Difference between revisions of "internal grammar of tags"

From Lojban
Jump to navigation Jump to search
m
 
m
Line 1: Line 1:
  
Sketch for a beginning of a position on Figurative Language:
+
'''Tense grammar simplification proposal''', by [[User:xorxes|xorxes]]
  
It seems to me that since there is ''pe'a'' & since ''broda brode'' is
+
=== Current grammar: ===
  
explicitly defined as a combination of ''broda'' with its places plus
+
^
  
''brode'' with its places, we are left with (A) using in formal speech
+
tense-modal = simple-tense-modal # | FIhO # selbri /FEhU#/
  
''broda brode'' or ''broda pe'a brode'' or ''broda brode pe'a'' or ''fu'epe'a''
+
simple-tense-modal = [[jbocre: NAhE|NAhE]] [[jbocre: SE|SE]] BAI [[jbocre: NAI|NAI]] [[jbocre: KI|KI]]
  
''broda brode fu'o'' regardless of how clumsy it seems to our ''glibau''
+
| [[jbocre: NAhE|NAhE]] (time [[jbocre: space|space]]  space [[jbocre: time|time]]) & CAhA [[jbocre: KI|KI]]
  
sensibility, in order to express the various things that a ''tanru''
+
| KI
  
can be; & (B) the informal possibly of occasionally dispensing with
+
| CUhE
  
''pe'a'' where it can be understood as elidable.
+
time = ZI & time-offset ... & ZEhA [[jbocre: PU [jbocre: NAI|NAI]] & interval-property ...
  
----
+
time-offset = PU [[jbocre: NAI|NAI]] [[jbocre: ZI|ZI]]
  
Note that sections 14 & 15 of Chapter 5 in The Book have such ''tanru''
+
space = VA & space-offset ... & space-interval & (MOhI space-offset)
  
as: ''snime nanmu'' for "snowman" & ''kensa bloti'' for "spaceship". (I
+
space-offset = FAhA [[jbocre: NAI|NAI]] [[jbocre: VA|VA]]
  
would add a ''pe'a'' to the second term of each, which were otherwise
+
space-interval = [[jbocre: VEhA & VIhA) [jbocre: FAhA [jbocre: NAI|NAI]]) & space-int-props
  
better ''nanmytai'' & ''velkla''. --''la maikl.'')
+
space-int-props = (FEhE interval-property) ...
  
''I don't know who's talking above. I don't agree that a snowman is not a nanmu: it is very typical to extend the tertau before restricting it further, as in "stone lion", where obviously all real lions are made out of meat. --[[jbocre: John Cowan|John Cowan]]''
+
interval-property = number ROI [[jbocre: NAI|NAI]] TAhE [[jbocre: NAI|NAI]] | ZAhO [[jbocre: NAI|NAI]]
  
* In the same vein, I like the example ''time travel''. "Travel" means "going to other places", but "time travel" means "going to other times". ''mi'e [[jbocre: jezrax|jezrax]]''
+
^
  
----
+
=== Proposed grammar: ===
  
mi'e xod .i tezu'e ma pilno zo pe'a .i di'e mupli ca'i mi
+
^
  
"do gerku ki'u le du'u do dukse gletu" .i do ca'a remna jena gerku .i zo
+
tense-modal = ([[jbocre: NAhE|NAhE]] [[jbocre: SE|SE]] tag-unit [[jbocre: NAI|NAI]] #) ...
  
pe'a sarcu
+
tag-unit = BAI
  
"mi pu tcidu le balcukta" .i pe'i la ueb. cu cukta .iki'ubo zo pe'a na
+
| CAhA
  
sarcu .i ju'ocu'i do tugni .o'u
+
| CUhE
  
Look, the fact that the Book tells us that "gerku zdani" can refer to the White House because a dog once chased Bill Clinton's cat shows the limit of figurativeness in tanru without invoking pe'a. If you can show ANY relationship between the parts, it's valid. You are advised, according to your desire to be understood, to try to pick understandable relationships. But it's not until we start ''really'' getting figurative that we ever need any pe'a. If I call your face a papri because I can easily read your emotions, then I should use a pe'a. My understanding of Helsem's statements is that he wants us to invoke pe'a too much.
+
| KI
  
(''sa'e banzu zo'o'')
+
| ZI
  
----
+
| PU
  
For another position, [[jbocre: put out your eyes as if they were a fire|put out your eyes as if they were a fire]].
+
| VA
 +
 
