In Lojban, everything that can be spoken can also be written. Therefore, these tones of voice must be represented by explicit words known as ``attitudinal indicators'', or just ``attitudinals''. This rule seems awkward and clunky to English-speakers at first, but is an essential part of the Lojbanic way of doing things.

Attitudinals make no claim: they are expressions of attitude, not of facts or alleged facts. As a result, attitudinals themselves have no truth value, nor do they directly affect the truth value of a bridi that they modify. However, since emotional attitudes are carried in your mind, they reflect reactions to that version of the world that the mind is thinking about; this is seldom identical with the real world. At times, we are thinking about our idealized version of the real world; at other times we are thinking about a potential world that might or might not ever exist.

Therefore, there are two groups of attitudinals in Lojban. The ``pure emotion indicators'' express the way the speaker is feeling, without direct reference to what else is said. These indicators comprise the attitudinals which begin with ``u'' or ``o'' and many of those beginning with ``i''.

The cmavo beginning with ``u'' are simple emotions, which represent the speaker's reaction to the world as it is, or as it is perceived to be.

	.ua	discovery			confusion
	.u'a	gain				loss
	.ue	surprise	no surprise	expectation
	.u'e	wonder				commonplace
	.ui	happiness			unhappiness
	.u'i	amusement			weariness
	.uo	completion			incompleteness
	.u'o	courage		timidity	cowardice
	.uu 	pity				cruelty
	.u'u	repentance	lack of regret	innocence

Here are some typical uses of the ``u'' attitudinals:

Splitting off the attitude into an indicator allows the regular bridi grammar to do what it does best: express the relationships between concepts that are intended, desired, hoped for, or whatever. Rephrasing these examples to express the attitude as the main selbri would make for unacceptably heavyweight grammar.

Here are the propositional attitude indicators beginning with ``e'', which stand roughly in the relation to those beginning with ``a'' as the pure-emotion indicators beginning with ``o'' do to those beginning with ``u'' --- they are more complex or difficult:

	.e'a	permission			prohibition
	.e'e	competence			incompetence
	.ei	obligation			freedom
	.e'i	constraint	independence	resistance to constraint
	.e'o	request				negative request
	.e'u	suggestion	no suggestion	warning

4. Attitudes as scales

In Lojban, all emotions and attitudes are scales. These scales run from some extreme value (which we'll call ``positive'') to an opposite extreme (which we'll call ``negative''). In the tables above, we have seen three points on the scale: ``positive'', neutral, and ``negative''. The terms ``positive'' and ``negative'' are put into scare quotes because they are loaded words when applied to emotions, and the attitudinal system reflects this loading, which is a known cultural bias. Only two of the ``positive'' words, namely ``.ii'' (fear) and ``.oi'' (pain/complaint), represent emotions commonly thought of as less ``virtuous'' in most cases than their negative counterparts. But these two were felt to be instinctive, distinct, and very powerful emotions that needed to be expressible in a monosyllable when necessary, while their counterparts are less commonly expressed.

(Why the overt bias? Because there are a lot of attitudinals and they will be difficult to learn as an entire set. By aligning our scales arbitrarily, we give the monosyllable ``nai'' a useful meaning and make it easier for a novice to recognize at least the positive or negative alignment of an indicator, if not the specific word. Other choices considered were ``random'' orientation, which would have unknown biases and be difficult to learn, and orientation based on our guesses as to which scale orientations made the most frequent usages shorter, which would be biased in favor of American perceptions of ``usefulness''. If bias must exist in our indicator set, it might as well be a known bias that eases learning, and in addition might as well favor a harmonious and positive world-view.)

In fact, though, each emotional scale has seven positions defined, three ``positive'' ones (shown below on the left), three ``negative'' ones (shown below on the right), and a neutral one indicating that no particular attitude on this scale is felt. The following chart indicates the seven positions of the scale and the associated cmavo. All of these cmavo, except ``nai'', are in selma'o CAI.

	  cai     sai     ru'e   cu'i   nairu'e naisai  naicai
	[carmi] [tsali] [ruble] [cumki]

Each of the attitude scales constitutes an axis in a multi-dimensional space. In effect, given our total so far of 39 scales, we have a 39-dimensional space. At any given time, our emotions and attitudes are represented by a point in this 39-dimensional space, with the intensity indicators serving as coordinates along each dimension. A complete attitudinal inventory, should one decide to express it, would consist of reading off each of the scale values for each of the emotions, with the vector sum serving as a distinct single point, which is our attitude.

Now no one is going to ever utter a string of 100-odd attitudinals to express their emotions. If asked, we normally do not recognize more than one or two emotions at a time --- usually the ones that are strongest or which most recently changed in some significant way. But the scale system provides some useful insights into a possible theory of emotion (which might be testable using Lojban), and incidentally explains how Lojbanists express compound emotions when they do recognize them.

