Zompist Bboard Again

Posted: **Sat Mar 18, 2023 12:24 pm**

Raphael wrote: ↑Thu Mar 16, 2023 4:59 am
bradrn wrote: ↑Thu Mar 16, 2023 4:30 am Did it work?
Yes, thank you. I wanted to know whether the ZBB can properly show exponents.

I notice you used a superscript Unicode character rather than a BBCode superscript like this: E=mc².

Posted: **Sat Mar 18, 2023 12:29 pm**

Several times over the past few days, I have read claims that Trump anticipates being arrested in the coming weeks. Given that, it sounds like he has all the warning needed to flee the country before facing formal charges that would make it impossible for him to appear publicly. Indeed, if facing charges at the state level, can't he simply avoid them by staying out of that state?

Posted: **Sat Mar 18, 2023 12:33 pm**

malloc: I think there's a dedicated US politics thread.

Travis B. wrote: ↑Sat Mar 18, 2023 12:24 pm
I notice you used a superscript Unicode character rather than a BBCode superscript like this: E=mc².

Yes, sorry about that. I didn't know about the [ sup ] tag. Thank you for telling me!

Posted: **Sat Mar 18, 2023 12:46 pm**

Completely different topic: what is the current scientific consensus on how old human kind is as a species? I know, of course, that ultimately any decision made by paleontologists about at which point in the past our ancestors start to be "officially" counted as homo sapiens is arbitrary, but I'm curious where most of them draw the line these days. I've seen numbers ranging from 30,000 BCE to 200,000 BCE.

I'm mainly asking just in case I want to calculate how long any given period in history is as a share of all of human history (and prehistory).

Posted: **Sat Mar 18, 2023 1:51 pm**

300,000 BC is the consensus now. 30,000 BC is way too late.

Posted: **Sat Mar 18, 2023 2:00 pm**

WeepingElf wrote: ↑Sat Mar 18, 2023 1:51 pm 300,000 BC is the consensus now. 30,000 BC is way too late.

Thank you!

Posted: **Tue Mar 21, 2023 6:34 am**

Is there any place that has reliable information on facial expressions in cats, and to which extent they do or don't mean similar things as similar facial expressions in human beings?

All those photos of cats looking sceptical, very sceptical, decidedly unimpressed, or outright shocked got me curious about that question.

Posted: **Tue Mar 21, 2023 7:02 am**

Raphael wrote: ↑Tue Mar 21, 2023 6:34 am Is there any place that has reliable information on facial expressions in cats, and to which extent they do or don't mean similar things as similar facial expressions in human beings?

Cats don't really have a wide range of facial expresions. It gets down to various degrees of having the mouth and eyes open or closed. Different cats have different constant (or resting) faces, which can be used for funny pics.

Posted: **Tue Mar 21, 2023 7:42 am**

I haven't hung out with cats much in years, but from what I can see there's a lot more *pics* of cats making funny faces than *videos* of cats making funny faces, which suggests to me that, likely, it's a matter of the face being sort of.... accidental? maybe the cat's sneezing or smelling something or in the process of licking something off his teeth or something.

Posted: **Tue Mar 21, 2023 12:57 pm**

I asked GPT4 what kind of conlang it would make, and this is what it suggested:

GPT4 wrote: As an AI language model, I don't have personal preferences or desires, but I can suggest an interesting set of features for a conlang based on various linguistic phenomena and structures found in natural languages. Here's a proposal for a conlang with some unique features:

Phonology: Choose a balanced set of consonants and vowels, but include some rare phonemes or phonological processes, such as ejective consonants or tone systems.

Morphological typology: Create a language that is primarily agglutinative, where words are formed by combining multiple morphemes with distinct meanings. This can make the language highly expressive and regular in its morphology.

Word order: Opt for a less common word order, such as VSO (verb-subject-object) or OSV (object-subject-verb), to distinguish your conlang from more widely-spoken languages like English or Spanish.

Nominal morphology: Implement an extensive case system to mark noun phrases for their grammatical roles, similar to Finnish or Hungarian, and/or include various types of noun classes or gender systems, as seen in Bantu languages.

