Survey of Chinese Historical Syntax Part II: Middle Chinese

Abstract

This is the second article in a two-part introduction to Chinese historical syntax. The previous article introduces aspects of pre-medieval grammar which differ markedly from modern Chinese varieties, specifically fronting of object NPs to preverbal position, the asymmetry between subject and object relative clause formation, and the encoding of argument structure alternations like active and passive. Each of these characteristics is related to morphological distinctions on nouns, verbs, or pronouns which are either overtly represented in the logographic writing system in Archaic Chinese or have been reconstructed for (Pre-)Archaic Chinese. In this second article, I discuss changes which took place in Middle Chinese and correlate these innovations with the loss of the (Pre-)Archaic Chinese morphology. The main goal of these articles is to highlight a common denominator, i.e. the morphology, which enables a systemic view of pre-medieval Chinese and the changes which have resulted in the striking differences observed in Middle Chinese and beyond.

1. Introduction

This paper is the continuation of an article on historical Chinese syntax covering the Pre-Archaic and Archaic periods. The current paper examines developments which took place in Middle Chinese, beginning with Early Middle Chinese of approximately the 1st century BCE. I draw particular attention to correlations between syntactic changes and the loss of verbal and nominal morphology. This allows in turn for the identification of morphological triggers for the acquisition of new syntactic forms.

2. Loss of Morphology for Case and Clausal Nominalization

In this section, I discuss the loss of case distinctions and show how this accounts for the concomitant neutralization of the subject/object relativization asymmetry.

2.1. Loss of Case Distinctions

The loss of case distinctions can be demonstrated by the mixed use of genitive and accusative pronouns in positions formerly restricted to accusative pronouns. In Late Archaic Chinese, subject position in a clause embedded by a causative verb received accusative case.¹ (1a) shows an example of the third person accusative pronoun in this position. From Early Middle Chinese, however, the genitive pronoun qi came to be felicitous in this position, as shown in (1b).

Later in Middle Chinese, qi could be found as the goal in a double object construction (2a). (2b) shows that this was an accusative position in Late Archaic Chinese.

I interpret the appearance of the genitive pronoun in positions historically reserved for the accusative pronoun as evidence that, at least in the spoken language, the morphological distinction between these two pronouns had been lost.

Further support for the proposal that genitive case was lost in early Middle Chinese comes from the fact that subjects of other types of embedded clauses were no longer required to appear with the genitive marking. (3a) shows a Late Archaic period sentential subject with genitive marking. (3b) shows a similar sentence from an Early Middle Chinese text, which does not show genitive case for the embedded subject.

2.2. Loss of the Subject/Object Relativization Asymmetry

The distinction between subject and object relative clause formation started to break down in the beginning of the Middle Chinese period. Whereas relative clauses formed on object position in Late Archaic Chinese required the morpheme suo to appear at the VP boundary, as in (4a), examples lacking suo can be found in Early Middle Chinese, as in (4b). Note further generalization of the subject relativization strategy employing zhe in (4b).

Suo relative clauses did continue to appear in written texts until modern times. They are even found on a limited basis in Modern Mandarin (see Chiu 1993, 1995; Ting 2003, 2010 for discussion and analysis). Suo also continued play a role in Middle Chinese in forming one of the passive constructions, as I discuss below in Section 4. But both subject and object relative clauses in modern Mandarin (particularly in the spoken language) now generally use the linker de.

In the first of this series of articles, I discussed the relationship between the nominalization of embedded clauses in Late Archaic Chinese and the need for suo in the edge of VP (vP) to form a relative clause on object position. It should not be surprising, then, that the loss of the obligatoriness of suo in object relative clauses correlates with the loss of genitive case and consequently the loss of the nominal layer in embedded clauses. Since it was genitive case on the embedded subject which signaled the nominalization, I assume that this case was the trigger or cue (in the sense of Dresher 1999; Lightfoot 1991, 1999; Roberts 1997; Roberts and Roussou 2003; and others) for acquiring the nominalized structure. Put simply, the change involved the acquisition of the default finite clausal structure including a CP layer, which could house an operator to form a relative clause. This broke down the distinction between the two types of relative clause, since an operator could move to the CP layer from either subject or object position. The earlier subject relativization strategy involving zhe was generalized to both subject and object relativization because of its high structural position, which allowed it to bind the operator in the edge of the relative clause CP.

