2.4 The Forms of Names: A name is any sequence of two or more sounds, including irregular ones, in any order except that the last must be a consonant and the whole followed by a pause. The formula for names is therefore (X)+[C/q/x]+.. There is one restriction and one exception to this formula. The restriction will be discussed in the next paragraph. The exception is that there is one very special class of scientific names, the so-called "Linnaean binomials" of biology, whose spellings are fixed by international convention and may not be altered. Occasionally these genus and/or species names end in vowels, e.g., Escherichia coli. Occasionally they end in consonants, E.g., Australopithecus afarensis. Thus the formula for Linnaean names is (X)X.. As these exceptional names are irregular beyond the hope of any ordinary morphological redemption, certain extraordinary arrangements must be made to place them in special contexts from which their resolution will be possible in Loglan. These arrangements are described in Sec. 2.13-15. In all intervening sections the word 'name' will refer to non-Linnaean names.
MRH: Names are in theory more restricted in form under my current PEG parser, as they are required to be built from syllables satisfying phonetic rules quite similar to the rules for syllables in borrowings. They are also of course required to end in a consonant followed by a pause (or silence). The lao names which include Linnaeans (and all other foreign names) are genuinely irregular. In practice, the only situation where a name in the corpus has had to be changed is when a syllabic consonant had to be written in doubled form as we now require.
2.5 The Pause Before Vowel-Initial Names: If a name commences with a vowel, then it must also be preceded by a pause. Thus the English names 'Ellen' 'Eileen' and 'Iona' go into Loglan as Elyn, Aili'n and Ai,onas. So 'Hello, Ellen', 'Come in, Eileen' and 'Give it to Iona' all require pauses in Loglan: /LOI.ELyn/ Loi Elyn, /nenKAa.aiLIN/ Nenkaa, Aili'n and /DONsula.aiONas.da/ Donsu la Ai,onas, da. (Note the final /s/ added to /aiONa/ to make it C-final; there is more on this move in Sec. 2.11.) If the preceding word is a "name-marker", like loi or la (see next section), or another name, then the pre-nominal pause is not represented by a comma in text. If the preceding word is anything other than a name-marker or a name, as in Nenkaa, Aili'n, then the pre-nominal pause is represented by a comma in text.
The pre-nominal pause is almost always "intervocalic"; see Sec. 2.36 for a discussion of intervocalic pauses.
MRH: this pause should be explicitly possible to write as a comma when desired.
2.6 The Name-Marker Restriction on Names: The restriction on the (X)+[C/q/x]+. name-making formula arises out of the fact that Loglan names are morphologically so irregular that in order for them to be uniquely resolvable in the Loglan speechstream they must not only be followed by pauses but surrounded by them—as in Takna, Djan, mi = 'Talk, John, to me!'—whether they are V-initial or not, unless they are initial in an utterance (Djan, takna mi = 'John, talk to me!') or preceded by a "name-marker" (Takna mi Hoi Djan = 'Talk to me, O John!'). Name-markers like Hoi are thus a morphologically privileged class of words. They serve to link C-initial names pauselessly to the rest of the utterance. For example, in the production /TAKnalaDJAN/ (TAAK-naa-laa-JAAN) the name Djan is not preceded by a pause. The reason it isn't is because it is preceded by the name operator la, which is another member of the class of name-markers that make pausing before C-initial names unnecessary. This last production resolves as Takna la Djan and means 'Talk to John!'. It is clear that the name-marker la is crucial to its resolution; see Sec. 2.17.
MRH: the name marker restriction has been abandoned. To make this possible, all names are marked either with a name marker word or by being preceded directly by an explicit pause (as after a name). In particular, all vocatives must be marked with hoi.
