Anyway, the vast majority of the phonologies I've seen recently have been very Standard Average European. I'm not a big fan of SAE phonologies (although I'll admit that they're a lot easier to work with than non-SAE phonologies in a lot of cases), so here's a guide on how to leave that box.
First, we have to define SAE. Wikipedia gives a list of traits, but I have to add some to more accurately describe the sort of phonology that I'm talking about. (I realize that some of these, especially the tone thing, don't actually apply to real-world SAE, but they do to the conlang equivalent.) I'll bold my additions.
Also, if I missed any conlang examples, tell me and I'll add them.
- absence of phonemic opposition velar/uvular
- phonemic voicing oppositions (/p/ vs. /b/ etc.) as the only distinctions between sounds with the same POA and MOA
- initial consonant clusters of the type "stop+sonorant" allowed
- no initial velar nasal
- a wide variety of allowable clusters that, except for those that contain a sibilant and a stop, all adhere to the sonority hierarchy
- only pulmonic consonants
- no phonation or secondary articulation contrasts on vowels
at leastthree degrees of vowel height (minimum inventory i e a o u)- lack of lateral fricatives and affricates
- two series of coronals
- lack of a tone or register system
- at least two of each of the following type of consonant: fricative, nasal, liquid, semivowel
So, from there, we have a rough idea of how the goal can first be approached: figure out how to realistically avoid SAE characteristics. Obviously I'm not saying that every one of your conlangs should avoid each of these features, but if you end up hitting all of them and you're not consciously trying to make a Euroclone, I'd advise rethinking your inventory. (Another thing that I'd have to advise against, after reading over some of these conlang inventories, is excessive regularity. Don't pull a Somali or a Dahalo, but don't be afraid to leave realistic gaps in your inventory.)
Anyway, here we go. A lot of this will just be my speculation, so correct me where I've gotten things wrong.
1. Absence of phonemic opposition velar/uvular
That may appear to be self-explanatory, but uvulars can be a bit tricky to handle. Essentially, they're like velars, only more so; gaps in uvular series are even more common than gaps in labial or velar series. The uvular nasal almost never occurs contrastively*, voiced uvular fricatives are uncommon, and holes in uvular stop series are frequent: Georgian only has /qʼ/, Avar has only two short uvular consonants, Uyghur has /ʁ/, and Chechen, one of the few languages with both uvulars and a large vowel inventory, is missing /ɢ/. According to WALS, it's slightly more common for a language to have uvular stops and continuants than only uvular stops, and a bit over three times more common for a language to have only uvular stops than only uvular continuants.
/ɢ/ is uncommon for articulatory reasons; see here. One point not mentioned on that page, however, is that, in addition to losing their voicing for greater ease of articulation, /g ɢ/ can drop a different feature toward the same end and lenite, as probably happened in Uyghur.
Uvulars tend to lower surrounding vowels (see: Quechua, German), and my distinctly unscientific impression is that, for this reason or something related to it, languages with uvulars will tend to have fewer vowels than languages without them. (Arabic and Quechua have three, many North American langs with uvulars have four, etc.) I also think languages will generally only add an uvular series once they already have four other consonant series (i.e. /p t c k/), although I've seen a lot of languages with incomplete palatal series (most commonly no stops, as in Ket and some Semitic languages, but a fricative gap is also possible, as in Nivkh**) and a mostly full uvular series.
How to derive uvulars, you ask? If vowel harmony-related alternation is any guide to diachronics, Uyghur derived them from velars around back vowels. There's also the pharyngealized velar route that I think Arabic took.
There are also some North American langs that, instead of velars and uvulars, have prevelars and postvelars, but I'm not sure what the difference is so I won't say much there. They tend to have very incompete prevelar series, though.
Uvulars are a somewhat areal feature, and ~17% of the languages in WALS have them.
* The only language I can find where it does is Greenlandic. Any other examples?
** However, Nivkh /s z/ pattern as palatals in stop-fricative alternation. The consonants that pattern with the dentals are /r̥ r/. Yes, trills pattern as fricatives in Nivkh. I want to see this done more.
Natlang examples: Aymara, Chechen, Greenlandic, Nivkh, Uyghur
Conlang examples: South Eresian, Kannow
Sound samples: French, Quechua
2. Phonemic voicing oppositions (/p/ vs. /b/ etc.) as the only distinctions between sounds with the same POA and MOA
As originally formulated, this is easy to avoid: just don't add a voicing opposition. But it's not that simple; in SAE inventories, there is usually a distinction between voiceless and voiced plosives and, somewhat less commonly, between voiceless and voiced fricatives. This is reasonable enough. According to WALS, 61% of languages have a voicing contrast in plosives and 35% have one in fricatives.
But what else can you do?
Most obviously, you can have only one series, as in Finnish*, or have some other sort of fortis-lenis distinction, as in English (aspirated vs. unaspirated), Nuxalk (aspirated vs. ejective), or, less realistically, some German dialects** (affricated vs. lenis). But it's also possible to add other series:
• Ejective
Ejectives were mentioned above, but I'm going to go ahead and mention them again; they occur in 17% of WALS, although they tend to be strongly areal features. I've seen them used in a few conlangs, although I don't think there are any so far with ejective fricatives.
