(Return to the Table of Contents)

Chapter 5
GRAMMAR 3: UTTERANCES

5.1

Utterances and Speeches 279

5.2

Utterance Types and Ingredients 283

5.3

Three Kinds of Modification 286

5.4

Sentence Modifiers 288

5.5

The Four Varieties of Relative Operators 293

5.6

Modal Operators lia and Kin 293

5.7

Causal Operators kou moi rau soa and Kin 296

5.8

The Seven Varieties of Free Modifiers 298

5.9

Salutations & Other Expressions of Direct Address 299

5.10

Expressions of Attitude 302

5.11

Questions with ie he ho hu ha 309

5.12

Relative Interrogatives with -hu 314

5.13

Discursive Modifiers 317

5.14

Parenthetic Remarks with kie ...kiu 319

5.15

Utterance Sequencers with -fi 321

5.16

Identity Sentences with bi bie 322

5.17

Identifying vs. Predicating Modifiers 327

5.18

Punctuation 334

5.19

Marking Main Predicates with ga 334

5.20

The Specific Terminators gue gui guo guu 342

5.21

Marking Right-Boundaries with gu 335

5.22

Abnormal Word Order with ga gi goi gio 342

5.23

Utterances and Their Modifiers: A Summary 346

5.24

Logically Connected Sentences with ica and Kin 347

5.25

Quantified Sentences 354

5.26

Negative Sentences 369

5.27

Causally Connected Sentences with ikou 378

5.28

The Connectives imoi irau isoa 386

5.29

Causal Questions and Answers 389

Notes 393

5.1 Utterances and Speeches

One of the ancient prejudices of grammarians is that grammar has to do with sentences, and that portions of speech that are not sentences are somehow not grammatical. While it may be true that some formal writing is composed entirely of sentences, speech is not. Greetings, questions, answers, commands, requests and replies bulk larger in the flow of actual speech than those neatly formed statements of fact or opinion that are the writer's stock-in-trade. Obviously, if Loglan is to be a spoken language, we must provide for the bits and pieces of ordinary speech as well as the more formal speech-forms with which ordinary grammar deals. Let us call these bits and pieces, whether they are sentences or not, the utterances of the language.

But the utterance is not the largest unit of speech. Some passages of speech--like reports, recipes or narrations--consist of strings of utterances each of which could have ended the message had the speaker chosen to stop there.

(1) I saw it there. It was under a tree. Joe was standing right next to it. I don't know what happened to it then.

In this report, there are three points at which the speaker could have stopped but didn't. In a certain sense, the report consists of a string of utterances each of which could have stood alone. Let us call such larger units of continued utterance speeches. A speech is not over until the speaker who is making it yields the floor, so to speak. On the other hand, what we are calling speeches might well be taken to correspond to what we mark off as paragraphs in writing. It is whatever occurs between the real starts and stops in either speech or writing.

Loglan is by design an isomorphic language. The forms of speech and writing must correspond to one another in sufficient detail so that the essential features of each can be generated from the other. So the first thing we require in the management of speeches is a sign by which to know where the constituent utterances in them end. We will sometimes need a sign to mark the end of the last (or only) utterance in a speech or paragraph, but usually we will not.

Every language has at least one end-of-utterance sign. English has several. The one it uses to mark the end of declarative sentences is a fall in pitch. The last syllable is also often lengthened or drawn-out, but is not necessarily louder. This pattern of tone and pace change is usually, but not always, followed by a pause before the speaker continues. If you speak speech (1) aloud you are very likely to hear all these signs in the neighborhoods of the words 'there.' 'tree.' 'it.' and 'then.' This is the most common of the English end-of-utterance signs. There are several others.

In Loglan the single end-of-utterance sign is the continuation connective I. Instead of being a feature of the utterance being ended, however, I is the first word of the following utterance. Like other vowel-initial words, I is always preceded by a pause. In Loglan we use the pause followed by an I-word to mark the ends of all utterances that are not final in their speeches. Since there is, of course, no utterance following the last one in a speech its end need not be marked.

(2)

Ia. I mi pa vizka da va. I da pa nilca ne tricu

Yes. I saw it there. (And) It was under a tree.

You may speak this little speech with any intonation whatever, or with none. That is to say, you may choose to accompany it with your English end-of-sentence sign, or with no tone or pace variation at all. All that is required for a listening loglanist--or a properly programmed computer--to understand that the utterances Ia and I mi pa vizka da va are over is that they are followed by pauses followed by the word I...or, as we shall see in a moment, by one of its many kin.

The lexemic kin of I are all little word compounds made with leading I followed pauselessly by one of a privileged set of words which adhere to it in this way. For example, Ibuo [ee-BOO-oh] is made from I + buo. By itself buo means 'however' or 'but'. So Ibuo is simply the utterance-initial and so capitalized form of the same word: 'However' or 'But'. I-words are of course also preceded by pauses and terminate the preceding utterances just as I does.

(3)

Ia. I mi pa vizka da va. Ibuo da pa nilca ne tricu

Yes. I saw it there. However, it was under a tree.

Notice that the pauses that precede I and Ibuo are represented in text by a period followed by a space. This is the pause-period, as we will call it to distinguish it from the pause-comma with which we are already familiar. Pause-periods tend to be a little longer in speech than pause-commas. However, pause length is a matter of personal style. No length difference need be maintained for the clarity of the language. The capital I-word that always follows the pause-period is what distinguishes it from the pause-comma.

I has more distant lexemic relatives--such as the logical sentence connectives like icanoi [ee-shah-noy] 'if'--that are not capitalized in text. The pauses that mark these "weaker" utterance breaks are pause-commas. In the typical idiolect the pauses that precede these lower-case i-words are somewhat shorter. They are always marked by commas instead of periods in text:

(4)

Mi pa vizka da va, icanoi da pa nilca ne tricu

I saw it there if it was under a tree.

This is counted as one utterance in English and two utterances--although closely connected ones--in Loglan. We shall see later that for certain logical purposes--for example, the reckoning of operator scope--the Loglan logic understander will treat this specimen as a single utterance. But the parser treats it as a pair.^R1

Thus the complete story of the end-of-utterance sign in Loglan is that it is always [. ee] in speech and either ', i-' or '. I-' in text. That is all. No tone or pacing change is required, although any may be used.

There is just one more thing to consider. As hinted above, the '. I' sequence in Loglan text does not translate all English periods, but only those which occur within speeches. For example, it translates the first two periods in the English portion of speech (3) but not the third:

(3') Yes. I saw it there. However, it was under a tree.

There is no Loglan word that signals the end of a speech, or the point where a speech might be continued but is not. The speaker simply stops. Certainly the mark for continuation could not be used, and that is what I really is. When an utterance is over in spoken Loglan, silence reigns. Or another voice begins. In written Loglan, that silence or the start of that "other voice" is often signalled by a carriage return followed by an indentation, or by a carriage return alone, or by a pair of carriage returns (which amount to a skipped line), or by some other textual convention. When such end-of-speech-or-paragraph cues threaten to be ambiguous, or to take too much space, a cross-hatch (#) may be used. Thus the Loglan of (3) could be shown textually as:

(3") Ia. I mi pa vizka da va. Ibuo da pa nilca ne tricu#

and the cross-hatch (the standard linguistic sign for silence) could then serve to concatenate speech into even larger units of language, say discourse, just as the period-plus-I-word concatenates utterances into speeches.

But notice that the cross-hatch does not stand for any spoken word or other sound in the speech-stream. It stands for silence, a voice-change, the end of a paragraph or message. Whether cross-hatches (or some other graphic sign) will gain currency as end-of-paragraph marks in written Loglan is difficult to predict. The important point is that Loglan speeches, whether composed of more than one utterance or not, are the natural units of both speech and writing, and therefore probably also of thought, and as such, their ends will be naturally known.

Logically, of course, the utterance concatenator I that forges utterances into speeches is just another version of the conjunction e ('and'). We might call it the strong conjunction that occurs between sentences as opposed to the weak conjunction that occurs within them. But apart from matters of scope the logical force of I is the same as that of e. For whoever uses I between the two utterances in a speech evidently means to utter both independently; and this, had da thought about it beforehand, is often what da could have said with e or ice or ke...ki.... Thus I is logically the sign of an afterthought, of an essentially unplanned continuation of speech. One may, of course, change one's mind about the truth of what one has already said. But this is not usually the sense of I.

A final point. We have noted that I cannot meaningfully terminate an utterance. For to signify to one's listener that one means to keep talking and then fail to do so is ungrammatical, in the broad sense that it sets up expectations that are not then fulfilled. But I may initiate an utterance. And it does so with a very clear sense indeed. Suppose John says:

(5)

Mi pa vizka da va

I saw it there.

and stops. And then Pete adds:

(6)

I da pa nilca le tricu

And it was under the tree.

Pete evidently means to supply a continuation, not of his own, but of John's original remark. And why not? But are (5) and (6) then one speech? Or two? It hardly matters...one speech with two voices or two speeches which might have been one. Arbitrarily we will say that they constitute two speeches and that the voice-change marks the break. But this admits the possibility that speeches may commence with I. And so they may.

We will in the main confine ourselves in the rest of this book to examining the structure of single-utterance speeches, or, what amounts to the same thing, to uncontinued utterances. This is because our understanding of the structure of human discourse, or the explication of the relationships between the utterances that comprise a speech--except for the logical ones between adjacent utterances--is not well-developed scientifically, and is in any case beyond the scope of at least this edition of this book. Accordingly we will use the word 'utterance' in what follows to designate either a single-utterance speech or an I-marked portion of a speech.¹

5.2 Utterance Types and Ingredients

The two principal ingredients of Loglan utterances are, of course, the predicates and arguments whose grammar was developed in Chapters 3 and 4, and the three principal types of utterances are sentences, answers and imperatives. As we have already seen, many utterances consist exclusively of these two ingredients. For example, the minimal sentence is one composed of an argument followed by a predicate:

(1)

La Djan, ditca

John is a teacher.

The minimal answer is an argument:

(2)

La Djan

John.

And the minimal imperative is a predicate:

(3)

Ditca

Be a teacher!

Both imperatives and sentences may of course be followed by one or more sutori (from Loglan su + to + ri) arguments:

(4)

Ditca lo numsensi [noom-SEN-see] lo friki

Teach mathematics (number-science) to the Africans.

(5)

La Djan, ditca lo numsensi

John is a teacher of mathematics

As suggested, single-argument utterances like (2) are normally the answers to questions. For example, Hu ditca numsensi lo friki = 'Who teaches mathematics to the Africans?' or Hu merji la Selis = 'Who is married to Sally?' The answer to either question might be La Djan = 'John.' Later in this chapter we will see how to make questions like these. But we already know how to make the answers.

Arguments spoken or written as utterances may also specify the topics of discussion or the titles of books. But notice that the latter are always answers to implied questions. Hu nu perti levi bukcu = 'What's this book about?' you ask yourself as you pick it up and look. (Literally, 'What is pertained to by this book?). Or in better loglandical word-order, Levi bukcu ga perti hu = 'This book is about what?' One answer is the one you might see printed on its cover in a loglandian bookstore: La mela Peloponesos po Dorja = 'The Peloponnesian War.' Thus titles are almost always constructed as argument forms.

Utterances like (3) and (4), however, which commence with, or are composed only of, predicate expressions, are always imperatives: Ditca = 'Be a teacher!'; Gudbi botci = 'Be a good boy!'; Gancu = 'Be a winner!'...in short, 'Win!' In other words, imperatives in Loglan are nothing but sentences without first arguments. No inflection or other marking is required.

The fourth type of utterance are those treated by the parser as strings of separate utterances but by our logic understander as strings of connected utterances. The utterances connected may be sentences or imperatives. The connectives involved may be either keks--which may be used to connect utterances as well as utterances parts--or the weak type of i-word in which i itself is not capitalized and which is preceded by a comma-pause. For example,

(6)

Ditca da, icanoi tu djano da

Teach it if you know it.

is logically a single utterance, a connected one in which the connectands could be separate utterances but aren't. In contrast,

(7)

Ditca da. Ikii [ee-KEE-ee] tu djano da

Teach it. Clearly you know it.

is a pair of utterances. They are concatenated--i.e, strung together as part of the same speech--by the I-word Ikii. So in (6) and (7) we encounter a third ingredient, the connectors, to add to the arguments and predicates of which minimal sentences, answers and imperatives are made.

So we now have four types of utterances and three types of ingredients. Any of these utterances may be embellished or elaborated in various ways, and the embellishments themselves may be spoken as separate utterances. Two further kinds of utterance ingredients are involved in this embellishing work. They are modifiers and punctuators. Examples of modifiers are vi, as in

(8)

Mu titci vi

We eat here.

and Ia, as in

(9)

Ia mu fa titci

Yes, we're going to eat.

and na la Ven, as in

(10)

Da pa kamla na la Ven

X came at nine.

All of these modifiers may be answers and so may serve as utterances:

(13)

Na la Ven

At nine.

The fifth ingredient is the punctuators. A very common one is the general comma-word gu. We have already seen it used in

(14)

Mu titci vi gu le supta

We eat, here, the soup.

Without its punctuator, (14) would mean

(15)

Mu titci vi le supta

We eat in/at the soup.

So the effect of gu is to close off certain structures so as to prevent them from absorbing new material. All these topics will of course be dealt with in some detail in later sections of this chapter.

In sum, there are four types of Loglan utterances: (1) sentences, (2) answers, (3) imperatives and (4) connected utterances. They are made out of five types of ingredients: (i) predicates, (ii) arguments, (iii) modifiers, (iv) connectors and (v) punctuators. We already know how to construct unmodified sentences, answers and imperatives. The rest of this chapter will be devoted to explaining the three things we do not know: (A) how modifiers are made and used; (B) how and when utterances need to be punctuated; and (C) how utterances may be connected to each other.

We commence with modification.

5.3 Three Kinds of Modification

There are three types of modifiers in Loglan, (a) free modifiers like Ia, which may be inserted at any point in a sentence, (b) sentence modifiers like Vi and Na la Ven, which may go nearly anywhere but which must be guarded grammatically from absorbing, or being absorbed by, other elements, and (c) argument modifiers, which may only follow arguments. To illustrate the difference between these three styles of modification, let us apply them to the same basic sentence:

(1)

Ia da godzi la Paris'

Certainly X goes to Paris.

(2)

Na lo dotra da godzi la Paris'

In the winter X goes to Paris.

(3)

Da godzi la Paris', ja sitci go nu cluva da

X goes to Paris, which is a city beloved by X.

The sense in which ia [yah] 'certainly' is grammatically "free" may be appreciated by noting that ia may go anywhere in its sentence--except, in fact, between la and Paris', for nothing may come between a name operator and its name without becoming part of that name--without causing parsing difficulties. The word 'certainly' is similarly free to move about in English. On the other hand, as we do this we notice that the meanings of the resulting sentences subtly change: 'X, certainly, goes to Paris (not Y)' 'X goes, certainly, to Paris (doesn't live there)' 'X goes to Paris, certainly (not to Rome).' In short, attaching a free modifier to a particular element of a sentence calls our attention to that element. In a sense it emphasizes it by suggesting a contrast between this sentence and another like it in which just this element has been changed. So non-initial free modifers highlight particular features of the host sentence. It is apparently only when free modifiers are initial, as in (1), that they modify the host sentence as a whole.

So much for free modification with little words like ia. The sentence modification that is going on in sentence (2) is quite different. Here, the whole sentence will be modified no matter where we put the modifying phrase na lo dotra. While these kinds of phrases may be placed nearly anywhere in their sentences, their contribution to the meanings of those sentences doesn't seem to change as they move. There seems to be no difference in the three claims that (i) in the winter X goes to Paris, (ii) X goes in the winter to Paris, and (iii) X goes to Paris in the winter. In all three cases, the entire predicated event (X's going to Paris from somewhere) is involved. It is that entire event that takes place in the winter. Whether you mention that fact late or early in the sentence in which you report it seems to make no difference in the claim. It is in this sense that na lo dotra is a sentence modifier. What it modifies is not the local neighborhood in which we find it but the whole sentence in which it is found. It is apparently only convenience, then, or the order in which they spring to mind, that determines the order in which sentence modifiers are uttered. Their meaning is so global, in fact, that sentence modifiers are often called simply modifiers. They seem to have about the same relationship to a sentence as an argument does.

The argument modifying phrase in sentence (3) represents still a third style of modification. Here the Loglan linking word ja [zhah] and the English relative pronoun 'which' firmly attach the predicate expressions sitci go nu cluva je da and 'is a city beloved by X' to the preceding arguments, namely to la Paris' and 'Paris'. The effect is clearly that of a local modifier. Moreover, unlike a sentence modifier, the ja-marked modifier is bound to remain in its place--or at least be moved about with its argument if that is moved--if it is to have this meaning at all. It is Paris and nothing else of which the incidental fact of its being a city beloved by X is said to be true.

In fact the only other place in this sentence where the ja-clause could conceivably occur and still be grammatical is right after Da. In this new position the same modifying clause would be as firmly attached to Da as it is now to la Paris'; and it would thus convey the cheerfully nonsensical message that X, which is incidentally a city beloved by itself, goes to Paris.

All three styles of modification may, of course, occur in the same sentence:

(4)

Ia na lo dotra da godzi la Paris', ja le sitci go nu cluva je da

Certainly in the winter X goes to Paris, which is a city beloved by X.

In the next half dozen sections we will take up, first, the construction of sentence modifiers, second, the many varieties of free modifiers, and finally the logically more intricate matter of argument modification. That done we will move on to the final topics of Loglan grammar, the punctuation and connection of utterances.

5.4 Sentence Modifiers

There are two kinds of sentence modifiers in Loglan. One of them is, in its simplest form, a prepositional phrase like na ti = 'at this time'; and the other has an adverbial meaning, like na = 'now', and is clearly an abbreviation or ellipsis of the first. The fully expressed sentence modifier is therefore either a phrase, like na lo dotra = 'in the winter', or a clause, like fa lepo la Djan, pa kamla = 'after the-event-of John's coming', or, less literally, 'after John came'. In either case the modifier is formed by putting something called a relative operator before an argument. In these last two cases the relative operators are na and fa and the arguments are lo dotra and lepo la Djan, pa kamla. Such arguments may be of any form whatever...including, as we've just seen, event-descriptions. When they are event-descriptions, then what we are calling a sentence modifier in Loglan will seem in English grammar to be a subordinate clause.

From adverbial na, to preposition-like na in na ti, to conjunction-like na in na lepo la Djan, pa kamla ('when John came'), the Loglan relative operator seems to be the pillar of this construction. Earlier in this book we encountered two varieties of relative operators, although we didn't call them that then. These were the tense and location operators of Sections 3.6 and 3.7, of which pa na fa and vi va vu were the elementary instances. But there are many other kinds of relative operators. We call them all "relative" because they relate the main event predicated in a sentence to something else: to the time, place, or other circumstances of speech when the operator is being used as an "inflector" of the predicate, to other designated things or events in the case of modifying phrases or clauses.

To appreciate the semantic range of these sentence-modifying constructions let's consider the following quartet of them: vi lo resra = 'in restaurants', kou tao [koh-oo-tow] = 'because of that (alluded-to event)', lia lo cinkau [sheen-KAU-oo] = 'like a puppy', and nia lepo da junti = 'throughout X's youth'. Notice that no grammatical distinction is drawn between relative phrases and relative clauses in Loglan. Both are sentence modifiers formed by joining a relative operator to an argument...no matter how elaborate that argument is. This will lead, as we'll see, to a considerable simplification of the grammar of modification.

Let's now apply this quartet of modifiers to the following unmodified sentence: Da pa clucea [shloo-SHEIGH-ah] = 'X fell in love (lover-became).' First, in solo passages:

(1)

Da pa clucea vi lo resra

X fell in love in restaurants.

(2)

Da kou tao pa clucea

X because of that fell in love.

(3)

Lia lo cinkau da pa clucea

Like a puppy X fell in love.

(4)

Da pa clucea nia lepo da junti

X fell in love throughout X's youth.

And then in concert:

(5)

Da clucea kou tao vi lo resra, lia lo cinkau, nia lepo da junti

X fell in love because of that, in restaurants, like a puppy, throughout da's youth.

I've put commas between the modifying phrases in the English translation of (5) to indicate that none is to be interpreted as a local modifier. Such punctuating moves are not necessary in Loglan--although I've used two optional phrasing pauses in the long production of (5) to make it easier to read aloud--because the fact that each of these modifiers modifies the sentence as a whole is unequivocally conveyed by the simple circumstance of their being unmarked. We will see in a later section how, if we want to say that it was a "restaurant like a puppy", we can do so...odd as that claim might be. But we must then mark the modifier lia lo cinkau with one of the ja-links of Section 5.17 to turn it into a local modifier.² When we say that these modifiers modify "the sentence as a whole" we mean, of course, the sentence as modified by all its other modifiers.

If we were engaged in semantical analysis, we might well argue that speaking the modifiers of a given sentence in a different order would deposit the features of the speaker's image in the listener's mind in a different narrative sequence...much as a story-teller might describe the same scene differently on different occasions by mentioning its elements in a different order. Yet the logical claim of sentence (5) is independent of the order of its modifiers. Moreover, that claim, once it is finally made, is identical to the claim made by the conjunction of sentences (1) through (4). For while each of these shorter sentences tells only part of this complex story, they make serially the same set of partial claims as are made collectively in (5).

To see that all the partial claims of sentences (1)-(4) are indeed order-independent, and that they probably do add up to (5), let us arbitrarily rearrange the elements of claims (1)-(4), in particular let us speak them in the inverse order in which they appear in (5):

(6)

Nia lepo da junti gu, lia lo cinkau, vi lo resra, kou tao, da pa clucea

Throughout X's youth, like a puppy, in restaurants, because of that, X fell in love.

Note that, now that the nia-clause is no longer final, the punctuator gu is required to prevent the sentence inside its lepo-clause from acquiring lia lo cinkau and vi lo resra as its sentence modifiers. The three comma-pauses, however, are optional...designed to break up this long sentence into five breathgroups, so as to make it easier for the newcomer to read aloud, and perhaps also for the listener to understand.

Sentence (6) is perhaps a bit more difficult to understand than (5) because one has to wait so long to learn about the main event. But apart from that, it is clear that (6) is merely one of the many possible retellings of the curious story told by (5).

The most compact form of a sentence modifier is simply the relative operator itself. For example, na, vi, kou, lia and nia, all of which have been used prepositionally above, are also capable of functioning as adverbs...that is, used without their customary arguments. Some English words are also capable of playing this dual role. For example, one may say 'They are flying above the house' but also 'They are flying above.' In English the word 'above' may thus be used as both adverb and preposition. While this is only rarely true in English, it is true of all the Loglan relative operators. These are neither adverbs, prepositions, nor conjunctions but a class of multi-function words that may play without confusion any of these syntactic roles when called upon to do so.

When used alone in this elliptical fashion na, for example, means 'at this unmentioned time (usually assumed to be the time of speech)', or simply 'now'. By a similar ellipsis, vi alone may be taken to mean 'in this unmentioned place (but usually assumed to be the locality of speech)', or simply 'here'. Analogously, kou means 'because of this unmentioned cause (inferably the event just alluded to in speech, or the event of speech itself)', or simply 'consequently'. Similarly interpreted, lia means 'like this thing or event (that has just been alluded to in speech)', or simply 'similarly'. Finally, when nia is used as an adverb--and nia is the sign of the continuous present tense--it means 'throughout this event (usually one alluded to in speech)', or simply 'meanwhile'. And when we consider that some of these relative operators--in principle, as we are about to discover, all of them--may also be used to inflect predicates as well, as, for example, in Da na sucmi ('X is now swimming') and Da vi farfu ('X is here a father'), it is evident that Loglan gets quite a lot of syntactic mileage out of its relative operators.

In fact, it does. The lexeme to which all these very different kinds of words belong--called the PA Lexeme by the machine grammar--is one of the "portmanteaux" lexemes of Loglan. It contains a huge congeries of semantically unlike elements, yet all have the same grammar. It contains, for example, tense operators, location operators, modal operators, and causal operators, and the meanings of these have little in common except their "relativity". Yet the remarkable fact is that all of them can be handled by the same small set of grammar rules. This is emphatically not true of similar words in natural grammars. So a large simplification of Loglan grammar is achieved by pooling them together. In the end the learner will discover that the same rules handle them all. As a consequence, to learn where pa and kin may go in Loglan, is to learn not only where vi and kin go, but also where kou and lia and all their diverse kin go as well. And remarkably enough this will include the so-called "inflecting position" before the predicate.

