<div dir="ltr"><br><br><div class="gmail_quote"><div dir="ltr">Am Fr., 21. Sep. 2018 um 23:12 Uhr schrieb <a href="mailto:goodman.m.w@gmail.com">goodman.m.w@gmail.com</a> <<a href="mailto:goodman.m.w@gmail.com">goodman.m.w@gmail.com</a>>:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div style="font-family:arial,helvetica,sans-serif">Returning to this thread so it doesn't fade away...<br></div><div style="font-family:arial,helvetica,sans-serif"><br></div><div style="font-family:arial,helvetica,sans-serif">I don't understand the complexities well enough to have an opinion on whether we should use YADU, Guy's variant, or something else.</div></div></div></div></blockquote><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div style="font-family:arial,helvetica,sans-serif"> I'm concerned, however, that if it is sufficiently complicated to work out the ramifications of the use of defaults, it would outweigh the benefits from the perspective of the grammar developer. With this in mind I would (...expect that a grammar developer would...) prefer something more predictable over something slightly more powerful. I'd even be happy with something more restrictive than YADU if it were still useful, easy to understand, and could give informative error messages at compile time. That is, can we somehow proscribe the complicated corner cases with a more restrictive definition?<br></div></div></div></div></blockquote><div><br></div><div> I think it would make sense to have a number of use cases (or unit tests) to see whether any variant of defaults would help the developer. I proposed my variant to deal with the use case Emily gave -- but there could well be other cases I haven't thought of, as well as side effects I haven't thought of! I'm not sure that a more restrictive version of YADU could deal with Emily's use case.<br></div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div style="font-family:arial,helvetica,sans-serif"></div><div style="font-family:arial,helvetica,sans-serif"><br></div><div style="font-family:arial,helvetica,sans-serif">Regarding the syntax of these things, Emily and Guy use /#coref, whereas Copestake 2002, the ParisDefeasibleConstraints wiki, and the LKB use /l (lowercase L for "lexical", which is specified by the *description-persistence* parameter in the LKB). I don't really understand the difference, but it seems the latter doesn't require a paired coreference, as shown in regverb in the standard example below:<br></div><div style="font-family:arial,helvetica,sans-serif"><br></div><div style="font-family:arial,helvetica,sans-serif"> verb := *top* &<br> [ PAST /l #pp,<br> PASTP #p /l #pp,<br> PASSP #p /l #pp ].<br><br> regverb := verb &<br> [ PAST /l "ed" ] .</div><div style="font-family:arial,helvetica,sans-serif"><br></div><div style="font-family:arial,helvetica,sans-serif">If these are in fact different instances of the same mechanism, then maybe "/#coref" is just "/ #coref" (i.e., a persistent-default operator followed by an actual coreference)?<br></div></div></div></div></blockquote><div><br></div><div>The YADU paper (Lascarides and Copestake, 1999) also uses just "/" rather than "/l" -- although that paper gives typeset AVMs, rather than machine-readable tdl. I'd be happy to stick to "/l" if that would reduce confusion!<br></div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div style="font-family:arial,helvetica,sans-serif"></div><div style="font-family:arial,helvetica,sans-serif"><br></div><div style="font-family:arial,helvetica,sans-serif">On a related note, I recently added to PyDelphin the ability to serialize TDL, so I can imagine writing code that would transform a grammar with defaults to one without as some kind of compile-time preprocessor. It would be great if we could agree on and provide implementations of defaults before the next summit.<br></div></div></div></div></blockquote><div><br></div><div>Yes! So far I haven't been able to find explicit documentation of how defaults are handled. My current understanding is that, the compile-time processor needs to expand type and instance definitions to include all inherited constraints -- and then the (possibly inherited) default constraints on instances are turned into hard constraints, while the (possibly inherited) default constraints on types are discarded. I would be happy to be corrected here if I'm wrong!<br></div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div style="font-family:arial,helvetica,sans-serif"></div></div></div></div><br><div class="gmail_quote"><div dir="ltr">On Thu, Sep 6, 2018 at 4:17 AM Guy Emerson <<a href="mailto:gete2@cam.ac.uk" target="_blank">gete2@cam.ac.uk</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>What I've proposed isn't credulous default unification, although it does use it as part of its definition, since DefFS does. The YADU paper does describe credulous default unification (definition 9, on page 69, or page 15 of the pdf), citing Carpenter.<br></div><div><br></div><div>Credulously unifying [ F #1, G #1 ] into [ F.P c, G.P d ] just yields the singleton set without any new constraints, { [ F.P c, G.P d ] }, because the only piece of information in [ F #1, G #1 ] that is not already in [ F.P c, G.P d ] is the re-entrancy, but this re-entrancy is incompatible. It's crucial that [ F #1, G #1 ] does not contain the paths F.Q and G.Q, or else these *would* be incorporated in credulous unification -- but this follows Emily's example with valence-min, where the relevant feature paths are similarly not there. This is why I proposed first expanding [ F #1, G #1 ] by credulously unifying in the opposite direction.<br></div><div><br><div class="gmail_quote"><div dir="ltr">Am Mi., 5. Sep. 2018 um 18:05 Uhr schrieb Ann Copestake <<a href="mailto:aac10@cl.cam.ac.uk" target="_blank">aac10@cl.cam.ac.uk</a>>:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> <div text="#000000" bgcolor="#FFFFFF"> <p>this seems like it might be credulous default unification? can't remember whether we described that in the YADU paper or not - I think the idea comes from Bob Carpenter. Sorry I am just procrastinating and don't have time to read/think properly, or check.