Gaps in ejective plosive series tend to be toward the front of the mouth; many languages have no /pʼ/, and some don't have /tʼ/. I don't think I've ever seen a natlang with only /kʼ/ (langs without /pʼ tʼ/ tend to have other ejective consonants, like /tsʼ/ or /qʼ/), although I'm sure it's possible, and I've seen it pulled off in at least one conlang inventory. (Klgapixo, I think it was called. This was several years ago, though, and I'm almost certain it's dead.)
Ejectives can also co-occur and pattern with glottalized (i.e. creaky-voiced) resonants. I think this is generally associated with North American languages, but 24% of langs in WALS with ejectives also have glottalized resonants.
Ejective sonorants and voiced ejective stops are impossible, although homorganic voiced-ejective clusters ([dtʼ] etc.) are attested in Taa, otherwise known as ǃXóõ. Ejective retroflex stops are very rare, although this might just be a statistical glitch. Note, however, that implosive retroflex stops are unattested, so maybe something is going on there. Ejective affricates are very common in languages with ejectives, and I'd guess noncoronal affricates are more common as ejectives than as non-ejectives.
A three-way distinction between unvoiced unaspirated, aspirated, and ejective consonants seems to be reasonably common in natlangs and popular in natlangs, more so than the North American two-way aspirated/ejective distinction.
Natlang examples: Archi, Dahalo, Gwich'in, Kabardian, Nuxalk, Tlingit
Conlang examples: Kannow, Na'vi, Sentalian, South Eresian,
Sound samples: Hausa, Zulu
• Implosive
Implosives are about as common as ejectives (13% of WALS), although they're much more areally concentrated; most langs with implosives are in Africa or China.
Gaps occur in the opposite pattern to implosives, although probably more so; I'd guess /ɠ/ is less common in langs with implosives than /pʼ/ is in langs with ejectives. Voiceless implosives occur in only a few languages, and those languages always have voiced implosives, except for (iirc) one where a voiceless implosive can occur as an allophone of an ejective. The voiced retroflex implosive is unattested, although that might be a statistical anomaly.
I think implosives mostly occur in languages with at least two other series, although I've seen things that are close to counterexamples: Vietnamese has five unaspirated consonants, two implosives, and only one aspirated consonant. (The other aspirated consonants shifted to fricatives.)
Wikipedia claims implosive fricatives and affricates are "extremely unusual", but I haven't been able to find any attestations of them.
Diachronically, these can be developed from voiced stops in a chain shift (I can't come up with any examples, but voiced stops in some languages are realized as slightly implosive) or clusters of a voiced stop and a glottal stop.
Natlang examples: Owere Igbo (contrasts all of /pʰ p ƥ bʱ b ɓ/), Sindhi, Tsou
Conlang examples: Proto-Vdangku
Sound samples: Sindhi
• Long (fortis?)
I'm not sure if this is attested outside Caucasian languages, but it's fairly common in them. There's probably some other articulatory thing going on, but I don't know what it is.
Natlang examples: Akhvakh, Avar, Lak
Conlang examples: ???
• Alternate phonation: (note: I'll deal with phonation on vowels separately)
• Slack voice
And now we get into alternate methods of phonation***, which I've never seen even touched in conlangs. (However, note that they're very uncommon, especially on consonants.) Slack voice involves holding the glottal opening slightly wider than in modal voice, but not as much as in breathy voice, with which I'm not sure if it's ever contrasted. Slack-voiced consonants tend to give the following vowel a slightly breathy quality, and in fact, this is the primary distinction between slack-voiced and stiff-voiced consonants in Javanese, and between slack-voiced and modal consonants in Wu Chinese.
Natlang examples: Javanese, Wu Chinese
Conlang examples: ???
• Breathy voice
These are mostly known as the 'voiced aspirates' in Proto-Indo-European and many Indo-Aryan languages, although they can be hard to work with since there's only one attested language that has them without aspirates. Breathy voice, or 'murmur', is holding the glottal opening slightly wider than in slack voice, but not as much as in voicelessness. These tend to lower the tone of surrounding vowels; this is their main distinguishing feature in some Bantu languages, and Punjabi lost breathy voice altogether and developed tone from them. Gujarati has contrastive breathy voice on both vowels and consonants.
These might be easier to develop than other nonmodal phonations; [ɦ] is commonly realized as breathy voice on a nearby vowel, so it could be developed from there, and then maybe transferred onto consonants. In my conlang Hathe, I developed breathy-voiced vowels from ɦV sequences, as in the name of the language: [a̤ðə].
Note that, in Zulu, one of the very few languages to contrast /h ɦ/, /ɦ/ is the breathy-voiced version of /ɦ/.
Natlang examples: Sindhi, Urdu, Zulu
Conlang examples: ???
