56. #d# before the surd #s# is pronounced #t#; the spelling #d# is
preserved for etymological reasons only: as, #adsum#, pronounced
_atsum_.
57. #g# always has the sound of English _g_ in _go_, never that of _g_
in _gentle_. #gu#, when it makes one syllable with the following vowel,
is pronounced like English _gw_: as, #sanguine# like _sanguine_.
58. #h# has a weak sound as _h_ in British English (Southern), and by
some was not counted as a consonant. Consequently the same uncertainty
existed as to initial #h#. The omission of initial #h# is recognized in
classical Latin for #ānser# (originally #*hānser#). Elsewhere the
omission of initial #h# in spelling, as #ostia# for #hostia#, is rare
until the third century A.D.
Very rarely #h# is written between two vowels to denote that each should
be pronounced separately (like our diaeresis in _coëxtensive_): as,
#ahēneus#, _bronze_, with #aē# separate (116 _a_); but #aes#, _bronze_,
with diphthongal #ae#.
59. Unsyllabic (22) or consonant #i# has the sound of English _y_ in
_year_.
60. There were two varieties of #l#. One was like the English _l_,
guttural in character, because in its pronunciation not only the blade
(front part) of the tongue touched the gums, but in addition to this the
rear part of the tongue was elevated toward the soft palate. The other
#l# was purely dental, and formed without such back elevation. This
second variety appeared in the combination #ll#, or whenever #l# was
followed by the front vowels (43) #e# or #i#, or when it was final.
Elsewhere #l# was guttural.
61. From the earliest times final m in unaccented syllables had a faint
sound or was even inaudible (Quint. 9, 4, 39). Consequently it is often
omitted in writing in the older inscriptions both before an initial
vowel or consonant: as, POCOLO for #pōcolom#; OINO for #oinom# (#ūnum#),
and the grammarian Verrius Flaccus proposed to write only half an #M#
for final #m# before a vowel. In prosody, therefore, final #m# did not
prevent elision (2493). The same is seen in prose in cases like
#animadvertō#, _I pay heed to_, from #anim^{um} advertō#, _I turn my
mind toward_ (395); #vēnīre#, _to be sold_ for #vēn^{um} īre#, _to go to
sale_ (1165). But in monosyllables where #m# closes the accented
syllable, it did not vanish (2494, 2495), and this difference in the
treatment of final #m# is reflected in the Romance languages.
62. #n# stands for two sounds. It represents the dental nasal, as _n_ in
English _now_. But before the gutturals #k#, #c#, #g#, #q#, and the
compound #x (= cs)#, it represents the guttural nasal which is written
_ng_ in English _sing_, _wrong_. This second n is sometimes called #n
adulterīnum# or ‘spurious #n#,’ thus: #nc# (in #avunculus#) as in
_uncle_; #ng# (in #angulus#) as in _angle_; #ngu# (in #sanguine#) as in
_sanguine_; #nqu# (in #inquit#) as _inkw_ in _inkwiper_; #nx#
(in #pīnxit#) as in _lynx_.
63. Dental #n# before #s# had a reduced sound, and is therefore
sometimes omitted in writing: as, CESOR for #cēnsor#; COSOL for
#cōnsul#, in older inscriptions; and #fōrmōsus# by the side of
#fōrmōnsus#; #vīcēsimus# by the side of #vīcēnsimus#, Cicero omitted the
#n# in the adjective suffix #-ēnsis#: as, #forēsia#, _of the forum_;
#hortēsia#, _garden plants_.
64. #q#, in classical Latin, appears only in the combination #qu#,
sounded like English _qu_ or _kw_ (27). #r# was trilled.
65. #s#, in classical Latin was always unvoiced (surd, 75) like English
_s_ in _so_, _sin_, never voiced (sonant, 75) as English _s_ in _ease_.
#su#, when it makes one syllable with the following vowel, is like _sw_
in _sweet_ (27).
66. In old Latin, final #s# after a short vowel and before a consonant
seems to have been reduced in sound or to have disappeared altogether.
In the older inscriptions it is often omitted in the ending of the
nominative singular #-us#, and in the pre-Ciceronian poets final #s#
often does not make position (2468). But such omission was considered
vulgar in Cicero’s time (Cic. _O._ 161; Quint. 9, 4, 38).
67. In the archaic period Latin #s# stood also for the voiced sibilant
(English _s_ in _ease_, _z_ in _zeal_), as in ASA, _altar_ (154).
68. #t# is always sounded as in _time_, never as in _nation_. The
pronunciation of #ci# and #ti# with the #c# and #t# as sibilants (as in
English _cinder_, _nation_) is very late.