“Again” and “Again”: A Grammatical Analysis of lại and nữa in Vietnamese

Abstract

This work examines the grammatical properties of lại and nữa in Vietnamese, both of which can express the repetition of an event. It has been observed that different syntactic positions of lại result in different readings, as noted in previous studies. When lại precedes a verb, it may assume either the repetitive reading or restitutive reading. When lại follows a verb, it can only assume the restitutive reading. Nữa can be used for the repetitive reading and the incremental reading as well, in the sense that an activity is incremented by adding subevents measured along some dimension, as discussed by Tovena & Donazzan (2008). We adopt Stechow’s (1996) structural analysis and the theory of focus semantics and propose that the preverbal lại is adjoined to vP, which can be focus-associated with an element within its c-command domain, i.e., vP or VP. This is the origin of the ambiguous readings of the preverbal lại. The postverbal lại is adjoined to VP, and this is the reason why it does not yield ambiguous readings. We also propose that nữa is adjoined to vP, along with the movement of vP to a higher functional projection. This results in the surface final position of nữa.

1. Introduction

This work investigates the adverbs of repetition in Vietnamese, lại and nữa, and proposes syntactic accounts for them. The adverb lại can be used alone or with the particle nữa to express the meaning of ‘‘again’’ (Nguyễn, 1997), as shown in (1).

(1)	Trời	lại	mưa	(nữa)	rồi.1
	sky	again	rain	more	PERF
	‘It is raining again.’ (Nguyễn, 1997, p. 147)

The word lại in Vietnamese can be used as a verb meaning “to come”, an adverb, a modal particle conveying the speaker’s attitude, or a sub-element in a sentence connector (Thompson, 1987; Nguyễn, 1997; Trần, 2023, and others). See (2a)–(2d).

(2)	a.	Em	hãy	lại	đây	với	anh!
		you	IMP	come	here	with	me
		‘Come here to me, please!’ (Trần, 2023, p. 252)
	b.	Tôi	lại	yêu	anh ấy	như	ngày		nào.
		I	again	love	him	as	day		which
		‘I fell in love with him again like before.’ (Trần, 2023, p. 255)
	c.	Vì sao	khi	con	kéo	đàn,	bà
		why	when	I	play	violin	grandma
		lại	khóc	vậy		mẹ?
		LAI	cry	such		mother
		‘Why does grandma cry when I play the violin, mom?’ (Trần, 2023, p. 261)
	d.	Ngược lại  /		Trái lại,	anh ta	rất		chăm chỉ.
		however		however	he	very		hard-working
		‘However, he is very hard-working.’ (Trần, 2023, p. 263)

In this work, we focus on the repetitive use of lại. It has been pointed out that different syntactic positions of lại result in different readings (Thompson, 1987; Nguyễn, 1997; Phan, 2013, etc.); see (3a,b). When the adverb lại precedes a verb, it has a repetitive reading. When the adverb lại follows a verb, it can only yield a restitutive reading. According to Phan (2013), in (3a), the entire event of the subject writing a letter is repeated, while, in (3b), only the result state of the event (i.e., the letter having been written) re-occurs.2^,3

(3)	a.	Ông	lại	viết	thư.
		he	again	write	letter
		‘He wrote another letter.’ (Phan, 2013, p. 98)
	b.	Ông	viết	lại	thư.
		he	write	again	letter
		‘He revised the letter.’ (Phan, 2013, p. 98)

Stechow (1996) postulates a structural analysis for the ambiguous readings of the adverb wieder, “again”, in German and argues that the ambiguity of wieder arises from different modifying scopes. Beck and Johnson (2004) further apply this analysis to the ambiguity of the adverb again in English (see also Beck & Snyder, 2001; Beck, 2006). Inspired by these analyses, we propose that the preverbal lại in Vietnamese adjoins to vP. The modifying scope of the preverbal lại can be the entire event or a result state of the event. This gives rise to the two readings, namely the repetitive and restitutive readings. In addition, we argue that the postverbal lại adjoins to VP, and it can only modify the result state of the predicate vP. Therefore, it only yields the restitutive reading. We also show that syntactic tests support the proposed analyses of the preverbal lại and the postverbal lại.

