A great deal of the machinery of language, which in natural languages is shared between the grammar and the vocabulary, is handled in gua\spi purely by words. Here is an explanation of how to say a wide variety of basic language patterns. Frequently I have thought that some form or meaning required a new primitive word, or even a change in the grammar, but it has turned out that existing words were more than adequate if creatively used. Make this fact your prejudice in similar situations.
A major difference between gua\spi and Old Loglan (and, I fear, Lojban) is that Loglan relies heavily on metaphor and on the human ability to understand metaphors, whereas gua\spi compound words mean what they do because specific rules say how words combine. To say it differently, the meaning of a Loglan compound arises from wishful thinking, not from rules designed into the language. Gua\spi can represent metaphors --- efficiently too. But in my Loglan writings I found that about 90% of all compound words could be interpreted according to rules I uncovered in the definitions, which Jim Brown was following unconsciously. I expect that even more compounds will turn out to be lawful in gua\spi, in which the definitions are tailored to make compounding easy, rather than taking the semantically slippery form of a true metaphor.
Gua\spi words have a relation between function and morphology. The CV pattern is for structure words, a few pronouns, and digits. CVV is for ``real'' relations, what in English are verbs. CCV is for abstract ``nouns'', normally used to denote objects. CCVV is used for more concrete ``nouns'' such as species, chemical elements, or household artifacts. The purpose of these assignments is to cater to the known proclivities of natural language speakers, who like to segregate nouns from verbs, and to simplify the process of making words for meanings. Nothing in the language depends on these assignments, and the language never actually distinguishes between nouns and verbs. If you find that some word has a morphological form other than what you expect, this is simply an effort to keep related words together, and it will have no effect on the efficiency of the language.
Here is the distribution of gua\spi words by functional and morphological category. There are 11 V's, and 14 regular C's. `:' is allowed only in CV words such as ``^:i'', and `#' does not count towards differences in words. There are 74 allowed CC digraphs out of 196 possible pairs; many CC's are too hard for people to recognize reliably.
Category Form Quantity Total Prefixes CV 61 61 Pronouns CV,CCVV 54 Digits CV 26 Other structure words 80 Primitives CVV 568 Primitives CCVV 532 Letterals CCVV 104 Predicates 1410 Total words 1551
Form Used Available Fraction Used CV 98 165 59% CCV 206 814 25% CVV 568 1694 34% CCVV 679 8954 8%
The words of natural languages appear to be arbitrary symbol strings of tremendous variety of sound. Gua\spi is similar in that its words were generated by a partially random process. To begin, the word lists of Loglan [L4] and Lojban [Lja] were merged and some additional words were added. For most words an English, Chinese and Latin translation was determined.
Then experimental phonetic data [NB2] was used to assess candidate words for the ease with which speakers could recognize them. For each gua\spi meaning, randomly generated word candidates were evaluated for recognizability, for distance from other gua\spi words, and also for similarity to their natural language equivalents. The final assignments were determined through a process of numerical annealing so as to maximize the summed quality scores.
Because the words are modestly similar to their natural language counterparts, learners are helped a bit in remembering them; and the sound patterns are anchored to forms known to please humans, since a prior attempt with purely random sounds was unacceptably ugly. The base natural languages included English because it is very widely spoken, Chinese (Mandarin) for the same reason, and Latin as a proxy for the other European languages, all of which have been influenced heavily by Latin. (Latinoid English words were avoided.) Both Loglan and Lojban use many more natural languages as word creation fodder.
CV structure words were assigned by hand; related structure words, like articles, have the same C and varying V's. Structure words pertaining to numbered cases have the same V's as the corresponding digits, but contrasting consonants, making learning easier.
A question often asked is, why create new words? Why not use Chinese or English words? First, some attempt has been made to keep gua\spi culturally neutral, and if Chinese words were used it would intimidate English speakers and vice versa. More important, Chinese words are designed for use with Chinese. Many required meanings, like articles, simply do not exist in Chinese, and similarly in English. And those meanings that are present are only approximations of the gua\spi meanings; while users have to invest a lot of effort to learn the new words and their definitions, they will find it even harder to keep straight what a word of their native language ``really'' means in gua\spi. That is why the approach was rejected of simply stealing natural language vocabulary.
