23 TOPIC 3: NATURAL CLASSES, NATURAL RULES, RULE ORDERING NATURALNESS IN PHONOLOGY Natural Classes Language sounds fall into natural classes according to their articulatory and/or acoustic properties. We name these properties using features like [labial], [voiced], [high], [strident]. We can define a natural class as follows: natural class: A group of language sounds that share a phonetic feature or (usually small) set of features that no other sounds in the language share. Examples that we have seen: • ENGLISH “VELAR FRONTING” [+velar]: - fronting applies not just to /k/, but to all velars, i.e. /k, g, ŋ/ - fronting does NOT affect other consonants, e.g. /t, d/ [+front]: - velar fronting takes place before all front vowels, not just /i/ - velar fronting does NOT take place before non-front vowels • PAPAGO “PALATALIZATION” !"#$%&++alveolarstop, !"#$%&++alalveopalatstop: palatalization turns /t, d/ into affricates [t!ʃ, d!ʒ] palatalization does NOT apply to [-stop] or [-alveolar] [t!ʃɨposid] ‘to brand’, [d!ʒusukal] type of lizard, not *[t!ʃɨpoʃid], *[d!ʒuʃukal] [d!ʒuki] ‘rain’, [t!ʃuagia] ‘net bag’, not *[ d!ʒut!ʃi], *[t!ʃuad!ʒia], [+high]: palatalization applies before all and only the vowels /i, ɨ, u/ • LANGO “SPIRANTIZATION” !!"#$$%&+long-voice-stop, !!"#$$%&'long-voice-stop: all simple voiceless stops become non-stops in / V___V Compare: [ʔàxóbò] ‘I said’, [gédo!] ‘to build’, [ʔàɾ"ód!ʑò] ‘I beat’, [bwògò] ‘young’ What do these forms tell us about our natural class? Where natural classes operate • “INPUT” SEGMENTS: English velars, Papago alveolar stops, Lango voiceless stops • “TRIGGER” SEGMENTS: English front vowels, Papago high vowels • “OUTPUT” SEGMENTS: English fronted velars, Papago alveopalatal affricates, Lango fricativesLinguistics 120A 3. Natural Classes, Natural Rules, Rule Ordering 24 Natural Processes = Natural Rules Question: Why do we not say that … • English /k!/  [k] / ___ ɑ, ʌ, ɹ, l? • Papago changes alveopalatal affricates to alveolar stops before non-high vowels? • Lango changes fricatives into stops when geminate or at word boundaries? Answer: In general, we expect phonological rules to have “natural motivations.” But how do we know what is “natural”? WHAT MAKES PHONOLOGICAL RULES “NATURAL”? Part of understanding phonology is understanding why certain processes are “natural” and others are not. It cannot be an accident that certain processes (= rules) operate in virtually the same way in thousands of languages and other processes are rarely, if ever found in languages. There is not one simple explanation for why this is so—all the articulatory and perceptual aspects of language come into play. Nonetheless, we can list some of the most common natural processes and provide a rudimentary explanation for their frequency and, by implication, the rareness of the opposite processes. Below are some widespread natural phonological processes. The opposite processes would (probably always) be unnatural, for example, it would not be natural for voiced sounds to become voiceless between vowels or for fricatives to become stops between vowels. CONSONANTS NATURAL PROCESS TOWARD EXPLAINING WHAT IS “NATURAL” (1) INTERVOCALIC VOICING {f, s, x}  [v, z, ɣ] / V__V (somewhat less common) {p, t, k}  [b, d, g] / V__V Voiceless fricatives become voiced between vowels. Somewhat less common, voiceless stops become voiced between vowels. Presumably the obstruents take on the voicing quality of the surrounding vowels. (2) INTERVOCALIC SPIRANTIZATION {b, d, g}  [β, ð, ɣ] (somewhat less common) {p, t, k}  [f/ɸ, θ/s, x] Voiced stops become fricatives between vowels. Somewhat less common, voiceless stops become voiceless fricatives. This has sometimes been characterized as assimilation to the [+continuant] feature of vowels, but more likely is failure to quite make the oral closure for the stop in transitioning from one vowel to the next. (In practical terms, maybe this is what “assimilation to the [+continuant] feature” means!) (3) INTERVOCALIC SONORIZATION {b, d, g}  [w, ɾ, w] / V__V (somewhat less common) {p, t, k}  [w, ɾ, w ~ h] / V__V Related to INTERVOCALIC SPIRANTIZATION is INTERVOCALIC SONORIZATION. Sometimes they will be mixed, for example, in Kuria, /b/  [β] whereas /d/  [ɾ]. The explanation for sonorization is probably the same as for spirantization, that is, failure to quite make oral closure between the vowel, but in sonorization the oral constriction is even “weaker” than for spirantization.Linguistics 120A 3. Natural Classes, Natural Rules, Rule Ordering 25 (4) FINAL DEVOICING {b, d, g, v, z, ɣ}  [p, t, k, f, s, x] / __]word Voiced obstruents become devoiced at the end of a word. Much less commonly, sonorant consonants devoice. No one has come up for a really good explanation for why this process is so widely attested. It seems to have to do with the extra effort required to sustain voicing of an obstruent when not transitioning to a following vowel. (5) OBSTRUENT VOICING ASSIMILATION [-sonorant]  [αvoice] / __ ! -sonorant"voice# $ % & ' ( Example: /b/  [p] / __t /p/  [b] / __d (An obstruent—stop, fricative, affricate—becomes voiced before a voiced obstruent and voiceless before a voiceless obstruent.) It is uncommon for languages to tolerate a sequence of two obstruents that disagree in voicing. If this takes the form of an active process, it is almost always the first one that assimilates to the second one (“regressive” assimilation). The motivation is clearly that it is hard to make a transition in voicing over the short duration that obstruents usually have, so the first consonant anticipates where the sequence will end up in terms of the [voice] feature. (6) (a) PALATALIZATION (CORONALS) {t, d}  [t!ʃ, d!ʒ] / __{i, e} {s, z}  [ʃ, ʒ] / __{i, e} (b) PALATALIZATION (VELARS) {k, g}  [t!ʃ, d!ʒ] / __{i, e} Both alveolar and velar consonants often become alveopalatal affricates