it turns out it's quite hard to loan words into Lojban. while for a natural language, or for most other conlangs, you just have to make phonotactical transformations so the word fits the syllable structure of the language, in Lojban you have to make sure the word also has the correct morphological shape, by guaranteeing that it has a consonant cluster near its beginning.
this means, for example, that things that are almost universally considered "easy to pronounce" or "easy to loan" across natural languages, such as Japanese sakura or the words bonobo, gecko, rap [the music genre] and melatonin, none of which have any consonant clusters, have to gain consonant clusters when they are Lojbanized:
tsakura cherry blossom [← JA sakura]
jbonobo bonobo
dalgeko gecko
ra'apxu rap (yes, seriously)
meltoni melatonin
sometimes this Lojbanization is done almost seamlessly, in a way that does, of course, distort the source word, but the distortions, if any, are unintrusive and "nice". (there is of course some degree of personal opinion on what is a "nice" distortion. these are just my guidelines.)
meltoni, for example, manages to maintain the proper morphology while still sounding very nice, and also fitting well among native Lojban forms (like sorpeka and nonseka). dalgeko is also completely fine, it's a great word — the dal- part is from danlu 'animal' — and so is tsakura, whose ts- I don't find that bad.
jbonobo, on the other hand, is just silly. the "epenthetic" j at the beginning of the word does not go unnoticed at all; worse, it makes you think that a jbonobo is some sort of Lojbanic monkey. similarly for ra'apxu, which would be completely unrecognizable if I didn't tell you that it means 'rap'.
the goal of this blog post is to show that it's possible to coin "good" loanwords, that aren't too far from their originals, but also fit within the phonaesthetics of native Lojban words most of the time. so, I'll go starting from the most unproblematic, easily-loaned examples, and move on to harder and harder words as we go on.
and yes, this article features vocaloid. after all, I made the words voksaroido 'vocaloid' and kanbaji 'pinning badge', which is the perfect excuse to show you this Kasane Teto pin I got the other day:
pe'i le vi kanbaji pe la tétos cu zabna sai je cimle sai
the recommended Mati soundtrack to listen to while reading this article is these two songs, obviously sung by Kasane Teto:
Tarinai Binary 足りないバイナリー (YT), by Honemaru, and
Pure-Nō ピュア脳 (YT), by Taidasaki (or is that Taida Saki, two words? if only Japanese had spaces.)
1. picking a source that loans well
making a good loanword starts with picking a good source for it. some words are specific cultural products originating in one place, and have the same name practically everywhere, so for those, there isn't much choosing of source words to do. an example of this is spageti — except for Chinese, which calls them 意大利面 yìdàlìmiàn 'Italian noodles', the word for them is spaghetti pretty much everywhere. the same thing applies to cidjrmotci 'mochi', or to 'tamale', 'açaí' and 'sashimi', if we were to make words for those.
most other names of things, including most plants and animals, though, have different names in different languages, and sometimes those names will be better or worse at adapting to Lojban morphology. a good example of this is 'carrot'. there are so many words for carrot on la jbovlaste, and some of them are better than others. other than the Linnaean-derived prefixed ones (genjrdauko and its variant gejrdauko), here's all of them:
ka'orta
gulrota
najgenja [a compound, 'orange-root'. not a loan.]
porkana
morkova
ŭortele
károt zei spati [are you kidding me?]
sargarepa
each of these words (the loanwords, at least, not the compound) has their etymology in a different language:
ka'orta is from Latin carota or its descendants
gulrota is from Norwegian gulrot
porkana is from Finnish porkkana
morkova is from Russian морко́вь morkóv' or one of its cognates
ŭortele is from Dutch wortel
sargarepa is from Hungarian sárgarépa, transcribed incorrectly (Hungarian ‹s› is [ʃ] and should be Lojbanized c)
out of all of these, sargarepa is bad because it has a transcription mistake [and also it's just longer], and ka'orta is bad because it distorts its source when better options are available (this jumbling of consonants and especially adding random apostrophes is genuinely a last-resort effort, not to be used unless no other options work).
this leaves us with gulrota, porkana, morkova, and ŭortele. all of these four are very nice words, and their shapes are generally true to their sources. they also sound very Lojbany, and fit in with native Lojban forms — except for maybe ŭortele, with its initial glide.
out of all those four, my preference is for morkova (maybe this is familiarity bias; I have studied Russian in the past). but all of them, especially the first three, are perfectly good Lojban loanwords, which both sound like their sources without too much distortion, but also sound "like Lojban words", not too divergent from the phonotactics of the native vocabulary of roots and compounds (both proper compounds and clipped ones).
