thinking a bit about Harley & Ritter's paper in Language just yesterday afternoon. When I wrote my dissertation I concluded that there were no such geometries, although I toyed with them quite a bit. I have not, but would like to get a chance to, think carefully about H&R's arguments and decide if I wish to recant or put up a fight. I too am a bit confused about the place such representations would occupy in the grammar. Sets of morphosyntactic features might be geometrically related in a syntactic node for which vocabulary items compete, or they might be geometrically related in vocabulary items which compete for insertion into those nodes. If the latter, we might expect that person and gender could not be encoded in a single vocabulary item without gender. In other words, we should never expect, for example, [2 pl f] to split into two pieces: [2 f] and [pl]. More generally, we should, I believe, expect that any given vocabulary item should encode a subgraph of the entire graph (= tree) of features. Notice that this doesn't mean "subtree" since, as I pointed out, if this were so one could not have a vocabulary item expressing [2] without also expressing gender and number at the same time, clearly a false prediction in many languages, as for example Arabic in which t-aktub-na spells out [2 pl f], with /t-/ as [2] and /-na/ as [pl f] If the former, we might expect certain agreement phenomena, which presumably examine the syntactic nodes themselves and not the vocabulary items which are inserted, to be subject to constraints which would find explanation in a geometry of the features involved. Martha says: So how does a given grammar determine the relation between representations and interpretations? It seems as though the answer is this: it uses the most specific morphosyntactic category compatible with the intended meaning. When I last thought about this my attitude was that one should view the matter of interpretation (just like everywhere else) as combining semantic entailments with functional inferences. The idea that plurals in a language with duals mean "three or more" does not strike me as being the result of a specifically interpreted "minus dual" value of some sort, but rather a blocking effect. If the speaker had meant two (s)he would have used a dual; thus using the plural implies three or more. As a pragmatic or functional inference however, one might expect this to be cancellable in the appropriate discourse context. Here my knowledge of the phenomenon ends; it would be helpful to find out if plurals can be used to refer to two individuals in a language with a dual category if the discourse context is suitably constructed, such as, for example, a situation in which all that was discourse-relevant was that the argument in question was not individual, but at least two. Since I am committed (as I think many of us are) to a semantics which interprets the syntax rather than generates it, I don't understand what it means to say "it uses the most specific morphosyntactic category compatible with the intended meaning" except as a kind of pragmatic rule of the Gricean sort "be as informative as possible". I'm not an expert on this literature but I seem to recall that Larry Horn has worked on this problem for number and certain quantifiers. So for example, asserting "three men are in the room" seems to strongly imply that "no more than three men are in the room" but it does not entail it. Obviously these remarks are really off the cuff but maybe we can get some dialogue going. Best regards, Rolf From mcginnis at UCALGARY.CA Sat Feb 28 19:50:45 2004 From: mcginnis at UCALGARY.CA (Martha McGinnis) Date: Sat, 28 Feb 2004 12:50:45 -0700 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Rolf, thanks very much for your response! It's true, the two possibilities (feature geometries = Vocabulary, feature geometries = syntax) should indeed make the different empirical predictions you mentioned. The syntactic agreement predictions might be hard to test, since agreement Vocabulary should be subject to the same feature geometry as pronoun Vocabulary... but one can imagine syntactic distinctions between e.g. number+gender agreement (as on French participles) vs. person+number agreement (as on French finite verbs) -- e.g. in French, the former type of syntactic agreement relation doesn't seem to block further syntactic movement of the nominal, while the latter type does. If H&R's geometry is correct, we might expect to find other syntactic evidence for these kinds of agreement splits, rather than a person+gender vs. number split. >The idea that plurals in a language with duals mean "three or more" >does not strike me as being the result of a specifically interpreted >"minus dual" value of some sort, but rather a blocking effect. If >the speaker had meant two (s)he would have used a dual; thus using >the plural implies three or more. As a pragmatic or functional >inference however, one might expect this to be cancellable in the >appropriate discourse context. Good point. It's important to know whether or not what we call "plural" really means "dual or plural" in languages with dual number morphology. I (blush) don't know the number literature well enough to know this. Reportedly though, what we call "1st exclusive plural" really doesn't mean "exclusive or inclusive plural" in languages with inclusive person morphology... so similar questions arise in the person domain in any case. >Since I am committed (as I think many of us are) to a semantics >which interprets the syntax rather than generates it, I don't >understand what it means to say "it uses the most specific >morphosyntactic category compatible with the intended meaning" >except as a kind of pragmatic rule of the Gricean sort "be as >informative as possible". Yes, me too -- that's what I find so disturbing about the notion that that's how these categories are used / interpreted. If the feature geometries are Vocabulary items only, then it makes sense to me that they could be inserted into fully specified syntactic nodes whose features straightforwardly determine the way the category is interpreted. So, for example, if it's true that 1st person exclusive Vocabulary can't be used in an inclusive context, then we could insert it into a node that's [-Addressee] -- even if the 1st exclusive Vocabulary item itself doesn't have this feature. But what if the feature geometries are syntactic? Then we can't work any interpretive magic with negative-valued syntactic features. That is, we really NEED to know if "plural" really does mean "not dual" in languages with dual number morphology (ditto "exclusive" and "not inclusive"). Looking at cancellable implicatures is a fascinating idea! That really made me think. But I think I've convinced myself that this is actually *not* the way to determine the syntactic/semantic features of morphosyntactic categories. At first, it seems promising. For example, suppose you're an Arabic speaker and you hear (1), which is designed to cancel the plural's implicature of "greater than 2". My guess is that the implicature CAN be cancelled -- that (1) in Arabic would not be a contradiction. (Any Arabic speakers out there to confirm?) (1) Only they (pl.) are lucky -- in fact, only they (du.) are lucky. Anyway, suppose, for the sake of argument, that the implicature IS cancellable. From this, we might conclude that the plural really can be used for groups of two. But actually, I don't think that would be the correct conclusion. Here's why: (2)-(4) all sound fine to me as well, given the right context. There's a strong contrast with (5), for example. But surely we DON'T want to say that first plural in English really can be used for a set consisting of just the speaker, or just the addressee, or just a 3rd person! (2) Only we are lucky -- in fact, only I am lucky. (3) Only we are lucky -- in fact, only you are lucky. (4) Only we are lucky -- in fact, only he is lucky. (5) #Only I am lucky -- in fact, only he is lucky. I think these implicatures actually tell us about semantic relations between the two pronouns -- that is, "we" is a set that can *include* (not consist of) the sets "I", "you", and "he", while "I" is not a set that can include the set "he". Likewise, "they (pl)" can include the set "they (du)". That doesn't tell us whether it can *consist of* such a set. So... oh well. But what about just finding out whether a verb with a subject like "Rolf and Martha" can be plural, or whether it has to be dual? Would that be a good enough way to find out if plural can refer to sets of two? Suppose the verb in this context really does have to be marked as dual. If so, it seems to me that we WOULD need negative feature values in order to ensure this (e.g. plural morphology is inserted into [+Group, -Minimal] number in languages with dual morphology). If there are negative feature values in the syntax, then H&R's privative feature geometry could only be in the Vocabulary. (Actually, I once tried to imagine what their feature geometry would look like with binary +/- feature values, but... my head exploded. Perhaps others can do better than I..!) Best, Martha From mcginnis at UCALGARY.CA Sat Feb 28 23:08:57 2004 From: mcginnis at UCALGARY.CA (Martha McGinnis) Date: Sat, 28 Feb 2004 16:08:57 -0700 Subject: morphosyntactic feature geometries In-Reply-To: <55939948-6895-11D8-BAEE-000393CE35FE@man.ac.uk> Message-ID: Hi Dan, >This is an interesting issue. In their new book on agreement, Wechsler >& Zlatic (2003, The many faces of agreement, CSLI) address some >correspondences between semantics and morphology. Thanks for the reference. I'll try to get hold of it. >As for the syntax-morphology connection, wouldn't 'unification' predict >that the morphology would either (i) fill in missing values or that >(ii) the syntax nodes unspecified for certain features could allow some >features to play a morphological role, but no syntactic one (i.e. that >feature-matching between syntax and morphology would not be forced if >there is no incompatibility)? Or is (ii) ruled out in DM? I'm