Run, Spot, run!


In most natural languages the imperative is the shortest form of the verb; but in LAN the plain imperative ‘Do!’ must be expressed as:


DESA MI E TU DO   my desire = you do

VA MI E TU DO     my wish = you do

KOM MI E TU DO    my order = you do

BEG MI E TU DO    my entreaty = you do, etc...


This makes it completely clear and uses only full words. The structures may be compressed just a little bit:


      DESA MI< DO TU, KOM MI< DO TU, etc...


or maybe ‘KOM DO TU’, but they are still verbose. So I added ‘IM’ = ‘let (me, him, and also you) ...’ as an imperative marker[1].

As for the vocative, it must be an aside. So eventually we end up with:


      {[SPOTU]} IM RUN !     


And we cannot even repeat ‘run’ without repeating ‘IM’ also. Well, not all of us are in first grade.


All words are full words, and, in principle, there are no bound forms. For instance, there is nothing like the English past tense marker,

‘-ed’, which cannot stand alone. The token ‘PA’ actually means ‘past’ and may be used as the noun/adjective ‘past’ or as the adverb ‘then’. More exotic adverbs, like ‘IFE’ = ‘completed in the past’ may be used to translate verbal aspects, but there is no grammatical difference between:


      GO SOLO = go slowly


      GO IFE = done with going.


Similarly, the English plural marker ‘-(e)s’ may be translated by the token ‘SE’= ‘several’. But SE is a full word, and may enter compounds like ‘SE+HA’ = ‘several have’ = ‘joint possession’. And there are other translations for the plural, e.g. ‘ZE’ = ‘all kinds of’, which may, e.g. used as the noun ‘variety’.




Damn brackets

In compounds, there are two levels of binding, + and #. The fact that # has precedence over + is of course equivalent to the use of brackets (at one level only, no embedding) Is this really needed? is it enough?


An example, from the Babel text:


scatter (intransitive) = become in many places

      = FI+SE+PE, or even better: FI+ZE+PE

      (FI = become; SE = many; PE = place; ZE = variety, all kinds of)


scatter (transitive) = cause to be in many places

            = ZO+SE+PE, ZO+ZE+PE, IGI+SE+PE, IG+ZE+PE

            (ZO,IGI,IG = cause to)


Only + is used, because in any of these compounds, the order of the commponents is immaterial; replacing the compound by a phrase, or using partial compounds will not change the meaning:

PE SE ( place modified by many ) = SE PE (many modified by                                              place)

FI+ZE = become various = diversify

FI+ZE PE = diversify place = scatter (place indifferently as      subject or object)

ZE+PE = all kinds of places (no single English word for this      compound)

FI ZE+PE = become various places = scatter


   become modified by variety modified by place = scatter

One may take advantage of this freedom of rearrangement, and of the synonyms IG=ZO, to produce (e.g.)


      zeigpe = ZE+IG+PE[2] 


which is lighter than its equivalent zyozyepe = ZO+ZE+PE.


Now replace one word by a compound:


many = not one = NO+ONE


and consider arrangements and bracketing (alas!) of the compound FI+PE+NO+ONE


FI+PE+NO#ONE -> FI PE NO+ONE = become modified by place modified by     many

FI+PE+NO+ONE -> FI PE NO ONE = become modified by place modified by     not modified by one



brackets (lisp,allnoun)

slots or places (lojban)




. . . The meaning of a word is all the occasions where it is used, versus all the occasions where it is not. For instance, the grammarian says that `to be’ has two meanings (more properly two usages): as a copula and as an equivalent of ‘to exist’. But I would say that these two are just one, because the same word is used. Or the philosopher may distinguish several senses of ‘have’ :

      I have two feet.

      I have a sister.

      I have a dollar.

      I have a responsibility to my family.


But since only one word is used... And then, obviously, ‘ashtray’ means ‘plateau de frene’.

So maybe the meaning of such a word is all the occasions where it is used in LAN, versus all the occasions where it is not. Used by whom? educated native speakers? forget it.

Then maybe words don’t have meaning, and one must consider full utterances? That seems nearer to the truth, but is even worse than lojban predication with places.



easy to describe vs easy to learn


I am a programmer.

I am fat.

I am dancing.


Poetry is what is lost in translation, meaning it what is conserved in translation. (again emphasis on foreign language)


memorizing rules vs memorizing cases


descriptive name vs non-descriptive names

identifiers = proper names

“foreign words”




The real-life model for LAN is the Bantu language family. Bantu nouns fall into classes, and each noun class employs certain prefixes on words agreeing with the noun. For instance, in Swahili:


a verb or adjective agreeing with a noun of the KI-class  will take KI- as a prefix


Widely generalized (ignoring such concepts as noun, verb or adjective, never mind noun classes) this becomes Rule #2.a:


a word modifying another word will take the head-word’s first open syllable as a prefix


Then Rule #2.b is added for symmetry: the prefix may be the first syllable of the head-word or of the modifier.

Family affairs

As another example of minimal vocabulary, consider compound words denoting family relationship:

grandparent = parent of parent

uncle = brother of parent

uncle = husband of sister of parent


This is an infinite open class! What are the minimal building blocks?


A possible choice:

parent = PAR

child = VO+PAR

spouse = POSU

These terms are sexless, as is politically correct; but still one must tell mother from father:

male = MA (don’t get mixed up with Ma = Mommy)

female = FE

and, less obviously:

same = SA, SAME

one = UNU

two = UDU


Using these one may build[3]:

SA+UNU+PAR = half sibling

SA+UDU+PAR = sibling

SA+PAR = sibling or half sibling

SA+MA#PAR = paternal sibling

MA+SA#UDU#PAR = (full) brother

POSU+SA#PAR = in-law

SA#PAR+PAR = uncle/aunt (blood relatives)

POSU+SA#PAR+PAR = uncle/aunt (not blood relatives)

SA+PAR#PAR = having grandparent(s) in common

VO#PAR+SA#PAR = nephew/niece

VO#PAR+VO#PAR = grandchild

and so on.

It is obvious that these terms are very clumsy (and much more precise) than the common English words. In another design of words for relatives, one could take POG = ‘progeny, son or daughter’ as a primitive instead of PAR, or one could take SIB = sibling as a primitive, or one could use special terms for elder/younger brother/sister, etc. There is a trade-off between simplicity (precision, a minimum set of primitives) and common usage; in addition, common usage differs greatly among cultures. So I take the easy way out and supply translations for the words in my English vocabulary.



