Relaxing the cmene-Forming Rules
OK, dragging out the old bugbear: permitting la/lai/doi in cmene. I found the discussions we had of it back in the old PHPbb, and it needs to be brought up in the new tiki forum. It was the cause of a lot of argumentation, which after a while started to get repetitive.
I'm going to try to simplify things by stating the supporting evidence and arguments and also as many of the arguments against it that I can find, with my answers to them (if applicable--maybe I can't answer everything).
For starters, see the page that started a lot of this.
To recap: not even the most skilled Lojbanists can remember to keep those pesky syllables out of cmene with any decent consistency. Yes, I know, people do catch it (rarely), but nobody has even a halfway-decent record at this, even after doing Lojban for many many years. (I know what you're thinking, and no, not even you.) The claim is that what's truly in line with usage is to pitch the rule. In order to do that and retain audio-visual isomorphism, we need to introduce another front-end delimiter for cmene, and obviously choosing a syllable for it hasn't worked, so we would have to require that all cmene (even those starting with consonants) must be preceded by a pause in all contexts (even after la/lai/doi).
Proposal
The syllables la, lai, and doi, heretofore forbidden/restricted in Lojban cmene, shall be permitted in all positions of cmene (not just when preceded by a consonant). In compensation therefor, a pause will be mandatory before all cmene, even when preceded by the words la, lai, or doi.
Counter-Arguments
1. Well, I wouldn't make that mistake!
When I see the mistake showing up in .i la lojban. mo, which I know was written and painstakingly proofread by some of the most careful and skilled Lojbanists out there, I think I'm justified in saying that people can't be relied upon to keep to this rule; it is a rule that cannot be observed. (Ref: page iii of the aforesaid book, in the paragraph which thanks people who contributed to the book, including la stivn. *laitl.) When it shows up (years earlier) in ju'i lobypli in a comic strip, I can conclude that not even the revered Founders can be so sure of their judgment. (Ref: comic from some year, someone please help me here. A cat was addressed as doi *lat. This motivated a rule-change to permit la/lai/doi when they preceded by a consonant, thus permitting the similar form doi mlat.) The rule is broken.
2. It would make people talk... like... William... Shatner.
Pauses before names are already mandatory in most grammatical situations. It's only after three words that they are optional (and would become mandatory). Admittedly, those three words are far and away the most common words preceding cmene in actual usage (partly for that reason). People need to work on their pauses anyway. In mi se cmene zo .mark., the pause before the name is not optional, however often people pronounce it so. All the more so in the more common construction of mi'e .mark. (other COIs can be shielded by adding doi, but this is not always possible). It would not actually break up fluid speech appreciably, no more than it does already with vowel-initial names. There are ways to make it small enough: I suggest that people try to train themselves that la ends with a glottal stop, like the Klingon word la'. Makes things pretty easy. Does it require some re-learning? Is it likely to cause errors of omitting the pause? Yes, unfortunately. But see below.
Another point worth remembering is that names are actually not all that common in Lojban. Once introduced by name, people and things are usually referred to by bound KOhA variables, or more commonly BY lerfu-strings. These latter are particularly convenient for the purpose, as frequently they require no binding, and are shorter than the name. More argument on this tangential point is available on another page.
3. It's just another rule that people will mess up. Who can remember to pause anyway?
See above for tricks on how to remember to pause. But also, the errors in question are different in their impact. A missed pause happens in speech. Your hearers understand you or they don't. If they don't, woops, you have a problem that will have to be cleared up, and serve you right for forgetting the rule. If they understand you (as is more likely, in all fairness), then it's in the past and isn't really an error anymore, since everyone's parse of the text contains the correct words. When errors of illegal cmene happen, they happen in writing, too. Writing persists. Even if you're understood, everyone's parse of the text contains an illegal word, and if the text is in writing, it remains wrong for all to see (and learn bad habits from) for ever afterward. Bad cmene get into books (see above!) into cmene-lists, into common parlance, and remain there in perpetuity because they're enshrined in writing.
Additionally, the mistakes are different in kind, even if misunderstood. An illegal cmene that is interpreted according to the strict rules will usually result in a perfectly valid—but semantically wrong—sentence. For example, la stivn. *laitl. cu xamgu mi breaks down into la stivn. lai tl. cu xamgu mi, which is perfectly grammatical—but doesn't say anything like what we wanted. Now we have two sumti instead of one, which is fine by the grammar, but all the place-filling is off. On the other hand, if a pause is missed, you basically wind up with a big ol' cmene sitting there in the middle of the sentence with nothing before it, which will usually be a syntax error. mi viska la .bab., if said without the pause, becomes mi viska labab. or possibly even miviskalabab. (or any of the various other possibilities, depending on where optional pauses are taken.) Either way, it's a cmene where we aren't expecting a cmene, and so the error is signaled right away, at parse-time and not at interpretation-time, so to speak. (Now, this also leads one to consider the possibility of not requiring la before the name in the first place, since it isn't required to detect the start of the name anymore. Such a proposal would be worth considering, but it does raise some other issues, and it's not what we're talking about now.)
It really boils down to what kind of disaster you're willing to chance, and how serious we are about Lojban being unambiguous. Do you want to chance having a dialect, even a common or dominant dialect, that is technically incorrect when spoken but correct when written, and which can be spoken correctly with a little extra concentration, or do you want to completely lose the Lojban audio-visual isomorphism? That's really what's at stake here. Because keeping la/lai/doi out of cmene will not last long, indeed, it has already failed miserably. And it's not the same amount of "extra concentration" that it takes to fix them. It's easy to remember to throw in an extra pause, at least to know what to do once you realize there's a problem. Reconstructing a cmeneassuming you notice itis a more involved affair. Keeping the la/lai/doi rule will result in the overwhelmingly dominant dialect of Lojban losing its audio-visual isomorphism. Pick the disaster you want to deal with: you have to have one or the other.