The loss genitive case also had consequences for changes in subject relative clauses. In Late Archaic Chinese, the genitive marker functioned as a linker between the modifying clause and the head NP in a headed relative clause, as in (6a). In Middle Chinese, we find a gradual increase in examples which have no linking element, as in (6b, c).

There are also examples of zhe being co-opted to fill various functions previously performed by the genitive marker zhi. As Feng (1990), L. Jiang (1999), and Lü (1943) point out, zhe can be found marking possessors in Middle Chinese texts.

The graph zhe was eventually replaced by di (Cao 1986; Feng 1990; L. Jiang (1999); Lü 1943; Ohta 1958; and others). The following examples show di with a possessor, modifier, and as the linker between a relative clause and the head nominal. In short, di at this time has all the functions of Modern Mandarin de. The graph di has been replaced in Modern Standard Mandarin with de.

In this section, I showed how a single relativization strategy emerged in Middle Chinese for relative clauses involving gaps in both subject and object position. The loss of the nominal layer in embedded clauses led to the addition of a CP layer to which operators could move from either subject or object position. Consequently, the subject relativization strategy was generalized to object relative clauses. This continues to be the strategy for forming relative clauses in Modern Mandarin.

3. Word Order

In this section, I discuss the loss of Archaic Chinese movement transformations and discuss possible connections with loss of morphology.

As suggested in part 1 of this series, pronoun fronting to negation might be analyzed as object shift motivated by the need to receive structural case. Recall that Late Archaic Chinese accusative pronouns fronted from their base positions to a position immediately following the negator, as in (9a). On the other hand, dative pronouns did not front, as shown in (9b). Pronoun fronting was lost in Early Middle Chinese. (9c) shows an Early Middle Chinese example in which an accusative pronoun does not front. If, as I have suggested in the earlier article, the motivation for pronoun fronting to negation was case, then loss of pronoun fronting was simply due to the loss of morphological case distinctions.

In the first part of this series, I also proposed (with Djamouri et al. forthcoming; Meisterernst 2010) that object focus fronting involved an embedded nominalization in a cleft construction. (10a) repeats an example, with the focused object preceding the genitive case marking the nominalization. Djamouri et al. (forthcoming) consider the Pre-Archaic Chinese clefts to be parallel to the focalization construction in Modern Mandarin clefts using the copula shi…de, as analyzed by Paul and Whitman (2008). As shown in (10b), the focused constituent follows the copula shi, and the rest of the clause is followed by de, which they analyze as heading an aspect projection.

An important difference between the (Pre-)Archaic and Modern constructions is that object fronting is allowed only in the former. The focused constituent following the copula in Modern Mandarin can only be a subject or adjunct. If we accept Meisterernst’s (2010) analysis in which the object moves to the projection headed by the genitive marker, then the loss of genitive morphology provides an account of the loss of fronting.

Wh-movement was likewise lost in Early Middle Chinese. Early examples of wh-in-situ involved phrasal categories, as in (11a). Monosyllabic wh-words continued to front, as in (11b).

However, even the fronting of monosyllabic wh-words was no longer the syntactic movement to the clause-medial focus projection that it was in Late Archaic Chinese. In Early Middle Chinese, the wh-word is merely reordered the left of the verb which selects it. In (12a), the wh-word appears left-adjacent to the embedded verb, even though it takes scope in the matrix clause. In Late Archaic Chinese, a wh-word taking scope in the matrix clause would move to the focus position in the matrix clause, as in (12b).

