The Dative Markers and Their Developments in Hunan Sinitic Languages

Abstract

This study is based on a sample of 30 Sinitic languages spoken in the Hunan Province. Its first objective is to explore the types of dative markers, comparing the form of the dative with allative, passive, benefactive, and differential object markers in these languages. Five patterns are identified: (I) DAT = ALL (II) DAT = GIVE = OM ≠ PASS; (III) DAT = GIVE = OM = PASS; (VI) DAT = GIVE = PASS ≠ OM; (V) DAT = BEN. Then, we reveal three main possible grammaticalization pathways that motivate the five synchronic patterns: (a) Allative > Dative; (b) (TAKE >) GIVE > Dative; (c) Benefactive > Dative. It concerns two distinct developments for the second pathway. Based on the areal distribution of the various types of dative markers, we can observe how the dative markers are developed in Hunan Sinitic languages.

1. Introduction

This paper discusses the dative (recipient) markers in the Sinitic languages spoken in the Hunan Province.

Before embarking on the examination of the dative markers, it is pertinent to offer a brief introduction to the languages spoken in the Hunan Province. Hunan is located in the south-central region of China. According to Bao and Li (1985), the Hunan Sinitic languages can be classified into five broad areas: Xiang varieties spoken in the center of Hunan; Southwestern Mandarin varieties spoken in the west and south; Gan and Hakka varieties spoken in the east; Waxiang spoken in the west; and Tuhua1 spoken in the south (both within the Mandarin-speaking regions). Additionally, there are several non-Sinitic languages spoken in Hunan: Tujia in the northwest; Miao in the west; Dong in the southwest; and Yao in the south. Of these languages, Tujia exhibits SOV word order, while other languages use SVO order. Hunan is classified as a transitional zone for Sinitic languages in Chappell (2015), such that an examination of the Sinitic languages in Hunan can shed light on the refinement of this linguistic area and contribute to a better understanding of the linguistic development within Sinitic languages.

Dative usually refers to a morphological case that prototypically marks the recipient or indirect object in a ditransitive construction (cf. Haspelmath 2016). In this paper, we use dative marker to indicate the element that introduces the recipient argument, as the morpheme to in I gave a pen to Paul in English. Most studies on the Sinitic ditransitive constructions focus on the sentence structures, especially on the relative word order between the recipient and the theme (see Hashimoto 1976; Zhu 1979; Yue-Hashimoto 1993, for example). Chin (2010) firstly identified two types of dative markers in Sinitic languages (i.e., the go-type and the give-type) and discussed the chronological development of the two types of dative markers. See also Li and Wu (2015) for the development of dative markers in Yichun Gan.

In this paper, we present five types of dative markers from a synchronic perspective on the one hand, and we explore four grammaticalization pathways underlying these various types of dative markers on the other. Grammaticalization is defined as the development from lexical to grammatical forms and from grammatical to even more grammatical forms (Kuteva et al. 2019, p. 3). Cross-linguistically, the dative markers are frequently derived from allative markers, GIVE verbs, and benefactive markers (Kuteva et al. 2019).

In the languages that we investigated, we observe all these three patterns: namely, the dative marker shares the same form as either the allative marker, the GIVE verb, or the benefactive marker in each language. However, in order to clarify how the dative use was developed from these sources, especially for the GIVE verbs which are generally TAKE verbs in origin, we take the passive markers and differential object markers which are closely related to GIVE/TAKE verbs into consideration and refine the three main patterns into five: DAT = ALL; DAT = GIVE = OM ≠ PASS; DAT = GIVE = OM = PASS; DAT = GIVE = PASS ≠ OM; DAT = BEN. The detail of our methodology is presented in Section 2.

The layout of this article is as follows: this introduction leads into Section 2, which presents our definitions for the markers and relevant constructions, our terminology and methodology. We take allative markers, GIVE verbs, benefactive markers, passive markers, and differential object markers into consideration, and classify the 30 languages into five types. They will be discussed in detail in Section 3 with a map showing their areal distribution. The relevant diachronic developments of dative markers are found in Section 4 and are followed by a conclusion in Section 5.

2. Methodology and Definitions

The sample of 30 languages spoken in the Hunan Province covers the five groups of Sinitic languages (i.e., Xiang, Gan, Southwestern Mandarin, Hakka, and Tuhua) and one unclassified one (i.e., Waxiang). Both fieldwork data and data from descriptive grammars and journal articles are used. Data are glossed and translated by the author when not provided in the original literature. The locations of all the languages can be found in Figure 1.

In this paper, only the dative marker in the main2 ditransitive construction of the language is considered. As indicated in Zhang (2011), the Southern Sinitic languages3 use the postverbal dative construction4 [V + T(heme) + DAT + R(ecipient)] to encode transfer. This is the case for most of our sample languages. However, the preverbal dative construction [DAT + R+V + T] is also attested as the main type in some languages spoken in northwestern Hunan, such as in Waxiang. For the languages that use postverbal dative markers, we can identify two primary patterns: DAT = ALL and DAT = GIVE; for the languages that use preverbal dative markers, we can find a third pattern: DAT = BEN.

The dative markers that are related to allative markers and benefactive markers are relatively straightforward to identify and analyze, but when the dative marker shares the same form as the GIVE verb, we need to further confirm how the dative use is developed from the GIVE verb. Regarding the GIVE verbs, it refers to the general-purpose verb of giving in this paper, such as gěi 给 in Standard Mandarin or give in English. GIVE verbs have been extensively discussed in the literature due to their polyfunctionality in Sinitic languages (see Lai 2001; Chin 2011; Ngai 2015; Lu and Szeto 2023). For instance, in Standard Mandarin, gěi can act as a benefactive marker, dative marker, causative verb, passive marker, and differential object marker. However, in the Southern Sinitic languages, the GIVE verbs are more diverse in forms, such as pa⁴¹ 把 in Changsha, te²² 得 in Hengyang, and lɛ⁴⁴ 拿 in Lianyuan from our sample, and they are TAKE5 verbs in origin. Furthermore, in most cases, they manifest different syntactic behaviors from gěi in Northern Sinitic languages. See the two example sentences below in Standard Mandarin and in Huarong.

(1) Standard Mandarin (Zhu 2009, p. 170)
给	我	一	枝	笔。
gěi	wǒ	yì	zhī	bǐ
give	1sg	one	clf	pen
‘Give me a pen.’

(2) Huarong (SWM; own fieldwork)
把	本	书	得	我。
pɑ21	pən21	ɕy53	tɛ13	ŋo21
give	clf	book	dat	1sg
‘Give me a book.’

The GIVE verbs which can take an R argument (in addition to a T argument) will be considered as a genuine GIVE verb in this analysis (cf. Zhang 2011), like the case in Standard Mandarin; while some GIVE verbs can only take a T argument but not an R argument, they solely have the semantic meaning of giving, like the case in Huarong6. When we consider that a dative marker is grammaticalized from a GIVE verb, it is generally the genuine GIVE verbs that we talk about, since it can precede an R argument and can be easily reanalyzed as a dative marker from V₂ position in a serial verb construction, as illustrated by (3).