not sure I understand (ii). Is the idea that some features could be present only in the morphology, not in the syntax? Dave Embick has argued (within a DM framework) that this is indeed the case. He calls such features "dissociated". His argument for such features are based on Greek NonActive morphology (in a MITWPL paper), and I believe also the Latin perfect (in an LI article). Dave, feel free to chime in here if my references are out of date. >Besides compositional analysis of morphosyntactic nodes, is anyone >looking at agreement from this Harley-Ritter perspective? This would >seem an ideal area to look for answers to Martha's queries. I am, but not in detail. It would be an interesting topic to cover systematically. Anyone looking for a thesis project? -Martha -- mcginnis at ucalgary.ca From dan.everett at MAN.AC.UK Sun Feb 29 00:05:37 2004 From: dan.everett at MAN.AC.UK (Daniel L. Everett) Date: Sun, 29 Feb 2004 00:05:37 +0000 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Martha, > I'm not sure I understand (ii). Is the idea that some features could > be present only in the morphology, not in the syntax? Dave Embick > has argued (within a DM framework) that this is indeed the case. He > calls such features "dissociated". His argument for such features > are based on Greek NonActive morphology (in a MITWPL paper), and I > believe also the Latin perfect (in an LI article). Dave, feel free > to chime in here if my references are out of date. > Thanks for this additional reference. This is indeed what I have in mind. >> Besides compositional analysis of morphosyntactic nodes, is anyone >> looking at agreement from this Harley-Ritter perspective? This would >> seem an ideal area to look for answers to Martha's queries. > > I am, but not in detail. It would be an interesting topic to cover > systematically. Anyone looking for a thesis project? > > -Martha > I am currently working on some interesting types of agreement in Amazonian languages (Bob Dixon documents a related case, but takes a very different approach to the problem in an article in the Journal of Linguistics, 2001, on Jarawara), in which there are two types of agreement on the verb, one semantically motivated and one morphosyntactically motivated (Dixon argues that both are morphosyntactically motivated, but only at the expense of allowing the possessor to be the head of the NP in some cases but not others). This is not number agreement, but it does, I believe, show independence between the semantics, morphology, and syntax of agreement (along lines predicted by the W&Z book, though not identical). -- Dan Any DMrs going to the numerals conference in Leipzig next month? Looks like a good conference. Some issues related to this discussion will likely arise, though numerals aren't number. ------------------------------------------ Daniel L. Everett Professor of Phonetics & Phonology Postgraduate Programme Director Department of Linguistics The University of Manchester Oxford Road Manchester, UK M13 9PL http://ling.man.ac.uk/info/staff/de Fax: 44-161-275-3187 Office: 44-161-275-3158 From ajohns at CHASS.UTORONTO.CA Sat Feb 28 23:26:33 2004 From: ajohns at CHASS.UTORONTO.CA (Alana Johns) Date: Sat, 28 Feb 2004 18:26:33 -0500 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Regarding the recent discussion of the dual and its interpretation, I thought I would mention something I know but have not investigated fully. This is that in at least one dialect of Inuktitut which is known to have a dual, I have a story from a monolingual speaker where the topic of the story is a couple (isolated), and they are consistently referred to by the plural. It is possible that other dialects (e.g. Labrador Inuttut) would not allow this so it has been on my agenda for some time to investigate this more thoroughly. Alana On Feb 28, 2004, at 2:50 PM, Martha McGinnis wrote: > Rolf, thanks very much for your response! > > It's true, the two possibilities (feature geometries = Vocabulary, > feature geometries = syntax) should indeed make the different > empirical predictions you mentioned. The syntactic agreement > predictions might be hard to test, since agreement Vocabulary should > be subject to the same feature geometry as pronoun Vocabulary... but > one can imagine syntactic distinctions between e.g. number+gender > agreement (as on French participles) vs. person+number agreement (as > on French finite verbs) -- e.g. in French, the former type of > syntactic agreement relation doesn't seem to block further syntactic > movement of the nominal, while the latter type does. If H&R's > geometry is correct, we might expect to find other syntactic evidence > for these kinds of agreement splits, rather than a person+gender vs. > number split. > >> The idea that plurals in a language with duals mean "three or more" >> does not strike me as being the result of a specifically interpreted >> "minus dual" value of some sort, but rather a blocking effect. If >> the speaker had meant two (s)he would have used a dual; thus using >> the plural implies three or more. As a pragmatic or functional >> inference however, one might expect this to be cancellable in the >> appropriate discourse context. > > Good point. It's important to know whether or not what we call > "plural" really means "dual or plural" in languages with dual number > morphology. I (blush) don't know the number literature well enough > to know this. Reportedly though, what we call "1st exclusive plural" > really doesn't mean "exclusive or inclusive plural" in languages with > inclusive person morphology... so similar questions arise in the > person domain in any case. > >> Since I am committed (as I think many of us are) to a semantics >> which interprets the syntax rather than generates it, I don't >> understand what it means to say "it uses the most specific >> morphosyntactic category compatible with the intended meaning" >> except as a kind of pragmatic rule of the Gricean sort "be as >> informative as possible". > > Yes, me too -- that's what I find so disturbing about the notion that > that's how these categories are used / interpreted. If the feature > geometries are Vocabulary items only, then it makes sense to me that > they could be inserted into fully specified syntactic nodes whose > features straightforwardly determine the way the category is > interpreted. So, for example, if it's true that 1st person exclusive > Vocabulary can't be used in an inclusive context, then we could > insert it into a node that's [-Addressee] -- even if the 1st > exclusive Vocabulary item itself doesn't have this feature. > > But what if the feature geometries are syntactic? Then we can't work > any interpretive magic with negative-valued syntactic features. That > is, we really NEED to know if "plural" really does mean "not dual" in > languages with dual number morphology (ditto "exclusive" and "not > inclusive"). > > Looking at cancellable implicatures is a fascinating idea! That > really made me think. But I think I've convinced myself that this is > actually *not* the way to determine the syntactic/semantic features > of morphosyntactic categories. At first, it seems promising. For > example, suppose you're an Arabic speaker and you hear (1), which is > designed to cancel the plural's implicature of "greater than 2". My > guess is that the implicature CAN be cancelled -- that (1) in Arabic > would not be a contradiction. (Any Arabic speakers out there to > confirm?) > > (1) Only they (pl.) are lucky -- in fact, only they (du.) are lucky. > > Anyway, suppose, for the sake of argument, that the implicature IS > cancellable. From this, we might conclude that the plural really can > be used for groups of two. But actually, I don't think that would be > the correct conclusion. Here's why: (2)-(4) all sound fine to me as > well, given the right context. There's a strong contrast with (5), > for example. But surely we DON'T want to say that first plural in > English really can be used for a set consisting of just the speaker, > or just the addressee, or just a 3rd person! > > (2) Only we are lucky -- in fact, only I am lucky. > (3) Only we are lucky -- in fact, only you are lucky. > (4) Only we are lucky -- in fact, only he is lucky. > (5) #Only I am lucky -- in fact, only he is lucky. > > I think these implicatures actually tell us about semantic relations > between the two pronouns -- that is, "we" is a set that can *include* > (not consist of) the sets "I", "you", and "he", while "I" is not a > set that can include the set "he". Likewise, "they (pl)" can include > the set "they (du)". That doesn't tell us whether it can *consist of* > such a set. > > So... oh well. But what about just finding out whether a verb with a > subject like "Rolf and Martha" can be plural, or whether it has to be > dual? Would that be a good enough way to find out if plural can > refer to sets of two? Suppose the verb in this context really does > have to be marked as dual. If so, it seems to me that we WOULD need > negative feature values in order to ensure this (e.g. plural > morphology is inserted into [+Group, -Minimal] number in languages > with dual morphology). If there are negative feature values in the > syntax, then H&R's privative feature geometry could only be in the > Vocabulary. (Actually, I once tried to imagine what their feature > geometry would look like with binary +/- feature values, but... my > head exploded. Perhaps others can do better than I..!) > > Best, > Martha > > _____________________________ Alana Johns, Dept. of Linguistics 130 St. George, Robarts Library University of Toronto Toronto, ON M5S 3H1 CANADA Tel. 416-978-1761 FAX 416-971-2688 From hharley at EMAIL.ARIZONA.EDU Sun Feb 29 03:22:39 2004 From: hharley at EMAIL.ARIZONA.EDU (Heidi Harley) Date: Sat, 28 Feb 2004 20:22:39 -0700 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Hi all! Been reading the discussion with great interest. Funnily enough I'd just been thinking of posting about a separate (but related) issue, but thought I'd get a couple of cents' worth in on this