Let us now take the mystery out of the inflecting role of the relative operator. Consider (i) Da kou sucmi. In contemporary Loglan usage, the claim of (i) does not differ substantially from that of (ii) Da sucmi kou, any more than Da na sucmi claims anything very different from Da sucmi na. In (ii) the causal relative kou is playing the role of an English adverb...a role which we seem to understand readily enough: 'X swims, consequently.' Hence (i) must mean something very like 'X is consequently a swimmer' in which the effective cause of the swimming--or of X's being a swimmer--is something which both speaker and listener may be assumed to be aware of...something current, perhaps, or recently alluded to, in this communication. Apparently the only, or at least the chief, difference between (i) and (ii) is syntactic and positional. In (i), kou is being used early in the sentence as an inflector, perhaps to mark the predicate off from some preceding description, and is therefore inferably forethoughtful. In (ii), kou comes along after the main statement of the sentence has been made, and so is being used afterthoughtfully, as it were, which is the very spirit of the adverbial construction. These are differences in meaning, alright, because they signify different things about the speaker and the speech situation. But it is a difference that seems to make no difference in the claim.

In short, the lesson to be learned in this section is that the Loglan relative operators--of which there are apparently four types: the temporal, the spatial, the modal, and the causal--are remarkably free, positionally. They are not quite "free modifiers" in the sense discussed earlier, for these, as we shall see, are even freer positionally than the sentence modifiers are. But the relative operators, as well as the phrases and clauses made with them, may be dropped into a sentence at almost any point at which an argument may be used. Sentence modifiers are, in truth, quasi-arguments; for they specify those temporal, spatial, modal and causal features of events which may be mentioned in connection with nearly any predicate. That is, whatever happens in nature takes place in time, but also in space, and in some sort of causal nexus with other events as well. And if that happening also happens in the human world--as the events reported in human sentences nearly always do--it will be found to have been accomplished with some instrument, for some purpose, in some mode, and according to some rule.

Thus, each of the many relative operators of Loglan may be used in all three ways: (i) as a preposition to form sentence modifying phrases or clauses; (ii) as an adverb functioning as a free-floating sentence modifier; and (iii) as an inflector of the predicate, in which position it performs the occasionally important syntactic function of marking the main predicate of the sentence to be such. In English we are used to seeing only tense words--the so-called "auxiliary verbs"--in this essentially syntactical, predicate-marking role. In Loglan, spatial words like vi are also capable of playing it, as are, we now learn with some surprise, causal and modal words like kou and lia. As a result of our constant effort toward grammatical simplification, another large increase in the semantical domain of the language has apparently been obtained.

5.5 The Four Varieties of Relative Operators

Two varieties of relative operators are already familiar to us, namely the tense operators consisting of na and kin, which relate events in time, and the location operators consisting of vi and kin, which relate them in space. The two remaining categories of relative operators are the causal operators, which relate the event or relationship described by the main predicate to other events according to how the speaker thinks they are causally related, and the modal operators, which place the main event in what might be called the human matrix. They allow the speaker to explain how the predicated event is similar to other events, who it is for, how it was brought about, or with what tools, agents, methods or according to what rules. Less systematic than the causal relatives, we'll discuss the modals first.

5.6 Modal Operators lia and Kin

Like the case tags of Section 4.31, the modal³ operators are all CVV-form words. Like them, too, each one is derived from a primitive predicate which serves as a clue to its meaning. At present there are but twelve of these modal words; they are shown in Table 5.1. We expect that, as the number of native languages of loglanists increases, other modal notions will be added from time to time.

Here are some examples of modally modified sentences. Most of them are modified with modal phrases:

(1)

Da pa dzoru lia la Djan

X walked like John.

(2)

Da pa kutla de tie leda najda

X cut Y with X's knife.

(3)

Da pa madzo de coi le bukcu

X made Y according to the book.

(4)

Lui le fumna da pa durzo de

For the woman, X did Y (i.e., in order to please the woman).

(5)

Da pa takna dii la Djan

X talked for (on behalf of) John.

These are modified with modal adverbs:

(6)

Da pa durzo ta dii

X did that as-a-representative (on behalf of some unmentioned principal).

(7)

Da pa takna rui

X talked accordingly (i.e., according to some unmentioned rule).

(8)

Da pa turka ciu

X worked as much (as some unmentioned person).

And this one is modified with a modal clause:

(9)

Da pa takna de lia lepo leda farfu pa takna leda matma

X talked to Y like X's father talked to X's mother.

Again, this is a phrasing pause.

It will repay us to take a moment to think about how the modals differ from the case tags of the BEU Lexeme described in Section 4.31, for these are the objects with which they are most likely to be confused. In a certain sense, modal phrases may be used to extend the place-structure of nearly any predicate. So they may be thought of as "itinerant cases", cases which are permitted to visit nearly any predicate on their far-flung rounds. But just because such itinerant features may turn up nearly anywhere, they may never be distinctive features of any predicate. It is for this reason that modal features are never, or rarely, part of the dictionary definition of a predicate...any more than time and place features are. It is for this reason, too, that modal prepositions may never be omitted from their arguments, as a case tag may be thought to have been when the argument it might tag is unmarked and in standard order. This is because a speaker may not rely on even the most knowing listener correctly to infer the modality of an argument from its context, as da may rely on that same knowing listener to infer the non-itinerant features of a predicate, namely the roles played by the occupants of its standard arguments. Thus modality must always be explicitly marked by the use of modal prepositions as the cases of arguments need never be.

Loglanists are exploring the uses of modals as inflectors and adverbs. We expect adventurous speakers to make many interesting discoveries in these uncharted waters. With a little thought, almost any of these strange new usages may be sensibly interpreted. For example, what does Da durzo de hea mean? In other words, what is the adverbial sense of hea? Well; if you think about it, hea used non-inflectionally and without an argument must at the very least mean that X did Y with the help of someone else, that is, "helpedly". So Da durzo de hea must mean 'X does Y with help'. What about hea as an inflector? Reasoning analogically from, let us say, the sense of pa itself in these three positions--as a preposition pa means 'before (this designated time)', as an adverb 'before some undesignated time (presumably inferable from context)', and as an inflector 'before this particular time, namely the point of speech'--Da hea durzo de must mean that X's doing Y was helped by someone or something current, perhaps a person present at the time of speech. Could we translate it as 'X does Y with your help', you the listener? Extending the currency principle to another case, what does Da sea durzo de mean? As a preposition sea means 'instead of'. So I would assume that the specimen means that X did Y instead of someone or something else, someone or something which was in some sense present at the time of speech. Could it mean that X does Y instead of your doing Y? But note that the speaker, too, is present at the point of speech...but perhaps less interestingly so, since the speaker is always present. As I say, loglanists are now exploring this vast new domain of meanings that has been opened up by the development of a machine-readable grammar.

5.7 Causal Operators kou moi rau soa and Kin

There are sixteen causal operators in Loglan, and their full significance will not be appreciated until, in the last sections of this chapter, we study the closely-related causal connectives that have been derived from them. In this section we consider how the sixteen elementary causal notions may be used prepositionally, adverbially, and even as inflectors.

The most common use of the causal relatives is as prepositions, and that is the sense of the English translations given here. Some of these meanings do not exist in the natural languages, and in these cases the Loglan meanings are often hard to think out.

Here are some examples of causally constructed sentence modifiers:

(1)

Mi pa godzi moi la Djan

I went because of John (for some motive involving John).

(2)

Mi pa godzi moi lepo vizka la Djan

I went to see John (so that I could see John).

(3)

Mi pa godzi moi

I went intentionally (with some unmentioned purpose).

(4)

Mi moi godzi

I intentionally/purposefully go.

Notice that when the prepositional sense is plain, the adverbial and inflecting senses are easily inferred. Note, too, that in Loglan we do not have to use constructions like 'because I wanted to see John' or 'because of my desire to see John' as operands of moi. Moi already says that whatever follows it is a designation of a motive, that is, a psychological state in which, by actively contemplating a desirable consequence of some action, an actor causes daself to undertake it. So we do not have to say all this over again in the operand. Because they imitate the spareness of the Loglan, I've used telegraphic phrases like 'to see John' as perhaps the best translations of moi-clauses.

Here's a sequence in which a reason is the analog of the cause, and a decision, or some justified action, is analogous to the effect:

(5)

Mi pa poltia [POHL-tyah] la Djan, rau lepo da nesta

I voted for (politically-chose) John because he is honest.

(6)

Mi pa poltia la Djan, rau tau

I voted for John because of that (previously alluded to reason).

(7)

Mi pa poltia la Djan, rau

I voted for John with reason (for some unspecified reason).

(8)

Mi rau poltia la Djan

I justifiably voted for John.

The semantical differences between the four levels of causation--the physical level with kou, the motivational with moi, the justificational with rau, and the inferential with soa--will be discussed in Sections 5.23-25.

5.8 Varieties of Free Modifiers

Modifiers that may be attached to anything at all in an utterance, including themselves, are a very common type of modifier in natural language. There are a number of distinct kinds of such modifiers in Loglan: (1) the vocatives, which, like the (marked) name hoi Djan, serve as addresses on the sentences to which they are attached; (2) the salutations, like loi [loy] = 'Hello', which serve as brief, stylized signals of certain universal human situations; (3) the attitudinals, like ia [yah] = 'I', which express the speaker's confidence, doubt or other such emotions toward what da is saying; (4) the relative interrogatives^R2, like vihu [vee-hoo], which means 'Where?', and which are built on the relative operators of the previous two sections; (5) the discursives, like buo [bwoh] or [boo-oh] which means 'however', and which allow the speaker to make comparisons with foregoing discourse; (6) parenthetic comments, made with kie...kiu [kyeh... kyoo], by which a speaker may express reservations or deviations from the main thrust of da's remarks; and finally (7) the utterance ordinals, like nefi [neh-fee], which means 'firstly' or 'primarily', by which the speaker may help daself and the listener keep track of some sequence of da's ideas.

The number of variations in some of these groups is small. There are only four salutations, for example. Some groups are middle-sized; there are about 20 discursives and 25 attitudinals. But others are essentially unlimited in number. Examples are the vocatives (any name may be one), the parenthetic comments (any utterance may be a parenthetic comment on another), and the utterance ordinals. These last, like any set of numerically-based words, are in principle infinite in number.

As mentioned earlier, a free modifier is always assumed to modify the word it immediately follows or, if initial, the whole utterance which it thus precedes. This will be illustrated in the sections that follow.

5.9 Salutations and Other Expressions of Direct Address

These are among the most primitive utterance forms of any language. With them one calls a listener's attention, singles out the person one wishes to talk to, or expresses the formal sentiments of gratitude, welcome, greeting or farewell. Sometimes these modifiers are collectively called vocatives, which is nothing more than Latin for "calls".

There are six special words of direct address in Loglan. These are Loi [loy] which means 'Hello'; Loa [loh-ah] which means 'Goodbye'; Sia [syah] which means 'Thanks'; Sie, which means "Sorry" (where an apology, not just regret, is intended); Siu [syoo] which means 'You're welcome' (in response to sia) or "Don't mention it" (in response to sie); and the general attention-caller Hoi [hoy] which, when used alone, has a sense rather like English 'Hey!' or the nautical shout 'Ahoy!, and, when used to introduce a phrase, turns that phrase into a vocative.

In addition, every name word of the language may be used as a call, when prefixed with one of the words of direct address (with nuance indicated by the word of address used).^R3 Thus hoi Djan, loi, Frans, loa, Far and loi, Lun [jahn frahns fahr loon] ('John', 'France', 'Dad' and 'Moon'), when used without the name-operator la and with a vocative marker, serve as simple calls.

Finally, any predicate expression whatever may be preceded by Hoi, and this move forms an address of any length whatever...in much the same way that the same kinds of expressions may be capitalized and used with la to form formal names. For example, on the pattern of the name la Ganbra Matma ('Noble Mother'), one may form the call Hoi Ganbra Matma ('O Noble Mother'), which affords a much more formal way of addressing one's female parent than shouting Mat. Similarly, Hoi Ganfua go Redro Nu Herfa means 'O Lady with the Red Hair' and also has all the requirements of formal address.⁴

Any call may be used alone:

So may any salutation:

Any call may be used with any salutation, and in either order:

(5)

Hoi Frans, sia

France, thank you!

(6)

Loa, Lun

Goodbye, Moon!

Any of these expressions may be used as an embellishment of any utterance in almost any position. Thus an address may be used to preface an imperative, as in

(7)

Hoi Djan, godzi

John, go!

or to follow one:

(8)

Godzi, hoi Djan

Go, John!

or to direct different parts of a message to different listeners:

(9)

Plizo ti, hoi Djan, e ta, hoi Meris

Use this, John, and that, Mary.

or to adorn declarative sentences with definite addresses:

(10)

Hoi Meris, la Djan, pa godzi

Mary, John went.

and to do this either forethoughtfully, as above, or afterthoughtfully as below:

(11)

La Djan, pa godzi, hoi Meris

John went, Mary.

Only two words in the above utterances were allowed to precede names pauselessly. These were la and Hoi. There are only a few words with that privilege in Loglan: these include the salutations, la and the little word ci which we have not yet encountered in connection with names. Any other word must be separated from a following name by a pause. This includes other names. Thus non-final names in a string of names must always be followed by pauses. In other words, /djanpolDJONZ/ will be heard in Loglan as the single name Djanpoldjo'nz, and not, as one had hoped, as Djan Pol Djonz. One is, in Loglandia, either Tam Braon or Tambrao'n depending on how one's callers speak one's name.

(14)

Djadou mi la Djan Pol, Hoi Djonz

Tell me about John Paul, O Jones

And now, observe, the Hoi-preceding comma does appear in text.

5.10 Expressions of Attitude

As we learned in Chapter 2 on word forms, the words whose main business it is to express feelings or attitudes in Loglan--especially toward what the speaker daself is saying--are all vowel diphthongs. Since there are only twenty-two of these ia-form words, we can now describe them all.

Grammatically, the attitude indicators, as we will call them, may either stand alone or embellish other forms. Thus Ia alone means 'Yes', Ia, Djan means 'Yes, John', No ia, Djan means 'No, certainly, John' or 'Certainly not, John' and Ia da pa kamla might be translated as 'Certainly X came' or 'I am sure that X came.' Like vocatives, indicators are free modifiers, for they may occur anywhere in a sentence. Like all other free modifiers, when indicators are initial they are taken to modify the utterance as a whole, but a non-initial indicator modifies only the immediately preceding element. For example, if ia follows no in some utterance, it is taken to modify that negative, that is, to express negation with conviction. Thus when it stands alone No ia [noh-yah] means 'No certainly' or 'I am certain that no.' We would express this sentiment in English the other way round, by saying 'Certainly not.' While the Ia no-order is possible in Loglan, it means 'Certainly it is not the case that' only when it occurs at the head of an utterance, for anywhere else ia will be occupied in modifying something else.

For example, in La Djan, ia no godzi the indicator ia modifies Djan, not no, giving the sense of 'John, certainly, didn't go.' It does not mean 'John certainly didn't go.' So if we do want to express the sentiment of English 'John certainly didn't go' we would have to say La Djan, no ia godzi in Loglan. Thus whenever negative attitudes are to be expressed, the preferred order in Loglan is negative-indicator, as in No ia.

Here are some possible uses of ia:

(1)

Ia mi ditca

Certainly I am a teacher.

(2)

Mi ia ditca

I, certainly, am a teacher.

(3)

Mi ditca ia

I am a teacher certainly.,

(4)

Mi ia no ditca

I, certainly, am not a teacher.

(5)

Mi no ia ditca

I am not, certainly, a teacher (I am certainly not a teacher).

We have put ia through its paces. Now let's look at the other indicators.

The twenty-two attitude indicators of Loglan are grouped into several distinct series. There is (i) the conviction series, of which ia is the head; (ii) the intention series of which ai ('I will') is the head; (iii) the obligation series, of which oa ('must') is the head; (iv) the request series of which ei ('Is that so?') is the head; and finally what we may call (v) the emotive series of which ua (expressing satisfaction) is an example. Let us consider these five series one at a time.

Degrees of conviction may be expressed in Loglan with the following series of indicators:

Three negative forms are also possible, namely no ia ('I disagree'), no io [noh-yoh] (I think not) and no ii [noh-yee] ('Perhaps not'). Thus seven distinct expressions comprise the conviction scale in Loglan.⁵

Degrees of obligation are similarly expressed:

Again there are three negatives: no oa [noh-OH-ah] ('must not do...'), no oe [noh-OH-eh] ('should not do...'), and no oi [NOH-oy] ('may fail to do...').

The word ia, we have said, means 'Yes.' But there are other kinds of affirmatives in Loglan. For example, the sense of 'Yes, you may' may be translated simply as Loglan oi. In addition there are several affirmatives which express willingness or consent. These belong to the intention series.

As usual we have three negatives: no ai [NOH-igh] ('I refuse'/'I will not'), no ao [NOH-ow] ('I don't want to'), and no ae [noh-AH-eh] ('I hope not'). (Remember that [igh] is the value of these letters in 'sigh'...which is always the sound of Loglan /ai/.)

Note the formal parallels between the three series. Each has three positives, three negatives and a zero point which takes no negative. Thus zero conviction is ignorance ('Who knows?') and is expressed with iu; zero obligation is ethical indifference ('It doesn't matter') and is expressed with ou; and zero intention is another kind of indifference ('I don't care if I do or not') and is expressed with au. Thus there are, in effect, seven positions on each of these three "attitude scales," and the zero-point on each is marked by final -u.

Only one of these scales will present serious difficulties to the speaker of English: the intention scale. This scale proves troublesome to English-speakers because in Loglan, unlike English, any of the seven intentional expressions may be used with sentences which do not involve the speaker, as well as (more comfortably for us) those that do. Thus both

(6)

No ai mi godzi la Romas

I refuse to go to Rome.

(7)

No ai tu godzi la Romas

I will see to it that you do not go to Rome.

are meaningful uses of No ai in Loglan. English forms like 'I insist that...', 'I refuse to permit you to...', 'I do not want you to...', and so on, are best treated in Loglan as intentional expressions and not as predications at all. They are therefore best translated by attitude indicators, and not by literal equivalents of the English words.

In fact, literal translation into Loglan is seldom the best choice where English expressions of attitude are involved. This applies to other kinds of attitudes as well as those of the intentional kind. For example, the sentence 'I believe the Earth is round' may occur with two quite different meanings in English. One involves an indication of moderate conviction (this is what 'I believe...' usually expresses in English):

(8)

Io la Ter, bamfoa [bahm-FOH-ah]

Probably the Earth is round (ball-form).

The other gives a self-report on the speaker's belief system such as da might make in response to an inquiry about da's beliefs:

(9)

Mi krido lepo la Ter, bamfoa

I believe that a certain state-of-affairs obtains namely that the Earth is round.

This is seldom the meaning of the English source. Both translations contrast with the unadorned and unqualified remark:

(10)

La Ter, bamfoa

The Earth is round.

It is clear that (8) and (9) make two very different claims about the world. Sentence (8) is true if and only if the Earth is round, never mind the speaker's feelings. Indeed, for most purposes (8) may be regarded as having the same truth-conditions as sentence (10) in which no attitudes whatever are indicated. Does the expression of a feeling about a sentence change the truth-value of that sentence? We suppose not. Sentence (9), on the other hand, is quite a different matter. It is true if and only if the speaker does in fact believe the condition asserted in sentence (10), never mind the Earth's geometry.

We are not often dismayed by this kind of ambiguity in English, possibly because we are seldom aware of it. But again, the Loglan speaker may wish to be utterly clear about what parts of da's speech merely express da's attitudes--and which are, therefore, neither true nor false but simply there--and which parts make claims about the world. And in expressing daself in this purely attitudinal way the Loglan speaker has great freedom. For let us note that claiming can be done with faint conviction,

(11)

Ii la Ter, bamfoa

Perhaps the Earth is round.

or faint intention,

(12)

Ae la Ter, bamfoa

I hope the Earth is round.

as well as strong conviction,

(13)

Ia la Ter, bamfoa

Certainly the Earth is round.

or strong intention,

(14)

Ai la Ter, bamfoa

I intend that the Earth shall be round(!).

Through it all, the roundness of the Earth persists. For the Earth's geometry is, of course, independent of all our hopes and fears and of nearly all but the most robust of our intentions. The logician's interest in all these sentences is substantially the same because the evidence that confirms or refutes La Ter, balfoa is in every case the same.⁶ Only after we know what something claims can we make use of whatever information we may have about the speaker's attitude toward that claim. But knowledge of claims and information about attitudes interact in important ways. Thus the startling thing about (14) is not the strength of the speaker's attitude or the geometry of his claim, but the enormity of such a claim when accompanied by such an attitude. For with (14) we suddenly have a vision of a speaker levelling mountains in the interests of geometry, while in (11), (12) and (13) the same approximate truth mixed with other attitudes may be more placidly entertained.

Note that in Loglan these two components of meaning--traditionally called the "cognitive" and the "emotive" components⁷--may be clearly separated. Again, we suppose that this arrangement will perform some service for the thinker.

The next list of indicators forms a series of requests:

Ei requests information about the truth of some matter:

(15)

Ei la Djan, pa kamla

Did John come? (Literally, 'Is it true that John came?')

(16)

Ei tu fa godzi

Are you going? (Is it true that you will go?)

Ea requests some cooperative act from the listener:

(17)

Ea mu godzi

Let's go.

(18)

Ea la Ter, bamfoa

Let's make the Earth round.

Eo requests permission, and presupposes that the listener is in a position to give it:

(19)

Eo mu godzi

Please, may we go?

(20)

Eo la Ter, bamfoa

Please, let the Earth be round.

Finally, the indicator eu is used to request the imaginative cooperation of the listener in entertaining the speaker's suppositions. Often such requests are expressed in English by using the word 'suppose' and/or putting the following verb in the subjunctive mood:

(21)

Eu mi bragai [BRAH-gigh]

Suppose I were king (born-ruler).

Remember that [igh] is as in 'sigh'.

(22)

Eu mu pa godzi

Suppose we had gone.

(23)

Eu tu fa felda

Suppose you do fall.

(24)

Eu la Ter, pilno

Let's assume the Earth is flat. (Suppose the Earth were flat.)

For certain purposes eu may be regarded as the sign of the Loglan subjunctive mood. It creates--or at least it asks for--a characteristic suspension of disbelief. This is the mood that fiction writers require of their readers as well as the one that scientists must be in when entertaining contrary-to-fact conditionals.

The final series of indicators is the emotive one. It contains the purely expressive words of the language:

No difficulty in using these words is likely to be experienced by the speaker of English.

Some compound attitudinal words have already been defined, and others may yet be; for the twenty-one simple indicators are, in a sense, merely the elementary vocabulary of attitudinal expression. Thus the compound word aiui [igh-wee] (ai + ui = 'I will' + 'happily') means 'I consent with pleasure', or simply 'Gladly'; uuoa [woo-OH-ah] (uu + oa = 'sorry' + 'must') might be translated 'I am sorry but I/you must'; and uiou [wee-OH-oo] (ui + ou) might mean 'Happily it doesn't matter'. The reader may wish to explore other possibilities of the attitudinal system.⁸

5.11 Questions with ie he ho hu ha

In addition to 'Yes-No' questions asked with indicator ei, and requests for various kinds of cooperation made with ea eo and eu, there are five more types of questions that may be formed with little word markers in Loglan. These question-marking words are ie he ho hu ha [yeh heh hoh hoo hah]; and although they are not themselves free modifiers, they are used, as we shall see, to create some. So it will be best to deal with them now before continuing with the rest of the free modifiers.

Unlike free modifiers, these five interrogative words have very definite, but largely familiar, grammatical distributions.

Ie, the only VV-word in the set, is the identity interrogative. With it one raises questions about the identity of the persons or things being designated by one's interlocutor. To use it, the speaker puts it immediately before the argument about which da wishes to raise this question. For example, suppose someone mentions la Djan Djonz and obviously assumes that you the listener know who this gentleman is. But you don't. So you say

(1)

Ie la Djan Djonz

Which John Jones?

or simply

So ie may be used alone with the sense of 'Who?' or 'Which?', or even as 'What?' in the sense of 'What did you say?', as well as to mark a failed identifier.

One answer to either of these questions might be:

(3)

La Djan Pol Djonz

John Paul Jones.

If, now, you know the identity of John Paul Jones, then (3) will be sufficient answer for you. But if you don't, you might continue with a different kind of question. 'What does this John Paul Jones do?' you might like to ask in English...in the hope of gathering more information about this still unknown person. That kind of question requires the use of the interrogative predicate he [heh] in Loglan:

(4)

La Djan Pol Djonz, he

John Paul Jones is/does what?

He is the question-asking predicate. It has the same grammar as any other predicate word. In fact, as far as the grammar is concerned, he is a predicate. But semantically, he is a place-holder: it asks us to replace it with a real predicate that will make a true and useful sentence out of the one in which it appears. Either of the following answers might give the question-asker the information da needs:

(5)

Da farfu la Ruprt Djonz

X is the father of Rupert Jones.