The word nữa, which occurs after the predicate, expresses the meaning “more, in addition, also”, as shown in (4). It can also denote the repetition of an event, as in (5).

(4)	Ông	dùng	cơm	nữa	thôi?
	you	have	rice	more	SFP
	‘Are you going to eat more rice?’ (Thompson, 1987, p. 271)
(5)	Hôm nay	trời	mưa	nữa	rồi.
	today	sky	rain	more	PERF
	‘It is raining again today.’

We propose that nữa adjoins to vP, and, furthermore, it triggers the merger of a FocusP on the phrase structure, to which vP moves. This results in the predicate-final position of nữa.

This paper is structured as follows. Section 2 presents a review (Stechow, 1996). Section 3 examines the preverbal lại and the postverbal lại. Section 4 discusses nữa and its syntactic position. Section 5 outlines our proposal. Section 6 contains the summary.

2. The Structural Analysis of the Repetitive/Restitutive Ambiguity

Stechow (1995) points out that wieder, “again”, in German, exhibits semantic ambiguity, as shown in (6). The adverb wieder, “again”, may assume the repetitive reading (the reading in (6a)) or the restitutive reading (the reading in (6b)).4 In (6a), the whole event “Randi had caught Bockhirsch” is repeated. In (6b), the result state of the event, “Bockhirsch is in the state of being a prisoner,” is repeated.5

(6)	Randi	den Bockhirsch	wieder	fing. (Stechow, 1996, pp. 94–95)
	Randi	Bockhirsch	again	caught
	a. ‘Randi had caught Bockhirsch before, and she caught him again.’
	b. ‘Bockhirsch had been a prisoner before, and Randi caused him to be a prisoner again.’

This repetitive/restitutive ambiguity is also observed in English. See (7). When the whole agentive event “Sally opened the door” is repeated, a repetitive reading is obtained. When the state of the door’s being open is repeated, the restitutive reading results instead. In other words, if the repetition includes the agent of the action, the repetitive reading is yielded.

(7)	Sally opened the door again. (Beck & Johnson, 2004, p. 106)
	a. ‘Sally opened the door, and that had happened before.’	(repetitive)
	b. ‘Sally opened the door, and the door had been open before.’	(restitutive)

Adopting the theory of Voice projection by Kratzer (1994), Stechow (1995, 1996) postulates a structural analysis for the ambiguity of wieder, “again”, as shown in (8) (from Stechow, 1996, p. 97).

(8)

There are two possible positions for wieder in the D-structure of the sentence in (6). When wieder is generated in a higher position, it scopes over the VoiceP, resulting in a repetitive reading. When it is in a lower position, it only scopes over the main predicate (XP here), resulting in a restitutive reading.

Words corresponding to wieder in other languages are also claimed to exhibit such semantic ambiguity. Some authors (Beck & Johnson, 2004; Beck, 2005) apply Stechow’s (1996) theory to again in English for its ambiguous readings in the double-object construction.

3. The Syntactic Position of the Adverb lại

In this section, we discuss the syntactic structures of the preverbal lại and the postverbal lại and how their syntactic positions are correlated with their potential readings.

3.1. The Preverbal lại

To begin with, we examine the syntactic position of the preverbal lại. The strategy that we use consists of comparing the syntactic occurrence of lại with different types of elements of different syntactic positions in the same sentence, thereby determining the structural position of the element lại.

First, the evaluative adverb quả nhiên, “indeed”, and the epistemic adverb có lẽ, “possibly”, must precede the preverbal lại.6 Evaluative and epistemic adverbs are usually assumed to be CP-level adverbials (see Cinque, 1999; Giorgi, 2010).