 +
| [[jbocre: MOhI|MOhI]] FAhA
 +
 
 +
| ZEhA
 +
 
 +
| VEhA
 +
 
 +
| VIhA
 +
 
 +
| [[jbocre: FEhE|FEhE]] number ROI
 +
 
 +
| [[jbocre: FEhE|FEhE]] TAhE
 +
 
 +
| [[jbocre: FEhE|FEhE]] ZAhO
 +
 
 +
| FIhO # selbri /FEhU/
 +
 
 +
^
 +
 
 +
=== Rationale: ===
 +
 
 +
==== SE-conversion ====
 +
 
 +
Every tag-unit can be used as a tag, and therefore as a connective. It is arbitrary and inconvenient that SE is currently disallowed with some tags.
 +
 
 +
==== NAhE ====
 +
 
 +
{NAhE PU}, {NAhE CAhA} and {NAhE PU CAhA} are all allowed, but{NAhE PU NAhE CAhA} is not. This is arbitrary and inconvenient. Similarly for other combinations.
 +
 
 +
==== Order of units ====
 +
 
 +
{co'a na'o broda} "starting to typically broda" is allowed, but {na'o co'a broda} "typically starting to broda" is not. It will be accepted by the parser, but parsed as {na'oku co'a broda}. Similarly for all other order restrictions. Note: arbitrary combinations of tag-units are already allowed in selbri-tags as long as there is an intervening {ja'a}, for example: {na'o ja'a co'a broda} is allowed, without ku's.
 +
 
 +
==== NAI ====
 +
 
 +
There's no good reason to allow it selectively here and there instead of everywhere.
 +
 
 +
==== Backwards compatibility ====
 +
 
 +
Fully compatible. Everything currently grammatical remains grammatical.
 +
 
 +
--------------------
 +
 
 +
=== Notes ===
 +
 
 +
#In the original proposal I had kept PU [[jbocre: ZI|ZI]], FAhA [[jbocre: VA|VA]], ZEhA[[jbocre: PU|PU]], VEhA [[jbocre: FAhA|FAhA]] and VIhA [[jbocre: FAhA|FAhA]] as separate forms because I thought their compound meaning might follow special compositional rules. I don't think that is the case, though. Just as the Imaginary Journey composition follows the ordinary left-to-right scope rule, these componds follow the rule too. For example {ze'u pu} indicates a long duration of an event in the past of some reference point, where the event is in the past for the whole duration.
 +
 
 +
#Originally I had only redefined the simple-tense-modal, but since the exclusion of FIhO-modals from the general case was due to the LR(1) restriction which no longer applies, we can now generalize the full tense-modal.
 +
 
 +
--------------------
 +
 
 +
(comments)
 +
 
 +
And's:
 +
 
 +
SE: Absolutely, yes.
 +
 
 +
NAhE: Yes.
 +
 
 +
Order: Is {lo na'o(ku) co'a broda} grammatical? If not, then that is an argument in favour of your proposal. If it is grammatical, then I think it would be better if all selbri tags were instead sumti tags, since otherwise we have a syntactic distinction with no semantic import.
 +
 
 +
*{lo na'oku co'a broda} is not grammatical. {lo na'o ja'a co'a broda} is grammatical. But having to remember for which combinations you need to insert ja'a is absurd.
 +
**OK, then. I am in favour.
 +
 
 +
NAI: Certainly the status quo seems arbitrary. But IMO NAI is a Bad Thing when it contributes to logical form, because it doesn't follow the usual scope rules. Allowing NAI everywhere is probably better than allowing it arbitrarily, but better would be to disallow it everywhere except for places where na can't do the job.
 +
 
 +
*NAI only affects the meaning of the previous word. For example {ru'inai} means "intermittently". It follows the usual scope rule for UIs, the scope is always the previous word. If you prefer, the complex word+nai is a new word with a new meaning. The new meaning is not strictly compositional, but it is usually easy to guess.
 +
**So it functions like NAhE, then? I agree the status quo is an ugly mess, but the risk of fixing it by allowing NAI anywhere is that we end up with a semantic mess. Are we (BF) really going to say for every cmavo what it means when followed by NAI? Or is it like a tanru, dependent on glorking?
 +
 
 +
***I will restrict it, for the purposes of this proposal, to words in tags. NAI is already allowed after most of them anyway, so we already have to do that.