(6)

Da ditca lo numsensi vi le ganta ckela

X teaches mathematics (number- science) at the high school.

Not nearly so informative but briefer is another possible answer:

(7)

Ditca

(X) Teaches.

Is this legitimate? It certainly looks like the imperative, 'Be a teacher!', and certainly the speaker does not mean that. On the other hand whatever may grammatically replace the interrogative predicate he must obviously be a legitimate answer to whatever question he has created. And Ditca may usefully replace he in La Djan Pol Djonz, he if John Paul Jones is indeed a teacher. So as an answer to an he-question Ditca will not be heard as an imperative but as an abbreviation of the longer answer Da ditca.

(7)

Seu ditca

(X) Teaches.

But let's say you're still not certain who this fellow Jones is. So you pursue the matter opened with sentence (6) with another interrogative, this time using the interrogative argument hu. Just as he is a member of Lexeme PREDA, so hu is the interrogative member of Lexeme DA:

(8)

Le ganta ckela je hu

The high school of what (community)?

If you are a loglanist, you know that high schools, like restaurants, are always parts of larger wholes, namely the communities they serve. Perhaps knowing the community that this high school serves will help you identify this particular math teacher. The answer you get might be:

(9)

Je la Nordi Spali

Of the North Side.

Or you might be given the longer, formal name of the school where Jones teaches:

(10)

La Nordi Spali, ge Ganta Ckela

The North Side High School.

Or our informant might respond by replacing hu in (8) with:

(11)

La Nordi Spali

The North Side.

Thus hu has the same grammar as any other argument variable. For example,

(12) Hu ditca hu hu

has the same grammatical structure as

(13) Da ditca de di

and happens to be a perfectly legitimate Loglan question which translates into 'Who teaches what to whom?' Thus, curiously enough, questions and declarative sentences have exactly the same grammar in Loglan. What distinguishes them is not their grammar but the words they contain.

But let us proceed with our inquiry. Our interlocutor now offers us the new information:

(14)

La Djonz, kapta

Jones is a captain.

We ask:

(15)

Da he, kapta

He's what kind of captain? (Literally, He's a what-captain)?

Our informant replies:

(16)

Da mursi kapta

He's a sea captain.

We pursue that with:

(17)

Da mursi kapta hu

He's a sea-captain of what?

And then we get our final answer, the one that finally identifies our man:

(18)

La Bonom' Ricar'

(Of) The Bonhomme Richard.

And to this, the only suitable reply of course is Ua. Or perhaps, if one is the least bit surprised, Ueua.

The last two interrogatives are ho and ha. Ho is the interrogative quantifier and may be used wherever any other quantifier may be used. For instance, we may ask

(19)

Ho le mrenu pa kamla

How many of the men came?

and the answers may be:

or

(21)

Ne le mrenu pa kamla

One of the men came.

Or we may ask

(22)

Hoba tugle tu

How many legs have you? (Literally, how many somethings x are legs of you?)

(23)

Hoba herfa letu hedto

How many hairs on your head? (Literally, how many somethings x are hairs of your head?)

Again, the answers might be Te, Ne, Ro or Sunenimo ('At least ten thousand'). For whatever may legitimately replace ho in a question is obviously a legitimate answer to that question. Thus most answers to ho-questions consist simply of quantifiers.

Ha is the Loglan interrogative connective. This is something that doesn't exist in natural language but is essential to the conduct of a logical one. Let us see why. Suppose you encounter a loglandian hostess who knows how foolish it would be to ask you if you'd like 'tea or coffee' since if you want either, your answer, being that of a loglanist, is bound to be Ia, sia. But such an answer would not advance her project at all. So to forestall such correct but useless answers she uses Loglan's interrogative connective to ask a different question:

(24)

Tu danza lo tcati, ha lo skafi

You want tea how-connected-to coffee?

Now, to answer her properly you must give her a replacement of ha that fits your case. If you want neither tea nor coffee, you will reply Noenoi sia, which is Loglan for 'Neither, thank you.' If you want tea but not coffee, Enoi sia ('And-not, thank you') will serve very well. If not tea but coffee, Noe sia ('Not-and, thank you' will signal that preference neatly. And if you want both, one or the other or both, one or the other but not both, one or the other or neither but not both, etc., etc., you would answer, E sia, A sia, Onoi sia, Noanoi sia, respectively, down the long list of logical possibilities which loglanists will tend to savor more than other people, perhaps.⁹

In sum, loglanists use ie when they want to know the identity of the designatum of some inadequate argument (Ie da = 'Which X?'), he when they want a predicate for an answer (Da he = 'X is/does what?' Tu he = 'How are you?'), hu when they want an argument for an answer (Hu gritu = 'Who sings?'), ho when they want a number for an answer (Ho da = 'How many X's?'), and ha when they want a connective for an answer (Da ha de = 'X is how connected to Y?').

The result is an utterly novel but very flexible system of interrogation that is capable of translating with great precision all the questions of natural language and a great many more besides. Questions like 'When?' and 'Why?' are still missing from our catalog. These will be supplied in the next section through the good offices of the argument interrogative hu.

5.12 Relative Interrogatives with -hu

The next group of words appear here because they were for some time free modifiers, but they have resumed their status as sentence modifiers in the 2013 updates. They are also question-forming words. These are the relative interrogatives made by postfixing the suffix -hu to any of the relative operators described in Section 5.5. This move produces words like nahu [nah-hoo] = na + hu = 'at' + 'what (time)?' or 'when?', vihu [vee-hoo] = vi + hu = 'in' + 'what (place)?' or 'where?', kouhu [koh-OO-hoo] = kou + hu = 'because of' + 'what (event)?' or 'why?' and all their diverse kin. Speaking one or more of these words in any sentence will turn it into a question. We are beginning to see why we do not use the question mark in Loglan text. Nor do we need, but of course may freely use, the rising tonal contour that is the sign of a question in many natural tongues. Utterances become questions in Loglan when they contain one or more interrogative words. So Loglan questions do not have to be marked by rising pitch in speech, nor by question-marks in text. A spoken word will have told the listener that it is a question da has heard.

The interrogative compounds made with -hu are, as mentioned, sentence modifiers.

(1)

Nahu tu pa mercea la Pol

When did you get married (married-become) to Paul?

(2)

Kouhu tu pa mercea la Pol

Why (from what external cause) did you marry Paul?

(3)

Moihu tu pa mercea la Pol

Why (from what motive) did you marry Paul?

(4)

Rauhu [rah-OO-hoo] tu pa mercea la Pol

Why (for what reason) did you marry Paul?

Obviously, there can be many varieties of why in Loglan. We will discover why that is so, and what kind of why that 'why' is, in the last sections of this chapter.

Let us now observe what happens when speakers choose to use one of these question-making words in a non-initial position. The result is still a question, but of a rather different kind (we insist that these are sentence modifiers as their form suggests, but we allow the emphasis rules of free modifiers to apply as a matter of style):

(5)

Tu rauhu pa mercea la Pol

Why did you marry Paul? (Literally, You-why married Paul?)

(6)

Tu pa mercea rauhu la Pol

Why did you marry Paul? (You married-why Paul?

(7)

Tu pa mercea la Pol, rauhu

Why did you marry Paul? (You married Paul-why?)

Any of the hundreds of relative interrogatives may be used in all these entertaining ways.

5.13 Discursive Modifiers

When a speaker wishes to call the listener's attention to some aspect of the flow of discourse, da uses one or more discursive modifiers. These are single Loglan words which function much as phrases like 'For example', 'In short', and 'And so forth' do in English. Generally Loglan speakers use discursives to call attention to a comparison, contrast, or inference that may be drawn or found between the current sentence--the one in which the modifier resides--and certain foregoing portions of the discourse.

There are presently twenty-seven of these discursive words in Loglan; see Table 5.3. They form an open class which may be added to at any time. All are CVV in form and all the current ones have been derived from primitive predicates selected to suggest their meanings.¹⁰

(1)

Bea da ditca

For example, X is a teacher.

(2)

Buo de cirna

In contrast, Y is a student.

(3)

Sui da turka le botsu

Moreover, X works on the boat.

This is their usual position. But non-initial positions in the utterance are also possible.

(4)

Da ditca bea

X teaches, for example (as an instance of a more general activity mentioned earlier).

(5)

De buo cirna

Y, in contrast (to some X), is a student.

(6)

Da turka leda gardi sui

X works on his garden, too (in addition to working on something else).

But note this. Sentences (1)-(6) must be initial in their speeches else they would be marked with I. So from the absence of a resident I-word we know that these are either the first or the only utterances in their speeches. But it is the role of a discursive word to comment on foregoing portions of a discourse. Therefore the sutori (from Loglan su + to + ri) utterances in speeches are much more frequently adorned with discursives than speech-initiating utterances are. Moreover, when discursives are initial in sutori utterances they are often compounded with the continuation operator I:

(7)

Ro le frasi ga smina turka. Ibea [ee-BEIGH-ah] la Pi,e'r, ditca

Many of the French (people) are mental workers. For example, Pierre is a teacher.

The compounding move is not obligatory, however. The speaker may choose to pause after the word I--which usually means stressing it as well--and that keeps the two operators, and their functions, separate:

(8)

Ro le frasi ga smina turka. I, bea la Pi,e'r, ditca

Many of the French are mental workers. And, for example, Pierre is a teacher.

The Ibea-usage is likely to be more frequently employed than the I, bea-one, however.

Grammatically, compound words like Ibea, Ibuo and Isui [ee-BEIGH-ah ee-BOO-oh EE-swee] are no longer strictly discursives. Specially modified as they are to be at the head of an utterance, they are no longer "free to go anywhere". Instead, like I itself, the I-prefixed discursives are members of the I Lexeme. They are I-words; so these words, too, mark the joints, in multi-utterance speeches, between one parsable unit of discourse and the next one.¹¹

5.14 Parenthetic Remarks with kie...kiu

Parenthetic remarks are also free modifiers in Loglan for they, too, may be used anywhere. No problems are likely to arise for speakers and writers of European languages in using Loglan parentheses. The written forms of these languages all use parentheses in ways that are similar to Loglan. So for Europeans, learning to use the Loglan parentheses in speech will be just another instance of learning to speak a familiar punctuation mark out loud.

There is a novelty introduced by this particular application of the spoken punctuation principle, however, and that is that, by using the two spoken words kie [kyeh] and kiu [kyoo] in ways that can be modeled on the speaker's customary use of opening and closing parenthesis in everyday writing--whatever da's custom is--Loglan speakers may learn to warn their listeners that they are about to embark on a qualification or an excursion, perhaps a long one, but that eventually they will return to the point at hand. This almost never happens in natural speech. Yet every user of a written opening parenthesis makes da's readers just this implicit promise: 'Permit me this aside, but I promise to return'; and to use a closing one is like saying 'There. I'm back. We can now get on with what we were talking about.' With the spoken parentheses of Loglan, the same sentiments can be spoken briskly; so the same courtesy, and the same social pressure to observe it, may become available in speech.

Pauses generally precede both the opening parenthesis and the closing one. As usual any such pauses are represented by commas in text. Neither the commas nor the pauses they represent are obligatory, however.

(1)

La Selis Djonz, kie Mi pa condi cluva da, kiu pa sorme lemi fremi

Sally Jones (I loved her deeply) was a sister of my friend.

Sometimes, as in this specimen, the initial letter of the enclosed remark is capitalized. In translating such spoken parentheses into written English, it is often just as appropriate to use a pair of dashes (--):

(1') Sally Jones--I loved her deeply--was a sister of my friend.

The choice between dashes and parentheses tends to be idiolectic in English, so the translator should follow da's own writing customs in this regard.

The parenthetic remark itself may be of any grammatical type whatever, and as simple or complex as the speaker or writer likes:

(2)

La Selis Djonz, kie Mi djadou ei tu da, kiu pa sorme lemi fremi

Sally Jones--I did tell you about her?--was a sister of my friend.

(3)

La Selis Djonz, kie Ueuiua, kiu pa sorme lemi fremi

Sally Jones--Wow!--was a sister of my friend.

Indeed any utterance or utterance string whatever may appear between parentheses.

When a parenthetic remark is initial it modifies, as usual, the entire utterance. So such a parenthetic comment at the head of a speech often has the effect of a label. For example, a playwrite might choose to identify da's characters in this way:

Read aloud--as, for example, by the prompter--this would be:

(4') Kie La Hamlet, kiu Lepo dzabi, onoi lepo no dzabi

whereas the actor himself might read it:

(4") Lepo dzabi, onoi lepo no dzabi

In Loglan text the parenthesizing words kie and kiu, together with any attendant commas, may of course always be replaced by their conventional symbols. Thus the biographer could write:

and the reader would read aloud

For as always, the written symbols '(' and ')' are pronounced like the words they replace.

5.15 Utterance Sequencers with -fi

This simple but, in principle, infinite category of words contains the last of the free modifiers and need not detain us long. Any quantifier, whether numerical or not--or any letter-variable, for that matter--may be postfixed with the suffix -fi to generate a compound little word, for example Nefi [NEH-fee] 'firstly' or 'primarily'. Such words may be used to remind the listener--or the speaker daself, for that matter-- of the position of some utterance, or some portion of an utterance, in some sequence of such elements in da's speech.

The quantifiers most frequently used as sequencers are, of course, the positive integers: ne to te, etc., giving Nefi Tofi Tefi [NEH-fee TOH-fee TEH-fee] 'firstly' 'secondly' 'thirdly', etc., as the sequencers. Often these words may be more simply translated as 'first' 'second' 'third', etc., or even as 'one' 'two' 'three'. Decimalized numbers may also be used, Nepinefi Nepitofi [neh-pee-NEH-fee neh-pee-TOH-fee], etc., as for the sections of this book. Alphabetic sequencers may also be used. They may be either Latin or Greek, or upper or lower case. For example, Amafi Baifi Caifi [AHM-mah-fee BIGH-fee SHIGH-fee], etc., illustrate the use of Latin capitals as utterance sequencers.

A common, but certainly not the only, position for these free modifiers is at the heads of the utterances they sequence. When this is the case the first sequencer is capitalized and the sutori ones are prefixed with the continuation connective I-, giving Itofi Itefi Ifofi [ee-TOH-fee ee-TEH-fee ee-FOH-fee], etc., as the markers of the sutori utterances in the sequence. This is exactly what happens to the discursives in these same positions (Section 5.13) and imparts a recipe-like structure to the speech:

(1)

Nefi, kutla le mitro su nera inca pisku

First, cut the meat into some (at least one) one-inch pieces.

(2)

Itofi, jarklu [ZHAR-kloo] le palto

And-second, slice (thin-cut) the potatoes.

(3)

Itefi, nenbromao [nen-BROH-mough] te frese negda

And-third, break in (in-break-make) three fresh eggs.

(4)

Ifofi, tensea [ten-SEIGH-ah] le krima, e le vinjo

And-fourth, add (increase put) the cream and the wine.

(5)

Ifefi, gudbi mismao [MEES-mough]

And-fifth, mix (mixture-make) well (literally, good mix).

(6)

Irafi, kokfa le miksa vi ne zavno nia nepife horto

And-finally, cook the mixture in an oven for (during) one-point-five hours.

We have now completed our survey of both sentence modifiers and free modifiers. We have just one more type of modifier to consider, and these are the argument modifiers ('John the teacher' = La Djan, ji le ditca). These, however, will require us to understand the distinction between "predication" and "identification". We can best do this by first understanding its application to sentences and questions. So it is these topics that we must now take up.

5.16 Identity Sentences with bi and bie¹²

In Section 5.11 we found that a speaker could inquire about the identity of someone or something simply by preceding any designation of that person or thing with the interrogative ie. Thus

(1)

Ie la Djan

Who is John? (Literally, Which the John?)

is a question about the identity of someone named John. An adequate answer would be an alternative, though better, designation. We can also ask the same kind of question with hu:

(2)

Hu farfu la Selis

Who is the father of Sally? (Literally, Who fathers Sally?)

For this, too, asks for a designation as an answer.

Usually such questions arise when the designation which some listener has been given fails to locate the designated thing for that listener. There are five Johns in the room, let us say. Someone tells you 'John is a chemist.' So you say 'Which John?' Or, with some ponderousness, perhaps: 'Whom do you mean to designate with the name 'John'?'

Such questions simply ask for better designations. Useful answers might be 'That tall man over there', or 'John Jones', or 'Mary's brother', and so on. You succeed in identifying something for someone when you provide a designation that really works for that someone, that is, one that enables da to single out the thing or set of things--perhaps only in da's mind--you are actually talking about. Thus

(3)

Leva langa mrenu

That long (i.e., tall) man.

(4)

La Djan Djonz

John Jones.

(5)

Le brudi je la Meris

Mary's brother.

are all possible answers to questions (1) and (2). They are therefore all possible utterances. Because they are nothing but argument forms, and are not, in themselves, sentences, we will sometimes call this form of utterance an identification. An identification is, of course, just one of the many kinds of answers. Its grammar is extremely simple; for clearly every designating argument form of the language is a potential answer.

But our answer to questions (1) or (2) may be even more explicit. Just as we can say in English 'John is Mary's brother' in answer to such questions, so we can produce what we will now call an identity sentence in Loglan:

(6)

La Djan, bi le brudi je la Selis

John is (identical to) the brother of Sally.

(7)

Da bi la Djan Djonz

X is John Jones.

Such sentences--and they are sentences--say nothing about anyone. They simply equate two different designations of the same thing. So what they are about is designations that have a common designatum. Thus the little word bi is a kind of predicate, but a very special kind. It means 'The expression '...' designates the same thing as the expression '...' does.' Usually, the argument to the left of bi is unfamiliar, and the argument to its right is a designation of the same object which is more familiar to the listener...or so the speaker hopes. Suppose, for example, you have never heard the name 'Samuel Clemens'. Hearing it for the first time you are likely to ask 'Who's Samuel Clemens?' And the speaker could well answer 'Samuel Clemens is Mark Twain.' Or in Loglan

(8)

La Semiul [SEM-yool] Klemenz, bi la Mark Tuein

Samuel Clemens is (the same person as) Mark Twain.

Have you been told anything about either Samuel Clemens or Mark Twain? No. But you have been told something about his two names, namely, that both of them designate the same person. This is useful--and indirectly it tells you that whoever he was he had (at least) two names--but it is not a sentence about Mark Twain.¹³

Identity sentences are obscure creatures in the natural languages. For they travel under the same grammatical guise as predication sentences. Suppose I tell you 'John is the father of Paul' in English. Am I identifying John? Or predicating something about him and Paul? Out of context it is impossible to tell. But if I have asked you who John is and you then tell me that he is the father of Paul, you are identifying him. For you are telling me that the man I can (presumably) locate as the father of Paul is also the person you mean to designate when you say 'John'. But if I know who John is and then ask you what he is or what John does, and you tell me that John is a teacher, or one of Mary's brothers, or that he is the father of Paul, you have told me something both testable and useful about John, and this requires that you use a predication sentence if you are speaking Loglan.¹⁴

Now in some cases we can sense this same difference between predication and identity in the English shift from 'the' to 'a', as in the following pair of sentences:

(9)

La Djan, bi le brudi je la Selis

John is the brother of Sally (that is the one I mean).

(10)

La Djan, brudi la Selis

John is a brother of Sally (she may have several).

And in these cases, the difference between the English predication sentence and the corresponding identity sentence is unusually clear. But some predicate relations connect individuals uniquely. Paul can have only one father, and in English we would therefore never say 'John is a father of Paul.' Thus in the case of the following pair of sentences

(11)

La Djan, bi le farfu je la Pol

John is the father of Paul.

(12)

La Djan, farfu la Pol

John is the father of Paul.

there is no perceivable difference between the English sentences. Yet a very clear difference exists in Loglan.

Identity sentences will not be easy for the English speaker to recognize, not even in his own speech. Therefore the Loglan difference between identities and predications--a difference which is fundamental to any logical language--will for a time seem arbitrary and unnecessary. But perhaps the following rule of thumb will help. When you are trying to help your listener locate something which da has failed to locate from some earlier designation, then you will probably do so with identity sentences, even in English. But when questions of identity are, at last, cleared up, and your listener knows who Samuel Clemens is, then your subsequent remarks about him--for example, that he informed a certain newspaper that the reports of his death were greatly exaggerated--will almost undoubtedly be delivered in predication sentences. For only these give information about the non-linguistic world.

There is a second identity operator which, in natural language, is used chiefly by mathematicians and logicians. This is the membership operator which means '...is a member of class...'; and in Loglan is bie. Membership sentences- -for so we can now call them--are formed with bie exactly as the more common type of identity sentences are formed with bi. Thus:

(13)

La Djan, bie leva te mrenu

John is one of those three men.

(14)

Da bie le natra numcu

X is one of the natural numbers (that is, a member of the set of natural numbers).

(15)

La Ter, bie le telfoa [tehl-FOH-ah] je la Sol

The Earth is one of the planets (Earth-forms) of the Sun.

Note that the objects designated on the right of bie are here assumed to be sets, not singular individuals. Thus (15) differs from (15a) La Ter, bi le telfoa je la Sol ('The Earth is the planet of the Sun (i.e., the one I have in mind)') only in its operator; yet the designation on the right now has a different meaning.

Like identities, membership sentences are useful chiefly in settling questions which arise from obscure designations. But unlike identity sentences, membership sentences do not single out familiar individuals, but familiar sets of individuals among which the individual in question is presumably to be found. Thus if you have never heard of either Samuel Clemens or Mark Twain, I may not be able to identify him for you, but I can at least give you one of his memberships:

(16)

La Mark Tuein, bie le grada ge merki srite go neveri heknirne

Mark Twain is one of the great American writers of the 19th Century (hectoyear).

(The last pause is a phrasing pause and optional.) I may not have succeeded in locating him for you; but at least you know where to look.¹⁵

The uses of identities and membership sentences in mathematics and logic are matters beyond the scope of this book. Perhaps it will suffice to suggest the importance of this topic for these disciplines to say that mathematics especially is almost entirely concerned with questions of identity. When you are asked in mathematics what number X is--and you are given, for example, the identifying information that X² + 2X = 15--you are being asked, not for a predicate which is true of X, but for another and more satisfactory designation of the particular (unknown) number that is designated by X. When you "solve" such an equation you find such a designation. And it will in truth help you locate X among the natural numbers.¹⁶

5.17 Identifying vs. Predicating Modifiers

We are now ready to take up the last variety of Loglan modifiers, the argument modifiers. These are the prepositional phrases, subordinate clauses, adverbs and even modifying arguments that are made local in their effects by being attached to the preceding argument. This is accomplished by one of the modifying linkers ji ja, jie jae or jio jao [zhee zhah, zhee-eh zhah-eh, zhyoh zhough]. Every argument modifier is marked with one of these six links, and all of them localize the effect of the linked modifier:

(1)

Lemi brudi ji vi levi hasfa, ga ditca lo helna

My brother--the one in this house--teaches Greek.

(The pauses in this and the following sentences are all phrasing pauses and would be eliminated in rapid speech.) If we remove the localizing link ji from (1), the phrase vi levi hasfa becomes of course a sentence modifier:

(2)

Lemi brudi vi levi hasfa, ga ditca lo helna

My brother, in this house, teaches Greek.

which, as we learned in Section 5.4, makes substantially the same claim as

(3)

Lemi brudi ga ditca lo helna vi levi hasfa

My brother teaches Greek in this house.

If, now, we link vi levi hasfa to the preceding argument in this new position, we get quite a different story:

(4)

Lemi brudi ga ditca lo helna ji vi levi hasfa

My brother teaches the Greek which is (say, spoken) in this house.

What my brother teaches is that particular dialect, perhaps, that is spoken in this house. All of the six modifying links localize modification in this way.

But the links have different senses and some take different kinds of operands. Ji is the identity link. It makes the information added by the modifier play a part in identifying the designatum of the argument to which it is attached. In (1), for example, it says that the one of my brothers I am talking about (presumably I have several) is the one who is in this house. In (4) ji narrows down the vast domain of all things (linguistically) Greek to that particular portion of it that is "in this house". Ji will link arguments, relative operators, clauses, and even sentences without first arguments to (other) arguments in this same restrictive way:

(5)

Lemi brudi ji le ditca ga gudbi mrenu

My brother the teacher is a good man.

(6)

Lemi brudi ji ditca lo helna ga gudbi mrenu

My brother who teaches Greek is a good man.

Both (5) and (6) add restrictions to the designation given by the argument so-modified, namely Lemi brudi. Sentence (5) does so by offering an alternative designation (le ditca) in case the first one (Lemi brudi) fails; (6) does so by adding a predication but in an identifying way. Perhaps the speaker has several brothers who teach, and this is the one who teaches Greek.

