[Corpora-List] Words ambiguous for sentiment

Krishnamurthy, Ramesh r.krishnamurthy at aston.ac.uk
Tue Apr 11 09:21:14 CEST 2017


hi keith

isn't there a contradiction or circularity in your recent post?

#1 "The corpus linguistics perspective probably starts with Firth: ?You shall know a word by the company it keeps?. This is in line with Ramesh?s comments about meanings arising from contexts. I would certainly agree with this but context is a big word. Context may be a horizon (or window) of words 5 to the left or 5 to the right of a node word. It may be a dynamic window so context can extend a lot further. It may be the whole document or the meaning of the word may be impacted on by the genre or register." ...

I would say, *not* 'may be'... but *is*! This is why the meanings of utterances can be, and are being, constantly *reinterpreted*... ie the meaning of the original utterance changes, in the light of the changed contexts of the reader/interpreter (eg more historical information has become known about the material or logistical circumstances of the original utterance, more biographical information has come to light about the participants in thge original interaction, or new ideas or methods of analysis or interpretation have been developed, etc etc)?

#2 "Word meanings can predict context and be predicted by context."

How can 'word meanings *predict* context', if the contributions that words make to meanings is ab initio and per se *dependent on context*?

best ramesh

---- Date: Sun, 09 Apr 2017 21:22:28 +0200 From: Keith Douglas Charles Stuart <kstuart at idm.upv.es> Subject: [Corpora-List] Words ambiguous for sentiment To: CORPORA at uib.no

Thanks to Valerio, Jorge and Ramesh. Sorry to be so slow getting back to you. The background to my question is that I am helping on a research project which involves developing software to carry out sentiment analysis. The project can be found here: https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Ftecnolengua.uma.es%2F%3Fpage_id%3D8&data=02%7C01%7CR.Krishnamurthy%40aston.ac.uk%7Ce64199f8fde7404b1fd908d4803d973a%7Ca085950c4c2544d5945ab852fa44a221%7C0%7C0%7C636274449429782526&sdata=X0TE9aMwYzrCWfbq3J1eh%2Fdv4lYSSbRjBMp4F1J0aBk%3D&reserved=0

And a paper describing the software here: https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fjournal.sepln.org%2Fsepln%2Fojs%2Fojs%2Findex.php%2Fpln%2Farticle%2FviewFile%2F5422%2F3186&data=02%7C01%7CR.Krishnamurthy%40aston.ac.uk%7Ce64199f8fde7404b1fd908d4803d973a%7Ca085950c4c2544d5945ab852fa44a221%7C0%7C0%7C636274449429782526&sdata=wGIvN%2FvBepjqO%2FBTJPpdxXXAjoc2Pqn2Q9o0Hrsj%2Bw8%3D&reserved=0

The problem that I was referring to can be seen from two perspective or traditions. One is a more linguistic tradition (corpus linguistics) and the other a more NLP/Computer Science tradition (computational linguistics).

The corpus linguistics perspective probably starts with Firth: ?You shall know a word by the company it keeps?. This is in line with Ramesh?s comments about meanings arising from contexts. I would certainly agree with this but context is a big word. Context may be a horizon (or window) of words 5 to the left or 5 to the right of a node word. It may be a dynamic window so context can extend a lot further. It may be the whole document or the meaning of the word may be impacted on by the genre or register. Word meanings can predict context and be predicted by context.

Of course, a parser (or a POS tagger) can pick up (as implied by Jorge) that ?smashing? is an adjective in one case and a verb in another. So, depending on its local grammar, the evaluative meaning of a word may change in the sense of the link between specific language patterns and evaluation as proposed in Hunston & Sinclair (2000).

As Sinclair showed many years ago, words like ?back? have many multiple senses and can be found in negative idioms such as ?back the wrong horse?, ?back against the wall? etc. but also in positive idioms such as ?back on track?. So, as Valerio has suggested, one could use synsets where, for different meanings of the same word, a different polarity score is given. They introduce the term polypathy as the property of a word of having different senses spread apart on the polarity scale (and the polypathy of a lemma being calculated as the standard deviation of the polarity scores of the possible senses of the lemma). This solution is lexicon-based and you need to plug in to the software system the positive and negative scores in SentiWordNet.

I was thinking more on the lines of research carried out in the computational linguistics tradition. Recent vector-based models (for example, https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fcode.google.com%2Farchive%2Fp%2Fword2vec%2F&data=02%7C01%7CR.Krishnamurthy%40aston.ac.uk%7Ce64199f8fde7404b1fd908d4803d973a%7Ca085950c4c2544d5945ab852fa44a221%7C0%7C0%7C636274449429782526&sdata=xMyCrOHE4ZzWYr1sU4buzG%2B5UMIdQYE6x6DXkPFetS4%3D&reserved=0) can capture the rich relational structure of the lexicon by encoding distributed numerical representations of word features, features such as the context of individual words. They take a text corpus as input and produce the word vectors as output. So, what I would be interested in working on, at least theoretically, would be how to use the vector representation of words to predict the sentiment annotations on the words from the contexts in which they appear. Best, Keith -------- From: Krishnamurthy, Ramesh Sent: 06 April 2017 10:20 To: kstuart at idm.upv.es Subject: [Corpora-List] Words ambiguous for sentiment

hi keith

i'm fine... how are you? :)

i saw this post of yours in corpora-list, and was a bit confused...

