*sr- roots in PIE

[Dennis Philps]

Dear IEists,

The silence has been deafening - it's good to see the list (and its
owner) back!

Question from a non-specialist : Given that the PIE initial consonant
cluster system contains (however sporadically spread) not only *(s)+
stop roots, i.e. *(s)p-, *(s)k-, *(s)t-, but also *(s) + sonorant
roots, i.e. *(s)m-, *(s)n-, *(s)l-, *(s)w-, and, plausibly *(s) +
laryngeal roots (Hoenigswald 52, Southern 99, etc.), how come there
appear to be no *(s)r- roots? Especially when *sreu- "flow" appears
to enjoy the same semantico-structural relationship with *ser- "flow"
as, for example, *skel- "cut" with *sek- "cut".

Have any publications dealt with this apparent anomaly in the system?
The only one I have found is Benveniste's paper ('Repartition des
consonnes et phonologie du mot, 1939:35) in which he states that "it
is probable that r and l were originally one single phoneme capable
of splitting into two phonemic units while often maintaining its
uniqueness" (my translation). There is also C. Friedrich's comment in
"Sprachen" that Hittite has no initial r- in its PIE vocabulary.

Many thanks for your thoughts,

Dennis.