Zompist Bboard Again

I've been meaning to make more conlangs, but when I get started I always have this nagging feeling that I'm being uncreative by working off of SAE principles in the conlang's phonology and to some extent grammar. However, I cannot find a source on just what features are genuine universals and what features are unique to Europe, aside from a few exceptions such as a general lack of tones and potential presence of front-rounded vowels. I genuinely want to break outside the box of SAE traits.

In any case, what are some otherwise-common features in grammar and phonology not common to most languages in Europe? What are some features common in Europe but not so much elsewhere? Thankya for reading.

I wrote up a page mostly on my own on the KneeQuickie but I seem to have either forgotten to save it, or I saved it on my other computer and cant access it now.

http://www.frathwiki.com/Standard_Average_European

is similar, but focuses more on grammar whereas the page I wrote focused more on phonology. One feature I can think of that isnt on the list is that European languages often allow /s/ in places where other fricatives, and sometimes even other consonants, are not allowed. This feature is quite rare elsewhere, .... Khmer has some clusters like that, but I dont know enough about the language to rule out the possibility that every single one of them is a loan from Sanskrit or another IE language.

edit: it looks like I saved it by posting it to a sandbox page .... now posting it here.

Do you know WALS? It can get pretty unreliable, but for quite a range of things it can give you a sense of what some of the options are.

Generally speaking, learning about various options is likely to help you more than just learning that certain things are a bit SAE-y. Like, I think you'll benefit more from learning that many languages use participles for relative clauses, or invariant particles, rather than just learning that relative pronouns are SAE-y. That sort of thing you can learn about by reading typological surveys, surveys of non-IE language families, and grammars of particular languages.

Since you’re asking about SAE, I feel that I should at least mention the SAE phonology and grammar tests, which give you a method for determining how close to SAE your phonology and grammar are.

akam chinjir wrote: ↑Sat May 02, 2020 12:03 am Do you know WALS? It can get pretty unreliable, but for quite a range of things it can give you a sense of what some of the options are.

I concur — I also really like WALS for this.

Generally speaking, learning about various options is likely to help you more than just learning that certain things are a bit SAE-y. Like, I think you'll benefit more from learning that many languages use participles for relative clauses, or invariant particles, rather than just learning that relative pronouns are SAE-y. That sort of thing you can learn about by reading typological surveys, surveys of non-IE language families, and grammars of particular languages.

I agree with this as well — I feel that my conlanging has gotten much better once I actually started reading a bit about linguistics. As for specific resources for doing this, you can find lots in the Stack collection of books and articles; my favourites in the Stack so far are Dixon’s Basic Linguistic Theory (particularly volumes 2 and 3), as well as his and Aikhenvald’s ‘Explorations in Linguistic Typology’ series (in the Stack they’re under the ‘Oxford Linguistics’ subfolder). (Admittedly those two sources can be a bit unreliable at times — I’ve actually been having a long discussion about their reliability recently in the Linguistic Miscellany Thread — so take them with a grain of salt, but generally they’re fairly good for getting an overview of non-SAE ways of doing things.)

Starbeam wrote: ↑Fri May 01, 2020 6:32 pmIn any case, what are some otherwise-common features in grammar and phonology not common to most languages in Europe? What are some features common in Europe but not so much elsewhere? Thankya for reading.

I'd like to push back against the whole idea a little.

First, though, I'll say it's understandable, and can be a lot of fun. I've tried to do this myself, creating "extremely non-IE" languages like Kebreni, Elkaril, and Wede:i.

But... well, "Standard Average European" is an amateur linguist's invention that isn't a big preoccupation of linguistics in general. The Frathwiki summary ends up admitting that it's largely a sprachbund focused on French, Dutch, German, and Italian. So it minimizes a good deal of variation within Europe. Plus I'd quibble with a good amount of Haspelmath's list even within that area. (E.g. it's rather misleading to say that either French or German is simply SVO. Plus, comparative inflection of adjectives in French and Italian...?)