The Lojban attitudinal system was designed by starting with a long list of English emotion words, far too many to fit into the 39 available VV-form cmavo. To keep the number of cmavo limited, the emotion words in the list were grouped together by common features: each group was then assigned a separate cmavo. This was like making tanru in reverse, and the result is a collection of indicators that can be combined, like tanru, to express very complex emotions. Some examples in a moment.

The most significant ``common feature'' we identified was that the emotional words on the list could easily be broken down into six major groups, each of which was assigned its own cmavo:

	ro'a	social		asocial		antisocial
	ro'e	mental				mindless
	ro'i	emotional			denying emotion
	ro'o	physical			denying physical
	ro'u	sexual				sexual abstinence
	re'e	spiritual	secular		sacrilegious

Using these, we were able to assign ``o'u'' to mark a scale of what we might call ``generalized comfort''. When you are comfortable, relaxed, satisfied, you express comfort with ``o'u'', possibly followed by a scale indicator to indicate how comfortable you are. The six cmavo given above allow you to turn this scale into six separate ones, should you wish.

For example, embarrassment is a social discomfort, expressible as ``.o'unairo'a''. Some emotions that we label ``stress'' in English are expressed in Lojban with ``.o'unairo'i''. Physical distress can be expressed with ``.o'unairo'o'', which makes a nice groan if you say it with feeling. Mental discomfort might be what you feel when you don't know the answer to the test question, but feel that you should. Most adults can recall some instance where we felt sexual discomfort, ``o'unairo'u''. Spiritual discomfort, ``o'unaire'e'', might be felt by a church-goer who has wandered into the wrong kind of religious building.

Most of the time when expressing an emotion, you won't categorize it with these words. Emotional expressions should be quickly expressible without having to think about them. However, we sometimes have mixed emotions within this set, as for example emotional discomfort coupled with physical comfort or vice versa.

Coupling these six words with our 39 attitude scales, each of which has a positive and negative side, already gives you far more emotional expression words than we have emotional labels in English. Thus, you'll never see a Lojban-English emotional dictionary that covers all the Lojban possibilities. Some may be useless, but others convey emotions that probably never had a word for them before, though many have felt them (``.eiro'u'', for example --- look it up).

You can use scale markers and ``nai'' on these six category words, and you can also use category words without specifying the emotion. Thus, ``I'm trying to concentrate'' could be expressed simply as ``ro'e'', and if you are feeling anti-social in some non-specific way, ``ro'anai'' will express it.

There is a mnemonic device for the six emotion categories, based on moving your arms about. In the following table, your hands begin above your head and move down your body in sequence.

	ro'a	hands above head	social
	ro'e	hands on head		intellectual
	ro'i	hands on heart		emotional
	ro'o	hands on belly		physical
	ro'u	hands on groin		sexual
	re'e	hands moving around	spiritual

The implicit metaphors ``heart'' for emotional and ``belly'' for physical are not really Lojbanic, but they work fine for English-speakers.

The following cmavo are discussed in this section:

	ga'i	hauteur/rank	equal rank	meekness/lack of rank
			[galtu]
	le'o	aggressive	passive		defensive
	vu'e	virtue (zabna)			sin (mabla)
			[vrude]
	se'i	self-orientation		other-orientation
			[sevzi]
	ri'e	release		restraint  	control
			[zifre]
	fu'i	with help      without help	with opposition
		easily				with difficulty
			[frili]
	be'u	 lack/need	presence/satisfaction	satiation
	se'a	 self-sufficiency			dependency
			[sevzi]

The grammar of indicators is quite simple; almost all facets are optional. You can combine indicators in any order, and they are still grammatical. The presumed denotation is additive; thus the whole is the sum of the parts regardless of the order expressed, although the first expressed is presumed most important to the speaker. Every possible string of UI cmavo has some meaning.

Within a string of indicators, there will be conventions of interpretation which amount to a kind of second-order grammar. Each of the modifier words is presumed to modify an indicator to the left, if there is one. (There is an ``unspecified emotion'' word, ``ge'e'', reserved to ensure that if you want to express a modifier without a root emotion, it doesn't attach to and modify a previous but distinct emotional expression.)

For example, ``.ieru'e'' expresses a weak positive value on the scale of agreement: the speaker agrees (presumably with the listener or with something else just stated), but with the least possible degree of intensity. But ``.ie ge'eru'e'' expresses agreement (at an unspecified level), followed by some other unstated emotion which is felt at a weak level. A rough English equivalent of ``.ie ge'eru'e'' might be ``I agree, but ...'' where the ``but'' is left hanging. (Again, attitudes aren't always expressed in English by English attitudinals.)