some more examples of thoughtful source-picking
(if a word in this article has an asterisk after it, it means that I made it. after all, these are my own "rules", and sometimes it's easiest to illustrate them with my own examples.)
kampaneto 'glockenspiel', from Esperanto kampaneto
glokenspile, or something of that sort, would be a compound ('Loglan-cancer-spile'!), so loaning that wouldn't have worked. distorting it to glokespile would, but the Esperanto word just happens to be exactly the right shape, so something like that tends to be the better choice.
impisi 'hyena', from Zulu impisi
the Latinate hyena would need a lot of distortion to become a valid word (xi'enma?), but the Zulu word simply loans perfectly.
ankabuta 'spider', from Arabic ʕankabūt (I really wish I could change the font to one with better IPA support on here)
the usual go-to for animal and plant names in Lojban is Latin, but that would also not work here, as the Latin word arānea has no clusters.
in the specific case of ankabuta, English spider (as spaida) would also have worked just as well, but the loan vocabulary already has a lot of Latin and English. some of this is because of sheer recognizability (Latin morphemes are, in a way, the language of science; English is the most spoken language in the world), but considering sources from outside the "usual suspects" — like Arabic for ankabuta, Zulu for impisi, Hindi for galtuni 'tonsil' [from गलतुंडिका galtuṇḍikā] can often give better-shaped words, as well as being in line with the Lojbanic ideal of cultural neutrality, as it makes the sources of our loans more diverse!
2. a little distortion is fine — just don't do too much
sometimes, though, picking a "good source" for your loanword just won't work. you have to work with the options you have — this is especially the case either for spageti-type "cultural product" words, or sciencey Greco-Latinate terms, which tend to coincide in many (but not all) languages. also, even when multiple natlang options are available, it's somewhat common to try to adapt a scientific Latin name when it has a good shape.
in those cases, if the option you have fits the shape of a Lojban word, you can use that: troxili is 'hummingbird' (short for Trochilidae), and a nimfaĭa is a water lily, from Nymphaea. sometimes pronouncing things differently, as different languages would do when loaning in Latin, helps with finding a word with better morphology. this is often the case with initial soft c and g, which can be loaned as tc (as in Italian) and dj (as in, well, English) to secure a cluster at the beginning of your loanword.
this is what is done with tceraso 'cherry' (from Cerasus) and djirafa 'giraffe', for example. this is what I did when I coined tcikero 'chickpea, garbanzo' — ×kikero isn't valid, but I felt like transcribing both c's as soft (tcicero) would make the word have too much c (which Lojban already has, mind you!), so the first c was transcribed the Italian way and the second one was transcribed as k. (the tcik- rendering was also inspired by the chick- in English chickpea.)
here's some more examples of creative, natlang-informed "distortions"/alternative renderings:
sekfoĭa* 'sequoia': my own attempt at making a word that's less distorted than tsekuŭoĭa, but also shorter than ku'ursekoĭa (with no pseudo-prefix). the rendering of ‹qu› between vowels as kf was an idea by Zearen, but I adopted it as a /kf/ pronunciation in these kinds of words is found, for example, in Polish.
tcitani* 'titanium': yet another shorter, prefixless version of a previously existing word (jinmrtitani). the tc at the beginning is from the pronunciation of titânio in most dialects of Brazilian Portuguese, and also from Japanese チタン chitan.
in general, initial ‹ti di› can be turned into tci dji in Latinate words to guarantee a cluster if one is needed, inspired by Portuguese pronunciation (and sometimes Japanese, if the word is present there). similarly, initial ‹z› (and ‹x›, which in some languages like English is /z/) → dz is another possible emergency repair — if I made a (prefixless) loan for xenon, I would make it dzenoni*.
tcikoni 'stork' (var. tcikona), from Ciconidae: yet another example of the "Italian" pronunciation of c being used to repair a word.
yet another word for 'stork' derived from this is on JVS, kiknoĭa, but kiknoĭa is too distorted — an example of what to avoid.