Aldridge (2012) proposes that syntactic wh-movement was reanalyzed as cliticization as an intermediate stage in the change to wh-in-situ. But it is at least as plausible to imagine that wh-movement was lost completely in the spoken language by Early Middle Chinese, though the appearance of movement was maintained in the written language by means of local reordering of the verb and monosyllabic wh-words.

It is difficult to identify a morphological trigger for the loss of wh-movement. Aldridge (2010) also does not provide a detailed analysis, only suggesting that the motivation for the movement may have become opaque to acquirers of the language. Aldridge analyzes Late Archaic Chinese wh-movement as focus fronting to a position between the subject and VP. Late Archaic Chinese also had focus fronting of an NP object, which was likewise lost in Middle Chinese, as discussed above. It is possible that the loss of NP focus fronting may have removed some of the motivation for learners to posit a focus feature driving movement to a clause-medial position. Without a robust trigger for acquisition of the movement, learners would have simply opted for the default parameter setting, i.e. for the lack of movement. Roberts (1997) proposes a similar explanation for the loss of object shift in English.

In this section, I have proposed that the loss of morphology for case and nominalization removed the trigger for the acquisition of several movement transformations, resulting in the loss of these object fronting operations.

4. Passive and Causative

Sound changes taking place in Middle Chinese served to simplify syllable structure and obscure earlier morphological processes (Feng 1998; Norman 1988). The causative/denominal *s- sometimes produced consonant clusters which survived into Middle Chinese, e.g. *sr- of shi‘send’. In other cases, sound change reduced the earlier cluster to a new, single segment, as in *xok‘black’ (< *s- + mok‘ink’).³ The *-s nominalizing suffix became the departing tone in Middle Chinese and can still be seen in Modern Mandarin, as in zhuàn‘record, biography’, derived from chuan‘transmit’.

4.1. Passive

Consequently, it is unsurprising that bare passives were no longer productive in Middle Chinese; passives in Middle Chinese are all overtly marked. The common passive form at the end of the Late Archaic Chinese period was the jian passive, as discussed in the preceding article.

Another type of passive, which emerged in the Late Archaic period and grew significantly in frequency in Early Middle Chinese, appears to embed an object relative clause using suo under the copula wei (Dong 1998; Wei 1994; Yan 1995).

Peyraube 1989b, Tang (1987), Wang (1958), and Yan (1995) assume that wei…suo passives developed from an earlier construction using just wei in the Late Archaic period. In wei passives, the agent appears between wei and the verb.

Chou (1961), Cikoski (1978), Ma (1898), Peyraube (1989b), Pulleyblank (1995), Tang and Zhou (1985), Wang (1958), Yang and He (1992), and others analyze both jian and wei as passive auxiliaries selecting the main VP. Wei (1994) takes a different tack and instead analyzes wei as a copula taking a nominal (rather than verbal) complement. There are even rare examples in which genitive marking can appear between the agent and verb following wei, clearly arguing for a nominal analysis of the constituent following wei.

As a replacement for the wei passive, the wei…suo passives are also argued to involve the copula wei taking a nominal complement, this time the complement being a headless relative clause formed by suo (Dong 1998; Wei 1994; Yan 1995). One possible reason for the replacement of the nominalization, as in (15) and (16), with a headless relative structure, as in (14), may have been the loss of morphology marking the embedded nominalization in Middle Chinese. In other words, the addition of the relativizer suo was necessary in Middle Chinese in order to mark the constituent following the copula as nominal.

Peyraube (1989b) analyzes wei as a preposition taking the agent NP as its complement rather than as a copula. Li (2011) counters, however, that wei could not have been a preposition, because it does not form a constituent with the following NP. The material following wei can be coordinated to exclusion of wei, clearly showing that wei does not form a constituent with the following NP.

