rant about exploiting the preparser: Difference between revisions
Jump to navigation
Jump to search
mNo edit summary |
mNo edit summary |
||
| Line 1: | Line 1: | ||
moved from [[jbocre: exploiting the preparser|exploiting the preparser]] | |||
---- | |||
I'm rather suspicious of suggestions coming from you as to how to modify the lexical or parsing stages of machine recognition, as I recall it taking a rather significant amount of effort to explain the parsing process to you. And there are always things which end up not being easy to say in a language. You cannot ensure that every utterance that ''you think might be popular in your broken version of Lojban'' is short. Further, you have no statistically significant evidence of anything needed to be made more concise, one way or the other. Without such evidence, suggestions of shortening things is just so much hot, unwanted air. | |||
---- | |||
I moved this from the original page, because this bit: | |||
;:"You cannot ensure that every utterance that ''you think might be popular in your broken version of Lojban'' is short. Further, you have no statistically significant evidence of anything needed to be made more concise, one way or the other." | |||
lowers the calibre of discussion insufferably. The least that can be expected is that the commentary actually address the content of what is commented on and not some travesty of it. For general ideological commentary, there are separate pages where that can be discussed. | |||
''I'm rather suspicious of suggestions coming from you as to how to modify the lexical or parsing stages of machine recognition, as I recall it taking a rather significant amount of effort to explain the parsing process to you.'' | |||
* If you don't understand the parsing process well, then this is fair comment, and you can wait until someone competent passes comment. If you are competent to judge the feasibility of exploiting the preparser in this way, then you can do so. | |||
''And there are always things which end up not being easy to say in a language.'' | |||
* That cannot fail to be true; how could it be otherwise? But you miss the point that there is in language a relationship between length and frequency: the more frequently something is said, the shorter it gets. The proposal to exploit the preparser was intended to provide a way to emulate this property of natural language, which is otherwise denied to Lojban. | |||
''You cannot ensure that every utterance that ''you think might be popular in your broken version of Lojban'' is short.'' | |||
* (am at a loss to find tactful response to this...) | |||
''Further, you have no statistically significant evidence of anything needed to be made more concise, one way or the other.'' | |||
* Indeed not, that is why "Some sort of statistical calculation could be performed on a large body of mature usage in order to ascertain what sort of string substitutions would yield the greatest gain in terms of concision." | |||
''Without such evidence, suggestions of shortening things is just so much hot, unwanted air.'' | |||
* What would it be with such evidence? Not hot unwanted air? The page you were commenting on is unmistakably located in a section where you cannot possibly be misled into thinking you are being offered something that you want. You know in advance that everything in that section is something you don't want or value. Furthermore, if you are Jay, then aren't you doing an MA in linguistics? I find it incredible that someone doing an MA in linguistics thinks it is unwanted hot air to point how Lojban's unambiguous grammar could emulate natural language processes of shortening. If you're not Jay, then apologies to Jay. | |||
---- | |||
I don't know if we have anything quite as silly as ''que est-ce que c'est'', | |||
but it is one of the charms of linguistic exoticism that languages differ | |||
wildly in what they feel a need for concision in (c.f. "They Have a Word For | |||
It"). | |||
* ''que est-ce que c'est'' = [[jbocre: keskse|keskse]] -- not so long. We can find very short words in English that are very rare. But how often do we find long words or phrases that are of very high frequency? In BrE, for example, the apparently long but common word ''particularly'' -- 5 sylls in citation form -- is reduced to /p(@)tIkli/ -- 2 sylls. | |||
( | |||
Revision as of 17:09, 4 November 2013
moved from exploiting the preparser
I'm rather suspicious of suggestions coming from you as to how to modify the lexical or parsing stages of machine recognition, as I recall it taking a rather significant amount of effort to explain the parsing process to you. And there are always things which end up not being easy to say in a language. You cannot ensure that every utterance that you think might be popular in your broken version of Lojban is short. Further, you have no statistically significant evidence of anything needed to be made more concise, one way or the other. Without such evidence, suggestions of shortening things is just so much hot, unwanted air.
I moved this from the original page, because this bit:
- "You cannot ensure that every utterance that you think might be popular in your broken version of Lojban is short. Further, you have no statistically significant evidence of anything needed to be made more concise, one way or the other."
lowers the calibre of discussion insufferably. The least that can be expected is that the commentary actually address the content of what is commented on and not some travesty of it. For general ideological commentary, there are separate pages where that can be discussed.
I'm rather suspicious of suggestions coming from you as to how to modify the lexical or parsing stages of machine recognition, as I recall it taking a rather significant amount of effort to explain the parsing process to you.
- If you don't understand the parsing process well, then this is fair comment, and you can wait until someone competent passes comment. If you are competent to judge the feasibility of exploiting the preparser in this way, then you can do so.
And there are always things which end up not being easy to say in a language.
- That cannot fail to be true; how could it be otherwise? But you miss the point that there is in language a relationship between length and frequency: the more frequently something is said, the shorter it gets. The proposal to exploit the preparser was intended to provide a way to emulate this property of natural language, which is otherwise denied to Lojban.
You cannot ensure that every utterance that you think might be popular in your broken version of Lojban is short.
- (am at a loss to find tactful response to this...)
Further, you have no statistically significant evidence of anything needed to be made more concise, one way or the other.
- Indeed not, that is why "Some sort of statistical calculation could be performed on a large body of mature usage in order to ascertain what sort of string substitutions would yield the greatest gain in terms of concision."
Without such evidence, suggestions of shortening things is just so much hot, unwanted air.
- What would it be with such evidence? Not hot unwanted air? The page you were commenting on is unmistakably located in a section where you cannot possibly be misled into thinking you are being offered something that you want. You know in advance that everything in that section is something you don't want or value. Furthermore, if you are Jay, then aren't you doing an MA in linguistics? I find it incredible that someone doing an MA in linguistics thinks it is unwanted hot air to point how Lojban's unambiguous grammar could emulate natural language processes of shortening. If you're not Jay, then apologies to Jay.
I don't know if we have anything quite as silly as que est-ce que c'est,
but it is one of the charms of linguistic exoticism that languages differ
wildly in what they feel a need for concision in (c.f. "They Have a Word For
It").
- que est-ce que c'est = keskse -- not so long. We can find very short words in English that are very rare. But how often do we find long words or phrases that are of very high frequency? In BrE, for example, the apparently long but common word particularly -- 5 sylls in citation form -- is reduced to /p(@)tIkli/ -- 2 sylls.