Printable View
Do think it would be easier to read it all nouns were blues, all verbs were red, all adjective green, articles, periods, commas, and question marks were bold, etc?? It might take some getting use to, but after that it might make an interesting cognitive experiment.
That's a rather bizarre suggestion. For a word-processing program, it may be okay when reading through a document to proofread it, but on the forums, there's practically no use for it.
How would inserting distracting, redundant information make things easier? If anything I think it would make it harder to read. Or are you saying you have trouble identifying whether a word is a noun or a verb without some kind of hint?Quote:
Originally Posted by thetinman
Highlighting a natural language (is that the right term?) such as English would not be like highlighting source code. When one reads English, one doesn't need highlighting to tell what is a verb and what isn't. We're so used to reading English that our brains don't need any clues or hints.
Unlike syntax highlighting for source code, I think that syntax highlighting for English would only be useful if you were doing some sort of analysis of text, like counting the average number of verbs per sentence. For ordinary reading it would just be annoying, unless the colours only contrasted slightly (which sort of defeats the purpose).
Another point: English is meant to be read in a linear fashion. When you read code your eye jumps around.
Besides, it would be a rather massive undertaking. You'd need a dictionary, not just a simple set of rules like you need for highlighting programming languages.
This is the worst idea in the history of the world.
1.Quote:
Originally Posted by Govtcheez
>Besides, it would be a rather massive undertaking.
Actually, part-of-speech tagging (POS Tagging) is a pretty common task in many areas of natural language processing. There are many freely available POS taggers.
It would require a bit more than just a dictionary. Take this very good example, almost identical to the title of a well-known book:Quote:
Originally Posted by dwks
"He eats shoots and leaves."
Is the word "shoots" a verb or a noun in this sentence? It could refer to shoots as in, "bamboo shoots," in which case it is a noun. Or it could refer to shoots as is "shooting a weapon." Without context, there is no way to determine which sense is meant. It rises above the level of grammar into semantics.
I suppose you could argue that the presence of commas in certain places might make it unambiguous. But commas are not a part of natural English -- they are invented to make ambiguous written statements easier to parse. In spoken English there is no comma (although there is usually a brief pause where one would be).
And what if you're parsing ungrammatical or imperfect English? The highlighting would be in contradiction to the intended meaning and only serve to confuse the reader.
Kind of like code getting out of sync with comments, whenever you have redundant information there is a risk of it not being self-consistent. Better to have a single but potentially ambiguous source of information than risk introducing paradoxes.
Considering cboard's recent past, I can't tell if you're serious or simply a master of biting sarcasm. Good job on that.Quote:
Originally Posted by thetinman
I think it depends, we don't need it, therefore it would actually be an annoyance. But if someone where to grow up learning english that way, then they would become dependent on it and would have a hard time without it. Again, it's all about conditioning. Unique idea though.
I think it would make reading it easier. I would suggest code as a good example. If many of the words had a color, I could determine what type of word it was with out looking at its letters.Quote:
Originally Posted by brewbuck
I could say the very same thing about code.Quote:
Originally Posted by dwks
Actually my eyes often jump around when I read English., and often I find my self reading code in a somewhat linear fashion. I'm not quite sure that this is relevant though.Quote:
Another point: English is meant to be read in a linear fashion. When you read code your eye jumps around.
Besides, it would be a rather massive undertaking. You'd need a dictionary, not just a simple set of rules like you need for highlighting programming languages.
It would not be that large an undertaking. Dictionary databases already exist, and the use of artificial intelligence could make some things easier.
Most people are able to determine the type of word it is by knowing the language they are reading.Quote:
Originally Posted by thetinman
Besides, how the hell do you propose figuring out what the word actually is besides "looking at its letters"?
I'm surprised that the forum seems to have such an aversion to the idea.Quote:
Originally Posted by Govtcheez
It is a horrible, horrible idea that will be of no use to anyone and would require considerable programming to accomplish. It's not going to happen.Quote:
Originally Posted by thetinman
I would say that commas are part of natural written English.Quote:
Originally Posted by brewbuck
If the english is ungrammatical or imperfect, then the system will not work as well. The same thing can be said about code. Another thing I could point out, is that each code editor has a different scheme for syntax highlighting. There is no standard highlighting scheme, yet one has little trouble reading code in an unfamiliar code editor.
> The same thing can be said about code.
The forums are not a compiler, you dope.
I think it would engage the senses more fully. In 1900 one could have said that adding audio to motion pictures would just be redundant and an annoyance. Today motion pictures with audio is the standard. I think most would agree with me that motion pictures are much better as a result.Quote:
Originally Posted by Queatrix
It makes reading code more pleasant. I think it would it would do the same for English. If you had access to a dictionary data base, it could be coded in a matter of hours.Quote:
Originally Posted by Govtcheez
I wasn't suggesting this for use on the forums.Quote:
Originally Posted by Govtcheez
This is off topic.Quote:
you dope.
