3. Grammaticalized function words in Cantonese and Sixian Hakka
3.1 Minimal tonal neutralization
3.1.3 Asymmetry between subclasses of sentence-final particles
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Sixian Hakka.
3.1.3 Asymmetry between subclasses of sentence-final particles
We now turn to the tonal behavior of sentence-final particles in Miaoli Sixian Hakka, which is similar but not identical to what we have seen in Cantonese.
The details of sentence-final particles are not as clear for Miaoli Sixian Hakka as for Cantonese, given the fact that this topic is less studied for Sixian Hakka. According to a brief introduction of partial sentence-final particles in Luo’s (1988) pioneering work, final particles in Miaoli Sixian Hakka appear to have inherent tones that can be recognized as one of the four lexical tones (see (11)), as shown below.
(28) Tonal realization of sentence-final particles in Miaoli Sixian Hakka (Luo: 206) ge4 Neutral assertion of relevance
le1 Realization of state lio1 Realization of state
la3 Smooth-alert
mo2 Used in yes or no questions
mang2 Used in yes or no questions about the realization of the state
a1 Interrogative
no3 Interrogative
na1 Adhortative
o1 Adhortative, smooth-alert honn3 Adhortative
hann3 Adhortative
How should we treat these apparent tones? Are they authentic lexical tones or only of localized intonation as what we have argued for Cantonese? Here I observe a clear difference. That is, in Miaoli Sixian Hakka, a proper subset of the sentence-final particles is tonic, having authentic, unneutralized, lexical tones, while the rest (namely the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
complement of the tonic subset) are atonic, which carry a boundary tone/localized intonation, and are therefore toneless, or neutral-toned by definition. The tonic class is quite a small subset, including only a few members, such as ge4, le1, lio1, mo2, mang2. In other words, most of the sentence-final particles are atonic. This difference can be justified in two ways.
First, based on J. Cheng (2007) and J. Wu’s (2018) research, there are pairs/groups of atonic final particles in Miaoli Sixian Hakka that are described as sharing the same segments but differing in the pitch realization only, just like the case in Cantonese. This is shown by the examples in (29), which also shows that the difference in pitch contour simply contributes pragmatic nuances in general. Besides, the semantic/pragmatic contrast between honn3 and honn4 in (29b) and (30) is in line with the discussion about universal boundary tones (reviewed in §3.2.2) in the sense that some specific pitch contours correspond to certain meaning/function. Specifically, high register boundary tone (i.e. T4) expresses look forwarding toward the hearer, while low register boundary tone (i.e. T3 and T2) conveys speaker-oriented committedness. Therefore, by the same token, we may treat each pair/group of the final particles as variant forms of the same sentence-final particle. Note that this kind of variation is hardly reported for the members of tonic class.
(29) Pairs/groups of tonally contrastive sentence-final particles a. Segmental form: a
a2 Smooth-alert, conversational liveliness
a4 Smooth-alert, conversational liveliness, more intensified than a2 a3 Smooth-alert, conversational liveliness, even more intensified than a2
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
b. Segmental form: honn
honn3 Adhortative, interrogative, speaker-oriented, more assertive than honn4 honn4 Adhortative, interrogative, hearer-oriented, seeking for confirmation
(30) Nuance between honn3 and honn4
The other way to justify the tonic vs. atonic subclassification is to draw a comparison with Hailu Hakka, the second dominant Hakka dialect in Taiwan. As is well-known, cognate morphemes in Sixian Hakka and Hailu Hakka normally have reversed tonal melodies. That is, a given cognate morpheme with a high pitch in one of the dialects carries a low pitch in the other, and if a cognate morpheme has a rising contour in one of the dialect, it must have a falling contour in the other, as illustrated in table (31).
(31) Tonal reversion in two Hakka varieties
Cognate Morphemes Sixian Hakka Hailu Hakka
Tien ‘sky’ 24 53
Fung ‘red’ 11 55
Dang ‘top, above’ 31 24
Kon ‘look’ 55 11
Gued ‘country’ 2 5
Pag ‘white’ 5 2
In the domain of sentence-final particles, one may expect that reversion like this would occur in every type of sentence-final particles if all the particles were really associated a. Tien1gong1ngid3 zang4 kon4=honn3
Tomorrow not.until watch=SFP
‘Let’s not watch it until tomorrow. (more assertive, with stronger presupposition)’
b. Tien1gong1ngid3 zang4 kon4=honn3 Tomorrow not.until watch=SFP
‘Let’s not watch it until tomorrow, shall we?’
‧
with a distinct lexical tone. However, according to J. Cheng (2007:164), only members of the tonic class show the tonal reversion. Members of the atonic class, on the contrary, basically have the same tonal melodies in both dialects. This discrepancy, as shown in the following tables, falls out naturally with the assumption made here that the “tone” carried by an atonic final-particle is in fact acquired from an intonational effect, thus involving
no genetically-related correspondence such as the reversion between the two varieties.
