Why is voicing considered lenition under phonological criteria?
To me voiced consonants seem stronger in articulation, therefore voicing should be considered fortification.
3 Answers
It's not based so much in what feels stronger, but on the overall patterns of sound changes. There is a tendency for voiceless consonants to become voiced, and voiced consonants to become continuants or approximants. Sometimes these changes are found in succession; for example, as late Latin developed into Spanish, the voiceless plosives p, t and k were first voiced (b, d, g) and later fricativized (β̞, ð̞, ɣ̞) (See here). These changes can all be seen as increases in sonorization. Voiceless plosives have the least amount of voicing, and are thus the least "vowel-like" of consonants. As we go from voiced plosives to fricatives, then approximants and finally vowels, each type is more vowel-like. We interpret vowels as weak, so any sound change to moving closer to vowels is interpreted this way, and called lenition.
A change that responds to contrasts among neighboring sounds and reduces those contrasts is a lenition. Assimilations are lenitions, while dissimilations are fortitions. Assimilatory voicing of consonants between voiced vowels lessens the constrast between the vowel preceding the consonant and the consonant, and also that between the consonant and the following vowel.
In the case of intervocalic voicing of obstruents, it may be that the assimilation winds up taking more articulatory energy, since the natural tendency of obstruents to devoice has to be overcome.
The distinction "lenis" vs. "fortis" was introduced into linguistics in the mid 19th century, to refer to the "air influx" of consonants, where German p is "hard" (harte) and German b is "soft" (weiche: von Raumer 1858 "Die sprachgeschichtliche Umwandlung und die naturgeschichtliche Bestimmung der Laute", who saw "aspiration" as being an important feature in describing Indo-European consonants). The Latin terms subsequently took over in linguistics, whence "lenition" and "fortition". The voiceless consonants of German happen to be aspirated, and it was common to describe them as having a forceful (strong) rush of air, which was lacking in the voiced consonants, which explains the feeling that voiceless stops are "strong".
The more generalized concept of "lenition" that prevails in contemporary linguistics is based on air pressure buildup differences associated with various manners of articulation, not on articulatory strength. Stops lead to greater air pressure differences than do fricatives, so stops are "harder" (more fortis) than fricatives. Geminate stop degemination (for example Finnish gradation) is also classified as a kind of lenition, and this is as you would expect based on the fact that a reduction in the closure duration will decrease the build-up of oral air pressure.