first. While Betsy and I didn't take any stand on the status of the geometries in any particular framework, I have a few guesses about how they fit in, but which I don't really have any justification for; armchair thoughts, for the most part. But I thought I'd throw 'em out anyway. Back in 1994 when I was first thinking about these questions, I assumed that VIs had to realize subgraphs of the geometry, as Rolf suggests, and were hence so specified, each with a particular subgraph. But I *also* thought that they were competing to realize a fully-specified geometry under a node provided by the syntax. In a footnote somewhere in H&R, the possibility that the independent features are assembled into the geometric shape by syntactic operations is mentioned. I still think this might well be right, given that there seems to be some degree of syntactic separation possible (NumP, etc.). W/r to the subgraph/subtree contrast mentioned by Rolf, if Arabic 't-' does realize just 2, it could do so by being specified for the H&R subtree headed by Hearer (since Arabic has the incl/excl distinction, Hearer is active there). Hearer doesn't dominate number, so there's no entailment relationship between 2 and number. But without negative features, and without a Hearer subtree, in 1994 I had to go for homophonic t- prefixes (I think your analysis did that too, though, right, Rolf?), while Halle's Impoverishment and Fission paper, with liberal deployment of negative feature values, manages not to do this. As for negative feature values, in 1994 I was dead agin' 'em; was thinking about the geometry in Avery & Rice terms. I *think* that betsy and I were also thinking that default interpretations arose at semantics, but that negative values in the syntax weren't possible. I'll have to go back & reconstruct the reasoning. But I am very interested to think about the various predictions! Finally, as for the specific problem of the semantic entailments of number, I highly recommend a short paper of Elizabeth Cowper's (think it's available on her website) about plural and dual. It contains a highly original suggestion about the interpretation of number features as 'greater-than', such that the feature-geometric representation of 'dual' and that of plural in a language without dual are the same: >2. Languages with 'dual' have a third dependent of the number feature, >3. Her theory makes different predictions about the possible interpretations of the features than ours does, and about the relative markedness of dual in a system that contains it, and hence also about possible syncretisms, which I actually suspect may be right. I really look forward to hearing about how the various predictions turn out! next posting in just a second, :) hh Heidi Harley Department of Linguistics University of Arizona (520)626-3554 http://linguistics.arizona.edu/~hharley/ hharley at email.arizona.edu From hharley at EMAIL.ARIZONA.EDU Sun Feb 29 04:02:49 2004 From: hharley at EMAIL.ARIZONA.EDU (Heidi Harley) Date: Sat, 28 Feb 2004 21:02:49 -0700 Subject: syncretism w/o paradigms In-Reply-To: <1078024959.a6bd4f0d9f865@www.email.arizona.edu> Message-ID: ... and for a second take, here's something I've been worrying about: I've just been rereading Bobaljik's 2000 reply to Williams' 1994 'remarks on lexical knowledge' article in which Williams argues that metaparadigmattic patterns of syncretism are a robust phenomenon whose existence a theory should predict. Bobaljik does a very convincing job of showing that an 'instatiated basic paradigm' isn't a requirement of UG, and hence that the metaparadigmatic effects are not traceable to any actual metaparadigm requirement. He argues that Impoverishment rules can do the same job that the metaparadigmatic points of entry that Williams proposes, which is certainly true, since they have their effect before Vocab Insertion and so should create the same syncretic patterns in all relevant paradigms, even ones whose VIs are not related to each other. While it's true that Impoverishment *can* do the job of creating metasyncretism in the right way, it's actually not really too great to have to do *all* syncretism that way. Impoverishment would be (un)doing exactly the same job that paradigm structure is supposed to be doing in lexicalist approaches. Worse, it would just mean that competiton between vocab items would almost never arise. Part of the attractiveness of the late insertion account is that the garden-variety syncretic effects we see in language can be taken care of via the subset principle, i.e. by regular competition of VIs for fully specified syntactic nodes, no additional operations necessary. But just VI competition by itself doesn't predict the metasyncretism effect at all, as bobaljik notes. In fact, it seems to me that it predicts that such effects should be rare to nonexistent, which they're definitely not. Given the robustness of the metasyncretic effect, it seems to me we want another account of why variation in syncretism across paradigms within a given language is so rare. The feature-geometric approach, like noyer's feature-hierarchy approach, does constrain possible syncretic patterns somewhat. But neither approach predicts why in two otherwise vocab-item-unrelated paradigms in a given language (e.g. the masculine nominal and adjectival declensions of Russian), the same set of contrasts are neutralized again and again. As long as the geometry or hierarchy is respected, the separate vocab items could easily realize one set of contrasts in one paradigm and another set in another. Given that languages DON'T require a metaparadigm (as Bobaljik shows), does anyone out there have any speculations about why metaparadigmatic effects show up so strongly? Are they historical accidents? Learnability effects? Do we want lang-specific 'entry points' marked on the feature geomtry, sort of like Williams suggests? or is Impoverishment the way that all syncretism is implemented, and we can just throw the subset/elsewhere principle out the window? or what? :) hh Heidi Harley Department of Linguistics University of Arizona (520)626-3554 http://linguistics.arizona.edu/~hharley/ From dan.everett at MAN.AC.UK Sun Feb 29 19:03:09 2004 From: dan.everett at MAN.AC.UK (Daniel L. Everett) Date: Sun, 29 Feb 2004 19:03:09 +0000 Subject: morphosyntactic feature geometries In-Reply-To: <8BC8529D-6A45-11D8-A168-000A959B3EAC@chass.utoronto.ca> Message-ID: On Saturday, Feb 28, 2004, at 23:26 Europe/London, Alana Johns wrote: > Regarding the recent discussion of the dual and its interpretation, I > thought I would mention something I know but have not investigated > fully. This is that in at least one dialect of Inuktitut which is known > to have a dual, I have a story from a monolingual speaker where the > topic of the story is a couple (isolated), and they are consistently > referred to by the plural. It is possible that other dialects (e.g. > Labrador Inuttut) would not allow this so it has been on my agenda for > some time to investigate this more thoroughly. Alana Alana, I think that number can often be used for different illocutionary effects (e.g. the obvious 'imperial we'). Thus its meaning is perhaps not always literal. Has anyone done a detailed study of number in Inuttut discourse? Dan ------------------------------------------ Daniel L. Everett Professor of Phonetics & Phonology Postgraduate Programme Director Department of Linguistics The University of Manchester Oxford Road Manchester, UK M13 9PL http://ling.man.ac.uk/info/staff/de Fax: 44-161-275-3187 Office: 44-161-275-3158 From dan.everett at MAN.AC.UK Sun Feb 29 19:10:01 2004 From: dan.everett at MAN.AC.UK (Daniel L. Everett) Date: Sun, 29 Feb 2004 19:10:01 +0000 Subject: syncretism w/o paradigms In-Reply-To: <1078027369.bb983b1cbb04f@www.email.arizona.edu> Message-ID: Heidi, Hey. The arguments against the status of paradigms don't seem to pay much attention to recent work on 'periphrastic paradigms', so far as I am aware. So work by Ackerman, Stump, myself, and others on languages in which paradigms can include phrases might be more supportive than perhaps Jonathan knew (this work is all pretty new). In a recent ms (Liminal Categories, on my website) - which I am currently splitting into two papers - the first part deals with pronominal paradigms in Wari that are 100% periphrastic, yet must be treated as paradigms. If phrases can occupy cells of paradigms, then principles of paradigm construction don't seem reduceable to morphological principles like 'impoverishment'. Dan ------------------------------------------ Daniel L. Everett Professor of Phonetics & Phonology Postgraduate Programme Director Department of Linguistics The University of Manchester Oxford Road Manchester, UK M13 9PL http://ling.man.ac.uk/info/staff/de Fax: 44-161-275-3187 Office: 44-161-275-3158 From hharley at EMAIL.ARIZONA.EDU Sun Feb 29 20:43:26 2004 From: hharley at EMAIL.ARIZONA.EDU (Heidi Harley) Date: Sun, 29 Feb 2004 13:43:26 -0700 Subject: syncretism w/o paradigms In-Reply-To: Message-ID: Hey dan! You wrote: Hey. The arguments against the status of paradigms don't seem to pay much attention to recent work on 'periphrastic paradigms', so far as I am aware. So work by Ackerman, Stump, myself, and others on languages in which paradigms can include phrases might be more supportive than perhaps Jonathan knew (this work is all pretty new). In a recent ms (Liminal Categories, on my website) - which I am currently splitting into two papers - the first part deals with pronominal paradigms in Wari that are 100% periphrastic, yet must be treated as paradigms. If phrases can occupy cells of paradigms, then principles of paradigm construction don't seem reduceable to morphological principles like 'impoverishment'. In fact, the phrase/word 'competition' that these kind of periphrastic paradigmatic effects show is in fact one of the strongest arguments, in my mind, in favor of a postsyntactic morphology, and against the independent existence of 'paradigms' (independent of the set of terminal nodes made available by the syntax, that