(A page from the Loglan web site.)

(From Lognet 96/1. Used with the permission of The Loglan Institute, Inc.)

An Introduction by Emerson Mitchell

The following article was written by James Jennings several months ago, summarizing the then state of discussion of a proposed subjunctive system for Loglan. A little later a somewhat different system, motivated by a different understanding of the natural language usage of subjunctives, was proposed by James Cooke Brown. Although the two proposals are compatible logically, they are sufficiently different linguistically that it seems unlikely Loglan will adopt both. James withdrew the paper at that time for the expressed reason that he did not want to publish something that was already obsolete, and that he did not have the time to revise it to match JCB's proposal.

Since then we have persuaded James that it might be a good idea to let the rest of you in on the discussion, even though it is only a preliminary proposal which may never be adopted. So I volunteered to do any required revisions. On rereading the article, I found only trivial revisions to make.

So here it is. JCB's proposed subjunctive system, which may make this one obsolete, merits its own article. Which I hope you will be reading in the next issue of Lognet.

--Hue E'mrsn

Mi Dui Foi Lepo Mi Fio Dui
(I Would If I Could)

by James Jennings

English-speaking writers of Loglan have long felt a lack when it came to the words would and could. What does the subjunctive look like in Loglan? How do you render the meaning of:

If I were king, I would be happy.

Most of the sense of the English is captured by the more logical sounding

If I am king, then I am happy.
Kanoi mi bragai ki mi hapci.

but it's a trap. The Loglan if (anoi = or-not) is a logical if, and that puts a fatal restriction on the meaning. If it so happens that I am not king, then the above claim is true, even if kingship would in fact make me miserable. The English if is a lot more flexible (and ambiguous).

There is another sense in which this attempt falls down. The English If I were king sort of implies that I am not king. It doesn't say that outright, but the average listener would usually assume that I'm not king, because if I were king I'd say it differently. We call this a "counterfactual statement", and the problem of rendering it into Loglan has been called the "counterfactual problem".

So what is the answer? Emerson Mitchel, a logician and a recent addition to the Lodgru [Logicians-Group, one of The Institute's several working groups.--JCB], has suggested borrowing the "possible world" semantics created by Saul Kripke that is used in a field called "modal logic". (Modal logic has nothing to do with the modal operators listed on page 294 of Loglan 1.) The Lodgru has argued about it in some depth, but most of the controversy, I believe, comes from people, including myself, not being familiar with what modal logic is like. I have found the book, Everything that Linguists Have Always Wanted to Know About Logic (but were ashamed to ask), by James D. McCawley to be very helpful. (In the 2nd edition, see Chapters 11 and 15.) It's reasonably readable, especially if you've ever read a college level math text before (and liked it, soi crano).

Here is the key idea. Imagine a map of "possible worlds". (Note: The phrase possible worlds is math jargon, and doesn't mean that the "worlds" are necessarily possible. The concept expressed here seems to me to be too general for any simple English phrase to cover well, so I'll be using the mathematical term throughout.) In the middle of the map is a dot labeled You Are Here. This is the factual world, the one we currently inhabit, where the speech is taking place. Scattered about the map are other dots, each of which are labeled by statements that are true in those worlds. For example, some of those dots are labeled I am king. Those are the (counterfactual) worlds in which I am king. A different collection of dots is labeled I am happy. If it so happens that every dot which is labeled I am king is also labeled I am happy, then it is true that if I were king, then I would be happy.

Emerson's proposal is to adopt the little word foi to mark "possible worlds". We would use it just like na or vi. For example:

If I were king, I would be happy.

would become

Foi lepo mi bragai guo, mi hapci.
In-worlds-where the-event-of I am king prevails, I am happy.

(Note that the mark of the subjunctive in English has two disjoint parts, were and would be, while in this proposed form, the Loglan subjunctive will have only the foi-phrase. At first glance, it may look like the would is missing from the Loglan.)

Grammatically, this is similar to

Na la Tordei, mi hapci.
During Tuesday, I am happy.
Vi lemi tcaro, mi hapci.
In my car, I am happy.

That is, foi, na, and vi all specify a point in some conceptual space where the action is occurring.

A nice example:

It would have taken an hour, were it feasible at all.
Su horto ga nerbi da foi lepo da nu durzo.
At-least-one hour is necessary to do X in-worlds-containing (labeled by) the-event-of X is-doable.