</p> <p>Cheers,</p> <p>Ann<br> </p> <br> <div class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958moz-cite-prefix">On 05/09/18 17:56, Guy Emerson wrote:<br> </div> <blockquote type="cite"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr">I'm going to adapt Alex's extended Nixon diamond example from <a href="http://moin.delph-in.net/StanfordDefaults" target="_blank">http://moin.delph-in.net/StanfordDefaults</a> , because it's a minimal working example that captures the essence of Emily's example.</div> <div dir="ltr"><br> </div> <div>We have:</div> <div><br> </div> <div>a := *top* &</div> <div> [ F /#1,</div> <div> G /#1 ].</div> <div><br> </div> <div>x := *top* &</div> <div> [ P *top*,</div> <div> Q *top* ].</div> <div><br> </div> <div>c := *top*.</div> <div>d := *top*.<br> </div> <div><br> </div> <div>b := a &</div> <div> [ F.P c,</div> <div> G.P d ].</div> <div><br> </div> <div>I take it that the behaviour we would like to have is that the default re-entrancy between F and G is pushed down (in b) to a re-entrancy between F.Q and G.Q. As I understand, YADU ( <a href="http://www.aclweb.org/anthology/J99-1002" target="_blank">http://www.aclweb.org/anthology/J99-1002</a> ) would discard the re-entrancy, because it's not compatible with the hard constraints. (Also, note that I have deliberately left the features P and Q outside of the definition of a, to mimic valence-min in Emily's example -- we can't represent the re-entrancy between F.Q and G.Q in a, but only in b.)</div> <div><br> </div> <div>I have a tentative proposal -- and I say tentative, because Ann has already warned that YADU can blow up, and what I'm about to suggest would be probably be even more likely to blow up. So here be explosive dragons!</div> <div><br> </div> <div>But with that warning aside... YADU represents a defeasible feature structure in two parts - a normal feature structure (containing the hard constraints), and a "tail" of default constraints. In the above example, only a has a non-empty tail, which just has one thing in it (the re-entrancy). With YADU, unifying two defeasible structures involves first unifying their hard structures, combining the tails, and then discarding anything in the tail that's incompatible with the hard structure. My proposed extension of YADU is not to discard incompatible constraints in the tail, but rather "expand" them and keep whatever compatible constraints are in this expanded set.</div> <div><br> </div> <div>This "expansion" of constraints doesn't require new algebraic operations, because we can use what's already in YADU -- in particular, the DefFS operation (Definition 12 on page 70, which is page 16 in the pdf). Essentially, DefFS takes a hard structure and a set of possibly contradictory default constraints, and adds as many of the default constraints as possible without creating a contradiction. My proposal is to apply DefFS *backwards* -- if we unify two hard structures and find that some default constraint in the tail is now incompatible with the unified hard structure, we can use DefFS to "expand" the default constraint. We treat the default constraint as the hard structure in DefFS, and we treat the unified hard structure as the tail in DefFS. This tells us how much of the unified hard structure is compatible with the default constraint -- in other words, we have "expanded" the default constraint in the context of the unified hard structure. We can now decompose this expanded structure into individual constraints, and add them to the new tail, as long as they don't contradict the new unified hard structure.</div> <div><br> </div> <div>This might be clearest looking at the above example. We would like to unify b's feature structure with a's feature structure. This gives us the following hard structure, and tail (I am ignoring the types of tail elements, for simplicity, because we aren't considering when one default overrides another default):<br> </div> <div><br> </div> [ F x & [ P c,<br> Q *top* ],<br> G x & [ P d,<br> Q *top* ] ]. <div>tail: { F=G }</div> <div><br> </div> <div>To apply DefFS, we first credulously add the hard structure's constraints to the tail element. This gives a pair of structures (which differ in the value of F.P):</div> <div><br> </div> <div>[ F #1 & x & [ P c,<br> Q *top* ],<br> G #1 ].</div> <div><br> </div> <div>[ F #1 & x & [ P d,<br> Q *top* ],<br> G #1 ].</div> <div><br> </div> <div>DefFS gives us the generalisation of these two structures, which is:</div> <div><br> </div> <div>[ F #1 & x & [ P *top*,<br> Q *top* ],<br> G #1 ].<br> </div> <div><br> </div> <div>We can break this up into individual constraints: { F:x, G:x, F=G, F.P=G.P, F.Q=G.Q }</div> <div><br> </div> <div>Of these five constraints, the first two are already part of the unified hard structure, and the second two are incompatible with it. So we are left with a single element to keep in the tail: F.Q=G.Q. So, expanding elements in the tail rather than discarding them gives:</div> <div><br> </div> <div>[ F x & [ P c,<br> Q *top* ],<br> G x & [ P d,<br> Q *top* ] ]. <div>tail: { F.Q=G.Q }</div> </div> <div><br> </div> <div>Finally, the last step (as with YADU) is to apply DefFS, which would give the following compiled structure used at runtime (supposing I've understood what is supposed to happen at compile time -- I wasn't sure where to look for this documentation):</div> <div><br> </div> <div>[ F x & [ P c,<br> Q #1 ],<br> G x & [ P d,<br> Q #1 ] ].