Bennett (1981), Sun (1996), Wang (1958), and (Wei 2003) propose that the use of bei in passives (the passive marker in Modern Standard Mandarin) arose as a replacement, first for jian (Wei 2003) and later for wei in wei…suo passives (Bennett 1981; Ohta 1958; Sun 1996; Tang 1987; Wang 1958; Wei 1994). (18) shows an early example from the 2nd century BCE. Note that early examples of bei passives do not contain an agent NP following bei. This fact clearly precludes an analysis of bei as a preposition, as proposed by Li and Thompson (1974) and Peyraube (1989b, 1996).

It was not until after the Han Dynasty (approximately from the 3rd century CE) that examples began to emerge in which an agent intervenes between bei and the main verb. According to Wei (1994), bei passves overtook wei...suo in passives in frequency in the Sui period (late 6th century). Wei (1994:310) suggests that the decline of the wei...suo passive is related to the replacement of wei with shi as copula and the loss of the object relativizer suo.

Evidence that bei, like its predecessor wei, is not a preposition comes first from Hashimoto (1987), who shows that bei does not form a constituent with the following NP. He proposes instead that modern bei should be analyzed as a verb which embeds a clause. The agent occupies subject position in the embedded clause. Analyses along these lines have been developed and defended for Modern Mandarin by Chiu (1995), Feng (1995), Huang (1999), Huang et al. (2009), Ting (1998), and others.

Wei (1994:321) provides additional evidence from historical sources for the biclausal analysis. In the following Western Jin (3rd century) example, a pronoun coreferential with the subject can appear in the embedded VP. If the subject preceding bei were the subject of the same clause containing the object pronoun, then a violation of Condition B of Chomsky’s (1981, 1986) Binding Principles should ensue. Therefore, the subject and object cannot be clausemates.

In this way, bei passives came to acquire the properties they have in Modern Mandarin. In particular, bei can be followed by the agent, and the matrix subject can be separated from its argument position by a clause boundary. These first two characteristics are illustrated in (22a). Finally, (22b) shows that a gap is not required.

(22)	a.	Zhangsani	bei	[Lisi	pai	jingcha	[zhuazuo-le e]]
		Zhangsan	pass	Lisi	send	police	arrest-perf
		‘Zhangsan suffered Lisi sending police to arrest him.’					(Huang et al. 2009:125)
	b.	Zhangsan	bei	Lisi	daduan-le	yi-tiao	tui.
		Zhangsan	pass	Lisi	hit-asp	one-cl	leg
		‘Zhangsan had a leg (of his) broken by Lisi.’

In this subsection, I have also suggested that the Middle Chinese wei...suo passive may have arisen as a replacement for an earlier copula construction involving wei and a nominal complement. If this is the case, this is yet another example of a change induced by the loss of nominalization morphology in Early Middle Chinese.

4.2. Verb-Complement Construction

In this subsection, I consider a possible connection between the development of verb-resultative compounds and the loss of causativizing morphology. Recall from the previous article in this series that Archaic Chinese had an unaccusative/causative verb alternation which was mediated by verbal affixes invisible to the writing system. The verb huai could be used transitively (23a) or intransitively (23b) in Archaic Chinese.

Feng (2005) proposes the following analysis of this alternation within the framework developed by Hale and Keyser (1993, 2002) for analyzing causative derivations from an unaccusative root as involving the addition of a causativizing light verb and its concomitant external argument causer. (24a) shows the unaccusative structure with no causative light verb and no external argument. The internal argument is the highest NP in the extended verbal projection, so it will raise to subject position, yielding SVO word order. In the causative derivation in (24b), the causative morpheme functions as the light verb. The external argument is selected in the specifier of the light verb. The verb root moves to the position of the light verb, and the causative morpheme is spelled out as an affix on the verb. For Archaic Chinese, this would be seen in the voicing alternation or the presence or absence of the sonorant prefix reconstructed by Baxter (1992), Baxter and Sagart (1998), and Pulleyblank (1973).

Mei (1991) relates the loss of causative morphology in part to the emergence of the verb-complement construction, or what is more commonly referred to in modern Chinese linguistics as V-V compounds (Y. Li 1990, 1993), resultative verb compounds (Chang 1998; Cheng and Huang 1994; Li and Thompson 1981), or simply resultatives (Sybesma 1999).