Or perhaps both speaker and listener are in the same room with the speaker's brother. So the speaker may improve da's designation by linking it to the simple locator va:

(7)

Lemi brudi ji va, ga gudbi mrenu

My brother over there is a good man.

Notice that in all these sentences, a modifier linked with ji identifies.

But just as we distinguish in Loglan between identity sentences and predication sentences, so we must now draw the same distinction between modifiers that identify and modifiers that make claims about the world. Thus instead of identifying something or someone, a speaker may wish to use a modifier to make a subsidiary claim about some already identified object in the course of making some larger claim about it:

(8)

Lemi brudi ja ditca lo helna, ga gudbi mrenu

My brother, who teaches Greek, is a good man.

In sentence (8) the predicating link ja performs the same grammatical function of attaching the predicate clause ditca lo helna to lemi brudi that ji performs in (6), but with the notably different message that my brother is, incidentally, a teacher of Greek. Such subordinate clauses may, of course, be much more elaborate than this. For example:

(9)

Lemi brudi, ja ditca lo gleca le friki na lo dotra vi la Ganta Silezias, ga gudbi mrenu

My brother, who incidentally teaches English to the Africans in the wintertime in Upper Silesia, is a good man.

And this in turn, can be turned into an identification very simply by replacing the little word ja by ji:

(10)

Lemi brudi ji ditca lo gleca le friki na lo dotra vi la Ganta Silezias, ga gudbi mrenu

My brother--the one who teaches English to the Africans in the wintertime in Upper Silesia--is a good man.

One sometimes has to make rather strenuous efforts in the English--using phrases like 'the one', 'incidentally', dashes, and the like--to make such distinctions clearly. But in Loglan, being a logical language, this important distinction is drawn phonemically by the linking words themselves.

Notice that there is no difference in truth-conditions between the claim of (9) above, with its subordinate predicating clause, and the same claim made with the same subject and a connected predicate:

(11)

Lemi brudi ga ditca lo gleca le friki na lo dotra vi la Ganta Silezias, e gudbi mrenu

My brother teaches English to the Africans in the wintertime in Upper Silesia, and is a good man.

Sentence (11) is not a legitimate transform of (10); for as we have noted several times, identifying information may be false and still be useful...that woman may still be a man. But the function served by predicating modifiers, as distinct from the locating function served by identifying ones, is to permit the speaker to claim less important things in a by-the-way fashion while making more important claims directly. Thus the hearer clearly understands from (9) that it is the fact that the speaker's brother is a good man that is the main burden of the speaker's message, and that the business of where and what the brother teaches is only incidental information. While from

(12)

Lemi brudi, ja gudbi mrenu, ga ditca lo gleca le friki na lo dotra vi la Ganta Silezias

My brother, who is incidentally a good man, teaches English to the Africans in the wintertime in Upper Silesia.

which is equivalent to (9) by way of (11), the listener would form the opposite impression.¹⁷

Note that both identifications with ji and predications with ja may be accomplished in the same sentence:

(13)

Lemi brudi ji la Djan, ja ditca ga gudbi mrenu

My brother John, who incidentally teaches, is a good man.

Or several identifications:

(14)

Lemi brudi ji la Djan, ji ditca ga gudbi mrenu

My brother John (the one) who teaches is a good man.

Or several predications:

(15)

Lemi brudi ja ditca lo gleca la Pidr, ja pa kamla la A'frikas, ga gudbi mrenu

My brother, who incidentally teaches English to Peter, who incidentally came from Africa, is a good man.

The word 'incidentally', while it translates no word in the Loglan expression directly, seems to express the predicative nature of ja-clauses quite adequately in English. Without such clarifying asides such matters are difficult to be clear about in English.

Another pair of modifier linkers are jio [zhyoh] and jao [zhough]. In this pair jio is the identifier and jao the predicator, and both have the sense of English 'such that'. Both allow supplementary information to be incorporated in the main sentence in the form of a subordinate sentence in which the argument shared by the two sentences is not initial:

(16)

Da jio la Djan, pa donsu le botci da, pa nigro horma

X, such that John gave the boy X, was a black horse.

In more usual English, we would say, 'The gift that John gave the boy was a black horse.' By converting donsu with fu and linking two of its arguments, the same thing can be said with much the same word order in Loglan:

(17)

Le fu donsu je le botci jue la Djan, pa nigro horma

The gift given the boy by John was a black horse.

Note that (17) makes no subsidiary claims; for the predicate relation between the gift, the boy and John, which might be thought to have been claimed in (16), is part of the identifying description in (17). It is for this reason that we interpret constructions formed with jio to be identifying rather than predicating clauses.

But the same clause can be predicatively linked to Da with jao:

(18)

Da jao la Djan, pa donsu le botci da, pa nigro horma

X--and by the way, John gave the boy X--was a black horse.

Now, because of jao, the Loglan listener will know that the ensuing clause is meant to convey, not an identification, but a subsidiary claim. The phrase 'and by the way' helps to express this intention in English. But while the intent of the Loglan is clear, it is not until we speak the two claims as separate English utterances that we realize fully what the English of (18) is really saying:

(19)

Da pa nigro horma. I la Djan, pa donsu le botci da

X was a black horse. And John gave the boy X.

The way to determine whether an argument modifier in a natural language sentence deserves a predicating link in its Loglan translation is to unwind the two clauses as separate assertions first and then judge whether their joint assertion is the intent of the original sentence. Thus (18) is clearly equivalent to (19) while (16) equally clearly is not.¹⁸

Both identifying and predicating clauses apply to the argument they immediately follow. Thus in

(20)

Le brudi je la Djan, ji le ditca pa gudbi mrenu

The brother of John the teacher was a good man.

the identification ji le ditca applies to John, not his brother. But in case we do wish to make an identification apply to an entire specified description we must first close off that description with the closing comma gu, as follows:

(21)

Le brudi je la Djan, gu ji le ditca, pa gudbi mrenu

The brother-of-John--who is the teacher--was a good man.

Now it is the brother who is being further identified as the teacher. But, as the translation suggests, it is difficult to make such distinctions in English. There will be more on the uses of punctuators in Sections 5.20-21.

Finally, the pair of modifier links jie and jae exists to permit class-membership clauses to be used in either an identifying or a predicating way. As usual, the one that contains the phoneme /i/ is identifying:

(22)

La Plutos, jie le la Sol, telfoa, ga kleda la Ter

The Pluto that is one of Sol's planets is colder than Earth.

while the one that contains the /a/ is predicating:

(23)

La Plutos, jae le la Sol, telfoa, ga kleda la Ter

Pluto, which is (incidentally one of Sol's planets, is colder than Earth.

In English it is often difficult to tell whether the speaker means to predicate or to identify with da's subordinate clauses. For example, without reading their parenthetic portions '(The)' and '(incidentally)' aloud, the sense of the English translations of (22) and (23) seem to differ only slightly if at all...despite the presence of commas in the latter and their absence in the former, a feature that is supposed to settle this matter in English text. But in fact, it rarely does. Pauses in speech are equally helpless to make these distinctions unambiguously.

But in Loglan the small phonemic difference in the links themselves makes this subtle difference clearly. We have learned to expect this. For the Loglan speaker must be clear about whether da is identifying something or claiming something to be true of it whenever da uses an argument modifier...if only because there is no way da can speak Loglan without doing the one thing or the other clearly. In Loglan there is thus no middle ground between predication and identification, as there apparently is in English.^19,20

5.18 Punctuation

To punctuate an utterance is to insert a syntactic marker at a point at which, if the marker were removed, the utterance would be perfectly grammatical but would mean something that one does not intend. The syntactic marker may be nothing more than a pause-comma, i.e., a pause in speech and a comma in text. Or it may be a g-initial word like ga or gu. Punctuation, therefore, is the avoidance of unintended meanings through the use of special markers whose entire function is syntactic, that is, which do nothing other than distinguish one structural arrangement of the words of an utterance from some other.

Punctuation is a surprisingly simple topic in Loglan. Apart from the morphological occasions for pausing--which are few enough: (i) the pauses after names, (ii) before those names that are not immediately preceded by la, hoi, or other "name markers", (iii) before all vowel-initial words (with minor exceptions related to VCV letterals), (iv) before all eks whether vowel-initial or not, and (v) before predicate words when the just-preceding syllable is stressed--there are relatively few types of Loglan constructions that need punctuating; and very often the punctuation they do need is the simplest of all punctuation, the pause-comma. We will list in the next four sections all the constructions that ever need punctuating in Loglan. But, as we'll see, not even these constructions always require it.

For completeness we begin the next section by reviewing the punctuating move with ga that we already know. Then in Section 5.20 we will deal with the default clause punctuator, which is gu. In Section 5.21 the uses of punctuators ga,gi,goi,gio in effecting abnormal word-orders will be described; and in Section 5.22 the four specific punctuators gue gui guo guu will be discussed.

5.19 Marking Main Predicates with ga

When the main predicate of a sentence is untensed and otherwise unmarked, and when the last preceding element that is not a free modifier is a predicate word, then ga is used just before that main predicate to prevent it from being absorbed into the preceding predicate string. For example,

(1)

Le mrenu ia, ga tsitoa [tsee-TOH-ah]

The man, certainly, is a thief (is one who takes criminally).

Note that ia alone does not perform this predicate-protecting function:

(2)

Le mrenu ia tsitoa

The man (type), certainly, (of) thief. (The one who is a man? Or steals men?).

Thus (2) is an answer, not a sentence.

Since free modifiers may go anywhere, their appearance at some particular point in an utterance is without grammatical significance. So, as far as the parser is concerned, they might as well not be there.²¹

Notice also that, while free modifiers can't do it, any relative operator can perform ga's role.

(3)

Le mrenu kou tsitoa

The man is "causedly" a thief.

In fact, ga, like kou, is just another member of the PA Lexeme, the lexeme that contains all the relative operators (Section 5.5). So ga is a relative operator, but the one which has no other meaning than the punctuational one.

5.20 The Specific Terminators gue gui guo guu

Three types of clauses sometimes require this kind of closely-targeted termination: je/jue-clauses, and these can be terminated by gue; ji/ja phrases or clauses (and their kin), and these can be terminated by gui; and lepo-clauses (and their numerous kin) (as well as abstract predicates), and these can be terminated by guo^R10. The fourth punctuator guu, introduced in the 1990's, terminates the "termsets" which follow predicates. Any of these punctuators may be used in place of a string of gu's--or in place of a single gu, for that matter--to reach back to the last-occurring instance of the type of clause which it is its business to close. In doing so it will ignore all other species of unterminated phrases and clauses, and zero in on its target. The difference between the specific terminators and gu is precisely that: gu will close the first unclosed clause or phrase it comes to in its leftward search for structures to close, regardless of their grammatical type. The translations of Appendix G contain several instances of these specifically targeted terminators at work. One can see how much more efficient they are than gu in deeply nested prose.

It is all right to use gu whenever a single gu will do the job. When there is any doubt (and a remarkable number of small structures you might overlook may be closed with a single gu) you should consider using one of the specific terminators in place of the string of multiple gu's (of length possibly difficult to divine!) that would otherwise be necessary. The machine could parse the latter just as well as a single specific terminator, of course; but humans find strings of "right parentheses"--and that is what a string of undifferentiated gu's amounts to--tedious to both produce and unravel. Sentences in which clause-specific terminators have been used are apparently much easier for humans to understand.

What is actually happening here is that there are four disjoint grammatical classes associated with the specific terminators, with gu being usable in place of each of them (gu will always close the smallest and nearest structure it can close) and another class of "gaps" for which gu alone is appropriate.

5.21 Marking Right-Boundaries with gu and Its Relatives

Gu is the all-purpose comma-like word.^R5 It can be used to close off a wide variety of constructions that would not end at just that place without it. It is thus a right-boundary marker. In fact gu is like an unmatched right-parenthesis. It can remain unmatched because the left-boundaries of these constructions are always known. For reasons just noted, in most cases it is better to use one of the specific terminators, where this is possible.

An example we have already seen is:

(1)

Mu titci vi gu le supta

We eat, here, the soup.

Without gu the sentence modifier vi would not end, as intended, with vi itself. Instead, vi would absorb le supta into a longer sentence modifier, an unintended prepositional phrase:

(2)

Mu titci vi le supta

We eat in the soup.

Notice that the beginning of this modifier with vi is in either case plain. This is a case in which gu is the only closer that will work.

A pause-comma may also accompany gu.^R6 It may precede it,

(4)

Mu titci vi, gu le supta

We eat, here, the soup.

or follow it,

(4')

Mu titci vi gu, le supta

We eat, here, the soup.

at the speaker's option. (There are reasons for both practices; this usage has not stabilized yet.)

Another place where gu might be needed is after a linked argument inside a predicate string. This is called "internal specification":

(5)

Da kukra je lo litla gu, grobou [groh-BOH-oo]

X is a faster-than-light ship (big-boat).

We do not allow a comma pause to achieve this result: in the 2013 reimplementation, a comma pause is a mere free modifier, not a punctuator. The terminator gue would also work here, by closing off the je-phrase.

Notice that no gu is required to close off internal specifications when the last element in the specification has a firm right-hand end:

(6)

Da kukra je de grobou

X is a faster-than-Y ship.

When the last element in such a specification is a name:

(7)

Da kukra je la Apolo'n, grobou

X is a faster-than-Apollo ship.

Gu would not be wrong here, just redundant. The right-hand edge of je la Apolo'n is already firm. It needs no further marking. In other words, loglanists seldom use more punctuation than is needed to keep their right-hand boundaries clear.

Another use of gu is to close up lepo-clauses to prevent them from swallowing their sequelae. These are often other lepo-clauses:

(8)

Mi pa rulkao [ROOL-kow] lepo santi, gu lepo helba la Bab

I was obliged (rule-acted) to be silent in order to help Bob.

Here the third place of rulkao ('...should/is obliged to do...in order to...') is the looked-for result of acting in an obligatory way. With gu in place, lepo helba la Bab occupies that third place. But without it, the helba-clause will occupy the second place of santi ('is quieter/more silent than') and produce the peculiar claim:

(9)

Mi pa rulkao lepo santi lepo helba la Bab

I was obliged to be more silent than helping Bob (was).

This could make sense. But it is likely to be nonsense in the situations in which (8) is intended. At least it is clear nonsense.

Here's another pair of lepo-clauses that needs careful punctuating:

(10)

Lo nimla ga cnida lopo cluva, gu lopo clivi

Animals need to love in order to live.

(Cnida = '...needs...for purpose/outcome...') Dropping this gu causes the lopo clivi clause to be incorporated in the lopo cluva one...and this will happen even with a comma:

(11)

Lo nimla ga cnida lopo cluva, lopo clivi

Animals need to love living.

Pause-commas are quite incapable of closing-off lepo-clauses. If a right-boundary mark is to be supplied, it must be either gu or (preferably) guo.

Here's a somewhat different case involving lepo-clauses. In the following specimen a clause modifying the main sentence would be engulfed by an earlier lepo-clause if the earlier one were not terminated by gu:

(12)

Da pa djano lepo ti fa crina guo, pa lepo le neri drida fa felda

X knew it was going to rain here, (and X knew this) before the first drops fell.

(The predicate crina means '...is rained on by (cloud)...' So Ti crina means 'It rains here', i.e., on this place.) If, in this situation, we omit gu, we get a sentence which is almost impossible to think about:

(13)

Da pa djano lepo ti fa crina, pa lepo le neri drida fa felda

X knew it was going to rain here before the first drops fell.

The Loglan of (13) says that X knew that it-was-going-to-rain-before-it-rained. Such claims are nearly impossible to entertain... or, what amounts to the same thing, to believe are being made. But we know that this is the claim of Loglan (13) for in it pa lepo le neri drida fa felda ('before the first drops fell') is plainly a modifier of the sentence ti fa crina ('it was going to rain here'). Like the kindred notion that someone Y was born before Y was born, such "knowledge" shakes our faith in the universe...or would do if we believed for a moment that anyone possessed it. But what we do with sentences like English (13) is interpret them "sensibly", that is, as if they said (12). In fact, it is almost impossible to speak the English sentences (12) and (13) in ways that even faintly suggest the remarkable difference that will be so plain in the Loglan once we understand it. But our minds want to stay in their comfortable plausibility tracks. So to show what Loglan (13) actually does mean we are obliged to construct another Loglan sentence, one with exactly the same grammatical structure as (13), but this time with a plausible meaning:

(14)

Da pa djano lepo ti fa nicycri [NEESH-uh-shree] pa lepo le neri clife fa felda

X knew it was going to snow here (snow-rain) before the first leaves fell.

Now pa lepo le neri clife fa felda is firmly (and plausibly) part of the larger lepo-clause about ti fa nicycri. This is knowledge we can believe someone actually might have...unlike the grammatically parallel claim of (13), which we can hardly entertain anyone's making.

Another type of structure which frequently requires closing with gu or its kin (in this case guu) is that of the connected predicate. The occasion for closing off such a structure with gu arises when a speaker has spoken a string of two or more connected predicates, and what da then has to say applies to all of them. For example,

(15)

Da fundi, e norfundi guu, la Pit

X likes, and dislikes, Pete.

(Again, JCB used gu here, but we regard use of guu here as stylistically mandatory.)

Because guu turns la Pit in this sentence into the joint argument of both predicates, we may infer both that X likes Pete and that X dislikes Pete...difficult to do, perhaps, but not impossible. The point is that without guu only the second of these inferences is allowable:

(16)

Da fundi, e norfundi la Pit

X likes (someone or something) and dislikes Pete.

To see that this is the only correct interpretation of (16), let us examine the following series of sentences. Some are punctuated by guu and some are unpunctuated:

(17)

Da fundi lo grato, e norfundi la Pit

X likes cake and dislikes Pete.

(18)

Da fundi lo grato, e norfundi guu la Pit

X likes cake more than, and dislikes, Pete.

Here we are suggesting that a joint argument may be a third argument of one connected predicate (fundi = 'likes...more than...') and, at the same time, the second argument of another (norfundi = 'dislikes...more than...'). One is not required to speak or write in this way, of course, but it is clear that one may do so...in both languages.

(19)

Da fundi lo grato, e norfundi la Pit, guu lo gutra sinma

X likes cake more than, and dislikes Pete more than, foreign movies.

Now the two predicate expressions are coordinate again. This time their joint argument, lo gutra sinma, is the third argument of both. Finally, we remove guu again and produce:

(20)

Da fundi lo grato, e norfundi la Pit, lo gutra sinma

X likes cake, and dislikes Pete more than foreign movies.

In other words, guu's appearance anywhere in the sequel of a set of connected predicates cuts off the argument and modifier sets of all the individual predicates as developed up to that point. Then whatever follows guu is taken as augmenting each of those separate developments. For example, by closing (20) with guu we may then add a joint modifier:

(21)

Da fundi lo grato, e norfundi la Pit, lo gutra sinma, guu vi levi telfoa

X likes cake, and dislikes Pete more than foreign movies, (and both take place) on this planet.

Without guu, each predicate in a connected set of predicates develops its own set of arguments and modifiers independently:

(22)

Da fundi lo grato vi leva telfoa, e norfundi la Pit, lo gutra sinma vi levi telfoa

X likes cake on that planet, and dislikes Pete more than foreign movies on this planet.

So the loglanist does not take for granted that someone who says (16) "really means" (15), as we do so often--and so forgivingly--in English. The price of saying what one means is so small in Loglan--often just the effort to write or speak the little word gu (or as in this case guu)--that most of us are willing to pay it. So in general, we loglanists make the extra effort to punctuate even our informal speech precisely. One consequence of this is that when we do misspeak we expect to be misunderstood...or, more precisely, we expect that what we said will be taken to have been meant. For we are speaking a language in which, unlike the natural ones, one can always say what one means. There thus arises among loglanists an obligation to mean what one says.

A final use of gu-like words (in this case gue). They may be used to effect the attachment of argument modifiers to larger and more distant units than the nearby small ones to which attachment is automatically made in the absence of a gu(V). For example, we saw this sentence in the previous section:

(23)

Le brudi je la Djan, gue ji le ditca, ga gudbi mrenu

The brother of John--the brother who's the teacher--is a good man.

Here, because of the use of gue to close the je-clause, the identifier ji will attach itself to the whole linked argument Le brudi je la Djan. Without gue, of course, ji will link the modifier to the last-mentioned argument, namely la Djan. Unfortunately, as the 1989 Loglan grammar was written, one gu wouldn't do it: one gu would close la Djan; only a second one would close the je link so that the subordinate clause would attach correctly.^R11

So (23) is a case of gu-type words being used to increase what might be called the "grammatical size" of the modificand to which an argument modifier is being applied. These "backward" references, while often intended, are difficult to convey with any accuracy in natural language, and so require ad hoc interventions...of which the awkwardly constructed English sentence in (23) is an example.

5.22 Abnormal Word Order with ga gi goi gio

These last four punctuators permit non-standard, i.e., non-SVO ("Subject-Verb-Object(s)"), word orders to be expressed in Loglan. Ga [gah] we have seen before; it is the "time-free" predicate marker introduced in Section 4.4. Here it will be used to help express sentences of VSO and VOS ("Verb-Subject-Object" and "Verb-Object-Subject") word-orders...in short, all those abnormal word-orders in which the predicate is spoken first. Another of these abnormalizing punctuators is gi [gee]. It might be called the fronting operator because it permits sentences to be spoken in the OSV and OVS word-orders, that is, when the sutori argument(s) of a sentence is (are) to be spoken first. Goi [goy] is really just a variety of gi that is used to mark off "sentence quantifiers". These are a special kind of logical construction which we will study in detail in Section 5.25.

(1)

La Mini,a'polis, la Seint Pol, gi da pa godzi

(To) Minneapolis (from) Saint Paul, X went.

(2)

Lopo clivi gi, la Djan, pa takna la Meris

(About) Life, John talked to Mary.

(3)

La Meris, lopo clivi gi, la Djan, pa takna

(To) Mary (about) life, John talked.

The original proposal was that when gi is being used, the argument that is in the last place of the standard, or dictionary, place-structure of the main predicate will always be mentioned; for gi always follows this argument. Unless the gi-sentence is an imperative--in which case the predicate will immediately follow gi--the normally first argument, too, will be spoken. So what the gi-sentence uniquely allows is leaving out the middle arguments of sutera-order predicates in incomplete utterances. We are very skeptical of this proposal, because one may not know all the sutori places of a predicate adorned with many arguments. We propose that the place of a first argument after gi can alternatively be specified using one of the numerical case tags zua, zue, zui, zuo, zuu, and the places of untagged arguments following can then be read in succession. In the examples that follow, we follow the original intention:

(4)

La Seint Pol, gi da pa godzi

(From) Saint Paul, X went (to somewhere).

(5)

Lopo clivi, gi la Djan, pa takna

(About) Life, John talked (to someone).

Imperative forms may also be spoken in this gi-inverted way:

(6)

Lopo clivi gi, takna

(About) Life, talk.

In Loglan-normal word-order, it is only the last elements in the argument-string that one is able to omit:

(7)

Da pa godzi la Mini,a'polis

X went to Minneapolis (from somewhere).

(8)

Takna la Meris

Talk to Mary (about something).

So gi increases the incompletion options of the speaker.

Goi is grammatically similar to gi in that it marks off a set of one or more argument-like entities that are always spoken first in the sentence. But the goi-entities are not really arguments at all but special argument-like constructions which logicians call sentence quantifiers:

(9)

Raba goi, ba samto ba

For every something x, x is the same as x.

The expression Raba goi is such a quantifier. We will study the uses of such constructions in Section 5.24.

Now let us consider the two predicate-first abnormal word-orders that are made with ga. These, recall, are VSO and VOS. Ga has two roles in such sentences: (a) The first instance of ga--or any other tense operator, for that matter--indicates that the predicate it just precedes, although initial in the sentence, is in the declarative mood. This advises the listener or reader that a deferred subject will be coming up. (b) The second instance of ga--and this one must really be ga, and not some other tense-operator, obviously--will identify that deferred subject when it does come up. So the two punctuation schemata for VSO and VOS sentences in Loglan are gaVgaSO and gaVOgaS. In each case ga marks both an initially-used predicate and a deferred subject.

Of the two predicate-initial word-orders, the VOS word-order is by far the most common in English...and therefore in Loglan translations from English as well. Here's an example of one:

(10)

Ga groda loe damlandi gotca, ga loe monca gotca

Larger than the typical lowland goat is the typical mountain goat.

Ga groda ('Larger than...') is the predicate. The fact that it comes first and is marked--in this case with ga, but any tense operator would do--tells the listener that it's not an imperative da's listening to, and that a subject will be coming later. Da also knows that that subject will be marked with ga. In this case an "object" (loe damlandi gotca) intervenes. But the listener is patient. Da waits for the second ga.