Given some set, a ``basis'' is a subset from which all its members can be derived, as with vectors. Each vector space has a specific dimension, or number of basis elements, but words are not so simple. Loglan has about 1000 primitive words and it was intended that virtually any meaning should be achievable as combinations of these words; that is, the primitive words form a basis of nearly all meanings. Experiment proved that this intention had been accomplished for the most part, but that as anticipated, some areas were incompletely or imprecisely covered. The Lojban project of LeChevalier [Lja] is a continuation of Brown's work on Loglan [L1], and he has added about 300 primitive words, mainly about human emotion and interaction. For gua\spi I took over LeChevalier's primitive word list, with his kind assistance. I rewrote all the definitions to match gua\spi usage. I also added and deleted a small number of words to deal with specific gua\spi issues, and I expanded the scientific vocabulary in mathematics, chemistry, zoology, botany and agriculture. As a result, gua\spi has about 1400 primitive words. Some people are interested to discover just how few basis words we can get by with. However, I have experience with the Loglan word list and I have confidence in its ability to handle the required meanings; and while I do not believe it is minimal, I think it is fairly close. Thus I chose to use existing word lists for gua\spi rather than to try for radical pruning or de novo creation.
The CCVV pattern is used for organized groups of noun-type words such as the phrase-relative pronouns. These words have a CCV part and a final V, which matches the V in the caselink and the conversion for the same case, as well as the corresponding digit.
1st part Phrase where antecedent of pronoun is located zdm Phrase that pronoun is in xdr Phrase restricted by phrase containing pronoun vgr Phrase to which replies are directed xgn Enclosing main-level sentence zgl Previous main-level sentence Last letter Which case is represented o 1st case u 2nd case a 3rd case e 4th case i 5th case y The whole phrase
Here is a list of all the phrase-relative pronouns. There is also ``zgln'' meaning ``the previous discourse in general'' and ``zglr'' meaning ``the event just finished or still continuing''.
Phrase 1 2 3 4 5th Case (converted) of phrase . . . zdmy zdmo zdmu zdma zdme zdmi That the pronoun is in xdry xdro xdru xdra xdre xdri Being restricted xgny xgno xgnu xgna xgne xgni Enclosing top level sentence zgly zglo zglu zgla zgle zgli Previous top level sentence vgry vgro vgru vgra vgre vgri Question to be answered
Letterals are words representing letters, which are built up on a regular pattern like the pronouns are. A letteral means ``X1 is an instance of the letter (whatever)''. ``zu -fma !xo -<letteral>'' is the right way to say that something has the shape of a letter. To spell a word, compound the letters start to end. The result means ``X1 is an instance of something spelled (whatever)''.
Among letterals there are different forms for V's and C's, and upper and lower case in two alphabets are supported. The first table shows stem forms arranged phonetically, while the second shows what to substitute for the asterisks to signal alphabets and cases.
p psl* f fsl* b bzl* v vzl* m **lm u **lu t tfl* s sfl* d dvl* z zvl* n **ln o **lo c cfl* q qfl* j jvl* x xvl* l **rl i **li k kfl* g gvl* w **lw y **ly e **le : :zla # #vla r **lr a **la
To select cases and alphabets, use these for * or **:
The relation of Greek letters to gua\spi Roman letters is basically phonetic; the arbitrary assignments below are marked by `*'. The `:' can be used for the aspiration mark, an apostrophe in Greek or `h' in English transliteration, as in ``Hellas''. In principle one can also use a form like ``tler -fn -:alfa'' (the letter alpha) to refer to Greek letters.
* ** Alphabet Case a bx Roman Lower e dx Roman Upper y vx Greek Lower o zx Greek Upper
p Pi f Phi b Beta v Psi* m Mu u Upsilon t Tau s Sigma d Delta z Zeta n Nu o Omicron c Chi* q Theta* j --- x Xi* l Lambda i Iota k Kappa g Gamma w Omega* y Eta e Epsilon : ' # --- r Rho a Alpha
In English we use acronyms freely, but in gua\spi the letterals for the acronym are as long or longer than the words themselves. It is better to make an ordinary compound word for the concept.