(9)	a.	Ông	quả nhiên	lại	viết	thư.
		he	indeed	again	write	letter
		‘He indeed wrote another letter again.’
	b.	*Ông	lại	quả nhiên	viết	thư.
		he	again	indeed	write	letter
(10)	a.	Ông	có lẽ	lại	viết	thư.
		he	possibly	again	write	letter
		‘He possibly wrote another letter.’
	b.	*Ông	lại	có lẽ	viết	thư.
		he	again	possibly	write	letter

Second, the preverbal lại must precede the subject-oriented adverb vô tình, “unintentionally”, and the vP-level adverb bình tĩnh, “calmly”, as shown in (11) and (12). Presumably, these adverbials are adjoined to vP.

(11)	a.	Nam	lại	vô tình	làm vỡ	chiếc	bình.
		Nam	again	unintentionally	break	CL	vase
		‘Nam unintentionally broke the vase again.’
	b.	*Nam	vô tình	lại	làm vỡ	chiếc	bình.
		Nam	unintentionally	again	break	CL	vase
(12)	a.	Nam	lại	bình tĩnh	trả lời	câu hỏi.
		Nam	again	calmly	answer	question
		‘Nam calmly answered the question again.’
	b.	*Nam	bình tĩnh	lại	trả lời	câu hỏi.
		Nam	calmly	again	answer	question

Third, the preverbal lại must precede the vP-level aspectual adverb liên tục, “continually” (Cinque, 1999), but it follows the generic-modal adverb thông thường, “normally”, which presumably is a TP-level adverb.

(13)	a.	Gần đây	Nam	lại	liên tục	mắng	Hoa.
		recently	Nam	again	continually	scold	Hoa
		‘Nam has been scolding Hoa continuously recently.’
	b.	*Gần đây	Nam	liên tục	lại	mắng	Hoa.
		recently	Nam	continually	again	scold	Hoa
(14)	a.	Nam	thông thường	lại	đến	phòng	tập
		Nam	normally	again	come	room	do
		thể dục	vào	buổi chiều.
		exercise	in	afternoon
		‘He normally goes to the gym again in the afternoon.’
	b.	*Nam	lại	thông thường	đến	phòng	tập
		Nam	again	normally	come	room	do
		thể dục	vào	buổi chiều.
		exercise	in	afternoon

Fourth, the preverbal laị must precede the passive maker bị, which, according to (Bruening & Tran, 2015), is the head of a predicate, equivalent to v in most current syntactic analyses.

(15)	a.	Nam	lại	bị	Hoa	đánh.
		Nam	again	PASS	Hoa	hit
		‘Nam was beaten by Hoa again.’
	b.	*Nam	bị	lại	Hoa	đánh.
		Nam	PASS	again	Hoa	hit

Fifth, the preverbal lại can take a dynamic or a stative predicate, but it requires the perfect aspect marker rồi when it occurs with a stative predicate.7 This phenomenon is similar to the adverb you, “again”, in Mandarin. Lin and Liu (2009) observe that the adverb you, “again”, in Mandarin requires the presence of the perfect aspect marker le when the main verb is stative. It appears that the same requirement is effective for the Vietnamese adverb lại as well. Compare the examples in (16) and (17).

(16)	a.	Nam	lại	hôn	Hoa.		[dynamic]
		Nam	again	kiss	Hoa
		‘Nam kissed Hoa again.’
	b.	Nam	lại	đánh	Hoa.		[dynamic]
		Nam	again	hit	Hoa
		‘Nam hit Hoa again.’
(17)	a.	Nam	lại	thích	Hoa	*(rồi).	[stative]
		Nam	again	like	Hoa	PERF
		‘Nam likes Hoa again.’
	b.	Nam	lại	béo	*(rồi).		[stative]
		Nam	again	fat	PERF
		‘Nam gets fat again.’