Verbal morphology: Incorporate a rich tense-aspect-mood system with various combinations of tense (past, present, future), aspect (perfective, imperfective, progressive), and mood (indicative, subjunctive, imperative) markings.

Evidentiality: Include evidential markers that indicate the source of information or the speaker's certainty about a statement, a feature found in languages like Turkish or Tuyuca.

Polypersonal agreement: Use polypersonal agreement, where the verb agrees with multiple arguments, such as the subject, object, and/or indirect object, as seen in Basque or Georgian.

Incorporation: Allow noun incorporation, where a noun is combined with a verb to create a new verb form, as in Mohawk or Nahuatl.

Syntax: Employ non-configurational syntax, where the order of constituents is relatively free, and grammatical relations are marked by morphology or pragmatic factors, as in Warlpiri or Hungarian.

Unique features: Invent a few unique features or structures that are not commonly found in natural languages, such as a novel way of expressing negation, possession, or relative clauses.

Remember, these are just suggestions to create an interesting and unique conlang. You can modify any of these features or incorporate additional ones based on your preferences and goals for the language.

Posted: **Tue Mar 21, 2023 1:18 pm**

Ok, by now I should know that these bots know all kinds of things about all kinds of topics, but I still seriously wonder how on Earth it learned that stuff.

Posted: **Tue Mar 21, 2023 1:22 pm**

Raphael wrote: ↑Tue Mar 21, 2023 1:18 pm Ok, by now I should know that these bots know all kinds of things about all kinds of topics, but I still seriously wonder how on Earth it learned that stuff.

Its understanding is shallower than the text suggests though. I've asked it a number of basic typological questions and it often claims examples demonstrate things they don't etc. It's memorised a basic fact and an example or explanation (or at least can generate the sequences of tokens), and there's enough tokens in common that it thinks the two things go together, only... the whole isn't consistent.

People overestimate how much of the output of these models is reasoning Vs just doing a kind of memorisation based on a vast corpus. The issue is then that regurgitating the bits that seem like they might go together sometimes works and sometimes doesn't.

Posted: **Tue Mar 21, 2023 3:33 pm**

I see that GPT4 is a fan of the kitchen sink language. Noob!

(I'm not sure, but I think extensive case marking is not characteristic of the polysynthetic languages it's recommending.)

Posted: **Tue Mar 21, 2023 3:59 pm**

zompist wrote: ↑Tue Mar 21, 2023 3:33 pm I see that GPT4 is a fan of the kitchen sink language. Noob!

I thought kitchen sink stuff was the second step in the evolution of a beginning conlanger, after the complete noob step of ciphering one's first language?

Posted: **Tue Mar 21, 2023 4:02 pm**

That recommandation looks a little like Quechua?

Posted: **Tue Mar 21, 2023 4:13 pm**

zompist wrote: ↑Tue Mar 21, 2023 3:33 pm I see that GPT4 is a fan of the kitchen sink language. Noob!

(I'm not sure, but I think extensive case marking is not characteristic of the polysynthetic languages it's recommending.)

Not necessarily? There are plenty of agglutinative synthetic languages with both polypersonal agreement and lots of cases. Basque, which it mentions, is one, although Basque lacks other features like productive noun incorporation.

Other possibilities: Yupik, Dalabon (polysynthetic, 7 case suffixes), Koryak, Adyghe (doesn't have a large case system, but some kind of erg/abs/obl system) etc.

EDIT: I asked it to name some natlangs that had the features on its list and it said:

GPT4 wrote: It's rare for a single natural language to have all of the features listed above, but some languages exhibit a significant number of them. Here are a few examples:

Basque: An agglutinative language with a rich case system, polypersonal agreement, and relatively free word order (though typically SOV). It lacks evidentiality, noun incorporation, and tone systems.

Georgian: A Kartvelian language with an extensive case system, polypersonal agreement, and a relatively free word order (with SOV as the most common order). It does not feature evidentiality, noun incorporation, or tone systems.

Tuyuca: A Tucanoan language spoken in the Amazon rainforest with a rich tense-aspect-mood system, evidentiality markers, and noun classes. It does not have polypersonal agreement, noun incorporation, or an extensive case system.