(3) Standard Mandarin (Zhu 1979, p. 83)
我	送	一	张	票	给	小李。
wǒ	song	yì	zhāng	piào	gěi	Xiǎolǐ
1sg	offer	one	clf	ticket	dat	Xiaoli
‘I offered a ticket to Xiaoli.’

In Hunan, most GIVE verbs are originally TAKE verbs, and they form postverbal dative constructions but not double object constructions, which means they take a T argument but not R argument. However, some of these GIVE_<TAKE verbs can also be used as a dative marker. Take Changsha, for example: the GIVE_<TAKE verb and the dative marker are both pa⁴¹ 把. The process of how the dative marker is developed from a GIVE_<TAKE verb will be discussed in detail in Section 4.2.

Zhang (2011) claims Southern Sinitic languages lack genuine GIVE verbs, and the ditransitive constructions in Southern Sinitic languages are formed by the combination of TAKE verbs and a directional element which can be templated as [TAKE + T + ALL + R]. It is exactly in this construction that TAKE verbs obtained the semantic meaning of giving. On the basis of this construction, when the T argument shows preverbally (could be topicalized or marked by an object marker) or simply mentioned in the previous context, we have [TAKE + ALL + R], and according to Zhang (2011), [TAKE-ALL] becomes a genuine GIVE verb, which can precede the R argument directly, and form a serial verb construction with the GIVE_<TAKE verb in ditransitive constructions, whence it can be reanalyzed as a compound dative marker: [GIVE_<TAKE + T + DAT_<[TAKE-ALL] + R]. Then, due to the frequent use, the allative marker might be omitted from the compound form, leaving the TAKE verb in the language as a dative marker alone: [GIVE_<TAKE + T + DAT_<TAKE + R].

These developmental stages can still be observed in many Southern Sinitic languages, such as in Shanghai Wu (Qian 1997), Yichun Gan (Li and Wu 2015), and Liancheng Hakka (Ye 2023). Take Yichun as an example. At first, the directional element ku⁴² 过 is used to mark the recipient:

(4) Yichun (Gan; Li and Wu 2015)
我	把	本		书	过	你。	[GIVE<TAKE + T + ALL + R]
ŋo34	pa42	pun42		ɕy34	ku42	ɲi34
1sg	give	clf		book	dat	2sg
‘I gave a book to you.’

Then, the GIVE_<TAKE verb combines with the directional element and forms a compound genuine GIVE verb as pa⁴²-ku⁴² 把过 which can take a recipient as its argument. This compound form is then reanalyzed as a dative marker in a serial verb construction:

(5)	我	把	本		书	把过	你。	[GIVE<TAKE + T + DAT<[TAKE-ALL] + R]
	ŋo34	pa42	pun42		ɕy34	pa42-ku42	ɲi34
	1sg	give	clf		book	dat	2sg
	‘I gave a book to you.’

Finally, the directional element dropped off and the GIVE_<TAKE verb alone becomes a new dative marker in Yichun:

(6)	把	本		书	把	你。	[GIVE<TAKE + T + DAT<TAKE + R]
	pa42	pun42		ɕy34	pa42	ɲi34
	give	clf		book	dat	2sg
	‘Give a book to you.’

This is the first possible diachronic development for the dative markers that have the same form as the GIVE_<TAKE verbs. Note that although the dative marker has the same form as the GIVE_<TAKE verb in the language, it does not mean the GIVE_<TAKE has become a genuine GIVE verb, because the dative marker is developed by dropping off the allative element from a genuine compound GIVE verb, but is not developed from the GIVE_<TAKE verb itself. (Nevertheless, it is possible for the GIVE_<TAKE verb to develops further into a genuine GIVE verb on this basis.)

The second possible explanation for a dative marker sharing the same form as the GIVE_<TAKE verbs is the that GIVE_<TAKE verb becomes a genuine GIVE verb through relexicalization (Güldemann 2012), and the dative use is developed from the genuine GIVE verb. This is a case mentioned in Shaowu (Ngai 2015; 2021, p. 384).

To help us tell if the GIVE_<TAKE has become a genuine GIVE verb in the language, we need to take two other makers into consideration: passive markers and differential object makers.

Chappell and Peyraube (2006) argued “GIVE > permissive causative > passive” is a common grammaticalization pathway in Sinitic languages, such as gěi in Beijing Mandarin or 俾 pei³⁵ in Cantonese. Below is an example of the bridging stage for the reanalysis. In (7), gěi can actually be interpreted as ‘to give’ in a pivot construction, a permissive causative marker, or a passive marker.

(7) Beijing Mandarin (Xu 1992)
车	给	小王	修好	了。
chē	gěi	Xiǎowáng	xiū-hǎo	le
car	give/let/pass	Xiaowang	repair-be.good	crs
‘(Someone) gave the car to Xiaowang (and he) repaired it.’
or ‘(Someone) let Xiaowang have the car repaired.’
or ‘The car was repaired by Xiaowang.’

The key point for the reanalysis is that the argument after the GIVE verb has to be an R argument, which can be considered as a causee in the causative construction or an agent in the passive construction. If the GIVE_<TAKE verb in a language can be used as a passive marker, we can consider that the GIVE_<TAKE verb in this language is a genuine GIVE verb, thus making it also possible to develop a dative use. On the contrary, if the GIVE_<TAKE verb in a language cannot be used as a passive marker, we can consider that it might still stay as a TAKE verb, or at least its GIVE use is not yet well developed.

As for the differential object marker, Chappell (2007) outlines two common grammaticalization pathways for the object markers in Sinitic languages: (i) TAKE (>instrumental) > OM; (ii) GIVE > benefactive > OM. If an object marker is developed from the GIVE verb, generally it has to undergo an intermediate stage as a benefactive marker. While the pathway from TAKE to object marker is much more common, such as bǎ 把 in many Northern Sinitic languages. For the languages in Hunan, if the object marker has the same form as the GIVE_<TAKE verb, and we cannot find an identical benefactive marker, then we can consider the object marker in these languages to be more likely to be developed from the TAKE use of their GIVE_<TAKE verbs in the history.

In summary, we take passive makers and differential object markers into consideration to help us decide whether the GIVE_<TAKE verbs in our sample are genuine GIVE verbs that can develop a dative use, or they are still a TAKE verb that has gained a dative use by dropping off the allative element in a compound GIVE verb. When DAT = GIVE = PASS ≠ OM, it is more likely to be the former case; when DAT = GIVE = OM ≠ PASS, it is perhaps the latter case especially when a compound GIVE verb form can be found in the language; finally for the syncretism DAT = GIVE = PASS = OM, both cases are possible. All the grammaticalization pathways will be explained in detail in Section 4.

Therefore, in this paper, we investigate six elements: dative markers, allative markers, benefactive markers, passive markers, differential object markers, and GIVE verbs. According to the data, we can classify the languages in our sample into five types: DAT = ALL; DAT = GIVE = OM ≠ PASS; DAT = GIVE = PASS = OM; DAT = GIVE = PASS ≠ OM; DAT = BEN.