To summarize, the hierarchical distribution of the adverbials discussed so far is shown in (18). Based on this observation, we propose that lại adjoins to vP as its outer-most layer.

(18)

3.2. The Postverbal lại

In this subsection, we turn to the grammatical properties of the postverbal lại. Remember that the postverbal lại only yields a restitutive reading. First, the postverbal lại is only compatible with dynamic predicates, as shown in (19). Ungrammaticality results when it co-occurs with a stative predicate, regardless of whether the perfect marker rồi occurs or not, as shown in (20).

(19)	a.	Nam	sơn	lại	nhà.			[dynamic]
		Nam	paint	again	house
		‘Nam repainted the house.’
	b.	Nam	làm	lại	một	cái	bánh.	[dynamic]
		Nam	do	again	one	CL	cake
		‘Nam made another cake.’
(20)	a.	*Nam	thích	lại	Hoa	(rồi).		[stative]
		Nam	like	again	Hoa	PERF
		Intended reading: ‘Nam likes Hoa again.’
	b.	*Nam	quên	lại	chuyện	này	(rồi).	[stative]
		Nam	forget	again	story	this	PERF
		Intended reading: ‘Nam forgot this story again.’

Second, the postverbal lại cannot occur after the object, as in (21b). It must precede the object, as in (21a).

(21)	a.	Ông	viết	lại	thư.
		he	write	again	letter
		‘He revised the letter.’
	b.	*Ông	viết	thư	lại.
		he	write	letter	again

Third, the postverbal lại must precede the postverbal manner adverb từ từ, “slowly”. The reverse order is ungrammatical.

(22)	a.	Làm	lại	từ từ!
		do	again	slowly
		‘Do it slowly again!’
	b.	*Làm	từ từ	lại!
		do	slowly	again

Fourth, the postverbal lại must precede the completive particle xông, “finish”. Phan (2013) argues that xông takes the “base predicate” (roughly equivalent to VP) as the complement.

(23)	a.	Nam	vừa	sơn	lại	xông	nhà.
		Nam	just	paint	again	finish	house
		‘Nam has just finished the repainting of the house.’
	b.	*Nam	vừa	sơn	xông	lại	nhà.
		Nam	just	paint	finish	again	house

Fifth, the postverbal lại must precede the dynamic modal được, “be able to”, as shown in (24). Ungrammaticality results if lại follows the postverbal modal được.

(24)	a.	Nam	viết	lại	được	lá	thư.
		Nam	write	again	be able to	CL	letter
		‘Nam was able to revise the letter.’
	b.	*Nam	viết	được	lại	lá	thư
		Nam	write	be able to	again	CL	letter

Sixth, the postverbal lại can co-occur with the preverbal lại, as shown in (25).

(25)	Ông	lại	viết	lại	thư.
	he	again	write	again	letter
	‘He revised the letter again.’

To summarize, the hierarchical distribution of the postverbal lại is shown in (26). The postverbal lại must precede postverbal manner adverbs, the completive marker, the dynamic modal, and the object in a linear order.

(26)

4. The Adverb nữa and Its Syntactic Position

In this section, we discuss the grammatical properties of the adverb nữa, “more, in addition, also” in Vietnamese. Before delving into the discussion of nữa, let us introduce the concept of “incremental reading” first. Tovena and Donazzan (2008, p. 91) point out that the repetitive adverb ancora in Italian can give rise to an incremental reading, which means that an activity is incremented by adding subevents measured along a specific dimension. See (27).

(27)	a.	Maria sta ancora leggendo.
		‘Maria is still reading.’
	b.	Maria sta leggendo ancora un libro.
		‘Maria is reading one more book.’

The word ancora in (27a) denotes a meaning comparable to “still,” and the one in (27b) denotes a meaning similar to “more.” The incremental interpretation can be thought of either as a repetition of events or as a continuation of an activity by adding more object units. The adverb nữa in Vietnamese is quite similar to ancora, as it yields an incremental reading, shown in (28) and (29).