My mistake then. It seems I wasn't the only person making that assumption, though.Quote:
Originally Posted by thetinman
If you're going to use it as a tool to teach ESL or something like that I can see a use for it, but for general use anywhere else it'd be very grating.
Nice hand-waving there. It could not be "coded in a couple of hours." Once again, "He eats shoots and leaves." Is shoots a verb or a noun? How the hell is a dictionary going to help?Quote:
Originally Posted by thetinman
Even in C there would be problems. The classic example is the hack between the parser and the lexer where an identifier is retyped to be a typename by a typedef. The compiler itself is unable to determine on a merely syntactic level whether a particular symbol is an identifier or a typename. Therefore some weird back-channels between the parser and the lexer are required.
Basically, you don't understand anything about languages.
that doesn't make sense. Art imitates life in most cases, audio to motion picture was portraying film the way plays, operas, ballet and others have been displaying stories hundreds of years before motion picture. Audio is a part of life and that needed to be reflected if motion picture was going to be worth a damn 100 years later. What does spell checking imitate? The annoying guy who interrupts your story so that he can say, "It's 'Exacerbate' not 'eggsasibate'" before I punch him in the face.Quote:
Originally Posted by thetinman
It is an interesting idea, actually. The implementation must be very robust, or else the result will only be confusing.
It might actually improve the speed of reading, but it would probably take a while before the brain got used to take advantage of the extra information.
EDIT: I agree that implementing it here would be a horrible idea, but the concept itself is interesting.
What extra information?Quote:
Originally Posted by Sang-drax
Umm, the colors and the boldness of the words. :rolleyes:Quote:
Originally Posted by brewbuck
Note that this thread was from May 16, 2 weeks ago, and was last replied on that same day before these new ones. I've already mentioned my thoughts on this - worthy only when proofreading a story or the like, but useless otherwise.
Which is extra information how? I think you don't understand what "information" is.Quote:
Originally Posted by Sang-drax
If the sentence can be highlighted using only the information available within the sentence, then BY DEFINITION the highlighting adds NO NEW information.
It's not highlighting based on only the info in the sentence, though. It's going to have to be cross referenced against an extensive dictionary. You can't just look at a string of words and determine the parts of speech without knowing things about grammar and how specific words act. That's the extra information.Quote:
Originally Posted by brewbuck
I don't know why I am answering you at all...Quote:
Originally Posted by brewbuck
Yes, it is extra information. A simpler example is source code. Nothing in the source code defines which words are keywords and which words aren't. The syntax highligher uses external information from a dictionary and adds new information to the source code from this dictionary by making certain words bold. I hope you understand now.
The LANGUAGE defines it. Strings of lexemes do not stand on their own, but are only meaningful with respect to a grammar. Just because the first line of a C program isn't "using language c;" doesn't mean this information is not available.Quote:
Originally Posted by Sang-drax
I understand that you obviously have no knowledge of either information theory or linguistic theory, if that's what you mean.Quote:
I hope you understand now.
Let's throw a bit of theory into this, shall we:
Computational processing of textual data
Interesting parts:
Part-of-Speech tagging (Brill, HMM)
Part-of-Speech tagging (HMM ctnd.)
Then I recommend "Parsing, formal grammars" and "Stochastic Parsing", as well as "Classification - Visualization". As long as we don't have the same theoretical background, there's no use discussing anything. Personally, I took computational processing of textual data and "natural language processing" (audio) and some opinions/beliefs in this thread amuse me.
How's about you and brewbuck drop a little science and enlighten us? That's certainly more productive than saying "I'm right hahaha"Quote:
Originally Posted by KONI
This idea is interesting.
brewbuck, if you encountered a language you've never seen before, for example Hindi, then I am sure it would be impossible for you to figure out which words that are nouns. You would probably not even recognice the characters used.
It is impossible for you to determine where the nouns are -- the information just isn't there.
But! After syntax highlighting using a human or an extensive database of words and grammar it would be very easy for you to see the nouns -- they would be bold. There's your extra information.
The problem here is that you don't seem to acknowledge the external database/human experience as additional information. I don't really understand why, but obviously you must have some deep insight in information theory that I don't. I skipped that course. Feel free to enlighten me with arguments other than what subjects I don't understand.
I don't get how that could ever be helpful... When I'm reading, I don't give a damn if a word is a verb or a noun, I just naturally know it. English isn't even my native language... Awful idea..