is). The phrase/word distinction has no pretheoretical status in such a theory, and so we expect to see such effects everywhere, as in fact we do. Indeed, the Impoverishment story can certainly predict syncretic effects that hold simultaneously across word-sized and phrase-sized realizations of given feature sets/terminal nodes. But this kind of effect doesn't have any bearing on whether Impoverishment is the right way to capture *all* metasyncretism effects, I don't think... ? Did someone at the paradigms workshop before the LSA talk about this kind of metasyncretism problem? :) hh From sbejar at CHASS.UTORONTO.CA Sun Feb 29 21:08:40 2004 From: sbejar at CHASS.UTORONTO.CA (Susana Bejar) Date: Sun, 29 Feb 2004 16:08:40 -0500 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Nice discussion. I thought I should pipe up, since my thesis is in fact an exploration of some of these very topics. In particular, I explored consequences for the theory of agreement if we assume a Harley+Ritter type system for formal features in syntax. I tried to show that feature hierarchies (which I see in terms of entailment relations) play a crucial role in agreement systems and can explain some otherwise confounding complex agreement patterns. I argued that conditions on agreement (i.e. whether Match and Value succeed or fail) are stated explicitly in terms of whether or not an entailment relation exists between features of the target and those of the controller. A consequence of this view is that the geometry turns out to be crucial to the determination of locality, (anti)intervention, etc. in agreement systems. This turns out to be extremely useful in accounting for languages where NPs exhibit phi-dependent asymmetries with respect to whether or not they can control agreement. To continue with the going example, if an agr-head has a [group] probe then the search for a goal can only be halted by an element in its domain that is itself specified for some feature that entails [group]. In a language with a plural/dual contrast, this means that both plural and dual NPs in the domain of the probe will Match and halt it (see (1) and (2)). However, a singular NP will not halt it (3), since no feature of a singular NP entails [group]; let's assume it has only and [individuation] feature. (Note that I'm assuming a particular view of the relation between [group] and [minimal], namely that [minimal] is a dependent of [group], i.e. [minimal] entails [group]. This is a controversial assumption, but see the Cowper paper mentioned by Heidi for good arguments in support of this). (1) ---------------- AGR NP1 NP2 [group] [group] ... (2) ---------------- AGR NP1 NP2 [group] [minimal] ... (3) --------------------------------- AGR NP1 NP2 [group] [individ] ... As far as where the geometry lives, I haven't come across reasons not to assume the simplest possible scenario: both syntactic and morphological objects are built from the same inventory of features constructed by the learner so terminal nodes and vocabulary items alike are built from the same features, some of which stand in intrinsic hierarchical/geometric/semantic relations to one another. I think we have to be careful not to coflate separate issues when thinking about feature systems. One question is privativity, another question is whether something like a geometry is a valid way to represent intrinsic relations between features. Likewise, with respect to underspecification, we need to distinguish logical underspecification (which is what the H&R system employs) vs. the kind of underspecification DM uses when assigning features to vocabulary items. The latter is not logical underspecification, but rather something extrinsically imposed in order to derive the observed blocking patterns between competing vocabulary items. In principle, I don't think having negative features precludes having geometric relations between features. So if we did have negative values in syntax, we could still have a (non-privative) feature geometry. The negative values would simply be another way of node labelling: (4) +Individ / \ +group (-group) / \ +min (-min) This does, however, preclude logical underspecification, so the question is whether or not this is a desirable outcome. As far as the hypotheses with respect to agreement that I outlined above, this would be a bad outcome. Locality patterns in syntactic agreement systems suggest to me that default values (or - values) are not specified in syntax. If they were we wouldn't expect any phi-asymmetries in the kinds of NPs that can control agreement. So we wouldn't expect the existence of languages where plural and dual NPs can match a probe, but singular NPs cannot, since the singular NPs would be specified as [-group] and therefore should technicaly be able to match a probe for [group]. (5) ---------------- AGR NP1 NP2 [group] [-group] ... (6) ---------------- AGR NP1 NP2 [group] [+group] ... Under this view, if the default specification is necessary at all, it must be supplied later. In her last posting, Martha considers this question with respect to the example of a conjoined subject like "Rolf and Martha". If I understand correctly, I think Martha's suggestion is that If a verb in this context needs to be marked as dual, not plural, then we need to assume negative features in th syntax. plural morphology <-> [+group, -minimal] dual morphology <-> [+group, +minimal] The implication, I think, is that without the possibility of a [-minimal] specification, nothing will prevent the incorrect use of a plural marker rather than a dual marker. I'm not sure I understand why this is a necessary conclusion. Couldn't we just assume that vocabulary competition ensures the correct outcome? (5) Dual AGR node matching vocabulary items winning VI [group], [min] plural <-> [group] dual <-> [group, min] dual <-> [group,min] (6) Plural AGR node matching vocabulary items winning VI [group] plural <-> [group] plural <-> [group] Perhaps I'm missing the point! I'd welcome any clarification! Best, Susana Bejar From mcginnis at UCALGARY.CA Thu Feb 26 19:13:12 2004 From: mcginnis at UCALGARY.CA (Martha McGinnis) Date: Thu, 26 Feb 2004 12:13:12 -0700 Subject: hello, DM-Listers! Message-ID: Dear DM-Listers, Whoa, has this list been quiet! I guess everyone is too busy doing research to post queries. I'd like to take this opportunity to encourage you to update your fellow listers on what you've been working on, morphologically speaking. Don't be shy! I know some of you have been writing busily away. Also, please take a moment to raise any questions you might have. Here's my contribution... I've recently been working on morphosyntactic feature geometries, a la Harley & Ritter (see their 2002 Language paper). This work has got me thinking about where morphosyntactic feature geometries should fit into the grammar. H&R don't focus on this, because they want to be as theory-neutral as possible. But once you start to think of it, it's not completely obvious. Let's begin with the observation that morphosyntactic feature geometries can be somewhat unspecified. For example, a language with a special "dual" number category would have a special featural representation (with both [Minimal] and [Group] specified), while the plural number category would have only [Group] specified. On the other hand, a language without a special dual category would also represent dual number as just [Group]. So how does a given grammar determine the relation between representations and interpretations? It seems as though the answer is this: it uses the most specific morphosyntactic category compatible with the intended meaning. If there's a [Minimal]+[Group] category, it uses that for the dual; if not, it uses just [Group]. On the other hand, if the meaning is plural, [Minimal] can't be used, so only [Group] is. Now, this sounds a lot like Vocabulary competition in DM, which inserts the most specific Vocabulary item compatible with the fully specified syntactic/semantic features of the syntactic node (which in turn determine the meaning). Suppose then that H&R's privative feature geometries are morphosyntactic representations of Vocabulary items. This is more or less Bonet's approach, if I understand it correctly. If so, then how are the features of *syntactic nodes* represented? (Bonet's dissertation leaves this issue somewhat open.) For example, how do we know that the plural representation can't be used to represent the dual meaning in a language that has a special dual category? Does the syntactic node have a binary-featured representation, like [+Group, -Minimal]? If not, then what? Does the semantics fill in default feature values like [-Minimal]? I'm sure some of you have thought this through much more carefully than I've been able to, as yet. Please let me know what you think. References welcome! All the best, Martha From mcginnis at UCALGARY.CA Thu Feb 26 19:25:02 2004 From: mcginnis at UCALGARY.CA (Martha McGinnis) Date: Thu, 26 Feb 2004 12:25:02 -0700 Subject: morphosyntactic feature geometries Message-ID: Dear DM-Listers, [Please forgive re-posting: it seems some mailservers are filtering out messages with "hello" in the title.] Whoa, has this list been quiet! I guess everyone is too busy doing research to post queries. I'd like to take this opportunity to encourage you to update your fellow listers on what you've been working on, morphologically speaking. Don't be shy! I know some of you have been writing busily away. Also, please take a moment to raise any questions you might have. Here's my contribution... I've recently been working on morphosyntactic feature geometries, a la Harley & Ritter (see their 2002 Language paper). This work has got me thinking about where morphosyntactic feature geometries should fit into the grammar. H&R don't focus on this, because they want to be as theory-neutral as possible. But once you start to think of it, it's not completely obvious. Let's begin with the observation that morphosyntactic feature geometries can be somewhat unspecified. For example, a language with a special "dual" number