halonia : languages & cultures

The Swadesh list is a classic set of 100 vocabulary items that are collected by linguists in different languages to compare the similarities and differences of those languages. The Swadesh list is by no means perfect, but it’s a generally recognised set of basic vocabulary. For different languages you may find that there’s one of two things that don’t quite work, maybe there aren’t many animals with horns, or maybe there are several different words that are used for ‘all’.

So what happens when you try and do a Swadesh list of emoji?

Well, if you needed even more proof that emoji aren’t language, here it is.

Firstly, I was relatively fussy, but not too fussy. Some are great, some are not. For things like ‘bark (of a tree)’ I could have just put tree up again, but the part-whole relationship gets difficult. Colours are there, but on particular objects. There are also very few emoji of people just standing, full-bodied?

And verbs… well verbs are tough. Adjectives are difficult as well, because they’re pretty abstract until applied to something (we know what a ‘good dog’ is, but what’s a ‘good’?). Pronouns are a challenge for emoji because they require context for ‘me’ to mean the particular person speaking at the time, very easy to do with language, difficult to do with emoji.

we used a lot of inspiration from ASL/ signed languages in general.

firstly, pronouns: since emoji are usually used on cell phones or message services like discord, we used ASL style pointing fingers. which ones you use for which person might depend on the program in use, but on discord we used 👈 for me, 👆👇 for you, and 👉 for third person.

all/everything: a) reduplication as a collective/plural indicator ; b) 👐 or 🙌

good, bad: 👍👎

name: 🏷

also handy was our generic verbalizer, 💢, with which we can derive eat 👅💢, bite 👄💢, see 👁💢, hear 👂💢, drink [X]💧/🥛/🍷/☕/🍵/etc💢, and others.

Clam glyphs: if you read this your gay

Suzy Styles of NTU’s BLIP Lab has put together this chart showing the most common speech sounds across languages. It’s set out like the IPA chart, but incorporates features such as unvoiced nasals that are represented with diacritics and not typically included in the main chart. You can see that only a few sounds occur in 80% or more of the world’s languages, and there are many sounds that only occur in a small handful (some of you may think of Zipf’s law when you see this distribution, as only a few sounds are in many languages, and many sounds are only in a few languages).

It’s useful if you’re building a conlang and you want to know how naturalistic it sounds. It’s also handy if you’re learning a language and want to know just how ‘weird’ those ‘weird’ sounds you’re learning are. For example, the trilled ‘r’ of Italian and Spanish [r] is actually much more common than the English ‘r’ sound [ɹ], and the sound at the start of thing in English [θ] is also pretty unusual.

From the figshare page for the chart:

Prevalence rates of speech-sounds across 1672 languages. Data from PHOIBLE Online. Colour scale indicates range from the listed percentage to the next higher percent. This figure first appeared in Styles SJ (2016) ‘Sensory worlds: Multisensory outcomes of sensory tuning to phoneme structure’ Presentation at the 5th Southern African Microlinguistics Workshop, Bloemfontein, South Africa, November 2016. Data source: Moran, S., McCloy, D., & Wright, R. (2014). PHOIBLE Online. Retrieved 2016-10-06, from Leipzig: Max Planck Institute for Evolutionary Anthropology http://phoible.org/

This took place in an Algic language known as Yurok, in northwestern California.  In Proto-Algic, there’s a reduplicative prefix *C©eː- indicating intensive action.  This became *C©e- in an ancestral form of Yurok.  Two crucial sound changes were involved in the reanalysis of this.

The first is the fact that stem-initial /h/ is lost after pronominal prefixes.  Thus, for example, the verb helomey- “to dance”, becomes ʔnelomey “I dance”, with the /h/ lost after the prefix ʔne-.  As a result, the historic stem-initial /h/ rarely (if ever) actually surfaces.  There are no underlying vowel-initial forms in the language.

However, when the reduplication occurs, the /h/ surfaces.  A form starting with hVC- would become hehVC-, but in both cases the initial /h/ is lost after the pronominal prefixes.  Hence, a form such as the ʔn-elomey above would become ʔn-e-helomey.