(25)	Zhangsan	qi-lei-le	liang	pi	ma.	(Cheng & Huang 1994: 188)
	Zhangsan	ride-tired-asp	two	cl	horse
	‘Zhangsan rode two horses tired.’

Object sharing by two transitive verbs was productive by the Han period (S. Jiang (2005); Liang 2005; Mei 1991; Ohta 1958; Shimura 1984; Wang 1958), as shown in (26a). If the second verb was intransitive, however, there could be no object in the Han period, as in (26b).

Mei (1991) and Shimura (1984) date the emergence of the resultative contruction of the type in (27) to the Six Dynasties period (5th century CE). An object was permitted following an intransitive verb only from the Tang period.

Mei posits a connection between the loss of the Archaic Chinese causativizing morphology and the neutralization of the transitivity alternation in V-V sequences seen in (26). Specifically, when the phonological distinction marking transitivity was lost, the intransitive variant remained and could assume the place of the former transitive variant in V-V sequences.

In this section, I have shown how the development of the Modern Mandarin passive construction, as well as verb-resultative constructions, may have been related to the loss of nominalizing morphology (in the case of the passive) and causativizing verbal morphology (in the case of VV compounds).

5. Conclusion

In the first of this two-part series of articles, I showed how many central features of Pre-Archaic and Archaic Chinese syntax were related to morphological alternations which have since been lost in the language. In this sequel article, I have followed up on this introduction by showing that many of the salient changes which took place in Middle Chinese were the direct or indirect consequences of the loss of this morphology. This approach to Chinese historical syntax firmly grounds the present work within the growing consensus that (Pre-)Archaic Chinese was a morphologically complex language despite the outward telegraphic appearance afforded by the writing system. Furthermore, this focus on the interaction between morphology and syntax provides a unifying source for a here-to-fore seemingly disparate collection of syntactic characteristics. Finally, the relationships drawn between morphology and syntactic processes and constructions helps to identify triggers responsible for the changes observed in Middle Chinese.

Let me conclude this series of articles by highlighting the advantages of the present proposal against the backdrop of an earlier global approach to syntactic change in Chinese. Li and Thompson (1974) proposed that basic word order in Chinese has been in the process of changing SVO to SOV since the end of the Archaic period. Most of the evidence for this shift comes from the positioning of adjunct PPs. This claim, if it could be substantiated, would identify a single characteristic correlating word changes with developments in other aspects of the grammar, for example the passive construction in which the agent surfaces in preverbal position. However, the Li and Thompson proposal suffers from numerous conceptual and empirical problems and consequently fails to identify a unifying factor accounting for multiple changes from Archaic to Middle and Modern Chinese.

The first problem is their very assumption that the position of (adjunct) PPs should be a main determinant is establishing basic word order in a language. As argued by Djamouri et al. (forthcoming), Light (1979), Sun (1996), Sun and Givon (1985), and many others, if the relative positions of verb and direct object are examined, basic word order in Chinese has been and remains VO.

Another problem with the Li and Thompson proposal is the fact that there is no fixed position for PPs as such. As noted by He (1989, 1992), Sun (1996), Zhang (2002), and others, a PP can be found preceding or following a verb in both Archaic and Modern Chinese. In Modern Standard Mandarin, adjunct PPs tend to surface in preverbal position, while argument PPs appear post verbally, within the VP, a pattern which was solidified by late Middle Chinese (Hong 1998, Zhang 2002). The preverbal locative in (28a) is contrasted with the postverbal goal in (28b).

Argumental PPs, especially those projected by the dative/locative preposition yu, were likewise postverbal in (Pre-)Archaic Chinese (He 1989, 1992; Hong 1998; Qian 2004; Shen 1992; Sun 1996; Wang 1958; Zhang 2002; and others).