It is the policy in gua\spi to use foreign words as-is (except for necessary mangling to make them fit the CV pattern), to represent the names of foreign people, places, flora, fauna, units of measure, foods, clothes, and so on. Certain of these word categories have a few members assigned CCV words. These are in gua\spi because they were in Loglan, and they were in Loglan because they occurred with high frequency in European literature. In Loglan it has proven impractical to manufacture in-language primitive words for every possible primitive, not for lack of word space but rather because those working on the language have other issues to attend to than making a continuous stream of predicates. For example, such an obvious animal as ``bear'' didn't make it into Loglan and probably never will. Gua\spi will have primitive words for more fauna and flora, but as in Loglan, the majority of species will never have gua\spi primitive words, and neither will most ethnic foods, foreign countries, provincial units and so on.
When foreign words fit neatly into the language, speakers are tempted to over-use them. Agglutinative languages like gua\spi and Loglan have great power to produce compound words with quite precise meanings, and speakers should try very hard to learn to produce such words. It is a fact that despite Loglan's limited set of nouns and, at that time, its lack of foreign words, I was able to write about 20,000 words of text while inventing only four unapproved primitive words (and a fifth was made for me after the fact): bear, torus, tape, noodle and oar. Speakers of gua\spi should try to emulate this performance and to use foreign words only for truly local concepts, such as the ``|mu'' (a Chinese area unit for land) or ``adobo'' (a Filipino chicken stew).
See the discussion under Pronouns: Foreign Words about attaching foreign words to gua\spi predicates to provide cases.
Sentences are usually connected by a retroactive downjump ``^:o''. With this operator the preceeding sentence is taken out of the discourse and is inserted in the first case of the word following ``^:o'' (which will need a default ``vo''). Thus the following two sentences are equivalent:
^:i -dae !kara ^:o -bal !crw |jro ^tara ^kseo If the box is open then maybe the rat will eat the cheese (With retroactive downjump) ^:i !vo -ge -dae !kara !fu -bal !crw |jro ^tara ^kseo (With explicit infinitive)
The form with the explicit infinitive is more natural in gua\spi. When retroactive downjumps are allowed you have to finish an entire sentence structure and hear the next word, possibly a downjump, before you can place the structure in the parse tree. However, all natural languages allow afterthought sentence connectives as in the above examples, and speakers rarely use forethought forms where the beginning of the antecedent sentence is marked. I wonder if users might resist an absolute requirement to put ``!vo -ge'' at the beginning of the antecedent sentence.
Speakers, please try to minimize the use of retroactive downjumps. We shall see if it is feasible to outlaw them completely.
Since the antecedent sentence starts out at the top level, it is asserted by default and remains thus after the downjump. The consequent sentence is an ordinary infinitive, which is not asserted. If you wish to assert it, or the antecedent when not using the retroactive downjump, put in a ``ge'' prefix as shown for the explicit infinitive antecedent.
A sentence start word ``^:u'' connects sentences with coordinated arguments, and such a group acts as a unit for retroactive downjumps. ``^:e'' also makes such units, but the arguments are not coordinated. More complicated groupings are best handled by making explicit infinitive arguments out of the component sentences and using ``ge'' to assert them, or by assigning names to the sentences and subsequently asserting a causal relation among the antecedents of the names.
The retroactive downjump ``fy'' relocates the sub-phrase before it at the same level, to become the first case of the following predicate at the same level. It roughly translates the infix ``and'' that connects a list of arguments in English. For example,
^:i !tara /crw !kseo ^fy -tla !qkao |spl /ftu =plyw The rat eats the cheese and the cookie and the apple (an unordered set) ^:i !ji /vlw !qo -:nuarko ^fy -stl !qo -trentn ^qo -prinstn I travelled to Newark and Trenton and Princeton (an ordered list)
This is one way to do multiple arguments (see also the section just below), and is just about the only use for ``fy'' --- but difficult for hard-core English speakers to give up. Needless to say, the retroactive downjump is not mandatory. ``tla-set'' is used when the arguments form an unordered collection, while ``stl-list'' is for arguments that have a sequence. Both words take as many arguments as you wish.