4. People won't be making up cmene off the cuff when they're all standardized.
That will be true--once marketers and manufacturers stop coming up with new products and new names. If we have to wait until then to speak Lojban, there won't be anything sentient left to speak it. But we can't provide The Official Lojban Name(TM) for every single named thing in the known universe in advance! It would take most of eternity just to finish fine-detail geography! And as soon as the next new company or product shows up, any list is incomplete again. I saw *laivdjurnal once on IRC; could we have provided an official name for that ten years ago?
5. This is like requiring cu before all selbri.
There are some formal similarities, but none that actually make any difference. Mainly that both cases involve making something optional into something required, which isn't much of a similarity. The main thing is, though, that the la/lai/doi rule has been demonstrated over years of experience to be unworkable, whereas cu is usually used pretty well by speakers after a little experience. Anyway, this is the wrong section to be talking about requiring cu; take that up in BPFK Section: Scope cmavo.
6. Why don't we compromise: any cmene that contains la/lai/doi has to have a pause before it, but otherwise it doesn't have to.
There isn't any middle ground, I'm afraid. Is the phoneme-stream .mivIskalamArk. a big long cmene (of the first type, preceded by a pause), or is it mi viska la mark., ending on a cmene of the other type? It's one or the other. Besides, if people could remember to distinguish the two kinds of cmene, we wouldn't have this problem in the first place.
7. Can we allow them followed by h?
This was referred to by (I think) Pierre Abbat as "the ala'um rule", with the word giving an example. The idea being that if the naughty syllable was followed by something that couldn't possibly be the start of another word (namely, an apostrophe, which I will write as h when I need to speak of it in isolation, so there), it should be allowed. This would yield a more inclusive set of legal cmene (i.e. a proper superset of the currently allowed cmene), but wouldn't really be any easier to remember. It gives us more substitutes for la/lai/doi, but we already have substitutes, and if we could remember to use them we wouldn't be in this mess. The rule instead becomes more complex, and thus even harder to remember and use. It won't solve the problem, and it might just muddle things even more.
8. How about only requiring a pause before cmene that contain la/lai/doi?
That won't work. See above.
8.1. How come you suggested that one twice?
Because it gets suggested four or five times every time this matter is discussed, despite being shown to be incomplete each time. I figured (a) I might as well model the way the discussions really go, and (b) maybe by repeating it people will pay attention and not try to suggest it again.
9. But the usage is so entrenched, to say laLOJban. and all...
Usage is also very entrenched to permit cmene with la/lai/doi. And we can decide to leave both usages alone--if we're willing to give up on audio-visual isomorphism, and if we're willing to say that Lojban is ambiguous in this matter (that you can't parse .mivIskalamArk. unambiguously). Otherwise something has to give. Choose your disaster. The pauseless-la usage may or may not be retrainable. The la/lai/doi-in-cmene usage has been amply demonstrated not to be retrainable. If we want to keep audio-visual isomorphism and unambiguity, one of those usages has be abandoned. I say we keep the one we know we can't abandon.
Besides, usage needs some help anyway. People tend to be sloppy with pronunciation, but we shouldn't let that determine the right way to do things. No, that's not the same as letting sloppy cmene-construction determine the right way to do things, see above! Sloppy pronunciation is ephemeral. Bad cmene get into literature and into the common lexicon. People need practice with the mandatory pauses already required by Lojban. Unless we have a way to remove them completely, we have to accept some level of hardship. Pause before names except after the magic words (by current rules). This affects zo and mi'e particularly, the rest of COI to a lesser extent (they can be "shielded" by doi, and actually mi'e can be shielded by la, but people don't do that for some reason). Pause after lerfu-names (unless followed by another lerfu). Pause around ZOI-quote guard-words (that one should be obvious, but probably still needs to be attended to). Pause before vowels is usually pretty well observed, because initial vowels are considered to be preceded by a glottal stop, though in some cases things like .ua.ui are an exception to that. Even those glide-to-vowel diphthongs must be remembered to have that pause before them; I pronounce them with glottal stops before. Pauses after cmene I think people try for, and nobody truly forgets they need to be there, but I suspect they aren't always being put there. Again, my trick would be to attach glottal stops (mentally) to the affected magic words, so la is always pronounced with a pause after it, etc.
10. You know, your own speech doesn't pause before cmene either!
(For whatever suitable value of "you") Well, first of all, this proposal is still only a proposal, so it isn't actually part of the language yet. This is otherwise the same objection as number 3 above, which see.
11. A missing pause is worse (harder to hear) than an illegal cmene
No it isn't. See number 3 above.
12. Aww, c'mon, how many illegal cmene really make it out there?
Some, and even then not as many as start to make it out there and are shot down (by people other than the original perpetrator). That is to say, in general people simply don't notice bad cmene, though sometimes they are caught before they get too far into common parlance. But by no means all the time, and the examples above demonstrate that. The long and short of it is that la/lai/doi in cmene is in line with usage, and the baseline is not.
13. People also mess up with illegal medial clusters in cmene
Strangely enough, not all that often, I've found. Partly this is because most of the illegal medial clusters were made illegal because they are hard to pronounce, so people tend to avoid them anyway. Also, even though illegal medials can't make the same claim to being ephemeral that missing pauses can, I still maintain that they are less destructive mistakes, in that they do not threaten Lojban's audio-visual isomorphism and unambiguity.
14. I know! Why don't we only have to pause before---
No. See number 6. Again.