Sometimes to undersand a sentence correctly yo HAVE to know if the word is noun or verb - it can greatly change the meaning of the sentence... Not all people can do it easely with the foreign language... especially if it is english that does not bother to distinguish verbs and nouns with some additional parts of the word...Quote:
Originally Posted by Desolation
You're completely forgetting the fact that natural language is by definition ambiguous.
Sometimes writers don't want you to know if something is a verb or a noun and they make full usage of the two very different meanings the phrase would have to express a more subtle message.
Go literature ! \o/
I agree that it's interesting, but it doesn't really have any other than trivial use.
Yeah, but seriously, how common is that?Quote:
Originally Posted by KONI
All the time, if you consider slang and colloquialisms (sp?)
"He hits a grounder to short."
"I need to code for awhile if I'm ever going to finish this project."
First of all, grounder isn't even a "proper" English word (I think). Second, "short" is usually an adjective; here it's a noun. The word "code" in the second sentence is usually a noun, but, as well all know, it's used as a verb quite often. Using a simple dictionary and assigning each word to a color would never work. One of the things I always hear about English is that it's a hard language to learn because it has so many "exceptions to the rule." Because of that, this is going to be an immense project.
You may have to implement it by looking at the general form, instead. By looking at the sentence structure, I think you can determine that "hits" is a verb because "He" is a pronoun. If "hits" is a verb, then "grounder" must be a noun, since "a" is a determiner (pretty sure, anyway). If all of this is true, we can assume that "short" is a noun, since "to" is a preposition.
I did not really double check to make sure all of that is correct, but you get the idea. Instead of focusing on what each word means, you'd probably have to start with the underlying structure, and determine where the determiners, prepositions, and the like are. Then you can build up to the nouns and verbs. Probably. Maybe.
Since there is a set of rules for grammar, it can probably be done. It will be no small task, though. As an ESL teaching tool, it might be pretty useful, actually, since it focuses on the abstract sentence structure, instead of something like "this word is a noun, except in the following cases..."
No, there "to short" may also be a verb.Quote:
Originally Posted by Decrypt
No, Decrypt's point about American English is correct. In baseball, the shortstop is sometimes called short (especially by the announcers). In the sentence "He hit a grounder to short," the shortstop is the indirect object. Provided an ESL student has any inkling of what those are, he should be able to understand it is a noun. But this is where the syntax highlighter could come in and clear things up, because to short is not a verb; to shorten is.Quote:
Originally Posted by Rashakil Fol
Ah, how baseball has impacted American culture.... :)
Play with electricity much? Or stocks?Quote:
Originally Posted by citizen
Fair enough. Still, there's context to wrestle with and I don't think that particular sentence makes sense in any other way. I could be wrong, but it doesn't go against his real point anyway. The machine would have to be able to understand context to work perfectly, and I think he knows that.
On it's own it can be, but, in context, I'm not sure that's the case. If we've determined that the sentence so far is <pronoun> <verb> <determiner> <noun>, I'm not sure that a verb in its general form "to short" makes sense, assuming complete sentences. You're right, the sentence could just as easily read "He sells a stock to short," but the sentence is incomplete; another noun is required, I think: "He sells a stock to short it." Due to that, I think something like the above analysis would work. This is English we're talking about, though, so nearly anything is possible.Quote:
No, there "to short" may also be a verb.
The idea is that, if this highlighter is to work, I think that it'd have to use general grammatical rules instead of a strict set of uses for each word as mentioned above. However, to start, you'd have to have some set of words whose use is iron-clad, and, in the end, you'd probably have to use both a set of grammatical rules and a database of words and their uses to implement it properly.
You're discussing part-of-speech tagging of natural language here, which is considered "solved" (almost solved) and for which there exist very robust programs already. I linked to some part of speech tagging methods in post #31, namely Brill and HMM. They basically have a forest and maximize the probability of several trees producing the given phrase.
This already works very well (depending on the vocabulary, slang, out-of-dictionary forms etc). What the discussion should focus on is, if in natural language, this a) is useful and b) takes away the subtlety of our language, like jokes, literature or even "sexual" ambiguous remarks.
Oops, I forgot to follow the links. As far as it's use, I think we agree it's pretty limited. It may help as a teaching tool, but, other than that, I don't think it provides much benefit.
p.s. Richard is my hero
But now we're way out of the realm of what could be coded up "in a few hours." We're talking full part-of-speech tagging. People spend years writing PhD theses on just a little tiny corner of the theory.Quote:
Originally Posted by Decrypt
The original argument was whether highlighted text would be easier to read. Part of speech tagging is possible and useful, but making sentences easier to read is not one of its uses.