There are just seven simple little words that have this pause-blanketing privilege, namely I /.i/ Hoilaloaloisiasiu, plus any of the compounds formed with initial /.i-/, e.g., Ice = 'And'. Notice that I-words including I itself are themselves preceded by a pause and so must be initial in their breathgroups. But the other six name-markers do not require pausing either before or after they are spoken and so may occur anywhere in a breathgroup. Only la, by the way, must be followed by a name; the other name-markers may or may not be. This set of seven simple words and the I- compounds are both all the words and the only words which may precede C-initial names pauselessly.
MRH: the current list is la, ci, hoi, hue, liu, gao (a letter constructor). The utterance terminator i does not need to be on it. There is a case for returning loa, loi, sia, siu, sie to this list, but they are not currently on it.
Restriction is the other side of privilege. Obviously no copy of a name-marker may occur either initially or with a resolvable "prequel" within the first N-2 phonemes of a name. (The last two phonemes of a C-final breathgroup are bound to be all or part of the resident-name, and so do not have to be included in the search for name-markers.) If *Taknaladja'n, for example, were offered to us as a name, we would protest that its first part is not a name, but the incomplete phrase Takna la…, because that is how we have already resolved it. What we would have to do if it were new to us is first spot the name-marker la or a copy of it, and then discover that its prequel, takna, resolved. The prequel of a copy of a name-marker is that portion of the embedding breathgroup that lies to its left. A prequel resolves if the resolver can resolve it completely into regular words, that is, into structure words and/or predicates. Obviously it can do this with /TAKna/.
So it is clear why this restriction is necessary. Indeed if we wanted to make a name out of the utterance Takna la Djan—as we are free to try to do in Loglan—all we have to do is remove that offending sequence /la/. Is ?Taknadja'n then a name? Yes it is. It meets all the requirements for names including the name-marker restriction. For no copy of any of the sequences /.i… la hoi loa loi sia siu/ occurs within its production as /taknaDJAN/. We might translate such a name into English as 'Talkerjohn'…which opens up some interesting possibilities.
MRH: an essay on how the current situation works is needed, but in fact all of this is out of date.
2.7 Working Around the Name-Marker Restriction: The name-marker restriction sounds as if it might have eliminated a large proportion of the objects otherwise usable as names in Loglan. This is apparently not the case. In the many years I have been living with this restriction, I have found only two proscribed sequences which occur in natural names with sufficient frequency to justify a routine response. These are /i/ initially and /la/ either initially or in the context /VlaCX/; and both problems are easily dealt with. All other proscribed sequences are exceedingly rare. In fact I remember only two occasions on which a transcription of a natural name had to be rejected because a copy of some name-marker other than /i la/ was resolvably found in it; but the transcriptions were easily modified and I have now forgotten what the problem words were.
The routine solutions I use for modifying /i/- and /la/-containing transcriptions of natural names are as follows. In names like 'Ibanez' 'Isabel' and 'Ibiza'—Spanish is especially rich in /i/-initial names—I replace the /i/ (ee) in the transcription with /ii/ (yee). Thus /iBANieq/ /isaBEL/ and /iBIQa/ are the transcriptions and Iibanieq (yee-BAAN-yeth), Iisabe'l (yee-saa-BEL) and Iibiqas (yee-BEETH-aas) are the Loglan names. /la/-containing names are perhaps a little more common and many come from French. I used to replace the /a/ in /la/ with /e/; but these days I find the /la/ to /ly/ ((laa) to (luh)) as somehow closer to the French. So I now import French names like 'La Fontaine' (laa-fon-TEN) and 'de la Roche' (duh-laa-ROSH) as Loglan Lyfonte'n (luh-fohn-TEN) and Dylyro'c (duh-luh-ROHSH).
No doubt other languages and problems will inspire other solutions.
MRH: out of date. There is no name marker restriction, but a section on ways to avoid name marker problems will be needed in an uptodate technical manual.
2.8 Derivations of Names: All Loglan names that have been coined so far have been derived from some pre-existing linguistic source. Some sources have been internal, from within the Loglan language; others external, from outside it. Internal names are discussed in the next section; external ones in Sec. 2.10.