Compounds and phrases

      One must assign meaning to a compound. To do so, the first step is to split it into components, e.g.

      pyisevi = PI+SEVI = person, serve

Replace the + bonds by word-breaks to obtain the background phrase of the compound:

      PI SEVI = person modified by serve

The meaning of ‘pyisevi’ should be a combination of ‘person’ and ‘serve’, but in principle, a bit different from the background phrase ‘person serves’; otherwise the phrase would be used instead of the compound. We say that the background phrase suggests the meaning of the compound. In this case, one obvious interpretation is person serve -> servant. In this particular compound, the order of the components is irrelevant:

      sevyipi = SEVI+PI -> SEVI PI = serve modified by person

still meaning ‘person serves’; ‘person’ is subject, by default. One may take advantage of the free ordering of components – when sense permits – to create better compounds,  according to the following rules:

  • the fewer syllables in the compound, the better
  • a loose bond + is better (lighter) than a tight bond #
  • a loose bond realized as null is better (lighter) than a non-null

However, ‘sevyipi’ and ‘pyisevi’ are equally good; both have 3 syllables, and one non-null light bond.


Let’s try now to turn our servant into a waiter:

            pyisevieta = PI+SEVI+ETA = person, serve, eat

The background phrase

      PI SEVI ETA = person modified by serve modified by eat

may, unexpectedly, also mean ‘an eating servant = person who serves while eating’! So, from the cleaner phrase:

      PI SEVI DI ETA = person modified by serve modified by thing                   modified by eat

one would get:

      waiter = pyisevyidieta = PI+SEVI+DI+ETA

This can be made even clearer, using the tight bond #

      PI#SEVI+DI#ETA = pwisevyidwieta

The meaning of this term is suggested by PI#SEVI = servant, DI#ETA = food (thing to eat). Then combine

      servant modified by food -> waiter.

This, by the way, shows that interpreting compounds with tight bonds is a two-stage affair. The component order matters in this case:

      DI#ETA+PI#SEVI = dwietyapwisevi = food for servants.


But, after all, why should one bother to create a compound for ‘one who serves while eating’? Maybe a German or a Sanskritist would – these languages are much freer with compounds. Without any better justification than my taste, I decree that ‘pyisevieta’ is the word for ‘waiter’!


Consider another example:

      monyegi = GI+MONE = give, money

with the background phrase:

      MONE GI = money modified by give.

Because of the sense of the words, this can only mean ‘giving money’; ‘money’ is clearly the direct object of the verb ‘giving’ and no other relation seems reasonable. One obvious interpretation is:


give money -> pay

      monyegi pyisevieta = pay  the waiter ( = the waiter pays !? )

      monyegi tova pyisevieta = pay to the waiter


But others are possible:     


give for money -> sell       

      moneygi boko = sell a book


give money for -> buy        

      moneygi boko = buy a book


In the examples above, ‘for’ is one of the little[4] words we inserted so freely “just to make tolerable English” from TOKEN chains. So we end up with a word which means ‘pay’ if modified by a word denoting a person, but which is badly ambiguous if modified by a word denoting an object!

This is not as shocking as it seems, because not every word may be modified by any other word, but it shows the dangers of word-building and the fact that only semantics count, so no rules[5] are possible.


In conclusion, when dealing with compounds:


  • Crack the compound into components, and build the background phrase.
  • If the background phrase make sense, then it suggests the meaning of the compound.
  • If the background phrase does not make sense, rearrange the components in a different phrase till it makes sense; this one suggests the meaning of the compound.
  • The whole procedure may need repetition if there are tight bonds.
  • After all that, the meaning deduced may still be ambiguous.


       ‘or as a function’

I have the cut (conjunction) ‘O’ to join phrases, but one may also say:

`I drink coffee, cocoa or tea’. The cleanest representation would be:

      or(coffee, cocoa, tea)

which I would call ‘or as a function’. One could have similar constructions for ‘and’, ‘not’, etc. But I really don’t want brackets in my language! Once you let them in, you soon start to lisp and then only a magnetic memory can keep track! One may have instead:


and here ‘OR’ serves as a full word meaning ‘replaceable by / in addition to’.


Transitive verbs.


      I send you to him.      =(?)  Mi sen tuns ma.


The LAN sentence simply means ‘sending , involving me,you and a male’; it could mean, just as well, ‘He sends me to you’. The meaning of ‘send’ can be formalized as the English grammatical description:


      a transitive verb taking indirect objects,


or, in lojban, a predicate with at least 3 places:


      x1 sends x2 to x3,


so one expects ‘SEN’ to have 3 or more modifiers, and if any of these are interchangeable, ambiguity will result. The three words ‘me’, ‘you’, ‘he’ are interchangeable here, and in many other English contexts (the same part of speech, and also very near semantically) so their meaning cannot resolve the ambiguity. An unambiguous sentence is:


      Mi do sen (su) tuns tova ma. =

            I active send (passive) you in direction male.


One could deduce from this a grammar rule:


      agent DO transitive_verb (SU) patient ... = mi do sen (su) tu


or maybe


      patient SU transitive_verb (DO) agent ... = tu su sen (do) mi


and both are anathema, I don’t want agents, patients, or even verbs in my grammar, just words and subordination. I don’t want ambiguity either, so certain words like ‘SEN’ will need ‘DO’ and ‘SU’ modifiers, but one may count on word sense:


      mi sen boko = sending, involving me and a book


is much more reasonable as ‘I send a book’ than ‘the book sends me’, so ‘DO’ and ‘SU’ are not needed. In the examples above, the words in parentheses are optional; when interpreting :


      Mi do sen tu se’tova ma.


it is clear that ‘mi’ is the agent; so ‘tu’ which is a modifier of ‘sen’ falls in the patient slot. How do I know that there is such a slot? (it won’t appear in the dictionary, no lojban[6]) Just from the meaning of ‘sen’.


      I would also like to add a ‘nexus[7] rule’ that


            sen man = man sen = man  sends


as opposed to ‘#send the/a man’, i.e. in a combination noun+verb, the noun is subject/agent by default. But I cannot even state the rule without the forbidden terminology ‘verb’, ‘noun’, ‘subject’, so... Maybe rephrasing:


Nexus rule: If word-A denotes an action, and word-B denotes a possible performer of that action, then the combination of the two words, in any order, means:

      word-B performing word-A


      This makes sense as a default, because any action has some agent/ subject[8], although it may not have a direct object/patient (e.g. ‘crawl’), or an indirect object/focus (e.g. ‘eat’). It avoids the terms ‘noun’ and ‘verb’, talking on ‘action’ and performer’ instead. Is this fair play? I would say so, because, of course, words may be classified – by sense – as action-words, without requiring that such words show tense, person, etc. as ‘verbs’ do.

      In practice, the nexus rule is the same as the SVO ~ VSO ordering; as soon as an action-word and a performer-word appear next to each other, the performer-word takes the subject slot, so only object slot(s) are left to be filled by subsequent words.



Damn lojban places, or subject/object revisited


      My idea of ‘meaning’, combined with the idea that there are no parts of speech, requires that the verbs ‘follow’, ‘precede’ and the prepositions ‘after’ and ‘before’ be related. Ideally the four of them will be, or at least contain, the same token. But the details of this relation are messy, and force one to consider syntactical roles[9] – subject and object.


