The 'Self-Movement' frame
Posted: Sat Jan 29, 2005 7:11 am
'Crawl' is, apparently, an instance of the "Self-Movement" semantic frame. Here are some of the frame elements that Fillmore and Atkins have identified (the table is a close-as-possible reproduction of the one appearing in Ravin and Leacock's (eds) Polysemy: Theoretical and Computational Approaches. Oxford, 2002).
...............................TABLE 5.4 Elements in the self-movement frame
-------------------------------------------------------------------------------------
Frame.............As exemplified in a corpus sentence
element
-------------------------------------------------------------------------------------
Mover.............David Attenborough was crawling up behind two copulating lions.
Area................Literally on hands and knees, his men crawled over every inch of the area
Path.................A cat can crawl through any hole it can get its head through
Source.............The two hedgehogs crawled from the nest at sunset
Goal.................They would never find her if she crawled into a cupboard and hid away
Distance...........From there they crawled the last ten yards to the edge of a rise
Manner............A small baby crawled out on its hands and knees, its face covered in grime
Speed...............I crawled smartly after him
-------------------------------------------------------------------------------------
Generally I have no problem with the semantic analysis here, but I was just wondering, how would you classify 'behind two copulating lions' and 'after him'? These seem to be also suited to the label "Goal", as 'in the cupboard' is, but "Goal" is still a quite vague label generally for what is, after all, the functional reason(s?!) for crawling 'up behind', 'into' and 'after' something; that is, these labels tell us nothing about e.g. "hunting" (shooting/capturing wildlife on film), escaping 'them', and escaping or perhaps chasing or just plain "keeping up with" 'him' (not that these meanings couldn't be made clearer with extra verbs/context, beyond the linking prepositions). I guess any linguistic analysis, no matter how "fine-grained", can never capture all the meanings we somehow so easily understand.
Anyway, I think the FrameNet approach is interesting in how it is trying to expand upon grammatico-semantic roles and relations, and be consistent with semantic labels by an examination of the complexities of real data. I've mentioned frames several times now on Dave's, and just thought you'd all be interested in the approach (if you hadn't checked it out already).
http://www.icsi.berkeley.edu/~framenet/
To access the frames, click on 'FN DATA'; you can then browse using the 'Index by Lexical Unit' (c - crawl) or the 'Index by Frame' (if you know the frame you are looking for).
...............................TABLE 5.4 Elements in the self-movement frame
-------------------------------------------------------------------------------------
Frame.............As exemplified in a corpus sentence
element
-------------------------------------------------------------------------------------
Mover.............David Attenborough was crawling up behind two copulating lions.
Area................Literally on hands and knees, his men crawled over every inch of the area
Path.................A cat can crawl through any hole it can get its head through
Source.............The two hedgehogs crawled from the nest at sunset
Goal.................They would never find her if she crawled into a cupboard and hid away
Distance...........From there they crawled the last ten yards to the edge of a rise
Manner............A small baby crawled out on its hands and knees, its face covered in grime
Speed...............I crawled smartly after him
-------------------------------------------------------------------------------------
Generally I have no problem with the semantic analysis here, but I was just wondering, how would you classify 'behind two copulating lions' and 'after him'? These seem to be also suited to the label "Goal", as 'in the cupboard' is, but "Goal" is still a quite vague label generally for what is, after all, the functional reason(s?!) for crawling 'up behind', 'into' and 'after' something; that is, these labels tell us nothing about e.g. "hunting" (shooting/capturing wildlife on film), escaping 'them', and escaping or perhaps chasing or just plain "keeping up with" 'him' (not that these meanings couldn't be made clearer with extra verbs/context, beyond the linking prepositions). I guess any linguistic analysis, no matter how "fine-grained", can never capture all the meanings we somehow so easily understand.
Anyway, I think the FrameNet approach is interesting in how it is trying to expand upon grammatico-semantic roles and relations, and be consistent with semantic labels by an examination of the complexities of real data. I've mentioned frames several times now on Dave's, and just thought you'd all be interested in the approach (if you hadn't checked it out already).
http://www.icsi.berkeley.edu/~framenet/
To access the frames, click on 'FN DATA'; you can then browse using the 'Index by Lexical Unit' (c - crawl) or the 'Index by Frame' (if you know the frame you are looking for).