The {lo'ai ... sa'ai ... le'ai} replacement construct is used to erase some recently uttered string of words and replace them with something else. It is comparable to the "s/foo/bar/" expression commonly used on IRC.

By saying {lo'ai srera sa'ai drani le'ai}, you are asking the listener to erase the most recent occurrence of {srera} in your text stream, replace it with {drani}, and reparse. For example, {lo'ai cakla sa'ai ckakla le'ai} can be translated as, "I didn't mean {cakla}, but {ckakla}."

It's possible to switch the order of the text to be replaced and the replacement by saying {sa'ai drani lo'ai srera le'ai}.

It's also possible to omit either the text to be replaced or the replacement, leaving it up to context. For example, {sa'ai ckakla le'ai} means, "I meant {ckakla}," and leaves it up to context what string of words should be replaced by {ckakla} (in this case, probably the word {cakla}). Similarly, {lo'ai cakla le'ai} means, "I didn't mean {cakla} (but something else)."

It is interesting to note that {lo'ai ... le'ai} can be thought of as an "error quote," which is the traditional (arguably poor) gloss for {lo'u} and {le'u}. In other words, {lo'ai} and {le'ai} are the real "error quote" cmavo.

Naturally, one can erase a specific string of words recently uttered by saying {lo'ai broda brode .ua .ue .ui sa'ai le'ai} — essentially replacing {broda brode .ua .ue .ui} with the empty string.

The converse case is somewhat special: {lo'ai sa'ai se setca le'ai}, rather than replacing the most recent occurrence of the empty string (which would be useless), actually tells the listener to insert the string {se setca} at the appropriate place in the recent text stream. Where the string is to be inserted is left up to context.

By saying just {le'ai}, omitting both the text to be replaced and the replacement, one can acknowledge the fact that one has made an error while leaving it completely up to context what the actual error was. Thus, {.i .ai mi cakla ... le'ai} means roughly, "I'm going to be chocolate. ...err, you know what I mean."

To summarize, the different forms of {le'ai} expressions are as follows.

Completely machine-understandable forms:

  • {lo'ai cakla sa'ai ckakla le'ai} replaces the most recent occurrence of {cakla} with {ckakla}
    • {lo'ai cakla sa'ai kukte cakla le'ai} inserts the word {kukte} before the most recent occurence of the word {cakla}
    • {lo'ai kukte sa'ai le'ai} erases the most recent occurrence of the word {kukte}.

Context-dependent forms:

  • {lo'ai sa'ai mutce le'ai} inserts the word {mutce} at the obvious place in the recently uttered text
  • {lo'ai cakla le'ai} replaces {cakla} with something that is obvious from context
  • {sa'ai ckakla le'ai} replaces something that is obvious from context by {ckakla}
  • {le'ai} declares that an error was made and asks the listener to make the obvious correction.

I have been experimenting with questions of the form {lo'ai ... sa'ai ... le'ai pei} (the {lo'ai} and {sa'ai} clauses being optional, as usual). Relevant answers to a {le'ai pei} question are 1) any {le'ai} construction or 2) any {le'ai} construction modified by {nai} (or, equivalently, simply {ja'ai} or {nai} alone, respectively). It should be quite clear what these expressions ought to mean; for example,

  • {le'ai pei} asks the listener whether they really meant to say what they just said
  • {lo'ai cakla le'ai pei} asks the listener whether they erred in saying {cakla}
  • {sa'ai ckakla le'ai pei} asks the listener whether they rather meant to say {ckakla}
  • {lo'ai cakla sa'ai ckakla le'ai pei} asks the listener whether they meant to say {ckakla} rather than {cakla}.

By answering a textual replacement question with {le'ai} (or {ja'ai}), you acknowledge that the suggested error was in fact made. If the question was simply {le'ai pei} (i.e., no particular error was suggested), then {le'ai} acknowledges that some error was made, but does not specify what the error was.

By answering with {le'ai nai} or {nai}, you deny that the suggested error was made (without necessarily asserting whether or not any other errors were made). If no error was specified and the question was simply {le'ai pei}, then answering with {le'ai nai} denies that any errors were made at all, asserting that what was said was entirely correct.

Presumably, it may be useful to answer {le'ai cu'i} when you don't have an opinion either way as to whether or not you want to make the replacement.

Naturally, it is possible to explicitly deny or affirm a specific error by answering with, e.g., {lo'ai cakla le'ai nai}, {sa'ai ckakla le'ai nai}, or {lo'ai calka sa'ai ckakla le'ai}.

This leads us to consider the expression {le'ai nai} used on its own, rather than as the answer to a question. This essentially becomes a way to metalinguistically assert that you really intended to say what you just said. You can even explicitly mark dubious language as actually intended by saying {lo'ai ... le'ai nai}. This saves the reader or listener from having to consider whether the doubtful expression was written or said in error or whether it was done on purpose, and the likeliness of either possibility.

Similarly, {le'ai cu'i} can be used when you are yourself doubtful of the correctness of what you are writing or saying. This is comparable to the expression "(sp?)" used by writers who are unsure of their spelling, but {le'ai cu'i} is also generally useful for grammatical uncertainties other than spelling. (Less directly, it could express semantic uncertainty, but this is probably too vague to be understandable and useful, since listeners will first attempt to interpret {le'ai cu'i} as grammatical uncertainty, and that will usually make sense some way or another.)

— Daniel Brockman

One way to express this using vanilla Lojban is {sei zo ckakla zo cakla basti}, {sei zo ckakla basti}, {sei zo cakla se basti} and {sei basti}. I and Eimi have been using this on IRC a few times and will interpret it identically to {sa'ai ckakla lo'ai cakla le'ai}, {sa'ai ckakla le'ai}, {lo'ai cakla le'ai} and {le'ai}, respectively.

— Daniel Brockman

Created by daniel. Last Modification: Monday 14 of July, 2008 17:41:16 GMT by daniel.