(a) in sonority

A voiced consonant, other than a sonant or a liquid, immediately preceding
a voiceless consonant is normally devoiced. Thus b, d, g, z, ž represent respec-
tively (p), (t), (k), (s), (š) in labs, bads, zirgs, aiztrenkt, mežs.
A voiceless consonant immediately preceding a voiced consonant other than
a sonant or a liquid is normally voiced. Thus c, k, p, s, š, t represent respectively
(dz), (g), (b), (z), (ž), (d) in piecgade, nākdams, kāpdams, pusdivos, trešdiena,

(b) in point of articulation

The dental nasal n occurs as a velar nasal before the velar plosives k and g.
Thus in banka, banga, etc., the sound heard is that of the ng of English sing,
followed by the k or g.
Before š or č, s is often assimilated to š; thus :
pusčetri is often pronounced (puščetri)
is often pronounced (viššrākais).
Before ž or , z is often assimilated to ž; thus :
izžūt is often pronounced (ižžūt)
aizdžinkstēt is often pronounced (aiždžinkstēt).
At the end of a word the combination -šs becomes (šš) and reduces to (š);
thus :
drošs is pronounced (droš).

c) in both sonority and point of articulation

A combination of (a) and (b) above explains such cases as
uzšūt pronounced (uššūt)
mežs pronounced (meš).

T.G. Fenell. H.Gelsen