(28)	Hôm nay	trời	mưa	nữa	rồi.
	today	sky	rain	more	PERF
	‘It is raining again today.’
(29)	Nam	có thể	ăn	một	bát	cơm	nữa.
	Nam	can	eat	one	bowl	rice	more
	‘Nam can eat one more bowl of rice.’

Now, we turn to the syntactic position of nữa. First, similar to the preverbal lại, the adverb nữa can occur with a stative or dynamic predicate; see (30) and (31). When it occurs with a stative predicate, the perfect aspect particle rồi is required, as shown in (31).

(30)	a.	Nam	hôn	Hoa	nữa.		[dynamic]
		Nam	kiss	Hoa	more
		‘Nam kissed Hoa again.’
	b.	Nam	đánh	Hoa	nữa.		[dynamic]
		Nam	hit	Hoa	more
		‘Nam hit Hoa again.’
(31)	a.	Nam	thích	Hoa	nữa	*(rồi).	[stative]
		Nam	like	Hoa	more	PERF
		‘Nam likes Hoa again.’
	b.	Nam	lại	béo	nữa	*(rồi).	[stative]
		Nam	again	fat	more	PERF
		‘He gets fat again.’

Second, nữa must follow postverbal manner adverbs, such as, for instance, từ từ, “slowly”, in (32). The reverse order is ungrammatical.

(32)	a.	Nam	ăn	từ từ	nữa.
		Nam	eat	slowly	more
		‘Nam ate slowly too.’
	b.	*Nam	nữa	ăn	từ từ.
		Nam	eat	more	slowly

Third, nữa must follow the completive marker xông, “finish”, as shown in (33).

(33)	a.	Nam	(còn)	ăn	bánh mì	xông	nữa.
		Nam	even	eat	bread	finish	more
		‘Nam even finished the bread.’
	b.	*Nam	(còn)	ăn	bánh mì	nữa	xông.
		Nam	even	eat	bread	more	finish

Fourth, nữa must precede the perfect aspect marker rồi. See (34). When the perfect aspect marker precedes nữa, the sentence is ungrammatical.

(34)	a.	Hôm nay	trời	mưa	nữa	rồi.
		today	sky	rain	more	PERF
		‘It is raining again today.’
	b.	*Hôm nay	trời	mưa	rồi	nữa.
		today	sky	rain	PERF	more

Fifth, nữa cannot scope over the epistemic modal adverb chắc chắn, “surely”. For the sentence in (35), only reading 1 is possible, where nữa is within the scope of chắc chắn, “surely”. In reading 2, nữa is intended to scope over the epistemic modal adverb chắc chắn. This reading is not available.

(35)	Nam	chắc chắn	ăn	cơm	nữa.
	Nam	surely	eat	rice	more
	1. ‘Nam surely ate rice again.’
	2. *‘Again, Nam surely ate rice.’

Sixth, nữa cannot scope over the negator không, “not”, as shown in (36). In this sentence, the negator scopes over the particle nữa in readings 1 and 2, and both readings are acceptable. In reading 3, the negator is intended to fall within the scope of nữa. However, this reading is unacceptable.

(36)	Nam	sẽ	không	đi	Mỹ	nữa.
	Nam	will	NEG	go	America	more
	1. ‘Nam doesn’t have any intention to go to the US now.’ (“will” > NEG > nữa)
	2. ‘Nam will not go to the US anymore.’ (NEG > “will” > nữa)
	3. *‘Nam again doesn’t have any intention to go to the US.’ (*nữa >NEG/“will”)

Seventh, nữa can scope over the quantificational subject of a sentence, as shown in (37). When the quantificational subject không ai, “nobody”, in (37) scopes over the particle nữa, reading 1 is obtained. When the particle nữa scopes over the quantificational subject, reading 2 is obtained. Both readings are acceptable.