The main difference between Archaic and Modern Chinese was the limited freedom in the former to place adjunct PPs in post verbal position. For example, PPs introduced by zi‘from’ could appear either preceding or following the verb, as noted by Hong (1998) and Zhang (2002). Zhang (2002) points out, however, that zi PPs could be post-verbal only when they were the sole constituent following the verb. Therefore, it is certainly not the case that post verbal placement of adjuncts was entirely unconstrained.

It is true that the proportion of post verbal adjunct PPs does decrease during the Middle Chinese period. But this is less the result of a shift in basic word order and more a consequence of independent changes taking place in the language. As noted by He (1992) and Hong (1998), locative constituents began surfacing as bare NPs in post-verbal position in Early Middle Chinese. This is illustrated by He’s (1992:240) following comparison of parallel passages in the two historical chronicles Zuozhuan (5th century BCE) and Shiji (1st century BCE). The preposition is present in the earlier text but missing in the latter. Clearly, this change represents a change in category rather than position.

Grammaticalization of verbs heading modifying (adjunct) VPs also led to the creation of prepositions which could only surface in preverbal position (Hong 1998; Huang 1978; Sun 1996). For example, the Modern Mandarin source preposition cong‘from’ grammaticalized from a verb meaning ‘follow’ in a construction like the one exemplified by (32a), in which cong heads an adjunct VP modifying another VP. Whitman (2000) proposes that the grammaticalization process involved changing the category of the head of the adjunct from V to P. Since no other structural changes were involved, the newly created PPs continued to surface in the same prevebal adjunct position, as shown in (32b). This grammaticalization process, then, also reflects a change in category and not a shift in basic word order.

Finally, let me point out that some of the basis for Li and Thompson’s assumption of a drastic increase in the occurrence of preverbal PPs is the result of misanalysis of the data. One such case is the passives in which the agent appears in preverbal position following the passive marker bei. I showed in Section 3 that bei does not form a constituent with the agent NP. Consequently, bei passives do not involve PPs and therefore do not constitute evidence for Li and Thompson’s (1974) claim.

Another case in point is the ba disposal construction. In the modern Mandarin disposal construction, the semantic object of a transitive verb appears in preverbal position following the functional morpheme ba (Bender 2000; Chao 1968; Cheung 1973; Goodall 1987; Huang et al. 2009; Y.-H. Li 1990; Li and Thompson 1981; Sybesma 1999; Zou 1993; among many others).

(33)	Ni	zenme	ba	yi	ge	zei	pao	zou	le?	(Bender 2000:109)
	you	how	BA	one	cl	thief	run	away	cl
	‘How did you let a thief get away?’

Until approximately the 7th century CE, ba was a lexical verb meaning ‘hold’ or ‘grasp’ (Wang 1958; Zhu 1957), as shown in (34a). It is commonly agreed that the modern disposal construction has its origin in an object sharing serial verb construction of the type in (34b).

Many linguists assume that ba grammaticalized from a verb into a preposition (Bennett 1981; Chao 1968; Her 1991; Jiang 2005; Y.-H. Li 1990; Li and Thompson 1974; Mei 1990; Peyraube 1989a, 1996; Sun 1996). However, it can be demonstrated that ba (like the passive marker bei) does not form a constituent with the following NP and consequently cannot be analyzed as a preposition. As Paul (2002) and Whitman (2000) point out, ba cannot be dislocated with the following NP, as shown in (35).

(35)	a.	Ta	ba	na		shi	ban	hao	le.
		she	ba	dem	cl	matter	do	complete	asp
			‘She took care of that matter.’
	b.	*Ba	na	jian	shi	ta	ban	hao	le.
		ba	dem	cl	matter	she	do	complete	asp
			‘She took care of that matter.’

Li (2006:382) further shows that the NP and VP following ba can be coordinated to the exclusion of ba, indicating that ba does not form a constituent with the following NP.

(36)	Ta	ba	[men	xi-hao],	[chuanghu	ca-ganjing]	le.
	he	BA	door	wash-finish	window	wipe-clean	asp
	‘He washed the door and wiped the windows clean.’