The definitions for the five grammatical markers are given in the following part.

A dative marker or an indirect object marker marks the recipient in a ditransitive construction. It can be a postverbal marker, as in (8), or a preverbal marker, as in (9).

(8) Ningyuan (SWM; Y. Wu 2009, p. 321)
他	送	一	杆	笔	给	我。
t’a33	soŋ213	i21	kan45	pi21	kə45	ŋo213
3sg	offer	one	clf	pen	dat	1sg
‘He offered a pen to me.’

(9) Ningyuan (SWM; Zhang 2009)
有	话	就	和	牛	讲。
iəu45	fa213	tɕiəu213	xo21	liəu21	tɕiaŋ45
have	speech	then	dat	buffalo	say
‘(He) talks to the buffalo when (he) has something to say.’

In (8), the dative marker is kə⁴⁵ 给, whereas in (9), the dative marker is xo²¹ 和. Note that the preverbal xo²¹ can have other interpretations as well, such as being a benefactive marker, as demonstrated by an example in (10).

(10) Ningyuan (SWM; Zhang 2009)
这里	景子	好，	你	和	我	照	个	相。
tɕi213li45	tɕin45tsɿ45	xau45	li45	xo21	ŋo45	tɕiau213	ko213	ɕiaŋ213
here	view	good	2sg	ben	1sg	take	clf	photo
‘The view is great here, take a photo for me.’

A benefactive marker is a preverbal marker, which marks the beneficiary. In this paper, we consider it as a distinct class of markers from dative markers. However, it is worth noting that a benefactive marker can develop into a preverbal dative marker, as is the case in Ningyuan.

An allative marker expresses “the meaning of motion ‘to’ or ‘towards’ a place” (Crystal 2003, p. 19). As such, it marks the goal in a theme-goal construction, for example, as tau⁴⁵ in Ningyuan.

(11) Ningyuan (SWM; Y. Wu 2009, p. 321)
我	给	毛毛崽	给	倒	床	高头	要	不	要得?
ŋo45	kə45	mau21mau21tsæ45	kə45	tau45	ts’uan21	kau33t’əu21	iau213	pu21	iau213tə21
1sg	om	baby	put	all	bed	on	ok	neg	ok
‘Is it ok that I put the baby on the bed?’

A passive marker is used to mark the agent in a passive construction. Sometimes, the passive marker shares the same form as the dative marker in the Sinitic languages. For instance, the dative marker kə⁴⁵ can also be used as a passive marker.

(12) Ningyuan (SWM; Y. Wu 2009, p. 321)
杯子	给	他	打烂	呱	了。
pei33tsɿ45	kə45	t’a33	ta45-lan213	kua21	liau45
cup	pass	3sg	hit-be.broken	cmpl	crs
‘The cup was broken by him.’

A differential object marker or disposal marker marks the object in a transitive sentence or the T argument in a ditransitive or theme-goal sentence. It might also share the same form as the dative marker. In Ningyuan, kə⁴⁵ is also used as a differential object marker, as already shown in (11). Another example is given in (13).

(13) Ningyuan (SWM; Zhang 2009)
给	那	本	书	拿	过来。
kə45	la213	pən45	ɕy33	la21	ko213-læi21
om	that	clf	book	take	pass-come
‘Bring that book over here.’

The forms of the dative markers of the 30 Hunan Sinitic languages are listed in

Table 1. Dative, benefactive, allative, passive, and object markers and GIVE verbs in the sample of 30 languages.

		DAT	BEN	ALL	OM	PASS	VGIVE
1.	Huarong SWM	tɛ13 得		tɛ13 得		pɑ21tɛ33 把得	pɑ21 把	Own field notes
2.	Changde SWM	kən55 跟	kən55 跟				pa21 把	Yi (2007); Zheng (1999)
3.	Yueyang Xiang	tə 得		得 tə		把得pa52tə	pa52 把	Fang (1999)
4.	Linxiang Gan	teʔ55 得		得 teʔ55			pa31 把	Li (2016)
5.	Cenchuan Xiang	pɑ3 把			pɑ3 把		pɑ3 把	Peng (2022)
6.	Yiyang Xiang	pa41 把			pa41 把	pa41 把 pa41tə45 把得	pa41 把	Cui (2009)
7.	Pingjiang Gan	pa35 把			pa35 把		pa35 把	Z. Wang (2009)
8.	Changsha Xiang	pa21 把			pa21 把	pa41tɤ24 把得	pa21 把	Y. Wu (2009); Wu (2011)
9.	Xiangxiang Xiang	xa4 给			xa4 给	xa4 给	po21 把	F. Wang (2009)
10.	Liuyang Gan	pa24 把			pa24 把	pa24 把	pa24 把	Bei and Xiang (2009)
11.	Xinhua Xiang	læ13 来					lɔ21 拿	Luo and Zou (2009)
12.	Lianyuan Xiang	xɑ45 哈			xɑ45 哈	xɑ45 哈	lɛ44 拿	Chen (2002, 2009)
13.	Loudi Xiang	sɿ5 赐				nõ44sɿ5 拿赐	nõ44 拿	F. Peng (2009)
14.	Longhui Xiang	bɑ13 把			pɑ13 把	bɑ13 把 bɑ13te55 把得	bɑ13 把	Ding (2009)
15.	Shaoyang Xiang	pa42 把			pa42 把	pa42 把	pa42 把	G. Li (2009)
16.	Shaodong Xiang	pa31 把			pa31 把	pa31 把	pa31 把	Sun (2009)
17.	Hengyang Xiang	te22 得				te22 得	te22 得	Peng (2005); Y. Peng (2009)
18.	Hengshan Xiang	tæ24 得				tæ24 得	tæ24 得	Mao (2009)
19.	Youxian Gan	te44 得				te44 得	te44 得	Dong (2009)
20.	Changning Gan	te33 得			te33 得	te33 得	te33 得	Q. Wu (2009)
21.	Xintian Tuhua	kəu35 □7		kəu35 □			oŋ35 安	Q. Xie (2009)
22.	Guiyang Tuhua	ta45 带	ta45 带		ta45 带		uã33 弯	Deng (2009)
23.	Rucheng Hakka	nɔ44 拿			nɔ44 拿	nɔ44 拿	nɔ44 拿	Zeng (2006)
24.	Ningyuan SWM	kə45 给			kə45 给	kə45 给	kə45 给	Zhang (2009)
25.	Suining SWM	pa55 把				pa55 把	pa55 把	Zeng and Li (2009)
26.	Chenxi Xiang	ko324 过					ko324 过	B. Xie (2009)
27.	Fenghuang SWM	paŋ55 帮kən55 跟	paŋ55 帮kən55 跟		paŋ55 帮kən55 跟		fən55 分	Q. Li (2009)
28.	Jishou SWM	paŋ55 帮	paŋ55 帮 kən55 跟		paŋ55 帮		fən55 分	Li (2002)
29.	Waxiang Unclassified	kai55 跟	kai55 跟		kai55 跟		tɤ55 得	Chappell et al. (2011)
30.	Tasha SWM	ke53 给	ke53 给				ko24 过	Own Field notes

. Except for the GIVE verbs, the other elements examined in this paper, namely benefactive, allative, object, and passive markers, generally have several forms that developed from different sources. Since we are investigating the dative markers in this paper, we only list the forms that are identical or related to the dative markers. However, note that an empty cell may also indicate that we have not found the relevant form in the literature.