Second, if you just try to reverse European features, you're still being largely influenced by European features. I guess my point is that I think you should adopt a feature because you like it and want to mess around with it, not simply because it's not used in French and German.

zompist wrote: ↑Sat May 02, 2020 2:41 am

Starbeam wrote: ↑Fri May 01, 2020 6:32 pmIn any case, what are some otherwise-common features in grammar and phonology not common to most languages in Europe? What are some features common in Europe but not so much elsewhere? Thankya for reading.

I'd like to push back against the whole idea a little.

First, though, I'll say it's understandable, and can be a lot of fun. I've tried to do this myself, creating "extremely non-IE" languages like Kebreni, Elkaril, and Wede:i. … Second, if you just try to reverse European features, you're still being largely influenced by European features. I guess my point is that I think you should adopt a feature because you like it and want to mess around with it, not simply because it's not used in French and German.

This is all true — I agree that you shouldn’t avoid SAE features just because they’re SAE, and you should feel free to use them if you want to. But I do think that the concept of SAE does serve a useful purpose in conlanging: you should at least be aware of which features are in SAE and how common they are outside SAE, so that you don’t end up accidentally duplicating them purely because you don’t know any better.

zompist wrote: ↑Sat May 02, 2020 2:41 am (E.g. it's rather misleading to say that either French or German is simply SVO.

I'd managed to forget how WALS fumbles German.

A related data point, Italian is also often analysed as VSO, at least a lot of the time, and some people have analysed Spanish that way too, I'm pretty sure.

More reasons that it's more important to learn about various ways languages do things than to come up with a list of things to avoid, imo.

akam chinjir wrote: ↑Sat May 02, 2020 3:02 amMore reasons that it's more important to learn about various ways languages do things than to come up with a list of things to avoid, imo.

Yeah, my suggestion for avoiding the trap of unwittingly reproducing SAE features was going to be learning (or at least studying thoroughly) a language that isn’t SAE, and preferably more than one. The more alternatives you’re familiar with, the greater your sense of what is possible (and what might truly be unworkable as opposed to merely strange). A list of non-SAE features doesn’t give you a real sense of these features are implemented—and how that might vary from language to language.

Linguoboy wrote: ↑Sat May 02, 2020 5:08 am

akam chinjir wrote: ↑Sat May 02, 2020 3:02 amMore reasons that it's more important to learn about various ways languages do things than to come up with a list of things to avoid, imo.

Yeah, my suggestion for avoiding the trap of unwittingly reproducing SAE features was going to be learning (or at least studying thoroughly) a language that isn’t SAE, and preferably more than one. The more alternatives you’re familiar with, the greater your sense of what is possible (and what might truly be unworkable as opposed to merely strange). A list of non-SAE features doesn’t give you a real sense of these features are implemented—and how that might vary from language to language.

I agree, but for most of us there's a limit to how many languages we can actually learn in detail given the demands of life. I can honestly say I've only ever really learned 3 natlangs in some depth apart from my native language, although I've dabbled with many more at some point. But I've also read a lot of typology books over the years and skimmed many grammars, which I think also help a lot getting a feel how things can work even if it doesn't soak in to the same extent.

Pabappa wrote: ↑Fri May 01, 2020 7:04 pm One feature I can think of that isnt on the list is that European languages often allow /s/ in places where other fricatives, and sometimes even other consonants, are not allowed. This feature is quite rare elsewhere, .... Khmer has some clusters like that, but I dont know enough about the language to rule out the possibility that every single one of them is a loan from Sanskrit or another IE language.

Such words can be native, such as ស្ងាប /sŋaːp/ 'to yawn', which is pretty firmly Austroasiatic and distinctly un-IE looking. In fact, I think they're mostly native, but I haven't checked.

Richard W wrote: ↑Sat May 02, 2020 8:14 am

Pabappa wrote: ↑Fri May 01, 2020 7:04 pm One feature I can think of that isnt on the list is that European languages often allow /s/ in places where other fricatives, and sometimes even other consonants, are not allowed. This feature is quite rare elsewhere, .... Khmer has some clusters like that, but I dont know enough about the language to rule out the possibility that every single one of them is a loan from Sanskrit or another IE language.

Such words can be native, such as ស្ងាប /sŋaːp/ 'to yawn', which is pretty firmly Austroasiatic and distinctly un-IE looking. In fact, I think they're mostly native, but I haven't checked.