Clearly, then ba cannot be analyzed as a preposition. Consequently, it, too, cannot be used as evidence for Li and Thompson’s (1974) claim regarding PPs and basic word order. Ba is generally analyzed as a functional category which takes the following NP-VP sequence as its complement (Bender 2000; Huang 1997; Paul 2002; Sybesma 1992, 1999; Whitman 2000; Whitman and Paul 2005; Zou 1993).⁴

In sum, other than a sharpening of the distinction between VP-internal argument and VP-external adjunct positions, Chinese basic word order has not undergone any fundamental changes. The claim made by Li and Thompson (1974) is therefore not tenable. In contrast, the present series of articles offers a systemic view of Chinese diachronic syntax based on careful investigation of the structural properties of the constructions involved. This approach was shown to illuminate interrelated properties of the synchronic grammars of different periods, as well as identify a primary unifying factor responsible for a series of changes that characterize Middle and Modern Chinese grammar.

Short Biography

Edith Aldridge is a generative syntactician with a research focus on language variation and change. She has published in the Journal of East Asian Linguistics and The Linguistic Review on interrogative constructions in Archaic and Middle Chinese. She has also published work in collected volumes on a variety of other topics in Archaic Chinese, including pronoun fronting to negation in Movement Theory of Control (Benjamins, 2010), DP structure in Historical Syntax and Linguistic Theory (OUP, 2009), and reflexive pronouns in the Proceedings of the 2nd Meeting of the International Conference on East Asian Linguistics (Simon Fraser University Working Papers in Linguistics, vol. 2, 2009). Aldridge also an interest in Japanese historical syntax, publishing an analysis of Old Japanese wh-movement in Japanese/Korean Linguistics, Volume 17 (CSLI, 2009) and the premodern Japanese writing system hentai kambun in the Journal of East Asian Linguistics and Language Change in East Asia (Curzon, 2001). Aldridge has presented work on Chinese Historical Syntax as an invited speaker at the 7th International Symposium on Ancient Chinese Grammar in Roscoff, France in 2010, at the City University of New York in 2009, and at the Workshop on Relative Clauses at the University Victoria in 2011. The other prong of Aldridge’s work focuses on comparative Austronesian syntax, especially ergativity and verb-initial word order in Austronesian languages, which was the topic of her 2004 PhD dissertation from Cornell University. She also has publications on aspects of Austronesian syntax in Linguistic Inquiry, Lingua, and Language and Linguistic Compass: Syntax and Morphology. Additional work has appeared in several collected volumes, including Structure of Clefts (Benjamins, in press), Grammatical Change: Origins, Nature, Outcomes (OUP, in press), and Continuity and Change in Grammar (Benjamins, 2010). Aldridge is currently assistant professor in the Department of Linguistics at the University of Washington. She was a Mellon postdoctoral fellow at Northwestern University from 2005 to 2007 and a visiting assistant professor in the Linguistics Department at Stony Brook University from 2002 until 2005.

Footnotes

• *

  Correspondence address: Edith Aldridge, Department of Linguistics, University of Washington, Box 354340 Seattle, WA 98195-4340, USA. E-mail: [e-mail]

• 1

  The case is assigned via exceptional case marking from v in the higher clause.

• 2

  The glosses used in this article are as follows:ACC = accusativeADV = adverbASP = aspectCL = classifierCONJ = conjunctionCOP = copulaDAT = dativeDEM = demonstrativeEXCL = exclamationGEN = genitiveHUM = humblePASS = passivePERF = perfectivePOT = potentialQ = question particleREL = relativizerSG = singular

• 3

  These reconstructions are based on, but slightly simplified from, Baxter and Sagart (1998).

• 4

  Huang et al. (2009) and Li (2006) propose that ba is ambiguous. One type of ba is a clausal head, and their analysis is very close to that of Whitman and Paul (2005) and Whitman (2000). The other type of ba is a verb which takes NP as its complement the ba phrase is a modifier adjoined to the main VP.