most of the time though, the "alternative consonant pronunciation" trick doesn't yield a good shape. (it's mostly restricted to things like ‹ce ci ge gi ti di›, initial ‹z x›, deaffrication of ts dz tc dj, and a few other oddballs.) in these cases, often a good choice is deleting the second vowel from the source word, as in:
meltoni melatonin
tokpona Toki Pona
pirmidi pyramid
almini* aluminum
foldota pangolin [← Pholidota]
aclotlu axolotl [the x is [ʃ] in the Nahuatl source]
ĭunkoda Unicode
another option, which isn't as ideal, but is still not very drastic, is taking the sonorant (that is, l, r, m, or n) in the second syllable of the word, if there is one, and turning it into lr, rl, mn, nm (or maybe ln) to form a cluster:
serlaxi shark [← Selachii(formes)]
lilneĭa Linnaean
xunmusu hummus [the m being geminated in Arabic does give some legitimacy to the -nm- here, even]
3. distortions you should avoid
I just showed some examples of mostly seamless distortions, but there are also distortions that should be avoided. other than scrambling around the letters too much, as in kiknoĭa, there are two main "loanword sins": bad intrusive consonants and random apostrophes.
bad intrusive consonants
it's not too uncommon for bad word coiners to add random consonants to their words for proper morphology. if it's done with a sonorant, as in xunmusu, this is usually fine, but for anything else it's usually the case that the extra consonant sounds glaringly intrusive inside the word, compared to its source.
this is most common when the consonant is an s, or especially a z, c or j added to the beginning of a word. an example of this, which I've already mentioned before, is turning bonobo into jbonobo, with a j that doesn't at all blend with the rest of the word. some other examples are:
jbatata sweet potato [← ES batata]
sfaile (computer) file; this one isn't as bad, as sf sounds less outlandish than jb or zg
zge'eno Zen [Buddhist]; this one also has a gratuitous apostrophe, a problem we'll see later
jvegana vegan
cketake cetacean (eg whale); the better alternative tcetace, using the Italian trick I mentioned, is also on JVS
in 99% of cases, avoid these kinds of distortions. if a word can't be saved by deleting vowels, add a pseudo-prefix to it. if that doesn't work, try to tweak one of its sonorants (m → nm, etc) instead, but inserting random c and j and such at the beginning of words should be avoided 99% of the time, basically.
(I have made this mistake before once: jgauzi* 'jiaozi, gyoza', which would have been far better as djauzi, since, well… that j in Mandarin is an affricate, Mati. what were you thinking with jg-? please ignore that word forever, and use djauzi* instead.)
gratuitous apostrophe adding
another way to "patch" words that is somewhat common among the coinages you see on JVS is randomly adding apostrophes to words so that they aren't shaped like roots or compounds, as in tre'endi 'trendy', skli'iparte for… uh, let me check, what even is that? oh, it's 'clip art', and vo'otka for 'vodka'.
in general, words like this are generally fixable by deleting one of the consonants (klipato* for clip-art, getting rid of the r), or otherwise by picking a different source, or adding a syllable to the end (tre'endi could be trendino*, with the -in- from the -ing in trending). for the vodka case in particular, I made ŭotka (since Russian doesn't have a phonemic /v/ vs. /w/ distinction anyway.)
(a similar thing happens with adding random glides, as in crimpŭui for 'shrimp', which is just an objectively worse way to say kreveta. anyway, please don't coin these, we've had enough of these "lo'oanwords". they just sound bad!)
4. pseudo-prefixing
sometimes, words of a certain shape just refuse to fit the Lojbanic morphological mold, because you didn't manage to find a good source, or deleting the second vowel didn't work. when this fails, you can usually look at adding a pseudo-prefix to your word (also known in more "cmevti and crimtu" jargon as a pseudo-stage-3 fu'ivla).
(I should also mention the fact that there's the option of making actual prefixed loanwords with syllabic consonants ["stage-3s"]. those are far easier to make though, as none of the morphological traps apply. in general, though, I'd avoid stage-3s for anything that isn't an obscure scientific term, a very specific cultural product, or sometimes a culture word that doesn't have a short root. they can often be the only way to fix root-shaped source words, though, as in cidjrmotci and cidjrpitsa — as djapitsa looks like a compound, and ×djarpitsa is a particle thief.)