2.9 Internal Names: These are the names that have sources within the language. For example, our family has long had a dog named Cimr (SHEEM-rr); her name is derived from the Loglan word for 'summer', which is cimra, and certainly fits her disposition. (Cimra means 'is the summer of year…'.) So a predicate was converted into a name of a puppy by dropping its final vowel, following it with a pause in speech or a comma in text, and capitalizing its initial letter in text. The result is Cimr as in Cimr, kamla (SHEEM-rr . KAAM-laa), or, for that matter, in 'Cimr, come!’ Another example. We once had a cat with the uninspired name Gro'katm, which means Big Cat (from groda katma). We pronounced it (GROH-kaat-mm); but we probably shouldn't have. We were deriving the name from the complex predicate grokatma (groh-KAAT-maa) in which (KAAT) is the stressed syllable; so the name could have been regularly stressed as Grokatm (groh-KAAT-mm). Still, the apostrophe in the written form Gro'katm would have made it clear in text—had it ever gotten into text—that we were stressing it irregularly…as one is free to do with names.
Any predicate may be used to yield a name in this way, Thus, if one wants to imitate the Caribbean custom of saying 'Man!' vocatively, Loglan Mren, used in such expressions as Mren, ea mu safgoi /MREN.eAmuSAFgoi/ = 'Man, let's sail!', does the job nicely. (Safgoi is derived from salfa godzi, which means 'sail-go'.) Fum is an equally useful vocative, coming as it does from fumna = 'is a woman'. Some other internally derived names come from structure words. These usually require that a consonant be added to the V-final word to make the name, and by convention the consonant added to make internal names is /n/. Thus Tun comes from tu, which, when used as a vocative, has the sense of 'You!' or 'Hey, you!'.
2.10 External Names: Externally derived names are the closest possible imitations in Loglan of either (a) the sounds or (b) the appearances of certain natural language names which are to be imported into Loglan. It is generally not possible to do both.
MRH: since we have imposed rules of pronounceability on names, we come down firmly on the side of reproducing the sounds. Use lao (which JCB only mentions as the Linnaean construction) to build foreign names with their original orthography.
2.11 Auditorily-Modeled External Names: When the things named are persons or places, it is conventional in Loglan to imitate the sounds of the natural name as closely as this may be done in Loglan phonetics. The preferred auditory model is the way the name in question is, or would have been, pronounced by the person named, or by the people who live in the place named. If that natural pronunciation does not end in a consonant, then it is conventional to add the phoneme /s/ to external names. Thus if Peter is an Englishman, his own pronunciation of 'Peter' will sound something very like (PEET-uh) in Loglan phonetics; whence Pitys is his Loglan name. But if Peter is an American, we will hear (PEED-rr) when he tells us who he is, so no /s/ will be required to complete his Loglan name. Pidr and Pitys are two very different-sounding names; and in Loglan they are spelled differently as well.
The procedure is similar for names of places and famous personages. 'Bach' is Bax and 'Berlin' is Berli'n in Loglan because (bahkh) is a good copy of the German name of that German composer and (behr-LEEN) is an almost perfect reproduction of how Berliners speak the name of their city. 'London', on the other hand, is Lyndn because (LUHN-dnn) is another very good copy of the sounds one hears on the lips of Londoners despite the unsettling departure from English spelling that is caused by writing it phonetically. (English, as everyone knows, is an unphonetically spelled language.) 'Paris' suffers hardly any visual change because in being phonetically rewritten as ?Pari' (paa-REE) it becomes clear to the word-maker that it requires the conventional addition of final /s/; thus Pari's is Loglan for 'Paris', an adventitious similitude. 'Roma' is Romas because again the local pronunciation (ROH-maa) is V-final and again /s/ is conventionally added to the original, this time without an adventitious matching of the local spelling.