(37)	Không	ai	đến	nữa.
	NEG	who	come	more
	1. ‘Nobody came again.’ (NP > nữa)
	2. ‘Again, nobody came.’ (nữa > NP)

Eighth, nữa can co-occur with the preverbal lai and the postverbal lai. When these three adverbial elements appear in the same sentence, as in (38), the sentence is fully grammatical.

(38)	Nam	lại	sơn	lại	nhà	nữa.
	Nam	again	paint	again	house	more
	‘Nam repainted the house again. (He also repainted other stuff)’

In summary, the following two sets of properties are observed with nữa. See (39) and (40). In linear order, the particle nữa must follow the postverbal manner adverb từ từ, “slowly”, and the completive marker xông, “finish”, and precede the perfect aspect marker rồi. Regarding the scope property of nữa, it must fall within the scope of the epistemic modal adverb chắc chắn, “surely”, and the negator không, “not”. In addition, nữa may scope over the quantificational subject of a sentence.

(39)	Linear order

(40)	Scope
	a.	chắc chắn, “surely” > nữa
	b.	không, “not” > nữa
	c.	Quantificational subject > nữa
		nữa > Quantificational subject

5. The Proposal

Inspired by the structural analysis of Stechow (1996), we propose an analysis for the preverbal lại, the postverbal lại, and nữa that is partially structural and partially focus-semantic (see Beck, 2006; Ippolito, 2007; and Csirmaz & Slade, 2020, for more focus-based accounts of repetitive adverbs). We agree with Stechow’s proposal that the different readings of a repetitive adverb result from its syntactic position, rather than lexical ambiguity (see Dowty, 1979; and Fabricius-Hansen, 1983, among others). However, the approach we adopt is more flexible, since syntactic adjacency does not completely determine the reading of the adverb in question (for details, see below). We argue that the repetitive adverb can target an element within its c-commanding domain and not only the constituent to which it is directly adjoined.

First, based on our observation of the preverbal lại (summarized in (18)), we propose that the preverbal lại adjoins to vP. Semantically, as a focus particle, it can be associated with the entire event, namely vP, or only with the result state of the event, namely VP. This results in two possible readings, i.e., the repetitive reading and the restitutive reading. See the example in (41) and its syntactic structure in (42). Note that, in Stechow’s theory, wieder (and also the English adverb again; see Beck & Johnson, 2004) can only modify the syntactic domain to which it is directly adjoined. However, in the case of Vietnamese (and Mandarin, too; see Lin & Liu, 2009), non-adjacent focalization is possible. In other words, when lại is adjoined to vP, focalization of the complement of vP, namely VP, is possible in Vietnamese (and Mandarin). This state of affairs is actually a normal case rather than an exception. An example is the focus adverb only in English. In the English sentence “John only bought books”, only may focalize the verb bought or the object NP books. For example, we can have the following two contrasts in mind: “John only bought books and did not buy other things”, in which case books is focus-marked; or alternatively, “John only bought books and did not borrow them”, in which case the verb bought is focus-marked. Having the English adverb only as a paradigm example, we claim that the preverbal lại in Vietnamese is such a focus particle. It can focalize a constituent that is within its c-command domain, immediately adjacent to it or otherwise. We assume that the focalization function of the preverbal lại is carried out by the probe–goal relation of current syntactic theory, which does not require adjacency.

(41)	Ông	lại	làm	một	cái	bánh.
	he	again	make	one	CL	cake
	‘He made a cake again.’ (repetitive or restitutive)

(42)	Preverbal lại (circled areas = possible focus targets)

The proposed analysis explains the position of lại when it co-occurs with other elements, as shown in (16) and repeated in (43).

