- Malayo-Polynesian:New Stones in the Wall
Many of the shared phonological innovations that define a Malayo-Polynesian subgroup of the Austronesian languages are either sporadic or apply in very restricted phonological environments, and so are easily overlooked. This squib adds two new pieces of evidence for a Malayo-Polynesian subgroup that are of this type. The first is an apparent conditioned change by which PAN *RVj became PMP *lVj, leaving no instances of the former sequence in MP languages. The second is a complex of changes that altered PAN *SayaN to PMP *anay 'termite'. Although some features of each innovation leave the direction of change indeterminate, others point clearly to the MP languages as the innovative set.
Some linguistic subgroups are established by exclusively shared innovations that affect scores or even hundreds of lexical items. This is true of the Oceanic subgroup of Austronesian, where the merger of PAN *p and *b, of the palatal obstruents *c, *z, *s (plus *j in all but the languages of the Admiralty Islands), and of *e (schwa) and the diphthong *-aw left effects that are widely attested in the lexicon. The preliminary (and now very much dated) list of roughly 700 POC forms in Grace 1969, for example, contains at least 93 items reflecting pre-POC *p or *b, and at least 18 reflecting pre-POC *mp or *mb. There are thus over 100 lexical items in this list alone that exemplify the loss of the voicing distinction between *p and *b or their prenasalized equivalents in Proto-Oceanic. The number of lexical items in the Grace list that are affected by the merger of the palatals (79) and of *e and *-aw (82) is smaller, but nonetheless substantial.
The phonological evidence for a Malayo-Polynesian subgroup containing all of the Austronesian languages spoken outside the island of Taiwan differs fundamentally from that for Oceanic in that most of the exclusively shared phonological innovations found in MP languages are either sporadic, or apply under very restricted conditions. There are at least two exceptions to this statement. First, many of the Formosan languages distinguish *C from *t or *N from *n (or both), and no non-Formosan language does. The apparent merger of *C with *t and of *N with *n in MP languages affects a fairly large number of lexical items, but it has been argued that these segmental distinctions are historically secondary products of earlier stress contrasts (Wolff 1991).1 Second, although PAN *S is reflected as a sibilant in nearly all Formosan languages (Blust 1999:43), it invariably appears as h, [End Page 151] or zero outside Taiwan. Because the Austronesian language family contains upwards of 1,000 languages, and only about 25 of these are or were spoken in Taiwan, this distribution constitutes significant evidence for a Malayo-Polynesian subgroup, even if the shift of a sibilant to a glottal fricative (and then to glottal stop or zero) is a natural change. However, by itself, this change hardly amounts to the powerful phonological evidence that has been used to justify the Oceanic group.
As a consequence of these problems, evidence of exclusively shared innovations that apply in restricted environments has assumed greater importance in establishing the highest-order subgroups of Austronesian. Blust (1995) summarizes those lexically more restricted changes that were known at the time. These include: (1) '*S metathesis' in 15 known examples; (2) merger of *a and *e before *-h in PAN *CumeS > PMP *tumah 'clothes louse', and a handful of other forms; (3) irregular loss of *S in PAN *Sepat to PMP *epat 'four', PAN *Si- to PMP *i- 'prefix of the instrumental focus', and PAN *Sipes to PMP *ipes 'cockroach' (three independent pieces of evidence); (4) change of PAN *Siwa to PMP *siwa 'nine'; (5) reduction of PAN *paNudaN to PMP *paŋdan 'pandanus'; (6) change of PAN *biRbiR to PMP *bibiR; (7) change of PAN *-mu '2PL genitive' to PMP *-mu '2SG genitive'; (8) change of PAN *iten 'ours (INCL)' to PMP *aten 'ours (INCL)'. Constructing the argument for a subgroup with such...