<html><body><div style="color:#000; background-color:#fff; font-family:Courier New, courier, monaco, monospace, sans-serif;font-size:14pt"><div><span>Dear Adam,</span></div><div style="color: rgb(0, 0, 0); font-size: 18.6667px; font-family: Courier New,courier,monaco,monospace,sans-serif; background-color: transparent; font-style: normal;"><br><span></span></div><div style="color: rgb(0, 0, 0); font-size: 18.6667px; font-family: Courier New,courier,monaco,monospace,sans-serif; background-color: transparent; font-style: normal;"><span>one question is why you need to assign semantic roles to arguments and adjuncts. Would you like to model alternations on the basis of lexical semantics (in the spirit say of LMT)? I would agree with Stefan that a simple interface would serve your purposes better, if this is your intention.</span></div><div style="color: rgb(0, 0, 0); font-size: 18.6667px; font-family: Courier New,courier,monaco,monospace,sans-serif;
background-color: transparent; font-style: normal;"><br><span></span></div><div style="color: rgb(0, 0, 0); font-size: 18.6667px; font-family: Courier New,courier,monaco,monospace,sans-serif; background-color: transparent; font-style: normal;"><span>Of course, one can always establish some (loose) correspondence between semantic roles and a simple interface (a similar idea was worked out by Anthony Davis back in</span><span> 2001). This approach leaves more room for using semantic role labels or something like that without relying on them for predicting syntactic behavior. <br></span></div><div style="color: rgb(0, 0, 0); font-size: 18.6667px; font-family: Courier New,courier,monaco,monospace,sans-serif; background-color: transparent; font-style: normal;"><br></div><div style="color: rgb(0, 0, 0); font-size: 18.6667px; font-family: Courier New,courier,monaco,monospace,sans-serif; background-color: transparent; font-style:
normal;">Best,</div><div style="color: rgb(0, 0, 0); font-size: 18.6667px; font-family: Courier New,courier,monaco,monospace,sans-serif; background-color: transparent; font-style: normal;"><br></div><div style="color: rgb(0, 0, 0); font-size: 18.6667px; font-family: Courier New,courier,monaco,monospace,sans-serif; background-color: transparent; font-style: normal;">Stella Markantonatou<br><span></span></div><div style="display: block;" class="yahoo_quoted"> <br> <br> <div style="font-family: Courier New, courier, monaco, monospace, sans-serif; font-size: 14pt;"> <div style="font-family: HelveticaNeue, Helvetica Neue, Helvetica, Arial, Lucida Grande, sans-serif; font-size: 12pt;"> <div dir="ltr"> <font face="Arial" size="2"> On Monday, February 3, 2014 1:34 AM, Adam Przepiorkowski <adamp@IPIPAN.WAW.PL> wrote:<br> </font> </div> <div class="y_msg_container">Dear All,<br><br>[Apologies for cross-posting.]<br><br>In the context of enriching the
Polish LFG grammar with semantic<br>representation, we are looking for a set of semantic roles (Agent,<br>Patient, Beneficiary, etc.) that could be used to mark arguments (and<br>possibly adjuncts) of verbs and other predicates. This set should be<br>exhaustive in the sense that it should be possible to assign – more or<br>less deterministically – a semantic role to any argument (and possibly<br>adjunct) of any predicate. For this reason the standard – in LFG<br>textbooks – sets of some 7 semantic roles do not seem sufficient.<br>Instead, we are looking at larger repertoires proposed in VerbNet, in<br>FrameNet and in John W. Sowa's work on Knowledge Representation.<br><br>We don't have any strong views about any particular set of semantic<br>roles, as long as it is exhaustive and applicable to real texts (as<br>opposed to being merely theoretically interesting). Has anybody in the<br>LFG community faced a similar task? If
so, what set of semantic roles<br>would you recommend? At the moment, we are wavering between VerbNet and<br>Sowa's system, both being more manageable than numerous roles offered<br>in FrameNet, but we are open to other solutions.<br><br>Many thanks, best regards,<br><br>Adam P.<br><br>-- <br>Adam Przepiórkowski ˈadam ˌpʃɛpjurˈkɔfskʲi<br><a href="http://clip.ipipan.waw.pl/" target="_blank">http://clip.ipipan.waw.pl/ </a>____ Computational Linguistics in Poland<br><a href="http://jlm.ipipan.waw.pl/" target="_blank">http://jlm.ipipan.waw.pl/ </a>___________ Journal of Language Modelling<br><a href="http://zil.ipipan.waw.pl/" target="_blank">http://zil.ipipan.waw.pl/ </a>____________ Linguistic Engineering Group<br><a href="http://nkjp.pl/" target="_blank">http://nkjp.pl/ </a>_________________________ National Corpus of Polish<br><br></div> </div>
</div> </div> </div></body></html>