In general, when borrowing place and person names, loglanists try to make the best imitations possible with their limited phoneme kit unless the result is thought to be too difficult for loglanists to use on a world-wide basis. Thus, should German 'München' (Munich) be exactly reproduced as Mwnxen (MEUN-khen) (as it can be by using the irregular phonemes /w/ = (eu) and /x/ = (kh) which happen to be German)? Or should we adopt the more forgiving Muncen (MOON-shen), which although quite close to the original is still pronounceable by that large population of potential loglanists whose tongues could not reproduce the German sounds exactly? This is not a morphological question but a broader linguistic one on which The Institute has no policy at present. Until it is settled by an international congress of loglanists, current loglanists are free to adopt whatever degree of imitation they can manage. Thus either Mwnxen or Muncen but certainly not *Miunek.
MRH: this I find quite agreeable.
2.12 Visually-Modeled External Names: When the name to be imported is a scientific or scholarly one, and a certain method of spelling it is in wide international use, then the model to be most closely copied is the visual appearance of the internationally shared portion of that name as spelled (or respelled) in the Latin alphabet. Take the names of the planets in the five major European languages:
From the appearances of these words, the choices all seem obvious. The only question would appear to be: Should we add /s/ or subtract the few final vowels when there are any? Clearly we should do the latter. The greatest variation in these natural names is precisely in these final vowels when they exist at all. This decision gives us either ?Mercur or ?Merkur for 'Mercury'; Venus for 'Venus'; Mars for 'Mars'; Jupiter for 'Jupiter'; Saturn for 'Saturn'; Neptun for 'Neptune'; and ?Uran or perhaps ?Uranus for 'Uranus'. Only the choices between ?Mercur and ?Merkur and between ?Uran and ?Uranus remain. ?Mercur would appear to imitate the slightly more widespread spelling. On the other hand, the pronunciation of [c] in the [Mercur-] sequence in all the languages that use it is Loglan /k/. This consideration tips the balance, in my opinion, in favor of Merkur. As for Uranus, Uran is clearly the best choice; it is the internationally-shared portion of these five written names, and its sequelae are extremely variable.
But now let us consider the pronunciation of these visually good copies of these low-variance portions of these six international words. The copies are so good, in fact, that in the astronomic or mythological contexts in which they are most likely to appear in print, they will surely remind any scholarly reader of what they mean. But how do they sound? Quite Loglandical, it turns out; (MEHR-koor), (VEN-oos)—not (VEEN-uhs) or (VEIGHN-oos)—(mars)—that's an /s/, not a /z/—(zhoo-PEET-ehr), (SAAT-oorn), (NEP-toon), and (OOR-aan). Only one, Neptun, sounds like the English word. But that is just as it should be. These words are international words. They have, like all the tools of shared scholarship, only international looks, not international sounds. If we in the Loglan community pronounce them in our special way, we are only doing what the rest of the international community of scholars has been doing all along.
MRH: we now prefer use of lao to build such names.
2.13 The Linnaean Polynomials: The Linnaean "binomials"—actually, polynomials because some of them have three parts—are a huge body of zoological and botanical terminology—probably numbering well into the early millions by this time—that have been made standard in spelling by international covenant among biological scientists. The covenant is very simple. Provided a proposed new Linnaean name is unique, then however it is spelled on its first appearance in print (including any adventitious errors) is the way it will then be spelled by all who use it afterwards. Even Japanese and Chinese scholars, including those who use the Latin alphabet in no other way, use Latin letters and Linnaean spellings to represent species and/or genera in the text of their own scientific writing…even if that writing is intended only for domestic consumption.
This stern system is called "Linnaean" after the great Swedish naturalist Carl von Linne who proposed this remarkably enduring system of biological nomenclature several centuries ago. Loglan has no choice but to follow it exactly. It is clear that the names of Homo sapiens neanderthalensis, Escherichia coli and Australopithecus afarensis must be at least written identically in all languages…even if they are pronounced in a great variety of ways. Phonological diversity doesn't matter; what matters is that readers will continue to be presented with the same words. So here the job of The Institute is to find a way of pronouncing these letters that is (1) consistent with the phonology of Loglan, (2) modelled on some natural language way of pronouncing these polynomials that is imitable by speakers of the widest possible group of other languages, and (3) does not—as far as can be foretold by examining reasonably large samples—render any two Linnaean words identically in speech. Once that is done, then a way must be devised for transforming the binomials spoken by loglanists into Loglan text that will not cause us to deviate from the audio-visual isomorphism of the language.