The second relevant change is that /h/ became /ɣ/ (written g) between vowels.  So that a forms such as ʔnehelomey became ʔnegelomey.  A reanalysis followed where the historically correct division ʔn-(h)e-gelomey was reanalyzed as ʔn-eg-elomey, with -eg- being interpreted as an infix placed after the onset of the first syllable.  This new analysis then spread to other verbs.  Thus, forms such as laːy- become legaːy- in this form.  It also applies to complex onsets, such as ɫkyorkʷ- → ɫkyegorkʷ-.  Not all verbs have adopted this new form, some preserve the original C©(e)-  prefix.  E.g., kelomen- “to turn” becomes kekelomen- “to turn several things”, not *kegelomen-, and ckɹckɹː- “to pierce” becomes ckɹckɹckɹː- “to pierce repeatedly”

A similar phenomenon occurred in Trukese, an Austronesian language spoken in Micronesia.  There is a reduplicative prefix CVC- used with most verbs indicating pluractionality (at least, that’s how Yu labels it, the examples he gives seems to be more of a general habitual or imperfective).  Thus, fætæn “walk” becomes fæf-fætæn “be in the habit of walking”, mɔt “sit” → mɔm-mɔt “be sitting”, etc.

However, in vowel initial verbs, the form is Vkk-, and in w-initial verbs, it’s an infix -Vkk-.  Thus, wɨn “drink” → w-ɨkk-ɨn “be in the habit of drinking” and ɔsɔmʷoːnu “pay chiefly respects to” becomes ɔkk-ɔsɔmʷoːnu “be in the habit of paying chiefly respects to”.

This reflects a historic phenomenon wherein word-initial /k/ was lost.  Thus, for example, that last example was, in an earlier stage, kasamʷóːnu.  The /k/ was lost in that form, along with various vowel changes, to produce ɔsɔmʷoːnu, but when the CVC- prefix was added, that /k/ became word-medial, and was thus preserved, while the initial /k/ of the prefix was lost.  Thus, kak-kasamʷóːnu became ɔkkɔsɔmʷoːnu.  Since that original /k/ was lost, the historic division ɔk-kɔsɔmʷóːnu would’ve been obscured, and it was reanalyzed as ɔkk-ɔsɔmʷóːnu. This new analysis was then spread to other vowel-initial verbs, so that all vowel-initial verbs used a prefix Vkk-, whether or not there was an etymological /k/ (it’s unclear from Yu’s description how original vowel-initial verbs form the pluractional before this reanalysis, or if they perhaps simply were unable to be inflected in that form at all)

Wiyot syllables always begin with consonants or consonant clusters, which are followed by a vowel. This vowel may be long or short. If the vowel is short, the syllable must end in the same consonant that begins the next syllable. Therefore, all non-final syllables are heavy, acquiring either a CVV or CVC structure. Word final syllables may or may not be heavy.

These syllable-final consonants are lengthened in speech, but do not appear as doubled letters in transcription. For example, in the word palógih, meaning ‘flounder’, the 'l’ is lengthened. Thus, the first syllable ends with 'l’, and the second begins with 'l’, and both syllables are considered heavy.

It depends on which system of ranking you use, of course (there are several), but in most systems there’s actually a rule covering this scenario: in the event that a consort’s courtesy title being of the same rank as their spouse’s would potentially create confusion over who holds the title by right and who by courtesy, the consort instead receives the next-highest title on the ladder.

So the husband of a prince would be a duke; the wife of a queen, a princess; and so forth.

(You actually see this rule in practice in the United Kingdom, albeit not in the context of a same-sex marriage; the Queen’s husband is styled a prince because if he were a king, folks might get confused about which of them was the reigning monarch.)

The only common situation where you’d expect to see, for example, two queens in the same marriage is if the reigning monarchs of two different realms married each other - and even then, you’d more likely end up with a complicated arrangement where each party is technically a princess of the other’s realm in addition to being queen of her own.

1. The system described above is, admittedly, merely one of the most common. Other historically popular alternatives include:

The consort’s courtesy title is of the same rank as their spouse’s, with “-consort” appended to it: prince and prince-consort, queen and queen-consort, etc. This is how, e.g., present-day Monaco does it.

The consort is simply styled Lord or Lady So-and-so, and receives no specific title. I can’t think of any country that still does it this way, off the top of my head, but historically it was a thing.

(Naturally, your setting needn’t adhere to any of these, but it would be highly irregular for it to lack some mechanism for clarifying the chain of command.)