In a simple sentence


      day follows night = DAG FOL NIT


one could just as well translate ‘night follows day’, if there is no such thing as SVO order. So, does ‘FOL NIT’ mean ‘before night’ or ‘after night’? ‘DO’ and ‘SU’ around ‘FOL’ are not quite what is needed, since neither day nor night ‘undergoes’ anything[10].


      It seems I must deal with strict order between the verb/preposition follow/before and its object, or the verb/preposition follow/after and its subject!


subject follows object = object before subject = subject before object[11]


So how about LAN words which may not be verbs, nor prepositions, nor subjects or objects?


      In this example there is no help from sense; unlike ‘I eat bread’ the two items in ‘day follows night’ are meaningfully interchangeable.  So the word ‘FOL’ has two ‘places’ a la lojban, and treats them asymmetrically. If only one such place is filled then … And one must consider words with only one place filled, because compounds often consist of such phrases.


      After which, I realize that any LAN word has precisely two places: its head and its tail (modifier). So one may translate differently according to the places filled:

·         both,

·         only one, by position head or tail,

·         none.


So a complete dictionary entry for ‘FOL’ will read:



H. FOL T. =       H. follows T., H. is after T., T. is before H., T.                      precedes H., etc.

H. FOL =          before H., preceding H.

FOL T. =          after T., following T.


Here, of course, H. and T. stand for head and tail. In addition, the dictionary may give examples of words compound with ‘FOL’, e.g.


      dagfol = DAG+FOL =      the day before, yesterday

      foldag = FOL+DAG =      the day after, tomorrow


The difference results from the fact that the two underlying phrases ‘DAG FOL’ and ‘FOL DAG’ respectively match the patterns ‘H. FOL’ and ‘FOL T.’ above. Which is precise, and Byzantine. Even stranger, the following compounds make sense,


      UNU+FOL : preceding, previous

      FOL+UNU : following, next


but conflict with equally reasonable  meanings:


      UNU+FOL =   zero

      FOL+UNU =   two, (the number 1’ constructed in accordance to                        Peano’s axioms)


      I don’t like this one little bit, and know in my heart that I would forever vacillate between ‘dagfol’ and ‘foldag’, not being able to remember which one means ‘tomorrow’.


all prepositions are verbs , all verbs are nouns, but some adjectives…


and a few words about ‘beautiful’



Better and better


  • the fewer syllables, the better
  • the low prefix \ is bad
  • the loose bond + is better than the tight bond #
  • a null loose bond is better than a semivowel loose bond
  • the combinations yi, iy, wu, uw are bad


Some more slogans


      Etymology is semantics.


Of course it ain’t, ‘nice’ does not mean ‘ignorant’, and ‘artery’ does not carry air. One must ignore usage, diachronic etymology – the fun part – and especially shun such compounds as ‘understand’, which has nothing to do with ‘under’ or ‘stand’, but clearly proceeds from both. Such self-explaining compounds, as ‘lukewarm’ = ‘body’ + ‘warm’ = ‘body temperature’ may be rare.

Then maybe derivation is semantics ?

Although poetry is lost, one can gain much flexibility from LAN compounds like:


      fat+disapprove = obese

      fat+approve = portly, embonpoint


Since there are only autosemantemes in LAN, morphological derivation is word-compounding (no slogan here) and LAN etymology is basically decomposing compounds – the reverse of word derivation.


      English word = LAN word.


That would really be great, especially for automatic translation. Unfortunately, it won’t even work in the form ‘English word = English word’; consider ‘the human race’ vs. ‘the rat-race’ and some rarer ‘race’ meaning ‘pluck’. Even extremely technical words carry this ambiguity: the biceps of the anatomist is not the biceps of the prosodist, although both terms are as precise as they get. So I will just take the easy way out, providing a simple-minded automatic translator, and lazily avoiding race1, race2 and race3; to ease my conscience somewhat, I also translate LAN formally via triplets, q.v.


      Sense beats grammar any time.


Or, if you prefer, poetic license. Although word order in LAN prescribes strict dependency, if that does not make sense, think of a different ordering. This, of course, may be more than exponentially ambiguous – that’s why we have human brains.






Consider first a few examples, involving intransitive verbs:


  1. Grandma relaxes.
  2. Grandma relaxes in the afternoon.
  3. Grandma relaxes in her room in the afternoon.


These sentences may be easily translated as:






Remember that each unmarked word modifies the preceding word; the mark < shows that ROOM and AFTERNOON both modify relax. For these cases, word reordering does not produce serious ambiguity:


1b.   RELAX GRANDMA = relax modified by grandma = the relaxation of Grandma


2b.   AFTERNOON GRANDMA RELAX = afternoon modified by grandma modified by relax = in the afternoon Grandma relaxes (change of emphasis?)


Even the drastic change of (3):




still translates as:


3c.   relax involving grandma, afternoon, room = Grandma relaxes in her room in the afternoon[12],


because the literal translation:


      relax modified by grandma modified by afternoon modified by room


does not really make sense – ‘afternoon modified by room’ or ‘grandma modified by afternoon’ are too strange, although ‘room modified by afternoon’ might be acceptable: the special room where the Victorian lady receives in the afternoon. Notice that (3c) is translated according to the “sense beats grammar” slogan, and contains ‘involving’ instead of ‘modified by’, i.e. something even vaguer than the usual modification filler. Since it is vaguer, it is more general, so maybe better?

Notice also that in (1b) ‘RELAX’ appears as ‘relaxation’; one can make this kind of sentence “all-noun” at will.


Then some transitive verbs:


  1. The girl likes raspberries.
  2. The girl never eats raspberries.
  3. The girl always eats raspberries at home.
  4. The girl teases her brother.
  5. The girl gives her sister a book.
  6. The girl sends her sister to her friend.



Example (4) may still be translated:


(4b)  LIKE GIRL RASPBERRY = liking involving girl, raspberries


and so can (5) and (6):


      (5b) eating involving girl, raspberries, never

      (6b)  eating involving girl, raspberries, always, home.


Translation using ‘involving’ seems quite general; besides it admits free word order, and can be set in “all-noun” form: `eating’ is a verbal noun, i.e. verb or noun as you please. However, there are problems with (7):


      (7b)  TEASE GIRL BROTHER = teasing involving girl, brother


is ambiguous; we would really like to know who teases whom (as English word order makes quite clear) but the ‘involving’ form won’t tell. Nor is there any help from meaning: it is quite plausible that the girl teases her brother, or the brother teases her, although the commonest situation is precisely ‘there is some teasing involving the girl, and her brother’. In the same way, sense cannot clarify (9):


      (9b) send involving girl, sister, friend =

            girl sends sister to friend,

sister sends girl to friend,

friend sends girl to sister, etc. (6 permutations, all plausible)      


However, sense helps somewhat in (8). The expression:


      (8b)  giving involving girl, sister, book


can only mean that the book is given. Therefore one cannot rely on sense to solve all the ambiguities which arise.

