Making Convoluted Text Simple Visually
January 26, 2014 7:01 PM Subscribe
Sometimes, some documents I read are so convoluted that I don't understand what they are telling me. I've found this to be true in for legal documents including terms of agreements and constitutions among others. Is there any kind of program that looks at the syntax of sections of text and converts them into block diagrams showing the relationships between subjects and objects with the verb, adverbs, adjectives, etc. showing how they are connected? For instance, if it was highlighting the sentence, "See Spot run", there would be two boxes, one labeled Spot and one labeled You with an arrow connecting the latter to the former. I'm thinking of something similar to sentence diagramming but graphically represented and not nearly as complicated. It seems to me that if something could lay out all of the relationships within a document, that would make it much easier for someone to understand what it means. Or is that magical thinking on my part?
Response by poster: I'm thinking the software would render a organizational chart of the document, showing hierarchy as well as common connections and those that deadend.
posted by CollectiveMind at 7:45 PM on January 26, 2014
posted by CollectiveMind at 7:45 PM on January 26, 2014
Not magical thinking, but, unfortunately nontrivial. From the standpoint of academic disciplines, what you're asking about falls in the realm of computational discourse analysis. I've drifted away from the field, but there has been a lot of work done in Rhetorical Structure Theory that tackles exactly the problem you describe (showing relationships between chunks of discourse). RST posits a number of relationships between discourse chunks, such as background, motivation, restatement, and elaboration. There is a list of relationships on the page linked above. There are some (not for the faint of heart!) RST software tools available as well that you might try experimenting with.
In addition to BungaDunga's suggestion of Montessori diagramming, you might consider mind-mapping. I use mind-mapping sometimes for working with/understanding convoluted/rambling academic papers, as well as for mapping out my own nonfiction and fiction. You can google around for mind mapping software (there is a lot of it) if you'd prefer not to work on paper.
posted by skye.dancer at 8:22 PM on January 26, 2014 [2 favorites]
In addition to BungaDunga's suggestion of Montessori diagramming, you might consider mind-mapping. I use mind-mapping sometimes for working with/understanding convoluted/rambling academic papers, as well as for mapping out my own nonfiction and fiction. You can google around for mind mapping software (there is a lot of it) if you'd prefer not to work on paper.
posted by skye.dancer at 8:22 PM on January 26, 2014 [2 favorites]
Not magical thinking, but naive. There's a good reason legalese is inscrutable: it's deliberately designed to be comprehensible only to those with legal training. Obfuscating text to the point where only a lawyer can pick it apart creates employment opportunities for lawyers, and that's exactly why it's written the way it is.
The most egregious examples of this kind of creative obfuscation are to be found in patents, which is exactly how a potentially sound idea (rewarding the public disclosure of innovations) has come to be the biggest brake on innovation ever devised.
Every now and again some well-meaning public official will get a push going to enforce the use of "plain English" in consumer-facing legal documents like insurance policies and contracts. It never lasts very long, because lawyers are patient, persistent, usually quite smart, and often well-connected within the machinery of government.
The best parsers that exist for natural language all run on the human brain. If you want to understand legalese, go get some legal training.
posted by flabdablet at 11:05 PM on January 26, 2014 [1 favorite]
The most egregious examples of this kind of creative obfuscation are to be found in patents, which is exactly how a potentially sound idea (rewarding the public disclosure of innovations) has come to be the biggest brake on innovation ever devised.
Every now and again some well-meaning public official will get a push going to enforce the use of "plain English" in consumer-facing legal documents like insurance policies and contracts. It never lasts very long, because lawyers are patient, persistent, usually quite smart, and often well-connected within the machinery of government.
The best parsers that exist for natural language all run on the human brain. If you want to understand legalese, go get some legal training.
posted by flabdablet at 11:05 PM on January 26, 2014 [1 favorite]
Guidance in how to read legal documents from the Government of New South Wales.
posted by epo at 5:41 AM on January 27, 2014
posted by epo at 5:41 AM on January 27, 2014
flabdablet's opinion is a little cynical, legal documents have to be amenable to a single interpretation otherwise it creates problems later on. This necessarily involves spelling things out in detail and avoiding ambiguity.
This is not to say that some documents e.g. online T&Cs are not deliberately verbose and convoluted. Even placing a human readable summary at the front is fraught, as doing so might tend give it legal strength.
I do agree that the real problem is lawyers and legal documents that often exist solely at the behest of lawyers not because the circumstances require them
posted by epo at 5:54 AM on January 27, 2014
This is not to say that some documents e.g. online T&Cs are not deliberately verbose and convoluted. Even placing a human readable summary at the front is fraught, as doing so might tend give it legal strength.
I do agree that the real problem is lawyers and legal documents that often exist solely at the behest of lawyers not because the circumstances require them
posted by epo at 5:54 AM on January 27, 2014
About 15 years ago, I was at a natural language processing conference in Cambridge where SRI demo'ed a pretty printer/indent(1) for legalese. Now, this was a demo, so it showed the project to its best advantage, but it was explained that part of it was simple grammar matching, and the rest was based on heuristics from part-of-speech tagging from a corpus.
I also distinctly remember the presenter saying that the system had been bought outright by a law multinational, and so SRI couldn't offer it as a product ☹
posted by scruss at 6:30 AM on January 27, 2014
I also distinctly remember the presenter saying that the system had been bought outright by a law multinational, and so SRI couldn't offer it as a product ☹
posted by scruss at 6:30 AM on January 27, 2014
The prosecution rests, Your Honour.
posted by flabdablet at 8:26 AM on January 27, 2014
posted by flabdablet at 8:26 AM on January 27, 2014
This thread is closed to new comments.
Getting a program to do this is... hard. I googled "automatic sentence analysis" and got this which will try to parse a sentence.
posted by BungaDunga at 7:44 PM on January 26, 2014 [1 favorite]