That handles would very nicely. What about could? Let's go back to the map of "possible worlds". Looking more closely, we find that the dots are connected by fine lines. These are "roads", many of them one-way, that connect the worlds. That is, if there is a line leading from one world to another, then there is some way in which the first world could evolve into the second. For example, if at least one world where I am a peasant, is connected to at least one world where I am happy, then it is true that

If I were a peasant, then I could be happy.

Unlike the foi case, I am not insisting that all peasant worlds are happy worlds, only that you can get to a happy world from some peasant world. We say that the happy world is "accessible" from the peasant world.

Emerson's proposal is to adopt the little word fio to mark accessible worlds. (This is mnemonically derived from fikco, is a work of fiction by, in honor of the often counterfactual nature of these claims.) Thus:

If I were a peasant, I could be happy.
Fio lepo mi lanpeu guo, mi hapci.
In-a-world-accessible-from-one-where the-event-of my being a peasant prevails, I am happy.

This is conceptually similar to

Fa la Tordei, mi hapci.
After Tuesday, I am happy.
Vu lemi tcaro, mi hapci.
Far away from my car, I am happy.

That is, fio, fa, and vu, each specify a relation between two points in some conceptual space. The first point is described by the argument of the PA-phrase. (My kingship, Tuesday, and my car, in the above examples.) The second point is determined by the scope of that phrase, often the rest of the sentence. (I am happy in all three examples.)

Foi and fio (and their required predicate eliminators) are the only new words in the proposal, but there are lots of implied usages. Let me run through some of those that have come up.

If foi or fio don't have an argument, the default, implied world is the factual world, the one in which speech is taking place.

Mi foi hapci.
I am (in the factual world) happy. (I'm really happy.)
Mi fio hapci.
I could be (starting from the factual world) happy.

These are similar to a tense marker.

Mi na hapci.
I am (now) happy.
Mi fa hapci.
I am (sometime after now) happy. (I will be happy.)

Here's another nice example.

Mi fio haispe lepo vizgoi la París.
In some world derived from this one, I am enjoying a visit to Paris. (Hint! Hint!)
I could enjoy a visit to Paris.

Note that foi haispe implies enjoyment (would enjoy), but fio haispe only says that enjoyment is possible (could enjoy). However, when one says I would enjoy it in idiomatic English, one usually means that one would enjoy it if one had it: one enjoys X in worlds where X prevails. (Mi haispe da foi da.)

Used by themselves, both na and foi can only add emphasis to a simple statement. In a longer story, however, some series of tense markers might have taken the narrative present far away from the actual present, and only na without an argument can bring you back. In the same way, foi can bring you back from the narrative world to the factual world. For example:

Mi hapci fio lepo mi bragai, ibuo mi foi lanpeu.
I could be happy, if I were king, but I am actually/really a peasant.

The "possible world" set up by fio lepo mi bragai might sweep over into mi lanpeu if you didn't have the foi. (The exact way that PA words like pa and fa will do this is still an open question, but it is assumed that, whatever they do, foi and fio will do the same things.)

Going back to the road map, if only some of the dots are accessible from a particular place, then there must be some dots that aren't accessible. That leads to claims like:

I couldn't be happy if I were a politician.
Mi hapci noifio lepo mi nu poltia.
My being happy is not accessible from a world where I am a politician.

where noifio is a compound of no and fio. Fionoi can be useful as well.

I could be happy if I weren't a politician.
Mi hapci fionoi lepo mi nu poltia.
My being happy is accessible but not from a world where I am a politician.

You can do similar things with foi.

Mi hapci noifoi lepo mi nu poltia.
I am happy, not in a world where I'm a politician.
Mi hapci foinoi lepo mi nu poltia.
I am happy, in a world where the-event-of I'm a politician doesn't prevail.

(It is interesting to note that noifoi is similar to fionoi, and noifio is similar to foinoi.)

Numbers can be compounded with PA words. Thus nena = once, topa = twice before, etc. We propose that the same privilege be extended to fio; this would give gradations of could that are quite difficult to render into English.

One can make similar compounds with foi; but note that, not only does foi not refer to accessible worlds, a bare fio has the same meaning as sufio, while a bare foi has the same meaning as rafoi.

Some predicates are world-building. That is, they refer to a "possible world" even though there is no foi or fio involved. In English these are verbs like dream, wish, think, intend, consider, order, and so on. Here is an example using expect.

I expect that it will rain this morning.
Mi fuikri lepo ba crina na levi monza,
I expect the-event-of something being rained on during this morning.

This is counterfactual, since it implies that it is not yet raining, but suggests that a world where it is raining is accessible.