<br> </div> </div> </div> </div> </div> </div> </div> </div> </div> </div> </div> <br> <div class="gmail_quote"> <div dir="ltr">Am Di., 4. Sep. 2018 um 23:10 Uhr schrieb Emily M. Bender <<a href="mailto:ebender@uw.edu" target="_blank">ebender@uw.edu</a>>:<br> </div> <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> <div dir="ltr">Dear Mike, <div><br> </div> <div>Thanks for bringing up this issue. At the 2010 DELPH-IN Paris Summit, Ann and I had a further conversation about this, from which I took the homework of typing up what it is I'd like to have (as a grammar developer, and especially from the point of view of the Matrix) wrt to defeasible constraints. Here's what I wrote down later that year (Oct 27):</div> <div><br> </div> <div><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">Dear Ann,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">Here, with much more delay than I intended, is the write up</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">I promised of my (reconstruction of my) understanding of where</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">we ended up in our discussion of<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">defeasible</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>identity<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">constraints</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">over crepes in Paris.</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">First, why I want it:</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">In lexical rules, we want to be able say (like in SWB) that</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">the value of certain features (HOOK, CAT, ARG-ST) is shared</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">between the mother and the daughter unless the rule contradicts</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">this. If the rule does contradict it, then we want only the information</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">specifically stated as such to change, and the rest "around" it,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">to be shared.</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">For a concrete example, take a hypothetical lexical rule that</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">changes the case on the first complement from acc to dat.</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">First, here's the general lex rule type:</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">lex-rule := phrase-or-lexrule & word-or-lexrule &</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> [ NEEDS-AFFIX bool,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> SYNSEM.LOCAL.CONT [ RELS [ LIST #first,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> LAST #last ],</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> HCONS [ LIST #hfirst,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> LAST #hlast ] ],</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> DTR #dtr & word-or-lexrule &</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> [ SYNSEM.LOCAL.CONT [ RELS [ LIST #first,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> LAST #middle ],</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> HCONS [ LIST #hfirst,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> LAST #hmiddle ] ],</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> ALTS #alts ],</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> C-CONT [ RELS [ LIST #middle,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> LAST #last ],</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> HCONS [ LIST #hmiddle,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> LAST #hlast ]],</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> ALTS #alts,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> ARGS < #dtr > ].</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">And a subtype with the<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">defeasible</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>identity indicated</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">(using /# for now):</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">defeasible</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">-identity-lex-rule := lex-rule &</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> [<span> </span></span><a href="http://synsem.local.cat/" rel="noreferrer" style="color:rgb(17,85,204);font-size:12.8px;background-color:rgb(255,255,255)" target="_blank">SYNSEM.LOCAL.CAT</a><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>/#cat,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> ARG-ST /#arg-st,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> C-CONT.HOOK /#hook,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> DTR [ LOCAL [ CAT /#cat,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> CONT.HOOK /#hook ],</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> ARG-ST /#arg-st ]].</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">The lex rule definition itself would just look like this:</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">acc-to-dat-obj-lex-rule := lex-rule &</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> [ SYNSEM.LOCAL.CAT.COMPS.FIRST.</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">LOCAL.CAT.HEAD.CASE dat,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> DTR.SYNSEM.LOCAL.CAT.COMPS.</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">FIRST.LOCAL.CAT.HEAD.CASE acc ].</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">The intended behavior is for that to compile into a rule that</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">includes<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">constraints</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>like these (I'm sure I'm missing some here):</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">acc-to-dat-obj-lex-rule (expanded):</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> [<span> </span></span><a href="http://synsem.local.cat/" rel="noreferrer" style="color:rgb(17,85,204);font-size:12.8px;background-color:rgb(255,255,255)" target="_blank">SYNSEM.LOCAL.CAT</a><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>[ HEAD #head,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> VAL [ SPR #spr,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> SPEC #spec,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> SUBJ #subj,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> COMPS [ REST #rest,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> FIRST [</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">NON-LOCAL #non-local,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> LOCAL [ CONT #cont,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> CAT [ VAL #val,</span></div> <div><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> AGR #agr,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> HEAD.CASE dat ]]]]]],</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> C-CONT.HOOK #hook,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> ARG-ST #arg-st,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> DTR [ SYNSEM.LOCAL [ CONT.HOOK #hook,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> CAT [ HEAD #head,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> VAL [ SPR #spr,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> SPEC #spec,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> SUBJ #subj,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> COMPS [ REST #rest,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> FIRST [ NON-LOCAL #non-local,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> LOCAL [ CONT #cont,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> CAT [ VAL #val,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> AGR #agr,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> HEAD.CASE acc ]]]]]]],</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"> ARG-ST #arg-st ]].</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">What I remember from Paris is that we decided it would be best to</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">encode these<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">constraints</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>not directly in the type definition as</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">I did above in<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">defeasible</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">-identity-lex-rule but in a collateral file that</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">instructs the LKB to do something special with certain feature paths</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">on instances of certain types at compile time.</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">We also worked out that we would only be able to "push down" the identity</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">constraint</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>to features that were necessitated by the types invoked</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">in the rule. Thus in the example above, we know that SPR, SPEC and SUBJ</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">need to be identified because the value of VAL is necessarily "valence"</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">(and not valence-min) as we've mentioned COMPS. But if CASE were appropriate</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">for both noun and comp (for example), then we wouldn't be able to know to</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">put in identity<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">constraints</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>for any other features of noun (or comp). If</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">the daughter in fact had a<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">constraint</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>on one of these other features,</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">it wouldn't </span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">be copied up to the mother. Relatedly, we lose the actual HEAD </span></div> <div><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">value because </span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">we can't identify HEAD while changing CASE. (So here, the grammar </span></div> <div><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">writer </span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">would need to stipulate [HEAD noun], say, on the mother.)</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">In Paris, I remember being convinced that the added simplicity in defining</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">lexical rules would out-weigh the lack of transparency noted above. And</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">I'm still pretty sure I agree with that. One thing in favor of that </span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">view is that </span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">if a rule </span></div> <div><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">defined using the<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">defeasible</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>identity type didn't have the expected </span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">behavior, the </span></div> <div><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">grammar engineer could always either add<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">constraints</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>or side-step</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">that type and hand-specify all the desired identities.</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">A further complication I noticed while writing out this example is the</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">interaction between<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">defeasible</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>and indefeasible identity tags. Two conditions</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">to consider:</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">1) The rule inherits a<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">constraint</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>(e.g., from the type of the DTR value) that</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">the REST of the ARG-ST is the same as the COMPS list.