Wikipedia lists all sorts of crazy consonant clusters in Khmer: lɓ-, mʔ-, ʔʋ-, cʰ- (pronounced as aspiration but phonologically a consonant cluster), tʰp- (but this aspiration is only a phonetic detail), lkʰ- (who knows what this aspiration is!). And they’re really quite frequent: a quick skim of the rest of the article easily reveals words like /sʔaːt/ to be clean, /cʰkae/ dog, /kʰɲom/ I, /lʔɑː/ handsome, /sdap/ understand. So really, out of all the southeast Asian languages, Khmer is the one I would most expect to have weird clusters with /s/!

The interesting thing about 2-consonant Khmer clusters is what isn't allowed. In this respect, the behaviour of /s/ is not at all 'SAE'.

The 3-consonant Khmer clusters are an interesting bunch. /lkh/ might be the most interesting. I suspect the three starting with /s/ actually are of IE origin.

I'm not sure how 'SAE' the behaviour of /s/ is. In modern languages, it has retained special behaviour in Germanic and Baltic, but how special is its behaviour in Italian? Italian has a prefix s-, and I wonder if that distorts the picture. Greek, Albanian and Slavonic have all developed quite a few new initial clusters. One can argue for sibilant plus stop being a pseudophoneme in IE, and especially /st/. It even has its own letter in (English?) Runic and Kharoshthi.

@Starbeam:

The Wikipedia article lists many good examples, and even distinguishes things that are more typical of SAE specifically (or, well, average core Western European, which is how Haspelmath seems to have defined it in practice) from those that are common in SAE but also common in some other areas. Maybe I'm misreading something but it sounds like you haven't read it...

I think the main revelation in WALS regarding weird characteristics of (modern) European languages is the usual lack of derived distributive numerals (meaning things like: "one by one, two by two...", "one at a time, two at a time...", "one for each person, two for each person..."). WALS says that around 3/4 of the world's languages have them, formed in various ways. Note that Latin and Ancient Greek did have them in a complete set/formation* (the former with distinctive stems even, and the latter with compounds of σύν- 'with' + number, although Greek tended to not use these as much as Latin).



Regarding the secondary topics in this thread, I think conlangers sometimes criticize other conlangers because making core-Western-European-like languages seems like a waste of the huge possibilities (rather as Linguoboy said), but hey, if that's what the latter group likes to be creative about, this devolves into dē gustibus nōn est disputandum territory. Knowing about non-European languages definitely gives you a nice perspective though.

Regarding word initial /sC/, it is also found in Classical Tibetan.


* Funnily, Latin and Ancient Greek had a great wealth in derived numbers overall. Besides cardinal, ordinal and distributive numbers, they also had complete formation of adverbs of number of times like 'twice, three times' (Latin -iēns, Grk -ιάκις). Note that to specify a time, you use a neuter ordinal in Latin (no idea how Greek does this): tertium 'the third time', quartum 'the fourth time'. There were also further minor types of numbers, generally used with only a few of the first integers and with quantifiers like 'many':
- multiplicative adjectives 'twofold, threefold' (prefix form + -plex, Grk -πλοῦς)
- proportional adjectives 'twice as great/big, three times as great/big' (prefix form + Lat -plus, Grk -πλάσιος or -ττός)
- Latin adjectives of rank 'of 2nd rank/secondary, of 3rd rank/tertiary' (ordinal stem + -ārius)
- Latin adjectives of legion 'of the 2nd legion, of the 3rd legion' (ordinal stem + -ānus)
- Latin adjectives of age 'of two/three/four years in age' (only bīmus, trīmus, quadrīmus, cf. patrīmus 'with a living father')
- Greek adverbs of relative day 'on the 2nd day, on the 3rd day' (ordinal stem + -αῖος)
- Greek adverbs of division 'in two ways/parts, in three ways/parts' (prefix form + -χῆ or -χα)
- Latin adjectives of division 'of two parts, of three parts' (distributive stem + -ārius)
This besides their great ability to use the prefix forms of numbers in compounds, as in bīga/trīga 'chariot of two/three horses, two/three-horse chariot', biennis/διετής 'lasting two years, two-year X', bilinguis/δίγλωσσος 'bilingual', bicolor/δίχροος 'of two colours, two-colour(ed) X', etc. As you can see in these examples, English can often do this too, but for example Spanish can't.

chris_notts wrote: ↑Sat May 02, 2020 6:01 amI agree, but for most of us there's a limit to how many languages we can actually learn in detail given the demands of life. I can honestly say I've only ever really learned 3 natlangs in some depth apart from my native language, although I've dabbled with many more at some point. But I've also read a lot of typology books over the years and skimmed many grammars, which I think also help a lot getting a feel how things can work even if it doesn't soak in to the same extent.