When a sentence is complicated, one can use ``fl-begin'' and ``fr-end'' to mark the beginning and end of a grammatical unit. A matching pair of ``fl'' and ``fr'' are supposed to be at the same grammatical level; if they are not, someone has made an error. They have no meaning beyond this checking function.
Another organizational transformation related to the retroactive downjump is error correction with ``fa''. When prefixed to a phrase, it causes the previous phrase at the same level to vanish and to be replaced by the one that follows. ``/fi -fa'' is a quick way to delete the current sentence. For example,
^:i !ji /vyu !gunu !ju /fa !diu -sui |zu -ken !ju I like your ass, I mean, your class ^:i -po !ju /crw !ftu =plyw |zey !ji /fi -fa Did you eat my apple . . . Forget it.
The conjunction ``fe'' has several functions depending on what it joins. First, when it joins phrase predicates it makes a collection of phrases with the same arguments; usually it is unseen in this function since the first tone `-' is usually sufficient to cue a parallel phrase compound.
^:i !jo /cyr -tai !sa-kqua Everyone get out of the water ^:i !qu -jaiw =tiri /cur -cie !qka -pne !qno =cail Tigereye cuts a hole through the piece of steel
Despite our monkeylike eagerness to treat these words as units, we should remember that in gua\spi there are phrases being conjoined. Thus everyone is told to go from the water and to be out of the water; and the result of cutting is both a hole and goes through the steel; the referent of this argument satisfies two predicates.
``fe'' also can join arguments. The first of the following examples shows a conjunction with ``fe'' in which successive arguments are joined into one. (For details, see Semantics: Cartesian Expansion.) The second example shows the same sentence rendered with ``tla-set'' in which the set members are listed. ``fe'' produces referents only in extension whereas ``tla'' produces a set in suitable context. There is a similar word ``stl-list'' when the arguments must have a specific order. For both words you may give as many arguments as you need.
^:i !kmau ^fe -kani /crw !tara ^fe -gara The cat and the cow eat the rat and the grass ^:i !tla !kmau ^kani /fi -crw !tla !tara ^gara (The same with sets)
Here is a particularly troublesome example of ``or'' in a set of arguments:
^:i |vi -pli ^jo /sle !xo -zu -co ^xo -psuw =fiei ^fe -xo -psuw =cawi ^fe -xo -qkmy Coffee, tea or milk? (Please choose one from the list)
Mixtures are expressed in several ways.
^:i !faia !do /fi -gai !xy -fpyl |sew !fkl ^qki Its face was covered by blue-gray fur --- in the overlap (sew) between blue and gray ^:i !flyl /zar !dvr ^vba ^fkl The flag is red, white and blue --- some parts in each color (zar) ^:i !jw /dmy !qkmy ^qtaw |zu -dma !xo -bimi This is a mixture (dmy) of milk and honey (sugar from bees)
Loglan has a concept of a ``mass individual''. According to Brown [L1] it is more characteristic of non-Western cultures. Here is my best explanation of it. Take the full referent set of an argument, and personify it so that, potentially at least, it is the same kind of thing as its members. For example, all sharks can be considered to be instances or manifestations of an archetypical shark god. This composite object is the mass individual. In Loglan, arguments in the ``serving or portion'' category, like ``cutri-water'', generally are used as mass individuals.
Concepts and features which in other languages seem unitary are revealed in gua\spi to be various. The mixtures above are one example, and mass individuals are another. Here are some examples related to mass individuals. Note that ``xo'' is the default first case article for servings and portions and so in argument sites with the normal default you need not say ``xo''.