Mohawk: An Iroquoian language with noun incorporation, a polypersonal agreement system, and a relatively free word order. It lacks an extensive case system and evidentiality markers.

Nahuatl: A Uto-Aztecan language with noun incorporation, an agglutinative morphology, and polypersonal agreement. It does not feature evidentiality or an extensive case system.

These languages display some of the features mentioned above, but none encompass all of them. When creating a conlang, you can draw inspiration from multiple natural languages to incorporate a diverse range of features.

Posted: **Tue Mar 21, 2023 6:05 pm**

chris_notts wrote: ↑Tue Mar 21, 2023 4:13 pm
zompist wrote: ↑Tue Mar 21, 2023 3:33 pm I see that GPT4 is a fan of the kitchen sink language. Noob!

(I'm not sure, but I think extensive case marking is not characteristic of the polysynthetic languages it's recommending.)
Not necessarily? There are plenty of agglutinative synthetic languages with both polypersonal agreement and lots of cases. Basque, which it mentions, is one, although Basque lacks other features like productive noun incorporation.

Polypersonal agreement is so common that I think it's not very useful to take it as a marker of polysynthesis. I don't think of Quechua, Swahili, Sumerian, and French as polysynthetic. Noun incorporation is a better test.

I don't know if it holds up, but there's a theoretical basis for my statement: polysynthesis is associated with head-marking, which means that you're not as likely to get elaborate case systems.

Posted: **Tue Mar 21, 2023 7:35 pm**

zompist wrote: ↑Tue Mar 21, 2023 6:05 pm I don't know if it holds up, but there's a theoretical basis for my statement: polysynthesis is associated with head-marking, which means that you're not as likely to get elaborate case systems.

Yes, I agree that complex noun case systems and other kinds of dependency marking are probably less likely in the most stereotypical polysynthetic languages, but I think there are examples.

If you insist on NI, then arguably Inuit/Yupik languages still count, although people do debate whether you can call what they have NI or not. Koryak has both NI and a load of cases.

Posted: **Wed Mar 22, 2023 2:40 pm**

I have a question about life in the USA. Basically, how "compact" or "solid" are suburban belts there? From what I've heard, I've got the impression that each major city in the US is surrounded by a ring of suburbs so that, if you're somewhere between the inner and outer edge of that ring, there's basically suburban middle- or upper class homes no matter where you turn.

In Germany, I have the impression that many major cities are surrounded by areas that more or less look like checkerboard patterns of suburban neighborhoods, agricultural land, and occasional patches of forest and grassland. For instance, there's the place where I live: my place is in a middle class suburban neighborhood (it's what in the US would be called an ADU; I'd never be able to afford an actual house here); if I walk a bit in one direction, things soon start to look very much like a major city; but if I walk a bit in another direction, I'm soon in a forest, and if I walk through that, in grassland.

So - how weird does that description sound like to people from the USA? Or does it not sound weird at all?

Posted: **Wed Mar 22, 2023 4:12 pm**

Raphael wrote: ↑Wed Mar 22, 2023 2:40 pm I have a question about life in the USA. Basically, how "compact" or "solid" are suburban belts there? From what I've heard, I've got the impression that each major city in the US is surrounded by a ring of suburbs so that, if you're somewhere between the inner and outer edge of that ring, there's basically suburban middle- or upper class homes no matter where you turn.

I suggest you pull up the satellite view of Google Maps and nose around a US city— Chicago, for instance. From the satellite view you can easily distinguish urban, suburban, forest, and fields.

I think your impression is a bit exaggerated. I'm in a suburb adjoining the city. If I walked west, I'd be on streets for a mile, then in the woods. But that woods is only half a mile wide. There are other green areas in the suburbs, some of them much larger.

When I was younger you could find some farms in between the suburbs— development here largely followed the train lines, so there were gaps between. Today most of those gaps are filled in with suburbs— you won't find farms till you're 40 miles from the city center. But again, there are parks and forest preserves all over.

But it also depends on the region.

Zompist Bboard Again

Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread

Re: Random Thread