category would have a special featural representation (with both [Minimal] and [Group] specified), while the plural number category would have only [Group] specified. On the other hand, a language without a special dual category would also represent dual number as just [Group]. So how does a given grammar determine the relation between representations and interpretations? It seems as though the answer is this: it uses the most specific morphosyntactic category compatible with the intended meaning. If there's a [Minimal]+[Group] category, it uses that for the dual; if not, it uses just [Group]. On the other hand, if the meaning is plural, [Minimal] can't be used, so only [Group] is. Now, this sounds a lot like Vocabulary competition in DM, which inserts the most specific Vocabulary item compatible with the fully specified syntactic/semantic features of the syntactic node (which in turn determine the meaning). Suppose then that H&R's privative feature geometries are morphosyntactic representations of Vocabulary items. This is more or less Bonet's approach, if I understand it correctly. If so, then how are the features of *syntactic nodes* represented? (Bonet's dissertation leaves this issue somewhat open.) For example, how do we know that the plural representation can't be used to represent the dual meaning in a language that has a special dual category? Does the syntactic node have a binary-featured representation, like [+Group, -Minimal]? If not, then what? Does the semantics fill in default feature values like [-Minimal]? I'm sure some of you have thought this through much more carefully than I've been able to, as yet. Please let me know what you think. References welcome! All the best, Martha From dan.everett at MAN.AC.UK Thu Feb 26 19:52:40 2004 From: dan.everett at MAN.AC.UK (Daniel L. Everett) Date: Thu, 26 Feb 2004 19:52:40 +0000 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: > Does the syntactic node have a binary-featured > representation, like [+Group, -Minimal]? If not, then what? Does > the semantics fill in default feature values like [-Minimal]? > This is an interesting issue. In their new book on agreement, Wechsler & Zlatic (2003, The many faces of agreement, CSLI) address some correspondences between semantics and morphology. Not specifically what Martha asks about, but an interesting discussion of the issues in any case. Their distinction between INDEX (semantic-based agreement) and CONCORD (morphological feature-based agreement - actually morphosyntactic features in their system) seems quite useful and could be relevant here as well. One would, ceteris paribus, expect number (in their system number is found in both CONCORD and INDEX) to match between morphology and semantics. So I wonder if this would predict different semantic representations for languages with/without 'dual'? As for the syntax-morphology connection, wouldn't 'unification' predict that the morphology would either (i) fill in missing values or that (ii) the syntax nodes unspecified for certain features could allow some features to play a morphological role, but no syntactic one (i.e. that feature-matching between syntax and morphology would not be forced if there is no incompatibility)? Or is (ii) ruled out in DM? Besides compositional analysis of morphosyntactic nodes, is anyone looking at agreement from this Harley-Ritter perspective? This would seem an ideal area to look for answers to Martha's queries. -- Dan ------------------------------------------ Daniel L. Everett Professor of Phonetics & Phonology Postgraduate Programme Director Department of Linguistics The University of Manchester Oxford Road Manchester, UK M13 9PL http://ling.man.ac.uk/info/staff/de Fax: 44-161-275-3187 Office: 44-161-275-3158 From rnoyer at BABEL.LING.UPENN.EDU Thu Feb 26 23:22:38 2004 From: rnoyer at BABEL.LING.UPENN.EDU (Rolf Noyer) Date: Thu, 26 Feb 2004 18:22:38 -0500 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Hi Martha et. al, Strangely coincidental: I too was looking through and thinking a bit about Harley & Ritter's paper in Language just yesterday afternoon. When I wrote my dissertation I concluded that there were no such geometries, although I toyed with them quite a bit. I have not, but would like to get a chance to, think carefully about H&R's arguments and decide if I wish to recant or put up a fight. I too am a bit confused about the place such representations would occupy in the grammar. Sets of morphosyntactic features might be geometrically related in a syntactic node for which vocabulary items compete, or they might be geometrically related in vocabulary items which compete for insertion into those nodes. If the latter, we might expect that person and gender could not be encoded in a single vocabulary item without gender. In other words, we should never expect, for example, [2 pl f] to split into two pieces: [2 f] and [pl]. More generally, we should, I believe, expect that any given vocabulary item should encode a subgraph of the entire graph (= tree) of features. Notice that this doesn't mean "subtree" since, as I pointed out, if this were so one could not have a vocabulary item expressing [2] without also expressing gender and number at the same time, clearly a false prediction in many languages, as for example Arabic in which t-aktub-na spells out [2 pl f], with /t-/ as [2] and /-na/ as [pl f] If the former, we might expect certain agreement phenomena, which presumably examine the syntactic nodes themselves and not the vocabulary items which are inserted, to be subject to constraints which would find explanation in a geometry of the features involved. Martha says: So how does a given grammar determine the relation between representations and interpretations? It seems as though the answer is this: it uses the most specific morphosyntactic category compatible with the intended meaning. When I last thought about this my attitude was that one should view the matter of interpretation (just like everywhere else) as combining semantic entailments with functional inferences. The idea that plurals in a language with duals mean "three or more" does not strike me as being the result of a specifically interpreted "minus dual" value of some sort, but rather a blocking effect. If the speaker had meant two (s)he would have used a dual; thus using the plural implies three or more. As a pragmatic or functional inference however, one might expect this to be cancellable in the appropriate discourse context. Here my knowledge of the phenomenon ends; it would be helpful to find out if plurals can be used to refer to two individuals in a language with a dual category if the discourse context is suitably constructed, such as, for example, a situation in which all that was discourse-relevant was that the argument in question was not individual, but at least two. Since I am committed (as I think many of us are) to a semantics which interprets the syntax rather than generates it, I don't understand what it means to say "it uses the most specific morphosyntactic category compatible with the intended meaning" except as a kind of pragmatic rule of the Gricean sort "be as informative as possible". I'm not an expert on this literature but I seem to recall that Larry Horn has worked on this problem for number and certain quantifiers. So for example, asserting "three men are in the room" seems to strongly imply that "no more than three men are in the room" but it does not entail it. Obviously these remarks are really off the cuff but maybe we can get some dialogue going. Best regards, Rolf From mcginnis at UCALGARY.CA Sat Feb 28 19:50:45 2004 From: mcginnis at UCALGARY.CA (Martha McGinnis) Date: Sat, 28 Feb 2004 12:50:45 -0700 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Rolf, thanks very much for your response! It's true, the two possibilities (feature geometries = Vocabulary, feature geometries = syntax) should indeed make the different empirical predictions you mentioned. The syntactic agreement predictions might be hard to test, since agreement Vocabulary should be subject to the same feature geometry as pronoun Vocabulary... but one can imagine syntactic distinctions between e.g. number+gender agreement (as on French participles) vs. person+number agreement (as on French finite verbs) -- e.g. in French, the former type of syntactic agreement relation doesn't seem to block further syntactic movement of the nominal, while the latter type does. If H&R's geometry is correct, we might expect to find other syntactic evidence for these kinds of agreement splits, rather than a person+gender vs. number split. >The idea that plurals in a language with duals mean "three or more" >does not strike me as being the result of a specifically interpreted >"minus dual" value of some sort, but rather a blocking effect. If >the speaker had meant two (s)he would have used a dual; thus using >the plural implies three or more. As a pragmatic or functional >inference however, one might expect this to be cancellable in the >appropriate discourse context. Good point. It's important to know whether or not what we call "plural" really means "dual or plural" in languages with dual number morphology. I (blush) don't know the number literature well enough to know this. Reportedly though, what we call "1st exclusive plural" really doesn't mean "exclusive or inclusive plural" in languages with inclusive person morphology... so similar questions arise in the person domain in any case. >Since I am committed (as I think many of us are) to a semantics >which interprets the syntax rather than generates it, I don't >understand what it means to say "it uses the most specific >morphosyntactic category compatible with the intended meaning" >except as a kind of pragmatic rule of the Gricean sort "be as >informative as possible". Yes, me too -- that's what I find so disturbing about the notion that that's how these categories are used / interpreted. If the feature geometries are Vocabulary items only, then it makes sense to me that they could be inserted into fully specified syntactic