In the next section, we discuss the five different types of dative markers and their distribution.

3. The Five Types of Dative Markers and Their Areal Distribution

Based on the analysis of 30 Sinitic languages spoken in Hunan, which include Xiang, Gan, Southwestern Mandarin, Hakka, Tuhua, and one unclassified Sinitic language, we identify five patterns for the dative markers, as detailed in

Table 2. Five types of dative markers.

	Patterns	Languages	Number of Languages
I	DAT = ALL	Huarong (tɛ13 得), Yueyang (tə 得), Linxiang (teʔ55 得), Xintian (kəu35 □)	4
II	DAT = GIVE = OM ≠ PASS	Cenchuan (pɑ3 把), Pingjiang (pa41 把), Changsha (pa21 把)	3
III	DAT = GIVE = PASS = OM	Longhui (pa13 把), Liuyang (pɑ24 把), Shaoyang (pa42 把), Shaodong (pa31 把), Yiyang (pa41 把), Ningyuan (kə45 给), Rucheng (nɔ44 拿), Changning (te33 得)	8
IV	DAT = GIVE = PASS ≠ OM	Hengyang (te22 得), Hengshan (tæ24 得), Youxian (te44 得), Suining (pa55 把)	4
V	DAT = BEN	Guiyang (ta45 带), Fenghuang (paŋ55 帮/ kən55 跟), Jishou (paŋ55 帮/ kən55 跟), Waxiang (kai55 跟), Tasha (ke53 给), Changde (kən55 跟)	6

. The dative markers are given after each language. Following this table, the areal distribution of these four types is presented in Figure 2.

3.1. Type I: DAT = ALL (4/30)

In Type I languages, the dative marker is the same as allative marker. In our sample, this type is mainly found in the northeast of Hunan: Huarong (SWM), Yueyang (Xiang), Linxiang (Gan), apart from Xintian (Tuhua) in the south. Take Huarong, for example.

(14) Huarong (SWM; own field work)
Dative construction
买	哒	个	褂子	得	我。
mai21	ta21	ko33	kuɑ24tsɿ33	tɛ13	ŋo21
buy	pfv	clf	coat	dat	1sg
‘(He) bought me a coat.’

(15) Allative construction
我	挂	哒	一	张	全家福	得	客厅	的
ŋo21	kuɑ24	tɑ21	i45	tsɑŋ53	tɕɦyn13tɕia53fu13	tɛ13	kɦɛ24tɦiɛn53	ti33
1SG	hang	PFV	one	CLF	family.photo	ALL	living.room	POSS
墙上。
tɕɦiɑŋ13 = sɑŋ21
wall = on
‘I hung a family photo on the wall in the living room.’

In this type of language, the GIVE verb is always BA 把, and three of the four languages use DE 得 as the dative marker, while one language (i.e., Xintian) uses kəu³⁵ whose etymological source is not clear. In Section 4.1, we will discuss the developmental path of DE as a dative marker in this type of language, using Huarong as an example.

3.2. Type II: DAT = GIVE = OM ≠ PASS (3/30)

There are three languages of Type II in our sample, they are Cenchuan (Xiang), Changsha (Xiang), and Pingjiang (Gan) which are in the north central region of the Hunan Province. They use BA 把 as the GIVE verb as well as the dative marker. Take Changsha as an example.

(16) Changsha (Xiang; Wu 2011, p. 188; Y. Wu 2009, p. 309)
Dative construction
他	送	哒	三	只	鸡	把	我。
tha33	sən45	ta21	san33	tsa24	tɕi33	pa41	ŋo41
3sg	offer	pfv	three	clf	chicken	dat	1sg
‘He gave me three chickens as a gift.’

(17) Differential object marking construction
把	窗户	打开。
pa41	tɕhyan33fu	ta41-khai33
om	window	make-open
‘Open the window.’

(18) Passive construction
杯子	把得/捞/听	他	打烂	哒。
pei33tsɿ	pa41tɤ24/lau33/t’in45	t’a33	ta41-lan21	ta21
glass	pass	3sg	hit-be.broken	crs
‘The glass was broken by him.’

Note that, as shown in (18), the compound form pa⁴¹tɤ²⁴ can be used as the passive marker, and tɤ²⁴ 得 is exactly the allative marker in Changsha. We can consider the dative use of pa⁴¹ 把 is based on the compound GIVE verb pa⁴¹tɤ²⁴, which gradually lost its allative element in it, hence pa⁴¹ alone is used as the dative marker. In Cenchuan and Pingjiang, we do not have the necessary data to tell if they have or had a compound form.

3.3. Type III: DAT = GIVE = PASS = OM (8/30)

The dative markers of the Type III languages are the same as their passive markers and differential object markers. In our sample, there are actually 10 languages mostly spoken in the central and south of Hunan that share the pattern. Nonetheless, as introduced in Section 2, the Type II, III, and IV languages are refined from the pattern: DAT = GIVE, so, for this reason, we exclude Xiangxiang and Lianyuan, which use the same marker for dative, passive, and object marking but have a distinct verb form for GIVE. As a result, we can only classify eight languages into this type. These are three Xiang varieties: Shaoyang, Shaodong, and Lianyuan; one Southwestern Mandarin variety: Ningyuan; two Gan varieties: Liuyang and Changning; and one Hakka variety: Rucheng. Take Shaodong, for example.

(19) Shaodong (Xiang; Sun 2009, pp. 105–12)
Dative construction
咯	只	衣衫	其	送	把	小明	哩。
ko31	tɕia55-31	iɛn55san31	tɕi31	səŋ35	pa31	ɕio31min12	li
this	clf	clothes	3sg	offer	dat	XiaomingNAME	sfp
‘He gave this clothing to Xiaoming.’

(20) Passive construction
其	只	脚	把	车子	闯	哩，	现唧	还	在
tɕi31	tɕia55-31	to55	pa31	t’ei55tsɿ	ts’aŋ31	li	ɣiɛn12·tɕi	ɣa12	dzei12
3SG	clf	foot	pass	car	hit	sfp	now	still	LOC
医院里			诊。
i55yɛn35-55 = li31			taŋ3
hospital = inside			treat
‘His foot was hit by a car, and he is still in the hospital.’

(21) Differential object marking construction
要	妹妹	把	屋	扫	一	下
io35	mei35·mei	pa31	u55	sau35	i55	ɣa24
make	little.sister	om	house	sweep	one	vcl
‘Make little sister to clean up the house.’

Generally speaking, the dative marker in Type III languages also shares the same form as the GIVE verb. The GIVE verbs in this type are not limited to the types that are originally TAKE verbs (e.g., pa³¹ 把 in Shaodong or te³³ 得 in Changning). In Ningyuan, it is kə⁴⁵ 给 (which shares the same etymology as gěi in Standard Mandarin) that is used as the GIVE verb, and it can also be used as a dative marker, a passive marker and a differential object marker.