Noun forms have the same problem: ‘my upbringing’ is what my parents did to me, or what I did to my children.


So what’s the big deal? In English

  • transitive verbs are different from intransitive verbs – that’s the reason for this classification
  • direct (and indirect) objects stand apart from other verb modifiers
  • certain ambiguities are resolved by word order
  • the fillers I supply so freely are actually quite meaningful:


      My upbringing by Mary ≠ My upbringing of Mary.


The point of all that is not to rediscover America, but to deal with ambiguity without involving all that grammar, and by using as little semantics as possible. I feel that ‘transitive’, ‘intransitive’, or even ‘verb’ are not fair play, but ‘action’, ‘performer’ and ‘patient’ are. Then one may say that there must be a way of distinguishing the patient from the performer of an action. What then about indirect objects? Shall I say that the action ‘give’ may have a performer, a patient and a focus? It may also have a host of other modifiers: temporal and spatial and whatever, and I do not want special terms for each of those (although they are distinguished in English). What I mostly don’t want is a lojban predicate with places[13]. Here is one with 5 places:


bevri 1     carry

bevri 2     carrier; cargo; carried

bevri 3     destination [for cargo]

bevri 4     source [of cargo]

bevri 5     carrying path


which, I think, fits something like:


The agent carries Enigma machines to Tokyo from Berlin over the pole.

            1         2                3          4            5


So why not a sixth place: by plane / by submarine? This is not to quibble about the places as listed; they show a great amount of judicious thought, and in most cases I would be hard put to add another place. But the fact that I (or the lojban designers) cannot think of other details to fit naturally into the predicate does not mean that nobody can. What I don’t accept is the finiteness of the place list – although I cannot give an example, deep in my heart I know that the list must be open.

      After which, I realize that any LAN word has precisely two places: its head and its tail (modifier). So one may translate differently according to the places filled:

·         both,

·         only one, by position head or tail,

·         none.


That is, I resign myself to consider each word as part of a triplet

head word tail  (or H W T )

and supply (possibly distinct) translations for the four subsets of

H W T which contain W. Alternatively, we may say that W appears in four possible environments: HWT, HW, WT, W. The last three ‘environments’ exist because the word may have no head or tail (coming as first word in a sentence, or after a break, or before a break, or as last word in a sentence, or as the only word in a sentence) or may appear as a component of a compound word.


As an example, a complete dictionary entry for ‘FOL’ will read:





H follows T, H is after T, T is before H,

T precedes H, etc.



before H, preceding H (the meaning of the missing T)



after T, following T (the meaning of the missing H)



cannot stand alone; would mean both before and after


This fits well with my idea of meaning: the verbs ‘follow’, ‘precede’ and the prepositions ‘after’ and ‘before’ are certainly related, and here are all translated by FOL. It also neatly sweeps some semantics under the rug: in the phrase ‘fall follows summer’, ‘follows’ is an transitive verb, and it is important to distinguish its subject from its object, but one could not really replace grammar by semantics: there is no action, no performer, and no patient[14].

The translation of the complete triplet takes care of the SVO (or other) ordering, which is so common in many natural languages. For most English transitive verbs the corresponding LAN words will be indeed HWT=SVO, but not always – see the ‘IG’ example below.

In addition, the two-term subsets HW and WT clarify the sense of compounds. For example consider the English verb ‘eat’





H eats T



H eats



T is eaten



eating, eat (intransitive)


Then one may build constructs with DI = thing:


DI+ETA = thing which eats = corrosive (?)

ETA+DI = thing eaten = food


The different senses are prescribed by matching DI+ETA to HW and ETA+DI to WT. Inquiring minds might want to know why ETA+DI is not matched to HW, with W = DI; the reason is that DI really stands alone, needing no triplet, i.e. is translated the same in any environment:





thing, concrete, object


As one more example consider the token IG. Originally, this is the Esperanto suffix –ig, forming causative verbs:


IG, ZO[15]



H causes T, H induces T, H renders  T, H makes … T, etc.



causes H, induces H



causes T, induces T



causation (intransitive)


The fact that in HW and WT both head and tail translate as direct objects is due to the fact that –ig is of such frequent use, and probably to its being a suffix to begin with. HW and WT will appear mostly in compounds. If a sentence consisting solely in ‘H IG’ should mean ‘H causes’ or ‘H is the cause’ one could say equally well ‘H IG TI’ (HWT), i.e. ‘H causes this’, with TI just as vague and general as ‘this’; alternatively, rely on sense: ‘MI IG’ can only mean ‘I cause’, not ‘cause me’. In any case, IG is not an English word, so I did not feel the need to adhere to SVO.

      Unfortunately, this does not settle the question. Any construction with IG is necessarily a double object phrase:

      X causes Y to do Z

So I must fit X,Y,Z into the two slots of H,T – somewhat impossible. I can say that, since IG T means “causes T”, a natural way to put it is 

      X IG Z Y = X causes the activity Z qualified by Y.

But what about

      Tom whitewashed the wall = [TOM] IG WHITE< WALL ?

Makes perfect sense, except that ‘white’ is not an `activity Z’, it is at best a stative verb, and verbs are a no-no. And even with a true activity Z, the sentence:

      X IG Z Y

has switched the positions of Y and Z from the original SVO:

Y performs activity Z on U

(compare with the SVO ordering X IG Z). This is precisely the difficult question of how the lojban places of a compound relate to the places of the components.

So maybe the solution is to declare that the “object” of IG is an action or a state, and any other object is incomplete?





H causes T, H induces T, H renders  T, H makes … T, etc.

T must be an activity or state; otherwise DO is understood after HWT.



causes H, induces H.  H must be an activity or state; otherwise DO is understood after HW.



causes T, induces T. T must be an activity or state; otherwise DO is understood after WT.



causation (intransitive)


Notice that I cannot say:


H IG T : H causes T, H induces T, H renders  T, H makes … T, etc.

T must be an activity or state; otherwise an activity or state must appear as additional qualifiers of IG.


because that would be tantamount to defining 3 places for IG. Then this kluge will work on multiple causatives:


      lady IG chef IG marinade venison =

      lady IG chef DO IG marinade venison =

      The lady caused the chef to marinade venison.


The first part ‘lady IG chef’ gets a default completion ‘DO’ because ‘chef’ is not an activity or a state.


Reasonable as it seems, this approach still leaves open great big pitfalls; the following compounds with UNU = one make sense,


      UNU+FOL = previous

      FOL+UNU = next


but conflict with equally reasonable  meanings:


      UNU+FOL = zero

      FOL+UNU = two, (the number 1’ built according to Peano’s axioms)


Now it’s true that ‘two’ and ‘the next’ are semantically related, and maybe some natural languages use related words, and yet…



Omen = sign of the future.  But

1     `sign `future

means just

2     ‘a future sign’ or ‘will (make a) sign’.

So, maybe `sign+`of+`future or `sign+`concerning+`future? Actually `sign+`future should be enough: it is a compound, not the plain phrase (1), so its translation is NOT (2).