2. The reason why the consort of a prince is historically a princess even though those titles are the same rank is basically sexism. This can go a couple of ways:

In many realms, there was no such thing as being a princess by right; the daughter of a monarch would be styled Lady So-and-so and receive no specific title, so the only way to be a princess was to marry a prince.

In realms where women could hold titles by right, typically a masculine title was informally presumed to outrank its feminine counterpart. So, e.g., kings outrank queens, princes outrank princesses, etc.

In either case, no ambiguity exists.

(Interestingly, this suggests that in a more egalitarian setting where masculine titles are not presumed to outrank their feminine counterparts, or vice versa, you’d need to explicitly disambiguate rankings even outside the context of same-sex marriages. Food for thought!)

3. It would also be possible to have two kings or two queens in the same marriage without multiple realms being involved in the case of a true co-monarchy. However, true co-monarchies are highly irregular and, from a political standpoint, immensely complicated affairs. If you’re planning on writing one of those, be prepared to do your research!

4. The next rank down from “countess” is either “viscountess” or “baroness”, depending on which peerage system you’re using.

The author discusses how the primary account given for the development of grammatical gender/noun class involves the grammaticalization of generic nouns/classifiers or demonstratives, with scholars allowing that the extension of such systems could proceed in different ways. And, course, PIE developed the feminine in such a different way - as many of you might know, it developed when the suffix -h₂, used to mark collective and abstract nouns, became so ingrained and widespread that it changed the inflectional patterns of the nouns it modified and began to trigger agreement on demonstratives, and then, out of analogy, on adjectives as well. (In fact, it may have contributed to the development of adjectives as a distinct word class.)

But the animate-inanimate system that preceded that clearly didn’t develop from generic nouns or classifiers or demonstratives either:

The scenario sketched in (3) cannot account for the rise of the PIE animacy-based two-gender system, in which no grammaticalization of generic nouns or demonstratives, and indeed no specific gender markers are involved. As well known, the difference between the two earliest genders of PIE, animate and inanimate, was indicated by absence of endings for nominative and accusative in the inanimate gender (Meillet 1921).

Which tbh for me is just like “hm okay I didn’t know that” (I’m no expert on IE case marking). But basically the author says that this difference was due to animate nouns being more highly individuated, thus taking case marking that clearly indicated agency, while less-individuated inanimate nouns did not.

I’ll just quote two sections where the author discusses her conclusions:

In the preceding sections I argued that gender systems can arise in at least two completely different ways. In the first place, they may arise from earlier systems of nominal classification, which become increasingly obligatory, and start triggering agreement, most likely starting with demonstratives or other pronouns. I call this type of process ‘gender from above’; it is described in (3), and called ‘lexico-semantic’ by Fodor (1959). In the second place, genders can arise from special patterns of case marking, following a development which I call ‘gender from below’. Such a process does not in principle imply the creation of gender markers, as it did not in the PIE two-gender system; it corresponds to the morphologically or syntactically motivated gender systems of Fodor (1959).

And:

I examined the possible origin of gender systems, and argued that gender systems can arise in two quite different ways, either from the grammaticalization of classifiers (gender from above), or from the establishment of agreement following different morpho(syntactic) behavior of groups of nouns (gender from below). Crucially, non-sex based gender systems with more than two genders seem to possibly arise only from former systems of classifiers. I argued that gender systems also have different primary functions depending on their origin: while genders ‘from above’ serve a classificatory function in the first place, genders ‘from below’ primarily fulfill the function of providing a means for referent tracking. For this reason, they tend to be sex-based, since male and female humans are equally discourseprominent and topic-worthy entities.

To explain why gender systems would be so typically so much smaller than classifier systems, you’d have to assume a rather drastic reduction in their sizes, either before becoming gender (which would raise the question of where all the 2- or 3-classifier languages are), or after becoming gender systems.  But if gender systems lost members fast enough to explain the relative rarity of 4+-member systems, then why would it suddenly stabilize when it hit 2 or 3?  And why would the large gender systems in families such as Bantu be stable over long periods of time?

tán unni taa tán aneilomi aliʃaipé tatikuwt w tapoós koonuy ittóniʃttsobih naab ya miwqheqohmt naab ya piigmné naab ya «han» poótalut paga tahállt osuóg w kéen pihat naunaab aun kamiál canayyata kamiál cagienellta ya wi óɂuhptana. tulq miw kaplaw atatsó tán lápiip ten w ɂip miáati ten  ya wi ɂukɂúhpmaqow.