!xo -kqua A portion of water (any one) !xe -kqua The portion of water !xe -psu -kqua The water molecule !xy -fpyl The hair (set) !xy -sur -fpyl The fur (emphasizing repetition)
^:i !qo -:mobi -dyk /kqnu |bir ^sa -qo -german -:melvyl Moby Dick was written by Herman Melville ^:i !qu -jaiw =tiri /fom !qu -kmaw =za -tye =tlme |va -zo -kqnu !sa -ji Tigereye is a character in The Welding Shop, which I wrote ^:i !qu -gua |xim !cfle /vu -zu -ziu !sa -kmi =xi -kmum /fi -pror |bir ^sa -xu -tla !qo -krnygan ^qo -:ryci The ``C'' Programming Language was written by Kernighan and Ritchie ^:i !qu -vo -tri -qtu =jy /zu -xim !xgnl |qe -zo -zymu !sa -qo -gabriel -fora Requiem by Gabriel Fauré (title on the score)
``Moby Dick'' is a bit ambiguous; it names both the book and the whale in it. The fault lies with the author for using one name for two referents. In any case, it is obvious what a foreign name means when referring to an object or a person. In the second example we have two gua\spi predicate names, conveniently all compound words so that they feel the same as the foreign names. But in the third example the predicate name is quite intricate, extends over three grammatical levels, and includes an imbedded name, ``C'', represented by a letteral. Nonetheless the principle is the same; ``qu'' converts the following phrase into a name.
The fourth example is perhaps the most difficult: the declaration of a title. The title is not part of the discourse but rather tells about it, hence it takes the form of a decoration with ``vi'' in a nonsentence. With it the by-line appears in the usual form for setting a modal case default.
Here are examples of the most important modal cases. However, virtually any word with two or more cases can be construed as a modal operator. Be alert for creative opportunities for expression.
|^:i !ji /za -ven !su -fkar ^vu -cnu !jrn !ji ^pra !fkar||I will buy the car when I have earned its price|
|^:i !ji /crw !kseo ^vu -bir !jun -vnl !tara||I ate the cheese before the rat came hunting|
|^:i !ji /qma -klo !kara ^vu -jro !crw !tara ^kseo |zu -vel !kara||I closed the box after the rat ate the cheese in it|
|jir||The location of an event|
|^:i !ji /fom !zu -plu |jir !pil -fn -:olimpu||I performed in the Olympic games, the games at Mount Olympus|
|zey||Genitive or possessive case: a relation of pertinence|
|^:i !trer =ji /dri !fkar |zey !ji||My brother is driving my car|
|^:i |jai !qo -kira /ju /dwu -csn -zu -jeu||Said Kira, ``You're a monster''|
|koy||The thinker, analogous to the speaker|
|^:i |koy !qo -kira /dri -fli /bni -siw -dan |qma||Thought Kira, ``The pilot needs to be rescued''|
|qnu||Vocative case: the listener paying attention|
|^:i |qnu !qo -jan /tara /jun !kseo !zey !ju||John, the rat is after your cheese|
|ciw||The holder of a subjective opinion|
|^:i !xi -tara /fpl |ciw !ji||Rats are beautiful, I feel|
|gae||The experiencer of an emotion (most have a case for this)|
|^:i !ju /csn |gae !ji||You're wierd, I feel|
|sen||The experiencer of an objective sensation|
|^:i !ftu -plyw /qke |sen !ji||The apple tastes [is] sour to me|
|brm||What something is part of (many parts have a case for this)|
|-xo -bryr |brm !dowu !ji||A brick of my house|
|tum||The instrument for doing an event|
|^:i !qo -kira /qma -pai !cana ^vu -tum !tuen||Kira bailed the boat [made it drain] with the bucket|
|zia||The way or manner of an event|
|^:i !xau =spa |cana /fi -fli |zia !fli !xi -bryr||The space shuttle (roundtrip spaceship) flies like a brick (flies)|
|fta||A standard or rule|
|^:i !xo -dair |xnu -syry -fn -:ameri-ka !co -ce /fi -pwo !ze -flu |xnu -fn -:ampere !co -cy ^fi -vu -fta !za -gul !xi -tyil ^se -zina||A wire of 14 AWG [American Wire Gauge] can carry 15 amperes according to the National Electrical Code|
|vrl||The reference frame|
|^:i !guna /xbr |vrl !qura |va -jmo -vjr||The snake is horizontal relative to the branch, which is almost vertical|
|plm||Such as: an example|
|-xi -pso |psi ^vu -plm !xi -kai |kei||Bad people such as thieves|
In English and all Indo-European languages, every sentence has a tense, that is, the syntax indicates (with great complexity) when the sentence occurred. Stories, for example, frequently have every narrative sentence in the past tense. Gua\spi uses instead the modal case default for tense, and explicit tense modal cases appear only for sentences off the default. The sentence start word ``^:a'' links sentences that occur in sequence. Here are some example sentences with tenses.