The subject could have come right after the verb, of course; and that would have generated the VSO word-order:

(11)

Ga groda ga loe monca gotca loe damlandi gotca

Larger is the typical mountain goat than the typical lowland goat

Notice that in both these sentences the deferred subject marker ga may be translated by the English copula 'is'. In fact, in both its predicate- and its argument-marking roles, ga often serves the clarifying function of the English copula.

Here is a more intricate example of a VOS sentence:

(12)

Ga seidjo [SAY-joh] lue no nu trecymou [tresh-uh-MOH-oo] bekti ji vi lo rardza [RAHRD-zah], ga lea suphernovi [soop-hehr-NOH-vee]

Among (included in the set of) the most interesting objects in the universe (all existing things) are the supernovae.

If, forgetfully, we neglect to speak the two ga's, we get an unintended and rather mysterious imperative:

(13)

Seidjo lue no nu trecymou bekti ji vi lo rardza, lea suphernovi

Be a member of the set of the most interesting objects in the universe (and relate them somehow to) the supernovae.

For without the two abnormal order markers we have a dangling argument, lea suphernovi, which the listener will not know how to fit into what is apparently an instruction, since the main predicate, seidjo ('set-member'), normally has only two places: '...is a member/element of set...'

This shows why both markers are required to speak a declarative sentence in Loglan in which the predicate comes first and the first argument is deferred...a more common occurrence in literary English than one might suppose.

In sum, both gi and ga are used to shift arguments that are unmarked in the Loglan-normal word-order into new positions...presumably positions that will better correspond to the natural order of some sentence being translated. Any of the six possible word orders--the six permutations of S, V, and O--is speakable in Loglan.²² So the ga/gi system of creating non-standard word orders to match the world's variety of sentence forms is, in a sense, the functional equivalent of the optional case-tags described in Section 4.31. With either system any word-order whatever may be achieved.

The delayed subject is optional. A sentence with no subject and a tensed predicate is read as a sentence of the previous type with omitted subject, not as an imperative. Na crina means "It is raining", not "Get rained on!".

A sentence with only modifiers appearing before an untensed predicate, on the other hand, is an imperative, and the grammar catches this.

A sentence with normal SVO order must have exactly one un-case-tagged argument before the predicate. An SOV (or SOV(O)) sentence with more than one un-case-tagged argument before the predicate must have the particle gio appearing between the first un-case-tagged argument and any subsequent ones. The reason for this is that unintended parses are facilitated by the unmarked SOV construction in long texts: if a final argument is left dangling for some reason, it will attach itself as an additional first argument to the next sentence!

5.23 Utterances and Their Modifiers: A Summary

In all previous sections of this chapter we have been concerned with utterance forms which were sentence-size or smaller. In the following section we will consider certain larger-scale utterances in which simple utterances occur as parts. These will be "connected utterances" of various kinds. But before turning to this final topic of Loglan grammar, it will repay us to retrace briefly the route by which we have come.

We have learned in this chapter that there are four forms of basic Loglan utterances: (i) answers (Da), (ii) imperatives (Godzi), (iii) predication sentences (Da godzi), and (iv) identity sentences (Da bi de). In addition, these basic forms may be embellished in numerous ways, and the embellishments themselves, when used alone, comprise a set of even simpler utterance forms. Thus there are (v) addresses (hoi Djan) and addressed sentences (hoi Djan, godzi = 'John, go!'); there are (vi) expressions of attitude (Ia), and attitudinally modified sentences (Ia da godzi), and, among these, many of the question-forms of Loglan are to be found (Ei da godzi = 'Does X go?'). Then there are (vii) the specific interrogatives Ie and Hu and kin and the many questions they create (Ie da = 'Which X?' Hu godzi = 'Who goes?' Da he = 'X is what?' Ho da = 'How many X?' Da ha de = 'X how connected to Y?'); (viii) the discursive operators (Kii = 'Clearly') and the sentences they embellish (Kii da godzi); (ix) the utterance sequencers (Nefi) and the utterances they order (Nefi, godzi = 'First, go'); (x) the parenthetic remarks (kie uu, kiu = '(Alas!)') and the sentences they embellish (Da, kie uu, kiu godzi = 'X (Alas!) goes'); (xi) the relative modifiers--which include the temporals Na, the spatials Vi, the modals Lia, and the causals Kou--and the sentences they embellish (Da na godzi = 'X now goes' Vi, da godzi = 'Here X goes' Da godzi lia = 'X goes similarly'); (xii) the relative interrogatives (Nahu) and the sentences they create (Nahu da godzi = 'When did X go?'); (xiii) the relative phrases (Na de = 'At Y') and the sentences they embellish (Da godzi na de = 'X goes at Y'; and finally (xiv) the relative clauses (Na lepo de godzi = 'When Y goes') and the sentences they embellish (Da godzi na lepo de godzi = 'X goes when Y goes'). Finally, we have just seen how (xv) potential ambiguities may be resolved by punctuation (Na lepo de godzi gu, da godzi = 'When Y goes, X goes') and how (xvi) certain unusual word-orders may also be obtained (Ga godzi de, ga da = 'Goes to Y, does X', Di gi, da godzi = 'From Z, X goes'). Standing apart from all these ways of forming, modifying, and punctuating utterances is (xvii) the mechanism by which utterances of any kind may be linked together in sustained discourse with the continuation connective I and its many kin (Ibuo) as in Ia. Ibuo da pa godzi = 'Yes. However, X went.'

In the rest of this chapter, we shall consider how sentences and imperatives may be linked together in two further ways: (a) logically and (b) causally, and finally how they may be (c) negated and (d) quantified with the logical quantifiers with which symbolic logic deals. With this last, and most formidable topic of Loglan grammar, we will complete our account of the grammatical structure of the language.

5.24 Logically Connected Sentences with ica and Kin

We saw in Chapters 3 and 4 how logical connections could be made between arguments and predicates in several ways. The simplest of these ways was with the unmarked series of connectives a, e, o and u and their elaborations (see Sections 3.14 and 4.22). We remarked then that connections made with these words were evidently "afterthoughts" in the sense that they do not have to be planned in advance, although, of course, they may be. The "forethought" method of making connections involved the use of the k-marked series of prefixed (or "Polish") connectives ka, ke, ko, ku and their elaborations, and these were used with the infixes ki or kinoi. Because they lead the pair of elements they connect, such connections do have to be planned in advance (Section 4.23). Finally, there was also a third series of connectives, namely ca, ce, co, cu and their elaborations, and these were used only to form connections among predicate words in a string of such words (Section 3.16). So we already have three series of 14 connectives each, and we shall now add a fourth. It is clear that the apparatus for making logical connections is rather elaborate in Loglan. But this is the price we pay for the machinery we need to make a language logical.

Before adding our final series of connectives, let us first note that the series of marked connectives, with their prefixes ka, ke, ko, ku, etc., and infixes ki and kinoi, may be used for connecting sentences in the forethought mode as well as arguments and predicates; for no uncertainties about what is connected to what can arise within this form. This is because whatever lies between any k-form leading mark and the inner k-marked infix with which (in well-formed speech) it can always be matched is the first connectand in a pair of grammatically similar connectands, and so the second connectand--the one that comes after the k-marked infix--must, therefore, be of the same grammatical type as the first. If the first connectand is an argument, the second must also be an argument; if a predicate, then a predicate; and if the first connectand is a sentence of any kind so also must be the second. Thus in

(1)

Ka da faltaa [fahl-TAH-ah] ki de ke tsitoa ki dupma

Either X is a liar (false-talker), or Y is both a thief and a deceiver.

what lies between ka and the first instance of ki is a sentence (da faltaa). So the second term of this connected-structure cannot be the argument de which is the immediate sequel of that ki but must also be a sentence of some kind. In this case that sentence is the entire remaining portion of the utterance, namely de ke tsitoa ki dupma = 'Y is both a thief and a deceiver.' This happens to be a sentence that contains a k-connected predicate. Had we thought--even for a moment--that

was the structure connected by ka and the first ki, we would have heard the elements da faltaa and de as discordantly dissimilar in grammatical type. This would have caused us--or rather, our parsers--to dismiss

this possibility as ungrammatical. We would then have listened for the rest of the sentence of which de therefore promised to be the first argument; and, in a well-formed sentence, we would always hear one ...as we did hear one in this case. Thus the right-hand boundaries of the second elements of such constructions, although unmarked, are as unambiguously determined as the marked left-boundaries are.²³ So the k-series of connectives may be used with any pair of elements so long as they are both of the same broad grammatical type. It is in this sense that the k-series are "context-free," while each of the other series of connective words of Loglan may occur only in a certain specific context or set of contexts.

Suppose, for example, we tried to use one of the unmarked a, e, o, u- series of connectives already in hand to connect sentences in an afterthought way. We might produce an ungrammatical string of words like the following:

(3)

*Da corta de, e di, a do faltaa

*X is shorter than Y and W or H is a liar.

Why is this ungrammatical? Because without some additional punctuation--or other marking--it is impossible to tell where one sentence ends and the other begins, in both languages. Are we saying that X is shorter than Y and W; or H is a liar? Or are we saying that X is shorter than Y; and either W or H is a liar? In English speech we settle such matters by using a certain distinctive pattern of stress, intonation and pause for each of the two interpretations. In written English we use punctuation to distinguish the sentence connective. But in Loglan, punctuation is spoken. Let us see, then, what we can do to mark one or the other of the two dubious connectives e or a phonemically so as to make one of them an afterthought sentence connective that will be plainly identifiable as such in either speech or writing.

We already have the continuation operator I, the mark of a parsable utterance. So I might be used as well to mark afterthought sentence connectives. Sentences to be connected in this way are not quite finished; but they are in principle parsable, as they would have been finished had the speaker not thought of this connection. Moreover, I is always accompanied by a substantial pause; and, in a sense, continuation is itself an afterthought mode of connection: one pauses, one thinks of something else, then one continues. Usually the logical import of the I-connection is the conjunction of some new assertion with all the old ones one has already spoken. But we need to be able to make implications, alternations, equivalences and independencies, too, in this same pause-before-continuing style of utterance connection.

But if we were simply to speak I before one of the unmarked a, e, o, u-words, we would soon be unwittingly speaking one of the attitude indicators of Section 5.5 whenever we did so. Thus I + a would soon become ia; I + e would become ie; I + o would become io, and so on. So, to prevent the production of unintended diphthongs, we put a consonant between the two vowels. There are several consonants that would perform this function adequately, but the one we use is c, a sound which is already associated with logical connection through its employment in the ca-series. So the last series of connectives we require, namely those which are used to connect sentences in the afterthought mode, is, in effect, formed by prefixing lower case i- to any member of the ca-series. Thus ica, ice, ico, icu, inoca, icanoi, [ee-shah, ee-sheh, ee-shoh, ee-shoo, ee-no-shah, ee-shah-noy] etc., are the marked connectives we require.²⁴ Using ice and ica to express the two interpretations which we noted for (3) above, we can now produce two distinct sentences, each with a clear meaning:

(4)

Da corta de, ice di, a do faltaa

X is shorter than Y; and Z or W is a liar.

(5)

Da corta de, e di, ica do faltaa

X is shorter than Y and Z; or W is a liar.

Thus, by choosing to adorn exactly one of the two connectives in (3) with ic-, we can speak one or the other of its two possible meanings quite unambiguously. We will sometimes refer to this new, and final, series of connectives as "eesheks".

What, then, is the difference in meaning between the sentence marked with ice ('and') and the same sentence marked with the utterance continuer I? Let's replace ice with I in (4) and see what happens:

(6)

Da corta de. I di, a do faltaa

X is shorter than Y. Z or W is a liar.

In both languages the sense of (6) is that these are two sentences separated by a full stop. The speech could have ended before the I but didn't. This contrasts with (4) in which, although there is a pause before ice, it is probably a very brief one because the speaker is obviously rushing on to give more information. Like I, the logical sense of ice is to express conjunction: the assertion of both claims. But ice does not launch the parse as I and its kin, the speech-continuing discursives like Ibuo, do. This is because an eeshek-connected string of utterances sometimes requires that later utterances in the string be understood before the sense of earlier ones can be fully taken in. This is particularly true when the eeshek is one that involves negation:

(7)

Mi fa stolo, icanoi da kamla

I will stay if X comes.

Notice that in English we do not punctuate the 'if'-clause. This is probably because the two clauses are so nearly a single idea. Thus even though Mi fa stolo is potentially a finished utterance, it would be a mistake to parse it too quickly. In this sentence, as the upcoming icanoi quickly shows, Mi fa stolo is not being asserted by the speaker but being made contingent on a condition which may or may not be realized.

In Loglan we must pause before the eeshek, as we do before all connections in this logical language; but we do not mark the eeshek-preceding pause with a period followed by a capital letter in text. Eesheks may, on occasion, initiate the first or even the only utterance in a speech. For example, Icanoi da kamla may be an answer to a question like Ei tu fa stolo = 'Will you be staying?' In such cases the initial i of the eeshek is always capitalized.

We might wish to make some logical use of the distinction between the I-words composed of I and its true semantic kin, the I-initial discursives, and the eesheks, saying for example, that the appearance of an I-word in a speech terminates the scopes of all preceding "sentence quantifiers" while an eeshek does not. (Sentence quantifiers are the topic of the following section.) But this is a matter of logic, not grammar, and is in any case still under study.

Afterthought sentence connections may, of course, be combined with forethought sentence connections in the same utterance. The afterthought may occur within the first term of a k-connected pair:

(8)

Kanoi da fa kamla, ica de fa godzi, ki mi fa stolo

If (X comes...or Y goes), then I will stay.

(I've used parentheses in the English to show the grouping, and triple dots to mark the point at which the afterthought occurs.) Or the afterthought may follow the kekked pair, in which case the original thought must be grammatically complete before the afterthought clause is spoken:

(9)

Ka da fa kamla ki de fa godzi, inoca mi fa stolo

(Either X comes or Y goes)...only if I stay.

Actually, sentences (8) and (9) make, in the end, the same claim...although they suggest that their speakers were in very different cognitive states as they were spoken. But as the parentheses show, both are left-associative; and the fact that the hesitations occur in different places does not alter their logical structure at all. In fact, the claim made with two hesitations, that is, with two afterthought connections,

(10)

Da fa kamla, ica de fa godzi, inoca mi fa stolo

(X will come...or Y will go)...only if I stay.

also makes the claim of (8) and (9).

So, like all afterthought connectives, the eesheks are left-associative. This means that it is the entire preceding portion of the utterance that is the antecedent term of the implication made with inoca in both sentences (9) and (10). These are thus equivalent to the same claim made forethoughtfully with kanoi...ki in (8).

The right-associated connection among these same three elements makes, of course, a different claim. The new claim may be expressed in Loglan only by kekking the last two elements:

(11)

Da fa kamla, ica kanoi de fa godzi ki mi fa stolo

X will come...or, if Y goes, then I will stay.

Of course this same right-associated claim may also be made with two forethought connections as follows:

(12)

Ka da fa kamla ki, kanoi de fa godzi ki mi fa stolo

Either X will come or, if Y goes, then I will stay.

The first kek used in (12), namely ka...ki, is of course redundant in the sense that a simpler eeshek could do the same connecting work...as it actually does in (11). Yet, because (12) is completely pauseless, it can probably be more rapidly spoken than (11). This may cause its redundance to be tolerated and the form itself to be preferred when no genuine afterthought is involved.

Finally, in anticipation of the matters to be dealt with in the next two sections, we must mention that kekked sentences, unlike eeshekked ones, carry the scope of any outside negative or "sentence quantifier" over the entire ki-connected structure. Thus,

(13)

No, kanoi da fa kamla ki mi fa stolo

It is not the case that if X comes then I will stay.

is the contradiction of the whole connected sentence-pair made with kanoi...ki; whereas the negative in

(14)

No, da fa kamla, inoca mi fa stolo

It is not the case that X will come...only if I'll stay.

applies only to the first of the two connected sentences. This follows, of course, from our decision to regard any sentence that precedes an eeshek as at least grammatically complete. Sentences (13) and (14) make, therefore, very different claims. In (13), the speaker Y is denying that Y's staying is conditional on X's coming. It turns out that this is equivalent to asserting unconditionally both that X will come and that the speaker Y will not stay.²⁵ How different this is from the claim of (14), in which the speaker de is saying that de's staying is conditional, but on X's not coming! Matters of scope are, therefore, essential to the interpretation of such sentence-long operations as quantification and negation, which are the subjects of the two following sections. In part such questions are settled in Loglan by the speaker's choosing between the forethought and the afterthought modes of making connections among his ideas.

5.25 Quantified Sentences

When we were considering quantified arguments in Section 4.22 we remarked that speaking a sentence like

(1)

Levi so mrenu pa kamla

These six men came.

is equivalent to speaking the same sentence without the quantifier,

(2)

Levi mrenu pa kamla

This man came.

six times. The number-word so is, in effect, a sentence quantifier. In any sentence containing a number-word n used as a quantifier, n specifies the number of copies of that sentence with n removed from it to which the original sentence is equivalent. We will now extend this apparatus in a logically very powerful direction. We will show how all the "universal" and "existential" sentences made with the non-designating variables ba, be, bo and bu of Section 4.21 are, in fact, implicitly quantified sentences. Our task in this section will be to show how all such quantifiers may be explicitly expressed when desired. In rendering sentence quantifiers explicit, we shall be imitating a style of Western speech normally used only by logicians. But we shall also show that, in all but the most intricate cases, the far simpler everyday speech style of implicit quantification suffices.²⁸

Consider the innocent-appearing claim

(3)

Ba brano

Something x is bread (There is some bread).

Readers familiar with symbolic logic know that for manipulative purposes it is often useful to express this existential claim explicitly:

(4) There is at least one x such that x is bread.

In this more elaborate expression the phrase 'There is at least one x such that...' is called the existential quantifier; and it is expressed symbolically, in the notation favored by most logicians, by putting upside-down 'E' before the variable to be quantified and placing the whole in parentheses: '(x)'. We need not use this notation in this book--except to make technical comments in the notes--for the Loglan verbal expression conveying the same idea is almost as compact. To form a Loglan existential quantifier we simply produce a copy of the variable to be quantified, mark it off with the special punctuator goi [goy], and speak or write it at the head of the sentence:

(5)

Ba goi, ba brano

There is at least one x such that x is bread.

In this sentence, the Loglan phrase Ba goi is an explicit existential quantifier. It means exactly what 'There is at least one x such that...' means when a logician is talking English.

But what is the meaning of such a claim? And in what sense is such a sentence "quantified"? From our discussion of non-designating variables in Section 4.21, the reader may recall that any sentence involving a claim about the existence of some undesignated someone or something is equivalent to an infinite number of demonstrative sentences such that if any one or more of them is true the sentence with the existential claim is also true. Thus sentence (5) is equivalent to

(6)

Ti brano

This is bread.

spoken an infinite number of times in the presence of an infinite number of objects. And just in case at least one of that infinite set of sentences is true of at least one of those objects, sentence (5) is also true. For all (5) says is that there is at least one object x somewhere such that 'x is bread' is true of that x. So sentence (5), and therefore (3), is quantified...and in a rather magnificent way. For we are apparently leaving the modest economies of numerical quantification, whereby sentence (1), for example, is worth six of sentence (2), and entering a new linguistic domain in which innocent-looking sentences like Ba brano turn out to be equivalent to infinite sets of sentences, and which make claims, therefore, which may not be practically encompassed in any other way. Note that both Ba brano and Ba goi, ba brano make the same vast existential claim; and that, in the first, the existential quantifier, while not present, is implicitly invoked.

Suppose there are several non-designating variables in a sentence. Rendering the sentence quantifiers explicit in that case means speaking or writing copies of all those variables at the head of the sentence and marking the last of them with goi:

(7)

Ba pa bloda be

Someone x hit someone y.

(8)

Ba be goi, ba pa bloda be

There is an x (and) a y such that x hit y.

Or with three non-designating variables:

(9)

Ba pa vedma be la Djek, bo

Someone x sold something y to Jack for some price z.

(10)

Ba be bo goi, ba pa vedma be la Djek, bo

There is an x, y (and) z such that x sold y to Jack for z.

Notice that the variables in the string of sentence quantifiers have been written in the same order as the order of their appearance as arguments of the original sentence. This is mandatory. We shall examine some of the consequences of this rule in a moment.

What about sentences which make universal claims? We learned in Section 4.21 that these are formed by quantifying a non-designating variable with ra ('all'). For example, the universally quantified argument Raba appears in:

(11)

Raba cluva la Espanias, anoi la Sol

Every someone x loves Spain if the Sun. (All who love the Sun, love Spain.)

Here, too, a sentence quantifier is implicitly invoked, this time the universal quantifier Raba goi, which, like the existential Ba goi, we may also write or speak explicitly at the head of the sentence. When we do so we strip the original argument of its quantifier ra, for it is now redundant:

(12)

Raba goi, ba cluva la Espanias, anoi la Sol

For every x, x loves Spain if (x loves) the Sun.

The phrase 'For every x' is the usual reading of the universal quantifier in logician's English; but it is usually written in text as the expression '(x)'. The Loglan verbal expression Raba goi is not quite so compact. But note that logicians use the unmarked form '(x)' for the universal and the longer marked form '(x)' for the existential while in Loglan we evidently prefer to let the existential be the unmarked form: ba as opposed to raba. This is because, in speech, existential quantification occurs far more frequently than universal quantification. The former appears implicitly in English, for example, in nearly every use of the indefinite article 'a' outside of predicates. But in the claims that traditionally interest logicians, universals may be the more common form.

Let us now mix these two types of quantifiers in the same sentences. We will do so first implicitly and in two different orders:

(13)

Raba nu cluva be

Every someone x is loved by someone y. (Everyone has a lover.)

(14)

Be cluva raba

Someone y loves every someone x. (Everyone has the same lover.)

And then explicitly in the same two orders:

(15)

Raba be goi, ba nu cluva be

For every x there is a y such that x is loved by y.

(16)

Be raba goi, be cluva ba

There is a y such that, for every x, y loves x.

(Notice that the /ra/ of raba is often stressed when it is part of an explicit quantifier followed by goi, but that raba elsewhere usually has level stress.)

These last two explicit forms make abundantly clear that the order in which the two quantifiers raba and be appear makes a very considerable difference, whatever one may think about the implicit formulations. But since (13) is strictly equivalent to (15), and (14) to (16)--the second in each pair being the rendering explicit or the "explicitation" of the first--it is clear that (13) and (14) are not permissible transforms of one another. We must apparently impose a restriction on the operation of conversion which we used so freely in Section 4.8. That restriction is this. Predicates may only be converted, and their arguments accordingly re-ordered, when the set of arguments either (a) includes no non-designative arguments, or (b) such non- designative arguments as do occur in it are all (i) of the same quantificational type (universal or existential) and also (ii) of the same sign (negative or positive). Conversion under any other circumstances than these will alter the original claim.

With this caveat in mind we now see that the basic sentence to which the explicit quantifiers are attached in (15) above, namely ba nu cluva be, may be legitimately re-converted since its two non-designative arguments are both of the same type and sign. Whence we may eliminate nu from sentence (15) and rewrite it as:

(17)

Raba be goi, be cluva ba

For every x there is a y such that y loves x.

And now the only difference between (16) and (17) is the order of the two sentence quantifiers. Thus the difference in sense between the two implicitly quantified original sentences, (13) and (14), apparently depends on our being able, in Loglan, to use a converse predicate (nu cluva) in order to speak the arguments raba and be in an order that will convey the order of the two sentence quantifiers which we implicitly intend. This is the reason why one of the rules for making quantifiers explicit is to write or speak copies of the non-designative arguments at the head of the sentence in the same order as they appeared in the implicitly quantified sentence.²⁹

The same very economical device operates in English. When we say 'Everyone wants something' we do not mean what the "same" sentence means in the passive voice, namely 'Something is wanted by everyone'; though there is a curious figure-ground phenomenon here that allows one to persuade oneself, fleetingly, of the equivalence of the two sentences. Not so in Loglan. Because of the existence of orderly rules by which any implicit quantification may be made explicit, and any explicit claim once suitably expressed be stated with its quantifiers suppressed,³⁰ we expect that the difference in meaning between:

(18)

Raba danza be

Everyone x wants something y.

and

(19)

Ba nu danza rabe

Something x is wanted by everyone y.

will always and everywhere be plain.

