On Fri, 1 Sep 2017 12:40:58 -0700, Leo Moser <[log in to unmask]> wrote:

>'Quirk' may be too strong a word, but it is a feature of English, the Latinate languages, etc.
>          -----There is no WHY. -----

I beg to differ!  There is no _universal_ why, no one size fits all why.  But there are reasons for all of the stuff below, and moreover these reasons are of kinds that recur from one language to the next.  If naturalism is your goal, then you refine your craft by acquiring a palette full of these reasons and learning how to paint them in layers, one over the next, to achieve the ultimate effects you want.

So for */tl-/ and */dl-/, Nathan Sanders' explanation looks good to me.  (I hadn't known this; in my mind I had vaguely filed it as a similar place avoidance sort of constraint, parallel to the also common */pw-/.)

>There need not be logic to phonemic patterns. 
>Check the following for logic:
>English has shriek but does not normally tolerate things like *sreek.  Yet it has sleek!
>   (Syllable initial sr- is very common in the languages of the sub-continent and elsewhere.)

Yes, on the way from PIE to Proto-Germanic there was epenthesis of *t in the sequence *sr.  This is articulatorily based -- trills are caused by airstream battering the tongue back and forth, which when it's in the position of /s/ doesn't work great -- although I couldn't say the details of why.

>English uses fl- and fr- very commonly ; but uses vl- only in certain foreign words—vr- hardly at all.

I dealt with this one last message.  On the long view, English uses initial /v-/ *at all* only in certain foreign words!  From the modern perspective, they're well integrated and not too few, but the phonotactics remembers that /v-/ is an invasive species through this limited combinatory potential.

(Interjections aren't governed by phonology as much as words of syntactically-productive classes are.  "vroom" is as unbothered by its phonotactic violation as are "mhm" or "meh" [final lax vowel] or "boing" [/oi/ before a non-coronal coda] or ...)

>English uses spr- and spl- very commonly ; but not sbl or zbl-, etc.

That you say it's /spl-/ and not /sbl-/ that English has is probably ultimately cued by the orthography.  What English has is neutralisation of the contrast /p/ vs. /b/ after an onset /s/.  The stop after /s/ isn't aspirated like /#p/ is, and in many lects /#b/ isn't particularly strongly voiced, so if you were committing English to writing for the first time you might very reasonably make your decision based on the aspiration and conclude it has /sb-/ rather than /sp-/.

Initial /z-/ is out of the question, really.  What goes for /v-/ goes tenfold for it.  

>English uses šr- rather commonly ; but not šl-, or zr-.

Yeah, *k was lost in the sequence *skl.  (English /S/ comes from Germanic *sk; unlike the other velars, this sequence palatalised unconditionally.)  This change I also don't understand particularly well.  I'm not even sure when it happened in English -- multiple times?  Proto-Germanic *sleutaną 'to shut' (no English descendant) had already lost the *k of its PIE antecedent; but "slave" was borrowed from French with a /k/ in early Middle English and dropped it maybe in the 16th century.