I still yearn for the ‘involving’ formulation: some action involving a heap of modifiers, which then get sorted (by sense or grammar) into subject, object, complements etc. (semantically: performer, patient, time modifier, etc.)






Interval vs series

Bounded vs unbounded

Both sides of neutral vs one side only


series: first to last, previous, next

                        ARE THERE OTHERS? ?causation

interval: old to new

bounded: probable to improbable

unbounded: hot to cold (in common usage)

one side: aged

both sides: hot to cold, beautiful to ugly, Latin altus = high or deep.
































































-M,-M step 1

P,P step 1







-M,0 step 1

0,P step 1



identical to oi?



-M,P step 1




identical to imi, ii?
























-inf,-M step 1

P,inf step 1







-inf,0 step 1

0,inf step 1






-inf,P step 1

-M,inf step 1






-inf,inf step 1


















This is the table of all possible gradations, showing:

  • bounded (P,M) vs unbounded (inf, i.e. ¥)
  • one-sided (1) , two-sided (i.e. 0+ or 0-), three valued (-101)
  • discrete vs interval items
  • in principle, op9 properly includes oi11; also dmp10 properly includes ii12 and imi13. Still I cannot think of any op9 which is not also oi11 (?days of the week), etc.
  • items that seem empty (any?) and the series item which seems to contain just one meaning: first to last.


The first two columns are interchangeable, if the basic meaning is exchanged with its antonym.


Now, what are the appropriate grades?


Are these really semantemes?

LAN MINO MUN = a land with relatively few mountains (MINO = minority)

LAN MINO+MUN = a hilly land (MINO+MUN = hill, a grade of MUN=mountain)







Translate Murphy’s law : “What can go wrong, will.”

This “will” is not really a future – of course it is.

BA PEHA E BA FU. = Possible bad is future bad. (not too bad)

BA PES+SIG E BA AL+SIG. = Barely probable bad is certain bad.

This one goes without “future”, which gets incorporated into “certain”.





Say something about engsemes



Birdseed aint tokhm-e-morg    ﺗﺨﻢ   ﻣﺮﻍ



This is the girl that you believe loves me.






Analyze all of those with triplets and see if one can say “her love” as well as “she loves” and “loves her”.





H loves T



H loves



T is loved



loving, love (intransitive)


My love is like a red red rose # LOV MI E LI ROSE RED.


(because “LOV MI” is actually “loves me”, and I think the poet meant “the one I love” which may be or may be not the same). Then:


That PI is too politically correct (or just plain bi).

Discuss a little bit; “make” is also the lexical function Real, Fact in some cases.



The noun make has 2 senses (first 1 from tagged texts)


1.    brand, make -- (a recognizable kind; "there's a new brand of hero in the movies now"; "what make of car is that?")



2.    shuffle, shuffling, make -- (the act of mixing cards haphazardly)


The verb make has 48 senses (first 29 from tagged texts)


1.    make, do -- (engage in: "make love, not war"; "make an effort"; "do research"; "do nothing"; "make revolution")


2.    make, get -- (give certain properties to something; "get someone mad"; "She made us look silly"; "He made a fool of himself at the meeting"; "Don't make this into a big deal"; "This invention will make you a famous physicist"; "Make yourself clear")


3.    make, create -- (make or cause to be or to become; "make a mess in one's office"; "create a furor")


4.    induce, stimulate, cause, have, get, make -- (cause to do; cause to act in a specified manner: "The ads induced me to buy a VCR"; "My children finally got me to buy a computer"; "My wife made me buy a new sofa")


5.    cause, do, make -- (give rise to; cause to happen or occur, not always intentionally; "cause a commotion"; "make a stir"; "cause an accident")


6.    produce, make, create -- (create or manufacture a man-made product: "We produce more cars than we can sell"; "The company has been making toys for two centuries")


7.    draw, make -- (make, formulate, or derive in the mind; "I draw a line here"; "draw a conclusion"; "draw parallels"; "make an estimate"; "What do you make of his remarks?")


8.    make -- (compel or make somebody or something to act in a certain way; "People cannot be made to integrate just by passing a law!"; Heat makes you sweat")


9.    create, make -- (create by artistic means; "create a poem; "Schoenberg created twelve-tone music"; "Picasso created Cubism"; "Auden made verses")


10.gain, take in, clear, make, earn, realize, pull in, bring in -- (earn on some commercial or business transaction; earn as salary or wages; "How much do you make a month in your new job?" "She earns a lot in her new job"; "this merger brought in lots of money"; "He clears $5,000 each month"), make -- (create or design, often in a certain way; "Do my room in blue"; "I did this piece in wood to express my love for the forest")


12.form, constitute, make -- (to compose or represent:"This wall forms the background of the stage setting"; "The branches made a roof"; "This makes a fine introduction")


13.reach, make, get to, progress to -- (reach a goal, e.g., "make the first team"; "We made it!" "She may not make the grade")


14.make -- (be or be capable of being changed or made into; "He makes a great host"; "He will make a fine father")


15.make -- (make by shaping or bringing together constituents; "make a dress"; "make a cake"; "make a wall of stones")


16.make -- (perform or carry out; "make a decision"; "make a move"; "make advances"; "make a phone call")


17.construct, build, make -- (make by combining materials and parts: "this little pig made his house out of straw"; "Some eccentric constructed an electric brassiere warmer")


18.make -- (change from one form into another; "make water into wine"; "make lead into gold"; "make clay into bricks")


19.make -- (act in a certain way so as to acquire; "make friends"; "make enemies"), nominate, make -- (charge with a function; charge to be; "She was named Head of the Committee"; "She was made president of the club")


21.have, get, make -- (achieve a point or goal, as in a sport; "Nicklaus had a 70"; "The Brazilian team got 4 goals"; "She made 29 points that day")


22.reach, attain, make, hit, arrive at, gain -- (reach a destination, either real or abstract; "We hit Detroit by noon"; "The water reached the doorstep"; "We barely made the plane"; "I have to hit the MAC machine before the weekend starts")


23.lay down, establish, make -- (institute, enact, or establish; "make laws")


24.make -- (carry out or commit; "make a mistake"; "commit a faux-pas")


25.make -- (form by assembling individuals or constituents; "Make a quorum"; "The branches made a roof")


26.hold, throw, have, make, give -- (organize or be responsible for; "hold a reception," "have, throw, or make a party", "give a course", etc.)