A discovery made in a remote mountain village high in the Peruvian Andes suggests that the ancient Inca used accounting devices made of knotted, colored strings for more than accounting.

The devices, called khipus (pronounced kee-poos), used combinations of knots to represent numbers and were used to inventory stores of corn, beans, and other provisions. Spanish accounts from colonial times claim that Inca khipus also encoded history, biographies, and letters, but researchers have yet to decipher any non-numerical meaning in the chords and knots.

Now a pair of khipus protected by Andean elders since colonial times may offer fresh clues for understanding how more elaborate versions of the devices could have stored and relayed information.

Anthropologist Sabine Hyland studies a khipu board, a colonial-era invention that incorporated earlier Inca technology.

“What we found is a series of complex color combinations between the chords,” says Sabine Hyland, professor of anthropology at St. Andrews University in Scotland and a National Geographic Explorer. “The chords have 14 different colors that allow for 95 unique chord patterns. That number is within the range of symbols in logosyllabic writing systems.”

Hyland theorizes that specific combinations of colored strings and knots may have represented syllables or words. Her analysis of the khipus appears in the journal Current Anthropology.

SECRET MESSAGES

Hyland made her discovery in the Andean village of San Juan de Collata when village elders invited her to study two khipus the community has carefully preserved for generations. Village leaders said the khipus were “narrative epistles about warfare created by local chiefs,” Hyland reports.

The khipus were stored in a wooden box that until recently was kept secret from outsiders. In addition to the khipus, the box contained dozens of letters dating to the 17th and 18th centuries. Most of the documents are official correspondence between village leaders and the Spanish colonial government concerning land rights.

Spanish chroniclers noted that Inca runners carried khipus as letters, and evidence suggests that the Inca composed khipu letters to ensure secrecy during rebellions against the Spanish, according to Hyland.

A khipu from the Andean village of San Juan de Collata may contain information about the village’s history.

“The Collata khipus are the first khipus ever reliably identified as narrative epistles by the descendants of their creators,” Hyland writes in her analysis. She notes that they are larger and more complex than typical accounting versions, and unlike most khipus, which were made of cotton, the Collata khipus were made from the hair and fibers of Andean animals, including vicuna, alpaca, guanaco, llama, deer, and the rodent vizcacha.

Animal fibers accept and retain dyes better than cotton, and so they provided a more suitable medium for khipus that used color as well as knots to store and convey information.

In fact several variables—including color, fiber type, even the direction of the chords’ weave or ply—encode information, villagers told Hyland, so that reading the khipus requires touch as well as sight.

Hyland cites a Spanish chronicler who claimed that khupus made from animal fiber “exhibited a diversity of vivid colors and could record historical narratives with the same ease as European books.”

THE BIG QUESTION

The Collata khipus are believed to date from the mid-18th century, more than 200 years after Spanish colonizers first arrived in 1532. This raises the question whether they are a relatively recent innovation, spurred on by contact with alphabetic writing, or whether they bear a close similarity to earlier narrative khipus.

“These findings are historically very interesting, but time is a big problem,” says Harvard anthropologist Gary Urton. “Whether or not we can take these findings and project them into the past, that remains the big question.”

A few years ago, Urton and Peruvian archaeologist Alejandro Chudiscovered a trove of khipus in what may have been a khipu workshop or possibly a repository of Inca records.

[VIDEO AT NAT GEO WEBSITE]

Deciphering patterns hidden within the devices may eventually become the work of computers, Urton says. He and his Harvard colleagues maintain a digital repository called the Khipu Database that categorizes images, descriptions, and comparisons of more than 500 of the artifacts.

The Inca at their height may have made thousands of khipus, perhaps even hundreds of thousands. But archaeologists suspect that natural deterioration and European colonizers destroyed most of the devices. Fewer than 1,000 are known to exist today.

Hyland plans to return to Peru in July to resume her research. Last summer, on her last day of fieldwork, she met an elderly woman who said she remembered using khipus as a young girl. But before Hyland could ask more questions, the woman darted away to tend to her livestock.