nodes whose features straightforwardly determine the way the category is interpreted. So, for example, if it's true that 1st person exclusive Vocabulary can't be used in an inclusive context, then we could insert it into a node that's [-Addressee] -- even if the 1st exclusive Vocabulary item itself doesn't have this feature. But what if the feature geometries are syntactic? Then we can't work any interpretive magic with negative-valued syntactic features. That is, we really NEED to know if "plural" really does mean "not dual" in languages with dual number morphology (ditto "exclusive" and "not inclusive"). Looking at cancellable implicatures is a fascinating idea! That really made me think. But I think I've convinced myself that this is actually *not* the way to determine the syntactic/semantic features of morphosyntactic categories. At first, it seems promising. For example, suppose you're an Arabic speaker and you hear (1), which is designed to cancel the plural's implicature of "greater than 2". My guess is that the implicature CAN be cancelled -- that (1) in Arabic would not be a contradiction. (Any Arabic speakers out there to confirm?) (1) Only they (pl.) are lucky -- in fact, only they (du.) are lucky. Anyway, suppose, for the sake of argument, that the implicature IS cancellable. From this, we might conclude that the plural really can be used for groups of two. But actually, I don't think that would be the correct conclusion. Here's why: (2)-(4) all sound fine to me as well, given the right context. There's a strong contrast with (5), for example. But surely we DON'T want to say that first plural in English really can be used for a set consisting of just the speaker, or just the addressee, or just a 3rd person! (2) Only we are lucky -- in fact, only I am lucky. (3) Only we are lucky -- in fact, only you are lucky. (4) Only we are lucky -- in fact, only he is lucky. (5) #Only I am lucky -- in fact, only he is lucky. I think these implicatures actually tell us about semantic relations between the two pronouns -- that is, "we" is a set that can *include* (not consist of) the sets "I", "you", and "he", while "I" is not a set that can include the set "he". Likewise, "they (pl)" can include the set "they (du)". That doesn't tell us whether it can *consist of* such a set. So... oh well. But what about just finding out whether a verb with a subject like "Rolf and Martha" can be plural, or whether it has to be dual? Would that be a good enough way to find out if plural can refer to sets of two? Suppose the verb in this context really does have to be marked as dual. If so, it seems to me that we WOULD need negative feature values in order to ensure this (e.g. plural morphology is inserted into [+Group, -Minimal] number in languages with dual morphology). If there are negative feature values in the syntax, then H&R's privative feature geometry could only be in the Vocabulary. (Actually, I once tried to imagine what their feature geometry would look like with binary +/- feature values, but... my head exploded. Perhaps others can do better than I..!) Best, Martha From mcginnis at UCALGARY.CA Sat Feb 28 23:08:57 2004 From: mcginnis at UCALGARY.CA (Martha McGinnis) Date: Sat, 28 Feb 2004 16:08:57 -0700 Subject: morphosyntactic feature geometries In-Reply-To: <55939948-6895-11D8-BAEE-000393CE35FE@man.ac.uk> Message-ID: Hi Dan, >This is an interesting issue. In their new book on agreement, Wechsler >& Zlatic (2003, The many faces of agreement, CSLI) address some >correspondences between semantics and morphology. Thanks for the reference. I'll try to get hold of it. >As for the syntax-morphology connection, wouldn't 'unification' predict >that the morphology would either (i) fill in missing values or that >(ii) the syntax nodes unspecified for certain features could allow some >features to play a morphological role, but no syntactic one (i.e. that >feature-matching between syntax and morphology would not be forced if >there is no incompatibility)? Or is (ii) ruled out in DM? I'm not sure I understand (ii). Is the idea that some features could be present only in the morphology, not in the syntax? Dave Embick has argued (within a DM framework) that this is indeed the case. He calls such features "dissociated". His argument for such features are based on Greek NonActive morphology (in a MITWPL paper), and I believe also the Latin perfect (in an LI article). Dave, feel free to chime in here if my references are out of date. >Besides compositional analysis of morphosyntactic nodes, is anyone >looking at agreement from this Harley-Ritter perspective? This would >seem an ideal area to look for answers to Martha's queries. I am, but not in detail. It would be an interesting topic to cover systematically. Anyone looking for a thesis project? -Martha -- mcginnis at ucalgary.ca From dan.everett at MAN.AC.UK Sun Feb 29 00:05:37 2004 From: dan.everett at MAN.AC.UK (Daniel L. Everett) Date: Sun, 29 Feb 2004 00:05:37 +0000 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Martha, > I'm not sure I understand (ii). Is the idea that some features could > be present only in the morphology, not in the syntax? Dave Embick > has argued (within a DM framework) that this is indeed the case. He > calls such features "dissociated". His argument for such features > are based on Greek NonActive morphology (in a MITWPL paper), and I > believe also the Latin perfect (in an LI article). Dave, feel free > to chime in here if my references are out of date. > Thanks for this additional reference. This is indeed what I have in mind. >> Besides compositional analysis of morphosyntactic nodes, is anyone >> looking at agreement from this Harley-Ritter perspective? This would >> seem an ideal area to look for answers to Martha's queries. > > I am, but not in detail. It would be an interesting topic to cover > systematically. Anyone looking for a thesis project? > > -Martha > I am currently working on some interesting types of agreement in Amazonian languages (Bob Dixon documents a related case, but takes a very different approach to the problem in an article in the Journal of Linguistics, 2001, on Jarawara), in which there are two types of agreement on the verb, one semantically motivated and one morphosyntactically motivated (Dixon argues that both are morphosyntactically motivated, but only at the expense of allowing the possessor to be the head of the NP in some cases but not others). This is not number agreement, but it does, I believe, show independence between the semantics, morphology, and syntax of agreement (along lines predicted by the W&Z book, though not identical). -- Dan Any DMrs going to the numerals conference in Leipzig next month? Looks like a good conference. Some issues related to this discussion will likely arise, though numerals aren't number. ------------------------------------------ Daniel L. Everett Professor of Phonetics & Phonology Postgraduate Programme Director Department of Linguistics The University of Manchester Oxford Road Manchester, UK M13 9PL http://ling.man.ac.uk/info/staff/de Fax: 44-161-275-3187 Office: 44-161-275-3158 From ajohns at CHASS.UTORONTO.CA Sat Feb 28 23:26:33 2004 From: ajohns at CHASS.UTORONTO.CA (Alana Johns) Date: Sat, 28 Feb 2004 18:26:33 -0500 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Regarding the recent discussion of the dual and its interpretation, I thought I would mention something I know but have not investigated fully. This is that in at least one dialect of Inuktitut which is known to have a dual, I have a story from a monolingual speaker where the topic of the story is a couple (isolated), and they are consistently referred to by the plural. It is possible that other dialects (e.g. Labrador Inuttut) would not allow this so it has been on my agenda for some time to investigate this more thoroughly. Alana On Feb 28, 2004, at 2:50 PM, Martha McGinnis wrote: > Rolf, thanks very much for your response! > > It's true, the two possibilities (feature geometries = Vocabulary, > feature geometries = syntax) should indeed make the different > empirical predictions you mentioned. The syntactic agreement > predictions might be hard to test, since agreement Vocabulary should > be subject to the same feature geometry as pronoun Vocabulary... but > one can imagine syntactic distinctions between e.g. number+gender > agreement (as on French participles) vs. person+number agreement (as > on French finite verbs) -- e.g. in French, the former type of > syntactic agreement relation doesn't seem to block further syntactic > movement of the nominal, while the latter type does. If H&R's > geometry is correct, we might expect to find other syntactic evidence > for these kinds of agreement splits, rather than a person+gender vs. > number split. > >> The idea that plurals in a language with duals mean "three or more" >> does not strike me as being the result of a specifically interpreted >> "minus dual" value of some sort, but rather a blocking effect. If >> the speaker had meant two (s)he would have used a dual; thus using >> the plural implies three or more. As a pragmatic or functional >> inference however, one might expect this to be cancellable in the >> appropriate discourse context. > > Good point. It's important to know whether or not what we call > "plural" really means "dual or plural" in languages with dual number > morphology. I (blush) don't know the number literature well enough > to know this. Reportedly though, what we call "1st exclusive plural" > really doesn't mean "exclusive or inclusive plural" in languages with > inclusive person morphology... so similar questions arise in the > person domain in any case. > >> Since I am committed (as I think many of us are) to a semantics >> which interprets the syntax rather than generates it, I don't >> understand what it means to say "it uses the most specific >> morphosyntactic category compatible with the intended meaning" >> except as a kind of pragmatic rule of the Gricean sort "be as >> informative as possible". > > Yes, me too -- that's what I find so