Revision as of 16:52, 4 November 2013

Tense grammar simplification proposal, by xorxes

Current grammar:

^

tense-modal = simple-tense-modal # | FIhO # selbri /FEhU#/

simple-tense-modal = NAhE SE BAI NAI KI

| NAhE (time space space time) & CAhA KI

| KI

| CUhE

time = ZI & time-offset ... & ZEhA [[jbocre: PU [jbocre: NAI|NAI]] & interval-property ...

time-offset = PU NAI ZI

space = VA & space-offset ... & space-interval & (MOhI space-offset)

space-offset = FAhA NAI VA

space-interval = [[jbocre: VEhA & VIhA) [jbocre: FAhA [jbocre: NAI|NAI]]) & space-int-props

space-int-props = (FEhE interval-property) ...

interval-property = number ROI NAI TAhE NAI | ZAhO NAI

^

Proposed grammar:

^

tense-modal = (NAhE SE tag-unit NAI #) ...

tag-unit = BAI

| CAhA

| CUhE

| KI

| ZI

| PU

| VA

| MOhI FAhA

| ZEhA

| VEhA

| VIhA

| FEhE number ROI

| FEhE TAhE

| FEhE ZAhO

| FIhO # selbri /FEhU/

^

Rationale:

SE-conversion

Every tag-unit can be used as a tag, and therefore as a connective. It is arbitrary and inconvenient that SE is currently disallowed with some tags.

NAhE

{NAhE PU}, {NAhE CAhA} and {NAhE PU CAhA} are all allowed, but{NAhE PU NAhE CAhA} is not. This is arbitrary and inconvenient. Similarly for other combinations.

Order of units

{co'a na'o broda} "starting to typically broda" is allowed, but {na'o co'a broda} "typically starting to broda" is not. It will be accepted by the parser, but parsed as {na'oku co'a broda}. Similarly for all other order restrictions. Note: arbitrary combinations of tag-units are already allowed in selbri-tags as long as there is an intervening {ja'a}, for example: {na'o ja'a co'a broda} is allowed, without ku's.

NAI

There's no good reason to allow it selectively here and there instead of everywhere.

Backwards compatibility

Fully compatible. Everything currently grammatical remains grammatical.


Notes

  1. In the original proposal I had kept PU ZI, FAhA VA, ZEhAPU, VEhA FAhA and VIhA FAhA as separate forms because I thought their compound meaning might follow special compositional rules. I don't think that is the case, though. Just as the Imaginary Journey composition follows the ordinary left-to-right scope rule, these componds follow the rule too. For example {ze'u pu} indicates a long duration of an event in the past of some reference point, where the event is in the past for the whole duration.
  1. Originally I had only redefined the simple-tense-modal, but since the exclusion of FIhO-modals from the general case was due to the LR(1) restriction which no longer applies, we can now generalize the full tense-modal.

(comments)

And's:

SE: Absolutely, yes.

NAhE: Yes.

Order: Is {lo na'o(ku) co'a broda} grammatical? If not, then that is an argument in favour of your proposal. If it is grammatical, then I think it would be better if all selbri tags were instead sumti tags, since otherwise we have a syntactic distinction with no semantic import.

  • {lo na'oku co'a broda} is not grammatical. {lo na'o ja'a co'a broda} is grammatical. But having to remember for which combinations you need to insert ja'a is absurd.
    • OK, then. I am in favour.

NAI: Certainly the status quo seems arbitrary. But IMO NAI is a Bad Thing when it contributes to logical form, because it doesn't follow the usual scope rules. Allowing NAI everywhere is probably better than allowing it arbitrarily, but better would be to disallow it everywhere except for places where na can't do the job.

  • NAI only affects the meaning of the previous word. For example {ru'inai} means "intermittently". It follows the usual scope rule for UIs, the scope is always the previous word. If you prefer, the complex word+nai is a new word with a new meaning. The new meaning is not strictly compositional, but it is usually easy to guess.
    • So it functions like NAhE, then? I agree the status quo is an ugly mess, but the risk of fixing it by allowing NAI anywhere is that we end up with a semantic mess. Are we (BF) really going to say for every cmavo what it means when followed by NAI? Or is it like a tanru, dependent on glorking?
      • I will restrict it, for the purposes of this proposal, to words in tags. NAI is already allowed after most of them anyway, so we already have to do that.