If every explicitly quantified sentence may indeed be formulated in such a way that its quantifiers may be suppressed, why use explicitly quantified sentence forms at all? There are two reasons. The first is that there are certain very useful logical transformations which may most easily be performed on quantifier strings, some of which we will apply to strings containing negatives in the next section. It is our hypothesis that, in Loglan, thoughtful speakers will be able to perform at least some of these operations on the speech-flow itself, without resorting to the pencil-and-paper techniques of logicians, simply because, in Loglan, the quantifier string is a regular flow of short, unitary words. The second reason is that there are certain connected sentence forms in which the quantifiers have scopes which cannot be expressed implicitly. While such sentences may always be transformed into other sentences whose quantifiers may be implicitly expressed, if we want to use such quantifiers in speech, then we must mention them explicitly. As an example of the latter phenomenon, let us be more extravagant about loving Spain:

(20)

Raba cluva be, noa la Espanias

Anyone x loves something y only if (x loves) Spain.

The more usual way of expressing this idea in English is to say, 'If anyone loves anything, then he loves Spain.' But that last sentence invites quite a different Loglan translation:

(21)

Raba cluva rabe, noa la Espanias

Anyone x loves everything y only if (x loves) Spain.

which obviously does not say the same thing. We will see in a moment how sentence (20) may be transformed into one with two universal quantifiers, but this, obviously, is not the way to do it. To see what is going on here let us first make the quantifiers of these two sentences explicit. To do so we adopt the convention that the scope of any implicit quantifier will be taken to run over all and only those clauses of any connected sentence in which its variable appears. Raba in both sentences has, by this convention, sentence-long scope; but be, in (20), and rabe, in (21), are both confined to the first of the two implicitly connected clauses that could be made explicit by expanding these last two utterances into connected sentences. For example, (20) could be expanded into Raba cluva be, inoca raba cluva la Espanias. It is clear, once (20) is expanded, that the scope of the implicit existential quantifier be is, by this convention, confined to the clause Raba cluva be, while the scope of Raba extends over both clauses. We may convey this difference in scope even in the compact form of the sentence (in which only the second arguments are connected) by using two goi's in the explicit quantifier string:

(22)

Raba goi, be goi, ba cluva be, noa la Espanias

*For every x such that there is a y such that x loves y only if (x loves) Spain.

(23)

Raba goi, rabe goi, ba cluva be, noa la Espanias

*For every x such that for every y such that x loves y, only if (x loves) Spain.

(I cannot produce a grammatical English sentence using the English afterthought form 'only if' as translations of (22) and (23); so I've starred them, and included parenthetically other English sentences which are grammatical, and which make the same claims as the Loglan ones, but in which the English connective is the forethought form 'if...then'.) To make sense out of these over-compact Loglan remarks, let us expand the connected arguments of these sentences into clauses. To do so we will use the marked connective kanoi... ki...--which is equivalent to ...noa...--in order to carry the scope of Raba over both clauses:

(24)

Raba kanoi be goi, ba cluva be ki ba cluva la Espanias

For every x, if there is a y such that x loves y, then x loves Spain.

(25)

Raba kanoi rabe goi, ba cluva be ki ba cluva la Espanias

For every x, if, for every y, x loves y, then x loves Spain.

These formulations make very clear that the scopes of the two inside quantifiers, be goi and rabe goi, are confined to the clauses they immediately precede, as our convention for interpreting implicit quantifiers requires. But now let us say with explicit quantifiers what cannot be said with implicit ones in Loglan. Let us move the inner quantifiers of sentences (24) and (25) ahead of the kekking prefix kanoi so that the quantifiers be and rabe will then have sentence-long scope explicitly despite the fact that their variables do not appear in the second clauses:

(26)

Raba be kanoi ba cluva be ki ba cluva la Espanias

For every x there is a y such that, if x loves y, then x loves Spain.

(27)

Raba rabe kanoi ba cluva be ki ba cluva la Espanias

For every x and for every y, if x loves y, then x loves Spain.

Now sentence (27) claims that, for endless pairs of x's and y's, if the first, x, loves the second, y, then the lover x will also be found to love Spain. But this is exactly what we claimed with the short-scope existential in (24), namely that if anyone loved something, then ba loved Spain! Sentences (24) and (27) do indeed make the same claim; and it is (27), with its two long-scope universals--one of which cannot be implicitly expressed--which best conveys the most usual English formulation of this idea, namely 'If anyone loves anything, then da loves Spain.'

But what does (26) mean? It happens that, just as (27) is equivalent to (24), so (26) is equivalent to (25). Sentence (25), recall, makes the trivial claim that anyone who loves everything (of course!) loves Spain. It is a bit difficult to see that sentence (26) also expresses this same banal truth. But the equivalence of the two forms can, perhaps, be suggested by considering that, in a universe in which all but one individual do not love everything, there will, for all those other individuals, be at least one something y they do not love; whence the implication of (26) will be true for them by the denial of the antecedent. And in the case of the one individual x who does love everything, the implication of (26) will be true by affirmation of the consequent; for "everything" includes Spain. In this same universe, sentence (25) will, perhaps more obviously, also be true. For the antecedent will be false, and the implication therefore hold, for all those individuals who do not love everything; and the consequent will be true in the case of that one individual who does love everything. It may be time to make clear that the truth of an implication in 'if...then' (kanoi...ki) order is guaranteed either by the falsity of its first term (its "antecedent") or by the truth of its second (its "consequent"). So to know that an implication is true one does not even have to look at its consequent if its antecedent is false, or at its antecedent if its consequent is true.

However this may be, the reader may take it for granted that (26) is equivalent to (25), and (27) to (24). In both cases an inside short-scope quantifier has been moved outside the connecting prefix kanoi, a move which gives it sentence-long scope, while at the same time, its quantificational type has been changed. It is this move, among others, which allows us to transform any expression which cannot be directly expressed implicitly--such as sentence (26) or (27), above--into one that can be. Thus sentence (20), above, with its short-scope existential, is the implicit expression of the same claim made explicitly in (27) with its two long-scope universals; while sentence (21), with its short-scope universal, is the implicit expression of the claim of (26) with its long-scope existential. This convertibility of quantifiers of one type into quantifiers of opposite type, with suitable modifications of scope or sign, is a recurrent principle in logic. And it is the existence of transformations that allow us to change sentences like (27), with their quantifiers reigning over clauses in which their variables do not appear, into sentences like (20), in which all quantifiers are confined to the clauses in which they do appear, that makes implicit quantification possible.³¹

Sentence quantifiers are not restricted to those formed with non-designating variables and the quantifier ra. For example, we may use a numerical quantifier in an explicit sentence quantifier, as in:

(28)

Neba goi, ba gandi

There is exactly one x such that x is a god. (There is just one god.)

Or we can express the same idea implicitly, as in:

(29)

Neba gandi

Exactly one something x is a god.

Or we may use a designating variable as the basis of the quantification, as in:

(30)

Ra da goi, da terjaofoa, a kurfa

For every X (in the set of X's) X is either a triangle or a square.

Or even a description, as in:

(31)

Re le mrenu goi, mei merji, anoi farfu

For most of (an understood set of) men, m is married if a father.

In this last sentence, mei ('m') is taken to stand for any member of the collection of men intended to be referenced by le mrenu, which may be assumed to be plural because it is quantified by re.^R8

Thus a sentence quantifier may be any indefinitely quantified argument form which reappears, or of which a valid replacement appears, in the sentence to be quantified. But it turns out that quantifiers based on designative arguments may not be implicitly expressed. Notice, for example, what happens to the sense of the sentence (31) when we remove the sentence quantifier and content ourselves with a quantified argument of the same form...hoping to express the same idea implicitly:

(32)

Re le mrenu ga merji, anoi farfu

Most of the men are married if (they are) fathers.

But this sentence now makes a slightly different claim. To see what the new claim is let us expand the connected Loglan argument into a similarly connected pair of Loglan sentences, and then translate the result back into English (because the Loglan is so transparent, this is often a useful technique when we want to see what a connected pair of English sentences is really claiming):

(33)

Re le mrenu ga merji, icanoi re le mrenu ga farfu

Most of the men are married if most of the men are fathers.

By replacing the second instance of le mrenu with either da or mei, this may of course be shortened to:

(34)

Re le mrenu ga merji, icanoi re da farfu

Most of the men are married if most of them are fathers.

Sentences (33) and (34) have clear interpretations in both languages. We now see that (32)-(35) can be false under circumstances under which (31) is plainly true. An example of such circumstances would be when the designated group of men is composed of a married father, an unmarried father, and a childless bachelor. In that case (32)-(35) are false (most are fathers, but most are unmarried), while the implication 'If X is a father then X is married' is true of the childless bachelor (by denial of the antecedent) and of the married father (by affirmation of the consequent) although it is false for the unmarried father. Whence sentences (31), which claim that the implication holds for more than half of them, is true.

Again, it is a matter of scope. The scope of the explicit quantifier in sentences (31) runs over the entire sentence it precedes even when expanded into a pair of clauses. But the scope of each of the quantified arguments of sentences (33) and (34) is, of course, confined to the constituent sentence in which it appears. This will always be true of sentences with designative arguments; whence we are denied the economies of implicit quantification in their case.

The same argument could, of course, be applied to sentences with non-designative arguments. We could, for example, have chosen to say that

(36)

Raba merji, anoi farfu

Every something x is married if a father.

may be properly expanded into the surprising claim that

(37)

Raba merji, icanoi raba farfu

Every something x is married if every some thing x is a father.

But sentences of this form would, if our logical usages allowed them, nearly always be trivially true. Since for any property P it will nearly always be false that everything is a P, the conditional that if every something x is a P (but not all somethings are) then something else is the case (but doesn't have to be) will nearly always be true. Therefore we deny the propriety of the transformation of (36) into (37) on the ground that the common argument of the connected predicates of (36) is of the non-designating variety. Instead we say that the proper expansion of (36) is as follows:

(38)

Raba merji, icanoi ba farfu

Every something x is married if (that same) something x is a father.

In this way we preserve the scope convention of implicit quantification mentioned earlier: namely that the scope of any implicit sentence quantifier is held to carry over just such clauses of a connected sentence in which its variable appears.

Of course we must, in a logical language, also be able to speak such nonsense as appears to be the claim of sentence (37) clearly when we want to. But we cannot do so in Loglan by using a common argument of a connected predicate. Instead, we must speak a connected sentence directly, using two different non-designating variables for the two short-scope universals:

(39)

Raba merji, icanoi rabe farfu

Every something x is married if every some thing y is a father.

And we now have two implicit universal quantifiers--either or both of which may be made explicit--the scope of each of which is confined to its own constituent clause. This arrangement now states clearly the nonsensical claim made unclearly in sentence (37). The reader will have sensed that the entire apparatus of implicit quantification depends on certain restrictions on, and conventional interpretations of, the usages of Loglan, one of which is that sentence quantifiers formed with designative arguments may not be expressed implicitly.

There is a final convention concerning the handling of sentence quantifiers which we have been observing all along and which we may now state openly. And that is that any incomplete sentence should first be completed before any of its quantifiers are made explicit. The reason for this rule is that all incomplete sentences have unexpressed implicit quantifiers, and these will not be among those made explicit if the sentence used as the basis of the explicitation is itself incomplete. Thus a sentence in which some but not all of its implicit quantifiers have been made explicit has a false air of exactness. Suppose, improperly--we will see why this is improper in a moment--we were to render the quantifiers of (38) explicit below:

(40)

Raba goi, ba merji, icanoi ba farfu

For every x, x is married if x is a father.

But as both merji and farfu are multi-place predicates, and as they are both treated here as one-place predicates, we are tempted to think that we have said something exact about marriage and fatherhood (in given societies) when in fact we haven't.

Let us go back to sentence (38), of which (40) is a logically improper (but not ungrammatical) transformation, and complete the argument sets of both its predicates. We recall that we are to supply existentials for the unexpressed arguments of positive predicates and universals for the unexpressed arguments of negative predicates. Both the predicates in (38) are positive; hence (38) may be completed as follows:

(41)

Raba merji be icanoi ba farfu bo bu

Every someone x is married to someone y if (that same) someone x is the father of someone z through some mother w.

Sentence (41), with its complete set of implicit quantifiers, may now properly be made explicit as follows:

(42)

Raba goi, be goi, ba merji be, icanoi, bo bu goi, ba farfu bo bu

For every x there is a y such that x is married to y if there are a z and w such that x is the father of z through mother w.

But having done this we may now ask whether the particular completion we supplied in sentence (41), and rendered explicit in (42), says what we wanted to say. Suppose what we are trying to express with some exactitude is a moral law of some society. (To express any claim as a moral law in which we ourselves believe we need only precede a sentence which asserts that claim with the indicator of strong obligation Oa.) To do so we must ask, Is it sufficient, in that society, that a man be married to someone y in order to have children by perhaps some other someone w? Very likely it is not. But that is what (42) allows with its two short-scope existentials be and bu. So completing a sentence and making its quantifiers explicit are linguistic acts which often force us to raise the question of what we actually mean by our incomplete or implicitly quantified remarks.

If, for example, we mean to say that if a man has a child by anyone y then we shall also find that he is married to that same y, then we have failed to say this with sentence (42)...nor, since one is a regular transform of the other, have we succeeded with sentence (38). Instead we require that the marital and maternal roles of the mother be linked by a long-scope universal quantifier in exactly the same way that the roles of the husband and father are universally linked in sentence (42); but unfortunately this cannot be said incompletely. It can, however, be said with implicit quantifiers, as below:

(43)

Raba merji rabe, icanoi ba farfu bo be

Every someone x is married to every someone y if that someone x is the father of someone z through that someone y.

And the quantifiers of this new sentence may now be stated explicitly as follows:

(44)

Raba rabe goi, ba merji be, icanoi bo goi, ba farfu bo be

For every x and every y, x is married to y if there is a z such that x is the father of z through y.

If we were in doubt about the adequacy of (43) as an accurate expression of our intent, (44) will dispel it. Sentence (44) says what we mean, and so, apparently, does (43). What it says, suitably moralized, is still a fairly broad rule since it permits polygamy; but it does rule out children whose fathers are not married to their mothers. This is, perhaps, the intent, at least, of that version of the human marriage law that is most common on this planet.

So we see that completing a sentence not only has the formal advantage of guaranteeing that any subsequent transformation of its quantifiers will also be complete, but the informal one of forcing us to deal with the exact nature of its claim. Thus (38) says exactly, if incompletely, what (41) and (42) say completely. So the speaker of (38) who thought da was saying something that was an abbreviated equivalent of (43) learns that da was not. Once this awareness is in hand, da may then go on, by adjusting da's quantifiers, to say exactly what da did mean.

We have in this section dealt with what is perhaps the most difficult set of linguistic usages engendered by our project to "make symbolic logic speakable". Yet we have barely scratched the surface. We will leave it to the logically-minded reader to carry the apparatus of implicit and explicit quantification further. Yet other readers will have glimpsed, perhaps, that in these elegant Loglan forms--all of which are far shorter and apparently far plainer than their horrendously intricate English equivalents--we may be on the road, at least, to making some kind of logic speakable. We do not yet know that even this little has been accomplished, of course. For it may be that we have simply compacted the intricacy of these quantificational ideas into denser forms which are even less understandable than their more attenuated English originals. Only experimental tests with authentic speakers of Loglan will settle this interesting question of whether we have, in fact, made logic speakable...or made it even more unspeakable than it was.

5.26 Negative Sentences

We saw in Section 4.24 on negative arguments that the negative operator no could be exported from any internal designative argument to a position at the head of the sentence where it is marked by a pause-comma. In this position and so-marked, no has the sense of the sentence-negating English phrase 'It is not the case that...' We also saw that, before non-designating arguments formed with ba, be or bo, such internal negatives could not be exported. We said that all such negatives are, in a sense, "stuck to" their arguments. Finally, we also noted that no before predicate expressions could be exported to the head of the sentence if and only if the sentence as a whole was (a) complete and (b) contained no non-designating variables or indefinite descriptions. We must now make sense out of all these complicated arrangements.

We hinted that the negative in sentence-negating position is stylistically preferred in Loglan. And so it is. But we can now state the matter more strongly. If a sentence contains non-designating variables or indefinite descriptions, and is, therefore, implicitly quantified, any negatives that come before any of its designating arguments must be exported before that sentence may be explicitly quantified. Let us see why.

Suppose we have said, in that quaintly archaic style using negative arguments that is permissible in Loglan,

(1)

La Djan, pa godzi no la Romas

John went not to Rome.

Since godzi is a three-place predicate here used positively, we may complete it with an existential:

(2)

La Djan, pa godzi no la Romas, ba

John went not to Rome from somewhere x.

This reveals that the sentence is implicitly quantified, something that wasn't plain before. But to make the quantification explicit, we must first detach the internal negative from the negative argument no la Romas and write it at the head of the sentence marked with a pause-comma:

(3)

No, la Djan, pa godzi la Romas, ba

It is not the case that John went to Rome from somewhere x.

This is phonemically distinct from the two-utterance speech,

(4)

No. I la Djan, pa godzi la Romas, ba

No. And John went to Rome from somewhere x.

in which No. is presumably an answer to some preceding question. Speeches (3) and (4) are virtual contradictories in import. So it is good to be attentive to one's I's in Loglan.

Returning to (3), we may now make the quantifier explicit by placing a copy of the non-designating variable between the sentence negative and the sentence it negates and following it with the marker goi:

(5)

No ba goi, la Djan, pa godzi la Romas, ba

It is not the case that there is an x such that John went to Rome from x.

Notice that the pause-comma marking No in (3) may be dropped when, as in (5), No precedes an explicit quantifier. In a sense No becomes part of the quantifier string and shares its final punctuator.

The rule, then, for making the implicit quantifiers of negative sentences explicit is to insert the explicit form between the negative and the sentence it negates; and this incorporates the negative operator into the quantifier string.

Now suppose we had attempted to make the quantifier in (2) explicit before exporting the negative:

(6)

Ba goi, la Djan, pa godzi no la Romas, ba

There is an x such that John went not to Rome from x.

But there is always such an x! Even if John is drawn so magnetically back to Rome that he goes there from Paris, London, Tokyo and DesMoines, Iowa, there will still be an x somewhere--in Antarctica, say--from which John doesn't go to Rome. Sentence (6) is, in fact, equivalent to

(7)

Ba goi, la Djan, no pa godzi la Romas, ba

There is an x such that John didn't go to Rome from x.

With predicates like godzi such sentences as (6) and (7) will always be true.³²

So, to avoid counter-intuitive transformation products like (6) we adopt the rule that the exportation of internal sentence negatives, like the completion of incomplete sentences discussed in Section 5.17, must be accomplished before rendering any implicit quantifier explicit. As always, one must be able to state trivial truths like that of sentence (6) clearly, of course, but in Loglan they must be stated with negative predicates, as in (7), the implicit version of which is:

(8)

La Djan, no pa godzi la Romas, ba

John didn't go to Rome from somewhere x.

Or, perhaps more clearly, with a universal quantifier and a negative argument as in:

(9)

La Djan, pa godzi no la Romas, raba

John went not to Rome from everywhere x.

From the latter, after exporting the negative, we get:

(10)

No, la Djan, pa godzi la Romas, raba

It is not the case that John went to Rome from everywhere x.

And this, once the quantifier has been made explicit, becomes:

(11)

No raba goi, la Djan, pa godzi la Romas, ba

It is not the case that, for every x, John went to Rome from x.

It happens that sentence (11) is transformationally equivalent to (7) by a logical rule called "reversal", which is useful for handling negative quantifiers and which we will discuss more fully toward the end of this section.

So the first rule for handling negative sentences is to recognize that a sentence with an odd number of negative designative arguments is a negative sentence, and should be so expressed before rendering any implicit quantifiers explicit. The second rule is that the explicit quantifiers of negative sentences must be inserted between the negative and its sentence.

But now let us consider the handling of negatives which are bound to non-designative arguments. Take the sentence

(12)

Ba corta no be

Something x is shorter than no something y. (There is a longest something.)

Such negatives may not be exported. Clearly sentence (12) is not equivalent to:

(13)

No, ba corta be

It is not the case that something x is shorter than something y. (Everything is the same length.)

which we would have obtained by incorrectly exporting the negative of (12). But to make the implicit quantifiers of sentences like (12) properly explicit, we move a copy of the entire argument, negative included, into the quantifier string leaving only a copy of the variable itself in the body of the sentence which is thus quantified:

(14)

Ba no be goi, ba corta be

There is an x (such that) there is no y such that x is shorter than y.

Similarly, the implicit negative quantifier of (13) may be rendered explicit by:

(15)

No ba be goi, ba corta be

There is no x (such that) there is a y such that x is shorter than y.

Unsurprisingly, the order in which negatives appear in a string of quantifiers makes some difference in both languages.

Let us consider a more intricate case:

(16)

Mi farfu no raba no be

I am the father of not everyone x through no someone y.

What have we said? So many negatives make it puzzling. So we first make the quantifiers explicit, making sure that we retain their order, and we get:

(17)

No raba no be goi, mi farfu ba be

Not for every x is there no y such that I am the father of x through y.

from which, after a moment of stubborn reflection, we might now gather that I am the father of someone! For, by a very useful logical rule for eliminating negatives in the quantifier string, we may rewrite (17) as below:

(18)

Ba be goi, mi farfu ba be

There is an x such that there is a y such that I am the father of x through y.

The rule that accomplishes the remarkable transformation of (17) into (18) states that any portion of an explicit quantifier string that reads ...no raba no... may be replaced by ...ba..., and any portion that reads ...no ba no... may be replaced by ...raba..., for any non-designating variable ba. In short, any quantifier surrounded by negatives may be replaced by a quantifier of opposite type after eliminating those negatives. This is a very convenient rule, often known as quantifier negation equivalence. In this case it could have been applied to sentence (16) directly. This would have given us the implicit version of sentence (18) immediately:

(19)

Mi farfu ba be

I am the father of someone x through someone y.

The English version of sentence (16) is almost impossible to understand no matter how carefully we try to phrase it...and so, probably, is the Loglan. It is not until the quantifiers in the English sentence have been rendered explicit, as in (17), that we begin to glimpse its claim. But again it is our hypothesis that Loglan speakers will be able to make these and other transformations directly on the speech-flow without going through the intermediate step of explicit quantification. After all, the string of sutori (second and subsequent) arguments in sentence (16) is identical to the string of explicit quantifiers in sentence (17); and, in each of them, any element surrounded by negatives is immediately replaceable by an element of opposite type. We surmise that this will be as easy to hear in Loglan as the double negative ('He didn't not come') is easy to hear--and eliminate--in English.

On the other hand, the rule that permits us to eliminate any pair of no's which surround explicit quantifiers does not apply to implicitly quantified negative first arguments if any subsequent arguments are non-designating. Thus in

(20)

No raba no farfu be rabo

Not everyone x is a non-father of someone y by everyone z.

the expression No raba no may not be replaced by Ba. The reason is that in the corresponding string of explicit quantifiers, raba is not surrounded by no's:

(21)

No raba be rabo goi, ba no farfu be bo

Not for everyone x is there a y such that, for every z, x is a non-father of y by z.

The reader will perhaps agree that this gem is even more opaque than the one we considered previously; and this time rendering the quantifiers explicit seems only to deepen the opacity. Fortunately there is a quite general transformation that allows us to eliminate the negatives from negative sentences that happen to have negative predicates, as this one does; and this operation applies to both implicitly and explicitly quantified forms.

That transformation is called reversal, and the reverse of a sentence with either explicit or implicit quantifiers is formed by changing (i) the signs of both the sentence and its predicate, and (ii) the type of each quantifier. Under reversal, all designative arguments and the signs of all internal quantifiers--in implicit forms, the signs of all arguments--remain unchanged. Applying the reversal rule to sentence (21) gives us:

(22)

Ba rabe bo goi, ba farfu be bo

There is an x (such that), for every y, there is a z such that x is the father of y through z.

which is still not exactly crystalline. But applying the reversal rule to the implicitly quantified formulation of sentence (20) directly gives us:

(23)

Ba farfu rabe bo

Someone x is the father of everyone y through someone z.

and we see, at last, that what is masked by the double negative in this case is the simple if preposterous claim that everyone has the same father.

To really understand reversal it will be useful to apply it to a simpler case:

(24)

No, ba brano

It is not the case that something x is bread.

The reversal of sentence (24) requires (i) that the sign of the sentence be changed from negative to positive, (ii) that the sign of the predicate be changed from positive to negative, and (iii) that its single quantifier be changed from an existential into a universal. Performing these three operations in any order produces:

(25)

Raba no brano

Everything x is non-bread.

Simplifying even further, we see that the reversal of

(26)

Ti no brano

This is non-bread.

is simply the same sentence with the signs of the predicate and the sentence reversed:

(27)

No, ti brano

It is not the case that this is bread.

But this is nothing but the familiar case of exporting the negative from a negative predicate in a sentence with no non-designating variables. Thus, the exportation of a negative predicate is evidently a special case of sentence reversal, namely the reversal of a positive complete sentence with a negative predicate in which no non-designating variables happen to occur. If they do occur, as in sentence (25), they too may be accommodated in the reversal procedure simply by changing their quantificational type.