27.make, make up -- (put in order or neaten: "make the bed"; "make up a room")


28.take, make -- (head into a specified direction; "The escaped convict took to the hills"; "We made for the mountains")


29.defecate, shit, take a shit, take a crap, ca-ca, crap, make -- (have a bowel movement; "The dog had made in the flower beds")


30.make -- (undergo fabrication or creation; "This wool makes into a nice sweater")


31.make -- (be suitable for; "Wood makes good furniture")


32.make -- (add up to; "four and four make eight")


33.make -- (amount to; "This salary increase makes no difference to my standard of living")


34.make -- (constitute the essence of; "Clothes make the man")


35.make -- (appear to begin an activity; "He made to speak but said nothing in the end"; "She made a if to say hello to us")


36.make -- (reach in time; "We barely made the plane")


37.make -- (gather and light the materials for: "make a fire")


38.cook, fix, ready, make, prepare -- (prepare for eating by applying heat; "Cook me dinner, please"; "can you make me an omelette?" "fix breakfast for the guests, please")


39.seduce, score, make -- (succeed in seducing; young men's slang; "Harry finally seduced Sally"; "Did you score last night?" "Harry made Sally")


40.make -- (assure the success of; "A good review by this critic will make your play!")


41.make -- (pretend to be; imitate; "She makes like an actress")


42.make -- (consider as being; "It wasn't the problem some people made it")


43.make -- (calculate as being; "I make the height about 100 feet")


44.make -- (cause to be enjoyable or pleasurable; "make my day")


45.make -- (favor the development of; "Practice makes the winner")


46.make -- (develop into: "He will make a splendid father!")


47.make -- (behave in a certain way; "make merry")


48.make, urinate, piddle, puddle, micturate, piss, pee, pee-pee, make water, relieve oneself, take a leak, spend a penny, wee, wee-wee, pass water -- (eliminate urine; "Again, the cat had made on the expensive rug")



Hiding too long my head in sand.


Even if I define triplets to define tokens, what will be the triplet structure for compounds?





H causative T

H causes T, H induces T, H renders  T, H makes … T, etc.


H causative

H causes, H induces, H is a cause


causative T

… causes T, … induces T



causation (intransitive), cause





H eat T

H eats T


H eat

H eats (intransitive)


eat T

… eats T; T is food



eating (intransitive)






Then:  causative + eat = feed, without further ado, matching BT. Translate:


(1)      I feed my children spinach =

me causative child many< me< eat spinach.


This sentence seems clean, because ‘causative child’ (unless it means ‘impregnate’ or so) can only be ‘cause the child to do/be …’.

Even cleaner:


(2)      me causative eat spinach< child many< me<


which matches the HBT pattern. But should the compound causative+eat have two heads and two tails?


Heat Hcausative causative+eat Teat Tcausative


Even admitting that Tcausative = eat, still there are three items left around the compound, and only two places for them H, T:


   Heat Hcausative causative+eat Teat


Or, in simpler words, the head of ‘feed’ is ‘me’, but it needs two tails, ‘children’ and ‘spinach’. Since it is clear that children eat spinach, but spinach doesn’t eat children, the sentence:


(3)      me causative+eat spinach< child many< me<


is just fine, but then I could feed sharks to the piranhas and viceversa. So? Instead of the compound causative+eat one must use the pure token sentences (1) or (2). Which somewhat defeats the purpose of compounds. (I am sure lojbanists have something[16] to say about that)



Beware of gradation words such as :

too_little +blood = anemia



a source of difficulty: one trouble after another delayed the job; what's the problem?

too_little + comfortable


a state of difficulty that needs to be resolved; she and her husband are having problems; it is always a job to contact him;

too_little + comfortable


hurried and brief; paid a flying visit; took a flying glance at the book; a quick inspection; a fast visit

too_little + duration


formality and propriety of manner

too_little + intimate


easily deceived or tricked; at that early age she had been gullible and in love

too_little + knowledgeable


relative darkness caused by light rays being intercepted by an opaque body

too_little + light


darkened with overcast; a dark day; a dull sky; a gray rainy afternoon; gray clouds; the sky was leaden and thick

too_little + light ; minor + light


an unilluminated area; he moved off into the darkness

too_little + light; privative + light


separated from or unfrequented by others; remote or secluded; a lonely crossroads; a solitary retreat; a trail leading to an unfrequented lake

too_little + populated; populated + one


(comparative of close') indicating the one of two that is the shorter distance away; take the near street and ten turn right



Actually, my usage is just fine in this case.








An interesting classification, although I don’t like the names


STATES (no change; unbounded; durative)

Actions (change)

ACTIVITIES (change; unbounded; durative)

Performances (change; bounded)

ACCOMPLISHMENTS (change; bounded; durative; prefaced)

Achievements (change; bounded; non-durative)

CLIMAXES (change; bounded; prefaced; non-durative)

PUNCTUALS (change; bounded; non-prefaced; non-durative)


Durative is self-explanatory; bounded means that the action has

an ending point; prefaced means that the action has a starting



{duration/0} {beginning/0} {ending/0} {change/0}


Should produce 16 cases (maybe not all meaningful)

But, if arranged in a tree:



<no change>  <change>

              /    \

             /      \


     <unbounded>    <bounded>

                      /   \

                     /     \


        <durative>         <non-durative>

                                /    \

                               /      \


                       <prefaced>     <non-prefaced>


If no change, then no beginning or end. But why distinguish end from start? Maybe : completion/trying? begin/while/end?



Obviously, since poetry is what gets lost in translation, sense is what is preserved. So the sense of whatever is established by translating (periphrase?) Translating into what? Let’s call it the standpoint language -- it would be a standpoint from which to connect to other languages or conlangs, like Kepler’s Mars.


That is, all the terms that the standpoint language uses need to be covered, and all those it ignores may be safely ignored. E.g. using Russian as the standpoint, I would have to bother about verb aspects, but could ignore articles.


My standpoint language should be without morphology, but with rich nuances and generalized terms too. To make it rich, a natural language would do -- it is politically correct to assume that all natural languages are equivalent, since they all express all that needs to be expressed, etc. Pity I don’t know Chinese, that would be a good choice.

On the other hand, English ain’t too bad. Just enough morphology to be understood when talking about it, and natural, of course, though not for me.

So if I can translate English, my conlang can actually express anything that needs expressed be.

And anything that won’t translate gets under the rug as poetry.




I played a lot lately with ‘semantic primitives’ and rules for combining them, to get self-explaining compound words. All that is supposed to reduce the need of memorizing vocabulary. It also has to be somewhat practical – not just the 60 Wierzbicka’s primitives, that makes every new word a short story. Rather something like Volapuk or Esperanto, somewhat tighter formalized. And self-explaining means, roughly, ‘about as evident as GUI icons are’. The test would be providing the compounds and asking for translation, maybe on a web-page.



So,it turns out I cannot avoid verbs and lojban places! The nexus rule may create problems:

      ZO+BO = causative+beauty

means 'what beauty causes' by the nexus rule, and not 'beautify', because the underlying phrase is ZO BO and BO is a possible performer of ZO: beauty causes...Then how can you say 'beautify'?

One possibility is to always use DO and SU (which makes ZO a bound word!)

      ZO+SU+BO = beautify, BO+DO+ZO = what beauty does = ZO+BO = BO+ZO

But the whole purpose of ZO (and IGI, which is just Esperanto -ig-) is to create easily compounds like beautify.