(43)

The analysis in (42) can account for the structural properties of the preverbal lại shown in (43). If the preverbal lại adjoins to vP, it will necessarily be lower than CP-level adverbs and TP-level adverbs, such as quả nhiên, “indeed”, có lẽ, “probably”, and thông thường, “normally”. In addition, since it is on vP and thus precedes the head v, it naturally precedes the passive maker bị, which is assumed to be the head of a predicate (equivalent to v) (see Bruening & Tran, 2015). And, since we assume that the preverbal lại adjoins to the outer-most layer of vP, it is higher than and, hence, precedes the vP-level adverbs vô tình, “unintentionally” (vP), and liên tục, “continually” (vP).

We also noted that the preverbal lại may occur with dynamic and stative predicates, but the perfect aspect marker rồi is required when it occurs with a stative predicate, as in (16) and (17), repeated below.

(44)	a.	Nam	lại	hôn	Hoa.		[dynamic]
		Nam	again	kiss	Hoa
		‘Nam kissed Hoa again.’
	b.	Nam	lại	đánh	Hoa.		[dynamic]
		Nam	again	hit	Hoa
		‘Nam hit Hoa again.’
(45)	a.	Nam	lại	thích	Hoa	*(rồi).	[stative]
		Nam	again	like	Hoa	PERF
		‘Nam likes Hoa again.’
	b.	Nam	lại	béo	*(rồi).		[stative]
		Nam	again	fat	PERF
		‘Nam gets fat again.’

This phenomenon is similar to the adverb you, “again”, in Mandarin. When you, “again”, occurs with a stative predicate, it also needs the presence of the perfect aspect marker le. Lin and Liu (2009, p. 1188) argue that, when you, “again”, adjoins to a static predicate, it turns the predicate into a dynamic one, namely a change-of-state predicate. We propose that the preverbal lại exhibits the same function. When it occurs with a static predicate, it turns the predicate dynamic. The presence of the perfect aspect marker rồi is therefore required to indicate that the predicate is now a dynamic one.8

As to the postverbal lại, we propose that it adjoins to the lexical VP as its outer-most layer. See (46) and (47) as a demonstration. In sentence (46), the verb làm, “make”, raises to the head of vP, resulting in the postverbal position of lại. Since the postverbal lại only c-commands the VP, its focus domain is limited to the result state of the predicate. Consequently, only the restitutive reading is available for the postverbal lai. This analysis is also compatible with Phan’s (2013) proposal, which posits a projection lower than the vP which denotes the result state of the asserted event, ResultativeP. For simplicity, we use the term VP here.

(46)	Ông	làm	lại	một	cái	bánh.
	he	make	again	one	CL	cake
	‘He made a cake again.’ (only restitutive)

(47)	Postverbal lại (circled area: possible focus target)

This analysis accounts for the structural properties of the postverbal lại shown in (24), repeated as in (48).

(48)

First, if the postverbal lại adjoins to VP, it should precede the object. This accounts for the fact that the postverbal lại must precede the object in a sentence. Second, if the postverbal lại adjoins to VP as its outer-most layer, other VP-level elements should be lower than it. This explains the fact that the postverbal lại precedes the postverbal VP-level manner adverb từ từ, “slowly.” It also explains the fact that lại precedes the completive particle xông, “finish”, and the dynamic modal được, “be able to.” Phan (2013) proposes that the particle xông and the modal được are heads of the projection CompletiveP and ResultativeP, respectively, both of which occur between a higher VP (roughly equivalent to vP) and a lower VP of a sentence. In our framework, the postverbal lại is the outer-most layer of the (generalized) VP; thus, it must be higher than these two elements and precede them. We take xông as an illustration.9

(49)	Nam	vừa	[sơn	lại	xông	nhà].
	Nam	just	paint	again	finish	house
	‘Nam just finished the house painting.’