We see at last what these objects we have been calling "predicate negatives" really are. They, too, are sentence negatives. But of course! What can be negated except a claim? And it takes a sentence, not a part of a sentence, to make a claim. But predicate negatives are that special kind of sentence negative that stand inside the quantifier string. Tucked away against the predicate as they are, their position guarantees that no quantification can occur inside them; such a negation will, therefore, always be the last operation to be performed after all the operations specified by the quantifier string have been performed. They permit us to use, in short, that elegant equivalence whereby we may say either 'It is not the case that all x are P' or 'It is the case that some x are not-P' without change or loss of meaning. And now, of course, we also see that if there is no quantifier string for the negative to stand inside of, as there will not be if the sentence has no non-designating variables, it makes no difference whatever if we speak our sentence negatives at the head of our sentences or in a way that makes them appear to be predicate modifiers. Stylistically, in a logical language, the sentence-head position is to be preferred. But there is no difference whatever between the claims of a negative sentence and that of the same sentence with a negative predicate provided that sentence is complete and no non-designating variables appear in it.

These reflections also clarify what may have seemed arbitrary, or even puzzling at first, about our methods of filling the holes in incomplete sentences. Why, for example, must we complete sentences that have negative predicates with universals, and sentences with positive predicates with existentials...even though the latter sentences may themselves be negative? The reason is now plain. If the sentence to be completed has a negative predicate, then all its quantifiers, when explicit, will stand outside this final negative. If it has a positive predicate and is negative, then all its quantifiers, when explicit, will stand inside this initial negative. But we have learned that to say 'It is not the case that some x are...' is simply another way of saying 'It is the case that all x are not...' So this method of completing incomplete sentences simply guarantees the equivalence of the two ways of making the same negative claim: one, with an inside negative; the other, with its negative moved outside:

(28)

La Djan, no pa godzi la Romas

John didn't go to Rome.

(29)

No, la Djan, pa godzi la Romas

It is not the case that John went to Rome.

The first, with its inside negative, we complete with a universal:

(30)

La Djan, no pa godzi la Romas, raba

John didn't go to Rome from every someplace x (i.e., from anyplace).

The second, with its outside negative, we complete with an existential.

(31)

No, la Djan, pa godzi la Romas, ba

It is not the case that John went to Rome from someplace x.

That the two sentences still make exactly the same claim may be seen by making the two quantifiers we have just supplied explicit:

(32)

Raba goi, la Djan, no pa godzi la Romas, ba

For every x, John didn't go to Rome from x.

(33)

No ba goi, la Djan, pa godzi la Romas, ba

There is no x such that John went to Rome from x.

For now, quite obviously, sentences (32) and (33) are simply reversals of one another; as are indeed, though perhaps less obviously, sentences (30) and (31).

5.27 Causally Connected Sentences with ikou

A class of sentence connectives which is of special interest to Indo-European learners are those formed by preceding the causal words kou moi rau soa and their compounds (see Section 5.7) with the prefix i-. This move generates the afterthought causal connectives of Loglan, words like 'because' and 'therefore' that figure so importantly in Indo-European thought. In Loglan these words are grammatically among the eesheks, but semantically they are more complex than the afterthought logical connectives that are their lexemic kin. Here are several sentences connected by ikou-type connectives:

(1)

Da pa rodja, ikou, tu pa cuidru [SHWEE-dru] da

It grew because you watered (water-did) it.

(2)

Tu pa cuidru da, inukou, da pa rodja

You watered it; therefore it grew.

(3)

Da pa rodja, inokou, tu no pa cuidru da

It grew although you didn't water it.

(4)

Tu no pa cuidru da, inunokou, da pa rodja

You didn't water it; nevertheless it grew.

Evidently the Loglan words ikou, inukou, inokou and inunokou [ee-koh-oo ee-noo-koh-oo ee-noh-koh-oo (and) ee-noo-noh-koh-oo] mean (approximately) what the words 'because', 'therefore', 'although' and 'nevertheless' mean in English. What kinds of claims do they express? Note that one must pause after these words to avoid potential phonetic ambiguities; alternatively, the 2013 reimplementation provides alternative forms of these words closed with -fi.^R9

From an examination of the Loglan forms we can guess several things immediately. The prefix i- strongly suggests that these are afterthought forms to be used without prefixed markers, and that strings made with them are therefore probably left-associative. Both are true. Stripped of i- the remaining (prepositional) forms kou, nukou, nokou and nunokou seem related to one another in two ways. For one thing, nukou is apparently the converse of kou; and this seems reasonable from the meanings of the English words 'because' and 'therefore' to which ikou and inukou apparently correspond. The second relationship that may be surmised is that nokou and nunokou ('although' and 'nevertheless') are in some sense the "negatives" of kou and nukou ('therefore' and 'because'); but the reason for this is not so plain. Finally, we might hesitantly infer that nunokou ('nevertheless') is the "converse" of nokou ('although'); and this is suggested not only by their forms but by the sense of the English words. Let us first examine the two converse pairs and return for the more elusive negative relationships later.

If we restore the prefix i- to these forms and re-examine the English sentences of (1) and (2), we see that they do in fact make the same claim and differ only in the order of their clauses. Thus E ikou, C--where C and E stand for sentences asserting some "cause" and some "effect," respectively--is evidently equivalent in meaning to C inukou, E. It is in this sense that inukou is indeed the converse of ikou. From sentences (3) and (4) we may draw a parallel conclusion. They too make the same claim with the same clauses in different orders. Apparently inokou and inunokou are also converse connectives in Loglan, just as 'nevertheless' and 'although' are in English. This much is plain, even in English.

But what about the negative relationships that we suspect exist between ikou and inokou in sentences (1) and (3), and between inukou and inunokou in (2) and (4)? We may note that sentences (3) and (4) do indeed involve a negative event, namely that you didn't water it, and that this is related somehow to the "negative feeling" of 'nevertheless' and 'although' in English. In particular, the forms of these two negative sentences are no C inunokou, E ('not C nevertheless E') and E inokou, no C ('E although not C'). In these expressions E is still some looked-for effect, namely its growth, but no C is now the absence (or negation) of some presumed cause C, namely your watering it. This suggests, but does not yet quite explain, the reason for the infix - no- in these two negative connectives.

We proceed now to examine the simplest and probably most basic claim made in these four sentences, namely the one made with Loglan inukou and English 'therefore':

(2)

Tu pa cuidru da, inukou, da pa rodja

You watered it; therefore it grew.

But what does this sentence claim? Clearly it claims more than a similar sentence with the logical connection of implication claimed to hold between these same two clauses:

(5)

Kanoi tu fa cuidru da, ki da fa rodja

If you water it, then it will grow.

For apart from the obvious difference in tense, the claim of (5) does not require that each of its constituent sentences be true. Thus it need not grow and you need not water it, and the claim of (5) that if you water it, then it will grow, may still be true. Not so with the causal claim in (2). Its growing and your watering it have, according to (2), actually taken place. For if you can show that either of these events has not taken place, you have refuted sentence (2). But is that all? Is (2) just an obscure form of logical conjunction? Is (2), in short, equivalent to (6) below?

(6)

Tu pa cuidru da, ice da pa rodja

You watered it, and it grew.

Clearly not. Some essential ingredient of the claim of (2) is still missing. Sentence (6) expresses part of that claim, to be sure, but not all of it. We sense that the missing ingredient is an implicit claim that one of these events caused the other one to occur, a claim which we can now make explicit with event-descriptions below:

(7)

Lepo tu cuidru da, gu pa ckozu lepo da rodja

The event of your watering it caused the event of its growing (to occur).

This sentence makes explicit use of the predicate ckozu, which is the source of the word kou. Ckozu [SHKOH-zoo] is a three-place predicate meaning '...is a cause of...under conditions...', and is therefore incompletely specified in (7). In other words we suppose that all English sentences involving 'therefore'--and by implication, 'because', 'although' and 'nevertheless' as well--always make a covert (and incomplete) claim (incomplete because the conditions so essential to a scientific understanding of a causal relationship are, in natural language, never specified when these words are used) that some event is related causally to some other event (under those unspecified conditions), and this in addition to claiming openly that those two events occurred.

Logically, this is a most unsatisfactory state of affairs. For if sentence (2) with its causal connective inukou, is equivalent to the conjunction of (6) and (7)--from which such connectives have been eliminated in favor of explicit predications of what the speaker is in fact claiming to be true--then there is every reason in a purely logical language to be unhappy with such words. They hide too much. And therefore transformations involving them will, I fear, always be obscure. Moreover, they encourage simplistic causal thinking, the kind in which singular events are taken to be the causes of singularized effects, while both are blindly ripped from the fabric of conditions in which we know, from modern science, causation always occurs.

These are serious intellectual defects of the Indo-European system of causal words. But Loglan is more than a purely logical language, or it is less. For in addition to providing a clear logical apparatus for the thinker, Loglan must also accommodate the grammatical machinery of the major natural tongues. And not only in English, but in all the Indo-European languages--which include six of the eight most widely spoken languages on the planet (the two non-Indo-European ones among the first eight being Chinese and Japanese), and is therefore the major human language family--the grammatical apparatus for making covert causal claims of just this kind is not only universal, it is in most of them a most exuberant growth. For the Indo-European mind is apparently fascinated by causation. (This may be one of the reasons why speakers of certain ancient Indo-European languages, in particular, the early Greeks, were the principal inventors of science.) We are not likely to give that fascination up. So if we were to eliminate causal connectives from Loglan in favor of the logically more transparent but more cumbersome forms typified by the conjunction of (6) and (7) above, we might find that Loglan translations of all but the simplest utterances of the Indo-European tongues might treble the originals in length. Think for a moment how indispensable the English writer regards his 'because'es, 'although's and 'therefore's, how densely they are scattered throughout any carefully reasoned piece of prose--the present text, for example--and how much, we have learned, such words secretly contain. It is clear we cannot ignore the problem of translating such telegraphic words into Loglan. The best we can do is provide similar encapsulations of these intricate causal relations but in verbal forms that make their logical freight a little easier to bear, and their transformations therefore a little easier to perform. This, then, is the limited objective of the apparatus of causal connection in Loglan, to the description of which we may now, rather more somberly, return.

Let us now consider the problem of the two negative causal connectives. If ikou and inukou ('because' and 'therefore') implicitly claim a causal relationship between events, their negatives, inokou and inunokou ('although' and 'nevertheless'), implicitly deny one. Let us see how. Consider the difference between the following sentences, one of which we have seen before:

(8)

Da pa rodja, ikou, tu no pa cuidru da

It grew because you didn't water it.

Sentence (3) has the form E inokou, no C in which a presumed causal agent C (the watering) does not occur. Sentence (8) has the form E ikou, no C in which the fact that some event C does not occur is taken to be a cause of E. Obviously these sentences make vastly different claims about the world.

But both sentences make the same pair of explicit claims, namely that it grew and that you didn't water it. But in addition (8) claims implicitly that your not watering it caused it to grow, an extraordinary claim but a clear one. Sentence (3), in contrast, apparently denies this causal connection between your not having watered it and its growth, for according to (3) it grew despite the fact that you didn't water it. Evidently the speaker of (3) believes that its growth was caused by something else besides your not having watered it. In (3) da expresses surprise but stubborn surprise. What da has observed (namely that it grew) was counter-expected precisely because you didn't water it. But da is apparently unwilling to give up the causal principle that watering usually causes growth. Da therefore opens the search for other causes of this strange effect by denying that your not watering it caused its growth.

The speaker of (8), in contrast, takes a very different view of the same two events. De refuses to be surprised, or to look for another cause. De simply says 'Oh well, it must be the kind of plant for which not watering is a cause of growth.' Thus (8) and (3) contradict each other; they cannot both be correct explanations of this corner of the world. Since they agree in their explicit claims, their disagreement must lie in what they claim implicitly. We have seen that the connective ikou, represents a positive claim with the predicate ckozu; its denial will therefore involve the negative predicate no ckozu. And it is from this relation that we derive the Loglan word for 'although', namely from i + no + kou = inokou.

Once we have seen this, it is an easy matter to derive inunokou from inokou; for it is nothing but the converse of the latter:

(4)

Tu no pa cuidru da, inunokou, da pa rodja

You didn't water it; nevertheless it grew.

And even in English it is clear that (3) and (4) make the same negative claim in converse ways; whence i + nu + nokou = inunokou for 'nevertheless'.

So much for the meanings and derivations of the four most commonly-used causal connectives, those built on the causal preposition kou, which is, in turn, derived from the predicate ckozu. As we have already surmised from the uniform appearance of the prefix i-, they are all members of the unmarked, afterthought forms of the sentence connective system, the eesheks. Strings of causal connections made with i-marked forms are indeed left-associative:

(9)

Tu pa danza lepo da rodja, inukou, tu pa cuidru da, inukou, da pa rodja

You wanted it to grow, therefore you watered it; therefore it grew.

Like all afterthoughtfully connected sequences, the first two of these three connected clauses are evidently to be joined together before the third is added. So we may infer that the speaker is claiming, first, that your wanting it to grow caused you to water it; and second, that this sequence of causally connected events was itself a cause of a further event, namely its growth.

So much can be achieved by afterthought connections. But if we now want to assert the right-associated claim that the first event, namely your wanting it to grow, was the cause of the sequence of events composed of your watering it and its consequent growth, we can easily say that, too. We need only make a forethought form of the connective inukou. To do this we remove the leading i- and add -ki as a trailing mark, getting nukouki [noo-koh-OO-kee]. Then, because this prefixed connective will now precede the cause ('you watered it') rather than the effect ('it grew'), as its analog inukou did in the afterthought mode, we deconvert it, ending up with kouki [koh-OO-kee] (read 'because') as the prefix element of the forethought connective. Finally, we will use this new kek with the usual infix ki or kinoi as appropriate to join two connectands causally as well as logically. Using kouki...ki... ('because..., ...') between the last two clauses of (9) gives the right-grouping required:

(10)

Tu pa danza lepo da rodja, ikou, kouki tu pa cuidru da, ki da pa rodja

You wanted it to grow; therefore because you watered it, it grew.

While (10) is now plainly right-associative, its meaning does not seem to differ substantially from that of (9) in English. But here is a pair of causally-connected sentences in which the left- and right-associated senses of the connections among the same trio of clauses differ markedly:

(11)

Da pa rodja, inokou, tu no pa cuidru da, ikou, to no pa danza lepo da rodja

It grew although you didn't water it, because you didn't want it to grow.

(12)

Da pa rodja, inokou, nukouki tu no pa cuidru da, ki tu no pa danza lepo da rodja

It grew, although you didn't water it because you didn't want it to grow.

There is no forethought form of the '...because...' connection in English. If there were one, it would be *'therefore..., ...' Therefore I have translated nukouki...ki... in (12) (whose literal translation would be the ungrammatical English form *'therefore..., ...') by its infixed afterthought equivalent '...because...', and tried to convey the sense of right-grouping required by shifting the comma ahead of 'although'...which is a move familiar to English writers under these awkward circumstances.

The second of these two claims makes modest good sense. It denies the causal influence of the antecedent chain of events--composed of your not watering it because you didn't want it to grow--on the growth of the plant in question. Sentence (11) is more interesting but probably false. For it asserts that the growth of the plant in the absence of your watering it was actually caused by your not wanting it to grow...a perversity of the adaptive function that would fascinate any botanist. Again we see that matters of scope are decisive in Loglan, and decisively plain even in situations where English relies on "common sense". As always, we wish to preserve the absolute clarity of statement that permits us to talk nonsense when we wish to.

The grammar of causal connectives, both after- and forethought, is very similar to, though not quite the same as, that of the logical connectives with which we dealt in Sections 4.23 and 5.16. In fact the syntactical distribution of the i-prefixed causal words (ikou,, inukou, etc.) is identical to that of the i-prefixed logical connectives (ica, ice, etc.) while the marked kouki...ki series of forethought connectives is identical in distribution to the marked logical series ka...ki, and so on, including their application to argument and predicate terms; in short, they are keks. There are, however, no causal connectives corresponding to either the eks or sheks, the a-series or ca- series of logical connectives. Even so, the range of grammatical expression of causal ideas in Loglan is probably greater than in English. Thus, by connecting arguments causally, I can say that I love John because I love Mary, briefly and yet precisely in Loglan, while the structurally parallel remark in English is apparently ungrammatical:

(13)

Mi cluva kouki la Djan, ki la Meris

*I love because John, Mary.

On the other hand, a grammatical expression of this same sentiment is certainly possible in English by using the prepositional form 'because of':

(14)

I love Mary because of (loving) John.

Mi cluva la Meris, kou la Djan

And this, as we see, goes into Loglan quite directly as the modifying phrase kou la Djan and not as a connection between arguments at all. In fact, the Loglan causal form in (14) is more flexible than that in (13) with its kekked connection. For, as a sentence modifier, kou la Djan may appear (nearly) anywhere in this sentence. For example:

(15)

Mi cluva kou la Djan, la Meris

I love, because of John, Mary.

(16)

Kou la Djan, mi cluva la Meris

Because of John, I love Mary.

And if this be thought too vague, then the argument of this phrase may be turned into a lepo-clause in which more detail can be supplied:

(17)

Kou lepo mi cluva la Djan, gu mi cluva la Meris

Because I love John, I love Mary.

This makes exactly the same claim as (13). Again, the loglanist has many options, and da may choose the one that seems most natural to da, given the customs of da's native language.

The reader may have noticed that, in designing a system of causal connectives for Loglan, we have followed the structural pathways already laid down for the logical connections of the language. Thus to learn about the one is to learn a great deal about the other.

5.28 The Connectives imoi irau isoa

All the connections formed in the preceding section were based on the implicit use of a single predicate, the predicate ckozu. But there are other causal predicates, e.g., modvi and raznu, with which one may claim other varieties of causal relations between events. Modvi, for example, relates the motivation of some actor to da's action, and raznu relates the reasons given by some actor for his act. It is certainly true that the relation posited by ckozu ('...is a cause of...under conditions...') is the most general causal predicate in the language in the sense that all others will be found to be species of its genus. But analysis of the many near-synonyms of 'because', 'therefore', 'although' and 'nevertheless' in English suggests that these other, weaker--or more sharply focused--varieties of causal claims are frequently the ones intended by the speaker who uses these words in English. For example, the word 'because' seems to have at least four distinct meanings in English, as the following sentences show:

(1) It grew because you watered it.

(2) I took it because I wanted it.

(3) He got first prize because he won the race.

(4) X is not divisible by 2 because it is an odd number.

Sentence (1) involves 'because' in the broad physical sense of causation with which we have been dealing so far. Therefore we will continue to use ikou to translate it. Sentence (2), however, seems to speak of that special kind of cause that is a human motive. Of course the speaker might wish to speak of even this causal relation in its most general sense, and hence use ikou. But let us give da a choice. The predicate modvi means '...is a motive of...to do... under circumstances...' Let us derive a new causal operator imoi from this predicate just as we derived ikou from ckozu. With this new connective we can then express the special motivational sense of English 'because'. And it is with imoi that most speakers will probably choose to translate the sense of 'because' which occurs in (2) above. ('My wanting it motivated my taking it; and both events occurred.')

Sentence (3) expresses still a third sense of causation, namely the kind we call justification. The act of giving a criminal a prison sentence or a winner a prize is justified (so we say), by the event of their having "earned" it, that is, by the one having fairly and truly won some contest and the other having been fairly and truly convicted of some crime. The winning and crime-doing are also causes of the prize- and sentence-giving, of course, and curiously enough, they are also motivations for it, but motivations of a special kind: namely the kind we feel righteous about. For these are the special kind of human motives by the mention of which we defend our actions to other humans. 'Why did you give him that sentence (or that prize)?' we ask the judge 'Because he earned it' says the judge piously. Impiously da might have said 'Because I wanted to.' Obviously these are different senses of 'because'. We will use the predicate raznu ('...is a reason for...to do...under circumstances...') to form the justificational sense of 'because', namely irau. And this is the sense of 'because' most probably intended in sentence (3). ('His winning the race was a reason for the judges giving him the prize; and both events happened.')

But note this. The Loglan speaker of a sentence meaning (3) now has three choices before da. If da wants to point to the broadest causal sense in which the winning of some contest led to the getting of some prize, da may still use ikou...because the race took place in the physical as well as in the social world. If da wants to point to the somewhat narrower sense in which the causal relation involved a human motive, for example the judge's desire to please the crowd or even to be fair, da may still use imoi. But if, as is most likely, da wishes to point to that very special and important kind of motive which we are willing to defend publicly, namely the judge's reason for judging justly, da will use irau. Again, we give options to the speaker of Loglan that are implicit, but very deeply buried, in the semantic structures of at least some natural languages.

The fourth variety of causation is more difficult to see. Sentence (4) clearly shows that we also use words like 'because' and 'therefore' in contexts in which no ordinary sense of causation is intended, and in the formal sciences like logic and mathematics we do so incessantly. At least our customary view of arithmetic truth, for instance, convinces us that there is no physical sense in which an "event" of something's being an odd number causes the impossibility of some other "event," namely dividing it by two. But this conviction may be mistaken. Words and numbers are tokens in the symbolic games we play, and among other things they are physical tokens. These games have rules: psychosocial, and hence biological, and hence, ultimately, physical rules. When we "obey" these social rules--for example, by not performing the physical act (one which physics allows us to do) of dividing a certain odd number by two and getting an integer for an answer--our behavior is in part caused by our having learned just those rules and not some others. Moreover, it is caused by them in a certain way, namely we are motivated by them. As we sometimes say, we are motivated by our "desire" to obey those rules. (Certainly no purely physical law prevents me from writing '3/2 = 1', for I have just done so. But my usual desire not to do so is itself a physical, that is, neuro-chemical, thing and a physically powerful restraint; but it is a thing, though I assume it is somewhere in my brain, which I cannot otherwise locate for you.)

But the existence of a social rule concerning the manipulation of symbols is not only part of my motive for obeying it, it is also one of my reasons for obeying it, that is, it is a motive for which I can expect to be rewarded when it is publicly displayed. Thus if I am formally correct in saying that because 3 is an odd number I may not divide it integrally by 2, teachers, mathematicians and other critical readers will applaud both my reasons and my reasoning. (If I am incorrect, it means I have disobeyed some other rule, which they may be quick to show me.) But these are not ordinary reasons, for the rules of logic and arithmetic are not ordinary rules. With other reasons, we brook disagreement; with these, we do not. They seem to be built into the symbolic game itself. Of course, if we stepped into another culture, with another set of symbols and other logics by which to conduct our verbal play, we might be surprised to find some very puzzling departures from our own rules lurking there. But in general we do not make this disconcerting step.³³ We accept, as we say, the "logical necessity" of certain conclusions given their premises. It is this relationship we covertly predicate when we say 'because' in this fourth and final way.

This fourth sense of 'because' is derived from the Loglan word snola which means '...entails...by rules/logic...' or, incompletely, '...is one of the premises of...' The derived sense of 'because' that means entailment is therefore isoa, and the corresponding sense of 'therefore' is therefore inusoa. This completes the set of causal connectives which Loglan provides: (i) the most general sense of causation with ikou, (ii) the motivational sense with imoi, (iii) the justificational sense with irau, and (iv) that special kind of justification called entailment with isoa. And just as there are three additional causal connectives built on the base -kou, namely inukou, inokou and inunokou, so there are nine more connectives built on the kernel elements -moi, -rau and -soa, and each of these has a forethought version made with -ki and used with infixes ki or kinoi. Their uses are straightforward extensions of the principles considered in Section 5.27.

5.29 Causal Questions and Answers

Just as we can use modal operators or tense or location operators to form relative interrogatives with hu (Nahu = 'When?'; Vihu = 'Where?'; Douhu [doh-OO-hoo] = 'How?' in the sense of 'By what method?'), so we can use causal operators to form the various question-asking words which translate as English 'Why?':

(1)

Kouhu [koh-OO-hoo]

Why? (For what cause?)

(2)

Moihu [MOY-hoo]

Why? (For what motive?)

(3)

Rauhu [rah-OO-hoo]

Why? (For what reason?)

(4)

Soahu [soh-AH-hoo]

Why? (From what premises?)

Like other -hu words, any of these causal interrogatives may be used anywhere in any sentence, and the effect is to turn that sentence into a question:

(5)

Kouhu tu pa felda

Why did you fall? (From what cause X did you fall?)

(6)

Moihu tu pa godzi

Why did you go? (From what motive X did you go?)

(7)

Rauhu tu pa faltaa [fahl-TAH-ah]

Why did you lie (false-talk)? (What reason justified your lying?)

(8)

Soahu tu fadpeo toi

Why do you conclude (end-think) that? (What premise entails that conclusion?)

Answers to such questions may be made in many ways. First, and most economically, a designation may be supplied as an answer to the inner question hu ('who?' or 'what?'). To question (5) about falling, a designative answer might be:

(9)

Le banhane

The banana.