disturbing about the notion that > that's how these categories are used / interpreted. If the feature > geometries are Vocabulary items only, then it makes sense to me that > they could be inserted into fully specified syntactic nodes whose > features straightforwardly determine the way the category is > interpreted. So, for example, if it's true that 1st person exclusive > Vocabulary can't be used in an inclusive context, then we could > insert it into a node that's [-Addressee] -- even if the 1st > exclusive Vocabulary item itself doesn't have this feature. > > But what if the feature geometries are syntactic? Then we can't work > any interpretive magic with negative-valued syntactic features. That > is, we really NEED to know if "plural" really does mean "not dual" in > languages with dual number morphology (ditto "exclusive" and "not > inclusive"). > > Looking at cancellable implicatures is a fascinating idea! That > really made me think. But I think I've convinced myself that this is > actually *not* the way to determine the syntactic/semantic features > of morphosyntactic categories. At first, it seems promising. For > example, suppose you're an Arabic speaker and you hear (1), which is > designed to cancel the plural's implicature of "greater than 2". My > guess is that the implicature CAN be cancelled -- that (1) in Arabic > would not be a contradiction. (Any Arabic speakers out there to > confirm?) > > (1) Only they (pl.) are lucky -- in fact, only they (du.) are lucky. > > Anyway, suppose, for the sake of argument, that the implicature IS > cancellable. From this, we might conclude that the plural really can > be used for groups of two. But actually, I don't think that would be > the correct conclusion. Here's why: (2)-(4) all sound fine to me as > well, given the right context. There's a strong contrast with (5), > for example. But surely we DON'T want to say that first plural in > English really can be used for a set consisting of just the speaker, > or just the addressee, or just a 3rd person! > > (2) Only we are lucky -- in fact, only I am lucky. > (3) Only we are lucky -- in fact, only you are lucky. > (4) Only we are lucky -- in fact, only he is lucky. > (5) #Only I am lucky -- in fact, only he is lucky. > > I think these implicatures actually tell us about semantic relations > between the two pronouns -- that is, "we" is a set that can *include* > (not consist of) the sets "I", "you", and "he", while "I" is not a > set that can include the set "he". Likewise, "they (pl)" can include > the set "they (du)". That doesn't tell us whether it can *consist of* > such a set. > > So... oh well. But what about just finding out whether a verb with a > subject like "Rolf and Martha" can be plural, or whether it has to be > dual? Would that be a good enough way to find out if plural can > refer to sets of two? Suppose the verb in this context really does > have to be marked as dual. If so, it seems to me that we WOULD need > negative feature values in order to ensure this (e.g. plural > morphology is inserted into [+Group, -Minimal] number in languages > with dual morphology). If there are negative feature values in the > syntax, then H&R's privative feature geometry could only be in the > Vocabulary. (Actually, I once tried to imagine what their feature > geometry would look like with binary +/- feature values, but... my > head exploded. Perhaps others can do better than I..!) > > Best, > Martha > > _____________________________ Alana Johns, Dept. of Linguistics 130 St. George, Robarts Library University of Toronto Toronto, ON M5S 3H1 CANADA Tel. 416-978-1761 FAX 416-971-2688 From hharley at EMAIL.ARIZONA.EDU Sun Feb 29 03:22:39 2004 From: hharley at EMAIL.ARIZONA.EDU (Heidi Harley) Date: Sat, 28 Feb 2004 20:22:39 -0700 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Hi all! Been reading the discussion with great interest. Funnily enough I'd just been thinking of posting about a separate (but related) issue, but thought I'd get a couple of cents' worth in on this first. While Betsy and I didn't take any stand on the status of the geometries in any particular framework, I have a few guesses about how they fit in, but which I don't really have any justification for; armchair thoughts, for the most part. But I thought I'd throw 'em out anyway. Back in 1994 when I was first thinking about these questions, I assumed that VIs had to realize subgraphs of the geometry, as Rolf suggests, and were hence so specified, each with a particular subgraph. But I *also* thought that they were competing to realize a fully-specified geometry under a node provided by the syntax. In a footnote somewhere in H&R, the possibility that the independent features are assembled into the geometric shape by syntactic operations is mentioned. I still think this might well be right, given that there seems to be some degree of syntactic separation possible (NumP, etc.). W/r to the subgraph/subtree contrast mentioned by Rolf, if Arabic 't-' does realize just 2, it could do so by being specified for the H&R subtree headed by Hearer (since Arabic has the incl/excl distinction, Hearer is active there). Hearer doesn't dominate number, so there's no entailment relationship between 2 and number. But without negative features, and without a Hearer subtree, in 1994 I had to go for homophonic t- prefixes (I think your analysis did that too, though, right, Rolf?), while Halle's Impoverishment and Fission paper, with liberal deployment of negative feature values, manages not to do this. As for negative feature values, in 1994 I was dead agin' 'em; was thinking about the geometry in Avery & Rice terms. I *think* that betsy and I were also thinking that default interpretations arose at semantics, but that negative values in the syntax weren't possible. I'll have to go back & reconstruct the reasoning. But I am very interested to think about the various predictions! Finally, as for the specific problem of the semantic entailments of number, I highly recommend a short paper of Elizabeth Cowper's (think it's available on her website) about plural and dual. It contains a highly original suggestion about the interpretation of number features as 'greater-than', such that the feature-geometric representation of 'dual' and that of plural in a language without dual are the same: >2. Languages with 'dual' have a third dependent of the number feature, >3. Her theory makes different predictions about the possible interpretations of the features than ours does, and about the relative markedness of dual in a system that contains it, and hence also about possible syncretisms, which I actually suspect may be right. I really look forward to hearing about how the various predictions turn out! next posting in just a second, :) hh Heidi Harley Department of Linguistics University of Arizona (520)626-3554 http://linguistics.arizona.edu/~hharley/ hharley at email.arizona.edu From hharley at EMAIL.ARIZONA.EDU Sun Feb 29 04:02:49 2004 From: hharley at EMAIL.ARIZONA.EDU (Heidi Harley) Date: Sat, 28 Feb 2004 21:02:49 -0700 Subject: syncretism w/o paradigms In-Reply-To: <1078024959.a6bd4f0d9f865@www.email.arizona.edu> Message-ID: ... and for a second take, here's something I've been worrying about: I've just been rereading Bobaljik's 2000 reply to Williams' 1994 'remarks on lexical knowledge' article in which Williams argues that metaparadigmattic patterns of syncretism are a robust phenomenon whose existence a theory should predict. Bobaljik does a very convincing job of showing that an 'instatiated basic paradigm' isn't a requirement of UG, and hence that the metaparadigmatic effects are not traceable to any actual metaparadigm requirement. He argues that Impoverishment rules can do the same job that the metaparadigmatic points of entry that Williams proposes, which is certainly true, since they have their effect before Vocab Insertion and so should create the same syncretic patterns in all relevant paradigms, even ones whose VIs are not related to each other. While it's true that Impoverishment *can* do the job of creating metasyncretism in the right way, it's actually not really too great to have to do *all* syncretism that way. Impoverishment would be (un)doing exactly the same job that paradigm structure is supposed to be doing in lexicalist approaches. Worse, it would just mean that competiton between vocab items would almost never arise. Part of the attractiveness of the late insertion account is that the garden-variety syncretic effects we see in language can be taken care of via the subset principle, i.e. by regular competition of VIs for fully specified syntactic nodes, no additional operations necessary. But just VI competition by itself doesn't predict the metasyncretism effect at all, as bobaljik notes. In fact, it seems to me that it predicts that such effects should be rare to nonexistent, which they're definitely not. Given the robustness of the metasyncretic effect, it seems to me we want another account of why variation in syncretism across paradigms within a given language is so rare. The feature-geometric approach, like noyer's feature-hierarchy approach, does constrain possible syncretic patterns somewhat. But neither approach predicts why in two otherwise vocab-item-unrelated paradigms in a given language (e.g. the masculine nominal and adjectival declensions of Russian), the same set of contrasts are neutralized again and again. As long as the geometry or hierarchy is respected, the separate vocab items could easily realize one set of contrasts in one paradigm and another set in another. Given that languages DON'T require a metaparadigm (as Bobaljik shows), does anyone out there have any speculations about why metaparadigmatic effects show up so strongly? Are they historical accidents? Learnability effects? Do we want lang-specific 'entry points' marked on the feature geomtry, sort of like Williams suggests? or is Impoverishment the way that all syncretism is implemented, and we can just throw the subset/elsewhere principle out the window? or what? :) hh Heidi Harley