For this type of language, there are three possibilities. The dative marker could be developed from a genuine GIVE verb, like the case of Ningyuan just mentioned; or either the GIVE_<TAKE verbs have been shifted to genuine GIVE verbs, and then developed a dative use; or the GIVE_<TAKE verb combines an allative form and becomes a compound genuine GIVE verb, then the compound form loses the allative element and, subsequently, the GIVE_<TAKE verb itself becomes the dative marker.

3.4. Type IV: DAT = GIVE = PASS ≠ OM (4/30)

In Type IV languages, the dative markers are the same as passive markers, but different from the object markers. This type is mainly found in southern Hunan, such as in Hengshan (Xiang), Hengyang (Xiang), Youxian (Gan), Changning (Gan), and Suining (SWM).

(22) Youxian (Gan; Dong 2009, pp. 36–38; Y. Wu 2009, p. 313)
Dative construction
其	送	支	笔	得	我。
tɕi51	səŋ11	tsɿ	pi44	te44	ŋo11
3sg	offer	clf	pen	dat	1sg
‘He offered me a pen.’

(23) Passive construction
小芳	得	爱婆	接走	哩。
ɕiau51faŋ44	te44	ŋəø11p’o213	tɕie44-tsei51	li
XiaofangNAME	pass	grandmother	pick.up-be.away	sfp
‘Xiaofang was picked up by her grandmother.’

(24) Differential object marking construction
我	把	毛毛	放	到	床上	要得	不？
ŋo11	pa51	mau213mau	faŋ11	tau	t’aŋ213=ɕiaŋ	iau11te	pu
1sg	om	baby	put	all	bed=on	ok	neg
‘Is it okay if I put the baby on the bed?’

In all the six languages of this type, the GIVE_<TAKE verbs have developed both dative and passive uses. In this group, except for Suining, which uses pa⁵⁵ 把 as a verb of giving, the other seven languages all use DE 得. Both the dative marker and the passive marker share the same form with the GIVE_<TAKE verb. In Section 4.2.2, we will take Hengyang as an example to discuss how DE develops into a dative marker from GIVE_<TAKE in this type of languages.

3.5. Type V: DAT = BEN (6/30)

This type of language is very easy to differentiate from the other types because it uses a preverbal dative construction. Note that, for some of these languages, the dative marker may also have the same form as the object marker, but since it is a preverbal marker, and it is not derived from the GIVE verb, we do not classify these languages into Type II languages. There are six languages in our sample that share this pattern, and they are found in the northwest of Hunan. Their GIVE verbs are more diversified, and sometimes the etymological sources for the GIVE verbs are not so clear. For instance, Guiyang (Tuhua) uses ta⁴⁵ 带 or uã³³ 弯 as the GIVE verb, but when it forms a ditransitive construction, it has to use [ta⁴⁵ + R + uã³³ + T]; Fenghuang (SWM) and Jishou (SWM) use fən⁵⁵ 分 as the GIVE verb, Waxiang (unclassified) uses tɤ⁵⁵ 得, Tasha (SWM) uses ko²⁴ 过, and Changde (SWM) uses pa²¹ 把.

For the six languages of Type V, the dative markers have not developed from the GIVE verbs. The main ditransitive constructions for them are formed on the basis of the benefactive construction. Take Jishou, for instance.

(25) Jishou (SWM; Li 2002, p. 318)
Benefactive construction
医生	倒	帮	他	看	病？
i55sən55	tau	paŋ55	t’a55	k’an35	pin35
doctor	prog	ben	3sg	see	illness
‘The doctor is treating him.’

(26) Dative construction
我	一	到	学校，	就	帮	屋里	打
ŋo42	i11	tau35	ɕio11ɕiau35	tɕiəu35	paŋ5	u11li	ta42
1SG	once	arrive	school	then	dat	home	make
了	个	电话。
lə	ko	tian35xua3
pfv	clf	phone.call
‘As soon as I got to school, I called home.’

The verbs that are used in the benefactive construction are transitive verbs, but they do not necessarily indicate transfer (Zhu 1979). The argument introduced by the benefactive marker is a beneficiary. As we can see from (25), the verb is ‘to see (the patient), to treat (the illness)’, and 3rd person singular is the beneficiary, while for the dative construction, the verbs are either intrinsically ditransitive verbs that express transfer or verbs of saying that concern an addressee. As shown in (26), the verb is ‘to call’, and the preverbal argument is an addressee.

In this group of languages, dative markers are developed from preverbal benefactive markers and have little to do with GIVE verbs in the given language. Four of these languages can use GEN 跟 as a preverbal dative marker, and one uses GEI 给. In Section 4.3, we will discuss the grammaticalization path of kai⁵⁵ 跟 in Waxiang as an example.

3.6. Interim Summary: The Areal Distribution

Except for Xiangxiang and Lianyuan, which we exclude from the Type III languages, there are other three languages in our sample that do not fit into any types that we have classified. They are three Xiang varieties: Loudi, Xinhua, and Chenxi. In Loudi, the dative marker is another verb of giving sɿ⁵ 赐, while the GIVE_<TAKE verb that is used in the ditransitive construction is nõ⁴⁴ 拿, which can also be used as a differential object marker. The postverbal dative construction in Loudi can be templated as [nõ⁴⁴ + T+ sɿ⁵ + R], and the passive marker is the compound form nõ⁴⁴sɿ⁵. Note that in Modern Chinese, the compound form bǎyǔ 把与 which combines a TAKE verb and a former GIVE verb has also existed (Mao 2022). sɿ⁵ no longer has a verbal use in Loudi, which suggests that it is probably a GIVE verb belonging to an older layer. In Xinhua at an earlier stage, the dative marker is læ¹³ 来, and it is different from the GIVE verb lɔ²¹ 拿or the allative marker tɔ⁴⁵ 到. In some varieties of Wu, 来 is identical to the allative marker that is also used as dative marker (Chin 2010). It is possible that læ¹³ was the allative marker in Xinhua before, but since we do not have any record in the literature, we cannot include Xinhua into Type I language. In Chenxi, the GIVE verb and the dative marker are both ko³²⁴ 过, producing the dative construction is [ko³²⁴ + T+ ko³²⁴ + R]. At the same time, the double object construction is also attested in Chenxi as [ko³²⁴ + R + T], but the detailed uses of ko³²⁴ and its developments cannot be found in the literature, meaning that we cannot draw a credible conclusion as to how it developed the dative use. So, in our discussion, we exclude these three languages.