One must then have the place definition for ZO:



      h. ZO .q : h. causes q., h. makes q. (as TU ZO HAPI MI = you make me happy), etc.

      h. ZO : h. is caused, (something) causes h.

      ZO q. : q. is caused, (something) causes q.


The one-place rules for ZO are exactly the opposite of the nexus rule; a noun near the verb ZO is its object, not its subject!


What about causative verbs (or multiply transitive) verbs? An example:

The mouse died. = MOSU DETA PA.

I killed the mouse. = MI ZO PA< DETA MOSU.

My wife made me kill the mouse. = FE+POSU ZO PA< MI ZO DETA MOSU.


As sentences, these are all clear (under the nexus rule, or SVO order) but ZO+DETA could just as well be 'bereavement' (what death causes) as 'kill'; ZO+ZO*DETA might be 'cause bereavement' instead of 'cause to kill'.



plain untagged two-words link is not enough, because many transitive verbs are asymmetric or antisymmetric:

      dog bites man vs. man bites dog

      allemande precedes courante vs. allemande is preceded by courante

(on the other hand there are some symmetrical verbs: boy meets girl, and many cases where only one meaning is possible: man eats sandwich)


to make these unambiguous, use the nexus rule: the 1st word which could be the performer IS the performer.


the nexus rule is sufficient in phrases, but not quite what I want in compounds, e.g. ZO which should have 'the 1st word which cold be the OBJECT is the object'


then I would have parts of speech: 'default subject', 'default object', maybe 'no default' forcing subject and object markers, etc.


or I would have semantic parts of speech: 'stative verbs', 'intransitive verbs', etc.


what about 'operative definitions' e.g. a Chinese verb is whatever takes negation (pu or mei)? which words can't take negation?

      this is a tree

      this is not a tree

(with the verb is, and even in Chinese with the verb shr) But in Russian, no verb, so 'tree' itself takes the negation. Operative definitions really wont work in a constructed language; they go by usage, and there ain't any.


'special usage' for shorties (making them a part of speech; not too bad) they might have different meanings according to the fill pattern

      h. CY q.

      h. CY

      CY q


To make it safer, use both:


      place definition for shorties


maybe some semantic parts of speech?

      performable action (e.g. eats ; most transitive verbs)

      static verb (e.g. stands , blue; intransitive verb)

      relation (e.g. precedes, before; needs two terms or more. THIS OPENS THE DOOR TO lojban PLACES)


maybe qualify by fill pattern:



needs h.

needs q.

neeeds h. and q. (that would be a relation)


or messier








1 at least


















I thought all words were 7bo; but Chinese Verb+Object are 5cq; DU is 3fh; and relations would be 8bc. Interjections might be 9bf : such a word stands alone, and 

words around it do not form links with it; etc.


I might use a marker (not just the cut I) to show there are no links.


English preposition are 8bc, English plural marker -s and past marker -ed are 2ch;

actually adjective and adverbs are 2ch -- they are defined as qualifiers, so they need a head.

those words with compulsory links (2,5,8,10) could be called incomplete; in extreme cases bound forms.


now go back to the translation rule: every word a noun. E.g.

      in = inside = interior

'the interior' is a bona fide English noun, but inherits incompleteness from 'in'; it begs to be followed by 'of ...'. Phrases where it stands alone 'the minister of the interior' [D2] are technical usage; 'the interior of a box' is strange wording for 'in the box', but is not technical or incomplete.

In the same way one might translate:

      on = surface

'surface' is less incomplete than 'interior'; one can talk about 'a polished surface' without saying to which object the surface belongs. On the other hand the prepositions 'in,on' float in the semantic space of the nouns 'interior, surface'; 'on the surface', 'in the interior' are probably commoner than any other prepositional phrases for these nouns. 'to the interior', which is also common usage, recreates the related preposition 'into'.

All of which is fun, but certainly I don't want this kind of detailed semantic analysis. All I should say is that after the translation:

      interior, surface

the prepositional phrases 'in/to the interior of'; 'on the surface of' will come automatically (to the English speaker) to be next replaced by the natural forms 'in,into,on'.


Any verb is a noun; one could always use the -ing form, beside the numerous nouns already found in English (seeing = sight, dying = death, rising = rise, etc.). Obviously, these words do not have equivalent meanings , but whatever they have in common is exactly what the LANGU word means.


eat vs food

food = DI+ETA = thing+eat

this means 'thing which eats' by the nexus rule, and might be translated as 'corrosive'. To make sure it means food, one should say DI+SU+ETA.


What if


instead of nexus I would have plain SVO? except that I don't want subjects and objects or verbs, it would solve most of the problems:

      DI+ETA = thing that eats

      ETA+DI = one eats that thing = food

and corrosive would be DI+ETA*GE.


I cut the bread with a knife = MI KUTU BADE< USE NIFE.

'cutting using knife' = 'cut with knife'

MI KUTU BADE is, of course, SVO; what about


now KUTU has two qualifiers, but the direct object does not follow the verb immediately; does this matter?

'me cutting using' doesn't fit SVO because of its meaning, so maybe some following word will be the object, and so it is: bread.


SVO rule: If one meets a transitive verb, the first preceding word that could be a meaningful subject is the subject, and the first following word that could be an object is the object.



the first HEAD of a transitive verb that could meaningfully be its subject is the subject, and the first QUALIFIER of a transitive verb that could meaningfully be its direct object is the direct object.


parts of speech: verbs


anything else? adverbs are adjectives, adjectives are verbs, prepositions are verbs;

                  pronouns, esp. personal





This is a possible solution of the “lojban places-SVO-nexus-transitive verb” mess.

Every WODU will be translated as an English verb! Adjectives are stative verbs:


      h. BO = h. is beautiful


and so are numerals:


      h. UDU = there are two h.s


nouns are nominal predicates:


      h. TERE = h. is a tree (containing the copula is)


verbs are verbs, prepositions are also verbs:


      h. RE q. = h. precedes q., h. before q. (translating before)


      h. TOVA q. = h. faces q., h. points towards q., h. is in the      direction of q., (translating to q., towards q.)


and the rest is a small closed class of pronouns, conjunctions, pure grammar words, etc. Even conjunctions may be made verbs:


      h. ORAW q. = h. may be replaced by q. (h. or q.)


Now, since they are verbs, some may be asymmetrical[17] transitive verbs so there may be different translations of WODU in the phrases:


  1. h. WODU q.

            h. RE q. = h. precedes q., h. before q. , q. after h., q.               follows h.

  1. h. WODU

            h. RE = after h. 

  1. WODU q.

            RE q. = before q., prior to q.


These are particularly important in compounds, which may fit one of the corresponding patterns


  1. h.+WODU+q.
  2. h.+WODU
  3. WODU+q.


where (4.) is probably the least frequent. And, instead of specifying for each WODU (with compatible semantics) the three patterns above, we shall say:


      If the meaning allows it, h. is the subject of the transitive verb (the agent of an action[18] verb) and q. is the direct object of the transitive verb (the patient of an action verb).