Department of Linguistics University of Arizona (520)626-3554 http://linguistics.arizona.edu/~hharley/ From dan.everett at MAN.AC.UK Sun Feb 29 19:03:09 2004 From: dan.everett at MAN.AC.UK (Daniel L. Everett) Date: Sun, 29 Feb 2004 19:03:09 +0000 Subject: morphosyntactic feature geometries In-Reply-To: <8BC8529D-6A45-11D8-A168-000A959B3EAC@chass.utoronto.ca> Message-ID: On Saturday, Feb 28, 2004, at 23:26 Europe/London, Alana Johns wrote: > Regarding the recent discussion of the dual and its interpretation, I > thought I would mention something I know but have not investigated > fully. This is that in at least one dialect of Inuktitut which is known > to have a dual, I have a story from a monolingual speaker where the > topic of the story is a couple (isolated), and they are consistently > referred to by the plural. It is possible that other dialects (e.g. > Labrador Inuttut) would not allow this so it has been on my agenda for > some time to investigate this more thoroughly. Alana Alana, I think that number can often be used for different illocutionary effects (e.g. the obvious 'imperial we'). Thus its meaning is perhaps not always literal. Has anyone done a detailed study of number in Inuttut discourse? Dan ------------------------------------------ Daniel L. Everett Professor of Phonetics & Phonology Postgraduate Programme Director Department of Linguistics The University of Manchester Oxford Road Manchester, UK M13 9PL http://ling.man.ac.uk/info/staff/de Fax: 44-161-275-3187 Office: 44-161-275-3158 From dan.everett at MAN.AC.UK Sun Feb 29 19:10:01 2004 From: dan.everett at MAN.AC.UK (Daniel L. Everett) Date: Sun, 29 Feb 2004 19:10:01 +0000 Subject: syncretism w/o paradigms In-Reply-To: <1078027369.bb983b1cbb04f@www.email.arizona.edu> Message-ID: Heidi, Hey. The arguments against the status of paradigms don't seem to pay much attention to recent work on 'periphrastic paradigms', so far as I am aware. So work by Ackerman, Stump, myself, and others on languages in which paradigms can include phrases might be more supportive than perhaps Jonathan knew (this work is all pretty new). In a recent ms (Liminal Categories, on my website) - which I am currently splitting into two papers - the first part deals with pronominal paradigms in Wari that are 100% periphrastic, yet must be treated as paradigms. If phrases can occupy cells of paradigms, then principles of paradigm construction don't seem reduceable to morphological principles like 'impoverishment'. Dan ------------------------------------------ Daniel L. Everett Professor of Phonetics & Phonology Postgraduate Programme Director Department of Linguistics The University of Manchester Oxford Road Manchester, UK M13 9PL http://ling.man.ac.uk/info/staff/de Fax: 44-161-275-3187 Office: 44-161-275-3158 From hharley at EMAIL.ARIZONA.EDU Sun Feb 29 20:43:26 2004 From: hharley at EMAIL.ARIZONA.EDU (Heidi Harley) Date: Sun, 29 Feb 2004 13:43:26 -0700 Subject: syncretism w/o paradigms In-Reply-To: Message-ID: Hey dan! You wrote: Hey. The arguments against the status of paradigms don't seem to pay much attention to recent work on 'periphrastic paradigms', so far as I am aware. So work by Ackerman, Stump, myself, and others on languages in which paradigms can include phrases might be more supportive than perhaps Jonathan knew (this work is all pretty new). In a recent ms (Liminal Categories, on my website) - which I am currently splitting into two papers - the first part deals with pronominal paradigms in Wari that are 100% periphrastic, yet must be treated as paradigms. If phrases can occupy cells of paradigms, then principles of paradigm construction don't seem reduceable to morphological principles like 'impoverishment'. In fact, the phrase/word 'competition' that these kind of periphrastic paradigmatic effects show is in fact one of the strongest arguments, in my mind, in favor of a postsyntactic morphology, and against the independent existence of 'paradigms' (independent of the set of terminal nodes made available by the syntax, that is). The phrase/word distinction has no pretheoretical status in such a theory, and so we expect to see such effects everywhere, as in fact we do. Indeed, the Impoverishment story can certainly predict syncretic effects that hold simultaneously across word-sized and phrase-sized realizations of given feature sets/terminal nodes. But this kind of effect doesn't have any bearing on whether Impoverishment is the right way to capture *all* metasyncretism effects, I don't think... ? Did someone at the paradigms workshop before the LSA talk about this kind of metasyncretism problem? :) hh From sbejar at CHASS.UTORONTO.CA Sun Feb 29 21:08:40 2004 From: sbejar at CHASS.UTORONTO.CA (Susana Bejar) Date: Sun, 29 Feb 2004 16:08:40 -0500 Subject: morphosyntactic feature geometries In-Reply-To: Message-ID: Nice discussion. I thought I should pipe up, since my thesis is in fact an exploration of some of these very topics. In particular, I explored consequences for the theory of agreement if we assume a Harley+Ritter type system for formal features in syntax. I tried to show that feature hierarchies (which I see in terms of entailment relations) play a crucial role in agreement systems and can explain some otherwise confounding complex agreement patterns. I argued that conditions on agreement (i.e. whether Match and Value succeed or fail) are stated explicitly in terms of whether or not an entailment relation exists between features of the target and those of the controller. A consequence of this view is that the geometry turns out to be crucial to the determination of locality, (anti)intervention, etc. in agreement systems. This turns out to be extremely useful in accounting for languages where NPs exhibit phi-dependent asymmetries with respect to whether or not they can control agreement. To continue with the going example, if an agr-head has a [group] probe then the search for a goal can only be halted by an element in its domain that is itself specified for some feature that entails [group]. In a language with a plural/dual contrast, this means that both plural and dual NPs in the domain of the probe will Match and halt it (see (1) and (2)). However, a singular NP will not halt it (3), since no feature of a singular NP entails [group]; let's assume it has only and [individuation] feature. (Note that I'm assuming a particular view of the relation between [group] and [minimal], namely that [minimal] is a dependent of [group], i.e. [minimal] entails [group]. This is a controversial assumption, but see the Cowper paper mentioned by Heidi for good arguments in support of this). (1) ---------------- AGR NP1 NP2 [group] [group] ... (2) ---------------- AGR NP1 NP2 [group] [minimal] ... (3) --------------------------------- AGR NP1 NP2 [group] [individ] ... As far as where the geometry lives, I haven't come across reasons not to assume the simplest possible scenario: both syntactic and morphological objects are built from the same inventory of features constructed by the learner so terminal nodes and vocabulary items alike are built from the same features, some of which stand in intrinsic hierarchical/geometric/semantic relations to one another. I think we have to be careful not to coflate separate issues when thinking about feature systems. One question is privativity, another question is whether something like a geometry is a valid way to represent intrinsic relations between features. Likewise, with respect to underspecification, we need to distinguish logical underspecification (which is what the H&R system employs) vs. the kind of underspecification DM uses when assigning features to vocabulary items. The latter is not logical underspecification, but rather something extrinsically imposed in order to derive the observed blocking patterns between competing vocabulary items. In principle, I don't think having negative features precludes having geometric relations between features. So if we did have negative values in syntax, we could still have a (non-privative) feature geometry. The negative values would simply be another way of node labelling: (4) +Individ / \ +group (-group) / \ +min (-min) This does, however, preclude logical underspecification, so the question is whether or not this is a desirable outcome. As far as the hypotheses with respect to agreement that I outlined above, this would be a bad outcome. Locality patterns in syntactic agreement systems suggest to me that default values (or - values) are not specified in syntax. If they were we wouldn't expect any phi-asymmetries in the kinds of NPs that can control agreement. So we wouldn't expect the existence of languages where plural and dual NPs can match a probe, but singular NPs cannot, since the singular NPs would be specified as [-group] and therefore should technicaly be able to match a probe for [group]. (5) ---------------- AGR NP1 NP2 [group] [-group] ... (6) ---------------- AGR NP1 NP2 [group] [+group] ... Under this view, if the default specification is necessary at all, it must be supplied later. In her last posting, Martha considers this question with respect to the example of a conjoined subject like "Rolf and Martha". If I understand correctly, I think Martha's suggestion is that If a verb in this context needs to be marked as dual, not plural, then we need to assume negative features in th syntax. plural morphology <-> [+group, -minimal] dual morphology <-> [+group, +minimal] The implication, I think, is that without the possibility of a [-minimal] specification, nothing will prevent the incorrect use of a plural marker rather than a dual marker. I'm not sure I understand why this is a necessary conclusion. Couldn't we just assume that vocabulary competition ensures the correct outcome? (5) Dual AGR node matching vocabulary items winning VI [group], [min] plural <-> [group] dual <-> [group, min] dual <-> [group,min] (6) Plural AGR node matching vocabulary items winning VI [group] plural <-> [group] plural <-> [group] Perhaps I'm missing the point! I'd welcome any clarification! Best, Susana Bejar