(32) Dialectal tonal reversion in sentence-final particles a. Tonic class: reversion applicable
Final Particles Miaoli Sixian Hakka Hailu Hakka
Ge4 55 11
Final Particles Miaoli Sixian Hakka Hailu Hakka
A2 11 11 division of sentence-final particles into three distributional classes, recast as a three-layered root CP by Paul (2014) ― (i) low C, the lowest root complementizer, including sentential tense/aspect particles, such as le, láizhe, and ne1 in Beijing Mandarin;
‧
Mandarin; (iii) Attitude C, the highest root complementizer, including speaker’s subjective attitude particles, such as ou, (y)a, ne3, etc. in Beijing Mandarin. This split-CP architecture is shown below. mang2, as the standard interrogative particles used in yes/no questions, namely equivalent to Mandarin ma, belong to the class of Force C. The remaining tonic particle ge4, though not in the lowest two classes of root complementizer, is not an Attitude C either. As an equivalent to the Mandarin de in the propositional assertion, ge4 should be in the non-root C position, according to Paul’s (2014) analysis. To sum up, the tonic vs. atonic distinction can be translated as follows: only sentence-final particles that belong to the highest root complementizer (i.e. the class of Attitude C) are atonic, or neutral-toned.The question that follows is why tonal neutralization targets sentence-final particles in Attitude C only. A possible account that I shall propose here is that attitude particles are more grammaticalized than final particles of the other classes, as shown in the
‧
sentence-final particles remain in the category of degree 6. This move is in line with the view of J. Cheng (2007) that, in Hailu Hakka, particles with tones, in reverse relation to the corresponding particles in Sixian Hakka (i.e. non-attitude particles), display more contentful properties. particles are the most subjective sentence-final particles. He argues that this is a logical deduction, given that attitude particles as such are always linked to subjective feeling, attitude and/or opinion of speakers rather than the utterance itself, and are therefore speaker-oriented. At the other end, particles in lower C, which convey tense, aspects, and/or modality, are tightly linked to the predication of the utterance itself. In other words, they are utterance-oriented, and are thus considered the least subjective complementizers.Particles in force C such as those expressing interrogative and imperative illocutionary force are, as V. J. Pan puts it, “situated between being subjective and being least subjective” (V. J. Pan, 2015: 821). Specifically, the force particles are more subjective than those particles in lower C because “they integrate the speaker’s expectation from her/his co-speaker” (loc. cit). At the same time, they are also less subjective than attitude particles, since they are still linked to the interpretation of the sentence itself.
The recognition of the high degree of subjectivity for attitude particles leads to the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
second assumption made here: the more subjective, the more grammaticalized. This assumption concerns the tight connection between subjectification and grammaticalization, as Traugott, one of the pioneer explorer of subjectivity, repeatedly show in her analyses of semantic/pragmatic change. She considers subjectification in grammaticalization a historical process by which meanings become increasingly based on the speaker’s subjective belief state, or attitude, toward what is said (Traugott 1989, 1995, 1997, 1999, inter alia). With a view to establishing general characteristics of grammaticalization, in one of her earlier influential works, Traugott (1982) has proposed unidirectional path of semantic change, characterized as ‘propositional > (textual >) expressive, and views this shift as involving subjectification , that is, an increase in the expression of subjectivity or speakers’ point of view. This shift is considered regular enough for Traugott to predict “paths of change, or constraints on the directionality of semantic change” (Traugott 1989:33), and the subjectification is suggested to be viewed as an intrinsically unidirectional phenomenon, that is to say, a trend towards a higher degree of encoding of the speaker’s point of view, not vice versa. This nonsubjective >
subjective tendency is considered an important role in semantic change and is supported by Traugott’s subsequent works (Traugott 1995, 1997, 1999, 2003b), as well as other studies on subjectification and grammaticalization (e.g. König 1991, Smith 1993, Carey 1995, Company 1995, Brinton 1996, Suzuki 1998, Chor 2013).
Having settled the difference in degrees of grammaticality between attitude particles and final particles of the other types, the asymmetry in neutralization between these subclasses in Miaoli Sixian Hakka can be analyzed in the same way as we did for the case of Cantonese. That is, function words with the degree of grammaticalization
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
lower than 6.5 (i.e. Fnc≤6), including sentence-final particles outside Attitude C, are all prosodically prominent enough to prevent their lexical tones from being neutralized.
Again, this means that the distinction between the categories with the degree lower than the highest grammaticalized category are all conflated.
(35) Conflation of Fnc≤6
In the same vein, the neutralization vs. non-neutralization distinction between grammaticality degrees 6 and 6.5 can be captured in the currently proposed framework by ranking ALIGN-R(ω, Fnc≤6) above SP-MAX-X0. This is shown in the tableau (36). In CAND 1 every syntactic word is parsed in a prosodic word ω, including both the non-attitude particle le1 and the attitude particle la0. The latter belongs to the degree 6.5 category on the grammaticality scale, and therefore the prosodic words built on the attitude particle incur one violations of ALIGN-R(ω, Fnc≤6). This problem gets fixed in CAND 2 by leaving the attitude particle unparsed, and in that way this candidate wins out as expected. The other candidates, again, though harmonically bounded by CAND 2, show that further unparsing of function words with lower grammaticality degree (i.e. Fnc≤6) is not allowed for the unnecessary violations of SP-MAX-X0 incurred.
Fnc=1 Fnc=2 Fnc=3 Fnc=4 Fnc=5 Fnc=6 Fnc=6.5
‧
‘As for the three books you brought here, I have read them already.
‧
國立 政 治 大 學