Second, an identity sentence might be used to answer the same question:

(10)

Le banhane bi le ckozu

The banana is the cause.

Third, a predication sentence:

(11)

Le banhane pa ckozu

The banana was a cause.

Fourth, a relative phrase may be produced using the causal operator embedded in the interrogative compound, in this case the one in Kouhu:

(12)

Kou le banhane

Because of the banana.

Fifth, a relative clause may be supplied using the same operator:

(13)

Kou lepo le banhane pa nu setfa ba ta

Because the banana was put by someone there.

Sixth, a partial sentence, a clause introduced by the corresponding causal connective, may be given as an answer:

(14)

ikou, la Djan, pa setfa le banhane ta

Because John put the banana there.

And this, of course, is only part of, seventh, the complete allegation:

(15)

Mi pa felda, ikou, moiki la Djan, pa vizka mi ja kamla, ki da pa setfa le banhane ta

I fell because (physical sense) because (motivational sense) John saw me coming, he (was motivated to) put the banana there.

Now it should be noted that nothing prevents the loglanist from asking da's causal questions the other way round, that is, with effects rather than causes as the objects of inquiry. In doing so, there will usually be no simple English parallels. So English translations of such questions are often circumlocutory:

(16)

Nukouhu tu pa felda

Your falling had what effect? (Literally, Therefore-what you fell?)

(17)

Nusoahu tu fadpeo toi

What follows from your concluding that? (Literally, Therefore-what you conclude that?')

But curiously enough, there is one fairly common English interrogatory remark--albeit a fairly impertinent one--that does run the causal inquiry in this direction, and that is 'So what?' This expression translates very neatly into Loglan as:

(18)

Nusoahu

What follows?

And in a logical language it may not be impertinent to inquire into the consequences of one's interlocutor's remarks.

It may also be observed that to put a question to someone ba with kou or nukou does not preclude ba's answering at some more specific level of causal inquiry. Thus, to question (5) above, about the general cause of your falling, answers made with moi or even rau are surely permissible.

(19)

Imoi, mi pa danza lepo tcaku tu

Because I wanted to startle you.

throws an entirely new light on this apparently externally-caused event by confessing that there was an inner motivation for it, while

(20)

Irau, raba pia felda

Because everyone was falling.

goes even further and suggests that, in addition to being motivated, your falling had a justification...a reason you would defend. It is difficult to think, however, of how an answer made with soa could be made to such a question (what entails a falling?). A question put at one of the finer levels of causal analysis is not, however, very adequately answered at a grosser level. To answer question (8), for example, about what entails a certain conclusion you have drawn--let us say, in writing--the reply:

(21)

ikou, lemi pinsi pa clidu

Because my pencil slid.

might well be regarded as irreverent. An explanation, possibly; but not an answer to a question about entailment.

In closing this discussion of the causal utterance forms of Loglan, it may be useful to remind the reader that the entire causal apparatus we have just described reflects the usages and grammatical provisions of one family of languages most particularly. This is the Indo-European language family, albeit a wide and important one as we have noted. But in this feature of its grammar Loglan is not truly cross-cultural. For the luxuriance of Loglan's 'because's, 'although's and 'therefore's will strike no welcoming chord in many non-Indo-European human breasts. In this, as in several other matters, Loglan accommodates the linguistic exuberance of certain important languages whose speakers have things to say which simply cannot be said (neatly) in any other way. The cost of doing so may be great. For the causal predicates which underlie these grammatical provisions are probably less necessary for human thought than most of us who speak European languages are accustomed to assume, and, as we have noted, they may even obscure thoughtful speech in some important ways. It is noteworthy that scientists use the causal mode of explanation less frequently these days, often preferring functional expressions relating one variable to another in which neither is labelled as the "cause". For nothing has proved more elusive, scientifically, than the very idea of causation, historically important for the early development of science though this concept may have been. It may therefore be a mistake to follow the European languages down this slippery but historic pathway. But if it is, it is a mistake that we have found unavoidable; and for the reasons we have given.

Even so, these are reasons that may be set aside one day. A more international generation of loglanists may eventually decide to simplify its language by replacing its Indo-European-derived causal apparatus with....

Only they will be able to tell us what.

Notes

1 In preparing text to be submitted to the Loglan parser, the user should know that both periods (.) and cross-hatches (#) are currently being treated by The Institute's parsing program as end-of-utterance signs. Thus the computer currently collapses the distinction between utterances and speeches which we have just developed. This is because, at the time of writing [1988], the Loglan parsing program is not designed to act on instructions in Loglan--for this to be possible, an understanding of the structure of its interlocutor's speeches will no doubt be essential--but simply to exhibit to the human learner how and whether the machine has been able to parse some utterance, thus informing its author what da actually conveyed to the machine by it. The specimen sentences parsed in this way will either have been submitted to the parser one at a time or they will have been extracted by the program from a submitted text and parsed in sequence. For such didactic purposes, then, we have found that it is best to limit the machine's parses to reasonably short units, such as single utterances.

2 In 1975 Loglan it was the local modifiers that were unmarked and the global ones that were marked. Nine years later James F. Carter pointed out (personal communication) that the latter were much more numerous in natural language than the former, citing the decisive evidence he had gathered by counting the incidence of the two types of modifying phrases in a large corpus of English sentences. Carter proposed that the unmarked Loglan status be reassigned to the commoner form. That unmarked forms have higher use-frequencies than marked ones is a general linguistic law, and one which we had applied repeatedly in engineering Loglan. So of course Carter's finding was accepted by The Institute and unmarked modifiers officially became sentence modifiers in 1984. The Carter finding had sweeping formal consequences, however, and it was not fully incorporated into the machine grammar until 1986 and 1987. The first published account of the new arrangements appeared in Brown (1987).

3 We here use the word 'modal' in a sense that differs from its usual meaning in logic.

4 This move was first suggested by James F. Carter (1981) and, for a time, these structures were called "Carter Vocatives".

5 Note that *iu no, or "negative ignorance," is undefined. This is because iu is, in effect, the zero-point on the conviction scale and is therefore neither positive nor negative.

6 Thus the reader may observe that the attitude indicators are not operators in the logical sense at all; in particular, they are not modal operators. I have taken Quine's strictures on modal logics (1961a, p. 156) to be sufficient to discourage any effort to incorporate one into Loglan.

7 Traditionally, that is, since Ogden and Richard's groundbreaking work (1938).

8 Already some further suggestions have been made. Reader Perry Smith, for example, suggests oueu = 'It doesn't matter' + 'we suppose that', or, as in mathematics, 'Without loss of generality we may suppose that ...'

9 The discovery that Loglan needed a connective interrogative was of course made by Prof. John Parks-Clifford, The Institute's logician at the time (personal communication, 1977).

10 Each word on this list except toe (from to) and nao (from English 'Now') is derived from a primitive predicate: coa from corta ('short'), dou from donsu ('give'), and so on. Such derivations are strictly mnemonic, however; for the functions of the discursive operators (attention-calling, etc.) are very different from the meanings of predicates (claiming, etc.). See Loglan 2, Chapter 8, The Theory of Loglan Semantics, for further discussion of this point.

11 There is something unsatisfactory to me about this indefinitely long and formless list of inference-shaping words. I suspect that they belong to a region of language behavior to which insufficient logical and semantical analysis has yet been applied. Zellig Harris's "discourse analysis" (1952), for example, only scratches its crusty surface; and certainly I do not claim to have broken through. But I also suspect that a proper analysis of the discourse-shaping transformations of speech would not only be difficult to perform on the natural languages, but that once performed could only lead, if Whorf is right, to a set of operators and a transformation schema that would be far more sophisticated than minds originally trained in the natural languages would know how to use. Perhaps such sophisticated developments are best left for Loglan₂; for the leap from Loglan₁ to such hyperlogical manipulations of the flow of discourse may, when it happens, not be so great as the one from English, say, to this only faintly-envisioned realm. All this suggests, of course, that any given logical language will prove to be only a temporary blessing. From its heights new logics will almost certainly be glimpsed; but we cannot hope that these transcendental modes of thought will be optimally facilitated in the language from whose rude platform they have first been barely seen. And from the point of view of thinkers attempting these new transcendental operations, what was once a "logical" language will then seem "illogical" indeed. This is the dynamic corollary of Whorf's hypothesis, of course: the "boot-strapping" operation to which I alluded in Section 1.2. It is also the historical loop-hole, in his essentially static conception of the constraints on language growth, through which the Western languages, anyway, with their proliferation of symbolic calculi, seem already to have in part escaped.

12 My treatment of identity throughout this section, and despite its linguistical air, owes much to the logical writings of Quine; especially his (1961) and (1961a).

13 We may express what sentence (7) is about by rewriting it as a predication sentence:

If one can agree that (7) and (7a) have the same truth-conditions, one must then agree that sentence (7) is not about Mark Twain. In fact, the expression la Semiul Klemenz in (7) does not designate anything at all. Instead, it is a specimen designation being exhibited--much as quoting it would do--to the eye (or ear) of the beholder as an example of something that is capable, on suitable occasions, of designating the same thing as some other specimen designation, namely la Mark Tuein, is capable of designating on perhaps quite different occasions. The Morning Star is, to be sure, the Evening Star, and both are the planet Venus; but when this object is in the evening sky it would make no sense to call it by the first of these three names.

14 The point here is that any (sensible) conversation requires that all parties agree (more or less) on who or what they are talking about. Once this is established--by effective new designations or by identities which call up old ones--the speakers can then proceed to make predications with some confidence that all parties know where to find, if necessary, the objects to which they apply. This is the sense in which every designation is contextual. For the effectiveness of a designation can only be judged in the context in which its production is intended to bring about this essential cognitive rapport between interlocutors.

15 Eventually a "partitive" operator may be required in Loglan with the sense of '...is a part of individual...' If this happens, sentences formed with this operator will be of the same grammatical class as the identity and membership sentences discussed here. Such an operator will be particularly useful in case scientific words should be formed as names; see Loglan 2, Chapter 12, Some Alternative Ways of Making Scientific Words.

16 Much of the mystery that surrounds numbers probably derives from the unclarity of our talk about them. Are they names? Or are they things? If things, where are they? And if names, of what things? On the view taken in this book numbers are, like Alice in Wonderland, imaginary things. The names of numbers, are of course, not imaginary, just as 'Alice' is not imaginary, but Alice, the girl in Wonderland, surely is. Still, we can ask sensible questions about Alice: Was she tall? Or was she fat? How old was she? By consulting the works of Lewis Carroll one can find out. Similarly, one may ask sensible questions about numbers: Is the number two odd? Or is it even? By consulting the designation-rules of mathematics--that is, the mathematical identities-- one can find out. For in mathematics one settles such questions not by examining Two--for like Alice, Two cannot be examined--but by examining its other names.

17 If logical considerations were our only concern we would certainly dispense with these subordinate predicate constructions in designing Loglan; for surely (10) and (12) are transformationally more obscure than (11). But we must also accommodate the natural languages; and, in the Indo-European languages at least, constructions which exhibit the author's views on the relative importance of his predications are essential to "good writing." Clearly we must provide translations which conserve these forms and exhibit these intentions. At the same time it is possible that forms like (11) will find uses in the language in contexts where transformational concerns are paramount.

18 (16) is not, because the identifying clause in (16) need not be true for the sentence as a whole to be true. X may not be a gift but stolen property and still be a black horse.

19 To ji ja, jie jae, jio jao we would have to add another pair of links for partitive modifiers if a partitive sentence operator were added to the language; see Note 15.

20 The identification-predication distinction is not quite strong enough to handle all such problems on its own. Take the charmingly ambiguous English sentence (a) 'He wants to marry a Norwegian.' The pair of translations (i) Da danza lepo mercea ba ji norgi ('He wants to marry someone Norwegian') and (ii) Da danza lepo mercea ba ja norgi ('He wants to marry someone, and that someone is, incidentally, a Norwegian') doesn't quite capture these two English meanings; for in (ii), the linked predication ja norgi, 'who is, incidentally, a Norwegian', falls within the scope of the Loglan lepo-clause. Thus a fastidious back translation of (ii) into English would produce (ii') 'He wants the event of (his) marrying someone x, who turns out (surprise!) to be a Norwegian, to take place.' This is, to be sure, a possible third interpretation of (a). But it is a quaint and unlikely one, given the structure of English. The real second horn of this dilemma in English cannot be accurately expressed until we have moved the ja-claim outside the lepo-clause, which can be done as in (iii) Da danza lepo mercea ba, ice ba norgi ('X desires the event of marrying someone x, and x is a Norwegian'). The new connective ice, which functions here as 'and' between two coordinate clauses, will be found with its numerous kin in section 5.24. I am indebted to Winograd (1984) for this delicious example.

21 The machine grammar treats free modifiers that are single words--that is, all free modifiers except the vocatives and parentheses, which may have internal grammatical development--as "grammatical noise". Thus a part of the parsing program called the "preparser" removes all such "noisy elements" from the specimen before passing it on to the parser. This and several similar strategies make it possible for an LR1 parser ("Left-generating, Right- reducing, 1-lookahead") to parse a human language which is clearly not LR1. There is more on the machine grammar strategy in Loglan 6 and Brown (1982a).

22 According to Greenberg (1966), three of the six possibilities, VSO, SVO and SOV, are the three dominant world word-orders, in that the three major divisions of human languages may be characterized by their use of them. The other three permutations--which may be generated by exchanging the positions of S and O in the dominant ones, thus VOS, OVS and OSV--occur only as infrequent variants of one of the three major types...much as VOS occurs as a literary variant of SVO in English. Apparently none of these minor forms is the preferred form in any language, however. Of the three dominant word-orders, SVO is much the most common, both in the number of speakers of the languages that employ it, and in the number of those languages. Chinese, English, Finnish, Greek, Russian and all the Romance and Germanic languages are SVO (Greenberg's Type II languages). SOV (Greenberg's Type III) is a close second as far as the number of languages is concerned, but gives no contest at all in number of speakers. Languages such as Hindi, Japanese, Basque and Turkish are SOV. VSO, which is the logician's preferred notation (and Greenberg's Type I), is a distant third in numbers of both speakers and languages. Languages such as Hebrew, Maori, Masai and Welsh are of this least common word-order type. Loglan can handle all six of these word-orders, including their imperatives as defined by the absence of a subject. In the declarative mood, the three dominant forms are handled by the punctuation schemata gaVgaSO, SVO and SOV--note that two forms, the most widespread two, are completely unmarked in Loglan--while gaVOgaS, OgiVgaS and OgiSV handle the three subdominant orders. The imperative forms are simpler. Two subject-less forms suffice for the six: the forms VO, VO and OgiV serve the three dominant patterns, and the forms VOgi, OgiV and OgiV again serve the subdominant ones.

23 It is for this reason that the Polish Notation, of which the marked series of Loglan connectives is a variation, is called "parenthesis-free." Of course it is not wholly parenthesis-free, since the leading marker functions as a left-parenthesis.

24 The phrases *i ca, *i ce, etc., with which these new words might be confused, do not occur grammatically in Loglan.

25 By the sequence of transformations -(p --> q) <--> -(-p v q) <--> (--p · -q) <--> (p · -q), the middle and crucial one being one form of DeMorgan's Law.

26 This exceeds the number of connective words in English by a factor of about 4. The 14 English connective words are 'and', 'both', 'but', 'either', 'if', 'neither', 'nor', 'not', 'only', 'or', 'possibly', 'then', 'unless' and 'whether'. They are used to construct longer expressions such as 'if and only if', 'whether or not', and 'and possibly both', as well as figuring alone and in pairs ('if...then...'). Needless to say the method of constructing these phrases is far from regular; neither is it transformationally transparent. There are, for example, speakers of English who do not know what the transforms of 'unless' and 'only' are. Thus 'Only the brave deserve the fair' is still a stumper in Freshman logic. Why such expressions should be so obscure in natural language is difficult to say. But there are no such uncrackable verbal nuts in Loglan.

27 These are the four kernel elements a, e, o and u, the three syntactic affixes c-, k-, and i-, and the three semantic affixes nu, no and noi. But these last three do not have to be learned afresh as they have similar meanings elsewhere. Knowing these ten elements, then, the hearer or reader will be able immediately to recognize every new connective as a variation of alternation, conjunction, equivalence or independence on the basis of its kernel vowel; the transformational character of the variation can then be gleaned from the affixes nu, no or noi, if any are attached; and finally the scope of the connection can then be determined by inspection of the affixes c-, k-, or i-, if any of these is used. Thus the Loglan system of connectives may actually be simpler to learn than the English one, despite its greater contextual explicitness.

28 Loglan is indebted to Prof. Herschel Elliott of the Department of Philosophy of the University of Florida at Gainesville for the system of implicit quantification described in this section. Prof. Elliott's objective in devising this system was to make symbolic logic teachable; mine in adapting it for Loglan, to make it speakable. The objectives are similar but do not everywhere dictate the same procedures. So I take complete responsibility for any errors of interpretation or of logic which I may have made in adapting Elliott's system to the purposes of speech.

29 Conversely, to express any explicitly quantified claim involving a multi-place predicate implicitly one must first find a converse operation, or a set of such operations, which orders the non-designative arguments of the predicate in question in the same order as their variables appear in the explicit quantifier string. In Loglan there will always be such a set of operations for every possible order of the arguments of multi-place predicates, although some permutations of the arguments of higher-order predicates require as many as four or five operations to express, and the compound conversion operator that results is probably unintelligible. If it were ever desired to actually manipulate such intricate claims implicitly, a set of auxiliary conversion operators capable of exchanging places among sutori arguments would have to be added to the language.

30 The existence of a complete set of such rules is as yet a conjecture; and as I have been, since 1975, occupied with matters of linguistic engineering, I have not had time to investigate the matter myself. However, it is known that algorithms exist for expressing the implicit quantifiers of any simple sentence explicitly, and for transforming any explicitly quantified simple sentence into a form in which its quantifiers may be implicitly expressed no matter what the length of its polyadic predicate or the order of its quantifiers; see Note 29 above. Moreover, the scope-problem for any two-clause connected sentence can probably also be solved; see Note 31 below. But the analysis of quantifier scope in third- and higher-order connected sentences has not yet been undertaken.

31 The following equivalences, in which 'p' is any sentence not containing 'x',

(x)(Fx --> p)	<-->	(x)Fx --> p
(x)(Fx --> p)	<-->	(x)Fx --> p
(x)(p --> Fx)	<-->	p --> (x)Fx
(x)(p --> Fx)	<-->	p --> (x)Fx

or, in Loglan,

suffice for shortening the scope of any quantifier of any implication in which the variable of the quantifier does not appear in exactly one of its clauses. The scopes of quantifiers of two-clause connected sentences made with other connectives, and in which an x-free clause p appears, may be similarly shortened. The general rule is that if either of the two negative affixes, no- or -noi appears in the Loglan connective word on the same side of that connective as the clause 'Fx' appears, then the short-scope quantifier will be of the type opposite to that of the long-scope quantifier; in all other cases it will be of the same type. This rule depends on nothing more mysterious than the equivalence of '-(x)' and '(Ex)-', and of '(x)-' and '-(Ex)'.

32 Notice that were we to encounter an explicitly quantified sentence, like (6), with a negative internal designative argument, the negative would have to be "exported" to predicate-negative position, as in (7). This is an awkward move. But if the logical usages here proposed for Loglan are followed, such sentences will not be encountered. They are not ungrammatical, however; they are merely invalid transformations of sentences like (2) on the conventions here proposed. One cannot, of course, legislate grammatically against bad logic any more than against bad arithmetic. Thus Te le to mrenu pa gotso = 'Three of the two men went' are grammatical but false sentences in both languages. Sentence (6) is not false, but it is an invalid transform of (2).

33 This is, of course, the step Whorf asks us to take.

R1 This particular I-connected sentence is actually a single sentence. JCB knows (and says explicitly later) that the ICA connectives do not necessarily connect separate utterances, so we are a little puzzled by what he says here.

R2 Relative interrogatives are no longer free modifiers, though they can be placed fairly freely; the current parser treats vihu exactly as the sentence modifier vi hu.

R3 In 1989 Loglan, an unmarked name word could be used as a vocative. This is disastrous in relation to the "false name marker" problem, and the 2013 reimplementation outlaws it. The original text lists la and hoi as the only words which can occur before a name word without an intervening pause: we have added the other salutations and a few other cmapua to this list.

R4 We are quite pleased with our invention of seu to resolve the problem of predicate answers: as a general label for answers to questions it may well have other uses.

R5 In fact, gu, though it can stand in for many different closure words, should seldom be used. The punctuator guu which closes termsets was added after 1989 L1, and with guu available the occasions on which one would normally choose to use gu are limited. This section should perhaps be rewritten with appropriate use of guu. Note that guu already appeared above as a preferred way to close kekpredunits in metaphors.

R6 However, a pause-comma may not replace gu or any of its specialized relatives. This pause-gu equivalence has been abandoned in the 2013 reimplementation, with one arguable exception involving sentence-initial no.

To illustrate the difficulties caused by attempting to parse pauses as gu's, utterance (4) actually doesn't parse under the last LIP; (4') does parse. I don't fathom the reasons. Both will parse under the current parser.

R7 Some of the strictures I have added are supported by remarks in Notebook 3 that certain term lists in sentence construction rules should consist entirely of modifiers or should contain exactly one argument. Some seem to be common sense. Forbidding more than one untagged argument before the predicate without explicitly marking this intention detected many bad parses in the Leith novel, and is in accord with a remark in Notebook 3. Restoring observatives makes me happy.

R8 The original form of this example was Re leu mrenu goi, mei merji, anoi farfu, which is simply wrong. This says "For most of the the understood sets of men, these sets (not the men) are...". The form to which I have corrected it seems to be the original form of the example. Because I removed this, I also removed a following transformation of the incorrect sentence, which leads to a missing line number. Note that the sentence could be said with the prefix Re mei ji bie leu mrenu... -- one could use a set form, but JCB has not correctly expressed the way that a restriction to a set would work.

R9 There is a pervasive problem with APA and IPA connectives: they are phonetically ambiguous with the A PA or I PA situation of a connective followed by a sentence modifier. This has to be resolved by requiring a comma pause after such a connective. JCB showed some signs of awareness of this issue in his examples in NB3, where he regularly inserted such pauses. Leith sometimes, inconsistently, does the same in his novel. There are no uses of APA connectives in this book, though there are plenty in NB3. The novel contains lots of APA and IPA connectives, and also lots of A PA and I PA situations; the issue must be addressed. We have a fairly stable solution using pauses (and also forbidding explicit pauses after the PA in the A PA or I PA situations where uncertainty would be created as to the intention); the best solution is to use the alternative APAFI and IPAFI forms we have provided, which do not require any pauses.

R10 A subtle point worth noting here is that (as of the update starting in 2013) the lepo descriptions do not have po abstract predicates as substructures. A guo following le po sentence, an abstract description, closes this abstract description. Similarly, a guo following po sentence, an abstract predicate, closes that predicate. These two constructions are separate. The one situation which requires care because of this approach is a description le (po sentence guo) predicate in which the intention is that the parenthesized abstract predicate is to modify another predicate: this is not an abstract description at all. This exact form does not parse: we insert ge to make this work: le ge (po sentence guo) predicate, so that one can tell that the po starts a predicate and does not participate in an abstract description in this case. In 1989 Loglan, abstract predicates built on sentences could not appear as modifiers; in fact, they could not participate as components in any complex predicate construction -- they could not even be tensed. In Lojban, there is a regular need of double closures of abstract descriptions, which we avoid by this device.

R11 The behavior of LIP, the old Loglan parser, with this example exhibits incoherence connected with pause/gu equivalence: LIP does succeed in parsing the example as JCB intends with gu instead of gue, but it does this by parsing the comma after Djan as a "gap". If one drops the comma, LIP parses it with the subordinate clause attached to la Djan, and yet the actual pause in speech must be present, so no change in the phonetic shape of the utterance occurs when the comma is dropped (this is a context where pause/gu equivalence, even if it were allowed in the grammar, could not legitimately apply). (My current parser never allows the comma after a regular Loglan name to be dropped.) This example manages to be an example of two of my objections to JCB's use of pause/gu equivalence at once: it also just cannot be the case that a comma adjacent to a gu could legitimately carry the semantic weight of another gu; one would tend to pause next to gu just because of its grammatical function as a "comma". There are examples in Notebook 3 where it appears crucial for things to parse as JCB reports that commas next to gu's be counted in ways that it is unimaginable that a speaker or hearer could manage.

I have fixed the current parser so that this example works as JCB intends -- by removing all closures of names with gu, since almost all serial names have unambiguous right closures anyway (a final predunit might occasionally have final links which need to be closed with gue.

(Return to the Table of Contents)