To conclude, we classify the languages in our sample into five main groups according to the dative constructions and the polyfunctionality of dative markers. The Type I languages have a dative marker that is identical to the allative marker, and they are mainly found in northeastern Hunan. In the Type II, III, and IV languages, the dative markers have the same form as the GIVE verbs. The dative markers in Type II languages are identical to their differential object makers. There was probably a compound GIVE verb in the history for these languages. As for the Type III languages, they are spread over central and southern Hunan. Their dative markers share the same form as the passive markers and the differential object markers. This pattern can be explained by two possible grammaticalization pathways. The type IV languages are mostly found in southern Hunan, and their dative markers are the same as the passive markers but different from the object markers. The Type V languages have various sources for GIVE verbs, and their main type of ditransitive constructions is the preverbal dative constructions. They are found in northwestern Hunan.

4. Diachronic Sources and Developments for the Dative Markers

In this section, we discuss the pathways of development for dative markers. There are three main grammaticalization pathways behind the synchronic distribution of different patterns of dative markers, as shown below. We will discuss these pathways with a language in our sample one by one.

• Allative > Dative
• (TAKE >) GIVE > Dative
  • [TAKE-ALL] > GIVE > Dative; TAKE > Dative
  • (TAKE >) GIVE > Dative
• Benefactive > Dative

4.1. ALL > DAT

Using the allative marker to mark the recipient is common in the languages across the world (Kuteva et al. 2019; Rice and Kabata 2007), such as to in English or à in French. Malchukov et al. (2010, p. 52) also show that the theme-goal construction and the ditransitive construction often overlap in a semantic map for many languages, such as in Finnish. Allative markers being the source for the dative markers in Sinitic languages has already been discussed in the literature. For example, Chin (2010) claimed there are two types of indirect object marker (i.e., dative marker) in Chinese: the go-type and the give-type. He mentioned that YU 于 in the oracle-bone inscriptions, LAI 来 in 17th century Wu, DU 度 in 16th century Min and GUO 过 in 19th century Yue are all directional verbs that can be used as dative markers. Li and Wu (2015) also mentioned that the directional element ku⁴² 过 can be used as a dative marker in Yichun Gan. For some languages spoken in northeastern Hunan, the theme-goal construction and the ditransitive construction can both be realized with the same syntactic template and share a same marker. What the literature lacks is an exploration of the developmental stages through which an allative marker acquires a dative use.

We take Huarong as an example and show how the allative tɛ¹³ 得 developed its dative use. In Huarong, tɛ¹³ can be used as a verb which means ‘get, obtain’, see an example in (27).

(27) Huarong (SWM; own fieldwork)
他	得	哒	头名。	(GET)
lɑ33	tɛ13	tɑ21	ɣou13miɛn13
3sg	get	pfv	first.rank
‘He won the first place.’

It can be used as an allative or dative marker either directly after the verb or after the T argument. The former case is more frequently found in my corpus of Huarong and in the literature of other languages using DE 得as an allative/dative marker.

(28)	他	把	车	停	得	路	边上	哒。	(Allative)
	lɑ33	pɑ21	tsɦɛ53	tɦiɛn13	tɛ33	lou33	pin53sɑŋ21	tɑ21
	3SG	OM	car	park	ALL	road	side	CRS
	‘He parked the car on the side of the road.’

(29)	他	捡	哒	两	百	块	钱	得	枕头
	lɑ33	tɕin21	tɑ21	liɑŋ21	pɛ45	kɦuai21	tɕɦin13	tɛ13	tsən21lou33
	3sg	hide	pfv	two	hundred	currency.unit	money	all	pillow
	下头。
	ɕia33lou21
	below
	‘He hid two hundred yuan under the pillow.’						(Allative)

(30)	他	书	把	得	我	哒。	(Dative)
	lɑ33	ɕy53	pɑ21	tɛ33	ŋo21	tɑ21
	3sg	book	give	dat	1sg	crs
	‘He gave the book to me.’

(31)	他	送	哒	一	包	烟	得	爷。	(Dative)
	lɑ33	sɤŋ24	tɑ21	i45	pɑu53	iɛn53	tɛ13	ia13
	3sg	offer	pfv	one	packet	cigarette	dat	father
	‘He offered a packet of cigarette to father.’

In the diachronic data, DE firstly appeared after the verbs having the meaning ‘to get’ in the Western Han Dynasty (202BC-8AD), as in (32). Then, it is found after the verbs which do not have the GET meaning; it can be analyzed as a phase complement which expresses the phase of an action in the first verb (Chao 1968), see (33).

(32)	赵	使	人	微	捕得	李牧
	zhào	shǐ	rén	wēi	bǔ-dé	Lǐ Mù
	Zhao	order	people	secretly	catch-get	Li Mu
	‘Zhao ordered people to secretly arrest Li Mu.’
	(史记·廉颇蔺相如列传 [Records of the Grand Historian: Biography of Lian Po and Lin Xiangru], cited from Ma 2003, p. 20, glossing is mine)

(33)	武丁	夜	梦得	圣人
	Wǔ Dīng	yè	mèng-dé	Shèngrén
	Wu Ding	night	dream-get	saint
	‘Wu Ding dreamed of a saint at night.’
	(史记·殷本纪 [Records of the Grand Historian: Biography of Yin Ben], cited from Ma 2003, p. 22, glossing is mine)

Note that in the historical documents for Mandarin, DE has not developed an allative use to mark a goal. However, Lamarre (2009) discussed the goal marker in the Northern Sinitic languages, such as de 的 in Beijing Mandarin (Zhu 2009, p. 115; Xu 1994), təʔ⁴ 得 in the Jin variety of Shenmu (Xing 2000; 2002, p. 596), or DE 得 in the Xiang variety of Changsha (Wu 2011, pp. 231–48). The origin of these goal markers is difficult to identify. Xu (1994) argues that de in Beijing Mandarin comes from ZHE 着, which is used as a preposition in the Dunhuang Variant Texts Collection (Dūnhuáng biàn wén敦煌变文), while Xing (2000, 2002) and Wu (2011) suggest that the source of the goal marker in Shenmu and Changsha is DE 得. Regardless of the etymological source, what these analyses have in common is that the goal marker comes from a bounded marker (yǒujiè biāojì有界标记 in Lamarre 2009) in the language. This is consistent with the use of DE 得as a phase complement that we mentioned in (33).

In fact, Wu (2001, pp. 53–54) proposed several phases for the diachronic development of the locative marker tɤ²⁴ 得 in Changsha:

• it was used as a main verb meaning ‘get’, ‘catch’.
• it was used as the second verb after a main verb with a meaning similar to ‘get’, such as ‘buy’, ‘have’.
• it was used as a verb complement to indicate the result or completion of an action. In this phase, it often took a locative noun as its object, as [V + DE + Place], and it could be reanalyzed as a preposition or an aspectual marker.

We agree with this analysis. We propose that from the phase complement in [V-DE], when DE is followed by a place, it is reanalyzed as an allative marker. Then, through metaphorical extension DE can be considered as a dative marker when it precedes a person instead of a place (a parallel development for 于 yú in Old Chinese is discussed in Ye 2020).