This may be called the SVO rule (which is plain and descriptive, but bothers me because I insist that there are no parts of speech and no syntactical roles) or the two-place rule[19] (which hides all this under lojban terminology, but is a fair description of the situation).

      So what did I get from all this terminology? One part of speech – verbs – is actually no parts of speech; ‘agent’ and ‘patient’ are purely semantic (as opposed to grammatical) categories, so one may say that knowing the sense of the word is all you need to use it (minimal grammar).


      Still one is left with the problematic terminology is ‘subject/object’ when dealing with transitive verbs which are not actions (e.g. precede). These may be called relations, i.e. a semantic object built on three WODU, with the second connecting[20] the first and the third. Maybe it boils down to using the WODU as preposition and postposition with different (actually converse) meanings:

      UNU RE : one precedes

      RE UDU : two follows

A different translation:

      UNU RE : one is before

      RE UDU : before two

manages to use a single English word, ‘before’ for the WODU ‘RE’; but the meaning is definitely not the same; an English preposition must come in front of the word it governs, and ‘is before’ is an incomplete, although intelligible phrase.


      Now as I got to converse meaning, I can probably include all transitive verbs by saying that the converse of an active verb is the passive. izzat so?

precedes -> is preceded = follows

but the converse of ‘gives’ is ‘takes’, not ‘is given’. Here lojban has the huge advantage of specifically inverting places 1 and 2 (passive) or

1 and 3 (converse) in the pattern:

      1 gives 2 to 3


The biggest difference between pre- and post- position will be in compounds:

      DI+ETA = thing which eats = ? corrosive

      ETA+DI = eat the thing, thing eaten = food

      UNU+RE = one before = 2

      RE+UNU = before one = 0

      SI+RE = the one who,precedes = previous

      RE+SI = the one who is preceded = next


Parts of speech:







at least one




















Boy meets girl. = BOGI METE GILA.

      METE : 8bc, symmetric

But consider meeting: team meeting = METE TEMA =  TEMA METE = the team meets

      METE : 10ac, symmetric


Boy loses girl. = BOGI LOSA GILA.

      LOSA : 8bc, asymmetric

the boy’s loss = BOGI LOSA

losing the girl = LOSA GILA


Boy gets girl. = BOGI GETAY GILA. (or BOGI ONTA GILA.)

      GETAY: 8bc, asymmetric



I do need an SVO rule, although it will be hidden under fancy terminology, as there are no verbs, nouns, subjects, objects.


A lojban predicate has a lot of places; a LANGU word has precisely two, one in front and one after; these, unless otherwise marked are the head (h.) and qualifier (q.).  Now comes the big semantic point:

some words show a special[21] relation between h. and q., others don’t.

For instance:

VODO = wood (the material):

      h. VODO q. does not have anything special, although it is quite

      reasonable: BOZO VODO HADA = a box of hard wood.

On the other hand

FO = precedes, before

      h. FO q. connects h. and q. : h. precedes q.

What I really should say: h. is subject, q. object of the verb precedes; ‘precedes’ is not an action word, but just a relation word, and its subject is not an agent, nor its object a patient.



As usual, it’s easier to criticize the competition. Here is a lojban predicate with 5 places:


bevri 1     carry

bevri 2     carrier; cargo; carried

bevri 3     destination [for cargo]

bevri 4     source [of cargo]

bevri 5     carrying path


which, I think fits something like:


The spy  carries code books to Tokyo from Berlin over the pole.

            1         2         3          4            5


So why not a sixth place: by plane / by submarine ? This is not to quibble about the places as listed; I think they show a great amount of judicious thought, and in most cases I would be hard put to add another place. But the fact that I (or the lojban designers) cannot think of other details to fit naturally into the predicate does not mean that nobody can. What I don’t accept is the finiteness of the place list – although I cannot give an example, deep in my heart I know that the list must be open.

So instead of the finite (but variable) lists of places prescribed by lojban, I would have precisely two: the head (h.) and the qualifier (q.) and I prescribe a default meaning for h. : if it makes sense, this is the agent of an action word, or the subject of a verb.


Essentially, I am against lojban places, because I dearly love word for word translations. The whole thing started this way: I picked a spelling list and munged it, so I got a list of

      WODU = English word

and this is how I would like to keep it; but

      RE = precede

      RE = follow

really won’t do. Eventually I will have to use pairs or triplets of words, but not always:

      TERE = tree

says all there is to it.




[1] I cannot find an English word to translate IM all by itself; does this make it a bound form?


[2]  ZE IG PE = variety modified by cause modified by place ?!

not really meaningful, because ‘cause modified by place’ does not really make sense. So rearrange, noticing that ‘variety modified by place’ does make sense, eventually reaching:

                cause modified by variety modified by place

[3] in all these compounds order is essential, i.e. the compound means the same as the phrase it replaces:

                SA#PAR +PAR = SA+PAR PAR = sibling of parent

                SA+PAR#PAR = SA PAR+PAR = same grandparent(s)

The situation is similar to the compound words denoting numbers.

[4] maybe not so small; it replaces ‘in exchange for’ which has a clearer, somewhat different meaning. Still, this use would be quite natural for an English speaker.

[5] i.e. general rules, as opposed to: ‘hilfen’ takes the dative, ‘bedarfen’ the genitive, verbs of motion the accusative, etc. An English example -- even muddier, because more dependent on meaning: the immutable order of the modifiers in   ‘three old cracked china cups’.

[6] yes it will! See triplets

[7] a NEXUS is the special combination SUBJECT+PREDICATE; properly, none is subordinate to the other (Jespersen).

[8] except when  it rains

[9] of which there aren’t any, either, in LAN

[10] The nexus rule will help somewhat, giving:

                DAG FOL = day follows ( as opposed to `#day is followed’)

[11] also ‘object precedes subject’

[12]  the literal LAN for that would be:  RELAX GRANDMA<  AFTERNOON<  ROOM.

[13] It’s always easier to criticize the competition.

[14] Cf. ‘The detective followed the suspect all day’.

[15] ZO is an alternative form.

[16] un-complimentary

[17] e.g. eat and  meet are both transitive, but eat is asymmetrical,  meet symmetrical.

[18] e.g. eat is an action verb, with a possible agent and patient, precede is not an action verb, just happens to be transitive.

[19] i.e., WODU may have up to two places, one before and one after, and these places have default meanings whenever such meanings make sense: before=subject/agent, after=object/patient.


[20] connecting in some organic way ? one may contrast RE in UNU RE UDU = one precedes two, with VODO in BOZO VODO HADA = a box of hard wood.  Give me a break.

[21] How fake can you get? special!? a transitive verb or  a preposition, and that’s it.

 [D1]usually; always; as a matter of habit etc. -- aspect

 [D2]Not English, Hebrew: sar hapnim.