As far as pronunciation is concerned, the habits of Romance-speaking scholars, especially Italian and Spanish ones, commend themselves on all three counts. Except for the pronunciation of [C c], Romance pronunciation is consistent with Loglan phonology—which is a phonological close relative—and it employs, like Loglan, a minimum set of widely used phonemes. Moreover, the Spanish and Italian communities of biologists have been pronouncing Linnaean words for several centuries; and it is unlikely that the pronunciation habits they have developed for that purpose produce homonyms with detectable frequency.
First, we will estimate this Romance-based pronunciation pattern; then in Sec. 14, we will consider how isomorphism in the neighborhood of Linnaean words may best be preserved.
2.14 Pronunciation of the Linnaean Polynomials: Here is our current estimate of how an Italian or a Spanish biologist, with some knowledge of how the French, German, English or Slavic proper names that are often celebrated in the polynomials are "really pronounced", as well as some feeling for the spelling peculiarities of Loglan, would try to regularize the pronunciation of the Linnaean vocabulary:-
We suppose that such a scholar will pronounce every letter and letter combination as in standard Loglan except these:-
Note that, in deference to our Loglan habits, our standard-setting Romance-speaking scholar will pronounce [C c] before [e i y] as Loglan /c/, that is, as written, and as /k/ elsewhere. Thus 'Cephalopoda' is to be pronounced (sheh-faa-loh-POHD-aa), while 'Coelenterata' is pronounced (koh-len-tehr-AAT-aa). This follows half the European custom of pronouncing [c] before "strong vowels" ([a o u]) like Loglan /k/, and before "weak vowels" ([e i y]) like /s/. Instead of /s/ we will ask our standard-setter to use our "native" Loglan /c/.
Double consonants, that is, pairs of identical consonant letters, will be pronounced as if they were a single instance of that consonant.
2.15 Writing Linnaean Names: Once Linnaean names are being pronounced in a standard way by loglanists, then the problem arises of how to furnish Loglan readers with clear indications of which names are Linnaean in a given text. Since some Linnaean names are vowel-final (Homo) and some are not (sapiens), their endings cannot be our clue; and while we are obliged like the rest of the scientific community to italicize, bold or underline Linnaean names in text, even these textual indications are not unique and are in any case difficult to represent unequivocally in speech. So in the interests of preserving isomorphism, a special Linnaean name operator is required. Instead of la we will use laa (LAA-aa) for announcing Linnaean names. We will further specify that laa always be followed by a pause in speech and that Linnaean names never be used without this operator, that is, as vocatives. (So you may not address your visitor as *Hoi Canis lupus even though da is one.)
Moreover, the variable number of terms in these polynomials—sometimes monomials (Australopithecus), most frequently binomials (Escherichia coli or E coli), but sometimes trinomials (Homo sapiens neanderthalensis)—would mean that we could not tell when a V-final breathgroup following a Linnaean term was simply another Linnaean term or a string of regular words. The threat of this ambiguity forces us to treat all Linnaean names as monomials in speech, looking to a separate algorithm to partition these transitional monsters, e.g., Homosapiensneandertalensis (hoh-moh-saa-PEE-ens- neigh-aan-dehr-taa-LEN-sees), Kanislupus (kaa-nees-LOOP-oos), back into their proper, and properly-spelled, polynomial parts. Because of their length we anticipate that some of these transitional monomials will be multiply stressed; but it is doubtful that the existence of these occasional multiple stresses will be more than adventitiously useful in partitioning.