A scale variable similarly modifies the previous emotion word. You put the scale word for a root emotion word before a modifier, since the latter can have its own scale word. This merely maximizes the amount of information expressible. For example, ``.oinaicu'i ro'ucai'' expresses a feeling midway between pain (``.oi'') and pleasure (``.oinai'') which is intensely sexual (``ro'u'') in nature.

The cmavo ``nai'' is the most tightly bound modifier in the language: it always negates exactly one word --- the preceding one. Of all the words used in indicator constructs, ``nai'' is the only one with any meaning outside the indicator system. If you try to put an indicator between a non-indicator cmavo and its ``nai'' negator, the ``nai'' will end up negating the last word of the indicator. The result, though unambiguous, is not what you want. For example,

Similarly, an indicator after the final brivla of a tanru may be taken to express an attitude about the particular brivla placed there --- as the rules have it --- or about the entire bridi which hinges on that brivla. Remembering that indicators are supposedly direct expressions of emotion, this ambiguity is acceptable.

Even if the scope rules given for indicators turn out to be impractical or unintuitive for use in conversation, they are still useful in written expression. There, where you can go back and put in markers or move words around, the scope rules can be used in lieu of elaborate nuances of body language and intonation to convey the writer's intent.

The following cmavo are discussed in this section:

	pei	attitude question
	dai	empathy
	bu'o	start emotion	continue emotion end emotion

You can ask someone how they are feeling with a normal bridi sentence, but you will get a normal bridi answer in response, one which may be true or false. Since the response to a question about emotions is no more logical than the emotion itself, this isn't appropriate.

The word ``pei'' is therefore reserved for attitude questions. Asked by itself, it captures all of the denotation of English ``How are you?'' coupled with ``How do you feel?'' (which has a slightly different range of usage).

When asked in the context of discourse, ``pei'' acts like other Lojban question words --- it requests the respondent to ``fill in the blank'', in this case with an appropriate attitudinal describing the respondent's feeling about the referent expression. As with other questions, plausibility is polite; if you answer with an irrelevant UI cmavo, such as a discursive, you are probably making fun of the questioner. (A ``ge'e'', however, is always in order --- you are not required to answer emotionally. This is not the same as ``.i'inai'', which is privacy as the reverse of conviviality.)

Most often, however, the asker will use ``pei'' as a place holder for an intensity marker. (As a result, ``pei'' is placed in selma'o CAI, although selma'o UI would have been almost as appropriate. Grammatically, there is no difference between UI and CAI.) Such usage corresponds to a whole range of idiomatic usages in natural languages:

Both ``pei'' and ``dai'' represent exceptions to the normal rule that attitudinals reflect the speaker's attitude.

Finally, we often want to report how our attitudes are changing. If our attitude has not changed, we can just repeat the attitudinal. (Therefore, ``.ui .ui .ui'' is not the same as ``.uicai'', but simply means that we are continuing to be happy.) If we want to report that we are beginning to feel, continuing to feel, or ceasing to feel an emotion, we can use the attitudinal contour cmavo ``bu'o''.

When attached to an attitudinal, ``bu'o'' means that you are starting to have that attitude, ``bu'ocu'i'' that you are continuing to have it, and ``bu'onai'' that you are ceasing to have it. Some examples:

The following cmavo are discussed in this section:

	ja'o	I conclude
			[jalge]
	ca'e	I define
	ba'a	I expect	I experience	I remember
			[balvi]
	su'a	I generalize			I particularize
			[sucta]
	ti'e	I hear (hearsay)
			[tirna]
	ka'u	I know by cultural means
			[kulnu]
	se'o	I know by internal experience
			[senva]
	za'a	I observe
			[zgana]
	pe'i	I opine	[pensi]
	ru'a	I postulate
			[sruma]
	ju'a	I state	[jufra]

The term ``discursive'' is used for those members of selma'o UI that provide structure to the discourse, and which show how a given word or utterance relates to the whole discourse. To express these concepts in regular bridi would involve extra layers of nesting: rather than asserting that ``I also came'', we would have to say ``I came; furthermore, the event of my coming is an additional instance of the relationship expressed by the previous sentence'', which is intolerably clumsy. Typical English equivalents of discursives are words or phrases like ``however'', ``summarizing'', ``in conclusion'', and ``for example''.

Discursives are not attitudinals: they express no particular emotion. Rather, they are abbreviations for metalinguistic claims that reference the sentence or text they are found in.

Discursives are most often used at the beginning of sentences, often attached to the ``.i'' that separates sentences in running discourse, but can (like all other indicators) be attached to single words when it seems necessary or useful.

The discursives discussed in this section are given in groups, roughly organized by function. First, the ``consecutive discourse'' group:

	ku'i	however/but/in contrast		[karbi]
	ji'a	additionally			[jmina]
	si'a	similarly			[simsa]
	mi'u	ditto				[mintu]
	po'o	the only relevant case