(50)	Co-occurrence of the postverbal lại with the other postverbal element

In Section 3.2, we observed that the postverbal lại can only be used with a dynamic predicate; it cannot be used with a stative predicate, not even with the perfect aspect marker rồi. See (19) and (20) above. This can be explained by the proposal that the postverbal lại adjoins to the projection denoting the result state of a verbal predicate, namely VP. Only a dynamic verbal predicate has a result state. Furthermore, we may assume that the result state that lại modifies must be the result of an agentive action. Since rồi only introduces a change-of-state meaning to a stative predicate and no agency is brought in, the addition of rồi cannot save the sentence from ungrammaticality. A piece of evidence for this proposal is that sentences (17a,b), where the presence of rồi makes a stative sentence with the preverbal lại grammatical, can only have the repetitive reading. They do not have the restitutive reading. The lack of an acceptable restitutive reading for (17a,b) clearly originates from the fact that stative predicates do not yield a result state, even when they become dynamic by the function of rồi.

Lastly, we turn to the particle nữa, which occurs in the predicate-final position. The particle nữa, in its lexical–semantic nature, is an additive particle that yields an incremental reading. The structural properties of nữa are shown in (39) and (40), now repeated in (51) and (52).

(51)	Linear order

(52)	Scope
	a.	chắc chắn, “surely” > nữa
	b.	không, “not” > nữa
	c.	Quantificational subject > nữa
		nữa > Quantificational subject

We propose that nữa adjoins to vP. We further propose that nữa triggers movement of the vP to a higher functional projection, Spec, FocP, resulting in its being stranded behind the predicate.10 Following the analysis of light predicate raising of Simpson (2001), we assume that the motivation for such movement is to defocus the constituent in question. Examples of such movement include the sentence-final particle kong, “to speak”, in Taiwanese (Simpson & Wu, 2002) and the sentence-final deontic modal in Cantonese and a number of Southeast Asian languages (Simpson, 2001). Let us use the sentence in (30a) as a demonstration, repeated as (53), with (54) as its structural analysis.

(53)	Nam	hôn	Hoa	nữa.
	Nam	kiss	Hoa	more
	‘Nam kissed Hoa again.’

(54)

This analysis can account for the structural properties of nữa shown in (51) and (52). First, if nữa adjoins to vP, VP-level or VP-internal elements such as từ từ, “slowly”, and xông, “finish”, should precede it because the whole VP moves to Spec, FocP, and becomes higher than vP. Second, nữa precedes rồi because rồi is the head of the projection AspP, which we assume triggers the raising of its complement (FocP in this case) to its specifier. See (55) and (56) for illustration.

(55)	Trời	mưa	nữa	rồi.
	sky	rain	more	PERF
	‘It is raining again.’

(56)

Third, if the particle nữa adjoins to vP, it cannot scope over TP-level elements such as the epistemic modal adverb chắc chắn, “surely”, and the negator không, “not”. In addition, as nữa adjoins to vP, it should fall within the scope of a subject quantifier. So, the scope relation “quantificational subject > nữa” in (52c) is obtained. On the other hand, a quantificational subject may undergo quantifier lowering (May, 1985) and assumes its scope position in Spec, vP. In that position, it falls within the scope of nữa since nữa adjoins to vP. In this way, the scope relation “nữa > quantificational subject” is obtained.

Fourth, our proposal can also account for the co-occurrence of nữa with the preverbal lại and the postverbal lại, as shown in (57) and (58). Sentence (38) is repeated in (57).

(57)	Nam	lại	sơn	lại	nhà	nữa.
	Nam	again	paint	again	house	more
	‘Nam repainted the house again. (He also repainted other stuff)’

(58)

6. Conclusions

Inspired by previous works (Stechow, 1996; Beck & Johnson, 2004), we propose that the ambiguity of lại in Vietnamese results from its adjunction to vP and VP. Additionally, the adverb nữa in Vietnamese adjoins to vP and yields an incremental reading. It also triggers vP movement to Spec, FocP, resulting in its predicate-final position.