#1 you asked about "a list of words (or phrases) that, in different contexts, change from having a positive to a negative sentiment"

#2 you then gave, as an example of what you were looking for, "It was a /SMASHING/ success. ...a new Islamic State video showed militants /SMASHING/ ancient artifacts at a museum in Mosul."

#3 you then said "I am aware that these are different senses of the word 'smashing' but I am still interested in getting a list of these kind of words."

Your examples are not only different *senses* of the root 'smash'... but also different wordclass usages (adj | verb), as well as different contexts.

A) I think meanings (semantics) arise from contexts (which includes all participants)... so individual words may contribute to meanings, but those meanings are not so much inherent in the words, but in their frequent usage in the same/similar contexts...

B) the 'sentiment' associated with the meanings arising from contexts are also variable, and depend on those contexts, not on the specific words used in them?

C) i am not sure what the relationship is between 'sentiment' and 'prosodic value'...

D) please *do not quote this* at the moment, but i have a couple of papers 'in press/forthcoming'... in which i have tried to address a similar issue, i think...

"Malinowski said language was a “mode of action” (see Section 2), and Firth said linguistics was “the study of ‘processes and patterns of life’” (cited by Lukin 2014), using ‘process’ to describe “both the linguistic and non-linguistic aspects of the context of situation”. Krishnamurthy (forthcoming) sees collocation as “the mechanism, process, and force that enacts language change”. Meaning arises from context, and context is continually changing, so collocational changes are actually the process of meaning formation, and language change merely the sum of the collocational changes. Krishnamurthy (2003) notes the changes in collocation from sleazy (a 17th century loanword used to describe places and activities in a mildly pejorative way (unattractive, cheap, disreputable) to sleaze, a c. 1980 back-formation, which soon becomes highly political (government, party, MP) and highly pejorative (scandal, corruption). More recent work on diachronic collocation, e.g. Alba-Salas (2007) and Kehoe and Gee (2009), may strengthen this view. Perhaps collocation affects semantics, and semantic prosody affects pragmatics, especially evaluation. The more frequently that words are used, the weaker or less specific their meaning seems to become. Krishnamurthy (2002) shows how sexy (first attested in c. 1890 with the negative meaning ‘engrossed in sex’; used positively in the 1920s, meaning ‘sexually attractive’) has increased rapidly in usage.... But the most striking feature of the data in general is actually the absence/loss of the connection with women, or indeed with sexuality at all. A large proportion of the occurrences have a general positive prosody (‘good, exciting, fashionable’), and journalists use it to describe anything and everything (e.g. cities, foods, sports, computers, politics, science, the arts, success, careers, pensions, price stability). Frequently used words also seem to tend towards one prosodic pole (whether positive or negative), and some may even flip polarity from time to time, e.g. funky, bad, wicked, sick."

E) have you read Patrick Hanks "Enthusiasm and Condescension"? (first google hit for: hanks enthusiasm condescension) Those "words" also appear to have flipped their prosodies in most contexts since the 18th C.

F) if you are including semantically auto-antonymous words (eg "sanction"? = to allow sb to do sth AND to prohibit sb from doing sth; according to context), then surely every word in any language is *capable* of such a switch? Another word like this i remember from the mid-1980s is "fearful".... (who experiences the fear?) ... he was fearful of the consequences of his action... his actions had the most fearful consequences...

I'm still not sure if I have understood the nature of the items that you are trying to collect, or want a list of... but do let me know if any of my points make sense, or are at all relevant to what you are trying to do... :)

best ramesh ----------

Date: Wed, 05 Apr 2017 20:28:34 +0200 From: Keith Douglas Charles Stuart <kstuart at idm.upv.es> Subject: [Corpora-List] Words ambiguous for sentiment To: CORPORA at uib.no

Dear Colleagues,

I was wondering if anyone had ever produced a list of words (or phrases) that, in different contexts, change from having a positive to a negative sentiment. For example,

It was a /SMASHING/ success. ...a new Islamic State video showed militants /SMASHING/ ancient artifacts at a museum in Mosul.

I am aware that these are different senses of the word 'smashing' but I am still interested in getting a list of these kind of words.

Thanks,

Keith Stuart -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/html Size: 13828 bytes Desc: not available URL: <https://mailman.uib.no/public/corpora/attachments/20170411/8dcd4c28/attachment.txt>



More information about the Corpora mailing list