^:i !tara /crw !kseo ^vu -cnu !vn -juw -qana The rat ate the cheese at midnight ^:i !tara /crw !qkao -spl ^vu -jro !xdry !su -kseo The rat ate the cookie after [it ate] the cheese ^:a !kmau /crw !tara Then the cat ate the rat ^:i |qe -bir |fto ^jn /vu -qe -jro |pql ^cyr -xyn !qo -kaesar ^qo -gal A long time ago, Caesar had just entered Gaul (a nonsentence, it asserts nothing but does set the tense default)
In Russian, every verb is formally assigned an ``aspect'': ``perfective'', meaning that the sentence's event is considered as a unit, including its completion, and ``imperfective'', meaning that the predicated relation is continuous. English has these aspects too, though each verb can have either aspect depending on a moderately complicated syntactic cue (``ing'' for imperfective and various others for perfective). Speakers also like to distinguish a ``completed'' versus ``aborted'' aspect, whether an event reached its usual conclusion. In gua\spi the unmodified predicates are perfective or imperfective according to their meanings, but subordinate clauses or compounding words can express whatever aspects are necessary. Here are a few aspects in gua\spi:
^:i !xo -kqua /tiu -flu !se -bil !bryo Water continuously flows under the bridge (imperfective) ^:i !tara /crw |scu ^kseo The rat completely eats the cheese (perfective) ^:i !kmau /qem -jun -xna !tara The cat tries to catch the rat (might or might not succeed) ^:i !kmau /qai -jun -xna !tara The cat fails to catch the rat (though it tries)
A decoration is a short subordinate clause. Sometimes it expresses the speaker's attitude about the sentence or the relation between sentences, in which case its prefix is ``vi'', or it can be a subordinate assertion with ``va'', or an actual part of the main assertion, with ``vu'' or `|' tone.
|bwy||On the other hand; Contrast|
|^:i !ji /pql -gal ^:i |vi -zo -bwy ^ji /bzu||I am short; on the other hand, I am wide.|
|^:i !ji /jie -crw ^:e |vi -zo -smy ^ji /jie -byw||I am hungry, and likewise I am thirsty|
|^:i |vi -csn ^tara /go -crw !kseo||Strangely, the rat didn't eat the cheese|
|^:i |spo ^kqu /zu -vel !kara -zgy |zey =ji||Maybe food is in my refrigerator|
|^:i |zba /fpu !kqu |zu -vel !kara -zgy |zey =ji||Probably the food in my refrigerator is spoiled|
|^:i |vi -zo -tfn ^jo /ser -sui||Importantly, you must study|
|^:i |qe -jai !do ^ve -tsu ^ve -fto -faw . . .||He says suddenly and very emphatically . . .|
|^:i |va -jiw ^cana -fer /vnl||Hey, the barge is coming|
|^:i |vi -pli ^jo /pin -dwo||Please be patient|
|^:i |vi -gny ^jo /gey !xo -kqua ^ji||Kindly give me some water|
|^:i |faw ^jw /pei -bil !ji ^kqua||He dunked me!|
|^:i |ql ^jo /pei -bil !jw |din /kqua||So dunk him back|
|gza||Start of Paragraph|
|^:i |vi -gza ^qu -jaiw =tiri /qou !dvu |jir =zna||¶ Tigereye watches the waves in the canal|
|qti||Thesis (Topic Sentence)|
|^:i |vi -qti ^gua =spi /pwo !xa jy |zu -pwo !gua =fn -:ewlan||Gua\spi can do everything English can do|
|^:i |vi -sku ^zge |xne !gua =vdm ^gua =cin /fi -klo |va -ge -qma !gua !spi||Thus the gap between human and machine languages is closed by gua\spi|
|^:i !vo -gi -vgry /gi -kau -tiw !vem |zu -tw ^ji ^vu -cnu ^va -zu -plm !qo -:mei -ci |tl -co||That would relieve me of many present troubles, Mei Chi being number one|
|^:i !ji /kio !xo -kqu |va -ftu =plyw||I have some food, that is, an apple|