4.2. (TAKE >) GIVE > DAT

Dative markers developing from a genuine GIVE verb is a very common pathway described in the literature, but as we mentioned before, the GIVE verbs in the Northern and Southern Sinitic languages are different from a syntactic perspective. Hashimoto (1976) has pointed out that the northern GIVE verbs form double object constructions of the form: [GIVE + R + T], while the Southern GIVE verbs generally form inverted double object construction: [GIVE + T + R]. Zhang (2011) claims that the main difference for ditransitive constructions between Northern and Southern Sinitic languages is that the Northern Sinitic languages use double object construction, while the Southern Sinitic languages use postverbal dative construction [GIVE + T + DAT + R], and the inverted double object construction is developed on the basis of the postverbal dative construction with the dative marker omitted.

We discuss how a GIVE_<TAKE verb which can only precede the T argument in the ditransitive constructions becomes a dative marker which introduces the R argument. We propose that the dative marker is developed from the GIVE_<TAKE verb that appears in the second position of a serial verb construction [GIVE_<TAKE + T+GIVE + R]. It concerns two possible developments.

4.2.1. Allative Element Dropping off from the Compound GIVE_<[TAKE-ALL]

This development has been discussed by Li and Wu (2015) in Gan languages, as already shown in Section 2. For this pathway, the GIVE verbs are TAKE verbs in origin. The TAKE verb can only take a theme argument and cannot precede the recipient directly. When it forms a ditransitive construction, it borrows the allative marker as a dative to introduce the recipient, as in [TAKE + T + ALL + R]. Next, the TAKE verb combines with the allative marker and becomes a compound GIVE verb [TAKE-ALL] which can be further reanalyzed as a dative marker in V₂ position of a serial verb construction. Then, the compound GIVE verb loses the allative element after a long period of frequent use and leaves the TAKE verb itself to become a dative marker.

We cannot present all the stages with our data of Type II languages. As we can see in (18), a compound form pa⁴¹tɤ²⁴ 把得 exists in Changsha, but we cannot find an example with pa⁴¹tɤ²⁴ being a genuine GIVE verb or a dative marker. Only its use as a passive marker suggests that it was probably a genuine GIVE verb before.

Nevertheless, we can observe this development with the data in Yueyang. The primary dative marker is tə 得, as shown in (34). pa⁴²tə 把得 can be used as a GIVE verb, as shown in (35). In fact, according to my own fieldwork, among the young speakers of Yueyang, pa⁴² 把 is also accepted as a dative marker, as shown in (36).

(34) Yueyang (Xiang; Fang 1999, pp. 222–24, own field work)
拿	件	衣	得	我。	[tə + R]
na55	tɕ’ian33	i45	tə	ŋo42
take	clf	clothes	dat	1sg
‘Give me a shirt.’

(35)	橘子	把得	他	哒。	[pa42tə + R]
	tɕy55tsɿ	pa42tə	la33	ta
	clementine	give	3sg	crs
	‘(Someone) gave the clementine to him.’

(36)	借	本	书	把	我。	[pa42 + R]
	tɕia324	pən42	ɕy45	pa42	ŋo42
	lend	clf	book	dat	1sg
	‘Lend a book to me.’

Changsha, Cenchuan, and Pingjiang probably share this pathway for their dative marker 把 BA, based on the geographic location of the languages and possible development.

4.2.2. Relexicalization

In this pathway, the TAKE verb gained a GIVE meaning through relexicalization. Güldemann (2012) discussed the polysemy of a ‘take/give’ verb in Tuu languages spoken in southern Africa. See two examples from Taa (Güldemann 2012, p. 73) with the verb /uM~!ãM (regularly followed by the dative preposition n/aM)8.

(37) Taa (Tuu languages; southern Africa)
Si	/oe	si n//au	/’ang	ʘuru
1p.e	hold.s:3>	problem.3	com:1s	offspring.p
‘We get/have problems with my children.’

(38)	suu	si	/ui	tuu
	feed.first.time	ipfv	?GIVE~TO:1>	people.1
	‘Purifying the people’ [lit.: feed to the people]

He proposed that the ‘obtainment-possession’ meaning is the original one, and the ‘give’ reading of the ‘take’ verb is triggered by its recurrent use in the ditransitive construction with a dative element. In other words, it is induced by a syntactically coerced semantic re-analysis.

Newman (1996, pp. 50–60) also demonstrated that TAKE and GIVE verbs are semantically close from a cognitive perspective: the participants (i.e., Giver, Theme, Recipient) for TAKE and GIVE remain the same within the same spatio-temporal domain.

From TAKE to GIVE has been reported in Sinitic languages, such as Shaowu (Ngai 2015, 2021), and Hengyang (Yang and Peng 2021). Take Hengyang, for example. te²² 得 can be used as a verb of GET or GIVE, dative marker, causative verb, and passive marker.

(39) Hengyang (Xiang; Peng 2005, pp. 143–45)
我	得	哒	一	百	块	钱	奖。	(GET)
ŋo33	te22	ta22	i22	pe22	k’uai33	tɕien11	tɕian33
1sg	get	pfv	one	hundred	currency.unit	money	prize
‘I won a prize of one hundred yuan.’

(40)	冇	得	一	分	钱	得	我。	(GIVE and Dative)
	mau213	te22	i22	fən45	tɕien11	te22	ŋo33
	neg	give	one	currency.unit	money	dat	1sg
	‘(Someone) didn’t give me a penny.’

(41)	你	就	得	其	骂	也	冇得	关系	唦。	(Causative)
	ni33	tɕiu213	te22	tɕi33	ma213	ia33	mau213te22	kuen45·ɕi	sa11
	2sg	just	let	3sg	scold	also	not.have	affect	sfp
	‘It is fine to just let him scold you.’

(42)	我	走	箇	来	冇	两	天	就	得	你
	ŋo33	tsəu33	ko33	lai11	mau213	lian33	t’ien45	tɕiu213	te22	ni33
	1sg	walk	here	come	neg	two	day	then	pass	2sg
	骂	一	顿，	我	还	不	如	不	来。
	ma33	i22	tən24	ŋo33	xai11	pu22	ɕy11	pu22	lai11
	scold	one	vcl	1sg	rather	neg	as	neg	come
	‘I haven’t even been here two days and you’ve scolded me. I might as well not have come.’
	(Passive)

Since there is no compound form mentioned in Hengyang from the literature, it is less likely that te²² develops a dative use through a compound GIVE verb, and the object marker in Hengyang is another verb of taking (i.e., lau⁴⁵ 㧯), which means te²² is more likely to turn into GIVE before developing an object marking use based on TAKE. We consider that te²² first became a genuine GIVE verb then developed the dative use from GIVE. Most languages that share this pathway use DE 得 as their GIVE verb.

4.3. BEN > DAT

Zhang (2011) mentioned that in some Sinitic languages spoken in Hunan and Hubei, instead of using the postverbal dative construction that commonly used in Southern Sinitic languages, they tend to use preverbal dative construction to express transfer. Huang (2021) investigated the sources for the preverbal dative markers in Sinitic languages, and she indicates that a common developmental chain for Southern Sinitic languages is: Comitative > Benefactive > Dative. From Map 96 of Linguistic Atlas of Chinese Dialects (Grammar Volume) (Cao 2008), we can also observe this tendency.