The algorithm that will complete the isomorphism has not been written. And when it is written it will probably involve repeated lookups in extremely long tables. But the principle of that algorithm is clear. Loglan can continue to be isomorphic, even in the region of the Linnaean polynomials, if a machine can be programmed to imitate the performance of the microbiologist (or the ichthyologist, or the botanist, or whatever, who hear Linnaeans from their specialties) who can listen to /eceRIKiaKOLi/ and know that da has heard a Linnaean name that da can confidently type as [Escherichia coli], including the bolding that is now conventional.
We know that such programs can be written; it is just a question of waiting for some competent loglanist with a biological bent to find the time.
MRH: I am much less concerned to incorporate Linnaean nomenclature into the language than JCB was, and I prefer to follow Steve Rice’s insight that lao is better viewed as a general foreign name constructor. This is not to say that this paragraph might not find its way into a future technical manual of the language!
2.16 The Post-Nominal Pause: Many loglanists have misunderstood the requirement that there be a pause after names to mean that, in order for Loglan speech to be understood at all, these "obligatory pauses" must be carefully produced every time. Not so. Sloppy Loglan has about as good a chance of being correctly understood—at least by a sympathetic human listener—as sloppy speech in any natural tongue…which is to say that the chances are pretty good. The difference is that non-sloppy Loglan speech can be impeccably understood by any properly instructed listener…even a machine.
MRH: I think they do need to be carefully produced, and I require in my parser that they always be written.
2.17 Resolving Names: The resolution of a Linnaean name from the speechstream is trivial. One comes upon the Linnaean operator laa followed by its pause; one scans right for the next pause or the end of the utterance; all that lies between the two pauses, or between the pause and silence, will be the Linnaean name. Partitioning the monomial thus generated into its proper Linnaean parts, and respelling those parts in ways that recapture the original spellings exactly, are non-trivial steps but beyond the scope of this book.
The resolution of non-Linnaean names is also non-trivial but quite easy to follow. Before the resolver can resolve names from the speechstream it must know how to resolve regular words from it; for a crucial step in the resolution of most names is the resolver's attempt to resolve as regular words the prequel of some copy of a name-marker that has been found in it; see Sec. 2.6. Thus the algorithm described in this section presupposes the existence of algorithms like those that will be described later for little words and predicates; Secs. 2.33 and 2.59-60. However, our discussion of the name-resolving algorithm can be followed without reference to those later algorithms. All that is required is the suspension of disbelief that they can be written.
Given that a regular-word resolver exists, then, the first step in the resolution of a name is to:
Find the first (or next) breathgroup in the utterance that is C-final. If there are none, then there are no names. (Being followed by obligate pauses, all non-Linnaean names are either right portions of their breathgroups or occupy them entirely. So all and only consonant-final breathgroups have resident names.) Suppose we locate a C-final breathgroup.
Search the first N-2 phonemes of the breathgroup for all copies of the sequences /.i la hoi loa loi sia siu/. If none are found, the entire breathgroup is a name and may be so resolved. (The last 2 phonemes in the breathgroup do not have to be searched; they will always be either part or all of the resident name.)
If one or more copies of the specified sequences are found, select the rightmost and discover whether its prequel will resolve into regular words. If its prequel does resolve, the name commences just to the right of that copy and stretches to the end of the breathgroup and may be so resolved. If the prequel of some copy does not resolve, select the next rightmost copy if any remain and repeat this step until no copies remain.
If no copy with a prequel that resolves into regular words can be found, then the entire C-final breathgroup is a name. Return to step (1) for further breathgroups until all C-final breathgroups have been processed in this way.
MRH: this is not the current situation. There is an obvious need to write up what the current rule is. This presupposes the name marker situation, which no longer exists. In general terms, names are identified by looking for consonant final words which syllabize correctly after name markers or after explicit pauses: the latter case generally occurs in serial names. A sequence which contains a name marker in such a way that a tail of it can also be read as a marked name *must* occur marked by an initial name marker to be read as a name.