Sound samples: Gujarati, Hindi, Sindhi
• Stiff voice
I'm switching directions of openness here; stiff voice involves holding the glottal opening more closed than in modal voice. Wikipedia claims Thai contrasts stiff-voiced stops with voiceless and aspirated stops, although on the page for Thai phonology it says those are actually just voiced stops. This might also be the third series in Korean.
Natlang examples: Javanese, Korean?, Thai?
Conlang examples: ???
Sound samples: Thai?
• Creaky voice
On consonants, this most frequently occurs on resonants as part of a glottalized series that also includes ejective stops. I don't think creaky-voiced stops are attested, although I could be wrong.
Natlang examples: Haida, Halkomelem
Conlang examples: ???
Sound samples: Hausa
• Mixed voicing
A mixed-voiced consonant is one where the voicing changes midway through the articulation of that consonant. Voiced-aspirated consonants are attested in Taa, along with voiced-ejective consonants (however, I think both of these are analyzed as clusters), and voiceless-voiced consonants are attested as allophones of voiced consonants in Tanacross.
Natlang examples: Taa, Tanacross
Conlang examples: ???
• Nonexplosive stops
Two nonexplosive bilabial stops, one voiced and one glottalized, are attested in Ikwerre. I know nothing about them. Here's the Wikipedia page.
Natlang examples: Ikwerre
Conlang examples: ???
• Secondary articulation / articulation modification:
• Palatalization
The most common. Usually contrasts with velarization, as in Russian and Irish. I shouldn't have to see much. Surprisingly underrepresented in conlangs, and where it does exist, it's usually to form another POA (i.e. /p t tʲ k f s sʲ x/ or /p t kʲ k f s xʲ x/) instead of another dimension (i.e. /p pʲ t tʲ k kʲ f fʲ s sʲ x xʲ/).
Natlang examples: Hausa, Irish, Russian
Conlang examples: ???
• Velarization/pharyngealization
I'm grouping them together because I've never seen them contrast. However, velarization usually contrasts with palatalization (an exception is Gilbertese, which contrasts velarized and plain /m p/, but has no secondary articulation contrast anywhere else), whereas pharyngealization can contrast with no secondary articulation, as in Arabic. There's no such thing as a velarized velar; it's possible that there could be a case where pretending they can exist would be useful for analysis, but usually that's just done as palatalized vs. normal, or, as in Irish( /p pʲ t tʲ k c/), analyzing palatalized consonants as taking up a whole different POA.
Also, note Ubykh, which patterns pharyngealized labial stops with rounded nonlabial stops, and languages like Arabic and Chilcotin, where uvulars pattern as pharyngealized velars.
Natlang examples: Arabic, Chilcotin, Gilbertese, Irish, Marshallese, Russian, Ubykh
Conlang examples: ???
• Labialization
Otherwise known as rounding. Can be used either as another dimension or as another POA or two; languages can distinguish labialization only on velars, only on velars and alveolars, only on velars and glottals, etc.
Labialized labials, unlike velarized velars, are possible, possibly because labialization usually involves a rounded offglide and some degree of velarization. Contrasts with both palatalization and velarization in Marshallese, although the only contrast in velars is normal and labialized. Labialized consonants can be realized allophonically as having simultaneous labial closure, or even a subsequent labial trill.
Some labialless American languages have something very similar to labialization, but that involves cupping the tongue instead of moving the lips. I can't remember what it's called, but it's apparently also attested in some dialects of British English on /ɒ/.
Natlang examples: Ikwerre, Lezgian, Proto-Indo-European
Conlang examples: Continental Kett, Kannow
• Simultaneous labial closure
Patterns with, and varies allophonically in some languages with, labialization (with one exception: there are languages with /tʙ̥/ but no analogous velar consonant), and doesn't contrast with it AFAIK. Mostly found in Africa, although the most famous example is spoken near Papua New Guinea.
Natlang examples: Ewe, Yoruba, Yeli Dnye
Conlang examples: Insular Kett (note that /p/ comes from /k͡p/, and is pronounced as such in some conservative dialects)
• Apical vs. laminal coronals
If you really want to do the palatalization-as-separate-POA thing, this is an alternative; according to The Sounds of the World's Languages, Bulgarian /t tʲ/ are normally realized as, respectively, apical and laminal alveolar. (Wikipedia analyzes them as /t tʃ/, but note that, according to SWL, laminal stops are more likely than apicals to be affricated.) Basque contrasts apical, laminal, and palatalized affricates and fricatives.
Natlang examples: Bulgarian, Basque, Dahalo
Conlang examples: Gadaye, although I write the laminals as palatalized alveolars
Sound samples: Mid-Waghi
• Whistled sibilants
Attested in some Bantu languages, most notably Shona, which appears to contrast non-labialized and labialized versions of them. Whistled dental clicks are apparently also possible, though I'm not sure how they'd be pronounced.
Natlang examples: Shona
Conlang examples: Enzielu
* I'm not sure how likely it is to have this and not have phonemic gemination.
** Actually, I don't know if there are any to make that contrast that don't have an unaffricated fortis series.
*** If anyone knows the diachronics behind any of these, post them. I can't think of any.
More later. (Holy fucking shit, I started this post at around 5:30pm. It's 2am now.)