Learning any language to a level of facility is a daunting undertaking, which is why I hesitated to suggest this, but there really is no substitute for learning a language quite unlike your own--and learning it at least well enough to comprehend spontaneous native-speaker usage, since formal registers are a different beast (and often highly influenced by non-native stylistic models).

You can probably get some of these benefits from reading a really well-written comprehensive grammar. Typology books are interesting, but unless they include a case study of a particular language (or group of languages), you aren't really going to gain an understanding of how this feature fits into the larger system of the language.

Linguoboy wrote: ↑Sat May 02, 2020 11:02 am Learning any language to a level of facility is a daunting undertaking, which is why I hesitated to suggest this, but there really is no substitute for learning a language quite unlike your own--and learning it at least well enough to comprehend spontaneous native-speaker usage, since formal registers are a different beast (and often highly influenced by non-native stylistic models).

I agree that you should if you can. But there's a limit to how many languages most people can learn really well, so I think you'll always also be relying on languages you've only read about (e.g. via grammars, ...) to add further breadth. Especially since many things people want to add into their conlangs are present in very few, if any, widely spoken languages. What language in the top 50 has a sophisticated system of switch reference, for example? And for languages not spoken by many people, materials and speaking partners can be a major issue.

You can probably get some of these benefits from reading a really well-written comprehensive grammar. Typology books are interesting, but unless they include a case study of a particular language (or group of languages), you aren't really going to gain an understanding of how this feature fits into the larger system of the language.

I think even general typology books have some value, but you're right that you need to read at least more general grammar summaries to see how a specific language feature fits with other features.

Doesn't Haspelmath's paper give some indication of the global frequency of each of the features he discusses, or am I misremembering?

Whew, this blew up a little bit while I was asleep. I can't respond to every single person, but I can address some common sentiments. I really thank you all for your patience; to find out SAE isn't completely sound almost is a load off of my back in a weird way.

I understand making everything un-European as possible just makes things still guided by European features, which is why I asked which features are genuine universals. I have nothing against European languages or European-like conlangs. The thing is, I feel limited by just sticking with langs from the region. I also only have so much patience and resources to learn languages, and many of the ones i find interesting feature-wise I don't per se want to learn but just know a good deal about.

I have read the Wikipedia article on SAE, but as with anything on that site, you need to vet it in some capacity.

chris_notts wrote: ↑Sat May 02, 2020 2:30 pmI agree that you should if you can. But there's a limit to how many languages most people can learn really well, so I think you'll always also be relying on languages you've only read about (e.g. via grammars, ...) to add further breadth. Especially since many things people want to add into their conlangs are present in very few, if any, widely spoken languages. What language in the top 50 has a sophisticated system of switch reference, for example? And for languages not spoken by many people, materials and speaking partners can be a major issue.

I've heard this before, but I often think it's a bit circular. I suspect that one important reason why people often want to include things like "switch reference" in their conlangs in the first place is because they see discussions of them where they're called interesting or implied to be interesting. This is especially true when phenomena receive a distinct name, I think, such as "fluid-S active-stative alignment", or "topic-prominent language", or "egophoricity", or "evidentiality". Meanwhile, many other little features tend to be ignored, features that really only jump at you once you actually try to learn a language while holding your typological knowledge in mind.



For example, the so-called "reflexive pronouns" of Classical Latin always refer back to the subject of the main verb no matter where they are in the sentence. This means that they aren't quite the stereotype of "reflexive pronouns" in modern Western European languages, but rather "subject-reference pronouns" if you will.