In our data, the Type V languages: Fenghuang, Jishou, Waxiang, Tasha, and Changde all share this pathway. Take Waxiang, for example.

(43) Waxiang (Unclassified; Chappell et al. 2011)
是	□	跟	我	担	的	水。	(Benefactive)
tshɛ25	zɤ13	kai55	u25	toŋ55	ti	tsu25
be	3sg	ben	1sg	carry	sp	water
‘He’s the one who carried the water for me.’

(44)	□	跟	我	得	件	衣。	(Dative)
	zɤ13	kai55	u25	tɤ55	tɕhia41	i55
	3sg	dat	1sg	give	clf	clothes
	‘He gave me a shirt.’

Preverbal dative constructions are very commonly used in the northwest of China, as shown in Map 96, such as in Lanzhou, Xining. There is a famous linguistic area in the northwest China, i.e., ‘Qinghai-Gansu Sprachbund’, and the Sinitic languages in this region have been influenced by Mongolian, Turkic, and other non-Han languages with SOV word order; as a result, some of the varieties of Sinitic languages have also shifted to SOV order and have even developed case markers (Xu 2015). Preverbal adpositional phrase is in harmony with OV word order (Dryer 1992). The use of preverbal dative construction in this region is reasonable. Zhang (2011) also tried to explain the similar situation in Hunan and Hubei by language contact, he indicated that these Sinitic languages with preverbal dative constructions are probably influenced historically by a Tibeto-Birman language with OV order, i.e., Tujia. It is possible that there existed a substratum of Tujia in Hunan, Hubei, and Guizhou. The Tujia people have lived in this area throughout many centuries, and many of the historical and cultural customs of the Tujia people have spread to the neighboring Han Chinese communities, but this remains to be proved by subsequent research.

4.4. Areal Distribution According to the Diachronic Developments

We can make three observations from the data. First of all, looking at the various types of patterns, we can note two different directions of development. From north to south, Type I languages can develop into Type II languages, and the dative markers of Type II languages are formed by the loss of allative element on the basis of a compound GIVE verb: [TAKE-ALL]. This process can actually be observed in Yueyang, as shown in Section 4.2.1. On the basis of Type II, the GIVE_<TAKE can further develop into a genuine GIVE verb which may subsequently develop into a passive marker, as the case of some Type III languages, such as in Yiyang (Shi and Wang 2009) and Longhui (Ding 2006, p. 84), a compound dative form ba¹³te⁵⁵ 把得 is reported. From the south to the north, the GIVE_<TAKE verbs of Type IV languages are directly transformed into genuine GIVE verbs through relexicalization, and further develop the dative and passive use. We note that most of the GIVE verbs in these languages are DE 得, only Suining uses pa⁵⁵ 把. Type IV languages may also develop further into Type III languages in the north, having object marker grammaticalized from the GIVE verb through the intermediate stage of benefactive. It might be the case that for Changning, the GIVE verb te³³ 得 is attested as a dative marker, a passive marker, a benefactive marker, and an object marker.

Another interesting phenomenon that we can observe is about the GIVE_<TAKE verb BA 把 is that from north to south, from the syntactic perspective, the TAKE use of BA is gradually weakened, while the GIVE use is gradually strengthened. For example, in Type I languages, e.g., Huarong, pa²¹ 把 can only be followed by the T argument, which has neither dative nor passive use, and the genuine GIVE verb is pa²¹tɛ³³ 把得, whereas in Type IV languages, e.g., Suining, pa⁵⁵ 把 is more like a GIVE verb, as it can act as both a dative and a passive marker, and the object marker is developed from another TAKE verb tan⁵⁵ 担.

Finally, we can find that DE 得 as a dative marker actually involves two different evolutionary pathways; one is represented by Type I languages, where the dative use is based on the allative use, and the other is represented by Type III languages, where the dative use is derived from DE being a genuine GIVE verb.

5. Conclusions

In this paper, we investigated the dative markers of 30 Sinitic languages in the Hunan Province, including 14 Xiang varieties, five Gan varieties, seven SWM varieties, one Hakka variety, two Tuhua varieties, and one unclassified Sinitic language in the Hunan Province. Five patterns have been observed, and the patterns show a regular distribution according to their geographic locations instead of the affiliations. We also identified three main grammaticalization pathways behind the patterns. See

Table 3. The types and developments of dative markers in Hunan.

Type I	DAT = ALL	a. ALL > DAT
Type II	DAT = GIVE = OM ≠ PASS	b. GIVE > DAT	(i) [TAKE-ALL] > GIVE > DAT; TAKE > DAT
Type III	DAT = GIVE = PASS = OM
			(ii) (TAKE >) GIVE > DAT
Type IV	DAT = GIVE = PASS ≠ OM
Type V	DAT = BEN	c. BEN > DAT

for the summary of the pathways behind each type of pattern.

For the Type II, III, and IV languages, the dative markers have the same form as their GIVE_<TAKE verbs, but it concerns two different grammaticalization pathways. One is that the original TAKE verb combines with an allative element and this compound form becomes a genuine GIVE verb, which can become a dative marker to introduce the R argument, then the allative dropped from the compound form and left the original TAKE verb as a dative marker. The second pathway is that the TAKE verb first became a genuine GIVE verb through relexicalization, then the dative use was developed from the genuine GIVE verb. Type III languages may involve these two different pathways.

We can see that the most used dative marker is DE 得 (attested in seven languages) and BA 把 (found in nine languages). The dative use of DE in northern Hunan is developed from its allative use, while in southern Hunan, DE grammaticalized into dative marker from the GIVE verb use. BA as a GIVE verb is widely spread in Hunan, but in northern Hunan, BA only gained the semantic meaning of giving and cannot be really considered as a GIVE verb which can take an R argument; however, in the south, the GIVE use of BA becomes mature: it can not only be used as a dative marker to introduce the recipient, but also develops a passive use. We claim that in the Hunan Province, from north to south, BA is gradually shifting from TAKE to GIVE.

Finally, we remarked on a pattern that might be induced by language contact from Tujia, i.e., in northwestern Hunan; different from other Sinitic languages spoken in Hunan, four languages in our data tend to use a preverbal dative construction to encode transfer, and the source for the dative marker is the benefactive marker.

Hunan has been regarded as a transitional zone in Chappell (2015) by three constructions: differential object marking, passive, and comparative constructions. How to refine the linguistic areas there relies on other features. Our examination of the distribution and origin of patterns with dative markers helps the further finer classification of the Sinitic languages in Hunan, as well as to probe the historical layers and developments for those languages. In addition, double object constructions as well as GIVE verbs have long been important in the research for linguistic geography, e.g., (Hashimoto 1976; Szeto 2019). However, the literature has not explored much the development of the dative markers. We hope this paper will be an important addition to the exploration of dative marking patterns by combining dative markers with GIVE verbs, allative markers, passive markers, and differential object markers.

Further extensive inquiry and investigation into the patterns of dative markers will undoubtedly serve to rigorously test and refine the grammaticalization chains outlined in our analysis, not only for other Sinitic languages and regions but also for broader linguistic contexts.