The logicians out there are probably champing at the bit, asking, but what, logically, does he really mean by that? It doesn't help that, according to McCawley's book, the term modal logic is often used rather vaguely. To finish up, I'd like to show them what the equivalent modal logic notation is. The rest of you can skip this bit.

(Note for text-only web brower users: The following has some mathematical symbols which were supposed to be rendered in Lognet using the font Zapf Dingbats. Alas, something went wrong between the proofreading and the printing. Here I have supplied them as graphics for those with graphical browsers. For the rest of you, please imagine that the text "[N]", meaning "necessary", is a small square, and the text "<P>", meaning "probable", is a small diamond.)

McCawley's book has this definition (page 376 of the 2nd edition):

A modal system M consists of

  1. a language L,
  2. a set W of worlds (= assignments of truth values to the propositions of L), and
  3. a binary relation R between the worlds of W.

[N]A (A is necessary) is true in a world w of a modal system M if and only if for all w' such that Rww', A is true in w'.

<P>A (A is possible) is true in a world w of a modal system M if and only if there is a world w' of M such that Rww', and A is true in w'.

The language (L) is Loglan. The worlds (W) are the dots (w) on the map. The binary relations (R) are the roads (the accessibility relations) on the map. The statement [N]A (A is necessary) motivates the word foi, as in

Mi hapci foi lepo mi bragai.
My being happy is necessary where I am king.

The statement <P>A (A is possible) motivates the word fio, as in

Mi hapci fio lepo mi bragai.
My being happy is possible where I am king.

There is no exact correspondence between the logical symbols [N] and <P>, and the Loglan words foi and fio, since the logical symbols operate on statements and the Loglan words make modifiers out of arguments. (We do seem to get pairs of similar equivalences, like '~<P>~ if and only if [N]' and 'fio is equivalent to noifoinoi', in the two notations.)

When I said that the lines of the map mean that there is some way in which the first world could evolve into the second, I was sweeping a lot under the rug. Modal logic can address many different accessibility relations, of which this is only one, and a vague one at that. It is assumed, however, that in everyday conversation the appropriate sense of accessible will be obvious from context, or at least can be made clear with an extra modifying phrase or two. In those rare cases where a logician wants to precisely specify a relation, he or she can use a predicate. Which predicate? Well, all PA words are supposed to have a corresponding eliminating predicate. Pa, for example has pasko. The pa in the sentence

Mi godzi pa la Tordei.
I go before Tuesday.

can be eliminated by expanding the sentence like this

Lepo mi godzi guo, pasko la Tordei.
The-event-of my going is earlier than Tuesday.

It is proposed that fio and foi be assigned the eliminators blikui and feorkui:

for fio: blikui <blicu kunci = possibly-related-to>
... is (occurs in a world/context/setting/possibility) related to ... by accessibility relation ...
for foi: feorkui <fekto kunci = factually-related-to>
... is actually related to (occurs in the same world as) ... according to point-of-view/observer/metaphysics/reality ...

For example:

Ba takna fio be.

is equivalent to

Lepo ba takna guo, blikui be.
So if you wanted to specify the kind of accessibility relation, you would first eliminate the fio, and then fill in the third slot of blikui.
Lepo ba takna guo, blikui be bo.
Some x's talking occurs in some world y according to some accessibility relation z.

By the way, in my opinion every other proposal for a Loglan subjunctive so far offered assumes a particular accessibility relation which it tries metaphorically to stretch over all cases. The fact that foi/fio doesn't may give it an elegant conceptual advantage.

Given all of this powerful new machinery, how would a logli say I would if I could if this version of the Loglan subjunctive were to be adopted? This is tricky because the English is quite compact. Let's expand it a bit.

Suppose you had been asked for help. (Eo, helba mi.) Let's build up the answer a bit at a time.

Mi helba tu foi.
I help you in this world. (I would help.)
Mi helba tu fio.
I help you in an accessible world. (I could help.)
Mi helba tu foi lepo mi kanmo lepo helba tu.
I help you in the world where I am able to help you. (I would help if I were able to help.)
Mi helba tu foi lepo mi fio helba tu.
I would help you if I could help you.

That's the right structure. The English compacts it down by omitting the understood verb. Loglan can't do that and stay grammatical, but it can use a predicate variable. The most compact Loglan form is therefore

Mi dui foi lepo mi fio dui.
I do-it in-a-world-where the-event-of my, in-an-accessible-world, doing-it prevails.
I would do it if I could do it.

Copyright © 1996 by The Loglan Institute, Inc. All rights reserved.

Send comments and corrections to:

djeimz AT megaseattle DOT com