Dux dūcēbat cīvēs sibi invīsūrōs.
leader.NOM thought.3S citizens.ACC 3S.REFL.DAT envy.FUT.PTCP.ACC.PL
'The military leader was thinking the citizens were going to envy him.'

Here, the pronoun referring back to the main subject (dux 'leader') is the "reflexive" sibi even though it is inside a subclause, a subclause with accusative cīvēs 'the citizens' as the subject. If you wanted to replace the object of envy with a true reflexive (the citizens envying themselves), you use the intensive pronoun ipse instead (ipsīs invīsūrōs). And if you wanted to replace it with yet another person, you'd use the anaphoric demonstrative is or the distal demonstrative ille instead (illīs invīsūrōs). This behaviour of reflexive pronouns is also true of subclauses with a finite verb.

This is not altogether too different from switch-reference, the difference being that classic switch-reference is morphologically more dedicated (while here Latin reuses the reflexive/intensive/demonstrative pronouns with a different purpose inside subclauses) and tends to deal specifically with changes in the subject (whereas Latin deals with any syntactic use).



So... I have this suspicion that if these Latin pronouns were called "subject-reference pronouns", and we talked about "specialized argument-reference pronouns" in general (to make up some terms), more conlangers would be interested in the topic... and create pronouns that have more stable reference throughout a whole sentence... and think of strange interactions/alignments between main clause pronouns and subclause pronouns... etc., etc.

(I'll give more examples of ignored features of Latin/Spanish/French later tonight...)

Ser wrote: ↑Sat May 02, 2020 7:56 pm For example, the so-called "reflexive pronouns" of Classical Latin always refer back to the subject of the main verb no matter where they are in the sentence. This means that they aren't quite the stereotype of "reflexive pronouns" in modern Western European languages, but rather "subject-reference pronouns" if you will.

This made me think of long distance anaphora, like you get in Mandarin, so I did some poking. Seems like it does get discussed as the same general phenomenon---so there's a name for it. One thing: it seems like it occurs mostly in reported thought and speech, and in that sort of context se can look like a logophor (that is, like a pronoun bound by the subject of the thought or speech verb).

I suspect that one important reason why people often want to include things like "switch reference" in their conlangs in the first place is because they see discussions of them where they're called interesting or implied to be interesting. This is especially true when phenomena receive a distinct name, I think, such as "fluid-S active-stative alignment", or "topic-prominent language", or "egophoricity", or "evidentiality".

You're probably onto something, though I don't think offering people the name "long-distance anaphora" would really change too much. I'd think it's got more to do with anaphora being a fairly heavy syntactic phenomenon---binding theory, locality, that sort of thing. My impression is that many conlangers find it easier and more interesting to sink their teeth into semantic phenomena like agency or control or animacy (and to probably a lesser extent pragmatic phenomena like topicality and definiteness). Anyway I know this has been true of me, and the linguistics literature that's generally most useful for conlangers (the functionalist literature) tends to highlight that sort of thing.

Another factor, maybe, is the tendency you sometimes see to approach language design templatically. This foregrounds certain kinds of issue, like how case-marking or agreement is aligned, or whether you've got evidentials, an issue that to begin with might just be a question of whether to add another slot to your template. (I suppose egophoricity could also go on that list, though actually I doubt many conlangers have heard of it.) But there are plenty of other issues you're never going to get to that way, including interesting binding stuff.

(I'll give more examples of ignored features of Latin/Spanish/French later tonight...)

Looking forward to it!

My current vote for coolest English thing that people could have a lot of fun playing with are the phrasal verbs.

Ser wrote: ↑Sat May 02, 2020 7:56 pm Here, the pronoun referring back to the main subject (dux 'leader') is the "reflexive" sibi even though it is inside a subclause, a subclause with accusative cīvēs 'the citizens' as the subject. If you wanted to replace the object of envy with a true reflexive (the citizens envying themselves), you use the intensive pronoun ipse instead (ipsīs invīsūrōs). And if you wanted to replace it with yet another person, you'd use the anaphoric demonstrative is or the distal demonstrative ille instead (illīs invīsūrōs). This behaviour of reflexive pronouns is also true of subclauses with a finite verb.

This is really neat, and it's immediately going into the file of "things I wish I'd put in my syntax book and will have to go into the 2nd edition".