</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">2) The rule doesn't inherit such a<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">constraint</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">, but the constituent that</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">serves as the daughter identifies its ARG-ST.REST and its COMPS.</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">I think (2) isn't a problem (this is very similar to things that confused</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">Tom, Ivan and I as we designed the lex rules in the textbook, though, so I'm</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">not feeling very confident just now!). As for (1), it could entail a similar</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">push down of identity inside the ARG-ST. But what if the ARG-ST to</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">DTR.ARG-ST identification were a non-</span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">defeasible</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><span> </span>identity<span> </span></span><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail-il" style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial">constraint</span><span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">?</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">Maybe that's just a broken grammar that either shouldn't compile or would</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">just have surprising behavior.</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">I hope you are still interested in this problem. Let me know if/when it would</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">be useful to have a grammar to play with.</span><br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <br style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"> <span style="font-size:12.8px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">Thanks!</span><br> </div> </div> <div class="gmail_extra"><br> <div class="gmail_quote">On Tue, Sep 4, 2018 at 10:42 AM, <a href="mailto:goodman.m.w@gmail.com" target="_blank">goodman.m.w@gmail.com</a> <span dir="ltr"><<a href="mailto:goodman.m.w@gmail.com" target="_blank">goodman.m.w@gmail.com</a>></span> wrote:<br> <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr"> <div dir="ltr"> <div style="font-family:arial,helvetica,sans-serif">Hello everyone,</div> <div style="font-family:arial,helvetica,sans-serif"><br> </div> <div style="font-family:arial,helvetica,sans-serif">I appreciate the feedback I've received in previous messages in my attempts to dust off neglected corners of TDL syntax, and I'd now like to bring up "defaults", or "defeasible constraints" (I believe these refer to the same thing). Are we prepared to start supporting defaults/defeasible-constraints in our processors and using them in our grammars? Or should we discard them as an undesired experimental feature (i.e., declare them to *not* be part of DELPH-IN TDL)?<br> <div style="font-family:arial,helvetica,sans-serif"><br> </div> <div style="font-family:arial,helvetica,sans-serif">Further information:<br> </div> <div style="font-family:arial,helvetica,sans-serif"><br> </div> Currently, only the LKB supports them (and maybe PET?). As I understand, they are a compile-time feature, meaning that they change how the grammar is compiled and that there is no longer a notion of "defaults" during run-time. I don't think the use of defaults causes any change in the competence or performance of a grammar.<br> </div> <div style="font-family:arial,helvetica,sans-serif"><br> </div> <div style="font-family:arial,helvetica,sans-serif">The benefit of defaults is for the grammar engineer as it can reduce the amount of boilerplate code and make the grammar source code more intuitive. I think any result that makes grammar writing easier is a big win. The differences it creates between the source-code form of the grammar and the compiled hierarchy, however, can complicate debugging (e.g., interactive unification).<br> </div> <div style="font-family:arial,helvetica,sans-serif"><br> </div> <div style="font-family:arial,helvetica,sans-serif">Some links:</div> <div style="font-family:arial,helvetica,sans-serif"> - <a href="http://www.aclweb.org/anthology/J99-1002" target="_blank">http://www.aclweb.org/anthology/J99-1002</a></div> <div style="font-family:arial,helvetica,sans-serif"> - <a href="http://moin.delph-in.net/ParisDefeasibleConstraints" target="_blank">http://moin.delph-in.net/ParisDefeasibleConstraints</a></div> <div style="font-family:arial,helvetica,sans-serif"> - <a href="http://moin.delph-in.net/StanfordDefaults" target="_blank">http://moin.delph-in.net/StanfordDefaults</a><span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155HOEnZb"><font color="#888888"><br clear="all"> </font></span></div> <span class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155HOEnZb"><font color="#888888"><br> -- <br> <div dir="ltr" class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155m_-2873427553648370275gmail_signature">-Michael Wayne Goodman</div> </font></span></div> </div> </div> </div> </blockquote> </div> <br> <br clear="all"> <div><br> </div> -- <br> <div class="m_-8593124814449095464m_4777295698704568155m_-4850224290236657958m_-2966151228754498155gmail_signature" data-smartmail="gmail_signature"> <div dir="ltr"> <div> <div dir="ltr"> <div> <div dir="ltr"> <div> <div dir="ltr"> <div> <div dir="ltr">Emily M. Bender<br> Professor, <span style="font-size:12.8px">Department of Linguistics</span></div> <div><span style="font-size:12.8px">University of Washington</span></div> <div>Twitter: @emilymbender</div> </div> </div> </div> </div> </div> </div> </div> </div> </div> </div> </blockquote> </div> </blockquote> <br> </div> </blockquote></div></div></div> </blockquote></div><br clear="all"><br>-- <br><div dir="ltr" class="m_-8593124814449095464gmail_signature" data-smartmail="gmail_signature">-Michael Wayne Goodman</div> </blockquote></div></div>