I've already shown one of these words at the beginning of the article, dalgeko. the -geko part is just from gecko, but the dal- part is from the combining form of the word danlu, animal. this is not a real prefixed loanword (that would be dalrgeko ~ danlrgeko, with a syllabic r), but it emulates the logic of those. some other examples include:
dalko'ati coati
spakiti broom plant (from Cytisus); spa- is from spati plant
dalmustela weasel (from Mustela)
cpikiŭi kiwi (bird; cpi- from cipni, bird), and jbakiŭi kiwi (the fruit, jba- ← jbari berry)
djatako taco (dja- ← cidja food); djakari curry, djatofu tofu, and so on follow the same pattern
since the prefixes in these kinds of words are not actual loanword prefixes, sometimes using a prefix that is not the actual combining form of the word can yield better results, although these types of words are moreso blends than prefixed loans. if the source allows for it, though, a blend can often be a great option:
zgilofoni xylophone, which blends -zgi- from zgike 'music' into the word — at the same time that it has zgi- as a pseudo-prefix, it also has zg- as a metathesized version of the /gz/ in, for example, French xylophone /gzilɔfɔn/.
kibzvaipi* … has IP address …, from kibzva 'be located online' and the English pronunciation of IP
voksaroido* vocaloid, from voksa 'voice' and Japanese ボーカロイド bōkaroido
kranbari* cranberry, from cranberry and jbari. in this case, kranberi would have worked just fine, but I decided to incorporate the Lojban word into it to make the word more… "fun", so to say
(you can see from almost all examples having an asterisk that 1. I really like blends, and 2. I can't really remember words that I didn't make myself. u'u!)
5. native freewords
since this article is about loaning words from other sources, native freewords (sometimes called jbofuvi, itself a native freeword, short for jbojbofu'ivla 'Lojban-Lojban loanword') aren't really part of the scope here, but I thought it was important to mention them anyway.
(I'll probably talk about native freewords in more detail in another post, about whether you should even loan specific words to begin with. sometimes the best option is simply making a compound, because God gave us rafsi so we could make lujvo.)
a native freeword is a word that uses the morphological freeword (zi'evla) or "loanword" shape, but instead of being borrowed from another language, is made from Lojban elements. most of them are shortened from longer compounds, but others are simply deformed roots. some examples are:
sorpeka 'bus' from sorprekarce, literally 'many-person-car'
pavjima 'unicorn', from pav- 'mono-' + jirna 'horn' + xirma 'horse'
torveki 'summary' (tordu ve skicu), stiseji 'consultant' (stidi se jibri), and quite a few others. this seven-letter shape (either CCVCVCV [like pavjima] or CVCCVCV [like stiseji]) is very common for these types of words.
one example of a native freeword I myself coined (with the help of Salarua in the Nijbo Discord server, who has recently been coining Lojban medical terminology) is gamdegi, meaning 'duodenum', the first part of the small intestine. the word duodenum itself comes from Latin duodēnum digitōrum 'space of twelve digits' (i.e. 12× the width of a finger), so I joined -gam- (from gai 'twelve') and -deg- (from degji 'finger') into a freeword shape, and made gamdegi.
native freewords can often be a good choice when none of your sources adapt very well, or just because they're clearer — they may be irregularly made, but they are made of Lojban elements, which makes them easier to remember, and also, they're just fun!
6. jokes are okay sometimes
there is somewhat of a tradition of making joke words in Lojban, or generally words with clever puns. the Lojban term for these is u'ivla — literally 'lolword' — itself a pun on fu'ivla and the interjection u'i. sometimes the jokes are absurd and create words that are way too long (as in pudlu'avalsi'ipo'ato, a garden-path word that means 'garden-path word', because it looks like a compound until you get to the -to at the end), but other times, a joke word can be both clever and usable. here's two of my favorites:
ci'onme'a 'red panda': the word for 'giant panda' is ci'onmau, from Mandarin xióngmāo. the -mau at the end is reminiscent of compounds like bramau 'larger' and nadmau 'more difficult', so if nadmau is 'harder' and nadme'a is 'less hard' (i ĭa nadme'a fa la itku'ile a'enai ♪), and if a giant panda is a "xiong-more", then we can also have "xiong-less", hence ci'onme'a.
xlodu'uxukau 'verbose, overwrought': this one comes from lo du'u xu kau 'whether' — something Lojban takes way too many syllables to say (unless you simply say lo jei). it's also a form of self-mockery, since this word started out in Toaq (loduhuhukao) to make fun of Lojban, and was then adapted into Lojban! (Toaq also has loseduhu, making fun of lo se du'u.)
generally, trying to make joke words should be used sparingly. but if you find an opportunity to make a clever joke word that isn't too long or generally too "forced", it can often be a nice option to pick! it adds color to the vocabulary of a language that is often thought by outsiders to be spoken by humorless pedants.
woah. I think I've never written anything this long before! the articles on my old blogs were all quite short, and I'd usually write them in one sitting — although some did take one or two days, give or take.
anyway, that's all for today. thanks for reading, and have a nice day!